CENTER FOR DRUG EVALUATION AND
RESEARCH

APPLICATION NUMBER:

2064880rig15000

SUMMARY REVIEW

 

ill"?
#3 Dirt}

~29 
DEPARTMENT OF HEALTH AND HUMAN SERVICES

I
150?? 

 

9+

.Q
Food and Drug Administration
Silver Spring, MD 20993

TO: Janet Woodcock, M.D., Director, CDER
Ellis Unger, M.D., Director, Of?ce of Drug Evaluation 1, CDER
Luciana Borio, M.D., Chair, Agency Scienti?c Dispute Process Review Board

FROM: Robert M. Califf, M.D., Commissioner of Food and Drugs

RE: Scienti?c Dispute Regarding Accelerated Approval of Sarepta Therapeutics?
Eteplirsen (NDA 206488) Commissioner?s Decision

DATE: September 16, 2016

Overview

The Chair of the Agency Scienti?c Dispute Process Review Board (SDR Board), Dr. Luciana
Borio, has forwarded to me the Board?s recommendation regarding the appeal by Dr. Ellis F.
Unger, the Director of the Of?ce of Drug Evaluation 1, of Dr. Janet Woodcock?s decision;
namely, that Sarepta Therapeutics? drug eteplirsen meets the standard for
accelerated approval for treatment of Duchenne muscular (Duchenne or DMD) in
patients who have a con?rmed mutation ofthe gene amenable to exon 51 skipping,
which affects about 13 percent of the population with DMD. Duchenne is a rare genetic disorder
characterized by progressive muscle deterioration and weakness and is often fatal. Dr. Unger
disagrees with Dr. Woodcock?s View and would decline to approve eteplirsen at this juncture.I

The SDR Board?s memorandum of August 8, 2016 concludes that under the standard for review
set out in Staff Manual Guide for formal appeals of internal FDA scienti?c disputes, Dr.
Unger had an adequate opportunity to present his scienti?c concerns within the Center for Drug
Evaluation and Research (CDER), and that Dr. Woodcock considered all relevant evidence in
making her decision.2 In her capacity as Acting Chief Scientist, Dr. Borio also conveyed her own
views on the dispute, stating among other things that she does not believe the available data and
information support accelerated approval of eteplirsen.3 As the SDR Board?s role was to conduct
a procedural review, the SDR Board further recommended that I either conduct a substantive
scienti?c review of the dispute or convene a panel of experts to conduct a scienti?c review and

 

1 Appeal at l.
2 SDR Board Recommendation at 2.
3 1d. at 2, 25-26.

advise the Agency on whether the available evidence for eteplirsen meets the standard for
accelerated approval.?4

I note that, in my understanding, it is highly unusual for a Center Director?s decision regarding a
product application to be appealed to the Commissioner?s of?ce. Decisions on medical product
approvals and clearances are delegated under provisions of Staff Manual Guide 1410 to the
appropriate Center Directors and Center staff. 5

I have read the documents pertinent to Dr. Unger?s appeal, Dr. Woodcock?s ?nal Center
decision, and the review and recommendation of the SDR Board. I have performed a thorough
review of the basic and clinical science, a review that also comprised a brie?ng with the review
team, including Dr. Unger, and discussions with Drs. Woodcock and Jenkins. Although under
the standard for review set forth in Staff Manual Guide 9010.1, I was not required to review the
science, I did so in order to properly evaluate the positions of Dr. Woodcock and Unger. My
decision following this review is to defer to Dr. Woodcock?s judgment and authority to
make the decision to approve eteplirsen under the accelerated approval pathway, in her
capacity as Director of the Center for Drug Evaluation and Research.

My reasoning and detailed considerations for my decision are given below.

Examination of the Scientific Dispute

Key Points of Agreement

There is agreement that Dr. Unger?s views and those of the review team were heard in
great detail in an environment of open discussion and dissent on multiple occasions. I agree
with the SDR Board that there is no basis for overturning Dr. Woodcock?s decision on
procedural grounds and Dr. Unger also agrees with this conclusion.

In addition, the science is not in dispute beyond the usual types ofdisagreement that occur
when experts review clinical evidence from different perspectives. As discussed in detail below,
the evidence evaluation is complicated by the unusual circumstance of the development program,
which involved a small subset of children with a rare disease and was characterized by major

 

4 Id. at 2, 27.

5 Although, as described in Staff Manual Guide the Commissioner retains authority to make decisions
on medical product applications, those decisions are most appropriately vested in the Centers and only the most
unusual circumstances would warrant overturnng the decision of a Center?including that of a Center Director?on
a medical product application. As is set out in the Staff Manual Guide 9010.1 applicable to this matter, FDA has
adopted a formal scienti?c dispute resolution process that emphasizes l) establishing and maintaining robust dispute
resolution processes within the Centers to ensure candid and comprehensive discussion of scienti?c issues, and 2) 3
Commissioner?s Of?ce review that focuses on the adequacy of the internal Center scienti?c dispute process 
whether the Center?s processes were followed, whether the Center considered all relevant evidence bearing on the
scienti?c question at issue, and whether the initiator of the appeal was provided an opportunity to express concerns
at all appropriate levels, prior to and including the Center Director). This established dispute resolution process does
not contemplate a substantive decision on the disputed issue by the Commissioner.

flaws in the clinical study design, making the judgment on science dif?cult. The points of
agreement include the following:

DMD is a rare, progressive disease characterized by the virtual absence of functional

0 There is no approved therapy for 

- The principal question in this dispute is not whether level is an appropriate
surrogate endpoint for the purpose of accelerated approval, but whether the quantity of
produced by eteplirsen is an effect that is reasonably likely to predict clinical
bene?t, as required under section 506(c)(1 of the Federal Food, Drug, and Cosmetic
Act and FDA regulations at 21 CFR 314.510.8

0 Available preclinical work supports the clinical development of eteplirsen, most notably
clear documentation of production of the transcript and the protein in a dose-dependent
manner in primates, without measurable toxicity.9

0 Major ?aws in both the design and conduct of the clinical trials using eteplirsen have
made it impossible to use much of the resulting trial data as reliable evidence in
regulatory decision-making, including for reasonable extrapolation to clinical care. 10

0 Despite the ?aws in the clinical development program noted above, both Dr. Woodcock
and Dr. Unger, as well as the review team, agree that eteplirsen produces measurable
increases in compared with control. All agree that the amount of 
produced is small compared with expectations at the outset of trials in humans. 12
Although the gap between expectations and measured increase is not directly relevant to
the ?nal decision, the log-order difference has added ?irther uncertainty. If the levels of
had reached those measured in the setting of Becker muscular 
there would be much less concern.

0 Although there is some disagreement about whether the data support the expression of

in one study (201/202), there is agreement that the other (study 301)
demonstrates clear evidence for production. Dr. Unger concludes that there is

 

6 Appeal at 2; SDR Board Recommendation at 2.

7 Appeal at 2. I note that there is a modest body of evidence suggesting that treatment with corticosteroids yields
relatively short-term improvements in function and muscle strength (see Matthews E, et al. Cochrane Database 
Rev. doi: 10.1002/ 14651858.CD003725.pub4), but there is no evidence from adequate and
well-controlled trials that this therapy has a major effect on the course of the disease.

3 Appeal at 3; SDR Board Recommendation at 4, l7.

9 Decisional Memo, July 2016.

'0 Woodcock Decisional Memorandum at 2?3, I 1; Appeal at 4-5, 8-9.

Woodcock Decisional Memorandum at 3; Appeal at 6.

'2 Woodcock Decisional Memorandum at 11.

evidence from one adequate and well?controlled trial, while Dr. Woodcock concludes
there is evidence ?om two adequate and well?controlled trials. [3

Key Points of Disagreement

There is disagreement about whether the amount of protein produced is
suf?cient to be ?reasonably likely? to predict a clinical bene?t. The basis for the judgment
about ?reasonably likely? is subjective and is not explicitly de?ned in statute or regulations.
Guidance on Expedited Programs for Serious Conditions Drugs and Biologics (May
2014) states at page 19 (internal citation omitted):

Determining whether an endpoint is reasonably likely to predict clinical benefit is a
matter of judgment that will depend on the biological plausibility of the relationship
between the disease, the endpoint, and the desired effect and the empirical evidence
to support that relationship. The empirical evidence may include  .
epidemiological, pathophysiological, therapeutic, pharmacologic, or other evidence
developed using biomarkers, for example, or other scienti?c methods or tools. 

I conclude that Dr. Unger and Dr. Woodcock have each exercised reasonable scienti?c judgment
in reaching differing conclusions on whether the effect on production seen in the
studies in Sarepta?s application reasonably predicts clinical bene?t for the relevant subpopulation
of Duchenne patients. There is also abundant evidence that Dr. Woodcock heard and read all the
scienti?c evidence, including the detailed views of Dr. Unger and the review team; yet she came
to a different conclusion. These differing conclusions are detailed immediately below:

I Dr. Unger, the review team, and Dr. Borio have concluded that the demonstrated levels of
are not ?reasonably likely? to predict clinical bene?t based on a chain of logic
that is well-described in Dr. Unger?s appeal and in Dr. Borio?s summary. 14 In essence,
based on the belief that Becker muscular provides a model for Duchenne
because the is similar to that produced with eteplirsen, Dr. Unger concludes
that functional protein in the range of 10% of normal levels would be needed to provide
evidence that would make it ?reasonably likely? that changes in the surrogate would
predict clinical bene?t. 15 The use of Becker muscular as a model for DMD is
an extrapolation, but a rational one in the absence of validation of as a
surrogate.

I Dr. Unger also ?nds that the results of the eteplirsen clinical trials to be inadequate for

supporting the conclusion that there is any relationship between measured levels of
and improvement in function within the trial data sets. Making certain

 

l3 Woodcock Decisional Memorandum at 2?5; Appeal at 4-6.
'4 Appeal at 4, 11-20; SDR Board Recommendation at 25.

'5 Appeal at 12-15.

assumptions, he provides plots and calculations that he interprets as showing that higher
levels are not associated with improved function in the eteplirsen trials. 16

I Dr. Borio fundamentally agrees with Dr. Unger?s conclusion, and Dr. Jenkins concurs
with Dr. Unger.?

0 Dr. Woodcock ?nds that using . .the greatest ?exibility possible for FDA while
remaining within its statutory framework,? eteplirsen is . .reasonably likely to predict
clinical bene?t?? Her conclusion is based on a view that the data ?om both Study
201/202 and Study 301 are from adequate and well-controlled trials and that, although
imperfect, they adequately meet criteria to include as af?rmation of a drug effect that is
reasonably likely to predict clinical bene?t. She points out the many uncertainties about
extrapolating from a particular level of a surrogate to clinical bene?t when that surrogate
is not yet proven, including complexities of assay validation, determining whether protein
is functional, and also the extraordinary dif?culty of knowing how the amount of protein
might affect functional outcome over time and within the context of the multidimensional
nature of protein interactions in complex cellular and subcellular ?anctions. She ?nds no
rational basis for identifying a speci?c threshold value for levels that would
be needed to support a determination that a particular level is ?reasonably likely? to
predict clinical bene?t. 19 Furthermore, she provides some post-hoe calculations from the
eteplirsen clinical trials that she regards as supportive, though not de?nitive, evidence
that higher levels of are associated with greater ?mction.

She is clearly employing and interpreting the full range of appropriate information,
comprising a ?totality of evidence? approach in determining that the clinical trials
demonstrated an effect on a surrogate endpoint that is reasonably likely to predict clinical
bene?t. Because of the uncertainties in this situation with a surrogate that has not been
validated, it is clear that Dr. Woodcock?s decision also utilized the ?exibility afforded
under the relevant statutory provisions, including consideration of the life-threatening
nature ofthe disease and the lack of alternative treatments.

Conclusions Regarding the Scienti?c Disnute

I conclude that quali?ed experts with extensive experience in FDA decision-making and
stellar track records can assimilate the same scienti?c evidence and disagree about the
extrapolation to whether the evidence supports a conclusion that the treatment has an
effect that is ?reasonably likely? to predict clinical bene?t. Both Dr. Unger and Dr.
Woodcock have drawn upon a deep level of knowledge and practical experience in deriving their
conclusions, and their documents re?ect strong convictions and profound concern both for the

 

mid. at 15-18.

17 SDR Board Recommendation at 15, footnote 94.
IS Woodcock Decisional Memorandum at 12.

'9 Id. at 9.

well?being of the patients and families with DMD and for the preservation of the mission
to protect and promote public health. That said, the history of the FDA includes a consistent
precedent of ?nal decision-making about medical products at the Center level. Overruling the
Center Director is exceedingly rare and, in my View, would be appropriate only if the Center
Director?s decision could not be supported by the available data and information. In the present
case, the scienti?c uncertainties lead to a situation in which the decision is a matter of reasoned
expert opinion and judgment.

Given that I do not have technical expertise beyond those already involved in this decision
and the record contains adequate evidence to support her conclusion, I defer to the
judgment of the Center Director to approve eteplirsen under accelerated approval with the
stipulations delineated in her Decisional Memo.

Additional Concerns

The SDR Board expresses concerns based on Dr. Unger?s appeal and on interviews the
Board conducted in its procedural review about the level of involvement of the Center
Director in the review of the New Drug Application for eteplirsen. The following four points
of concern were raised by Dr. Unger and cited by the SDR Board:

1. Intense involvement of the Center Director in early stages of review;

2. Extensive involvement in planning and participating in the Advisory Committee meeting
to consider eteplirsen, held on April 25, 2016;

3. Initial decision by Dr. Woodcock on May 4, 2016 to approve before the review team had
?nalized its process for decision-making; and

4. Final decisional memorandum by Dr. Woodcock completed before Dr. Unger finalized
his own decisional memo.

The SDR Board expresses concern about Dr. Woodcock?s ?extensive, early involvement in the
review process? and states that ?her involvement here appears to have upended the typical
review and decision-making process.?20 Furthermore, the Board cautions that . .care should be
taken to avoid the appearance of interfering with the integrity of scienti?c reviews at the lower
levels of a Center?? The SDR Board concludes that ?Dr. Woodcock herself was the one who
conducted that review and resolved the con?ict in her own favor??2

 

20 SDR Board Recommendation 27.

The concerns raised by Dr. Unger and by the SDR Board require that I evaluate the role of
the Center Director. While there are many aSpects of this situation that are unusual,
leading to a rare formal diSpute between a review team and a Center Director, I note the
following points:

Dr. Woodcock?s management style has been ?hands on? during her entire career at the
FDA. She is well?known for interacting with staff at all levels and expressing her
opinions. Indeed, complex and dif?cult cases often lead to early interactions with leaders,
and as described below, frequent meetings and discussions along the way are common in
CDER. This development program may represent a point of particular focus for Dr.
Woodcock with the result that her involvement was more intensive than usual and
correspondingly had an effect upon the review team, but such a focus does ?t within a
longstanding management approach.

At the same time, under Dr. Woodcock?s leadership and in conjunction with the
leadership of Dr. John Jenkins in the Of?ce of New Drugs, multiple mechanisms have
been established for ongoing and interim discussions of evolving drug development
programs and decision?making. These sessions, such as the Medical Policy Council,
weekly senior staff meetings, the use of frequent Center Director brie?ngs (in this case at
least 14), and external advisory committee meetings are known for their exchange of
ideas and welcoming of differences of opinion. Within this context it is also
understandable that when so much is at stake with this degree of uncertainty, passionate
debate can occur.

An additional factor in this situation is the emergence of patient?centered drug
development and the extensive interactions with the patient community as part of the
overall environment for development and decision?making. While the appropriate
methods for patient-centered drug development are evolving, the fact that DMD involves
vulnerable children with a life?threatening illness and understandably concerned parents
produces signi?cant pressure on all involved. This dynamic is well re?ected in Dr. Unger
and Dr. Woodcock?s documents. With a signi?cant history dating back to the
development of drugs for patient?focused drug development is not an entirely
new component of regulatory process, and it remains an explicit CDER priority in
the current era.

While others might direct and manage the organization and interact with patient and
family communities in a different manner than Dr. Woodcock, it is dif?cult to argue that
CDER has been unsuccessful under her leadership. A vast array of new drugs approved
during her tenure is bene?tting patients, including a host of signi?cant advances using a
variety of expedited programs. Further, this litany of therapeutic successes has been
accomplished without compromising overall standards. Under Dr. Woodcock?s
leadership, commitment to due process and mutual review of scienti?c data have been
major parts of CDER culture and are making a signi?cant impact on drug development.

 It is inevitable that in some of these situations, highly quali?ed experts will disagree. In
the face of profound changes in science and social interactions related to drugs, under Dr.
Woodcock?s leadership CDER enjoys broad public support, largely due to the constant,
healthy atmosphere of transparency and debate that characterizes the documentation in
this appeal. A review of Dr. Woodcock?s record reveals extraordinary courage in the face
of extreme pressure on many occasions, including from Congress, the press, patient and
patient advocacy groups, and industry. She has taken and supported unpopular decisions
when appropriate and is well~known for not relenting to pressure. Further, I ?nd no
pattern that indicates that this decision is part of a trend for lowering the standard for drug
approval?a trend that would be unacceptable in any event.

Conclusions Regarding Additional Concerns

Overall, while I recognize the strain created by political and public pressures, given Dr.
Woodcock?s well-documented history ofnot bowing to such in?uences and a record in this case
showing her close consideration of all relevant scienti?c evidence, I do not ?nd that she
deviated from her reSponsibilities as Center Director, nor do I ?nd that she succumbed to
pressure from the patient community, the public, the press, or others.23 Further, I do not
?nd the general pattern of discourse and involvement to be atypical for Dr. Woodcock?s
management of the Center, nor do I ?nd that her conduct was in con?ict with the job
requirements for Center Directors at the FDA.

Additional Context: Accelerated Approval

It is important to note that the debate about this application is taking place in the context of an
accelerated approval. Even under the best circumstances, accelerated approval may lead to
decisions that are not veri?ed upon further examination. These limitations are a reason
accelerated approval is available only for a limited group of drugs: those intended to treat serious
or life-threatening illnesses when the drug is expected to provide a meaningful bene?t over
existing therapy. In accelerated approvals, the surrogate is not a validated surrogate. As noted in
Guidance on Expedited Programs for Serious Conditions Drugs and Biologics, page 17:

For purposes of accelerated approval, a surrogate endpoint is a marker, such as
a laboratory measurement, radiographie image, physical sign, or other measure,
that is thought to predict clinical bene?t, but is not itself'a measure of clinical
bene?t. Depending on the strength of the evidence supporting the ability of a
marker to predict clinical bene?t, the marker may be a surrogate endpoint that is

 

23 I was troubled by statements on page 16 of the SDR Board memo that Dr. Woodcock?s decision to approve
eteplirsen may have been inappropriately motivated by concerns over the sponsor?s ?nancial well?being. To address
to my own satisfaction any questions that Dr. Woodcock?s statements in the SDR Board memo might raise, I have
discussed this issue directly with Dr. Woodcock, who said that she was aware of the ?nancial pressures on the
company, but that her decision was based on the science. Based on the record and our conversation, I am satis?ed
that her decision is indeed based on her scienti?c evaluation of the evidence.

known to predict clinical benefit (a validated surrogate endpoint that could be
used for traditional approval), a surrogate endpoint that is reasonably likely to
predict a drug?s intended clinical benefit (and that could therefore be used as a
basisfor accelerated approval), or a marker for which there is insufficient
evidence to support reliance on the marker as either kind of surrogate endpoint
(and that therefore cannot be used to support traditional or accelerated approval
of a marketing application)?

Dr. Unger?s and Dr. Borio?s signi?cant concerns about the implications of approving eteplirsen
on the basis of the current data, even under the accelerated approval pathway, are evident.25 As
noted previously, Dr. Woodcock has explained that approval here will rely on . ..the greatest
?exibility possible for The record amply re?ects that the evidence here is not as strong
as everyone involved in this dispute wishes it were. The record also re?ects, however, that Dr.
Woodcock has fully considered the patient population, lack of alternative therapies, and relevant
information and analyses, and has concluded that the data are suf?cient to meet the accelerated
approval standard. Therefore, while I understand the concerns expressed by Drs. Unger and
Borio regarding the implications of lowering of the standard for approval, I defer to Dr.
Woodcock?s conclusion that accelerated approval in this case does not represent such a lowering
of the bar.

Opportunities for Process Improvement

I agree that there are many aspects of the history of eteplirsen that we must avoid in the future.
Given the distinguished careers, time?proven expertise, and sincere concerns of Drs. Unger,
Borio, Jenkins, and Woodcock, it is critical to review the following key lessons from this
experience:

0 All reviewers point to serious ?aws in the eteplirsen drug development program. I
personally have had many experiences in drug development and the assessment of
therapeutics; it is common for problems to emerge that can be identi?ed easily in

 

24 More expansive discussion of these important issues on de?nitions of biomarkers and surrogates is included in a
joint document from FDA and the National Institutes of Health, which can be found at:

http: itu-wvw . ncbi .n [m .ni .govibook sr'N BK3 26791 .

25 In her capacity as Acting Chief Scientist, Dr. Borio notes: ?Granting accelerated approval here on the basis of the
data submitted could make matters worse for patients with no existing meaningful therapies?both by discouraging
others from developing effective therapies for DMD and by encouraging other developers to seek approval for
serious conditions before they have invested the time and research necessary to establish whether a product is likely
to confer clinical bene?t.? Dr. Unger concludes: ?If we were to approve eteplirsen without substantial evidence of
effectiveness, or on the basis of a surrogate endpoint with a trivial treatment effect, we would quickly ?nd ourselves
in the position of having to approve a myriad of ineffective treatments for groups of desperate patients. In essence,
allowing marketing based on desperation, patient lobbying, and the desire and need of hope. If we were to turn the
clock back to the days prior to the 1962 Kefauver-Harris Amendments to the Federal Food, Drug, and Cosmetic Act,
the damage to society and the ?eld of evidentiary medicine would be enormous.? Dr. Jenkins expressed the same
concerns, both in written documents and in follow-up discussions Ihave had with him.

26 Woodcock Decisional Memorandum at 12.

retrospect. Clinical development involves numerous human decisions and compromises
engendered by the fact that clinical trials involve human research participants who must
be respected independently of the many others who design, conduct and judge the
research. Furthermore, the diverse interests of sponsors, patients and families,
investigators, and regulators must be melded into a few protocols for human studies, so
that no protocol represents 100% agreement. Accordingly, any retrospective criticism
should be expressed with humility. However, the ?aws in the eteplirsen program are
serious and worthy of attention by others developing drugs for serious rare diseases.

0 Speci?cally, the poor quality of many of the biopsies and the failure of the sponsor to
implement a high-quality procedure for assay validation led to a situation in which only a
?action of the data could be used to make the regulatory decision. Given the serious
nature of the disease and the invasiveness of the procedures, any studies must be
conducted according to the highest standards, so that each child and parent who
volunteers can be con?dent that they have contributed as much as possible to the
generalizable knowledge needed to provide effective treatment based on high?quality
evidence. Such attention to standards of quality in study conduct is a critical element of
respect for research participants.

0 Further, all reviewers agree that had the sponsor conducted properly-controlled,
randomized trials with dose escalation from the beginning, we would by now likely have
access to definitive information that would have resolved the disagreement. Dr.
Woodcock points out that .. [t]here is no such thing as an ?exploratory study? for a
serious, life-threatening disease without therapeutic options. Randomization should be
performed very early in the development program and open-label studies should be
avoided.? 27 Time after time, we see the over?hyping of preclinical results playing into
misguided arguments that prevent the generation of high?quality evidence because of
unrealistic expectations of bene?t, which ultimately fosters continued uncertainty about
the clinical value of therapeutics.

It is unfortunate that the sponsor touted an academically based study that had unreliable
measures of the assay, thereby greatly overstating the degree of protein expression in the
follow?up biopsies. Blinded experts assembled by the FDA fundamentally debunked this
study, which has yet to be retracted and continues to be cited.28 Dr. Woodcock identi?ed
the lack of a reliable assay as the primary defect in this ?seriously deficient? development
program, and Dr. Unger comments multiple times on the uncertainties in the assay and
the manner in which poor methodology undermined the results presentation.29 It is
critical that we continue to improve the entire pipeline of translational research to
introduce more rigorous and reliable methods, even in academically?conducted

 

27 Woodcock Decisional Memorandum at 11.
28 In View of the scienti?c de?ciencies identi?ed in this analysis, I believe it would be appropriate to initiate a
dialogue that would lead to a formal correction or retraction (as appropriate) of the published report.

2" 1d.

10

translational research. Likewise, it is equally important that we continue to work on ways
to purge discredited research in a manner that reduces inappropriate citation and follow-
on research based on the incorrect assumption of the accuracy of the precedent studies.

Dr. Unger also points out that data ?om primate studies suggest that a substantial
increase in dose (approximately a log order) might well produce the 10% of normal
levels that Dr. Unger feels would meet criteria for an effect that is ?reasonably
likely to predict clinical bene?t.3530 This View that much higher doses are likely to be
bene?cial is shared across the entire FDA hierarchy, yet studies have not been done in
humans at higher doses. However, the FDA must make a decision based on available
data.

As Drs. Unger, Woodcock, and Borio all point out, the risk of lowering the bar for new
drug approval is a serious concern. We must remember that over 90% of drugs that enter
clinical trials do not make it to market, usually due either to inadequate bene?cial effect
or unexpected toxicity.31 Lowering the regulatory hurdle could expose unsuspecting
patients to the harmful effects of these drugs that would have failed if proper
development had been done. But the statute and regulations clearly demarcate accelerated
approval as a special situation in which a range of evidence from various sources and
unvalidated surrogates may be used to determine whether the demonstrated effect on a
surrogate endpoint is reasonably likely to predict clinical bene?t, requiring an additional
judgment on the part of the FDA. Thus, I am confident that this unique situation will
not set a general precedent for drug approvals under the accelerated approval
pathway, as the statute and regulations are clear that each situation must be
evaluated on its own merits based on the totality of data and information.

The Path Forward

Given the approval status of eteplirsen, it is critical for the well-being of patients with DMD that
the course of action ful?lls the intent of accelerated approval and that the very best methods are
used. As required for post-marketing studies for products approved under the accelerated
approval program, the sponsor must conduct the required con?rmatory trial with due diligence to
evaluate whether eteplirsen has the predicted clinical bene?t. It is notable that Dr. Unger and Dr.
Woodcock agreed on several key points regarding further study of eteplirsen, regardless of the
approval decision:

I The low doses currently employed are regarded as a starting point, as it seems likely that

signi?cant increases in dose could lead to desirable ranges of production.
Accordingly, a dose??nding study at much higher doses (as much as a log order increase)

 

30 Appeal at 27.
3 Hay M, Thomas DW, Craighead JL, Economides C, Rosenthal J. Clinical development success rates for
investigational drugs. Nat Biotechnol. 

11

is needed to ?nd the dose most likely to lead to maximal clinical improvement, as will be

re?ected in the approval letter for eteplirsen?:

In order to verify the clinical benefit ofeteplirsen, conduct a 2-year
randomized, double-blind, controlled trial of eteplirsen in patients who have a
confirmed mutation of the DMD gene that is amenable to exon 5] shipping.
Patients should be randomized to the approved dosage of eteplirsen (30 mg/kg
weekly) or to a dosage that provides signi?cantly higher exposure, e. 30
mg/kg daily. The primary endpoint will be the North Star Ambulatory
Assessment.

Dr. Unger points out that Study 301 could be converted to a randomized trial in which
one arm remains on the current dosing regimen and the other(s) receive higher doses in
the range that might be expected to produce much higher levels.33 A study
with only a modest increase in dose that found no difference would be non?informative.
Accordingly, the difference in doses in a post?approval trial should be enough to create
clear separation in levels.

0 Dr. Woodcock notes that the sponsor is also conducting a randomized trial of eteplirsen
versus no therapy in other forms of DMD.34 The utmost attention should be paid to
optimizing the methodological rigor of this trial and the trial evaluating higher doses
discussed above.

0 It is essential that the clinical and patient communities not be misled by exaggerated
claims, given the poor quality of the studies in the sponsor?s application and the fact that
the approval is based on a non-validated surrogate and that clinical bene?t has not yet
been established. The applicable statutory and regulatory provisions governing
accelerated approval, which require sponsors to submit promotional materials to FDA for
pre?dissemination review, should be helpful in this regard.

Summary

In conclusion, I defer to Dr. Woodcock in her role as Center Director to make the decision to
approve eteplirsen under the accelerated approval provisions. I ?nd merit and reason on both
sides of the disagreement, and despite the intensity of the argument, I believe that the quality of
thinking on both sides re?ects the importance of clinical and scienti?c expertise coupled with
due process Within the FDA. I ?nd no basis for a view that Dr. Woodcock was unduly in?uenced
by involvement with the patient community or other external pressures, and note that our
understanding about how to include patients in the regulatory process is evolving. In addition,
serious shortcomings present in the eteplirsen development program should not be allowed to
establish a broad precedent for therapeutic development in rare diseases. In the end, We have

 

32 Draft Approval Letter for eteplirsen, as of September 9, 2016.
33 Appeal at 27.
34 Woodcock Decisional Memorandum at 10.

12

signi?cant agreement on the science, but a disagreement on the integration of all the evidence as
to whether the small changes in levels induced by these doses of eteplirsen are
?reasonably likely? to predict clinical bene?t. Considering that a substantially ?awed
development program contributed to the dif?culty coming to resolution in this case, we must
redouble our efforts to move the therapeutic development ecosystem to use methods that will
produce high?quality evidence from the outset.

For the reasons set forth above, ?nal decision, as set forth in Dr. Woodcock?s July 14,
2016 Memorandum, is upheld. This matter is remanded to CDER for action consistent with this
decision.

Wm M. meow

Robert M. Califf, M.D., Commissioner ofFood and Drugs

13

DEPARTMENT OF HEALTH & HUMAN SERVICES

Date:

August 8, 2016

To:

Robert Califf, M.D.
Commissioner of Food and Drugs

From:

Luciana Borio, M.D.
Acting Chief Scientist

Luciana Borio A

Food and Drug Administration
Office of the Commissioner

Digitally signed by Luc ana Bor o A
DN c=US o=U S Government ou=HHS
ou=FDA ou=People cn=Luc ana Bor o A
0 9 2342 19200300 100 1 1=2000101586
Date 2016 08 08 20 00 37 04'00'

Subject: Scientific Dispute Resolution Appeal regarding Eteplirsen
This matter is before the Office of the Commissioner on an appeal submitted by Ellis Unger,
M.D., Director of the Office of Drug Evaluation I (ODE-I) (the initiator), under Staff Manual
Guide 9010.1, “Scientific Dispute Resolution at FDA” (the SDR-SMG). In his scientific dispute
resolution (SDR) appeal, dated July 18, 2016, Dr. Unger challenges the basis for a decisional
memorandum issued by Janet Woodcock, M.D., Director of the Center for Drug Evaluation and
Research (CDER). Dr. Woodcock’s decisional memorandum concludes that a new drug
application (NDA) submitted by Sarepta Therapeutics Inc. (Sarepta) for eteplirsen, a drug
intended to treat Duchenne muscular dystrophy (DMD), meets the standard for accelerated
approval under 21 CFR § 314.510. Specifically, Dr. Woodcock’s memorandum states that the
data submitted in support of the NDA establishes “increased dystrophin protein production, a
surrogate endpoint [for DMD] that [she] conclude[s] is reasonably likely to predict clinical
benefit.” 1 Dr. Unger states that he disagrees with Dr. Woodcock’s decisional memorandum
because he does not believe “the findings on the dystrophin surrogate endpoint are reasonably
likely to predict clinical benefit.” 2
Upon receipt of the appeal from Dr. Unger, in accordance with the SDR-SMG, the Office of the
Chief Scientist convened the Agency Scientific Dispute Process Review Board (the SDR Board),
a standing committee, which I chair, whose role in evaluating the appeal is to conduct a review
of the processes used in the Center to render a decision on the scientific dispute at issue. 3 Under
the SDR-SMG, “The goal of this review is to determine if the processes followed in the Center
fully considered all relevant evidence and provided the initiator with an opportunity to express
his or her concerns at all appropriate levels, prior to and including the Center Director.” 4 My
role in the process, as Chair of the SDR Board, is to provide a recommendation to you, as
Commissioner of Food and Drugs, with respect to “whether a Center failed to follow its
processes and/or did not provide an adequate opportunity to the initiator to express his or
concerns; [whether] all relevant evidence bearing on the scientific question at issue has been
considered; and[] whether the dispute should be remanded to the Center Director.” 5 The written
1

Woodcock Decisional Memorandum at 1.
Appeal at 3.
3
SDR-SMG at 3. (“The Agency Scientific Dispute Process Review Board (hereafter Board) is a standing committee comprised of
representatives of the Office of Accountability and Integrity, Ombudsmen from all Centers and the agency (or officials so designated)
and representative(s) from the Office of the Chief Scientist. The Board is chaired by the Chief Scientist.”).
4
Id. at 12.
5
Id. at 5.
2

 recommendation must reflect the SDR Board’s underlying rationale, along with minority views
among the members, for those findings. 6
In conducting its evaluation, the SDR Board reviewed pertinent aspects of the Center’s
administrative file for the eteplirsen NDA and interviewed Dr. Unger, Dr. Woodcock, one
member of the review team for the NDA, who requested anonymity, and Virginia Behr, the
Ombudsman for CDER. Based on its review, the SDR Board has determined that the processes
followed by CDER provided Dr. Unger with an adequate opportunity to present his scientific
views and that CDER considered all relevant evidence. As Chair of the SDR Board, I therefore
recommend that you do not remand this matter to the Center Director for further action. 7
However, there are additional considerations meriting your attention, which I describe below.
Furthermore, the SDR Board encourages you to conduct a thorough substantive review of the
scientific dispute in this matter or, in the alternative, to convene a panel of relevant experts to
conduct such a review and provide advice to the agency and you, as Commissioner, on whether
the evidence of the effect of eteplirsen on the surrogate endpoint is reasonably likely to predict
clinical benefit.
BACKGROUND
1. Eteplirsen and DMD
Dr. Unger provides an overview of eteplirsen and DMD in his appeal. 8 In short, DMD is a genetic
disorder with catastrophic effects on its sufferers:
[DMD] is an X-linked recessive neuromuscular disorder caused by mutations of
the dystrophin gene[,] . . . [which] disrupt the messenger ribonucleic acid
(mRNA) reading frame [and] lead[] to the absence or near-absence of dystrophin
protein in muscle cells. . . . Absence of dystrophin leads to muscle damage, with
replacement by fat and collagen. . . [and a concomitant] loss of physical function
in childhood and adolescence, with premature death from respiratory and/or
cardiac failure in the second to fourth decade. 9
There are no FDA-approved therapies for DMD. 10 Sarepta has designed eteplirsen to target the premRNA transcripts of the dystrophin gene so that exon 51 is excluded from the resulting mRNA: 11
[B]y restoring [] the mRNA reading frame, a ‘truncated’ but nevertheless partially
functional form of the dystrophin protein can be produced by muscle cells,
delaying disease progression. Similar truncated dystrophin is found in a less
severe form of muscular dystrophy, Becker Muscular Dystrophy (BMD). In
essence, the drug is hoped to induce production of sufficient Becker-type
dystrophin to slow the progression of the disease. This drug is specific for exon
51 mutations, a subset of the mutations that cause DMD. If approved, the drug
6

Id.at 13.
See id. (“The Commissioner will review the [SDR Board’s] recommendation and render a final decision on . . . whether the dispute
should be remanded to the Center Director for corrective action” and “work with the Center Director to determine what corrective
actions must be taken, if any.”).
8
Unless otherwise indicated, Drs. Unger and Woodcock appear to agree as to the background provided in this section.
9
Appeal at 2.
10
Id.
11
The charity, Muscular Dystrophy UK, has a nice description of the technology underpinning eteplirsen, which can be accessed at:
http://www.musculardystrophyuk.org/progress-in-research/background-information/what-is-exon-skipping-and-how-does-it-work/.
7

2

 would be indicated for ~13% of the overall DMD patient population. Eteplirsen
has not received marketing authorization from any regulatory authority, and no
similar drugs are approved. 12
In attempting to establish that eteplirsen is safe and effective for the treatment of DMD, and thus
meets one of the standards for approval in the Federal Food, Drug, and Cosmetic Act (FD&C Act),
Sarepta has submitted data from three clinical studies:
Study 201 was a single-center, double-blind, placebo-controlled study in 12
patients with DMD. Patients were randomized (1:1:1) to eteplirsen 30
mg/kg/week, eteplirsen 50 mg/kg/week, or placebo (4 patients per group). After
24 weeks, the 4 patients originally randomized to placebo were re-randomized to
eteplirsen 30 mg/kg/week (n=2) or eteplirsen 50 mg/kg/week (n=2). The trial was
eventually extended to an open-label phase (Study 202) where all patients
received eteplirsen, although investigators and patients remained blinded to dose.
These patients have continued to receive eteplirsen for more than 4 years. This
continuous study is referred to as Study 201/202. Study 301 is an externally
controlled study where all patients are receiving open-label eteplirsen, 30 mg/kg,
by weekly infusion. The study is ongoing and still accruing patients. Interim data
were obtained from 13 patients in this study. 13
Dr. Unger further explains:
The endpoints for [the three] studies can be broadly divided into those that aim to
show changes in physical performance, e.g., walking speed, rise time from the
floor, muscle function; and those that aim to show effects on production of
dystrophin in skeletal muscle – the surrogate endpoint. Dystrophin was quantified
in this development program using two methods: Western blot and
immunohistochemistry. 14
Immunohistochemistry (IHC) analysis looks at thin slices of muscle biopsies to see if dystrophin
is present or absent. Each muscle fiber that shows any amount of dystrophin is counted as
positive, regardless of the actual quantity of dystrophin present. Western blot analysis assesses
how much dystrophin is present.
For Study 201/202, Sarepta submitted Western blot and IHC analysis evaluating proteins in
muscle samples obtained from the twelve patients before the study and then again at twelve, 24,
and 48 weeks. 15 “The Western blots submitted by the applicant for Study 201 were
oversaturated, unreliable, and uninterpretable.” 16 Because CDER also determined that the
conditions under which the original IHC analysis was performed were inadequate, including that
the reader was not masked to sequence and time, the Center requested a re-reading of the stored
images by three masked pathologists under different conditions. 17 The IHC results from the
reread were not nearly as favorable, as compared to the initial IHC results reported by Sarepta.

12

Appeal at 2.
Id.
14
Id.
15
Id. at 4, 8.
16
Id. at 4.
17
Unger Decisional Memorandum at 12-13.
13

3

 The re-read showed a nominally statistically significant increase in dystrophin in
response to eteplirsen for the low dose group, but not the high dose group[] (. . .
[T]he type-I error rate was not controlled for multiplicity.) 18 Moreover, for the 4
patients who had received placebo through Week 24 and then switched to
eteplirsen, there was no increase in dystrophin at Week 48. 19
For Study 201/202, CDER also worked with Sarepta to improve the Western blot assays, and
researchers performed repeat biopsies on eleven of twelve patients at Week 180. 20 Only three of
the eleven patients had stored baseline samples that were adequate for evaluation, and so
baseline samples were obtained from six additional patients external to Study 201/202. 21 Dr.
Unger also notes that all baseline samples were obtained from a different muscle group than the
samples obtained at Week 180. 22 Based on its own analysis of the IHC data, Sarepta claimed a
remarkable increase of dystrophin immunostaining at Week 180: from 1.1% ±1.3% positive
muscle fibers at baseline to 17.4% ± 10.0% positive fibers at Week 180. 23 The Western blot
analysis resulted in Week 180 dystrophin levels that were small, with a mean increase of only
0.93% of normal dystrophin levels in the muscle fibers. 24 Dr. Unger remarked that the lack of
concordance between the IHC and the Western Blot results is “striking” and also noted that FDA
did not verify the integrity of the IHC results. 25 As previously noted, each muscle fiber that
shows any amount of dystrophin is counted as positive in IHC, regardless of the actual quantity
of dystrophin present.
As noted above, Study 301 is an ongoing study. For purposes of its review of the NDA, CDER
requested that Sarepta perform Western blot analysis on samples obtained from 13 patients
enrolled in the study. 26 The analysis compared paired biceps samples: baseline samples and
samples obtained at 48 weeks, after 48 weeks of treatment with 30 mg/kg of eteplirsen
infusion. 27 Dr. Woodcock told the SDR Board that representatives from CDER were present in
the laboratory for the Western blot analysis and oversaw the procedures and controls. The
Western blot analysis showed a statistically significant increase in dystrophin, ranging in an
increase from 0.22% to 0.32% of normal. 28 It should be noted, however, that a statistically
significant increase in dystrophin, the surrogate endpoint, of an exceptionally small magnitude
does not imply clinical benefit, which is the issue at the core of Drs. Unger and Woodcock’s
scientific disagreement.

18

That is, with respect to time points of assessment and the 2 doses tested.
Appeal at 8-9. Of note, in her decisional memorandum, Dr. Woodcock rejected the findings in both the original and second
evaluation of the images: “Much of the controversy over the adequacy of these assessments relates to the fact that rigorously validated
assays were not used to evaluate the initial 3 muscle biopsies, apparently resulting in overestimation of the various readouts and some
irreproducibility of IHC and Western blot dystrophin assays. For these reasons, I do not discuss or rely upon the results of these
earlier assays, or on re-reads of them.” (Woodcock Decisional Memorandum at 2). She explained to the SDR Board that, after
consultation with others in CDER, she does not view IHC results standing alone as a valid method to evaluate dystrophin levels.
20
Appeal at 5.
21
Id. at 5, 9.
22
Id at 5. Dr. Unger clarifies in his decisional memorandum that the baseline biopsies were from the biceps muscle, the Week 180
biopsies from the deltoid muscle. (Unger Decisional Memorandum at 17).
23
Appeal at 9. As discussed below, however, Dr. Unger does not believe that those results are reliable.
24
Id at 5.
25
Id. at 9-10.
26
Id. at 6. Dr. Unger states that the biopsies were obtained from 13 patients but only reports the data as to 12 patients. “There was
one patient for whom none of the values met the acceptance criteria [for the Western blot assay].” (Unger Decisional Memorandum at
21).
27
Appeal at 6.
28
Id.
19

4

 2. Legal Standard for Accelerated Approval and Patient Perspectives
On December 11, 1992, on the basis of its broad statutory authority to approve drugs under the
FD&C Act, FDA issued regulations providing for accelerated approval of drugs. 29 Under 21
CFR § 314.510, FDA may grant accelerated approval for a drug based on a surrogate endpoint
under certain circumstances:
FDA may grant marketing approval for a new drug product on the basis of
adequate and well-controlled clinical trials establishing that the drug product has
an effect on a surrogate endpoint that is reasonably likely, based on
epidemiologic, therapeutic, pathophysiologic, or other evidence, to predict
clinical benefit or on the basis of an effect on a clinical endpoint other than
survival or irreversible morbidity. Approval under this section will be subject to
the requirement that the applicant study the drug further, to verify and describe its
clinical benefit, where there is uncertainty as to the relation of the surrogate
endpoint to clinical benefit, or of the observed clinical benefit to ultimate
outcome. Postmarketing studies would usually be studies already underway.
When required to be conducted, such studies must also be adequate and wellcontrolled. The applicant shall carry out any such studies with due diligence. 30
The preamble to the proposed rule defines “surrogate endpoint” as follows:
A surrogate endpoint, or “marker,” is a laboratory measurement or physical sign
that is used in therapeutic trials as a substitute for a clinically meaningful endpoint
that is a direct measure of how a patient feels, functions, or survives and that is
expected to predict the effect of the therapy. For example, elevated cholesterol
and hypertension, two surrogate endpoints, are important because they are risk
factors for coronary and cerebral artery disease; but it is the impact of the diseases
(e.g., angina, congestive heart failure after a heart attack, paralysis after a stroke,
or sudden death) that is important to the patient. 31
In 2012, Congress passed the Food and Drug Administration Safety and Innovation Act (FDASIA).
Section 901 of FDASIA amended the FD&C Act to provide FDA with specific authority to grant
accelerated approval to drugs for serious conditions. 32 Section 506(c) of the FD&C Act now largely
tracks language in the regulations issued by FDA in 1992. Section 901 of FDASIA also added
current section 506(e) to the FD&C Act, which clarifies that the amendments were “intended to
encourage [FDA] to utilize innovative and flexible approaches to the assessment of products under
accelerated approval” but that “[n]othing in this section shall be construed to alter the standards of
evidence under subsection (c) or (d) of section 505 (including the substantial evidence standard in
section 505(d) of [the FD&C Act].” 33
Section 901 of FDASIA also directed FDA to issue guidance to industry on the development of

29

57 Fed. Reg. 58942 (Dec. 11, 1992).
Emphasis added.
31
57 Fed. Reg. 13234, 13235 (Apr. 15, 1992).
32
FDASIA, PL 112-144, July 9, 2012, 126 Stat. 993.
33
Id.
30

5

 drugs for accelerated approval and required consideration of the following:
In developing the guidance . . . [FDA] shall consider how to incorporate novel
approaches to the review of surrogate endpoints based on pathophysiologic and
pharmacologic evidence in such guidance, especially in instances where the low
prevalence of a disease renders the existence or collection of other types of data
unlikely or impractical. 34
Section 1137 of FDASIA further directs FDA to:
develop and implement strategies to solicit the views of patients during the
medical product development process and consider the perspectives of patients
during regulatory discussions, including by—(1) fostering participation of a
patient representative who may serve as a special government employee in
appropriate agency meetings with medical product sponsors and investigators;
and (2) exploring means to provide for identification of patient representatives
who do not have any, or have minimal, financial interests in the medical products
industry. 35
In May 2014, FDA finalized a guidance on “Expedited Programs for Serious Conditions —
Drugs and Biologics.” The Guidance provides general information on the evidence that the
agency considers in determining whether to grant accelerated approval. 36 The Guidance clarifies
that assessing a surrogate endpoint hinges on understanding both the disease process and the
relationship between the drug’s effect and the disease process. 37 With respect to the latter, the
Guidance states:
The extent to which a drug’s effect on the surrogate endpoint is known to predict
an effect on the disease either because the effect is on the causal pathway or
correlates with clinical outcomes is critical. Sometimes this relationship can be
assessed epidemiologically[,] but it is most persuasively established by knowing
that a drug that affects the surrogate endpoint also affects a clinical outcome. 38
The Guidance also provides some insight on how the agency exercises its judgment in evaluating
surrogate endpoints when little is known about how an effect on a surrogate endpoint might affect
clinical endpoints:
Particularly in rare diseases, there may be limited information in the literature,
lack of in-depth epidemiological or historical data, and little or no experience with
other drugs to inform the interpretation of surrogate endpoints or intermediate
clinical endpoints. FDA may consult with external experts on surrogate endpoints
and intermediate clinical endpoints where there is a lack of historical data for a
given disease. 39

34

Id.
Id.
36
Expedited Programs Guidance at 19-22.
37
Id. at 20-22.
38
Id. at 21.
39
Id. at 21-22.
35

6

 FDA obtains patient perspectives through a variety of avenues, “such as open public hearings on
specific diseases or drug development issues, and as speakers at FDA-sponsored conferences and
workshops.” 40
3. SDR-SMG and CDER’s SDR-SOPs
The Office of the Commissioner issued the SDR-SMG on January 13, 2009. Its stated purpose is
“to improve the process of internal scientific dispute resolution[] and to encourage open
communication throughout the agency.” 41 The SMG “encourages the resolution of scientific
disputes at the working level in the organization, starting with the frontline employees and their
immediate supervisors or team leaders” and cautions that the “agency’s appeals process for
scientific disputes is not a replacement for robust and fair Center-level processes.” 42 As noted
above, the SDR-SMG provides for submission of SDR appeals to the Office of the
Commissioner and outlines the process and standards for evaluating such appeals. Under the
SDR-SMG, the SDR Board evaluates whether “the processes followed in the Center fully
considered all relevant evidence and provided the initiator with an opportunity to express his or
her concerns at all appropriate levels, prior to and including the Center Director.” 43 As Chair of
the SDR Board, the Chief Scientist then provides a written recommendation on those issues to
the Commissioner, who renders a final decision on whether the scientific dispute should be
remanded to the Center for further action. 44
In addition to outlining the process for elevating scientific disputes to the Office of the
Commissioner, the SDR-SMG details the agency’s “requirements for the minimum standards
for scientific dispute resolution processes in the Centers” and provides a collection of nonmandatory “best practice[s]” for such dispute resolution. 45 The SDR-SMG’s requirements for
resolving scientific disputes at the Center-level begin with an obligation on the part of Center
management to ensure open scientific debate on controversial issues:
Center management shall create an atmosphere in which consultation and open
discussion on controversial issues are encouraged. When disagreements occur, it
is necessary to follow appropriate procedures for resolving them. Informal
methods, using good management practices for resolving conflict, should be
employed prior to instituting the more formal procedures described here.
Notwithstanding informal good management practices used to try to resolve the
conflict, timely written reviews of the scientific matter in dispute should be
completed by all members of a review group, including initiator and supervisors,
to enable as open and complete a discussion of the issues as possible at the
working level of the organization.46
The SDR-SMG then goes on to require the Centers to have in place written standard operating
procedures for formally resolving scientific disputes (SDR-SOPs) in the event that such informal
attempts at resolution are unsuccessful. 47 In contrast to the procedural review contemplated by
40

79 Fed. Reg. 65410, 65411 (Nov. 4, 2014).
SDR-SMG at 1.
42
Id. at 2.
43
Id. at 12.
44
Id. at 12-13.
45
Id. at 2-3.
46
Id.at 6.
47
Id.
41

7

 the SDR-SMG, Center-level SDR-SOPs should provide for substantive review of the scientific
disputes at issue within the Center. 48
At CDER, there are three interrelated chapters of the Center’s Manual of Policies and Procedures
(MAPPs) that serve to implement the SDR-SMG’s requirements. The first, MAPP 4151.8,
“Equal Voice: Discipline and Organizational Component Collaboration in Scientific and/or
Regulatory Decisions,” sets forth CDER’s principles for resolving scientific disputes informally
and requires “a collaborative environment for decision-making.” 49 According to the MAPP,
“[s]uch an environment requires open communication and exchange of ideas in a mutually
respectful professional environment[] and the full and open participation of all relevant
disciplines and organizational components in the decision-making process.” 50 MAPP 4151.8
states that “[e]ach individual who contributes to the decision-making process” must “be sure the
position represented is consistent with the scientific, regulatory, and/or administrative policies of
that . . . organizational component” and that “[o]pinions of staff should be documented and
supported by data in a matter commensurate with the magnitude of the decision being made.” 51
The second and third MAPPs at issue directly relate to CDER’s formal SDR process. MAPP
4151.1, “Scientific/Regulatory Dispute Resolution for Individuals Within a Management Chain,”
provides for raising a scientific issue to the “Next Highest Management Official” (NHMO) if
alignment on an issue cannot be reached by the staff on a team or through discussions with a
team leader or first-level supervisor. The individual who disagrees with the decision (the
disputant) “. . . may initiate a dispute resolution process by writing a statement (called a dispute
statement) describing the position, concept, opinion, or recommendations with which the
disputant disagrees . . . as well as the proposed changes and rationale for the changes in
recommendations and/or conclusions.” 52
The disputant submits the statement to the NHMO, i.e., “the management official one level
above the management official who made the decision being disputed.” 53 The NHMO then
issues a written decision on the issue, and any disputant may then appeal the written decision up
the chain of command all the way to the Center Director through use of the same process. 54
MAPP 4151.2, “Resolution of Differing Professional Opinions: Review by Ad Hoc Panel and
CDER Director,” provides for further formal review under certain circumstances if alignment
cannot be reached under the process in MAPP 4151.1. 55 A CDER employee may initiate the
process by submitting a written package, which must include “[a]n assessment of the possible
significant negative consequences to the public health” at issue in the dispute, to the CDER
Ombudsman. 56 The CDER Ombudsman and the Center Director then “determine whether the
consequences of the decision in question are potentially serious enough to warrant” additional
review. 57 If so, the Center Director appoints a chairperson to lead an Ad Hoc review panel for
purposes of evaluating the scientific dispute and providing a recommendation to the Center

48

See id.; see also footnote 136.
MAPP 4151.8 at 2.
50
Id.
51
Id. at 2-3.
52
MAPP 4151.1 at 3.
53
Id.
54
Id. at 4.
55
Id. at 5; MAPP 4151.2 at 1-2.
56
MAPP 4151.2 at 5.
57
Id.; see also id. (“In most cases, the Ombudsman will ensure that all other avenues for resolution (e.g., dispute resolution process,
Advisory Committee discussion, CDER regulatory briefing) have been exhausted . . . .).
49

8

 Director, who renders the final decision. 58 The Ad Hoc panel typically includes one member
with relevant technical expertise, one member chosen from a list provided by the person
requesting review, and, if possible, one member with relevant expertise who is external to the
agency. 59
4. Procedural History of the Dispute in CDER
Sarepta submitted its NDA for eteplirsen (#206488) on June 26, 2015. 60 CDER assigned it for
review to the Division of Neurology Products (DNP) within ODE-I, the office for which Dr.
Unger serves as Director. 61 Even before submission of the NDA, however, representatives from
the Office of New Drugs (OND), DNP and ODE-I (the review team) regularly briefed Dr.
Woodcock on issues related to the ongoing study of eteplirsen pursuant to an investigational new
drug application (IND) and the anticipated NDA. 62 The discussions at these briefings included
among their topics: the suitability of eteplirsen for accelerated approval, an overview and
background for eteplirsen, study design, a clinical site inspection report for Sarepta, general
brainstorming, and planned communications. 63 Dr. Unger told the SDR Board both that there
were far more briefings of the Center Director than is typical and that the scope of those
briefings included an unusual level of detailed discussion.
During the SDR Board’s separate interviews of Dr. Unger and the review team member (RTM),
the SDR Board learned that, at Dr. Woodcock’s direction, the review team also joined her in
meetings with patient advocacy groups for DMD on multiple occasions—anywhere from six to
twelve times—from very early on in the review process. The RTM described the meetings with
the patient advocacy groups, which frequently included boys with DMD and their parents, as
“intense,” “personal,” and “intimidating.” Dr. Unger and the RTM both thought that Dr.
Woodcock’s early interest and involvement in DNP’s approach to guiding the development of
eteplirsen was based in part on the enthusiasm in the DMD community in relation to an article
published about the initial findings for Study 201/202, which Drs. Unger and Woodcock now
agree are misleading and unreliable. Indeed, Dr. Woodcock told the SDR Board that she became
involved because of the broader public interest the article generated, along with encouragement
from the Commissioner of Food and Drugs at the time and her long-held belief that OND has
been very conservative in evaluating drugs for accelerated approval. In his decisional
memorandum, Dr. Unger explains the excitement surrounding eteplirsen at the time as follows:
[The initial findings for Study 201/202] were substantially reported in a 2013
publication, which claimed that eteplirsen markedly increased functional
dystrophin production: “…the percentage of dystrophin-positive fibers was
increased to 23% of normal; no increases were detected in placebo-treated
SDWLHQWV  S”        (YHQ JUHDWHU LQFUHDVHV RFFXUUHG DW ZHHN         DQG    
58

Id. at 6-7.
Id. at 6.
60
Unger Decisional Memorandum at 1.
61
Id. at 2.
62
Appeal at 24-25; Chronology prepared by Virginia Behr and submitted to the SDR Board (Behr Chronology) at 1-2. In his appeal,
Dr. Unger consistently refers to the representatives from OND, OND-I and DNP who were involved in the review of the eteplirsen
NDA as the “review team” or as “the division,” even though he appears to be referring to senior management within OND on
occasion. Dr. Woodcock has also used the same terminology on occasion, though not as consistently. For the sake of efficiency, this
memorandum refers to everyone at CDER who was involved in the review of the eteplirsen NDA, besides Dr. Woodcock herself, as
the review team. Nonetheless, the SDR Board notes that, within FDA, “review team” is often used to reflect the core team of
individuals within a division who are directly engaged in the review of the science underlying a regulatory submission.
63
Appeal at 24-25; Behr Chronology at 1-2.
59

9

 in the 30 and 50 mg/kg cohorts, respectively), suggesting that dystrophin
increases with longer treatment. Restoration of functional dystrophin was
confirmed by detection of sarcoglycans and neuronal nitric oxide synthase at the
sarcolemma.” The publication also stated that dystrophin expression was
confirmed by Western blot, with a figure showing what were termed
“representative” results.
Publication of this paper was followed by a Sarepta press release, which also
claimed a remarkable treatment effect from eteplirsen and raised wildly
unrealistic expectations in the DMD community. 64
In their interviews with the SDR Board, Dr. Unger and Dr. Woodcock stated that FDA also
received significant correspondence from the public and Congress, much of which urged
approval of eteplirsen. 65 Some of the correspondence used vulgar language and was abusive to
the review staff. 66
The briefings of Dr. Woodcock began again five to six months after submission of the NDA for
eteplirsen. 67 The focus of these briefings was on preparation for a planned meeting of the
Peripheral and Central Nervous System Drugs Advisory Committee (AC meeting) to provide
advice on the review of the eteplirsen NDA, which meeting was initially scheduled for January
2016 but then rescheduled for April 25, 2016. 68 The preparation involved discussions of the
ongoing review of the data, including the “strengths, limitations, and uncertainties of the data,
particularly with respect to the comparison between the open-label eteplirsen group and a
contemporary untreated external control group.” 69 During their respective interviews with the
SDR Board, both Dr. Unger and the RTM conveyed their belief that Dr. Woodcock was inclined
to grant approval from very early on in the process. But the RTM stated that Dr. Woodcock’s
views were not always clear during discussions throughout the review of the science—
sometimes she seemed to agree with external constituents, sometimes not. The RTM told the
SDR Board that, in his or her view, the review team was never sure whether they were
discussing science, policies, or politics. According to both Dr. Unger and the RTM, Dr.
Woodcock frequently conveyed that she thought the review team was being unreasonable and
encouraged DNP to find a way to approve the eteplirsen NDA. Both Dr. Unger and the RTM
told the SDR Board that Dr. Woodcock seemed focused on the external pressures, from both
patient advocacy groups and Congress, and that she frequently talked about the effects of a
decision regarding eteplirsen in terms of overarching policy (e.g., the need to be more flexible
for ultra-rare diseases). The RTM highlighted to the SDR Board that at least two members of the
review team were leaving FDA or had left the agency in the wake of both the decision-making
process within CDER and the pressures exerted by outside forces.
Dr. Woodcock conceded to the SDR Board that she was leaning toward granting approval in
light of the available data as early as 2014. She said that her goal throughout the discussions
64

Unger Decisional Memorandum at 11 (emphasis in original), citing Mendell JR, et al: Eteplirsen for the treatment of Duchenne
muscular dystrophy. Ann Neurol 2013;74:637-47 and Sarepta press release, dated 8/8/13
(http://investorrelations.sarepta.com/phoenix.zhtml?c=64231&p=irolnewsArticle&ID=1846052). Dr. Unger also notes, “It was these
perceptions and expectations that led the applicant to declare that a placebo-controlled study was no longer feasible.” Unger
Decisional Memorandum at 11.
65
See also Appeal at 23.
66
See, e.g., id. at 23-24.
67
Id. at 25; Behr Chronology at 2-3.
68
Appeal at 25; Behr Chronology at 2-3.
69
Appeal at 25.
10

 with the review team was to convince them to come around to her more flexible way of thinking
about the data. According to Dr. Woodcock, she recognized that there were serious and
significant flaws in the study design for Study 201/202 and the data it generated but that she did
not “want to hold” those flaws “against the patients.” She conceded that the results produced by
Studies 201/202 and 301 were always less than anyone in CDER had hoped.
In their respective interviews with the SDR Board, both Dr. Unger and the RTM focused to some
extent on Dr. Woodcock’s involvement in the planning stages for the AC meeting. They
expressed some surprise at the extent of her involvement. Dr. Unger indicated in his interview
with the SDR Board that Dr. Woodcock even advocated, unsuccessfully, for changing the order
of the questions to be posed to the committee and wanted the question on conventional approval
to come before the one on accelerated approval.
The RTM told the SDR Board: (1) that Dr. Woodcock made it clear in one or more of the
meetings leading up to the AC meeting that she intended to speak at the meeting but (2) that the
substance and purpose of her participation were never communicated. Although the RTM
affirmatively stated that the review team was free to develop its own presentation to the
committee, uncertainty with respect to Dr. Woodcock’s role made doing so more difficult. The
RTM also noted that Dr. Woodcock requested a longer than is typical Open Public Hearing
portion of the AC meeting that, as a result, the review team thought there would insufficient time
for them to make their presentations during a one-day meeting. The RTM stated that the review
team asked to extend the advisory committee to two days but that they were overruled.
On April 25, 2016, CDER held the AC meeting. The meeting focused on the data from Study
201/202. 70 Dr. Woodcock spoke at the meeting several times. At the meeting she made a
presentation that was intended to “provide a framework within which to consider [the] data
[underlying the eteplirsen NDA] based on [her] 30 years of experience at FDA and really
extensive experience in implementation of the legal standards for drug approval.” 71 She
highlighted many of the difficulties in interpreting the data. 72
At the AC meeting, Dr. Woodcock also described the standards for both conventional and
accelerated approval of drugs but mentioned that the agency had not “articulated an evidentiary
standard for determining if a surrogate endpoint is reasonably likely to predict clinical benefit.” 73
She concluded her presentation with the following remarks:
I would note that much of the effort in evaluating a drug development
program goes into avoiding a specific mistake, that is erroneously approving a
drug that is not effective.
There often is little consideration of another error, which is failing to
approve a drug that actually works. In devastating diseases, the consequences of
this mistake can be extreme, but most of these consequences are borne by
patients who traditionally [] have little say in how the standards are
implemented.
The accelerated approval program includes a requirement for
confirmatory studies for efficacy, so as you've heard from the sponsor, you have
to do further studies to explore and confirm effectiveness. An inherent
70

Sarepta had not yet submitted the data from Study 301.
Advisory Committee Transcript at 151.
72
Id. at 151-155.
73
Id. at 155-156.
71

11

 presumption in this program of accelerated approval, which is written in the
preamble to our regulation about it, is that more uncertainty is going to be
tolerated initially and that in fact sometimes we will collectively get it wrong,
otherwise accelerated approval would really have no different standards than
regular approval. 74
During the questions to the committee members, Dr. Woodcock restated the standard for
accelerated approval and emphasized that, with regard to the surrogate endpoint of dystrophin,
there has never been a “threshold established [to show a reasonable likelihood of predicting
clinical benefit] because there's never been a drug to do this.”75 When later asked for
clarification of the extent to which the committee members were to incorporate the testimony of
the boys and their families into their evaluation of clinical outcomes for Study 201/202, Dr.
Woodcock stated:
Well, we are instructed, as people said, to take the use of the patient community
into account, more on the benefit and the risk. * * * So the statutory standard is
more or less as described there, but there is flexibility, and that's where we should
take the views of the community into account. 76
During his SDR Board interview, the RTM stated that, notwithstanding Dr. Woodcock’s
emphasis on accelerated approval and the standard of “reasonably likely to predict clinical
benefit,” “[s]urrogacy was not discussed in any genuine scientific way” during the AC meeting
because it had not been framed that way by Sarepta through its presentation to the committee.
The RTM specifically stated that there was no discussion of “substantial evidence” in the context
of accelerated approval, nor what might constitute “interpretable evidence.” The RTM believed
that, by the end of an emotional AC meeting, the framework for evaluating the data under the
appropriate regulatory standards, as provided by the review team toward the start of the meeting,
had been forgotten by the committee members.
Dr. Woodcock explained to the SDR Board that she thought both that the review team did a
poor job framing the issues during their presentations and that the questions were confusing and
poorly worded. Indeed, during her interview with the SDR Board, Dr. Woodcock opined that the
review team “did not put its best foot forward.” She speculated that the confounding factor was
the number of interested persons attending both in person and by webcast. She stated that she
did not interfere with either aspect of the AC meeting because she knew she disagreed with the
review team and Dr. Unger had already signaled that he would file an SDR appeal if she decided
to grant accelerated approval to eteplirsen. She thought that the review team’s presentation of
the IHC data, in particular, was confusing. She further opined that the review team’s failure to
highlight the clinical data made the questions on conventional approval and accelerated approval
difficult for the committee members to understand. Dr. Woodcock also criticized the review
team for how it downplayed and undercut the views of the patient advocates.
At the conclusion of the AC meeting, the committee voted against accelerated approval by a
margin of 7-6. 77 Three of the members who voted in favor of accelerated approval were the
consumer representative and the two patient representatives. 78
74

Id. at 158-59.
Id. at 484.
76
Id. at 548-549.
77
Id. at 486-95.
78
Id. at 2-7, 486-88.
75

12

 On May 4, 2016, Dr. Woodcock met with the review team to discuss the AC meeting and plan of
actions for the NDA. 79 In his appeal, Dr. Unger contends that Dr. Woodcock “made clear her
intent to approve the drug” at this meeting, even though she had not yet reviewed drafts of
DNP’s final review memorandum or his review memorandum. 80 According to Dr. Unger, Dr.
Woodcock explained that she had already “reached a different conclusion” than the review
team. 81 Dr. Woodcock explained to the SDR Board that the memoranda were discussed during
the Center Director briefings and that she felt she understood the views of the review team and
did not see the point of an “exchange of reviews.”
On May 24, 2016, Dr. Unger met privately with Dr. Woodcock to discuss the eteplirsen
decision. 82 On May 31, 2016, Dr. Woodcock met with representatives from the review team to
discuss their reviews and her initial draft of a decisional memorandum based primarily on the
data from Study 201/202. 83 Dr. Woodcock received comments back from the review team at the
same meeting. 84 Dr. Unger told the SDR Board that he and members of the review team—
including Dr. Robert Temple, Deputy Center Director for Clinical Science and Dr. John Jenkins,
Director of OND—discouraged Dr. Woodcock from finalizing the decisional memorandum and
granting accelerated approval for eteplirsen until the additional data from Study 301 could be
obtained.
On June 3, 2016, in response to an email from Sarepta, a letter signed by Dr. Woodcock issued to
the sponsor. 85 The letter requested the additional data from Study 301, which was to include
comparisons of any biopsy samples obtained at Week 48 to the respective baseline samples for
those patients. 86 The letter stated,
If you are successful in showing, to FDA’s satisfaction, a meaningful increase in
dystrophin by Western blot analysis between the paired pre-and post-treatment
samples, we expect to be able to grant an accelerated approval within four
business days of receiving the data (assuming all other aspects of the application
are approvable). 87
Dr. Woodcock explained that Dr. Unger and the review team essentially agreed to the timeframe
of four business days, though they pushed instead for six. She felt that there was general
agreement that data from only twelve patients could be reviewed quickly, especially given that
representatives from CDER would be overseeing the Western blot analysis and ensuring that it
was done properly.
On June 27, 2016, Sarepta submitted the requested data. 88 Dr. Woodcock explained that
accelerated approval was not granted within four business days of that date precisely because the
results of the analysis were disappointing in that they provided evidence of only a minimal
increase in dystrophin at 48 weeks. Dr. Unger sent an email to Dr. Woodcock that read:
79

Appeal at 25; Behr Chronology at 2.
Appeal at 26.
81
Id.
82
Behr Chronology at 2.
83
Appeal at 26; Behr Chronology at 2.
84
Behr Chronology at 2.
85
June 3, 2016, General Advice letter.
86
Id. at 1.
87
Id. at 1-2.
88
Unger email to the SDR Board, dated July 22, 2016.
80

13

 I don’t have to tell you how difficult the eteplirsen decision has been for many of
us in ODE-I. As you know, we have reached different scientific conclusions on
the strength of the data, and in particular, the likelihood that the small increase
observed in Becker-type dystrophin is reasonably likely to predict clinical benefit.
This decision could be precedent setting with respect to accelerated approval, i.e.,
where the bar should be set for changes in a pharmacodynamic biomarker that are
deemed “reasonably likely to predict clinical benefit.” Moreover, to my
knowledge, this could be the first time a Center Director has overruled a review
team (and an advisory committee) on a question of whether effectiveness has been
demonstrated.
I know that Dr. Jenkins has mentioned the possibility of involving Dr. Califf in
the eteplirsen decision on at least one occasion, and I would like to request a
formal appeal to the Commissioner on this matter.
I’m aware that the Commissioner’s official role is to consider the administrative
aspects of review decisions and not the science. But given the potential for
setting a precedent here, I think he should be aware of the various points of view
and consider the potential ramifications of the matter at hand.
I’m also aware that you advised Sarepta that we would be prepared to grant
accelerated approval of their NDA within 4 business days of receiving their new
data, but there was a provision in the letter that the increase in dystrophin had to
be meaningful, and we do not have agreement on this point. Thus, it is my hope
that a Commissioner Briefing can be held before an action is taken.
I have discussed the above with Dr. Jenkins, and he supports this course of action.
I propose that we reserve a few minutes at the briefing tomorrow to discuss this
matter. 89
On July 6, 2016, Dr. Woodcock met with the review team one final time. 90 During the meeting,
Dr. Woodcock “indicated to the review team that [she] had read their memoranda that had been
updated to reflect the new [Western blot] data, and that [she] maintained [her] position that the
application should receive accelerated approval based on dystrophin production.” 91 She
discussed her rationale, which—based on her notes—appears to have tracked the rationale in her
final decisional memorandum. 92
On July 8, 2016, in light of Dr. Unger’s stated intention of filing an appeal with the Office of the
Commissioner, Virginia Behr, CDER Ombudsman, began working with him and Dr. Woodcock
to determine whether the institution of any formal appeals under CDER’s SDR-SOPs was
warranted. 93 Ms. Behr had determined that the procedure outlined in MAPP 4151.1,
89

Unger email dated July 5, 2016.
Appeal at 26; Behr Chronology at 3.
91
Woodcock’s handwritten notes, dated July 6, 2016, at 1.
92
Id. at 2. Also of note, on July 7, 2016, Dr. Unger briefed you on his rationale for disagreeing with Dr. Woodcock’s underlying
scientific reasoning for granting accelerated approval for eteplirsen (Behr Chronology at 3).
93
See “Agreement to utilize FDA Staff Manual Guide 9010.1 for internal appeal related to NDA 206488, eteplirsen injection” (SDRSOPs Agreement).
90

14

 “Scientific/Regulatory Dispute Resolution for Individuals Within a Management Chain” did not
apply because the disagreement was between the Center Director and a subordinate two levels
below her. 94 She also questioned the utility of using MAPP 4151.2, “Resolution of Differing
Professional Opinions: Review by Ad Hoc Panel and CDER Director.” 95 She reasoned that “the
CDER Director ha[d] already fully evaluated the issues and [was] one of the parties involved in
the dispute” and that “utilizing this MAPP could potentially extend this already lengthy NDA
action another 50 business days.” 96 She nonetheless consulted with both Drs. Unger and
Woodcock, who both agreed to bypass the Ad Hoc panel process in favor of the process outlined
in the SDR-SMG. 97 During his presentation to the SDR Board, Dr. Unger also indicated that he
thinks referring the matter to an Ad Hoc panel would have been pointless because Dr. Woodcock
had already made up her mind and a new process would not have changed the outcome.
On July 11, 2016, Dr. Woodcock provided a draft of her final decisional memorandum to the
review team. 98 She received comments back from Dr. Unger; Dr. Jenkins, the Director of OND;
and Dr. Ashutosh Rao, of the Office of Biotechnology Products, who was also on the review
team. 99 The comments from Drs. Unger and Rao do not debate the action proposed in Dr.
Woodcock’s draft decisional memorandum or its underlying scientific conclusions. 100 Instead,
they focus on clarifying certain facts asserted in the memorandum, and Dr. Unger provided
information regarding the clinical course of 11 patients enrolled in Study 201/202 to 240
weeks. 101 Dr. Jenkins provided more detailed analysis on and critique of some of Dr.
Woodcock’s findings and he expressed concern about her conclusions. However, he made no
attempt in his written comments to dissuade her from her ultimate conclusion regarding
accelerated approval. 102 By email on the afternoon of July 13, Dr. Unger stated, “I’ve canvassed
the Division, and we have no additional comments.” 103 Dr. Unger told the SDR Board that he
and the review team understood that Dr. Woodcock had already made up her mind and that thus
they did not see a point in criticizing Dr. Woodcock’s draft decisional memorandum.
Furthermore, the RTM told the SDR Board that some of the positions taken by Dr. Woodcock in
the draft decisional memorandum were brand new to him but that he did not feel any feedback he
could provide would receive due consideration by Dr. Woodcock. The RTM expressed concern
that Dr. Woodcock’s analysis for “reasonably likely to predict clinical benefit” raised new issues
and information that should have been presented at the beginning of the review and that had not
been addressed by the review team or, perhaps more importantly, presented by the sponsor in
support of the NDA. The RTM specifically discussed with the SDR Board the section of the
finalized version of the memorandum addressing whether the data for eteplirsen is adequate to
show a reasonable likelihood of predicting clinical benefit. 104 As an example of his concerns, the
RTM pointed to section (B)(5) of the decisional memorandum, which details the findings in the
94

Id. at 1. It is also clear from the record before the SDR Board that the supervisor between Drs. Unger and Woodcock, Dr. John
Jenkins, agreed with Dr. Unger.
95
Id.
96
Id. at 2.
97
Id.
98
Behr Chronology at 3.
99
Unger emails dated July 13, 2016 and sent at 12:57 AM, 9:26 AM (including attachments), and 11:19 AM (including attachment);
Jenkins email dated July 12, 2016; and emails (including attachments) from Rao dated July 12 and 13, 2016.
100
Unger emails dated July 13, 2016 and sent at 12:57 AM, 9:26 AM (including attachments), and 11:19 AM (including attachment);
emails (including attachments) from Rao dated July 12 and 13, 2016.
101
Unger emails dated July 13, 2016 and sent at 12:57 AM, 9:26 AM (including attachments), and 11:19 AM (including attachment);
emails (including attachments) from Rao dated July 12 and 13, 2016.
102
Jenkins email dated July 12, 2016.
103
Unger email dated July 13, 2016 and sent at 3:19 PM.
104
Woodcock Decisional Memorandum at 5-10.
15

 scientific literature regarding “the relationship of dystrophin expression to clinical status.” 105
The RTM indicated that he or she knows the scientific literature at issue very well and that he or
she could have provided significant input into the evaluation of the literature and the underlying
data and analysis. The RTM conveyed that he did not do so because he felt Dr. Woodcock had
already made her decision.
On July 14, 2016, Dr. Woodcock finalized her decisional memorandum. She explained to the
SDR Board that her conclusion regarding whether the increase in dystrophin production
identified by Studies 202 and 301 was reasonably likely to predict clinical benefit was based on
her own “medical/scientific judgment.” She emphasized that she has thirty years of experience
at FDA and that she has far more experience in assessing this type of evidence for an “ultra-rare
rare” disease than the review team. She thought that the review team was unreasonable in its
position on a threshold for predicting clinical benefit in this case. Her stated goal for the
decisional process was to move the review team toward what she viewed as a more reasonable
approach. She acknowledged that there were clear weaknesses in the data but that accelerated
approval should not be limited to “sure bet” drugs and that confirmatory trials are required for a
reason. Dr. Woodcock emphasized her view that the agency needs to accept more uncertainty
when granting accelerated approval. She also criticized OND for not issuing clear guidance on
what constitutes a sufficient drug effect to be “reasonably likely to predict clinical benefit,” as
she had suggested for an extended period of time. She also thought that the review team’s views
on balancing the mean results of a clinical study with a targeted evaluation of responsive patients
were misplaced, particularly in a DMD population, where additional genetic mutations or
deficiencies could have a profound effect on the outcome.
In her presentation to the SDR Board, Dr. Woodcock suggested that, in making the decision, she
was looking at the broader picture for the development of these types of drugs for very limited
patient populations in the United States (between 600 and 1300) and that there needed to be
some path forward for such innovative products. She opined that Sarepta in particular “needed
to be capitalized.” She noted that the sponsor’s stock went down after the AC meeting and went
up after FDA sent the June 3, 2016 letter. Dr. Woodcock cautioned that, if Sarepta did not
receive accelerated approval for eteplirsen, it would have insufficient funding to continue to
study eteplirsen and the other similar drugs in its pipeline. She stated that, without an approval
in cases such as eteplirsen, patients would abandon all hope of approval for these types of
products and would “lapse into a position of” self-treatment.
On July 16, 2016, Dr. Unger finalized his own decisional memorandum. In her own decisional
memorandum, dated July 14, 2016, Dr. Woodcock indicated that she had read Dr. Unger’s
decisional memorandum, 106 although she could not have done so given the timing of the two
memoranda. She explained to the SDR Board that she did not feel she needed to see a finalized
version of Dr. Unger’s decisional memorandum because she was already familiar with his views
on the data and the decision. She also stated that there was nothing in Dr. Unger’s appeal, which
is based largely on his finalized decisional memorandum, that would have changed her mind on
her decision or the underlying rationale. She stated, “He is entitled to his own opinion.”
5. Dr. Unger’s SDR Appeal
In his appeal, Dr. Unger focuses his arguments almost exclusively on the substance of his
scientific disagreement with Dr. Woodcock. Indeed, Dr. Unger makes clear in his appeal that he
105
106

Id. at 7-10.
Id. at 1.
16

 seeks “a scientific review on the matter of whether or not there is substantial evidence of a
quantitative effect on dystrophin protein that is reasonably likely to predict clinical benefit.” 107
Insofar as he explicitly addresses potential procedural issues under the review process
contemplated by the SDR-SMG, he does so in two paragraphs toward the end of the appeal. 108
He first states that Dr. Woodcock’s “direct involvement with this drug, compared to other
development programs, has been unprecedented.” 109 He states further that “[s]he also attended
the April meeting of the Peripheral and Central Nervous System Drugs Advisory Committee,
where she spoke and interjected a number of important comments.” 110 After conceding that
“[t]here is no question that there has been adequate time and place for the discussion of various
views,” Dr. Unger notes that he found it unfortunate that “the Center Director made clear her
intent to approve the drug at a briefing with the review team on May 4, 2016, before she had
seen drafts of the Division’s final review memorandum or my review memorandum.” 111 As
noted above, Dr. Unger indicates that Dr. Woodcock conveyed that she had “already ‘. . .
reached a different conclusion . . .’ than the review team.” 112
In his presentation to the SDR Board, Dr. Unger highlighted that Dr. Woodcock had never seen
the charts on page 10 of his appeal. Those charts show: (1) a comparison of the original IHC
results for baseline samples in the three patients whose biopsies were available at 180 weeks to
the IHC results for those same samples when they were re-evaluated after 180 weeks and (2) a
comparison of the IHC and the Western blot results at 180 weeks. 113 Dr. Unger stated, however,
that those charts were consistent with his earlier positions and would likely not affect Dr.
Woodcock’s analysis or decision. In a follow-up email to the SDR Board, Dr. Unger also
contended that Dr. Woodcock diverted from protocol when she finalized her decisional
memorandum on July 14, 2016, two days before his.
In his appeal, Dr. Unger frames his scientific disagreement with Dr. Woodcock as follows: “The
disagreement is over the question of whether the findings on the dystrophin surrogate endpoint
are reasonably likely to predict clinical benefit.” 114 Nonetheless, Dr. Unger explains his
disagreement with Dr. Woodcock through multiple challenges to the reliability of the underlying
data and specific issues he has with her rationale or the evidentiary basis for such rationale. Of
note, he makes the following scientific arguments:
x
x

As noted above, Study 201 showed only “a nominally statistically significant
increase in dystrophin in response to eteplirsen for the low dose group…”; 115
Study 201/202 was fundamentally flawed in several respects:
o “[T]he baseline biopsies were obtained from [external controls] . . . who
could differ in unknown ways from the subjects in Study 201/202”; 116
o “[T]he Week 180 biopsies were obtained from different muscles than the
baseline biopsies”; 117 and

107

Appeal at 26.
Id.
109
Id.
110
Id.
111
Id.
112
Id.
113
Id. at 10.
114
Id. at 3.
115
Id. at 9.
116
Id. at 5.
117
Id.
108

17

 x

o “The baseline biopsies for the three subjects with Week 180 data had been
stored for several years and the protein may have degraded, leading to a
falsely low baseline value, and a greater apparent increase from
baseline….” 118
Although the available data generated by Study 301 were the product of an
adequate and well-controlled study and showed a statistically significant increase
of dystrophin, the drug effect (i.e., an increase from 0.22% to 0.32% of normal) is
not reasonably likely to predict clinical benefit. 119
o “The treatment effect observed cannot be compared or related to levels of
dystrophin measured by other laboratories and reported in various
publications”; 120
o “Dr. Woodcock never provides a rational argument – based on reliable
data – to support the concept that ‘…low-level increases in dystrophin
production are reasonably likely to predict clinical benefit.’ She provides
no rationale – no link between a mean increase in dystrophin of 3 parts per
thousand and clinical benefit”; 121 and
o “No evidence of a clinical effect was demonstrated in the eteplirsen
development program, and there is no correlation between dystrophin
levels as determined by Western blot and clinical outcome.” 122

He also makes several overarching policy and legal arguments that call into question the
appropriateness of Dr. Woodcock’s decisional memorandum. His key arguments focus on the
effects that Dr. Woodcock’s decision would have on the pathway for accelerated approval and
the standard for “reasonably likely to predict clinical benefit.” 123 He also highlights the negative
effects that accelerated approval would have on the patients themselves, including false hope,
abandonment of other therapies, and a decline in drug development for DMD. 124 He further
questions “the ethics of approving or prescribing a drug for a fatal disease at a dose that is very
likely to be sub-therapeutic[] when the consequence of a sub-therapeutic dose is clinical
deterioration and death.” 125 Finally he worries that approving eteplirsen based on the data
submitted by the sponsor “would send the signal that political pressure and even intimidation—
not science—guide[] FDA decisions.” 126
ANALYSIS
1.

Whether CDER followed its own processes.

The first issue for the SDR Board to consider is whether CDER followed its own processes in
addressing Dr. Unger’s scientific dispute. Dr. Unger does not contend that there were any issues
with respect to how CDER chose to address and implement its own formal appeals process under
the SDR-SOPs in this case. In his appeal, Dr. Unger points instead to four deviations from
typical Center process: (1) Dr. Woodcock’s involvement in the early stages of review of the
eteplirsen NDA; (2) her extensive involvement in planning the AC meeting and her participation
118

Id.
Id. at 7.
120
Id. at 13.
121
Id. at 15.
122
Id.
123
Id. at 21-22.
124
Id.
125
Id. at 23; see also Unger Review Memorandum at 4, 5.
126
Id.
119

18

 in the meeting; (3) her initial decision (on May 4, 2016) to approve the eteplirsen NDA before
the review team had completed even their draft review memoranda; and (4) her issuance of her
final decisional memorandum before Dr. Unger finalized his own decisional memorandum as
Director of ODE-I. In its review of the administrative file and the surrounding circumstances,
the SDR Board has also identified below other potential deviations from process at the Center
level.
The agency-wide SDR-SMG directs the SDR Board to focus on the Center’s SDR-SOPs in
evaluating whether the Center followed its own processes in evaluating a scientific dispute. In
this case, however, both Drs. Unger and Woodcock have agreed that the only applicable SDRSOP, MAPP 4151.2, provides for a review by the Center Director in consultation with an Ad Hoc
panel and that going through such a process at this stage would be futile. The SDR Board has
determined that, absent the second aspect of that agreement regarding futility and the underlying
unusual circumstance of this scientific dispute, there would be reason to refer the matter back to
the Center for further review by an Ad Hoc panel.
The interplay between MAPP 4151.1 and 4151.2, the former of which provides for supervisory
review of scientific disputes all the way to the Center Director, suggests that MAPP 4151.2
actually calls for additional review of a scientific dispute by the Center Director under certain
circumstances even if she has already made a decision on the dispute. Although MAPP 4151.2
provides for bypassing review of the scientific dispute up the chain of command under MAPP
4151.1 if such exhaustion would impede the timely resolution of a serious public health issue,
MAPP 4151.2 also emphasizes that it should not be used before other means of resolution have
been attempted. 127 However, the key consideration for obtaining review by an Ad Hoc panel
under MAPP 4151.2 is “whether the consequences of the decision in question are potentially
serious enough to warrant [additional review],” not whether the resort to the process would be
futile. 128 It appears that Dr. Woodcock has never made a determination regarding the
seriousness of the decision in question, but it would be surprising if she determined that the
dispute in this case did not meet the standard, as reflected in the statement she signed. 129
In this case, however, it is clear from the record before the SDR Board that Dr. Woodcock was
so involved in the underlying scientific dispute—including direct and extensive personal review
of the data and analyses offered in support of the NDA—that we agree with the conclusion in the
agreements signed by Drs. Unger and Woodcock that “the CDER Director has already fully
evaluated the issues.” 130 Indeed, she has already received advice from an advisory committee
and had substantial conversations with her staff over an extended period of time with respect to
the dispute in question. There is no reason to believe that receiving additional advice from an Ad
Hoc panel would alter Dr. Woodcock’s views of the scientific issues. As the agreement between
her and Dr. Unger reflects, the process would be time-consuming and delay an important

127

MAPP 4151.2 at 5. (“In most cases, the Ombudsman will ensure that all other avenues for resolution (e.g., dispute resolution
process, Advisory Committee discussion, CDER regulatory briefing) have been exhausted before a [request for review under 4151.2]
is filed. However, in some cases, an individual may believe that his or her professional opinion will not be considered by his or her
supervisors or that there is not time to exhaust other options for dispute resolution without seriously endangering the public health. In
this case, the submitter should include . . . a written request to bypass these other mechanisms. . . .).
128
Id.
129
SDR-SOPs Agreement at 2 (“The difference of opinion between Drs. Unger and Woodcock could be considered to meet the criteria
for filing an appeal under MAPP 4151.2 because the drug indication sought is one for a serious and life-threatening disease that has
limited treatment options.”).
130
Id.
19

 regulatory decision unnecessarily. 131 Dr. Unger also told the SDR Board that he thought going
through the Ad Hoc panel process would have been pointless for the aforementioned reasons.
The difficulty for the SDR Board is that the agency-wide SDR-SMG is predicated on some level
of formal scientific dispute resolution within the Center, particularly a decision by the Center
Director regarding the formalized scientific dispute. 132 For that reason, the focus of the SDRSMG with respect to the process followed is on whether the Center followed its own SDR-SOPs
in resolving the scientific dispute. 133 Yet, the SDR-SMG also directs the Centers to adopt
“[i]nformal methods” for resolving scientific disputes, “to create an atmosphere in which
consultation and open discussion on controversial issues are encouraged,” to use “good
management practices for resolving conflict,” and “to enable as open and complete a discussion
of the issues as possible at the working level of the organization.” 134 As a result, the SDR Board
has determined that reviewing the processes used by a Center to resolve a scientific disagreement
is appropriate under the SDR-SMG even when, as here, the initiator has not availed himself of
the Center’s formal process for resolving scientific disputes and the Center Director has
explicitly agreed to that approach.
Whether the Center followed its own processes for resolving a scientific disagreement cannot be
viewed in a vacuum, however. Indeed, the SDR-SMG itself—at its most concise and in its
clearest voice—states, “The goal of [the SDR Board’s] review is to determine if the processes
followed in the Center fully considered all relevant evidence and provided the initiator with an
opportunity to express his or her concerns at all appropriate levels, prior to and including the
Center Director.” 135 Particularly in the context of a scientific dispute that did not go through a
formal SDR process at the Center but nonetheless received extensive review by the Center
Director, focusing on deviations from process without any regard to whether they affected the
initiator’s opportunity to present his views of the science (and to some extent whether those
views and the evidence were considered) would seem to miss the point of that review.
Accordingly, the SDR Board finds that it is more appropriate to address Dr. Unger’s arguments
regarding the Center’s deviations from appropriate process under the second prong of its
analysis: whether the Center provided Dr. Unger an adequate opportunity present his scientific
concerns.
The SDR Board’s one caveat is that, as noted above, the SDR-SMG does appear to assume that
there has been both at least some use of the formal dispute resolution within the Center and,
accordingly, a formal substantive review of the initiator’s scientific concerns before reaching the
Office of the Commissioner. 136 The limited scope of the SDR Board’s review under the SDRSMG—i.e., an evaluation of the Center’s decision-making process—means that Dr. Unger will
also not receive a substantive review of his scientific concerns under the SDR-SMG. In fact, at
the conclusion of the SDR Board’s review, Dr. Unger will not have received a substantive
review of his scientific concerns under any formal process at any level. Particularly in light of
131

Id. (“[U]tilizing this MAPP could potentially extend this already lengthy NDA action another 50 business days.”).
See SDR-SMG at 6 (requiring as a mandatory process for formal scientific dispute resolution a written opinion by the Center
Director and stating that such a written opinion as a step in the process is a “central criterion for advancement to the agency-level
appeals process.”).
133
See, e.g., id. at 12 (requiring the SDR Board to “obtain the full administrative record of the Center’s processes for the dispute and
review the Center’s published SOP(s)” and to “review that information to determine whether written Center processes were
followed.”).
134
Id. at 6.
135
Id. at 12.
136
See id. at 6 (referring to SOPs for resolution of Center-level scientific disputes without limiting them to procedural reviews and
contemplating the Center SOPs as a continuation of the informal SDR process).
132

20

 Dr. Unger’s explicit request for scientific review of the matter within the Office of the
Commissioner, therefore, the SDR Board recommends additional substantive review at this level,
as is discussed below.
2. Whether CDER provided an adequate opportunity to Dr. Unger to present his
scientific concerns.
In his appeal, Dr. Unger admits, “There is no question that there has been adequate time and
place for the discussion of various views.” 137 In so doing, he appears to concede away most of
his arguments with respect to whether he had an adequate opportunity to present his scientific
concerns, notwithstanding the procedural deviations he identifies. The SDR Board, however, has
not taken Dr. Unger’s concession at face value and has instead looked beyond it to evaluate the
administrative file and the surrounding circumstances to identify additional procedural issues.
We conclude nonetheless that Dr. Unger had an adequate opportunity to present his scientific
concerns to Dr. Woodcock before she issued her decisional memorandum.
As noted above, Dr. Unger identified four deviations from Center’s typical decision-making
process for the eteplirsen NDA: (1) Dr. Woodcock’s involvement in the early stages of review of
the eteplirsen NDA; (2) her extensive involvement in planning the AC meeting and her
participation in the meeting; (3) her initial decision (on May 4, 2016) to approve the eteplirsen
NDA before the review team had completed even their draft review memoranda; and (4) her
issuance of her final decisional memorandum before Dr. Unger finalized his own decisional
memorandum as Director of ODE-I. In reviewing this matter, the SDR Board—which includes
among its members Ombudsmen from other Centers that oversee reviews of medical products—
also considered other departures from the typical processes used by Centers in reviewing
applications for pre-market approval or clearance. 138
The SDR Board agrees with Dr. Unger that it was unusual for a Center Director to be so
involved in the early stages of reviewing an NDA, but the consensus on the SDR Board was that
Dr. Woodcock went several steps further than mere involvement and thereby departed from
typical practice among the Centers. By her own admission, Dr. Woodcock had a direct hand in
reviewing the data submitted in support of the NDA, even before the review team had written
their draft review memoranda, and actively encouraged the review team—including Dr. Unger—
to come around to her way of thinking in their own reviews. Specifically, she wanted the review
team to agree with her that the limited increase in dystrophin production established by the data
in Studies 201/202 was sufficient to show a reasonable likelihood of predicting clinical benefit.
At several points during the decision-making process for what is clearly a critical scientific issue
for the agency, Dr. Woodcock also provided a very limited amount of time for Dr. Unger and the
review team to provide feedback on additional data or her own scientific conclusions—most
notably when Sarepta submitted the data from Study 301 and when she provided two separate
draft versions of her decisional memorandum to the review team.
Notwithstanding the foregoing procedural shortcomings, the SDR Board finds that Dr. Unger
had an adequate opportunity to present his scientific views. Not only does he admit in his appeal
that he had an opportunity, but the record before the SDR Board demonstrates that he did. He
and the rest of the review team met with Dr. Woodcock on multiple occasions both before and
after the AC meeting. Drs. Unger and Woodcock both told the SDR Board that those meetings
involved substantive and detailed discussions of the data and science and the appropriate
137
138

Appeal at 26.
See SDR-SMG at 3 (defining the SDR Board to include Ombudsmen from all of the Centers).
21

 conclusions to be drawn from them. Although Dr. Unger complains that Dr. Woodcock was
involved in aspects of the NDA that went far beyond the norm for a Center Director at CDER,
including her role in the AC meeting, and that she reached or finalized decisions before
reviewing review or decisional memoranda, he does not maintain that those procedural
deficiencies compromised his ability to present his views. In fact, his own final decisional
memorandum—which Dr. Woodcock apparently saw in draft form before she finalized her
own—discloses that he felt empowered to push back on both Dr. Woodcock’s scientific
conclusions and their basis, despite the fact that he believed his efforts would be futile. Indeed,
he conceded to the SDR Board that nothing in his decisional memorandum or appeal submission
would have affected Dr. Woodcock’s decision on the scientific issue in question (including the
charts that he created for the first time in preparing his appeal submission under the agency-wide
SDR-SMG). He further conceded as much when he agreed not to pursue further review through
the Ad Hoc panel process under CDER’s SDR-SOPs. In short, through his own perseverance,
confidence in his own scientific expertise, and perhaps dint of personality, Dr. Unger ensured
that he himself had an adequate opportunity to present his scientific views despite the procedural
irregularities in the decision-making process within CDER.
The SDR Board nonetheless remains concerned about Dr. Woodcock’s extensive involvement in
the review of the eteplirsen NDA, including her degree of participation at the AC meeting, and
the limited timeframe she provided for feedback on the data from Study 301 and her own
scientific conclusions on that data. We fear that those actions could have chilled scientific
debate within CDER and reduced the level of participation by the review team during the final
stages of the decision-making process. By all accounts, Dr. Woodcock made clear her views that
CDER should lean toward finding that eteplirsen met the standards underlying accelerated
approval nearly from the outset of her involvement. By May 4, 2016, she had orally
communicated her intention to grant accelerated approval for eteplirsen, even though she had not
yet seen even the draft review memoranda from the review team or a decisional memorandum
from Dr. Unger. Then, when she requested data from Study 301 from Sarepta, she
communicated to the sponsor a compressed timeframe for CDER’s review. Although she later
expanded the timeframe for review when the data proved to be disappointing, she apparently
analyzed the data on her own, conducted her own additional search of the scientific literature,
and took only six or seven business days to orally communicate to the review team her decision
to grant approval.
To complicate matters further, Dr. Woodcock subsequently circulated a draft decisional
memorandum but provided only a limited amount of time for comments, even though the draft
decisional memorandum was the first time some on the review team had apparently seen key
elements for the basis of her decision on “reasonably likely to predict clinical benefit.” The
response from the review team is telling. As noted above, only Drs. Jenkins and Unger and
another reviewer outside of DNP provided comments. Except for Dr. Jenkins, no one made any
effort to make substantive comments beyond tips on how to make factual clarifications or to
supplement her analysis with additional data. It appears that, because the review team knew Dr.
Woodcock’s views by then, they saw no point in providing any additional substantive review or
meaningful feedback on any new issues raised by Dr. Woodcock’s memorandum. Indeed, Dr.
Unger and the RTM conveyed as much to the SDR Board.
There is no doubt that a Center Director should have wide latitude in leading the direction of the
Center in a manner consistent with her priorities and vision. The SDR Board also believes that
Center Directors have a role to play not only with respect to the resolution of scientific disputes
at issue in individual applications for pre-market-authorization by FDA, as evidenced by both the
SDR-SMG and CDER’s own SDR-SOPs, but also with respect to the ultimate decision on
22

 scientific issues that are not the subject of a dispute. It is also clear from Dr. Woodcock’s
presentation to the SDR Board that she firmly believes in the correctness of her scientific
decision in this case and that her involvement in the review of the eteplirsen NDA was always
motivated by the best of intentions. However, the SDR Board finds Dr. Woodcock’s extensive,
early involvement in the review process troubling. Indeed, her involvement here appears to have
upended the typical review and decision-making process.
Rather than ensuring that the scientific reviews started at the bottom of the chain of command,
Dr. Woodcock made clear from her position at the top that she was pushing for a particular
outcome from the very early stages. As a consequence, the regulatory reviews did not start at the
staff level with scientific reviews and then proceed through the chain of command for
concurrence or non-concurrence at all appropriate levels within the management structure, as
would be the typical course of decision-making for a regulatory decision grounded in science.
Indeed, before the reviewers had even completed their draft scientific reviews, Dr. Woodcock
had told them—on May 4, 2016—that she intended to grant accelerated approval. This sort of
top-down review does not, in the SDR Board’s view, “create an atmosphere in which
consultation and open discussion on controversial issues are encouraged,” as reflected in the
SDR-SMG’s requirements for resolution of scientific disagreements by the Center. 139 By the
time Dr. Woodcock issued her draft decisional memorandum on what she herself acknowledged
was a difficult scientific issue of incredible magnitude for the agency—i.e., whether the evidence
regarding dystrophin production was reasonably likely to predict clinical benefit—the review
team had decided it was pointless to challenge her ultimate conclusion or its basis. 140 Review
teams should have the opportunity to conduct their reviews without preemption by the Center
Director. As noted above, the SDR Board believes that Center Directors should have a role in
shaping policy, expressing concerns, and resolving issues once they are ripe for their review, but
we caution that care should be taken to avoid the appearance of interfering with the integrity of
scientific reviews at the lower levels of a Center.
3. Whether the Center Director considered all relevant evidence bearing on the
scientific question at issue.
The third issue for the SDR Board is whether CDER, including Dr. Woodcock, fully considered
all relevant evidence in resolving the scientific dispute at issue, i.e., whether the evidence of
eteplirsen’s effect on dystrophin production is reasonably likely to predict clinical benefit. In
this case, both Drs. Unger and Woodcock appear to agree that she considered all relevant
evidence. As noted above, Dr. Unger does not believe that any additional data or evidence
available to him could persuade Dr. Woodcock that she has reached the wrong scientific
conclusion. For her part, Dr. Woodcock does not feel that she has disregarded any relevant
evidence. Moreover, in her interview with the SDR Board, she demonstrated an awareness and
command of all of the evidence weighing against the scientific decision she has made, including
the arguments and analysis of the evidence presented in Dr. Unger’s appeal.
Whether Dr. Woodcock has addressed all of the relevant evidence in her decisional
memorandum is a more difficult question. In concluding that the minimal increase in dystrophin

139

Id. at 6.
In this regard, it is also worth noting again the language quoted above in the background section: “Each individual who contributes
to the decision-making process” must “be sure the position represented is consistent with the scientific, regulatory, and/or
administrative policies of that . . . organizational component” and that “[o]pinions of staff should be documented and supported by
data in a matter commensurate with the magnitude of the decision being made.” (MAPP 4151.8 at 2-3).
140

23

 production seen in the data is reasonably likely to predict clinical benefit, Dr. Woodcock has
provided a very limited rationale.
At the risk of oversimplification, Dr. Woodcock found, in essence, that the studies attempting to
correlate levels of dystrophin with clinical benefit, as have been reported in the scientific
literature, are unreliable in this context for variety of reasons, including: (1) the subjectivity of
the clinical evaluation, (2) the difficulty in correlating IHC results with Western blot results, (3)
the influence of anti-dystrophin antibodies, (4) the lack of information on dystrophin quality (as
opposed to quantity) in the different studies, (5) deficiencies in Western blot techniques from
earlier studies, and (6) the wide range of findings with respect to the correlation of dystrophin
levels with clinical benefit. 141 She concluded, therefore, that “protein in the range between
undetectable and 10% of normal is likely to be very important for clinical presentation, all other
things being equal, i.e. mutation status and non-dystrophin-related factors affecting phenotype,”
and that the “biochemical data strongly support the idea that low-level increases in dystrophin
production are reasonably likely to predict clinical benefit.” 142 She then attempted to bolster that
conclusion with a theory regarding the effect of exon 52 deletion and her reanalysis of the
intermediate clinical outcomes for a subset of subjects in Study 201/202. 143 She further
explained to the SDR Board that she was exercising her “medical/scientific judgment” in
reaching the scientific conclusion that she did.
It is easy for the SDR Board to understand why Dr. Unger’s appeal expressed such frustration
with this explanation of Dr. Woodcock’s rationale. He states:
I believe the burden is on Dr. Woodcock to show or explain why production of a nearzero quantity of dystrophin (0.3%) is reasonably likely to predict clinical benefit, and I do
not believe her July 14, 2016 memo makes this case. I believe that the available evidence
leaves open the possibility that some patients could benefit from a small increase in
dystrophin, but this possibility does not reach the threshold of being reasonably likely to
predict a clinical benefit. 144
Of course, considering the relevant evidence and addressing the relevant evidence in a manner
satisfactory to Dr. Unger or the SDR Board are two different propositions. The SDR Board
finds, based on the record before us, that Dr. Woodcock has considered all relevant evidence in
reaching her scientific conclusion. Based on her own medical judgment, she simply has a
difference of opinion with Dr. Unger—both with respect to the scientific conclusion and the
sufficiency of the underlying rationale.
4. Whether the dispute should be remanded to the Center Director.
Inasmuch as the SDR Board has concluded that Dr. Unger had an adequate opportunity to
present his scientific concerns during the decision-making process at CDER and that Dr.
Woodcock considered all relevant evidence in making her decision, the SDR Board does not
recommend returning this matter to the Center Director for corrective action. We also believe
that, for reasons discussed above, remanding this matter to the Center Director would be futile.

141

Woodcock Decisional Memorandum at 5-9.
Id. at 9.
143
Id. at 10.
144
Appeal at 20 (emphasis in original).
142

24

 CONSIDERATIONS FROM THE ACTING CHIEF SCIENTIST
In my capacity as Acting Chief Scientist, I feel the responsibility to convey some comments
regarding the underlying science for the decision being challenged by Dr. Unger in his appeal. I
cannot begin to understand the depth of pain and suffering that patients with DMD and their
families endure. As an experienced physician, I struggle to identify any other diseases associated
with this degree of suffering, not only to patients but to their families. Nevertheless, my
assessment is that the data presented by the sponsor to date are not adequate to support
accelerated approval of eteplirsen
Studies in animals showing that eteplirsen leads to “exon 51 skipping” are an important first step
in assessing whether eteplirsen might work for a subset of patients with DMD because skipping
exon 51 is necessary for the production of dystrophin in these patients. 145 The next step is to
assess whether eteplirsen actually leads to the production of dystrophin in patients with DMD
and, if so, whether such an increase in dystrophin confers clinical benefit. Despite the promising
animal studies demonstrating exon 51 skipping, both Drs. Woodcock and Unger, as well as the
review team in CDER, agree that the amount of dystrophin produced in the clinical studies
conducted at doses of up to 50mg/kg per week is very low. Animal data suggest that the doses
studied in humans is too low; in animals, exon 51 skipping was detected in a nonlinear, dosedependent manner (that is, higher doses led to significantly more exon 51 skipping).
Specifically, with a 1-log increase in dose (from 5 to 40 mg/kg), there was little change in exon
51 skipping. With a second log increase in dose (from 40 to 320 mg/kg), however, there was
more than a log increase in response. These dose-dependent responses are important because it
is wholly conceivable that higher doses would lead to a much greater amount of dystrophin
production, which could be important for clinical benefit. Because the drug appears to be safe,
the review team recommended evaluation of much higher doses of eteplirsen, of at least
200mg/kg per week. Approving a drug at a dose that does not show a meaningful increase in
dystrophin (when the drug could theoretically achieve one at higher doses) is concerning.
As for accelerated approval, the regulatory standard at issue requires a sponsor to show that the
drug under review leads to an effect on the surrogate endpoint (in this case, the production of
dystrophin) and that the effect is reasonably likely to predict clinical benefit (in this case,
improving, or slowing down decline in, muscle function). The term “reasonably likely to
predict” acknowledges the potential for doubt in the outcome of interest. Indeed, nobody knows
the minimum level of dystrophin that is likely to confer clinical benefit in patients with DMD.
The critical scientific and regulatory issue at stake in CDER’s decision here is whether such
minute amounts of dystrophin are reasonably likely to predict clinical benefit at the dosage of the
drug subject to approval. In this case, both Drs. Woodcock and Unger have attempted to provide
a rationale, based on scientific and professional judgment, for whether or not such small levels of
dystrophin are reasonably likely to predict the clinical effect of interest. By any meaningful
objective standard, however, the overall evidence derived from eteplirsen’s limited clinical
development program does not support that the levels of dystrophin produced by eteplirsen at the
doses studied are reasonably likely to provide clinical benefit. As pointed out in Dr. Unger’s
appeal, “Study 201 did not show a treatment effect on its 1° clinical endpoint, change in 6minute walk distance at Week 24. Study 202 failed on the same endpoint at 48 weeks. The
course of these Study 201/202 patients, having received eteplirsen for some 3.5 years, was not
distinguishable from external control patients.” 146
145
146

Eteplirsen targets a subset of patients with DMD who are amenable to exon 51 skipping.
Appeal at 16.
25

 Some may argue that it would be reasonable to proceed with accelerated approval based on
eteplirsen’s safety profile, even where there are significant doubts about the drug’s effectiveness.
That argument does not take into account the risks of treatment with indwelling catheters to
maintain vascular access in young patients, who would otherwise not need one and who often
receive adjunct chronic corticosteroids, or, even more importantly, the detrimental impact on
their quality of life.
I would be remiss if I did not note that the sponsor has exhibited serious irresponsibility by
playing a role in publishing and promoting selective data during the development of this product.
Not only was there a misleading published article with respect to the results of Study 201/202 147
–which has never been retracted—but Sarepta also issued a press release relying on the
misleading article and its findings. As determined by the review team, and as acknowledged by
Dr. Woodcock, the article’s scientific findings—with respect to the demonstrated effect of
eteplirsen on both surrogate and clinical endpoints—do not withstand proper and objective
analyses of the data. Sarepta’s misleading communications led to unrealistic expectations and
hope for DMD patients and their families. It is very disappointing that the findings did not hold
up to careful review.
FDA must remain steadfast in its commitment to alleviating pain and suffering, approach the
most challenging problems with absolute determination, and apply maximum flexibility to
facilitate the development and availability of effective treatments. The agency’s value centers on
its ability to do all of the above while maintaining objectivity, even in the face of political
pressure. FDA should never mislead patients by granting even accelerated approval to products
that are not shown to offer the prospect of meaningful benefit to patients under the appropriate
regulatory and scientific standard.
I acknowledge that there are currently no specific drugs available to treat patients with DMD and
that issuance of a complete response letter would cast uncertainty on whether eteplirsen would
continue to be developed, based on business and financial decisions that are external to FDA.
However, approving products based on hope, on subjective clinical judgment, or on theoretical
constructs that are not anchored in data leads to irreparable damage to patients. Approval at this
time could deter others from pursuing the development of truly effective treatments, both for
DMD and other serious, life-threatening conditions. Granting accelerated approval here on the
basis of the data submitted could make matters worse for patients with no existing meaningful
therapies—both by discouraging others from developing effective therapies for DMD and by
encouraging other developers to seek approval for serious conditions before they have invested
the time and research necessary to establish whether a product is likely to confer clinical benefit.
I remain deep in my conviction that, through science and a flexible, sound regulatory approach,
good therapies will emerge to provide meaningful clinical benefit to patients with DMD and
other rare serious diseases.
THE SDR BOARD’S ADDITIONAL RECOMMENDATION
Although the SDR Board acknowledges that the scope of our review, as prescribed by the SDRSMG, is limited to procedural questions, we nonetheless feel duty-bound to make one additional
recommendation. As noted above, Dr. Unger seeks from the Office of the Commissioner a
substantive, scientific review of Dr. Woodcock’s decision to grant accelerated approval to
147

See Mendell JR, et al. Ann Neurol 2013;74:637-47.
26

 eteplirsen. The SDR-SMG presumes that an initiator such as Dr. Unger has received some
substantive review of the scientific dispute at issue as part of a formal appeals process in the
Center. Dr. Unger has never received any such formal review of his scientific arguments or the
underlying evidence. To the extent he has ever received any substantive review of his scientific
disagreement with Dr. Woodcock, Dr. Woodcock herself was the one who conducted that review
and resolved the conflict in her own favor. Neither the SDR-SMG nor CDER’s SDR-SOPs
contemplate a scientific disagreement that arises between a Center Director and another manager
in that same Center—partly because no one has ever anticipated the unique circumstance of this
case. Especially given the SDR Board’s concerns regarding the decision-making process at
CDER, we think additional review within the Office of the Commissioner is appropriate.
The SDR Board encourages you to conduct a thorough substantive review of the scientific
dispute in this matter or, in the alternative, to convene a panel of relevant experts to conduct such
a review and provide advice to the agency and you, as Commissioner, on whether the evidence
of the effect of eteplirsen on the surrogate endpoint is reasonably likely to predict clinical
benefit. If you choose the latter, in light of the public and political pressure evident during the
entire review process at CDER, as detailed in this recommendation, we believe that delegating
this critical evaluation to a panel of experts would help ensure that the agency makes the most
appropriate decision from the perspective of protecting patients and the public health, especially
for DMD patients. Knowing as we do that you value cross-Center collaboration with respect to
medical product development, we recommend that you include on the panel experts from other
Centers devoted to the regulation of medical products. Doing so would not only help ensure
diverse expertise on the panel but also provide insights on the effects that any proposed
regulatory decision on eteplirsen might have on products regulated by those other Centers. We
further recommend that you consider whether to include experts from other components within
the Department of Health and Human Services and whether, consistent with applicable laws and
the appropriate timeframe for a decision, you should also include outside experts on the panel.

27

 Agency Scientific Dispute – Appeal
Date:

July 18, 2016

To:

G. Matthew Warren
Director
Office of Scientific Integrity, FDA

From:

Ellis F. Unger, M.D. (initiator)
Director
Office of Drug Evaluation-I
Office of New Drugs
Center for Drug Research and Evaluation
U.S. Food and Drug Administration

Re:

NDA #
206488
Drug:
eteplirsen (Exondys 51)
Applicant:
Sarepta Therapeutics
Indication: Treatment of Duchenne muscular dystrophy (DMD) in patients
who have a confirmed mutation of the DMD gene that is amenable to exon 51
skipping

1.

Background

The Office of New Drugs within the Center for Drug Evaluation and Research (CDER)
oversees regulation of new drugs, and is responsible for making regulatory decisions for
approval/non-approval of new molecular entities. Within the Office of New Drugs, there are 6
sub-offices, including the Office of Drug Evaluation-I. The Office of Drug Evaluation-I oversees
the Division of Neurology Products, which regulates drugs for the central and peripheral
nervous systems, as well as drugs for muscular disorders. Typically, a new drug application
(NDA) for a new molecular entity for a neurology indication is reviewed by the Division of
Neurology Products in concert with review staff from other offices in CDER. 1 The regulatory
decision is typically rendered by Office of Drug Evaluation-I, i.e., the signatory authority.
NDA 206488 for eteplirsen was reviewed by the Division of Neurology Products, and members
of the review team reached the unanimous conclusion that the NDA should receive a complete
response action. This view was shared by the Office of Biometrics, which performed the
statistical review, as well as the Office of Clinical Pharmacology, which performed the
pharmacology review. Dr. John Jenkins, Director, Office of New Drugs, also supports a
complete response action for this NDA (verbal communication).
This memo is meant to explain the salient arguments around the scientific disagreement here;
additional details are available in my memo recommending a complete response and Dr.
Woodcock’s memo recommending approval, and the reader is referred to those memoranda.
Disease Background:
1

Reviews are typically provided by Office of New Drug Quality Assessment, Division of Medication Error
Prevention and Analysis, Office of Biometrics, Office of Scientific Investigations, and others.

Scientific Appeal – NDA 206488 – Page 1 of 27

 Duchenne muscular dystrophy is an X-linked recessive neuromuscular disorder caused by
mutations of the dystrophin gene. These mutations disrupt the messenger ribonucleic acid
(mRNA) reading frame, leading to the absence or near-absence of dystrophin protein in muscle
cells. The disorder affects 1 in ~3,600 boys.
Dystrophin protein is thought to maintain the structural integrity of the muscle cell, cushioning it
from the stress and strain of repeated contraction and relaxation. Absence of dystrophin leads
to muscle damage, with replacement by fat and collagen. With progressive degeneration of
skeletal muscle (including breathing muscles) and cardiac muscle, there is loss of physical
function in childhood and adolescence, with premature death from respiratory and/or cardiac
failure in the second to fourth decade.
No specific therapies are approved for DMD. Steroids are currently the cornerstone of
management, widely believed to delay loss of ambulation and respiratory decline by several
years.
Drug Background:
Eteplirsen is a phosphorodiamidate morpholino oligomer (PMO) designed to target the premRNA transcripts of the dystrophin gene so that exon 51 is excluded, or skipped, from the
mature, spliced mRNA. Theoretically, by restoring of the mRNA reading frame, a ‘truncated’
but nevertheless partially functional form of the dystrophin protein can be produced by muscle
cells, delaying disease progression. Similar truncated dystrophin is found in a less severe form
of muscular dystrophy, Becker Muscular Dystrophy (BMD). In essence, the drug is hoped to
induce production of sufficient Becker-type dystrophin to slow the progression of the disease.
This drug is specific for exon 51 mutations, a subset of the mutations that cause DMD. If
approved, the drug would be indicated for ~13% of the overall DMD patient population.
Eteplirsen has not received marketing authorization from any regulatory authority, and no
similar drugs are approved.
Drug Development Background:
Three studies are germane to the issues here. Study 201 was a single-center, double-blind,
placebo-controlled study in 12 patients with DMD. Patients were randomized (1:1:1) to
eteplirsen 30 mg/kg/week, eteplirsen 50 mg/kg/week, or placebo (4 patients per group). After
24 weeks, the 4 patients originally randomized to placebo were re-randomized to eteplirsen 30
mg/kg/week (n=2) or eteplirsen 50 mg/kg/week (n=2). The trial was eventually extended to an
open-label phase (Study 202) where all patients received eteplirsen, although investigators and
patients remained blinded to dose. These patients have continued to receive eteplirsen for
more than 4 years. This continuous study is referred to as Study 201/202. Study 301 is an
externally controlled study where all patients are receiving open-label eteplirsen, 30 mg/kg, by
weekly infusion. The study is ongoing and still accruing patients. Interim data were obtained
from 13 patients in this study (see below).
The endpoints for these studies can be broadly divided into those that aim to show changes in
physical performance, e.g., walking speed, rise time from the floor, muscle function; and those
that aim to show effects on production of dystrophin in skeletal muscle – the surrogate
endpoint. Dystrophin was quantified in this development program using two methods: Western
blot and immunohistochemistry.

Scientific Appeal – NDA 206488 – Page 2 of 27

 2.

Description of How My Position Differs from the Center’s Perspective

Dr. Janet Woodcock, Director, CDER, disagrees with some of the findings of the review team,
and has reached the conclusion that the NDA should be approved. She finds that the data
meet the standard for accelerated approval under 21 CFR 314. 510, based on the change in a
surrogate endpoint of dystrophin protein production – a change she concludes is reasonably
likely to predict clinical benefit. The disagreement is over the question of whether the findings
on the dystrophin surrogate endpoint are reasonably likely to predict clinical benefit. The
decision of approval vs. complete response hinges on this question.
a.

Clinical/Statistical Efficacy

Accelerated Approval:
Dr. Woodcock has reached the conclusion that eteplirsen should receive accelerated approval
based on a small effect on the surrogate endpoint of dystrophin production.
The relevant statutory and regulatory framework (section 506(c) of the FD&C Act and 21 CFR
part 314, subpart H) states that a drug can receive accelerated approval if 3 factors are
satisfied:
1. If the drug treats a serious or life-threatening disease or condition,
2. if FDA takes into account the severity, rarity, or prevalence of the condition and the
availability or lack of alternative treatments, and
3. if the drug demonstrates an effect on a surrogate endpoint that is reasonably likely to
predict clinical benefit OR demonstrates an effect on an intermediate clinical endpoint
that can be measured earlier than irreversible morbidity or mortality (IMM) that is
reasonably likely to predict an effect on IMM or other clinical benefit. As noted in
section 506(c)(1)(B) of the FD&C Act, the evidence to support the concept “…that an
endpoint is reasonably likely to predict clinical benefit may include epidemiological,
pathophysiological, therapeutic, pharmacologic, or other evidence developed using
biomarkers, for example, or other scientific methods or tools.”
In terms of the prospect for accelerated approval for eteplirsen, DMD is clearly a serious,
severe, and rare condition with no approved treatments; therefore, factors 1 and 2, above, are
satisfied. There is no disagreement.
The critical issue is whether factor 3 is satisfied, and factor 3 can be subdivided into three
parts: 1) whether the surrogate endpoint is appropriate for the disease; 2) whether there is
substantial evidence of an effect on the surrogate endpoint; and 2) whether the effect
demonstrated meets the test of being “reasonably likely” to predict clinical benefit. Importantly,
there is no regulatory definition of “reasonably likely.”
For the first part of factor 3, whether the surrogate endpoint is appropriate for the disease, the
review team has agreed that the near-lack of dystrophin is the proximal cause of DMD, and
that the level of dystrophin in skeletal muscle is an appropriate surrogate endpoint that could
predict efficacy. There is no disagreement here.

Scientific Appeal – NDA 206488 – Page 3 of 27

 The second part of factor 3 is whether an effect has been demonstrated; the legal standard is
‘substantial evidence’ based on adequate and well-controlled clinical investigations. Typically,
such evidence would be two studies, both achieving a p-value < 0.05, but in some situations
FDA has the flexibility to interpret data from a single trial, or a single trial with supporting
evidence, as substantial evidence of effectiveness. 2 Dr. Woodcock believes that “…there is
evidence from adequate and well-controlled trials, and supportive evidence, that exposure to
eteplirsen increases dystrophin protein production in muscle cells.” I agree that there is
evidence from a single adequate and well controlled trial, Study 301, that eteplirsen induces
dystrophin production in muscle cells, but do not agree that there is reliable quantitative
evidence from the other trial, Study 201/202.
The third part of factor 3, the conclusion that the demonstrated effect is “reasonably likely” to
predict clinical benefit, is where there is disagreement.
A. Are the Data on Dystrophin Protein Production from One or More Adequate and WellControlled Studies?
Dr. Woodcock cites 3 lines of evidence pertinent to the conclusion that eteplirsen increases
dystrophin production:
1. Production of an appropriate mRNA transcript
2. Quantitative assessment of dystrophin content in muscle biopsies by Western blot
3. Semi-quantitative assessment of dystrophin in muscle tissue by immunohistochemistry
1. Production of an appropriate mRNA transcript
I agree that the applicant has shown expression of mRNA following treatment with eteplirsen.
As noted by Dr. Woodcock, this finding establishes proof of concept, but does not by itself
mean that there is increased dystrophin production.
2. Quantitative assessment of dystrophin content in muscle biopsies by Western blot
Western blot is a standard laboratory technique used to quantify proteins in body tissues. In
Sarepta’s development program, Western blot was used to assess dystrophin protein levels in
skeletal muscle in Study 201, in Study 202 (again, these were Study 201 patients who were
maintained on treatment), and finally in Study 301.
a.

Study 201:

The original Western blot analyses from Study 201 were intended to show that dystrophin
levels were greater in eteplirsen-treated patients than in patients in the placebo group, and
analyses were planned to compare the effects of the lower vs. higher eteplirsen doses on
dystrophin production. The Western blots submitted by the applicant for Study 201 were
oversaturated, unreliable, and uninterpretable.
b.

Study 202:

2

See: “Guidance for Industry Providing Clinical Evidence of Effectiveness for Human Drug and Biological
Products;” May, 1998.

Scientific Appeal – NDA 206488 – Page 4 of 27

 With FDA’s assistance, the applicant improved the assays and performed repeat biopsies on
11 of 12 patients of the Study 201/202 patients at Week 180. These were to be compared to
stored baseline (pre-treatment) samples; however, evaluable tissue was available for only 3 of
the 11 patients. The baseline samples are germane to the determination of the treatment
effect because the Week 180 biopsies showed only a small quantity of dystrophin (mean =
0.93% of normal). Thus, for the purpose of computing the change in dystrophin resulting from
eteplirsen treatment, even small differences in the baseline level are critical.
As noted by Dr. Woodcock, the review team and I had concerns about these controls, leading
us to conclude that Study 201/202 was not adequate and well controlled:
1. The goal was to assess the change in dystrophin with treatment, i.e., pre-treatment vs. posttreatment, but most of the baseline biopsies were obtained from subjects external to Study
201/202, who could differ in unknown ways from subjects in Study 201/202.
2. For all patients, the Week 180 biopsies were obtained from different muscles than the
baseline biopsies, and studies of both normal human muscle and non-clinical DMD models
have shown that dystrophin levels vary among muscles.
3. The baseline biopsies for the three subjects with Week 180 data had been stored for several
years and the protein may have degraded, leading to a falsely low baseline value, and a
greater apparent increase from baseline, accordingly.
Dr. Woodcock believes that “…these issues increase the uncertainty around the results, but do
not necessarily render them an inadequate basis on which to draw a conclusion.” She notes
that the external control patients were similar in age and mutation site to the patients in Study
201/202. She found little difference between dystrophin results across different muscle groups,
and little difference based on storage time, leading her to believe that these factors “…did not
result in large differences in the findings.”
Although I agree that these factors are not likely to lead to large differences, even small
differences would affect the calculation of the change in dystrophin at Week 180, because the
Week 180 values were quite small (mean only 0.93% of normal). At issue is how much of the
dystrophin detected at Week 180 was newly produced, vs present at baseline. For example, a
difference in the baseline level of only 0.30%, although minute, is substantial compared to
0.93%.
Dr. Woodcock notes that at Week 180, 2 subjects had dystrophin levels between 2 and 3%, 2
had a level between 1 and 2%, and 2 had a level of ~1%. She notes that 2 of these subjects
had both baseline and Week 180 samples, and there were clear increases in dystrophin in
these 2 patients. Of note, Dr. Woodcock points out that although some subjects had Week 180
dystrophin levels similar to the baseline (i.e., close to zero), she would expect this because she
would not predict that all individuals would to respond to a drug intervention.
She explains that the issue “…is whether the dystrophin levels found at 180 weeks were within
the variability expected for this assay in such patients and, thus, could have arisen by chance,
or whether they could have been caused by differences from the controls or from sample

Scientific Appeal – NDA 206488 – Page 5 of 27

 storage as outlined above, or whether they reflected a drug effect, and, thus, whether these
data could be seen as adequate and well-controlled.”
In the end, taking Dr. Woodcock’s arguments into consideration, my view is that the data from
Study 202 are suggestive of an increase in dystrophin in response to eteplirsen, but the study
was not adequate and well controlled. If we accept that there is a difference, Study 202 does
not reliably speak to the amount of dystrophin produced by eteplirsen, given the concerns
above. There is only certainty that the largest possible amount was 0.93% of normal (on
average), and <3% in any individual (if we assume that the quantity was zero at baseline).
Below I will present another concern that leads me to question the veracity of the Western blot
data from the Week 180 biopsies from Study 202, based on an issue that Dr. Woodcock did not
address in her memo.
c)

Study 301:

With the May 26, 2016 goal date approaching, OND and CDER could not reach agreement on
the regulatory action for this NDA. In order to gain additional information that might provide
evidence of an effect on a surrogate marker that was reasonably likely to predict clinical
benefit, we requested that the applicant perform Western blot analyses of skeletal muscle
biopsy samples that had been obtained in an ongoing study (Study 301, PROMOVI). These
samples were originally planned to be analyzed at the end of the study; however, we requested
an interim analyses of a subset of samples. Western blot analyses were performed on paired
biceps samples from 13 of the patients. For each of these patients, samples obtained at
baseline (prior to treatment) were compared to those obtained at Week 48, after 48 weekly
infusions of eteplirsen 30 mg/kg.
The data are shown in Table 1 and the distribution of these changes is shown graphically in
Figure 1. Of these 12 patients, 8 (two-thirds) had a change of 0.25% or less; only 1 patient
(8%) had a change greater than 1%. The applicant used 3 methods to consider the numerous
values below the limit of quantification, but irrespective of the method used, the mean
treatment effect was similar, ranging from 0.22% to 0.32% of normal, a change of
approximately 2 to 3 parts per thousand that was nevertheless statistically significant.

Scientific Appeal – NDA 206488 – Page 6 of 27

 Table 1: Study 301: Pre- and Post-treatment Values of Becker-Type Dystrophin
Patient

Time
Baseline

1
Week 48
Baseline
2
Week 48
Baseline
3
Week 48
Baseline
4
Week 48
Baseline
5
Week 48
Baseline
6
Week 48
Baseline
7
Week 48

status

value (%)

pass
pass
pass
pass
pass
fail
pass
fail
pass
pass
pass
pass
pass
fail
pass
fail
fail
pass
fail
pass
pass
fail
pass
fail
fail
pass
fail
pass

0.15
0.11
0.22
0.29
0.35
0.26
0.36
0.12
0.06
0.06
0.5
0.24
0.04
0.06
0.1
0.19
0.1
0.17
0.92
1.02
0.37
0.46
0.3
0.29
0.04
0.17
0.22
0.42

mean (%) delta (%)

Patient

0.13

Time
Baseline

0.13

8

0.26

Week 48
Baseline

0.35
0.01

9

0.36

Week 48

0.06

Baseline
0.31

10

0.37

Week 48

0.04

Baseline
0.06

11
Week 48

0.1
0.17

Baseline
0.85

12

1.02

Week 48

0.37

Baseline
-0.07

0.3

13
Week 48

status

value (%)

fail
fail
fail
fail
fail
pass
fail
pass
pass
fail
pass
fail
pass
pass
pass
pass
pass
fail
pass
fail
fail
pass
fail
pass

0.08
0.14
0.08
0.05
0.14
0.24
1.17
1.57
0.11
0.05
0.12
0.11
0.01
0.08
0.31
0.63
0.02
0
0.09
0.01
0.34
0.18
0.34
0.21

mean (%) delta (%)

0.24
1.33
1.57
0.11
0.01
0.12
0.05
0.43
0.47
0.02
0.07
0.09
0.18
0.03
0.21

0.17
0.25
0.42

All parties agree that these data were obtained from an adequate and well controlled study,
and that there is a statistically significant effect of eteplirsen. The disagreement is whether or
not the dystrophin production is at a meaningful level that is reasonably likely to predict clinical
benefit.
To the extent that one can compare results across studies, these changes in dystrophin are
even lower than the values obtained from Study 201/202 (the latter represent the quantity
detected at Week 180, not the treatment effect). Dr. Woodcock wrote that “Only 2 of 12
patients achieved a level over 1% of normal control.” Her characterization refers to the amount
of protein detected at Week 48, not the change in protein. In fact, only a single patient out of
12 had a treatment effect that exceeded 1%.

Scientific Appeal – NDA 206488 – Page 7 of 27

 Figure 1: Study 301: Distribution of Changes in Becker-type Dystrophin in 12 Patients

N=8

N=2

N=1

N=1

3. Semi-quantitative assessment of dystrophin in muscle tissue by immunohistochemistry
Study 201/202 – Data through Week 48
Dystrophin production was assessed in Study 201 using immunohistochemistry, a standard
laboratory procedure used primarily to localize proteins in tissue sections, but also used as a
semi-quantitative method to measure dystrophin levels. Muscle samples were analyzed at
baseline, and at Weeks 12, 24, and 48.
Dr. Woodcock notes “A finding of increased dystrophin was also seen in several IHC assays
performed by the applicant.” She explains that several baseline and other pre-Week 180
assays were performed (from Study 201/202), but the validity of the results was questioned at
the FDA inspection because of methodological issues, and so she does not consider these
data further.
I do not agree with Dr. Woodcock’s outright rejection of these data. In fact, FDA requested a
re-reading of the stored images by 3 masked pathologists under improved viewing conditions.
We did not request any changes in immunohistochemistry methods or techniques, other than a
different approach for selecting microscopic fields for image capture and analysis. Thus, we
stressed that their stored images could provide useful data if properly read. The re-read

Scientific Appeal – NDA 206488 – Page 8 of 27

 showed a nominally statistically significant increase in dystrophin in response to eteplirsen for
the low dose group, but not the high dose group. (The p-value is nominal because the type-I
error rate was not controlled for multiplicity.) Moreover, for the 4 patients who had received
placebo through Week 24 and then switched to eteplirsen, there was no increase in dystrophin
at Week 48.
Study 201/202 – Week 180 Data
The applicant performed immunostaining along with Western blot analyses from the skeletal
muscle biopsies obtained at Week 180.
Importantly, prior to performing these analyses, the applicant made changes to the
immunohistochemistry protocol with the intent of decreasing non-specific staining. Dr.
Woodcock details the technical factors in her memo. Their aim was to determine the treatment
effect for each patient, by comparing dystrophin levels at baseline and Week 180. Frozen
archived baseline tissue was available for only 3 of the patients, however, and so the applicant
supplemented these samples with muscle tissue from 6 untreated external DMD patients,
together to be compared to the Week 180 levels. Images were read by the same 3
pathologists, masked to treatment group.
Because external controls were used, the comparison of pre- vs. post-treatment values suffers
from the same problems described for the Western blot analyses (i.e., different patients,
different muscles, and possible loss of immunoreactive dystrophin with long-term storage).
These concerns notwithstanding, the applicant claimed a remarkable increase in dystrophin
immunostaining at Week 180: the 9 baseline samples (from 3 patients in Study 201/202 and 6
external controls) showed 1.1% ± 1.3% positive fibers (mean ± SD), whereas the Week 180
samples (from 11 patients in Study 201/202) showed 17.4% ± 10.0% positive fibers. I will note
that FDA made no attempt to inspect or oversee these analyses.
Given that the original analysis showed, at baseline, 13% positive fibers for patients in Study
201/202, it is important to understand why the results from a new immunostaining protocol
provided results of 1.1%, an order of magnitude lower.
As noted above, there were 3 patients in Study 201/202 with adequate archived tissue from
baseline, which permitted a new immunohistochemistry analysis and a comparison of results
between the old and new methods. Figure 2 shows how the two methods compare.
These are essentially replicate analyses of a single tissue sample using the two
immunohistochemistry methods. There is an inexplicable difference of more than an order of
magnitude between results of the old and new immunohistochemistry protocols. Such marked
differences raise concerns with respect to the validity of the applicant’s methods, and make
interpretation impossible.
The disparity also underscores the difficulty of comparing results of immunohistochemical
analyses for dystrophin across laboratories, or, for that matter, within the same laboratory.

Scientific Appeal – NDA 206488 – Page 9 of 27

 The integrity of the applicant’s
data is further called into
question by lack of agreement
between the
immunohistochemistry and
Western blot methods, i.e., a
lack of internal consistency.
The applicant claims to have
enhanced both the
immunohistochemistry
methods and the Western blot
methods in preparation for
processing the Week 180
biopsies. Following these
methodological improvements,
single tissue blocks were
subjected to both analyses –
analyses considered to be
complementary. Yet the lack
of concordance between these
two assessments of dystrophin
levels is striking (Figure 3).
It is simply not possible to
determine whether the
immunohistochemistry
methods are inaccurate, the
Western blot methods are
inaccurate, or both methods
are inaccurate. In light of the
discordance between methods,
the issues with the control
samples, and the order-ofmagnitude discrepancy
between the old and new
immunohistochemistry
protocols, these data provide
little confidence that the study
was designed well enough so
as to be able “to distinguish the
effect of a drug from other
influences, such as
spontaneous change…,
placebo effect, or biased
observation” (§314.126).

Figure 2: Comparison of Results from the New and Old
Immunohistochemistry Protocols – Lack of Agreement for 3
Patients in Study 201/202

Figure 3: Study 201/202 Week 180 Dystrophin Assessment –
Lack of Agreement between Immunohistochemistry and
Western Blot

A critical point is that results of immunohistochemistry analyses are method-dependent, and
results from different laboratories are not directly comparable. Here we see a striking
difference between results of different methods within a single laboratory.

Scientific Appeal – NDA 206488 – Page 10 of 27

 Dr. Woodcock concluded “Although the IHC assays provide only semi-quantitative
assessments of dystrophin content, they do support an effect of eteplirsen on the proposed
surrogate endpoint (an increase of dystrophin production as a result of drug exposure).”
Although this statement does not constitute an important part of her argument in favor of
dystrophin production, I do not agree that the immunohistochemistry data show an increase in
dystrophin as a result of drug exposure. Given that changes in the immunohistochemistry
protocol led to remarkably disparate results, and in light of the lack of correlation between
dystrophin results as determined by immunohistochemistry and Western blot, I question the
accuracy and interpretability of the Week 180 immunohistochemistry data. Moreover, the
results from the properly blinded re-reading of the original data through the first 48 weeks of
Study 201/202 are negative. I do agree, however, that the immunohistochemistry images
appear to show dystrophin in the proper location, which helps support proof-of-concept.
In summary, I agree that there are data on dystrophin production from one adequate and well
controlled study, Study 301, by Western blot. The amount of dystrophin produced and the
likelihood of a clinical effect are discussed below.
B. Is the Effect on the Surrogate Endpoint “Reasonably Likely to Predict Clinical
Benefit?”
As noted by Dr. Woodcock, “The usual way to address this question would be to rigorously
evaluate what is known about the correlation between dystrophin levels in muscle and
expression of disease.”
Without restating the details of Dr. Woodcock’s discussion, I generally agree with her basic
summary of the many challenges of interpretation (quoted below). Most of her discussion
speaks to the uncertainties inherent in correlating dystrophin levels with disease severity. I
strongly agree that we lack a sound basis upon which to relate dystrophin levels observed in
this development program to observations in the literature.
“1. The clinical classification of disease severity (i.e., phenotype) in the literature appears
broad, variable, and somewhat subjective.”
I agree. And importantly, as Dr. Woodcock notes, “the zone of real interest for this
discussion, between DMD and intermediate presentations, is not rigorously
categorized.”
“2. Much of the prior data reporting the relationship of dystrophin protein levels to phenotype
have been from immunohistochemistry studies using a variety of techniques and antibodies.”
I will add that the applicant’s own data show a striking difference between results of two
somewhat different immunohistochemistry protocols conducted at the same laboratory
(Figure 2). Thus, it would be treacherous to try to relate various levels of dystrophin,
determined by immunohistochemical methods at various laboratories, to a particular
clinical course.

Scientific Appeal – NDA 206488 – Page 11 of 27

 “3. Both IHC analyses and WB results are influenced by the anti-dystrophin antibodies used, as
well as other experimental conditions”
Agree. Thus, is not feasible to relate levels of dystrophin determined by older Western
blot methods, which lacked, for example, appropriate internal controls, to levels of
dystrophin reported in these eteplirsen studies.
“4. The phenotype is significantly influenced by dystrophin isoform quality as well as dystrophin
quantity.”
Agree. It is difficult to predict a protein’s function from its structure; even small
changes in dystrophin structure can be important.
“5. The literature contains various findings on the relationship of dystrophin expression to
clinical status, including the low levels of dystrophin protein of interest in this case.”
Agree. There is little consensus on the relationship between dystrophin expression and
clinical course at the low levels observed in eteplirsen-treated patients.
I also agree with Dr. Woodcock on the following points, and I paraphrase here:
•

Dystrophin levels >10% on Western blot are usually associated with a BMD phenotype.
Within the BMD phenotype, the relation between disease severity and protein expression is
not clear. Protein quality, rather than quantity, may play a key role in determining
phenotype in BMD.

•

Patients with DMD are usually found to have undetectable levels of dystrophin, or very low
levels. Dr. Woodcock notes that she believes the conventional threshold of <10% protein
resulting in DMD was based on immunohistochemistry data. She tries to make a
conversion between values observed from immunohistochemistry (~10% points higher on
immunohistochemistry than Western blot in DMD) and those observed from Western blot,
but I caution that immunohistochemistry results, in particular, are highly method-dependent,
as noted above.

•

Rarely, dystrophin levels in the 3 to 10% range have been associated with Becker Muscular
Dystrophy phenotypes. Dr. Woodcock found no evidence of a threshold value for protein
content and expression of a DMD phenotype.

Despite the absence of reliable data, Dr. Woodcock concluded that evidence from Western blot
and other experiments shows that protein in the range between undetectable and 10% of
normal is likely to be very important for clinical presentation, all other things being equal, i.e.,
mutation status and non-dystrophin-related factors affecting phenotype.
Because of the lack of reliable evidence, I do not agree that the small increase in
dystrophin shown in Study 301 is ‘reasonably likely’ to predict clinical benefit. This is
the central issue in this appeal.
The “reasonably likely” question hinges on whether the protein is functional, and whether the
quantity is adequate.

Scientific Appeal – NDA 206488 – Page 12 of 27

 These two uncertainties, protein function and protein quantity, are separate issues that must be
considered in series. The function of the Becker-type dystrophin detected in Study 301 cannot
be assessed. Nevertheless, the review team has been willing to assume that whatever
Becker-type dystrophin is produced would function as well as it does in the Becker form of the
disease. Although there can be no certainty on this point, the question of function seems small
relative to the uncertainty regarding the adequacy of the quantity of protein, and so function is
less germane to the question of “reasonably likely.” In short, it is the quantity of Becker-type
dystrophin produced that is central to the question of ‘reasonably likely,’ and central to the
approvability of this NDA under accelerated approval.
At the outset, it must be stated that the minimum quantity of Becker-type dystrophin that is
reasonably likely to predict clinical benefit in patients with DMD is unknown.
There are two ways to consider the quantity of dystrophin produced: as a binary responder
analysis and as a mean response. The former has the advantage of considering the possibility
that some patients may respond to the treatment whereas others do not; the latter does not
allow for this type of consideration.
The problem with a responder analysis is that there are no data upon which to define a
threshold for a ‘response.’ Various cut-points could be selected, but their selection would be
arbitrary, and the particular threshold chosen would have a major influence on the effect size.
Here I provide 3 lines of reasoning to support my view that there is not an adequate basis to
believe that the small increase in dystrophin shown in Study 301 is reasonably likely to predict
clinical benefit: 1) the treatment effect observed cannot be compared or related to levels of
dystrophin measured by other laboratories and reported in various publications; 2) the effect
size is inadequate on its face; and 3) no evidence of a clinical effect was demonstrated in the
eteplirsen development program, and there is no correlation between dystrophin levels as
determined by Western blot and clinical outcome.
1) The treatment effect observed cannot be compared or related to levels of dystrophin
measured by other laboratories and reported in various publications.
In order to place these small quantities of Becker-type dystrophin into a clinical perspective,
many have considered publications from laboratories that attempt to relate particular levels of
Becker-type dystrophin protein to clinical course, e.g., maintenance of physical function, age at
loss of ambulation. Ideally, as suggested by Dr. Woodcock, there would be reliable data
showing that Becker-type dystrophin levels in excess of a particular level are associated with a
more benign clinical course.
Realistically however, the use of such a framework would be contingent on the ability to make
interpretable cross-laboratory comparisons of dystrophin levels, which would require
standardized methods to measure dystrophin levels in muscle specimens. Unfortunately, the
methods have differed greatly, and the methods in the literature have lacked critical internal
controls such as dilution-series. As stressed above, comparison of dystrophin values across
laboratories seems unreliable.

Scientific Appeal – NDA 206488 – Page 13 of 27

 With respect to immunohistochemistry analyses, Figure 2 provides ample basis for concern
regarding comparability of results using different methods. Results of separate
immunohistochemical analyses of skeletal muscle dystrophin, conducted by the same
laboratory on single blocks of tissue, differ by more than an order of magnitude. These results
underscore the inherent methodological variability of immunohistochemistry assays, and the
futility of attempting to compare dystrophin levels across assays/laboratories.
Even with respect to more recent Western blot methods, reproducibility across laboratories is
low. As discussed by Dr. Woodcock, Anthony K et al (Neurology 2014;83;2062) compared
results of Western blot analyses from 6 patients (3 with DMD; 3 with Becker Muscular
Dystrophy) across 5 experienced laboratories, and found a high degree of variability. Only one
of the 5 laboratories had a coefficient of variation (CV = SD/mean X 100) below 0.3%. The
authors found that variability was particularly pronounced with low levels of dystrophin –
precisely the area of interest here.
During the applicants’ presentation at the April 25, 2016 meeting of the Peripheral and Central
Nervous System Drugs Advisory Committee, Dr. Kaye, a pediatric neurologist and interim Chief
Executive Officer of Sarepta, could not have been more clear in warning us not to make
comparisons between their Western blot results and reported data in the literature:
“Our validated Western blot method, optimized to detect low levels of dystrophin, is
arguably the first dystrophin Western blot to be truly quantitative. This was achieved by
use of a 5 point calibration curve on each gel and prespecified loading and exposure
limits to avoid signal saturation. Furthermore, samples were randomized, blinded and
run in duplicate on separate gels. In contrast, the Western blot methods in the majority
of historical publications referenced by FDA were performed using older methodology
that is semi-quantitative at best. Given these significant methodological differences, it
is inappropriate to compare our data to literature approximations.” (Source: Official
transcript of the meeting; underlining for emphasis.)
In summary, the field has not achieved adequate standardization of methods for dystrophin
quantification at the very low levels observed in eteplirsen-treated patients; therefore, it is not
valid to compare an increase in Becker-type dystrophin of, at best, 2 to 3%, with dystrophin
values cited in the literature for other mutations/patient populations, assessed at other
laboratories. If the applicant’s results cannot be compared to results in historical publications,
then there is simply no way to determine whether the low dystrophin levels in eteplirsen-treated
patients are reasonably likely to predict clinical benefit.
2) The effect size is inadequate on its face.
If one were to assume that it is possible to make cross-laboratory comparisons of dystrophin
levels, the largest change reliably demonstrated in Study 301, 1.3%, is an order of magnitude
less than the minimum dystrophin levels cited to be important in affecting the course of patients
with Becker muscular dystrophy (at least 10%).
Some of the better data come from Van den Bergen et al, who studied the relation between
dystrophin levels (quantified by Western blot) and clinical severity in 33 patients with Becker
Muscular Dystrophy (J Neurol Neurosurg Psychiatry 2014; 85:747). Although the authors did
not find a linear relationship between dystrophin levels and disease severity, all 4 of their

Scientific Appeal – NDA 206488 – Page 14 of 27

 patients with dystrophin levels <10% showed poor muscle strength and early symptom onset.
As discussed by the review team, DMD experts have proposed that “induction of approximately
10% of normal dystrophin levels sets a minimum level to confer measurable clinical benefit.”
Initially, the applicant reported results from immunohistochemistry analyses purportedly
demonstrating that eteplirsen caused 50 to 60% positive staining of muscle fibers for
dystrophin. This seemingly unprecedented achievement aroused much excitement in the field
of DMD research and in the DMD patient community. Upon proper re-analysis, however, the
numbers were far lower, and rigorous statistical analyses showed that the changes weren’t
statistically significant. The Western blot analysis from Study 201/202 showed a mean
dystrophin level of only 0.93% (range 0 to 2.5%), but these values are of questionable
reliability. Finally, an adequate and well controlled study (Study 301) showed a mean change
of 3-tenths of a percent (range 0 to 1.3%). Given that dystrophin is a structural protein, it
seems highly unlikely that such changes would translate to a clinical effect.
Here are Dr. Woodcock’s assertions on this topic:
“The broad phenotypic distinctions made in the clinic (e.g., DMD vs IMD vs BMD) are
different from the prediction of benefit to an individual patient who has a specific
baseline dystrophin level and whose mutation and external factors do not change preand post-drug. For example, extending ambulation by six months to a year would not
normally move a patient from one to another of these categories, but could be very
important to quality of life (e.g., as suggested in the Bello study). This is also true for
other functional improvements.
For these reasons, incorporating the analysis of dystrophin content discussed above, I
conclude that the biochemical data strongly support the idea that low-level increases in
dystrophin production are reasonably likely to predict clinical benefit.”
I agree that broad phenotypic distinctions made in the clinic (e.g., Duchenne vs. Intermediate
vs Becker Muscular Dystrophy) are different than trying to predict benefit to an individual
patient on the basis of a particular change in dystrophin. And I agree that extending
ambulation by 6 months to a year (or similar improvements in other functional areas) would be
extraordinarily important.
But Dr. Woodcock never provides a rational argument – based on reliable data – to support the
concept that “…low-level increases in dystrophin production are reasonably likely to predict
clinical benefit.” She provides no rationale – no link between a mean increase in dystrophin of
3 parts per thousand and clinical benefit.
3) No evidence of a clinical effect was demonstrated in the eteplirsen development program,
and there is no correlation between dystrophin levels as determined by Western blot and
clinical outcome.
Dr. Woodcock states:
“Additional support for “reasonably likely” comes from the long-term experience with the
drug. The sponsor’s comparison of the experience of the treated cohort to natural
history data does not reach the level of substantiation required for traditional approval

Scientific Appeal – NDA 206488 – Page 15 of 27

 based on the clinical data. However, it is highly suggestive of improvement in some
parameters, in some patients, over natural history. My conclusion is informed by all the
caveats expressed in the reviews about the pitfalls of nonrandomized comparisons.
Given that the two exon 52 deletion patients in the study had fairly good long-term
results in terms of rate of disease progression, the question arises as to whether exon
52 is a prognostic factor that could have skewed the results.”
The review team analyzed the clinical data in great detail, and could not reach the conclusion
that there was any reliable evidence of improvement relative to the expected natural history of
the disease. Study 201 did not show a treatment effect on its 1° clinical endpoint, change in 6minute walk distance at Week 24. Study 202 failed on the same endpoint at 48 weeks. The
course of these Study 201/202 patients, having received eteplirsen for some 3.5 years, was not
distinguishable from external control patients (see my review memorandum for more details).
The Advisory Committee voted (7 to 3 with 3 abstentions) that the clinical results of Study
201/202 did not provide substantial evidence that eteplirsen is effective for the treatment of
DMD, and their vote was in the face of extraordinary pressure from patients and patient
advocates to vote for approval. Two of the 3 “yes” votes were from patient representatives.
Correlation between dystrophin production and clinical effect
A correlation between dystrophin production (or with less certainty – dystrophin detected) and
clinical function could provide some support for a conclusion that dystrophin production is
reasonably likely to predict clinical benefit.
The applicant collected data on both dystrophin production and physical performance in Study
201/202. On the basis of the data presented in the NDA, the Division concluded that no patient
in Study 201/202 clearly deviated from the natural history of the disease. The Division
reasoned, therefore, that whatever the quantity of Becker-type dystrophin detected, it did not
predict clinical benefit. Thus the Division opined that the clinical data weaken, and do not
strengthen, the “reasonably likely” argument.

Scientific Appeal – NDA 206488 – Page 16 of 27

 The Division’s view
notwithstanding, it is worth
considering patients on an
individual basis to assess the
correlation between the
quantity of Becker-type
dystrophin detected and
changes in physical
performance.

Figure 4: Study 201/202 – Lack of Correlation between
Quantity of Dystrophin Detected and Preservation of Physical
Function (6-Minute Walk Distance)

As noted by Dr. Woodcock,
the 6-minute walk test results
do not show a strong
correlation (Figure 4). For the
9 patients in Study 201/202
who remained ambulatory at
Week 180 and agreed to
undergo a fourth muscle
biopsy, the figure shows little
correlation between the
quantity of dystrophin detected
(x-axis) and preservation of
physical function as assessed
by the change in 6-minute
walk distance from baseline (yaxis) after weekly infusions of eteplirsen for 3 to 3.5 years. For the 5 patients whose 6-minute
walk performance was best preserved (red arrows), 2 had the highest dystrophin levels
detected in the study (upper right), but 3 had levels that were near-zero (upper left).
Dr. Woodcock also evaluated the North Star Ambulatory Assessment (NSAA) as a function of
dystrophin detected in boys who could still walk and who had a dystrophin result at Week 180.
She obtained the data from the applicant’s briefing document for the Advisory Committee
meeting, and found a correlation between dystrophin detected at Week 180 by Western blot
and rate of decline in NSAA score through 180 weeks. Her graph is reproduced below:

Scientific Appeal – NDA 206488 – Page 17 of 27

  

With respect to
the correlation,
Dr. Woodcock

Figure 5: Study 201/202: Analysis of Change in NSAA vs. Expression of
Becker-type by Western Blot (Analysis by Dr. Woodcock)

 

explained: ?This

 

adds additional
support to the

30;

 


U1

idea that




 

0

production is
?reasonably likely

 


U1

Charm-In NSAA

to predict clinical
benefit.?

9 Patient



 


0

Given that the

Linear (Patient)

 

U1

correlation was

 

 

 

 

 

driven by the
patient depicted
at the lower right
(blue arrowWestern Blot

 

 



2.5

J-

5 3

 

 

 

level 

 

change in NSAA 3), I considered the NSAA data from that patient (Figure 6). I found
that his course was less benign than would be inferred from a change in NSAA of only 3 units.

Speci?cally, using linear regression (red line in Fig
worsened by a mean of 2.7 units per year.

ure 6), his NSAA score has, instead,

I reasoned that inclusion of a_H of the NSAA data for each patient would provide a more reliable
representation of their course than calculating the change between single pre-treatment and

post-treatment data points, because of the test-
to-test variability short-term swings of 4 to
5 points for patient 006). Thus, using linear
regression, I calculated the slope of the
relationship between NSAA and time for each
patient (as per the red line in Figure 6) and
plotted the slopes as a function of the
detected at Week 180. (Slopes were
calculated as loss of NSAA units per year.)

Using this method, there was no correlation (R2
0.36), Figure 7. lmportantly, the slight trend
apparent here is driven by one or two data
points.

Scientific Appeal NDA 20

 

 

Figure 6: Patient 006: NSAA vs Time

 

25,

 

ml?
P'Lll 100 150 200 250

 

time (weeks)

 

 

6488 Page 18 of 27

Summary:
In summary, I find no evidence that the
increase in dystrophin demonstrated in
Study 301 is reasonably likely to predict
clinical benefit (mean 0.3%, range 0 to
1.3%). The levels of dystrophin linked to
various Becker Muscular Dystrophy
phenotypes in publications are largely not
comparable to dystrophin levels measured
in this development program. The
applicant’s interim CEO correctly urged us
not to compare data from their Western blot
analyses to historical approximations from
the literature. And extremely low levels of
dystrophin, as found here, seem particularly
difficult to quantify and compare across
laboratories. Nevertheless, to the degree
that findings can be compared across
studies, dystrophin levels of 10% of more
would need to be achieved to impact the
clinical course. The finding in Study 301 is
an order of magnitude below this level.

Figure 7: Study 201/202: Analysis of Change in
NSAA (Linear Regression) vs. Expression of
Becker-type Dystrophin by Western Blot (My
Analysis)

Based on protein levels in other deficiency diseases, the effect size here appears to be too
small to provide benefit. If dystrophin were an enzyme that catalyzed a biochemical reaction in
myocytes, one might posit that a very small quantity could produce a substantial proportion of
the minimum necessary reaction product, and that the increase over baseline might be
important because levels are so low in untreated patients. But given that dystrophin is a
structural support protein that helps prevent myocyte injury due to stress and strain, I find it
difficult to conceive how a treatment effect of 3 parts per thousand could confer clinical benefit.
If there were 10 inches of snow on a sidewalk that needed to be cleared, 3 parts per thousand
would amount to 1/32nd of an inch. We must also recognize that a treatment that increases
dystrophin by 0.3% would seemingly have far less impact than being born with 0.3% more
dystrophin, and even that seems unlikely to matter.
I can find no precedent of an accelerated approval for a marketing application where the effect
size on the surrogate endpoint is as small as 0.3%.
Dr. Woodcock concludes:
“…my conclusion to rely on the surrogate endpoint described above represents the
greatest flexibility possible for FDA while remaining within its statutory framework. In
this case, the flexibility is warranted because of several specific factors, including: the
life-threatening nature of the disease; the lack of available therapy; the fact that the
intended population is a small subset of an already rare disease; and the fact that this is
a fatal disease in children. Of note, the therapy has been relatively safe in the clinic,
although intravenous administration always carries risk.…Therefore, I find that the

Scientific Appeal – NDA 206488 – Page 19 of 27

 probable benefits outweigh the foreseeable risks and that this application should be
approved under 21 CFR 314.510.”
As noted in 506(f)(1), the amendments made by the Food and Drug Administration Safety and
Innovation Act (FDASIA) “…are intended to encourage the Secretary to utilize innovative and
flexible approaches to the assessment of products under accelerated approval for treatments
for patients with serious or life-threatening diseases or conditions and unmet medical needs.”
Some have interpreted this “flexibility” as a lower standard for demonstration of effectiveness,
but this is not true.
Section 506(f)(2) of the FD&C Act specifically notes that drugs granted accelerated approval
must meet the same statutory standards for safety and effectiveness as those granted
traditional approval, notably the substantial evidence standard of section 505(d) with respect to
the drug’s claimed effect on a surrogate or intermediate endpoint. These facts have not been
altered by FDASIA.
To be clear, 506(f)(2) states: “Nothing in this section shall be construed to alter the standards
of evidence under subsection (c) or (d) of section 505 (including the substantial evidence
standard in section 505(d)) of this Act or under section 351(a) of the Public Health Service Act.
Such sections and standards of evidence apply to the review and approval of products under
this section, including whether a product is safe and effective. Nothing in this section alters the
ability of the Secretary to rely on evidence that does not come from adequate and well
controlled investigations for the purpose of determining whether an endpoint is reasonably
likely to predict clinical benefit as described in subsection (b)(1)(B).”
I believe the burden is on Dr. Woodcock to show or explain why production of a near-zero
quantity of dystrophin (0.3%) is reasonably likely to predict clinical benefit, and I do not believe
her July 14, 2016 memo makes this case. I believe that the available evidence leaves open
the possibility that some patients could benefit from a small increase in dystrophin, but this
possibility does not reach the threshold of being reasonably likely to predict a clinical benefit.
Finally, there was no clinical benefit demonstrated in the development program, and the
correlation between dystrophin and clinical effect was poor – not surprising given that the
applicant provided analyzable data from only 11 patients.
3.
Assessment of Possible Impact to Public Health Should My Position Not be
Adopted
The approval of this NDA in its present form would have far reaching negative consequences
for the public health.
1. Eteplirsen’s risks are certain, whereas its efficacy is not. Having considered Dr.
Woodcock’s line of reasoning and her desire to approve eteplirsen, the position of the
review team in the Division of Neurology Products, the Office of Biometrics, the Office of
Clinical Pharmacology, the Office of Drug Evaluation-I, and the Office of New Drugs (verbal
acknowledgement from Dr. John Jenkins) is that the applicant has not provided evidence
that this drug is effective at the dose studied.

Scientific Appeal – NDA 206488 – Page 20 of 27

 Dr. Woodcock notes that “…the therapy has been relatively safe in the clinic.”
The reality is that only a few dozen patients have been exposed to the drug, such that the
safety profile is not well characterized. A closely related drug being studied under a
(b) (4)
With
additional experience, important toxicity may emerge for eteplirsen. It is known that many
patients in these studies are now receiving infusions through indwelling catheters.
Maintenance of vascular access in patients on chronic corticosteroids poses a certain risk
of infections. Although we are not yet aware of any infection-related adverse reactions,
there would definitely be serious infections and possibly deaths if this drug is marketed, yet
evidence of efficacy is lacking.
2. By allowing the marketing of an ineffective drug, essentially a scientifically elegant placebo,
thousands of patients and their families would be given false hope in exchange for hardship
and risk. I argue that this would be unethical and counterproductive. There could also be
significant and unjustified financial costs – if not to patients, to society.
The prospect of providing false hope to desperate patients from a promising but ineffective
therapy recalls the experience with transmyocardial laser revascularization (TMLR). In the
1990s, patients with coronary atherosclerosis and severe angina who were poor candidates
for conventional revascularization procedures (“no-option” patients) underwent a
thoracotomy (opening of the chest cavity) to enable use of a laser to create channels
through the heart muscle. Ostensibly, these channels provided conduits for blood to flow
from inside the left ventricle to the myocardium. Conduct of sham-controlled studies was
impossible; studies were essentially baseline-controlled or historically-controlled. Large
treatment effects were reported by a number of investigators, generally from small studies.
There were marked increases in treadmill exercise time and relief of angina, with effects
sustained for more than a year in some cases. Although many in the cardiology community
raised concerns about expectation bias and were highly skeptical of the results, to some
the effects seemed larger and more durable than could possibly be explained by
expectation bias, i.e., a placebo effect. Thousands of patients underwent this invasive
procedure with the hope of angina relief. Some years later, with improvements in
technology, the conduct of sham-controlled studies became feasible, and TMLR was not
found to be effective. The false hope was ultimately dispelled with the publication of two
Cochrane Reviews. 3 These reviews found the appearance of a marked treatment effect,
but 30-day mortality was 6.8% in the TMLR group vs. 0.8% in the no-treatment group.
They noted “The assessment of subjective outcomes, such as improvement in angina, was
affected by a high risk of bias and this may explain the differences found.” In this case, the
cost of false hope was ~6% mortality in the first 30 days post-op.
I will also note that the primary endpoint of these laser studies was generally exercise
capacity – the same type of endpoint used in the eteplirsen DMD development program,
also for “no option” patients.
3. The accelerated approval pathway is designed to expedite the availability of promising new
therapies to patients with serious conditions, especially when there are no satisfactory
3

Cochrane Database of Systematic Reviews 2015, Issue 2. Art. No.: CD003712. DOI:
10.1002/14651858.CD003712.pub3

Scientific Appeal – NDA 206488 – Page 21 of 27

 alternative therapies, while preserving standards for safety and effectiveness. For drugs
granted accelerated approval, postmarketing confirmatory trials are required to verify and
describe the anticipated clinical benefit, and FDA may withdraw approval of a drug if a trial
required for verification of the predicted clinical benefit fails.
In reality, it is difficult to withdrawal a drug that is deemed to be effective, or possibly
effective, by patients with severe diseases and limited treatment options. FDA has not
succeeded in withdrawing the marketing of a single drug for lack of verification of clinical
benefit following accelerated approval. The reality is that if eteplirsen is given accelerated
approval, it is highly likely to remain on the market indefinitely, irrespective of whether or
not efficacy is verified.
4. With the false perception that eteplirsen is effective, patients who are gaining benefit from
steroids but experiencing untoward side effects might be inclined to taper or stop them,
which could lead to more rapid disease progression.
5. False scientific conclusions have the potential to mislead the field of medicine, slowing
progress in finding and developing therapies that actually are effective. For example,
consider the scenario of a related drug with far greater potential to promote dystrophin
production in patients with DMD. In order for a sponsor to study such a drug, patients
would likely have to agree to discontinue eteplirsen, and few patients may be willing to do
so. In short, approval of an ineffective therapy has the potential to discourage or inhibit the
development of other drugs that are effective, and this impact can be significant.
6. Accelerated approval would lower the evidentiary standard for effectiveness to an
unprecedented nadir. The amount of dystrophin produced in Study 301 is so meager that it
could be considered to be tantamount to any increase in dystrophin. In other words, if a
statistically significant change of 0.3% – a mere 3 parts out of a thousand – is considered
adequate to support accelerated approval here, then the question arises as to whether
there would be any statistically significant change that would be too small to be considered
“reasonably likely” to support accelerated approval. Similarly, if a ‘responder’ had been
defined as a patient with an increase in dystrophin of ≥1% (and there is no basis to accept
such a low threshold), there would have been only a single responder in Study 301. If we
were to adopt the concept that, for rare diseases, accelerated approval could be supported
by any statistically significant change in an appropriate surrogate, or a response in a single
patient, we would enable accelerated approval of a myriad of drugs for rare diseases. No
doubt there are some who would applaud this as an advance. But a standard this low
would undercut FDA’s ability to ensure that drugs that are approved are effective; it would
call into question much of what we do. Lowering the bar to this level would be tantamount
to rolling back the 1962 Kefauver-Harris Drug Amendments to the Federal Food, Drug and
Cosmetic (FD&C) Act, which have served Americans well for some 54 years.
7. With accelerated approval of this NDA, there would be highly detrimental effects on drug
development. Traditional drug development for rare diseases might be replaced by a
system where small, baseline-controlled, proof-of-concept studies designed to show any
change in a surrogate marker would provide a basis for accelerated approval, assuming
that the pathogenesis of the disease was well understood and that the surrogate was
directly on the causal path. There would be little reason to pursue adequately controlled
clinical trials to support efficacy prior to accelerated approval; in fact, the possibility of

Scientific Appeal – NDA 206488 – Page 22 of 27

 failure would provide a disincentive to conduct such trials. For example, a gene therapy
designed to produce a missing clotting factor could receive accelerated approval on the
basis of a tiny yet inconsequential change in levels of the factor, or a more robust response
in a single patient. In short, the precedent set here could lead to the approval of drugs for
rare diseases without substantial evidence of effectiveness.
8. Even if the 30 mg/kg/week dose
were considered to have a
meaningful effect on the surrogate
endpoint, we already know this dose
is sub-therapeutic. We know this
because patients who have been
receiving this eteplirsen dose for
some 3.5 years have been
progressing at a rate that is similar to
that expected, based on the natural
history of the disease (Figure 8). I
question the ethics of approving or
prescribing a drug for a fatal disease
at a dose that is very likely to be
sub-therapeutic, when the
consequence of a sub-therapeutic
dose is clinical deterioration and
death. The figure shows the
unremitting progression in the
patients in Study 201/202, based on
changes in NSAA.

Figure 8: Study 201/202 – Individual NSAA
Performance by Age

9. Approval of this NDA would send the signal that political pressure and even intimidation –
not science – guides FDA decisions, with extremely negative consequences (See Grainger
D., 11/30/15. “DMD Drugs: an existential threat to FDA,” Forbes 4). The public is well aware
of this development program: the meager size of the study population, the marginal (at
best) effect size, the Division’s dim view of the efficacy data, and the robust activism of
some members of the DMD community. Many would be amazed at an approval action,
because other DMD drugs, recently turned down for approval, appeared to provide stronger
evidence of efficacy.
FDA and Congress were bombarded with correspondence – pleas urging approval of this
NDA. More than 50 speakers registered to speak at the April Advisory Committee meeting.
I received 2,792 emails urging approval. Here is an example of the body of an email I
received last week:
“Dear Dr. califf: How is it that everyone in and around DMD understands this simple
Idea and the science geniuses at FDA don't? You stupid f_ _ _ers are costing each and
every DMD kids days of their lives with your Moronic Dystrophin dance. Time to get a

4

downloaded 7/18/16 at http://www.forbes.com/sites/davidgrainger/2015/11/30/dmd-drugs-an-existential-threatto-the-fda/#5ffc712455f7

Scientific Appeal – NDA 206488 – Page 23 of 27

 f_ _ing clue

(6)

The ramifications here are profound. The public will perceive that it was their
unprecedented lobbying efforts that made the difference and earned eteplirsen its
accelerated approval. For the future, this will have the effect of strongly encouraging public
activism and intimidation as a substitute for data, which is one of the worse possible
consequences for communities with rare diseases. This type of activism is not what was
envisioned for patient-focused drug development.

4. Detailed Description of the History of the Dispute, Including My Description of the
Center SDR Procedures Followed and/or Not Followed, Dates of Meetings, and
Decisions Rendered Throughout the Process

The following table shows the dates and main activities for 15 Center Director Briefings
associated with the development of this drug: 8 Center Director Brie?ngs took place during the
IND phase of development, prior to submission of the NDA, and 7 Center Director Brie?ngs
took place during review of this NDA.

 

DATE

I MEETING

I DETAILS

 

 

7/17/2013

Center Director
Bne?ng

Follow up on Action Item from 3/13/13 EOP2 Meeting: Sarepta
has submitted a comprehensive discussion of the issues from
the EOP2 mtg. To discuss the suitability to ?le the NDA for
Subpart approval.

 

10/18/2013

Center Director
B?e?ng

Dr. Unger presented an overview and Dr. Farkas had a slide
presentation on drisapersen and eteplirsen data.

Discussion:

1. Plan to have a manufacturing facility visit by ONDQA - to
observe process and obtain yield calculation. Sponsor is
expecting to have 2nd batch in Dec 2013. Determine how much
product the sponsor has.

2. OBP: recommended to establish specificity of the antibody
and variability of the assay.

3. Next trial - plan to have OSI group to observe the conduct.

4. Need data from the trial. DNP has
previously requested the data from but did
not get any response. Dr. Woodcock will initiate an inquiry to the
sponsor (raw data).

5. The Agency needs to assist Sarepta (characterize biomarker,
CMC facility, observe 6MWT, etc.)

6. 2nd Internal Meeting (Drs. Woodcock, Temple, Jenkins,
Unger and Neuro) before the 11/8/13 sponsor meeting. Discuss
further what to convey to Sarepta.

 

 

10/28/2013

Center Director
B?e?ng

(continuation of
10/18/13 meeting)

 

Suggestions/Recommendations for DNP to Consider:

-- We have concluded that we will not ask for biopsy until (we
understand the histopathology and are) we?re certain what is a
quantitative measure and identi?ed the surrogate marker for the
study.

-- Tell the sponsor that we have changed our view for the
quantitative measure of truncated as a surrogate PD
marker used in their study, because of the recent natural history

 

 

Scientific Appeal - NDA 206488 Page 24 of 27

 

1/17/2014
2/6/2014

Center Director
Briefing
Center Director
Briefing

3/5/2014

Center Director
Briefing

3/19/2014

Sponsor Meeting,
with Center Director
Center Director
Briefing

4/2/2014

6/26/2015
12/9/2015

Center Director
Briefing

1/13/2016

Center Director
Briefing

2/10/2016

Center Director
Briefing

4/15/2016

Center Director
Briefing

4/25/2016

Advisory Committee
Meeting
Center Director
Briefing

5/4/2016

(b) (4)
study and
.
-- Dr. Woodcock wants to have a comprehensive literature
review to fully understand what’s this mean of the deletions,
mutations, or duplications in the dystrophin gene, or this exon 51
of dystrophin mRNA ((Office of Translational Science)I believe
this task was assigned to a different group ).
-- To ask the Sponsor to provide their production schedule. I
believe Dr. Woodcock wants to understand the amount of
production and determine if the company can provide the drugs
to those DMD patients in the future.
-- To suggest that the Sponsor consider enrolling patients
younger in age (like starting with 5yrs) in their clinical study.
-- To ask the Sponsor if they could provide drugs for
compassionate use to patients (who are very sick or those were
in the drisapersen trial previously).
-- Schedule a T-con with (b) (4) to discuss (b) (4)
data
Request: Team to present DMD drugs study design to Dr.
Woodcock – Path forward for Sarepta (& (b) (4) )
DMD drugs study design (Discuss Sarepta path forward)
Action items: (a) Request biomarker data from the sponsor done TC on 2/7/14(b) If data interpretable, meet with sponsor for
a brainstorming session. Then follow-up with Advice Letter
Dr. Ash Rao presented biomarker data findings (including Drs.
Woodcock, Jenkins, Temple, Unger, Moscicki)
Team discussed path forward. Action Item: to invite Sarepta for
a brainstorming discussion.
brainstorming discussion - study design and path forward
Action: Sarepta to submit proposed studies and next steps
Drs. Woodcock, Moscicki, Temple, Unger
Discuss proposal & comments to sponsor
~Advice Letter-include previous meeting discussions
~FDA workshop – biomarker
~Work w/ sponsor on dystrophin biomarker
~Natural history raw data - primary investigators
SUBMISSION OF NDA
To brief on the current status of eteplirsen review in advance of
the planned Jan 22, 2016 AC meeting.
To discuss the application and the plan of action.
To review the slide presentation and plan of action for eteplirsen,
that will be presented during the Advisory Committee Meeting on
January 22, 2016 to senior leadership.
To discuss the ongoing review of the NDA, and what will be
presented during the Advisory Committee Meeting in April. To
discuss the strengths, limitations, and uncertainties of the data,
particularly with respect to the comparison between the openlabel eteplirsen group and a contemporary untreated external
control group.
To discuss the statistical review of the CINRG data. To discuss
the review of data on DMD that was conducted by the
Cooperative International Neuromuscular Research Group

Discuss the outcome and plan of actions for the application post
advisory committee meeting

Scientific Appeal – NDA 206488 – Page 25 of 27

 5/31/2016

Center Director
Briefing

7/6/2016

Center Director
Briefing

Discuss reviews conducted by the review team and leadership
along with any additional information obtained from the
sponsor. Discussed Dr. Woodcock’s memo. Timeline for
reviews due to Dr. Woodcock.
1. The levels of dystrophin observed in 12 DMD patients from
the recent interim analysis of an ongoing trial and whether the
levels seen can be interpreted to be “reasonably likely to predict
clinical benefit” and used as a surrogate endpoint to support
accelerated approval.
2. The design of one or more PMR trials to confirm clinical
benefit of eteplirsen if it is approved under accelerated approval.
3. Description of the available clinical data in the drug label if
approved.

Based on my years of experience in Office of Drug Evaluation-I, the Center Director’s direct
involvement with this drug, compared to other development programs, has been
unprecedented. She also attended the April meeting of the Peripheral and Central Nervous
System Drugs Advisory Committee, where she spoke and interjected a number of important
comments.
There is no question that there has been adequate time and place for the discussion of various
views. I will note, however, that I found it unfortunate that the Center Director made clear her
intent to approve the drug at a briefing with the review team on May 4, 2016, before she had
seen drafts of the Division’s final review memorandum or my review memorandum. Prior to
reading our reviews, Dr. Woodcock stated that she had already “…reached a different
conclusion….” than the review team.
5. Action, Decision or Remedy Sought
Although the above paragraph could be considered grounds for an appeal based on process, I
seek instead a scientific review on the matter of whether or not there is substantial evidence of
a quantitative effect on dystrophin protein that is reasonably likely to predict clinical benefit. I
maintain, along with the Division of Neurology Products, Office of Biometrics, Office of Clinical
Pharmacology, Office of New Drugs, and the majority of the members of the Peripheral and
Central Nervous System Drugs Advisory Committee, that substantial evidence is lacking to
support either a conventional or accelerated approval, and that a complete response should be
issued for this NDA.
The unprecedented finding of an increase in dystrophin protein in response to eteplirsen
establishes proof-of-concept and provides great promise that this drug, or other therapies, will
eventually be capable of ameliorating the fundamental genetic defect of DMD, but the effect
size here is insufficient at the tested doses.
6. Path Forward
Based on the quantity of Becker-type dystrophin produced in Study 301 and the clinical
findings in Study 201/202, additional studies at this dose are unlikely to support any type of
approval, i.e., the data obtained for eteplirsen at doses of 30 and 50 mg/kg/week are fairly
solid, but they do not support efficacy.

Scientific Appeal – NDA 206488 – Page 26 of 27

 I remain comfortable with the concept that substantial evidence of dystrophin production from
adequate and well controlled trials could support accelerated approval, but it is clear that higher
doses are needed, and greater quantities of dystrophin would need to be produced. The path
to a conventional approval would require a double-blind, placebo-controlled (or multi-dose)
study, at least one year in duration, using some measure of physical performance as the
primary endpoint, again, testing higher doses.
The applicant is continuing to enroll Study 301 (PROMOVI), an open-label, multicenter, 48week study in patients with DMD amenable to skipping exon 51. All patients are receiving
eteplirsen, 30 mg/kg/week as an IV infusion.
My suggestion for a path to approval is to randomize patients in the ongoing Study 301 to:
1) either remain on 30 mg/kg/week; or
2) have their dose significantly increased. This could be done through use of a higher
dose, through more frequent dosing intervals (with dummy infusions), or both. Given
that many patients receive eteplirsen through indwelling IV lines and no significant
infusion reactions have occurred, perhaps these infusions could be performed at home.
For example, the study could compare 30 mg/kg weekly to 30 mg/kg daily. Patients
who do not tolerate more frequent dosing could have their doses decreased, as
needed. Based on non-clinical findings, monitoring would need to be in place to assess
renal toxicity.
Patients and investigators would be blinded to treatment group. For accelerated approval, the
primary endpoint would be dystrophin production, comparing the higher and lower doses. For
standard approval, the primary endpoint would be a test(s) of physical performance such as
NSAA or rise time.
Such a trial would be methodologically sound and ethical. Virtually everyone, patients and
physicians alike, would want to know whether higher eteplirsen doses would increase
dystrophin production, and would have equipoise for participation. Although there is concern
regarding performance of muscle biopsies in patients randomized to placebo, this would not be
a concern here with all patients receiving active drug. And I would recommend that the
applicant forego immunohistochemistry studies in favor of Western blot analyses, such that
needle biopsies with local anesthesia would be sufficient (rather than open biopsies with more
intensive anesthesia and greater morbidity).
I also believe that it would be desirable for the company to provide access to eteplirsen for
DMD patients through expanded access programs, with cost recovery, while an adequate
dose-finding study is conducted.
FDA is charged with the responsibility of ensuring that drugs are shown to be effective prior to
marketing, based on substantial evidence. If we were to approve eteplirsen without substantial
evidence of effectiveness, or on the basis of a surrogate endpoint with a trivial treatment effect,
we would quickly find ourselves in the position of having to approve a myriad of ineffective
treatments for groups of desperate patients, in essence, allowing marketing based on
desperation, patient lobbying, and the desire and need of hope. If we were to turn the clock
back to the days prior to the 1962 Kefauver Harris Amendments to the Federal Food, Drug,
and Cosmetic Act, the damage to society and the field of evidentiary medicine would be
enormous.

Scientific Appeal – NDA 206488 – Page 27 of 27

 CENTER DIRECTOR DECISIONAL MEMO
NDA#

206488

Drug Name

EXONDYS 51 (eteplirsen)

Indication

Duchenne Muscular Dystrophy (DMD)

Sponsor

Sarepta

Author

Janet Woodcock, M.D.
Director, Center for Drug Evaluation and Research (CDER),
Food and Drug Administration

SUMMARY
This memorandum explains the CDER’s final decision on the above application. I have read the
reviews and recommendations by Drs. Unger (Office level), Bastings (Division level), Farkas (CrossDiscipline Team Lead), Breder and Rao (Clinical Reviewers), Ling (Statistical Reviewer), and
Bhattaram, Wu, and Rogers (Clinical Pharmacology Reviewers). In addition to the review memoranda,
I have also reviewed the Advisory Committee briefing materials, pertinent portions of the sponsor’s
submission, and multiple scientific statements submitted by the public, including a letter from a large
number of DMD experts.
The review team has done an exemplary job in performing a detailed evaluation of the data submitted
with the application. Nevertheless, I disagree with certain of their findings and come to a different
conclusion, as discussed below.
I find that the data contained in NDA 206488 meet the standard for accelerated approval under 21 CFR
314. 510 based on the surrogate endpoint of increased dystrophin protein production, a surrogate
endpoint that I conclude is reasonably likely to predict clinical benefit.
DISCUSSION
Extensive analyses have been performed by the team on the clinical results of the long-term experience
of 12 patients administered the drug, and I will not recapitulate these.
Approval under 314.510 is based, among other things, on adequate and well-controlled clinical trials
establishing that the drug product has an effect on a surrogate endpoint that is reasonably likely to
predict clinical benefit. Below, I discuss how both of parts of this standard are met.

1
Reference ID: 3959035

 A. Are the Data on Dystrophin Protein Production From One or More Adequate and WellControlled Studies?
The characteristics of adequate and well-controlled studies are laid out in 21 CFR 314.126. Three lines
of evidence are pertinent to the conclusion that eteplirsen results in increased dystrophin production.




Production of an appropriate mRNA transcript
Quantitative assessment of dystrophin content in muscle biopsies by Western blot
Semi-quantitative assessment of dystrophin in muscle tissue by immunohistochemistry (IHC)
techniques

The sponsor provided data demonstrating an increase in mRNA expression following treatment with
eteplirsen. The drug’s proposed mechanism of action is to bridge a section of the pre-RNA to result in a
shorter mRNA with an open reading frame, e.g., “exon skipping.” In this case, the production of an
appropriate mRNA transcript has been documented by PCR and Sanger sequencing. Although this
establishes proof of mechanism, it does not mean that there is increased protein production.
In the following, I discuss the assessments related to dystrophin protein production (2. and 3.) in some
detail. Much of the controversy over the adequacy of these assessments relates to the fact that
rigorously validated assays were not used to evaluate the initial 3 muscle biopsies, apparently resulting
in overestimation of the various readouts and some irreproducibility of IHC and Western blot dystrophin
assays. For these reasons, I do not discuss or rely upon the results of these earlier assays, or on re-reads
of them. With FDA’s assistance, the sponsor improved the design and conduct of the assays and
performed repeat biopsies on 11 of 12 patients at week 180. The control samples for these week 180
biopsies were stored baseline tissue (in 3 of 11 subjects) and baseline biopsies from subjects with exon
51 amenable mutations enrolled in another trial by the sponsor. FDA reviewers had the following
concerns about these controls, leading them to conclude that the studies were not adequate and well
controlled.
1. Most of the baseline biopsies were not from the same subjects as the week 180 biopsies (as the
original tissue had been used up for the previous assays). Given this, the control subjects could
differ in unknown ways from the test subjects.
2. The biopsies taken at week 180 were from different muscles in the upper extremity than the
baseline biopsies, including subjects with baseline tissue as well as for control samples. It is
hypothesized that there may be differences in dystrophin protein content among various muscles
in DMD patients.
3. The existing baseline biopsies for the three subjects with 180 week data had been stored frozen
for several years and may have changed (apparent decrease in dystrophin protein content) over
time.
In my judgment, these issues increase the uncertainty around the results, but do not necessarily render
them an inadequate basis on which to draw a conclusion. The non-treated control subjects were very
similar in age and dystrophin mutation site to the treated subjects (sponsor Appendix 10, AC briefing
package). The single deltoid muscle biopsy in the untreated control group (subject 7, sponsor Appendix
14, AC briefing package) had replicate dystrophin levels of 0.3% and below the limit of quantification,
averaging out at below 0.3%, and not different than biceps biopsy results in other patients, suggesting

2
Reference ID: 3959035

 that variations in upper extremity biopsy site (concern b above) did not result in large differences in the
findings. There was little difference in the dystrophin protein content found in the stored baseline
samples and the frozen samples, as discussed below.
The data submitted with the original application, supporting the finding that eteplirsen increases the
production of dystrophin protein, come from the quantitative assessment of (internally truncated)
dystrophin in muscle tissue by Western blot using the controls described above. Much of the
controversy around this method relates to the fact that the apparently achieved dystrophin levels are very
much lower than originally hoped (and previously claimed by the sponsor and investigators).
In the 180 week assessment, the three subjects with baseline biopsies available had baseline dystrophin
levels (reported as % of normal) below the level of quantification of the assay used (0.25%). These
results were similar in magnitude to the baselines of the six additional control biopsies drawn from
subjects in another study (highest level 0.37%). At week 180, two treated subjects had (an average of
replicate) dystrophin levels above 2%, two had over 1%, and two additional had about 1%. Of these
individuals, two subjects having both baseline and week 180 samples had clearly increased levels at
week 180 compared to baseline. (The third subject with a baseline sample did not consent to a week 180
biopsy). Unsurprisingly, some subjects had week 180 dystrophin levels similar to the overall baseline
control levels. Not all individuals are expected to respond to a drug intervention. The issue is whether
the dystrophin levels found at 180 weeks were within the variability expected for this assay in such
patients and, thus, could have arisen by chance, or whether they could have been caused by differences
from the controls or from sample storage as outlined above, or whether they reflected a drug effect, and,
thus, whether these data could be seen as adequate and well-controlled. The following data are relevant
to this issue.
Because the original data on the presence of dystrophin by Western blot suffered some difficulties in
interpretation because of lack of availability of baseline samples from most patients, the sponsor of this
application submitted, subsequent to the Advisory Committee meeting on this drug, additional Western
blot data from 12 patients with baseline and 48 week eteplirsen exposure, using baseline and posttreatment muscle biopsies from the same patients and muscle groups. This experiment clearly shows,
using adequate controls, that the drug increases dystrophin protein production in some of the patients.
The mean baseline dystrophin values in this study were very similar to the mean baseline values in the
180 week study. The achieved levels of dystrophin in these patients are lower than those seen in the
Western blots from the week 180 patients. Only 2 of 12 patients achieved a level over 1% of normal
control. It is not known if this result is due to a shorter duration of drug exposure or to other factors.
Putting together the 180 week data and the additional 48 week data, I conclude that there is substantial
evidence from Western blot experiments of increased dystrophin protein production, albeit at a low
level.
A finding of increased dystrophin was also seen in several IHC assays performed by the sponsor. Both
assays were originally performed with baseline and several pre-180 week assays by the sponsor as a part
of the clinical trial. The validity of the results of these assays were questioned by FDA because of
methodological problems in their conduct, as documented in the primary clinical review and in the
inspection report. Therefore, I will not further consider the results of these original assays. As
discussed for the Western blot above, the sponsor responded by performing an additional 180 week
biopsy and repeating the assays. Baseline tissue was available, as for Western blot, from recut samples

3
Reference ID: 3959035

 in only three cases. In one of these, the subject did not consent to a biopsy at 180 weeks. To
supplement the three baseline samples the sponsor included six other untreated patients from a different
trial, as discussed above for the Western blot. In both assays, greater staining or intensity was observed
after drug exposure at week 180 compared to controls. The results are described in more detail below.
A Percent Dystrophin Positive Fibers analysis was a semi-automated evaluation performed at 180 weeks
and compared to the controls used for the 180 week study as discussed above. The percentage of
positive fibers was assessed using a blinded read by Nationwide Children’s Hospital and by three
independent pathologists through Flagship Biosciences. The technique used to assess percent positive
fibers was modified from the original assay in the following ways:
1. A computer algorithm (MuscleMap from Flagship) that performs non-linear mapping of all
fibers was used for consistent and automated analysis of low intensity values, in contrast to a
manual and non-standardized fiber counting technique in the prior assay.
2. The images were inverted and amplified to score the total fibers (the denominator for the percent
positive fiber scoring).
3. An isotype matched secondary antibody staining step was incorporated to confirm lack of nonspecific staining and reduce background noise. The background signal was subtracted from test
sample values in calculation of percent intensity.
4. 8% of the images for re-analysis were blinded, renamed, randomized, and rotated 180 degrees.
5. A rejection factor for the inter-rater analysis score of <4 was established.
6. The images were acquired in a more systematic and random fashion to minimize bias, with
predefined rules for random sampling of fields and avoiding artifacts.
These changes were likely to result in a more conservative reading of Percent Dystrophin Positive
Fibers, and indeed the results, including the new untreated baseline controls, were read at 1.1% positive
fibers (in contrast to a higher result in the prior baseline using the original technique). The 180 week
cohort had a score, using this technique, of 17.4% positive fibers, showing a statistically significant
difference. Now, these results are subject to the same caveats as discussed for the Western blot (1-3
above), in that there were only two baseline to 180 week pairs, that the baseline samples had been frozen
for years, and that the external controls might differ in some way. So, these results cannot stand alone.
Other reviewers have pointed out that the (much higher) baseline values for Percent Positive Fibers from
the original experiment are not very different from the 180 week values in this new experiment.
However, I would point out that experimental conditions changed quite a bit, and very low values for all
the external controls, statistically comparable to the frozen baseline results, were obtained in this recent
experiment, suggesting that it returned a more conservative result. I do not believe that comparison of
the original baseline data, obtained under one set of experimental conditions, can be compared to the
later 180 week results, done under different, more optimized conditions and yielding very different
results for new (external control) baseline samples.
The sponsor also performed a Mean Relative Fluorescence Intensity assay for dystrophin. This assay is
commonly performed by laboratories evaluating DMD patients and is intended to be a semi-quantitative
evaluation of dystrophin content. Using the six external baseline samples and the three stored study
patient baseline samples, the mean intensity approximately doubled from baseline to 180 weeks. The
technique for this assay did not change significantly from the technique used in the assay done as part of

4
Reference ID: 3959035

 the original protocol, and the baseline means for the patient samples were roughly comparable to the
baseline means obtained in the new experiment.
Although the IHC assays provide only semi-quantitative assessments of dystrophin content, they do
support an effect of eteplirsen on the proposed surrogate endpoint (an increase of dystrophin production
as a result of drug exposure). The accompanying microscopy images also demonstrate correct
localization of the molecule within the muscle fibers, a very important factor in any translation to
clinical benefit.
In summary, I conclude that there is evidence from adequate and well-controlled trials, and supportive
evidence, that exposure to eteplirsen increases dystrophin protein production in muscle cells.
B. Is the Effect on the Surrogate Endpoint “Reasonably Likely to Predict Clinical Benefit”?
In this case, the standard for clinical benefit does not require “cure” or “conversion to Becker MD
(BMD) phenotype.” Clinical benefit encompasses improvements (including slowing of disease
progression) in how an individual feels or functions, or an improvement in survival. There is no
question that, for DMD patients and their families, small improvements in function or delays in loss of
function are meaningful benefits. Therefore, the question is:
What amount of increase in dystrophin production is reasonably likely to predict clinical
benefit (even small benefits)?
The usual way to address this question would be to rigorously evaluate what is known about the
correlation between dystrophin levels in muscle and expression of disease. The following summarizes
the existing scientific literature on this topic and the challenges in interpreting it.
1. The clinical classification of disease severity (i.e., phenotype) in the literature appears broad,
variable, and somewhat subjective.
Experts usually classify patients clinically as DMD (severely affected at a young age); intermediate
MD (also called DMD/BMD); or BMD, which can range from severe BMD to asymptomatic
individuals with biochemical abnormalities, usually increased creatine phosphokinase (CPK). There
is clearly a wide spectrum of disease wherein the ends of the spectrum are easily distinguishable, but
the zone of real interest for this discussion, between DMD and intermediate presentations, is not
rigorously categorized. In part, this is because “intermediate muscular dystrophy” (IMD) is less
common, due to the consequences of having either in-frame mutations with a truncated protein
expressed (leading to BMD) or out-of-frame mutations with little-to-zero protein expressed (leading
to DMD), as discussed below.
2. Much of the prior data reporting the relationship of dystrophin protein levels to phenotype have
been from IHC studies using a variety of techniques and antibodies.
Anthony, et al., (Neurology, 83, 2014) in a collaborative cross-laboratory study, investigated the
variability of techniques used to quantify dystrophin in individuals with muscular dystrophy.
Blinded tissue sections from three DMD and three BMD muscle biopsies were tested in five
5
Reference ID: 3959035

 different laboratories accustomed to performing dystrophin quantification. Estimates of dystrophin
expression using a somewhat standardized IHC technique were about 20%, 11% and 10% of normal
for the three DMD samples, on average among the laboratories. Corresponding estimates of
dystrophin content by Western blot, using an actin antibody to normalize for loading, but not a
serially diluted standard control, resulted in dystrophin estimates of about 11%, 0, and 0.4%
respectively, with fairly high CV’s. Therefore, in this small sample, repeated across five
experienced laboratories, IHC estimates were about 10 percentage points higher than Western blot
estimates.
Significantly higher estimates by IHC by fluorescence intensity (overall about 23% of normal) than
by Western blot were also seen in the evaluation of week 180 muscle biopsies in the Sarepta trial.
Because much of the historical data on protein content vs phenotype has been reported using IHC
analysis, extrapolating these findings to the current trial data is challenging. Additionally, Anthony
et al., found that the inter-laboratory variability was greatest for the low levels of dystrophin found in
the DMD patients. Western blot data in the literature quantifying dystrophin and relating it to
phenotype is often from experiments that were not designed to distinguish among dystrophin levels
below 10% of normal. These may have been reported out as “less than 10%.” From this sponsor’s
well-controlled studies, the analytically accurate dystrophin baseline for many DMD patients might
be in the range of 0.02-0.35 % normal, hence previous estimates of 5-10% might be an overestimation using non-standardized and semi-quantitative methods.
3. Both IHC analyses and WB results are influenced by the anti-dystrophin antibodies used, as well
as other experimental conditions
Significantly, if the epitope recognized by the antibody is modified by the deletion, the dystrophin
isoform may not be recognized and a result read out as zero. For this reason, recent studies use
multiple antibodies against known regions. Additionally, muscle biopsies in patients with BMD and
DMD may be quite variable in degree of fibrosis and fatty replacement; this may decrease the
reproducibility and representativeness of muscle biopsy estimates of dystrophin content by Western
blot. Additionally, imaging methods, choices for normalization, biopsy handling, background
standing, and a multitude of other experimental conditions can influence results.
4. The phenotype is significantly influenced by dystrophin isoform quality as well as dystrophin
quantity.
Dystrophin is a very large protein with multiple functional domains. Generally, DMD results from
an out-of-frame mutation (often a deletion) that leads to an unstable or unreadable mRNA transcript.
Thus, DMD patients usually have zero or very low levels of dystrophin, but the DMD phenotype can
also result from in-frame mutations that result in a unstable transcript or dysfunctional dystrophin
isoform. BMD usually results from an in-frame mutation (often an exon deletion) that affects the
functional quality of the protein and also the quantity produced. It remains unclear what role protein
function plays vs quantity in leading to the wide range of variability in BMD phenotypes. There are
a vast number of mutations that can lead to each of these phenotypes (Tuffery-Giraud, et al., Hum
Mutat, 30, 2009), all of which can have different effects on protein function as well as protein
production. This micro-heterogeneity is common in genetic diseases and is highly germane to

6
Reference ID: 3959035

 evaluation of interventions targeting the gene, gene expression, or protein function. There are also
non-dystrophin-related factors that can modulate phenotype.
5. The literature contains various findings on the relationship of dystrophin expression to clinical
status, including the low levels of dystrophin protein of interest in this case.
I note that in the decades since 1988, much technical progress has been made in standardizing
Western blot techniques, and the results from early studies may not be fully comparable to those
from recent experiments.
a. The seminal 1988 paper on this subject (Hoffman et al., NEJM, 318(21)) found that the
majority of patients with DMD had undetectable levels of dystrophin using their Western
blot technique and that 35 of 38 had levels below 3% in their assay. They also reported that
one of seven “intermediate” patients had dystrophin levels below 3% of normal, as did one of
the 18 patients with a BMD phenotype.
b. Beggs et al., (Am J Hum Genet, 49, 1991) published one of the early studies on the
correlation between the level of dystrophin on Western blot and clinical features of BMD.
Western blot was performed using a polyclonal serum and had about a 20% variability
between blots according to the authors. In this study a number of patients with BMD or
intermediate phenotype (DMD/BMD) were found to have dystrophin contents that
overlapped with those of the DMD patients. Of four patients included with DMD phenotype,
two had less than 5% dystrophin, and two had 10%, by their assay. Of patients with
BMD/DMD phenotypes, eight were found to have 10% of normal dystrophin, two had 15%,
one had 50%, and one had 100%. Three BMD patients with dystrophin levels of 10% were
found; two of these had relatively mild disease.
c. Nicholson et al., (J Med Genet, 30, 1993) studied patients across a wide range of DMD and
BMD phenotypes. They used loss of ambulation as a criterion to establish five functional
groups, grouped from one (most severe, LOA before age 9) to five (LOA past age 40) (presteroid era). They found a linear relationship overall between dystrophin levels (Western
blot with Dy4/6D3 antibody, using myosin for a loading control) and their five categories,
with more dystrophin protein translating to better function. They found no significant
difference between any two adjacent groups however, which they interpreted as showing
considerable overlap, as reflected in their patient level data (Appendix 1), which showed a
number of less severe patients (e.g., Group 2 or 3) registering no or very low dystrophin
abundance on their Western blot assay. Of note, they reported a higher average level of
dystrophin protein in severe DMD patients than other investigators, partly resulting from 5 of
their 21 severe patients reported to have dystrophin protein levels above 20.
d. Neri et al., (Neuromuscular Disorder 17, 2007) reported on families with X-linked Dilated
Cardiomyopathy. In these families, mutations give rise to absent dystrophin in heart muscle,
but only reduced levels of nearly normal dystrophin in muscle tissue. One patient in their
series had a normal neurological exam at age 23, an elevated CPK, and 29% of normal
dystrophin protein in skeletal muscle by Western blot. This example can contribute to
understanding the role of abundance of dystrophin protein vs compromised function.

7
Reference ID: 3959035

 e. Anthony et al., (JAMA Neurology, 71, 2014) evaluated the correlation between phenotype
and mRNA and protein expression in patients with both in-frame and out-of-frame mutations
amenable to exon 44 or 45 skipping. Studying a group of patients with closely related
deletions could diminish variability due to differences in function of the truncated protein.
Five samples from patients with clinical “mild” BMD and in-frame mutations underwent
Western blot analysis using the Dys-2 antibody. Their mean protein expression was 17%
(normalized to actin) with a standard deviation of 7.5%. Two of the “mild” patients had
dystrophin levels in this assay of around 10%. Based on comparisons of IHC experiments
with various antibodies, the authors found “no clear correlation between the level of
dystrophin transcript or protein expression with clinical severity” in 13 patients with inframe mutations leading to BMD. The finding of Neri et al., above, along with this report,
reinforce the concept that protein function (i.e., quality) is an important determinant of
clinical severity and undermine the concept that 10% dystrophin protein content is a
threshold, since these patients had “mild” BMD.
f. Van den Bergen et al., (J Neurol Neurosurg Psychiatry, 85, 2014) compared dystrophin
levels by Western blot with clinical severity in 27 patients with a clinical diagnosis of BMD.
Dystrophin expression ranged from 4-71% and 3-78%, depending on the antibody used. The
authors found no linear relationship between dystrophin expression by Western blot using
newly acquired muscle biopsies and clinical severity, muscle strength, or fatty infiltration on
MRI. Although this was the case for the majority of the patients, who had dystrophin levels
above 20% of normal, four patients had levels at or below 10%. These patients generally had
a more severe phenotype: one patient with a dystrophin level of 10% was wheelchair
dependent at 45 years; one patient with a level of 7% developed trouble with stair walking at
age 21; one patient with a level of 4% had a DMD phenotype with wheelchair dependency at
age 10, one patient with a level of 3% had wheelchair dependency at age 25.
g. Anthony et al., (Brain, 134, 2011) studied 17 BMD patients with exon 51 or 53 skippingamenable mutations by IHC methods. These patients primarily had very mild or
asymptomatic disease; the one patient classified as severe was ambulatory at age 25 but
unable to run. There was a statistically significant difference in dystrophin expression by
IHC when patients classified as mild disease were compared to asymptomatic patients.
h. Bello et al., (Neurology 87, 2016) published a detailed study of loss of ambulation in DMD
patients with particular exon deletions, using the CINRG-DNHS, a prospective natural
history study. They found patients with exon 44 amenable mutations to have a two-year
delay in loss of ambulation compared to the overall comparison group. This finding had
previously been reported by another group (van den Bergen, et al., J Neuromuscul Dis, 1,
2014). The mutations studied (primarily single-exon deletion of exon 45) are known to
undergo spontaneous skipping with production of some dystrophin. According to the Bello
report, of six patients previously tested by IHC, three showed traces of dystrophin production
and 0/four (possibly other patients) had dystrophin detectable by Western blot. These
authors suggest that the observed differences in loss of ambulation (LOA) could be due to
small amounts of spontaneously induced dystrophin that slightly ameliorate the ordinary
DMD phenotype.

8
Reference ID: 3959035

 i. Cirak et al., (Lancet, 378, 2011) published a study (AVI-4658) using intravenously
administered eteplirsen that showed a detectable increase in dystrophin protein levels using
both Western blot and immunofluorescence in 3/19 patients. The authors reported that the
functional properties of restored dystrophin were confirmed by assessing increased levels and
co-localization of neuronal nitric oxide synthase (nNOS) and sarcoglycan with dystrophin.
Such a protein assembly is suggested to be indicative of functional restoration of the
dystrophin-associated glycoprotein complex in muscle fibers (Molza et al., JBC, 290, 2015;
Wells KE et al., Neuromuscul Disord, 2003). Cirak et al., reported that the restoration was
more so in patients with exon 49-50 deletions than in those with 45-50 deletions, which is
consistent with a previous observation that nNOS binding domain is located in dystrophin
exons 42-45 (Lai Y et al., J Clin Invest, 2009). These studies suggest that important
functional domains are included in the dystrophin protein induced by eteplirsen.
To summarize what is known about the association between dystrophin levels and phenotype,
dystrophin content above about 10% on Western blot is usually associated with a BMD phenotype,
except in patients with higher levels of dystrophin (including above 50%) who potentially have
functionally deficient protein leading to a DMD phenotype. Within the BMD phenotype, a
proportional inverse relationship between disease severity and protein expression has not generally
been demonstrated (i.e., between 10-100%), although there may be a broad association, as seen in
the Anthony study (Brain, 134, 2011). This may be due to the fact that protein quality, rather than
quantity, plays a key role in determining phenotype in BMD. Patients with DMD are usually found
to have no detectable, or very low levels of, dystrophin. Dystrophin content in the 3-10% range has
been associated with DMD, DMD/BMD, and BMD phenotypes. I find no evidence of a threshold
value for protein content and expression of a DMD phenotype, although the majority of DMD
patients reported in the literature have dystrophin that is undetectable by the Western blot assays
used. Generally, the divide between DMD and BMD, in terms of protein, is the result of the
consequences of an OOF or an in-frame mutation, respectively. I believe that the conventional
threshold, at or below 10% protein, was derived from the IHC data that seem to estimate low-level
protein content about 10% percentage points higher on IHC than on Western blot, so that the
majority of DMD patients would read out at 10% of normal dystrophin on IHC. I believe that
evidence from Western blot and other experiments discussed above show that protein in the range
between undetectable and 10% of normal is likely to be very important for clinical presentation, all
other things being equal, i.e., mutation status and non-dystrophin-related factors affecting phenotype.
These findings are germane to the determination of “reasonably likely to predict clinical benefit.”
The broad phenotypic distinctions made in the clinic (e.g., DMD vs IMD vs BMD) are different
from the prediction of benefit to an individual patient who has a specific baseline dystrophin level
and whose mutation and external factors do not change pre- and post-drug. For example, extending
ambulation by six months to a year would not normally move a patient from one to another of these
categories, but could be very important to quality of life (e.g., as suggested in the Bello study). This
is also true for other functional improvements.
For these reasons, incorporating the analysis of dystrophin content discussed above, I conclude that
the biochemical data strongly support the idea that low-level increases in dystrophin production are
reasonably likely to predict clinical benefit.

9
Reference ID: 3959035

 Additional support for “reasonably likely” comes from the long-term experience with the drug. The
sponsor’s comparison of the experience of the treated cohort to natural history data does not reach
the level of substantiation required for traditional approval based on the clinical data. However, it is
highly suggestive of improvement in some parameters, in some patients, over natural history. My
conclusion is informed by all the caveats expressed in the reviews about the pitfalls of nonrandomized comparisons. Given that the two exon 52 deletion patients in the study had fairly good
long-term results in terms of rate of disease progression, the question arises as to whether exon 52 is
a prognostic factor that could have skewed the results.
Several facts militate against this conclusion. First, one of the exon 52 deletion trial subjects
(subject 6) had a fairly low score on the 6MWT at entry and a very low score on the NSAA,
compared to other subject around his age. He also was the only subject in the trial noted to be
unable to rise without external support at baseline. Additionally, the Italian external cohort had exon
52 deletion representation.
Questions have been raised about the correlation of dystrophin levels from Western blot with clinical
outcomes. The 6 Minute Walk Test does not show a strong correlation. I evaluated the NSAA in
children who could still walk (because the NSAA primarily scores activities related to walking) and
who also had a dystrophin result at 180 weeks (Table 1). I did this because the NSAA includes
multiple measures and therefore might have some noise averaged out. I looked at the absolute
decline in NSAA in patients since study initiation, and did not correct for the initial time some
patients spent on placebo. I only evaluated patients who were ambulatory. There was a positive
(inverse) correlation between dystrophin by Western blot and rate of decline in NSAA score, .
(Figure 1) This adds additional support to the idea that dystrophin production is “reasonably likely
to predict clinical benefit.” In totality, I find that the comparative disease course data provide
additional support for the use of the surrogate endpoint of an increase in dystrophin expression as
“reasonably likely to predict clinical benefit.”
Therefore, both the biochemical data and the clinical data lead me to conclude that an “increase in
dystrophin production” is reasonably likely to predict clinical benefit in DMD.
CONFIRMATORY TRIALS
The sponsor is currently conducting a nonrandomized, concurrently controlled trial in patients with
mutations amenable to exon 51 skipping compared to untreated DMD patients with other exon deletions.
Because of the relatively low level of protein induced, additional doses should be aggressively pursued
and, if successful, a dose-comparison trial could be confirmatory. The sponsor has also planned to
initiate a randomized trial with a related compound in other exons. The clinical results from these trials
can inform the predictive value of the surrogate endpoint.

10
Reference ID: 3959035

 EXPLORATION OF ADDITIONAL DOSES, REGIMENS, AND DRUG-MUTATION
INTERACTION
The dystrophin levels achieved in this development program are well below those initially hoped for. I
agree with Dr. Farkas and other reviewers that the sponsor should aggressively explore higher doses or
more frequent administration of eteplirsen. It appears that this is possible given the toxicology data and
the clinical safety profile observed to date.
Because patients in the Sarepta 180 week cohort had a range of deletions in the dystrophin gene,
variability in the pharmacodynamic response among deletions is of great interest. The two patients with
over 2% dystrophin in the 180 week Western blot both had exon 52 deletions. These patients also fared
fairly well, clinically. This raises the question of whether patients with this exon deletion naturally
produce more dystrophin. One of these subjects had a baseline sample available. It was found to be
below the limit of quantitation. There was an exon 52 subject included in the added baseline controls.
This subject’s assay had replicate results of 0.3% and below the limit of quantification, respectively, as
discussed above. This suggests that baseline dystrophin levels are not higher in exon 52 deletion
subjects and that there may be a drug-deletion interaction, wherein subjects with this deletion may have
a more robust pharmacodynamic response to the drug. There were a number of apparent non-responders
to the drug. It will be important to find out if this is mutation specific. It is likely that more detailed
knowledge about each patient’s specific mutation will have to be generated to study this in detail.
COMMENTS ON THE DEVELOPMENT PROGRAM AND REVIEW
The development program for eteplirsen was seriously deficient in a number of respects that may have
led to delay in broad access and certainly led to difficulties in regulatory review. In my assessment, the
most egregious flaw was the lack of robust and high-quality assays early in the development program.
Inaccurate conclusions from the assays used led to a flawed development program. Additionally, the
entire drug development field must recognize that there is no such thing as an “exploratory study” for a
serious, life-threatening illness without therapeutic options. Randomization should be performed very
early in the development program, and open-label studies should be avoided. When possible, seamless
adaptive dose-finding and early efficacy studies should be carried out with the goal of most efficiently
generating the data needed to demonstrate safety and effectiveness.
The flaws in the eteplirsen development program led to severe challenges in regulatory review. 21 CFR
312.80, concerning drugs intended to treat life-threatening or severely-debilitating illness, states that
FDA has determined “that it is appropriate to exercise the broadest flexibility in applying the statutory
standards, while preserving appropriate guarantees for safety and effectiveness…Physicians and patients
are generally willing to accept greater risks or side effects from products that treat life-threatening and
severely-debilitating illnesses than they would accept from products that treat less serious illnesses.” I
note that the acceptable risks include greater uncertainty about the effects of the drug. The Peripheral
and Central Nervous System Drugs Advisory Committee met on this application on April 25, 2016.
There was a split vote (7 against, 6 for) on the question of accelerated approval for this drug, reflecting
the greater than usual uncertainty about the application. This vote was taken before the additional data
on protein expression were submitted.

11
Reference ID: 3959035

 To conclude, the studies used in this analysis to support the effect of eteplirsen on dystrophin were
adequate and well-controlled as specified in 314.126. In addition, the surrogate of increased dystrophin
production is reasonably likely to predict clinical benefit. Given the deficiencies that have been
identified in the development program, my conclusion to rely on the surrogate endpoint described above
represents the greatest flexibility possible for FDA while remaining within its statutory framework. In
this case, the flexibility is warranted because of several specific factors, including: the life-threatening
nature of the disease; the lack of available therapy; the fact that the intended population is a small subset
of an already rare disease; and the fact that this is a fatal disease in children. Of note, the therapy has
been relatively safe in the clinic, although intravenous administration always carries risk. In addition,
adequate confirmatory studies are underway and planned and are capable of further refining our
understanding of the biomarker and providing evidence about the nature of the clinical benefit. The
approval does not create any risk of compromising the confirmatory trials because of their nature.
Therefore, I find that the probable benefits outweigh the foreseeable risks and that this application
should be approved under 21 CFR 314.510.

12
Reference ID: 3959035

 Table 1 Patient Data on Change from Baseline in 6MWT and NSAA
180 Weeks
Change from
Baseline
Subject

Baseline
WB

002
003
004
005
006
007
008
009
010
012
013
015

N/A
N/A
N/A
0
N/A
N/A
N/A
N/A
N/A
N/A
0
0

180
Week
WB
0.14
0
0.96
N/A
2.47
0
0.98
0.52
1.62
0.38
1.15
2.05

Fiber
Intensity

PDPF

∆
6MW

∆
NSAA

MD
7
28
N/A
29
12
26
23
21
26
32
30

4.54
1.42
29
N/A
21
7
12
22
24
33
19
18

-67
-174
-168
-231
-23
-197
-291
0
0
-114
-170
-1

-14
-27
-17
-19
-3
-19
-17
0
0
-19
-16
-13

Data from Sarepta
Therapeutics , Inc. PCNSD
Advisory Committee Briefing
Document, Appendix 5, p.
149 (6MW and NSAA0
Appendix 11, p. 155, (Percent
Positive Dystrophin Fibers
(PPDF), Appendix 12 p. 156
(fiber intensity) 14, p. 159.
(Western blot),

13
Reference ID: 3959035

 Figure 1. Decline in NSAA by
on Western

blot

Reference ID: 3959035

 

 

8



Change in NSAA
pWestern Blot

 

0 Patient

Linear (Patient)

 

14

 

--------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed
electronically and this page is the manifestation of the electronic
signature.
--------------------------------------------------------------------------------------------------------/s/
---------------------------------------------------JANET WOODCOCK
07/14/2016

Reference ID: 3959035

 Office of Drug Evaluation-I: Decisional Memo
Date
From
Subject
New Drug Application (NDA) #
Applicant Name
Date of Submission
PDUFA Goal Date
Proprietary Name/
Established (USAN) Name
Dosage Forms/ Strengths
Indication originally sought by
applicant (see page 29 for final)

Action:

July 15, 2016
Ellis F. Unger, MD, Director
Office of Drug Evaluation-I, Office of New Drugs, CDER
Office Director Decisional Memo
206488
Sarepta Therapeutics
June 26, 2015
May 26, 2016 (post-3-month extension for major amendment)
EXONDYS 51™
eteplirsen injection
2 mL single-use vials containing 100 mg (50 mg/mL) eteplirsen
10 mL single-use vials containing 500 mg (50 mg/mL) eteplirsen
“EXONDYS 51 is indicated for the treatment of Duchenne muscular
dystrophy (DMD) in patients who have a confirmed mutation of the
DMD gene that is amenable to exon 51 skipping. This indication is
approved on an intermediate endpoint demonstrating delayed
disease progression as measured by the 6 minute walk test [see
Clinical Studies (14)]. Continued benefit will be evaluated through
confirmatory trials.”
Complete response

Material Reviewed/Consulted - Action Package, including:
Project Manager
Yuet (Fannie) Choy, Laurie Kelley
Medical Officer/Clinical
Clinical Pharmacology/Pharmacometrics
Statistical Review
Pharmacology Toxicology
Office of Biotechnology Products

Christopher Breder
Atul Bhattaram, Ta-Chen Wu, Hobart Rogers, Kevin Krudys,
Angela Men, Christian Grimstein, Mehul Mehta
Xiang Ling, Kun Jin, Hsien Ming (Jim) Hung
David Hawver, Lois Freed, Paul Brown

Method Validation

Ashutosh Rao, Amy Rosenberg
Joseph Leginus, Mari Chelliah, Sung Kim, Denise Miller,
Zhong Li, Dahlia Woody, Martha Heimann, James
Laurenson, Donna Christner, Neal Sweeney, Edwin Jao,
Zhihao Peter Qiu, Wendy Wilson-Lee, M. Scott Furness
Antoine El Hage, Cara Alfaro, Susan Thompson,
Kassa Ayalew, Ni Aye Khin
Michael Hadwiger, Michael Trehy

Statistical Review – Stability data

Zhuang Miao, Xiaoyu Dong, Meiyu Shen, Yi Tsong

Office of Prescription Drug Promotion
Division of Medication Error Prevention
and Analysis

Aline Moukhtara
Deborah Meyers, Justine Harris, Danielle Harris,
Todd Bridges
Robert Pratt, Jamie Parker, Kellie Taylor,
Cynthia LaCivita
Tracy Peters

Office of New Drug Quality Assessment
Office of Scientific Investigations

Division of Risk Management
Associate Director for Labeling
Cross-Discipline Team Leader
Deputy Director, Division of Neurology
Products

Ronald Farkas
Eric Bastings

Office Director Decisional Memo – NDA 206488 – Page 1 of 42
Reference ID: 3959961

 1.

Introduction

Sarepta Therapeutics is seeking accelerated approval for eteplirsen for the proposed indication:
“EXONDYS 51 is indicated for the treatment of Duchenne muscular dystrophy (DMD) in
patients who have a confirmed mutation of the DMD gene that is amenable to exon 51
skipping. This indication is approved on an intermediate endpoint demonstrating
delayed disease progression as measured by the 6 minute walk test [see Clinical
Studies (14)]. Continued benefit will be evaluated through confirmatory trials.”
I agree with the views of the Division of Neurology Products (DNP), the Office of Biometrics, and
the Office of Clinical Pharmacology that the applicant has not provided substantial evidence of
effectiveness from adequate and well controlled trials to support conventional approval. I also
agree that the applicant has not provided support for accelerated approval, i.e., evidence from
adequate and well controlled trials of an effect on a biomarker that is reasonably likely to predict
effectiveness. Thus, I agree with the DNP recommendation to issue a Complete Response for
this application.
2.

Background

Description:
Eteplirsen is a phosphorodiamidate morpholino oligomer (PMO) designed to target the premRNA transcripts of the dystrophin gene so that exon 51 is excluded, or skipped, from the
mature, spliced mRNA. Theoretically, restoration of the mRNA reading frame would permit
translation of an internally truncated, but nevertheless functional form of the dystrophin protein.
The drug is targeted specifically for patients with DMD “who have a confirmed mutation of the
DMD gene that is amenable to exon 51 skipping.” It is not clear which of the specific mutations
are amenable to exon 51 skipping.
PMOs are a class of synthetic molecules based upon a redesign of the natural nucleic acid
structure. They are distinguished from native DNA and RNA because of a 6-membered
morpholino ring that replaces the 5-membered ring found in native DNA and RNA. Each
morpholino ring is linked through an uncharged phosphorodiamidate moiety rather than the
negatively charged phosphate linkage that is present in native DNA and RNA. Each morpholino
subunit contains one of the heterocyclic bases found in DNA (adenine, cytosine, guanine, or
thymine). Eteplirsen contains 30 linked subunits. The molecular formula of eteplirsen is
C364H569N177O122P30 and the molecular weight is 10.3 kilodaltons.
Disease Background:
Duchenne muscular dystrophy is an X-linked recessive neuromuscular disorder caused by
mutations of the dystrophin gene located on the short arm of the X chromosome. These
mutations disrupt the mRNA reading frame, leading to the absence or near-absence of
dystrophin protein in muscle cells. The disorder affects 1 in ~3,600 boys (~1 in 10,000 to
14,000 males). Patients who are amenable to skipping exon 51 constitute ~13% of the DMD
patient population.

Office Director Decisional Memo – NDA 206488 – Page 2 of 42
Reference ID: 3959961

 Dystrophin is thought to maintain the structural integrity of the muscle cell membrane by
connecting the cytoskeleton to the underlying extracellular matrix, and acting as a scaffold for
several molecules that also contribute to normal muscle physiology. Absence of dystrophin
leads to mitochondrial dysfunction and damage, with inflammatory processes also appearing to
contribute to muscle pathology. Muscle fibers ultimately undergo necrosis with replacement by
adipose and connective tissue. Principal disease manifestations include progressive
degeneration of skeletal and cardiac muscle, leading to loss of physical function in childhood
and adolescence with premature death from respiratory and/or cardiac failure in the second to
fourth decade.
No specific therapies are approved for DMD. Currently, glucocorticoid therapy is the
cornerstone of clinical management, and is widely believed to delay loss of ambulation and
respiratory decline by several years. Ventilatory assistance and physiotherapy are also thought
to improve survival for DMD patients.
3.

Product Quality

From a product quality perspective, NDA 206488 is recommended for approval. Eteplirsen
would be marketed as a sterile, aqueous, preservative-free, concentrated solution for dilution
prior to IV administration, to be supplied in single-use glass vials containing 100 mg or 500 mg
eteplirsen (50 mg/mL).
OPQ recommends the following post-marketing commitments (PMCs), to be fulfilled no later
than one year following NDA approval:
1. Investigate the root cause of the increasing assay trend observed in the drug product
stability study.
(b) (4)
2. Revalidate the accuracy of the in-process
method used during drug product
manufacture.
(b) (4)
(b) (4)
3. Revalidate the robustness of the in-process
method in terms of
(b) (4)
.
4. Investigate the consistent bias in the in-process(b) (4) results and the release (b) (4)
results.
4.

Nonclinical Pharmacology/Toxicology:

From a nonclinical perspective, NDA 206488 is recommended for approval. Pivotal toxicology
studies were conducted in male monkeys (39-week study) and juvenile male rats (10-week
study). A 26-week study was conducted in male transgenic mdx mice using a mouse-specific
surrogate (AVI-4225). In all 3 species, the kidney was identified as the 1° target organ, with
dose-dependent renal tubular cytoplasmic basophilia and/or vacuolation and tubular
degeneration and necrosis, primarily at the highest doses tested.
Dilatation of the lateral ventricles of the brain was observed at mid and high doses in the mdx
mouse study. The mechanism of this effect and its relevance to humans are unknown. In
juvenile rats, slight reductions in bone length, width, area, mineral content, and mineral density
were observed at the high dose. These concerns could lead to recommendations for long-term
monitoring in patients.

Office Director Decisional Memo – NDA 206488 – Page 3 of 42
Reference ID: 3959961

 Mean eteplirsen plasma exposures (AUC) at the no observed adverse effect levels (NOAELs)
for monkeys and juvenile rats were 20- and 6-fold, respectively, higher than that of patients who
received the to-be-marketed dose of 30 mg/kg/week by the intravenous route.
The applicant presented data on the exon skipping activity of eteplirsen in cynomolgus monkeys
(“Exon skipping activity of AVI-4658 in cynomolgus monkey tissue samples from (b) (4) study
(b) (4)
;” applicant study 4658-ssa-005). Samples of quadriceps muscle, heart, and
diaphragm tissues were collected on Day 79 from cynomolgus monkeys after 12 weekly doses
of eteplirsen at 0, 5, 40, or 320 mg/kg IV, or 320 mg/kg SC. Muscle samples were analyzed for
exon 51 skipping of the dystrophin gene using polymerase chain reaction (PCR).
Exon skipping was detected in a nonlinear, dose-dependent manner (Table 1, Figure 1). With a
1-log increase in dose (from 5 to 40 mg/kg), there was little change in exon 51 skipping. With a
second log increase in dose (from 40 to 320 mg/kg), however, there was more than a log
increase in response. As noted below, the applicant studied doses of 30 and 50 mg/kg/week in
the clinic (6 patients at each dose), and there is significant question as to whether the plateau of
the dose-response curve was reached. It is possible that much higher doses could lead to
substantially greater effects on dystrophin production – effects that could be important for
efficacy.
Table 1: Average Percentage of Exon 51 Skipping in Intact Monkeys (N=8 for Each Group)

Tissue

Average % Exon 51 Splicing ± SD

Quadriceps muscle
Heart
Diaphragm

0 mg/kg IV

5 mg/kg IV

40 mg/kg IV

0.0 ± 0.0
0.0 ± 0.0
0.0 ± 0.0

0.5 ± 0.5
0.1 ± 0.1
0.2 ± 0.2

0.6 ± 0.3
0.2 ± 0.2
0.9 ± 0.7

320 mg/kg IV 320 mg/kg SC
8.2 ± 7.4
4.5 ± 2.9
6.1 ± 3.5

1.3 ± 0.5
1.4 ± 0.5
2.2 ± 0.9

Figure 1: Evidence of Exon Skipping in Quadriceps Muscle in Intact Monkeys (N=8 for Each Group)

Office Director Decisional Memo – NDA 206488 – Page 4 of 42
Reference ID: 3959961

 With respect to the advisability of evaluating higher doses in humans, this subject is well
summarized by Dr. Bastings in his Division Memo: “Considering the seriousness of DMD, the
unmet medical need, and the nature of the toxicities observed in animals, I believe that the
nonclinical data would support, with proper monitoring, dosing in DMD patients at least up to
200 mg/kg, a dose expected to provide exposure similar to the most sensitive species NOAEL
for the toxicities seen in animals. If the human safety experience at these doses is acceptable,
further dose escalation is possible in DMD patients.”
Finally, the nonclinical review team provided insight that is relevant for the interpretation of
clinical data with respect to production of dystrophin protein: “The most robust finding among
the studies provided and referenced in this submission was the wide variability in the extent of
PMO-induced dystrophin expression within a single muscle and among different muscles,
suggesting that caution is warranted in generalizing from the results of biopsies taken from only
one or a few sites, muscle types, or patients.”
Carcinogenicity:
Carcinogenicity studies have not been conducted with eteplirsen. The nonclinical review team
opined that carcinogenicity studies in 2 species should be conducted as a post-marketing
requirement. Dr. Bastings agrees, and I agree, that for this serious indication with unmet need,
carcinogenicity studies can be deferred until after marketing.
5.

Clinical Pharmacology

The Clinical Pharmacology team does not recommend approval; they recommend generation of
robust evidence of effectiveness prior to approval. Specifically, the team is recommending a
double-blind, placebo-controlled study in patients with mutations amenable to exon-51 skipping
who are likely to be ambulant for 1 year, with use of appropriate endpoints based on upper or
lower body strength in patients between 4 and 12 years of age. They also suggest study of
doses greater than 50 mg/kg administered weekly, or alternate regimens that would include
loading and maintenance doses, for example, twice-weekly administration for 6 months followed
by weekly administration for 6 months. Their recommendations are based on the 3- to 4-hour
half-life of the drug, urinary excretion of 60-70% of the drug within 24 hours, and the absence of
known toxicity at doses of 50 mg/kg. The immunogenicity of eteplirsen can be further assessed
in future clinical trial(s) as well.
Summary of Pharmacokinetics:
•
•
•
•
•

Pharmacokinetics was approximately dose-proportional and linear from 0.5 to 50
mg/kg/week, with insignificant accumulation in this dose range.
Following single or multiple intravenous infusions, peak plasma concentrations (Cmax)
occurred near the end of infusion.
Plasma concentration-time profiles showed multi-phasic decline, with virtually all drug
eliminated within 24 hours (24 hours after completion of infusion, eteplirsen concentrations
were 0.02% of Cmax).
At doses of 30 and 50 mg/kg, the elimination half-life is ~3.5 hours, with ~65% of the drug
excreted unchanged in the urine. The drug is not metabolized.
Protein binding of eteplirsen in humans is relatively low, ~6% to 17%, and is independent of
concentration.
Office Director Decisional Memo – NDA 206488 – Page 5 of 42

Reference ID: 3959961

 •
•
•
•
•
•
•

The volume of distribution data suggest distribution or cellular uptake into peripheral tissues.
Inter-subject pharmacokinetic variability is moderate, generally in the range of 20 to 55% for
exposure measures (Cmax and AUCs) as well as other key pharmacokinetic parameters.
Intrinsic factors were not studied (typically, in a larger development program, age, gender,
body weight, geographic region, hepatic impairment, renal impairment, and other potentially
significant covariates would be studied).
In vitro investigations on major CYP isozymes and transporters did not reveal the need for
additional investigation in humans.
Eteplirsen was not a significant inhibitor or inducer of CYP.
Eteplirsen was not a substrate or inhibitor for any of the key human transporters tested.
Eteplirsen is expected to have a low potential for drug-drug interactions.

Finally, the clinical pharmacology team noted that if eteplirsen were found to be safe and
effective, it would likely benefit all mutations amenable to exon-51 skipping and should be
labeled accordingly.
QT Effects:
QT effects were not formally investigated in man.
6.

Clinical Microbiology

Not applicable.
7.

Clinical/Statistical Efficacy

Sarepta is seeking accelerated approval for eteplirsen for the proposed indication:
“EXONDYS 51 is indicated for the treatment of Duchenne muscular dystrophy (DMD) in
patients who have a confirmed mutation of the DMD gene that is amenable to exon 51
skipping. This indication is approved on an intermediate endpoint demonstrating
delayed disease progression as measured by the 6 minute walk test [see Clinical
Studies (14)]. Continued benefit will be evaluated through confirmatory trials.”
In this section, I provide an explanation of how accelerated approval might be used as a
potential pathway to approval, based on production of dystrophin in skeletal muscle. I then
discuss the evidence that eteplirsen produces dystrophin in skeletal muscle, based on
immunohistochemistry and Western blot analyses. Finally, I discuss the clinical data that could
serve as the basis for a conventional approval.
Accelerated Approval:
The applicant has requested accelerated approval based on an endpoint of 6-minute walk
distance. The proposed indication states that 6-minute walk test is considered to be an
intermediate endpoint demonstrating delayed disease progression.
There is little in the NDA to explain the applicant’s thought process here. In Sarepta’s briefing
materials for the April 25, 2016 meeting of the Peripheral and Central Nervous System Drugs
Advisory Committee, they stated (page 16):
Office Director Decisional Memo – NDA 206488 – Page 6 of 42
Reference ID: 3959961

 “The accelerated approval pathway means that there will be an acceptable degree of
uncertainty about whether the therapy will actually result in the anticipated clinical
benefit. This uncertainty is addressed by the requirement that ‘appropriate post-approval
studies to verify and describe the predicted effect’ would usually be underway at the time
of approval.”
The applicant appears to misconstrue the intent of the accelerated approval pathway. They
purport to show that, after 36 months of treatment, eteplirsen improves physical performance as
assessed by the 6-minute walk test. We consider the 6-minute walk test to be a valid and
meaningful measure of how well a patient functions – i.e., a clinical endpoint that would be a
basis for full approval – not a surrogate endpoint or an intermediate endpoint. For slowly
progressive diseases, an intermediate clinical endpoint, a clinical endpoint that can be
measured earlier than an effect on irreversible morbidity or mortality and is considered
reasonably likely to predict the drug’s effect on irreversible morbidity or mortality or other clinical
benefit, can be used to support accelerated approval. But all would agree that showing an
improvement on a clinically meaningful endpoint at 36 months would be adequate to support a
conventional approval in DMD, a position we have taken with other DMD drugs.
Thus, the applicant has provided study results that purport to show improvement in a
meaningful clinical endpoint after a relatively long duration of treatment, but they appear to
propose accelerated approval as a means to deal with uncertainty about whether the therapy
has actually been shown to provide a clinical benefit in the trial.
Clearly, if the review team had reached the conclusion that the applicant had provided
substantial evidence of an effect on 6-minute walk distance during some 3 to 3.5 years of
treatment, they would recommend a conventional (full) approval, and not accelerated approval.
As noted in the reviews, however, for a number of reasons the review team does not believe
that the applicant has provided substantial evidence of an effect on 6-minute walk distance, or
any measure of physical performance (see below). Importantly, accelerated approval is not
intended to enable use of less than substantial evidence of a treatment effect as a basis for
approval, to be bolstered by more compelling evidence to be developed in the post-marketing
setting.
Despite the lack of substantial evidence of clinical efficacy from Study 201/202 (see below), it is
important to consider whether accelerated approval, based on an effect on a surrogate
endpoint, could provide a viable alternative pathway to approval. The relevant statutory and
regulatory framework (section 506(c) of the FD&C Act and 21 CFR part 314, subpart H) states
that a drug can receive accelerated approval if 3 factors are satisfied:
1. If the drug treats a serious or life-threatening disease or condition,
2. if FDA takes into account the severity, rarity, or prevalence of the condition and the
availability or lack of alternative treatments, and
3. if the drug demonstrates an effect on a surrogate endpoint that is reasonably likely to
predict clinical benefit OR demonstrates an effect on an intermediate clinical endpoint
that can be measured earlier than irreversible morbidity or mortality (IMM) that is
reasonably likely to predict an effect on IMM or other clinical benefit. As noted in section
506(c)(1)(B) of the FD&C Act, the evidence to support the concept “…that an endpoint is
reasonably likely to predict clinical benefit may include epidemiological,

Office Director Decisional Memo – NDA 206488 – Page 7 of 42
Reference ID: 3959961

 pathophysiological, therapeutic, pharmacologic, or other evidence developed using
biomarkers, for example, or other scientific methods or tools.”
In terms of the prospect for accelerated approval for eteplirsen, DMD is clearly a serious,
severe, and rare condition with no approved treatments; therefore, factors 1 and 2, above, are
satisfied.
The critical issue is whether factor 3 is satisfied, and factor 3 can be subdivided into three parts:
1) whether the surrogate endpoint is appropriate for the disease; 2) whether there is substantial
evidence of an effect on the surrogate endpoint; and 3) whether the effect demonstrated meets
the test of being “reasonably likely” to predict clinical benefit.
As noted in 506(f)(1), the amendments made by the Food and Drug Administration Safety and
Innovation Act (FDASIA) “…are intended to encourage the Secretary to utilize innovative and
flexible approaches to the assessment of products under accelerated approval for treatments for
patients with serious or life-threatening diseases or conditions and unmet medical needs.”
Some have interpreted this “flexibility” as a lower standard for the demonstration of
effectiveness, but this is not correct. Section 506(f)(2) of the FD&C Act specifically notes that
drugs granted accelerated approval must meet the same statutory standards for safety and
effectiveness as those granted traditional approval, notably the substantial evidence standard of
section 505(d) with respect to the drug’s claimed effect on a surrogate or intermediate endpoint.
These requirements have not been altered by FDASIA.
To be clear, 506(f)(2) states: “Nothing in this section shall be construed to alter the standards of
evidence under subsection (c) or (d) of section 505 (including the substantial evidence standard
in section 505(d)) of this Act or under section 351(a) of the Public Health Service Act. Such
sections and standards of evidence apply to the review and approval of products under this
section, including whether a product is safe and effective. Nothing in this section alters the
ability of the Secretary to rely on evidence that does not come from adequate and well
controlled investigations for the purpose of determining whether an endpoint is reasonably likely
to predict clinical benefit as described in subsection (b)(1)(B).”
Again, the critical issue here is whether factor 3 (above) is met, in light of these considerations.
For the first part of factor 3, whether the surrogate endpoint is appropriate for the disease, the
review team has agreed that the near-lack of dystrophin is the proximal cause of DMD, and that
the level of dystrophin in skeletal muscle is an appropriate surrogate endpoint that could predict
efficacy. (Of note, the best-case scenario for eteplirsen is the production of an abnormal
Becker-type dystrophin, not normal dystrophin, but that will be discussed later.)
The second part of factor 3 is whether an effect has been demonstrated, and the standard
remains ‘substantial evidence’ based on adequate and well-controlled clinical investigations.
Typically, such evidence would be two studies, both achieving a p-value < 0.05. 1

1

In some situations, FDA has the flexibility to interpret data from a single trial, or a single trial with supporting
evidence, as substantial evidence of effectiveness. See: “Guidance for Industry Providing Clinical Evidence of
Effectiveness for Human Drug and Biological Products;” May, 1998.

Office Director Decisional Memo – NDA 206488 – Page 8 of 42
Reference ID: 3959961

 The third part of factor 3, the determination that the demonstrated effect is “reasonably likely” to
predict clinical benefit, is a matter of judgment. Thus, once there is substantial evidence of a
treatment effect, the determination of whether the effect size is “reasonably likely” to predict
clinical benefit is an area where flexibility can be applied. Presumably there is some threshold
effect that would have to be achieved in order to satisfy this criterion, but this is not described in
the regulations.
Is There a Basis for Accelerated Approval: Production of Dystrophin Protein in Skeletal
Muscle?
The applicant assessed skipping of the messenger RNA exon using reverse transcriptase
polymerase chain reaction (RT-PCR), a standard laboratory technique to detect RNA
expression. Exon 51 skipping was confirmed by RT-PCR analysis in all patients treated with
eteplirsen, establishing proof of concept that eteplirsen can cause at least some degree of exon
51 skipping, as intended. Because PCR is a highly sensitive technique that can detect even a
few copies of messenger RNA, the findings do not support efficacy.
Dystrophin production was assessed by two widely-used and complementary methods:
immunofluorescence (immunohistochemistry) and Western blot. Immunofluorescence is
generally used to assess the presence or absence of proteins in tissue sections, and is
particularly useful for cellular localization of protein (by light microscopy). Western blot provides
quantitative analysis of protein, but no information on cellular localization.
Originally, the applicant evaluated the effect of eteplirsen on dystrophin expression in Studies
28, 33, and 201/202.
Of note however, as the May 26, 2016 goal date was approaching, the Office of New Drugs
(OND) and the Center for Drug Evaluation and Research (CDER) could not reach agreement on
the regulatory action for this NDA: the Office of New Drugs favored issuance of a complete
response whereas CDER favored approval.
Thus, in order to obtain definitive data on dystrophin production to support accelerated approval,
we requested that the applicant perform Western blot analyses of skeletal muscle biopsy
samples that had been obtained in the ongoing Study 301 (PROMOVI). The applicant was told
by CDER that if they were “….successful in showing, to FDA’s satisfaction, a meaningful
increase in dystrophin by Western blot analysis between the paired pre-and post-treatment
samples, we expect to be able to grant an accelerated approval….” Thus, data from Study 301
were included in this NDA and discussed below.
A. Immunohistochemistry
The applicant used immunohistochemistry in cross-sections of skeletal muscle biopsies to
distinguish and count “dystrophin-positive” and “dystrophin-negative” muscle fibers. The
methods are described in detail in Dr. Breder’s review. Briefly, following immunostaining of
tissue sections for dystrophin, 4 fields were manually selected from the 4 quadrants of each
slide, and images were captured (digitized) at 20X magnification. The contrast of each image
was manipulated to enhance background staining so that most of the muscle fibers became
visible, making it possible for the reader to perform a manual count of the total number of fibers.
Image contrast was returned to normal, and positive fibers – fibers with at least some degree of

Office Director Decisional Memo – NDA 206488 – Page 9 of 42
Reference ID: 3959961

 positive staining – were manually counted. For each field, the number of positive fibers was
divided by the total number of fibers to calculate the percentage of positive fibers. Various rules
were prospectively established to define “positive” fibers; in essence, a fiber could be classified
as “positive” if its staining intensity was only slightly perceptible over background. Importantly
therefore, a reading of 50% “positive” fibers in a tissue field is not tantamount to 50% (normal)
dystrophin. A 50% figure means only that half the fibers exhibited staining that was at least
barely perceptible over background.
Immunofluorescence data were also analyzed using Bioquant software. For these analyses, the
user determined a brightness threshold for each digitized image, in essence selecting all pixels
where staining intensity exceeded a particular user-selected value. Once selected, the software
calculated the mean intensity of the selected pixels. Given that the region of interest for these
analyses was limited to the pixels that exceeded a threshold rather than the total image, I do not
consider the Bioquant analyses to be readily interpretable.
Study 33 was a 7-patient, exploratory, phase 1 study, initiated in 2007 at the Hammersmith and
Saint Mary’s Hospitals, London, UK. Two subjects received a single 0.09-mg dose of eteplirsen
in the extensor digitorum brevis (EDB) muscle of one foot and placebo in the contralateral foot.
Five subjects received a single 0.9-mg dose of eteplirsen in the EDB muscle of one foot and
placebo in the contralateral foot. After 14 to 28 days, dystrophin was detected adjacent to the
needle tracks by immunohistochemistry and Western blot. Western blot analyses were not
carried out for control muscles injected with placebo.
Study 28 was a 19-patient, exploratory, phase 1 study, initiated in 2009 at 2 sites in the UK.
Patients had DMD amenable to exon 51 skipping. Eteplirsen was administered weekly by the
intravenous route for 12 weeks at doses ranging from 0.5 to 20 mg/kg. There were up to 4
patients per dose level. After FDA expressed concerns about the reliability of the procedures
and methods, the applicant responded that "Study 28 was an exploratory phase 1b study which
was only intended to generate proof of concept data to guide future studies. For this reason,
quality controls for the dystrophin data in Study 28 were not properly optimized." Some data
were missing, and after considering all of this information, the review team did not deem the
results to be interpretable.
Study 201 was a single-center, double-blind, placebo-controlled, parallel-dose study in 12
patients with DMD. The study was initiated in 2011. Patients were randomized (1:1:1) to
eteplirsen 30 mg/kg/week, eteplirsen 50 mg/kg/week, or placebo (n=4 per group). After 24
weeks, the 4 patients originally randomized to placebo were re-randomized to eteplirsen 30
mg/kg/week (n=2) or eteplirsen 50 mg/kg/week (n=2). The trial was eventually extended to an
open-label phase (Study 202) where all patients received eteplirsen, although investigators and
patients remained blinded to dose. The extension trial is well described in other reviews.
The 1° endpoint of Study 201 was the percentage of dystrophin-positive fibers in muscle
biopsies as assessed using immunohistochemistry. The main comparison was planned to be
the 50 mg/kg/week group at Week 12 and the 30 mg/kg/week group at Week 24 to the
combined placebo group. The applicant’s original results are shown in Table 2, adapted from
their clinical study report. As will be noted below, these results are not deemed to be reliable.

Office Director Decisional Memo – NDA 206488 – Page 10 of 42
Reference ID: 3959961

 Table 2: Adapted From Table 11-1 of Applicant’s Clinical Study Report: Effect of Eteplirsen on
Dystrophin-Positive Fibers Detected by Immunohistochemistry with MANDYS106
Placebo
N=4

Time point
Baseline

Mean
SD (SE)
Min, Max
Mean

On-Treatmentb

SD (SE)
Min, Max

Change from Baseline

Mean
SD (SE)
Min, Max

30 mg/kg/wk
Eteplirsen N = 4

50 mg/kg/wk
Eteplirsen N = 4

15.64

18.19

11.00

10.742 (5.371)

5.501 (2.751)

4.668 (2.334)

3.2, 28.2

11.9, 25.3

5.4, 15.6

11.59

41.14

11.79

7.130 (3.565)

10.097 (5.049)

4.456 (2.228)

5.7, 21.7

32.7, 54.3

6.4, 17.2

-4.05

0.79

5.834 (2.917)

22.95c
5.792 (2.896)

7.099 (3.549)

-8.5, 4.5

15.9, 29.0

-9.3, 7.4

It should be stressed again that the figures in the table represent the percentage of dystrophinpositive fibers, but in no way correspond to the percentage or quantity of dystrophin relative to a
normal individual. Muscle fibers displaying virtually any staining intensity above background
were considered “positive.” As noted above, therefore, a reading of 50% positive fibers means
only that 50% of fibers exhibited staining that was perceptively above background.
These results were substantially reported in a 2013 publication, 2 which claimed that eteplirsen
markedly increased functional dystrophin production: “…the percentage of dystrophin-positive
fibers was increased to 23% of normal; no increases were detected in placebo-treated patients
(p≤0.002). Even greater increases occurred at week 48 (52% and 43% in the 30 and 50 mg/kg
cohorts, respectively), suggesting that dystrophin increases with longer treatment. Restoration
of functional dystrophin was confirmed by detection of sarcoglycans and neuronal nitric oxide
synthase at the sarcolemma.” The publication also stated that dystrophin expression was
confirmed by Western blot, with a figure showing what were termed “representative” results.
Publication of this paper was followed by a Sarepta press release, 3 which also claimed a
remarkable treatment effect from eteplirsen and raised wildly unrealistic expectations in the
DMD community. It was these perceptions and expectations that led the applicant to declare
that a placebo-controlled study was no longer feasible (see below).

2

Mendell JR, et al: Eteplirsen for the treatment of Duchenne muscular dystrophy. Ann Neurol 2013;74:637-47
Sarepta press release, 8/8/13, at http://investorrelations.sarepta.com/phoenix.zhtml?c=64231&p=irolnewsArticle&ID=1846052 [downloaded 5/10/16]
3

Office Director Decisional Memo – NDA 206488 – Page 11 of 42
Reference ID: 3959961

 The original data from Nationwide Children’s Hospital submitted to FDA are plotted in Figure 2.
Immunostaining for dystrophin appears to increase markedly in all 4 groups with time, with
some 50 to 60% of fibers staining positive for dystrophin at 48 weeks. For reasons explained
below, the review team disagrees with the veracity of these data.
Figure 2: Original Results of Dystrophin Immunostaining Using MANDYS106 Antibody: Percent
Positive Fibers as a Function of Time – Results Not Verified on Re-read

I was part of an inspection team that conducted (May 29 and 30, 2014) a site visit to Nationwide
Children’s Hospital in Columbus, OH, where Study 201 was conducted. We found the analytical
procedures to be typical of an academic research center, seemingly appropriate for what was
simply an exploratory phase 1/2 study, but not suitable for an adequate and well controlled
study aimed to serve as the basis for a regulatory action. The procedures and controls that one
would expect to see in support of a phase 3 registrational trial were not in evidence.
Although the technician had been blinded to treatment group, access to the treatment code was
not protected with the kinds of safeguards and firewalls that one would ordinarily put in place for
an adequate and well controlled trial. The immunohistochemistry images were only faintly
stained, and had been read by a single technician using an older liquid crystal display (LCD)
computer monitor in a windowed room where lighting was not controlled. (The technician had to
suspend reading around mid-day, when brighter light began to fill the room and reading became
impossible.) These issues are well described in a summary of inspectional findings in Dr.
Breder’s clinical review (page 27). There was also concern that the reader, although masked to
treatment assignment, was not masked to sequence/time (see below). Importantly, in a trial
where all patients eventually received the active drug, knowledge of sequence could lead to the
false appearance of a treatment effect, i.e., the appearance of increasing dystrophin expression
Office Director Decisional Memo – NDA 206488 – Page 12 of 42
Reference ID: 3959961

 with time, simply by having a lower threshold for calling fibers “positive” at later time points in
the study.
Having uncovered numerous technical and operational shortcomings in Columbus, our team
worked collaboratively with the applicant to develop improved methods for a reassessment of
the stored images. We suggested a re-read of all images by 3 independent masked readers,
such that blinding could be assured and inter- and intra-observer variability could be
characterized. We also suggested the use of better equipment, specifically, high-quality lightemitting diode (LED) computer monitors, in darkened rooms.
The applicant undertook a blinded re-analysis of the images on the server as FDA suggested.
Unfortunately, the re-analyses failed to show a significant increase in dystrophin-positive fiber
counts in eteplirsen treated patients (Figure 3). Note also that for patients who switched from
placebo to eteplirsen at Week 24 (dashed red and black lines), there was no response between
Weeks 24 and 48.
Figure 3: Blinded Re-read of Dystrophin Immunostaining Using MANDYS106 Antibody:
Results through Week 180 – Percent Positive Fibers as a Function of Time

This re-analysis, along with the study published in 2013,2 provides an instructive example of an
investigation with extraordinary results that could not be verified. The publication, now known to
be misleading, should probably be retracted by its authors.

Office Director Decisional Memo – NDA 206488 – Page 13 of 42
Reference ID: 3959961

 Figure 4 shows the correlation between the dystrophin immunohistochemistry data as read by
the technician at Nationwide Children’s Hospital and the 3 blinded pathologists. Each point
represents data from a single patient at a single time point (an analysis of 24 images), as read
by Nationwide Children’s Hospital (y-axis) and the group of 3 blinded pathologists (x-axis).
Readings from the 3 pathologists are averaged. Perfectly correlated readings would lie along
the blue line of unity. In most cases, the reading from Nationwide exceeds the reading from the
pathologists, i.e., above and to the left. Thus, despite less-than-optimal lighting conditions that
should have favored reduced reading of positive fiber counts at Nationwide Children’s Hospital,
there was a striking tendency for the reporting of higher counts at that institution.
Figure 4: Comparison of Positive Fiber Counts at Nationwide Children’s
Hospital to Re-read of Fiber Counts by 3 Independent, Masked Pathologists

One might reasonably ask why the original readings were not reproduced by a blinded re-read.
Figure 5 shows the same scatterplot between readings by Nationwide Children’s Hospital and
the 3 blinded pathologists. In this display, however, readings from samples obtained at the
disparate time points are shown with unique markers.
It is striking that the deviations between the readings of Nationwide and the re-read by the
blinded pathologists differ substantially by study time point. Thus, at Week 1 (●) and Week 12
(▲), time points before increased dystrophin production would be expected, there is reasonable
agreement between Nationwide and the pathologists, i.e., the points lie close to the blue line. In
contrast, for the Week 24.5 time point (+), readings from Nationwide Children’s Hospital are
much higher than those of the 3 pathologists, suggesting that blinding to sequence (i.e., time
Office Director Decisional Memo – NDA 206488 – Page 14 of 42
Reference ID: 3959961

 point) was not achieved. At the time the Week 180 samples were read at Nationwide Children’s
Hospital, the technician was aware that the images would be re-read by 3 pathologists, which
could explain why there is less exaggeration (i.e., the Week 180 readings are closer to the blue
line of unity than the Week 24.5 readings).
Figure 5: Comparison of Positive Fiber Counts at Nationwide Children’s Hospital to Reread of Fiber Counts by 3 Independent, Masked Pathologists: Apparent Interaction with
Time

Week 180 Data
As noted by the review team, the extension phase of the study (Study 202) has continued
through the present. Eleven (11) of the 12 patients consented to undergo a fourth skeletal
muscle biopsy at Week 180 (3.5 years), and these samples were analyzed using
immunohistochemistry.
Prior to the analysis of the Week 180 samples, however, the applicant made changes to the
immunohistochemistry protocol with the intent of decreasing non-specific staining. Their aim
was to compare the Week 180 dystrophin level to baseline for each patient. Frozen archived
baseline tissue was available for only 3 of the patients, however, and so the applicant
supplemented these with samples from 6 untreated external DMD patients, all to be compared
to the Study 201/202 patients at Week 180. Images were read by the same 3 pathologists,
masked to treatment group.

Office Director Decisional Memo – NDA 206488 – Page 15 of 42
Reference ID: 3959961

 For this analyses, the applicant claims a remarkable increase in dystrophin staining: the 9
baseline samples (including samples from 3 patients in Study 201/202 and 6 external controls)
showed a mean percent positive fiber count of 1.1 ± 1.3% (mean ± SD), whereas the Week 180
samples showed a mean percent positive fiber count of 17.4 ± 10.0%. I will note that FDA
made no attempt to inspect or oversee the new immunohistochemistry methods.
Given that the original baseline percent positive fiber count for patients from Study 201/202 was
13.0 ± 6.2%, it would be important to understand why the results from a new
immunohistochemistry protocol provided results more than an order of magnitude lower (1.1 ±
1.3%).
As noted above, there were 3 patients in Study 201/202 with adequate archived tissue for
separate immunohistochemistry analyses using both the old and new methods. Figure 6 shows
how the two methods compare. These are essentially replicate analyses of a single tissue
sample using the two methods.
There is an inexplicable
difference of more than an
order of magnitude between
results using the new and old
immunohistochemistry
protocols. These marked
differences raise concerns with
respect to the validity of the
applicant’s methods, and
make interpretation
impossible.

Figure 6: Comparison of Results from the New and Old
Immunohistochemistry Protocols – Lack of Agreement for 3
Patients in Study 201/202

The disparity also underscores
the difficulty of comparing
results of
immunohistochemical
analyses for dystrophin across
laboratories, or, for that matter,
within the same laboratory.
Commentary:
The review team provided much thoughtful discussion regarding the relative merit of
immunohistochemistry for the quantitative assessment of dystrophin in skeletal muscle. My
view is that such analyses, if properly blinded and controlled, can yield semi-quantitative
information that could show differences in dystrophin production, e.g., 50% is more than 25%,
although the method does not allow correlation of particular values of “percent positive” fiber
counts with quantitative measures of muscle protein. Moreover, comparisons of fiber counts
across centers, across experiments, or, for that matter, across staining or reading runs within a
single laboratory, do not seem likely to be informative.

Office Director Decisional Memo – NDA 206488 – Page 16 of 42
Reference ID: 3959961

 Recognizing that Study 201/202 was a small exploratory phase 1/2 study that was not powered
to show a small change in dystrophin, the study provides no evidence of increased dystrophin
production by immunohistochemistry.
It is unfortunate that the original readings from Nationwide Children’s Hospital, purporting to
show a marked effect of eteplirsen on dystrophin-positive fiber counts – counts now known to be
unreliable – led to the perception that the drug produces large amounts of dystrophin. These
results fueled the public perception that eteplirsen is highly effective as well as the DMD
community’s reluctance to participate in placebo-controlled trials. Only recently, an unauthored
report in the Wall Street Journal stated: “The trial turned up evidence that eteplirsen makes
good on pumping out dystrophin, a feat no treatment has managed.” 4 Presumably this
misperception has been carried over from the initial 2013 reports.
B. Western blot
1) Data analyzed prior to the PDUFA goal date
A second, more important line of evidence regarding dystrophin production is Western blot, a
standard, widely-used, analytical technique to assess levels of protein in biological tissues.
Western blot was used to quantify dystrophin protein directly, and the methods are described by
others.
For a variety of reasons discussed by Dr. Rao, the Western blot analyses originally conducted
by the applicant were technically unsatisfactory. The Western blots from the first 3 time points
had oversaturated bands, lacked appropriate controls, and were essentially uninterpretable.
After conducting a site visit to the Columbus OH laboratory, FDA rendered advice to the
applicant with the goal of improving technical aspects of the assay for future use.
The applicant amended the study protocol to allow for an additional skeletal muscle biopsy at
Week 180 (3.5 years), potentially enabling pre- to post-treatment comparisons of Becker-type
dystrophin after prolonged eteplirsen treatment. As noted above, 11 of the 12 patients in Study
201/202 consented to undergo a fourth skeletal muscle biopsy at Week 180. Of note, the
baseline samples had been obtained from biceps muscle, whereas the Week 180 samples were
obtained from deltoid muscle.
Two blocks were prepared from each patient sample. Sections from both blocks were pooled
during homogenization for lysate preparation, and Western blots were run in duplicate.
The individual (anonymized) values for the Western blot analysis are shown in Table 3. As
reported by the review team, the analysis for 11 of the 12 original patients showed a mean
dystrophin value of 0.93% ± 0.84% of normal (mean ± standard deviation) after 3 to 3.5 years of
eteplirsen treatment (3 years in patients initially randomized to placebo; 3.5 years in the other
patients). Mean values were virtually the same for the lower (30 mg/kg/week) and higher (50
mg/kg/week) dose groups; there is no suggestion of a dose-response.

4

A Legal Test for the FDA: Black letter law dictates approval for a muscular dystrophy drug; Wall Street Journal,
May 9, 2016.

Office Director Decisional Memo – NDA 206488 – Page 17 of 42
Reference ID: 3959961

 Of note, the Western blot values are quite variable, both between patients and between
duplicate runs within patients (i.e., repeatability; intra-assay precision), Table 3.
Mean values ranged from a maximum of 2.47% in Patient J, to near-zero in Patient H, and to
zero in 2 patients (E and G). For some patients, there were considerable discrepancies
between duplicate runs (the intra-assay difference was >0.5% in Patients B, C, D, and J). Aside
from patients with zero or near-zero dystrophin, only 3 patients showed reasonable intra-assay
agreement: Patients F, L, and K.
Given that these numbers represent duplicate runs from tissue homogenates, intra-assay
differences suggest limited precision/reproducibility of the method, heterogeneity of the
samples, or both.
Table 3: Individual Week 180 Western Blot Analyses – Study 201/202
Subject

Dose

Group Mean ± SD

Western blot
gel 1

gel 2

Mean
Mean
(arithmetic) (per protocol)

L
K
J
H
G
F

30 mg/kg
30 mg/kg
30 mg/kg
30 mg/kg
Placebo to 30 mg/kg
Placebo to 30 mg/kg

0.58
1.45
2.83
0.02*
0.17*
0.93

0.46
1.78
2.11
0.28
0.15*
1.02

0.52
1.62
2.47
0.15
0.16
0.98

0.52
1.62
2.47
0.14
0
0.98

0.96 ± 0.95

E
D
C
B
A

50 mg/kg
50 mg/kg
50 mg/kg
50 mg/kg
Placebo to 50 mg/kg

0.19*
0.75
1.22
2.43

0.16*
0.24*
0.69
1.67
1.15

0.18
0.50
0.96
2.05
1.15

0
0.38
0.96
2.05
1.15

0.91 ± 0.79

* below limit of quantitation

Change in Dystrophin with Treatment:
The critical question, of course, is whether the value of 0.93% is meaningfully greater than the
value at baseline, or even meaningfully greater than zero. Assuming that one considers this
value greater than zero, the baseline pre-treatment levels of dystrophin in these 11 patients are
critical in determining whether eteplirsen was responsible for the dystrophin detected at Week
180.
Unfortunately, adequate pre-treatment tissue samples were available for only 3 of these 11
patients. Thus, the applicant supplemented these data with muscle biopsies from 6 untreated
patients with DMD amenable to exon 51 skipping who were external to the study.
Whereas the Week 180 samples were obtained from deltoid muscle, 8 of 9 of the controls were
obtained from biceps muscle (the other one was obtained from deltoid). As noted above, the
non-clinical review team found “…wide variability in the extent of PMO-induced dystrophin
expression within a single muscle and among different muscles, suggesting that caution is

Office Director Decisional Memo – NDA 206488 – Page 18 of 42
Reference ID: 3959961

 warranted in generalizing from the results of biopsies taken from only one or a few sites, muscle
types, or patients.” Use of disparate muscle groups between patients in Study 201/202 and
controls was, obviously, ill advised. The finding of a difference between patients in Study
201/202 and the external controls could simply represent a difference between muscles.
FDA’s advice to the applicant (March 30, 2015) is still germane: "The control biopsy tissue that
you propose to use is from a number of different muscle groups, such that differences that may
exist in dystrophin expression among muscle groups may affect your results. However, in the
context of other major sources of variability among biopsies (including both intra- and interindividual differences even within the same muscle group), it appears reasonable for you to
proceed with these controls, with the understanding that dystrophin changes would need to be
robust to be interpretable as a drug effect."
Averaging Western blot data from pre-treatment biopsies of the 2 patients from Study 201/201
and the external treatment-naïve patients, the applicant reported a baseline dystrophin value of
0.08% ± 0.13% (mean ± standard deviation). Obviously, all but 2 of these controls are external,
such that the comparison to the treated patients in Study 201/202 is non-randomized and
indirect.
Table 4: Individual Untreated DMD Control Samples, Western Blot Analysis (% Normal
Dystrophin)
Study; Subject

Dose

Group
Mean ± SD

Western blot
gel 1

gel 2

Mean
Mean
(arithmetic) (per protocol)

201/202; X
201/202; A
201/202; B

0
0
0

0.05*
0.19*
0.13*

0.07*
0.08*
0.07*

0.06*
0.14*
0.10*

0
0
0

0±0

external;
external;
external;
external;
external;
external;

0
0
0
0
0
0

0.12*
0.03*
0.37
0.04*
0.20*
0.40

0.14*
0.12*
failed
0.30
failed
0.09*

0.13*
0.08*
0.37
0.17*
0.20*
0.25*

0
0
0.37
0.15
0
0.20

0.12 ± 0.15

A
B
C
D
E
F

All Mean ±
SD

0.08 ± 0.13

* below limit of quantitation

In determining whether there is substantial evidence that eteplirsen produced dystrophin in the
patients in Study 201/202, the critical questions are whether these values, near the lower limit of
quantification of the assay, are actually interpretable, and whether the comparison between
these subjects and a predominantly external group of untreated patients is valid.
The review team has pointed out important limitations with respect to comparability of the
Western blot results from the untreated controls, summarized below:

Office Director Decisional Memo – NDA 206488 – Page 19 of 42
Reference ID: 3959961

 •

Biopsies from controls were obtained from biceps, whereas Week 180 biopsies from
eteplirsen-treated patients were obtained from deltoid. There is some evidence that
dystrophin concentrations differ by muscle group, and the study does not account for this
possibility. Because the study is not well controlled, the difference between these groups of
patients cannot be attributed to a drug effect.

•

Two-thirds (6 of 9) of the control patients were from Study 301, and were external to study
201/202. There is no way to know how these particular patients were selected for the
purpose of this comparison.

•

Degradation of dystrophin or loss of immunoreactivity might occur during prolonged storage
of tissue samples. If so, it could have affected the baseline samples from the 3 patients in
Study 201/202, which were frozen for over 3 years prior to analysis. Note that the data are
consistent with loss in immunoreactivity over time (Table 4). The per-protocol values for all
3 patients from Study 201/202 whose samples were stored for 3 years are 0 (top), whereas
3 of 6 of the samples from the external controls (bottom) are greater than zero. Although
the numbers of samples are small and the comparison is non-randomized, the data
nevertheless support the concept that immunoreactive dystrophin decreases during storage.

For these reasons, the review team questioned the comparability of these two groups of
patients, and I agree. Having compared samples from different muscle groups in independent
groups of patients, the study was not adequate and well controlled; therefore, the validity of the
comparison is uncertain. The data provide little confidence that the study was designed well
enough so as to be able “to distinguish the effect of a drug from other influences, such as
spontaneous change…, placebo effect, or biased observation” (§314.126).
Having heard arguments and opinions from both the applicant and the review team, the
Advisory Committee, despite extraordinary public activism and pressure to vote favorably, voted
7 to 6 that the applicant had not provided substantial evidence from adequate and well
controlled studies that eteplirsen induces production of dystrophin to a level that is reasonably
likely to predict clinical benefit. Moreover, 2 of the Committee members who voted “yes” were
patient representatives.
Correlation between the applicant’s two methods to assess dystrophin
The discussion of the Week 180 dystrophin analyses would not be complete without a
comparison of the results of the two complementary methods used by the applicant. Of note,
the improved immunohistochemistry analyses and Western blot analyses were performed on
the same blocks of tissue, and one should expect a reasonable correlation between the two
methods if in fact the data are reliable.
Of note, there is a striking lack of correlation between these two methods of dystrophin
assessment (Figure 7). It is simply not possible to determine whether the
immunohistochemistry methods are inaccurate, whether the Western blot methods are
inaccurate, or whether both methods are inaccurate. My view is that is it not possible to render
a positive regulatory decision on the basis of unreliable data from these 11 patients. Internal
consistency is lacking.

Office Director Decisional Memo – NDA 206488 – Page 20 of 42
Reference ID: 3959961

 2)
Data analyzed after the
PDUFA goal date
As noted above, as the May 26,
2016 goal date was approaching,
OND and CDER could not reach
agreement on the regulatory action
for this NDA.

Figure 7: Study 201/202 Week 180 Dystrophin Assessment
– Lack of Agreement between Immunohistochemistry and
Western Blot

In order to gain additional
information that might provide
evidence of an effect on a surrogate
marker that was reasonably likely to
predict clinical benefit, we requested
that the applicant perform Western
blot analyses of skeletal muscle
biopsy samples that had been
obtained in an ongoing study (Study
301 [PROMOVI]). These samples
were originally planned to be
analyzed at the end of the study;
however, we requested an interim analyses of a subset of samples. As described by Drs. Rao,
Farkas, and Bastings, Western blot analyses were performed on paired biceps samples from 13
of the patients. For each of these patients, samples had been obtained at baseline (prior to
treatment) and at Week 48, after 48 weekly infusions of eteplirsen 30 mg/kg.
The age of these 13 patients ranged from 7 to 13 years. Paired pre- and post-treatment
samples were run in side-by-side lanes on the gels, and each gel was run in duplicate. A
muscle sample from a healthy 14 year-old boy with no pathologic diagnosis served as the
reference sample; values from the DMD patients were reported as percent of normal.
Dr. Ashutosh Rao from the Office of Biotechnology Products reviewed the methodology and the
technical reliability of the Western blot assay. Dr. Rao also conducted an inspection with Young
Moon Choi, Ph.D. (Office of Study Integrity and Surveillance) and Mark Babbit (Office of
Regulatory Affairs) as the analyses were being run. Xiang Ling, Ph.D., from the Office of
Biostatistics, performed the statistical review on the data.
According to the protocol, acceptance of the result from each gel was contingent on two factors:
1) the R2 value for the linearity of the standard curve of the normal control had to be > 0.9; and
2) the dystrophin band for the negative control DMD sample on the gel had to have a density
lower than the lowest sample of the standard curve (0.25%). Samples that did not meet both
criteria were deemed ‘failed’ and were not considered in the analyses. As it turned out, 22 of
the 52 gels (42%) failed, such that many of the values represent single readings rather than the
average of two. There was one patient for whom none of the values met acceptance criteria.
Thus, the applicant reported pre- and post-treatment data for 12 of the 13 patients.
The applicant used 3 methods to consider values below the 0.25% lower limit of quantification:
1) consider such values to be zero; 2) analyze such values as actually reported; and 3) consider
such values to be 0.24%.

Office Director Decisional Memo – NDA 206488 – Page 21 of 42
Reference ID: 3959961

 The review team believes the most appropriate analysis is the second: analysis of all values as
reported, but the results were similar for all 3 methods.
Reporting values below the limit of quantification as 0, pre- and post-treatment values are
0.06% ± 0.14% and 0.38% ± 0.50%, respectively (mean ± standard deviation), p<0.05. For the
‘as reported’ analysis, pre- and post-treatment values are 0.16% ± 0.12% and 0.44% ± 0.43%,
respectively, p<0.05. Reporting all values below the limit of quantification as 0.24%, pre- and
post-treatment values are 0.26% ± 0.05% and 0.48% ± 0.41%, respectively, p<0.05. Individual
data for the ‘as reported’ analysis are shown in Table 5, adapted from listing 1.1 of the
applicant’s “Preliminary Report: Western Blot Interim Analysis of Novel Dystrophin Expression in
Muscle Biopsy Samples from Week 48 of the Clinical Study 4658-301,” submitted June 27,
2016.
Irrespective of the method used to express data below the limit of quantification, the mean
change is similar, ranging from 0.22% to 0.32% of normal, a treatment effect of approximately 2
to 3 parts per thousand.
Table 5: Study 301: Pre- and Post-treatment Values of Becker-Type Dystrophin
Patient

Time

Baseline
1
Week 48
Baseline
2
Week 48
Baseline
3
Week 48
Baseline
4
Week 48
Baseline
5
Week 48
Baseline
6
Week 48
Baseline
7
Week 48

status

value (%)

pass
pass
pass
pass
pass
fail
pass
fail
pass
pass
pass
pass
pass
fail
pass
fail
fail
pass
fail
pass
pass
fail
pass
fail
fail
pass
fail
pass

0.15
0.11
0.22
0.29
0.35
0.26
0.36
0.12
0.06
0.06
0.5
0.24
0.04
0.06
0.1
0.19
0.1
0.17
0.92
1.02
0.37
0.46
0.3
0.29
0.04
0.17
0.22
0.42

mean (%) delta (%)

Patient

0.13

Time

Baseline
0.13

8

0.26

Week 48

0.35

Baseline
0.01

9

0.36

Week 48

0.06

Baseline
0.31

10

0.37

Week 48

0.04

Baseline
0.06

11

0.1

Week 48

0.17

Baseline
0.85

12

1.02

Week 48

0.37

Baseline
-0.07

0.3

13
Week 48

status

value (%)

fail
fail
fail
fail
fail
pass
fail
pass
pass
fail
pass
fail
pass
pass
pass
pass
pass
fail
pass
fail
fail
pass
fail
pass

0.08
0.14
0.08
0.05
0.14
0.24
1.17
1.57
0.11
0.05
0.12
0.11
0.01
0.08
0.31
0.63
0.02
0
0.09
0.01
0.34
0.18
0.34
0.21

mean (%) delta (%)

0.17
0.25
0.42

Office Director Decisional Memo – NDA 206488 – Page 22 of 42
Reference ID: 3959961

0.24
1.33
1.57
0.11
0.01
0.12
0.05
0.43
0.47
0.02
0.07
0.09
0.18
0.03
0.21

 The distribution of these changes is shown graphically in Figure 8. Of these 12 patients, 8 (twothirds) had a change of 0.25% or less; only 1 patient (8%) had a treatment effect greater than
1%.
Figure 8: Study 301: Distribution of Changes in Becker-type Dystrophin in 12 Patients

N=8

N=2

N=1

N=1

Commentary: Study 301 was a baseline-controlled study, where each patient served as his own
control: pre- and post-treatment biopsies were obtained from the same muscle and Western blot
analyses were run concurrently. An FDA inspection team observed the performance of the
assays and considers the results to be reliable. Thus, unlike the data obtained from Study
201/202, the Study 301 data are considered by the review team to have been generated from
an adequate and well controlled study. Study 301 provides substantial evidence of an effect of
the surrogate endpoint – Becker-type dystrophin.
The critical question is whether the quantity of dystrophin produced here – a mean of 2 to 3
parts per thousand – is reasonably likely to predict clinical benefit.
With levels of Becker-type dystrophin higher in Study 201/202 (at Week 180) than in Study 301
(at Week 48), the applicant speculates that there is greater dystrophin accumulation with longer
durations of treatment. These differences, however, could also be due to cross-laboratory
methodological differences or play of chance; therefore, such an interpretation is highly
speculative.

Office Director Decisional Memo – NDA 206488 – Page 23 of 42
Reference ID: 3959961

 The Question of “Reasonably Likely to Predict Clinical Benefit”
As discussed above, the accelerated approval of eteplirsen hinges on: 1) whether Becker-type
dystrophin is an appropriate surrogate endpoint for the disease; 2) whether there is substantial
evidence that eteplirsen produces Becker-type dystrophin in skeletal muscle, and 3) whether
such dystrophin produced is reasonably likely to predict clinical benefit, i.e., whether it is
functional, and whether the quantity produced is adequate.
1.

Is dystrophin an appropriate surrogate endpoint for Duchenne muscular dystrophy?

The review team believes that dystrophin is on the causal pathway of the disease, and there is
no debate about the appropriateness of dystrophin as a surrogate endpoint for Duchenne
muscular dystrophy.
2.

Is there substantial evidence that eteplirsen produces dystrophin in skeletal muscle?

Prior to receiving the new Western blot data from Study 301 on June 27, 2016, the review team
did not believe that substantial evidence from adequate and well controlled trials had been
submitted to support an accelerated approval.
Study 201/202: Immunohistochemistry analyses were performed to assess and compare
percent dystrophin-positive fibers at various time points before and during treatment. This is a
standard technique that has been used by many laboratories for decades to assess dystrophin
levels in DMD and Becker’s patients. Importantly, the analysis showed no evidence of
dystrophin production through 48 weeks of treatment with eteplirsen. This information is
particularly germane, because, unlike the Western blot analyses from Study 201/202, the
immunohistochemistry analyses are adequately controlled. The lack of a positive finding from
the blinded re-read of the immunohistochemistry data with proper controls undercuts the
evidence of dystrophin production from Western blot analyses.
The applicant supplemented these data with new analyses from Week 180 that purport to show
a remarkable increase in dystrophin from pre-treatment levels. Unfortunately, an altered
immunostaining protocol was used, and there was an inexplicable difference of more than a log
between results from the new and old protocols, rendering interpretation impossible.
The Western blot data from Study 201/202 were largely externally controlled, and there were
questions with respect to the proper selection of control patients, differences in the specific
muscles analyzed, and concerns regarding the possible degradation of immunoreactive
dystrophin in tissue samples that might occur during long-term storage and lead to a falsepositive result. Importantly, ignoring the baseline data and focusing only on the Week 180
samples, there is a striking lack of correlation between the immunohistochemistry data and the
Western blot data, i.e., there is no internal consistency. Thus, these data provide no basis to
believe that the study was adequate and well controlled.
Study 301: The new data submitted on June 27, 2016 were obtained from an adequate and
well controlled study. This baseline-controlled study shows a statistically significant increase in
Becker-type dystrophin with treatment, the surrogate endpoint. Thus, there are now data
showing Becker-type dystrophin production, albeit at a small level, from one adequate and well
controlled trial (Study 301), with inconclusive data from Study 201/202.

Office Director Decisional Memo – NDA 206488 – Page 24 of 42
Reference ID: 3959961

 The question of “reasonably likely” is, therefore, an issue of the quantity of protein produced. As
noted above, Study 301 showed a treatment effect of 2 to 3 parts per thousand in Becker-type
dystrophin after 48 weeks. Study 201/202, although not adequate and well controlled,
nevertheless suggested a treatment effect of 8 to 9 parts per thousand after 3.5 years.
3.
Is the dystrophin that was produced reasonably likely to predict clinical benefit, i.e., is it
functional, and is the quantity adequate?
These two uncertainties, protein function and protein quantity, are separate issues that must be
considered in series. The function of the Becker-type dystrophin detected in Study 301 cannot
be assessed. Function is therefore a matter of judgment for which regulatory flexibility can be
extended. The review team has been willing to assume that whatever Becker-type dystrophin is
produced would function as well as in the Becker form of the disease. Although there can be no
certainty on this point, the uncertainty is small relative to the uncertainty regarding the adequacy
of the quantity, and so function is less germane to the question of “reasonably likely.” In short, it
is the quantity of Becker-type dystrophin produced that is central to the question of ‘reasonably
likely,’ and central to the approvability of this NDA under accelerated approval.
It must be stated that the minimum level of Becker-type dystrophin that is reasonably likely to
predict clinical benefit in patients with DMD is unknown. The raw data are shown in Figure 9,
but this is an area where we must consider what is known about the disease and apply medical
judgment.
There are two ways to consider the quantity of Becker-type dystrophin produced: as a binary
responder analysis, and as a mean response. The former has the advantage of considering the
possibility that some patients may respond to the treatment whereas others do not; the latter
does not allow for this type of consideration.
The problem with a responder analysis is that there is no rational basis upon which to define a
threshold for a ‘response.’ Various cut-points could be selected, but their selection would be
arbitrary, and the particular threshold chosen would have a major influence on the effect size.
Drs. Farkas and Bastings have tried to provide a framework to help put these small increases
into perspective. The applicant’s data show that dystrophin levels in treatment-naïve DMD
patients range from 0 to approximately 0.4% by Western blot; the applicant has not detected
values > 0.4% in treatment-naïve patients.
DMD experts, including those involved with the development of eteplirsen, have stated that
levels < 3% are generally associated with the typical DMD phenotype, and no patient has been
found to have or produce a level of Becker-type dystrophin > 3% in response to treatment.
In order to place these small quantities of Becker-type dystrophin into a clinical perspective,
many have focused on publications from a number of laboratories that attempt to relate
particular levels of dystrophin protein to clinical course, e.g., maintenance of physical function,
age at loss of ambulation. Some have also cited non-clinical data to relate dystrophin levels to
maintenance of physical function. It is important to recognize, however, that many
methodological factors affect the results of these assays, and comparison of values across
various laboratories could lead to erroneous conclusions.

Office Director Decisional Memo – NDA 206488 – Page 25 of 42
Reference ID: 3959961

 Figure 9: Studies 301 and 201/202: Expression of Becker-type Dystrophin by Western Blot

Van den Bergen et al studied the relation between dystrophin levels (quantified by Western blot)
and clinical severity in 33 patients with Becker muscular dystrophy (van den Bergen JC, et al. J
Neurol Neurosurg Psychiatry 2014;85:747). Although the authors did not find a linear
relationship between dystrophin levels and disease severity, all 4 of their patients with
dystrophin levels <10% showed low muscle strength and early symptom onset. As discussed
by the review team, DMD experts have proposed that “induction of approximately 10% of
normal dystrophin levels sets a minimum level to confer measurable clinical benefit.”
Chamberlain, who stated at the open public session at the advisory committee meeting that very
low levels of dystrophin may be beneficial, discussed in a published paper (Basic Appl Myol. 7
[3&4]: 251, 1997) that “…a majority of fibers must accumulate approximately 20% of wild-type
levels of dystrophin for a significant correction of the muscle pathology,” a view that seemingly
contradicts the comments he made at the advisory committee meeting.
Anthony K et al (Neurology 2014;83;2062) compared results of Western blot analyses from 6
patients (3 with DMD; 3 with Becker Muscular Dystrophy) across 5 experienced laboratories,
and found a high degree of variability. Only one of the 5 laboratories had a coefficient of
variation (CV = SD/mean X 100) below 0.3%. Variability was particularly pronounced with low
levels of dystrophin.

Office Director Decisional Memo – NDA 206488 – Page 26 of 42
Reference ID: 3959961

 During their presentation at the April 25, 2016 meeting of the Peripheral and Central Nervous
System Drugs Advisory Committee, Dr. Kaye, a pediatric neurologist and interim Chief
Executive Officer of Sarepta, stated:
“Our validated Western blot method, optimized to detect low levels of dystrophin, is
arguably the first dystrophin Western blot to be truly quantitative. This was achieved by
use of a 5 point calibration curve on each gel and prespecified loading and exposure
limits to avoid signal saturation. Furthermore, samples were randomized, blinded and
run in duplicate on separate gels. In contrast, the Western blot methods in the majority of
historical publications referenced by FDA were performed using older methodology that
is semi-quantitative at best. Given these significant methodological differences, it is
inappropriate to compare our data to literature approximations.” (Source: official
transcript of the meeting; underlining for emphasis.)
It appears, therefore, that reproducibility of assays among academic centers has not been
established, such that it would not be feasible to compare an increase in Becker-type dystrophin
of 0.2 to 0.3% (or even far greater increases) with dystrophin values cited in the literature for
other mutations/patient populations, assessed by other laboratories.
Do the clinical data bolster the question of “reasonably likely?”
The applicant collected data on both dystrophin production and physical performance in Study
201/202. Such data have the potential to support the concept that the dystrophin level predicts
clinical response, and would support the ‘reasonably likely’ premise. Despite detailed
testimonials from patients in Study 201/202 claiming improvements in clinical performance, the
Division concluded, on the basis of the data presented in the NDA, that no patient in Study
201/202 clearly deviated from the natural history of the disease. They reasoned, therefore, that
whatever the quantity of Becker-type dystrophin detected, it did not predict clinical benefit. Dr.
Bastings opines that the clinical data weaken, and do not strengthen, the “reasonably likely”
argument.
Within Study 201/202, it is also reasonable to consider the correlation between the quantity of
dystrophin detected and maintenance of physical function in individual patients. The presence
of a correlation would help support the “reasonably likely” question.
For the 9 patients who remained ambulatory at Week 180 and agreed to undergo a fourth
muscle biopsy, Figure 10 shows little correlation between the quantity of dystrophin detected (xaxis) and preservation of physical function as assessed by the change in 6-minute walk distance
from baseline (y-axis) after weekly infusions of eteplirsen for 3 to 3.5 years. For the 4 patients
whose 6-minute walk performance was best preserved (red arrows), 2 had the highest
dystrophin levels detected in the study, but 2 had levels that were close to zero. Importantly,
therefore, these data do not show a quantitative correlation between the surrogate endpoint
deemed reasonably likely to predict clinical benefit, i.e., Becker-type dystrophin levels, and the
clinical benefit, i.e., maintenance of walking velocity. In Dr. Bastings’ memorandum, he
provides careful documentation of the trajectories of physical performance for each patient,
comparing their changes in performance to the quantity of dystrophin detected. After careful
consideration, he finds no correlation whatsoever.

Office Director Decisional Memo – NDA 206488 – Page 27 of 42
Reference ID: 3959961

 Figure 10: Study 201/202 – Lack of Correlation between Quantity of Dystrophin Detected
and Preservation of Physical Function (6-Minute Walk Distance)

Although it should be obvious that changes on the order of a percent or two are small, it is
nevertheless worthwhile to view these data at full scale to gain perspective (Figure 11). The
figure is identical to Figure 9, except for the scale on the y-axis.
If dystrophin were simply an enzyme
responsible for biochemical activity in
myocytes, one could posit that a very
small quantity of the protein could exert a
substantial treatment effect, especially
because levels are so low in untreated
patients. But given that dystrophin is a
structural support protein that helps
prevent myocyte injury from stress and
strain, I find it difficult to conceive how a
treatment effect of 3 parts per thousand
could confer clinical benefit. If there were
10 inches of snow on a sidewalk that
needed to be cleared, 3 parts per
thousand would amount to 1/32nd of an
inch. Finally, we must recognize receiving
a treatment that increases dystrophin by
0.3% is not that same as being born with
0.3% more dystrophin.

Figure 11: Studies 301 and 201/202: Expression of
Becker-type Dystrophin by Western Blot (full
scale)

Office Director Decisional Memo – NDA 206488 – Page 28 of 42
Reference ID: 3959961

 3.

Dose-response

Although the issue is somewhat peripheral to the “reasonably likely” question, the presence of a
dose-response in Study 201/202 would have provided supportive evidence that the dystrophin
that was detected was produced by eteplirsen. A dose-response was not evident, although one
could reasonably argue that the trial was very small and that the difference between 30 and 50
mg/kg/week was unimportant.
In a monkey study conducted to assess the pharmacodynamic effects of eteplirsen, a 1-log
increase in dose (from 4 to 40 mg/kg) caused minimal increase in exon 51 splicing as detected
by PCR (Section 4, Table 1). However, with a 2-log increase in dose (from 4 to 320 mg/kg),
there was a log increase in exon 51 splicing. As noted in Section 4 of this memorandum, it is
possible that much higher doses of eteplirsen could have a substantially greater effect, which
might translate to clinical benefit.
Advisory Committee
The Advisory Committee was asked to discuss: a) the strength of evidence that eteplirsen
increased the amount of dystrophin in muscles of treated patients relative to their baseline, and
b) the clinical meaning of the amount of dystrophin observed in the muscles of eteplirsentreated patients, taking into consideration the range of amounts of dystrophin known to be
typically present in patients with DMD and in patients with Becker muscular dystrophy. (Of note,
the data from Study 301 were not known/available to the Advisory Committee.)
Although the Committee failed to reach consensus on these questions, the discussion,
summarized below, is of interest.
With respect to production of dystrophin, about half of the committee members found evidence
that eteplirsen increased the amount of dystrophin produced in skeletal muscles. Among those
who were not convinced, two members cited issues with the controls (lack of pre- and posttreatment biopsies in the same patients; differences in muscle groups biopsied), two had
concerns about inconsistencies between dystrophin levels and clinical response (Figure 10),
and one cited concerns about the lack of a dose-response (Table 3).
Only four Committee members had explicit comments with respect to the clinical
meaningfulness of the amount of dystrophin detected in treated patients, and their opinions
were split. One member opined that the amount of dystrophin needed to impart clinical benefit
is unknown, but could be very low, or very low in a subset of patients. One of the patient
representatives felt strongly that dystrophin was produced, and that the amount was sufficient to
produce clinical benefit. One committee member, having opined that some dystrophin was
produced, stated that there is no basis to determine the quantity of dystrophin that would be
clinically significant, or whether the dystrophin is functionally active. Another committee
member, one who had not opined on whether dystrophin was produced, noted that whatever the
amount of dystrophin produced in the study, the amount was not clinically meaningful, based on
the lack of correlation between dystrophin levels and clinical results (Figure 10).
The Committee voted on whether the applicant had provided substantial evidence from
adequate and well controlled studies that eteplirsen induces production of dystrophin to a level
that is reasonably likely to predict clinical benefit.

Office Director Decisional Memo – NDA 206488 – Page 29 of 42
Reference ID: 3959961

 Ultimately, 7 members voted “no” and 6 voted “yes,” after one member changed his vote from
“no” to “yes.” In explaining their “no” votes, 5 committee members opined that the studies were
not adequate and well controlled; they questioned the techniques used to measure dystrophin
as well as the appropriateness of the controls. Four committee members expressed concern
about the lack of correlation between the dystrophin levels and clinical measures. They agreed
that even if some dystrophin was produced, there was no evidence that dystrophin production
was at a level that would be reasonably likely to predict clinical benefit. The 6 “Yes” votes
included the consumer representative and 2 patient representatives. These individuals believed
that there was some difference in dystrophin production and some evidence of improvement in
endpoints. One of the members who voted “Yes” stated that he was very troubled by not
understanding what constitutes a clinically significant amount, but was impressed by the
patients’ observations. Two members who voted “No” stated that their vote was justified by the
way the question was phrased, but that the patient testimonies suggested the drug works.
Is There a Basis for a Conventional Approval Based on Clinical Data?
The clinical data have been well described by the review team. The development program
consisted of one trial (Study 201/202) with a relatively short (24-week) placebo-controlled
portion (Study 201) followed by a long-term extension study (Study 202). Although the applicant
submitted biopsy data from the ongoing Study 301, no clinical data have been submitted from
that study.
As noted above, for Study 201, patients were randomized (1:1:1) to eteplirsen 30 mg/kg/week,
eteplirsen 50 mg/kg/week, or placebo (n=4 per group). After 24 weeks, the 4 patients originally
randomized to placebo were re-randomized to eteplirsen 30 mg/kg/week (n=2) or eteplirsen 50
mg/kg/week (n=2) and followed for 4 additional weeks. The trial was extended to an open-label
phase (Study 202), where all 12 patients continued to receive eteplirsen without interruption,
although investigators and patients remained blinded to dose.
The 1° endpoint of Study 201 was the percentage of dystrophin-positive fibers in muscle
biopsies as assessed using immunohistochemistry, but there were numerous exploratory
endpoints.
When the data from Study 201 were originally analyzed, the applicant found that eteplirsen
caused a striking and unprecedented increase in dystrophin production, based on the reading of
the immunohistochemistry data at Nationwide Children’s Hospital, with supportive data from
Western blot analyses.
The clinical data, too, were interpreted as positive. As discussed by the review team, 2 patients
in the 30 mg/kg/week treatment group became unable to ambulate soon after the trial began,
and there were no significant differences in 6-minute walk distance among the groups. Despite
clearly negative results, the applicant performed a post hoc analysis that omitted the 2 patients
in the eteplirsen group who became unable to ambulate. They represented these results as
positive, and publically promoted both the immunohistochemical dystrophin results and the 6minute walk data as positive (see clinical review).
Although FDA would later determine that the analyses underlying these data were not valid, the
publicity from the paper2 and Sarepta’s press release3 raised unrealistic expectations of efficacy

Office Director Decisional Memo – NDA 206488 – Page 30 of 42
Reference ID: 3959961

 in the DMD community. It was these perceptions that led the applicant to conclude that a
second placebo-controlled study would not be feasible.
FDA strongly suggested a second, larger, adequately-powered, placebo-controlled trial, but the
applicant was reluctant to run such a trial, in part because their supply of drug was limited, and
in part because of their insistence that the DMD community would not agree to participate in a
trial where there was a chance of receiving placebo. Faced with the applicant’s unwillingness to
conduct a second placebo-controlled trial, FDA agreed to an externally-controlled trial: a
comparison between patients in the ongoing Study 202 and patients in an external control
group. The Division expressed strong concern, however, with respect to the interpretability of
such a trial with 6-minute walk distance as the endpoint, given that physical performance is not
a “hard” endpoint, but can be influenced by motivation and other factors. Citing FDA Guidance, 5
the Division noted that the treatment effect would have to be dramatic for the results from an
externally-controlled study to be interpretable. Details of the interactions between FDA and
Sarepta are well documented by the review team.
International guidelines, adopted by the FDA as guidance, stress caution with respect to the
interpretation of data from externally-controlled trials. As noted in the International Conference
on Harmonization (ICH) E10 Guideline,5 blinding and randomization, used to decrease bias in
randomized controlled trials, are not utilized in externally-controlled trials; the inability to control
bias is a critical limitation of externally controlled trials. Groups can be dissimilar with respect to
a wide variety of factors that could influence outcome – factors that are both known and
measurable as well as factors that are unknown. As explained by Dr. Robert Temple at the
April 25, 2016 meeting of the Peripheral and Central Nervous System Drugs Advisory
Committee, it has been well documented that untreated historical-control groups tend to have
worse outcomes than apparently similarly chosen control groups of randomized studies,
possibly reflecting a selection bias.
The ICH E10 Guideline explains: “A consequence of the recognized inability to control bias is
that the potential persuasiveness of findings from externally controlled trials depends on
obtaining much more extreme levels of statistical significance and much larger estimated
differences between treatments than would be considered necessary in concurrently controlled
trials. The inability to control bias restricts use of the external control design to situations in
which the effect of treatment is dramatic and the usual course of the disease highly predictable.
In addition, use of external controls should be limited to cases in which the endpoints are
objective and the impact of baseline and treatment variables on the endpoint is well
characterized.” In essence, in order to be interpretable, the finding of a difference between
groups should be large – so large that the difference is patently obvious without the need to rely
on inferential statistics.
Having heard FDA’s concerns regarding the potential difficulty in interpreting an externallycontrolled trial, the applicant nevertheless obtained access to individual data from patients with
DMD from Professor Eugenio Mercuri at the Catholic University in Rome on behalf of the Italian
DMD Registry database (n=97) and from Professor Nathalie Goemans at the University
Hospitals in Leuven (n=89). From these 186 patients, 50 had a genotype amenable to exon
skipping therapy, were using corticosteroids at baseline, had 6-minute walk data available at
baseline, and were ≥ 7 years old. Among these 50 patients, 13 had a genotype amenable to
5

Guidance for Industry: E10 Choice of Control Group and Related Issues in Clinical Trials, May 2001

Office Director Decisional Memo – NDA 206488 – Page 31 of 42
Reference ID: 3959961

 exon 51 skipping therapy. I will note that the review team has been unable to gain an
understanding of how dates of inception were determined for registry patients, i.e., when
patients were considered to have ‘enrolled.’
Study 202 was continued, therefore, with patients continuing to receive either 30 or 50
mg/kg/week eteplirsen. Numerous comparisons of physical function were planned between
these 12 patients and the 13 patients in the external control group. Measures included 6-minute
walk, rise time, timed 10-meter run, and North Star Ambulatory Assessment (NSAA).
With two small groups of patients, there was no way to match patient pairs. Fortuitously, the
mean ages and 6-minute walk distances were well matched at baseline, although the review
team found that initial age of steroid use and baseline NSAA scores were dissimilar between
groups – and both of these differences favored the eteplirsen group.
It is clear that some patients exited the registry to enroll in clinical trials. Thus, DMD patients
who remained in the Italian and Belgian registries (the control group): 1) did not seek knowledge
(or lacked knowledge) regarding applicable clinical trials into which they might have enrolled; 2)
sought enrollment in trials but did not qualify; or 3) qualified for enrollment in a trial(s) but made
a conscious decision not to participate. Obviously, such patients could differ substantially from
patients in Study 201/202. The point is that there can be unknown factors beyond baseline age,
weight, length of steroid use, and 6-minute walk distance that importantly affect outcomes.
The applicant presented the data by time-on-treatment, but because physical abilities change
significantly with age in patients with DMD, the review team believes that the more meaningful
way to display the longitudinal 6-minute walk data is by age (recognizing that both analyses
have advantages and limitations, and that there is no ideal way to present these data). The 6minute walk data are shown in Figure 12 as a function of age. The review team stresses that,
Figure 12: Patients in Study 202 vs. Patients in External Registries: 6-Minute Walk Distance by Age

Eteplirsen
Control

Office Director Decisional Memo – NDA 206488 – Page 32 of 42
Reference ID: 3959961

 by simple visual inspection, the two groups show little difference in performance.

There are 4 patients in the eteplirsen group, ~14 to 15 years of age, who continue to retain good
walking ability (inside the oval). There are 2 control patients in this age range who had been
maintaining similar walking ability, but appear to have experienced a precipitous loss of
ambulation between ages 14 and 15. As explained by the review team, there are concerns
regarding the comparability of the assessments of these patients, and concerns about
comparability of the groups in general.

The applicant?s argument for accelerated approval is based on this comparison of 6-minute walk
distance between the patients in Study 202 and the patients in the external control group from
Italy and Belgium. The difference in 6-minute walk distance is certainly statistically signi?cant.
The problem is that the study was externally-controlled, and the statistical test was based on a
non-randomized comparison.

Data from the Cooperative International Neuromuscular Research Group (CINRG) provide an
additional source of information on the natural history of patients with DMD. Figure 13 is a
Kaplan-Meier (K-M) survival curve from CINRG showing time-to-loss of ambulation. Of note,
25% of patients remain ambulatory at age 17; their course seems quite consistent with that of
patients from Study 201/202.

 

 

 

 

Figure 13: Data on DMD Patients from the Cooperative International Neuromuscular Research
Group (CINRG) on Loss of Ambulation (Source: Dr. Farkas? Review, 5/17/16: Fig. 21, Pg. 66)

8 K-M Estimate for Loss of Ambulation





Age (years)

 

 

 

In summary, the review team strongly believes that patients on eteplirsen in Study 201/202 do
not demonstrate a substantial treatment effect on walking velocity that clearly differentiates their
course from the natural history of the disease. For a more complete description with
comprehensive patient profiles, see the reviews of Drs. Breder and Farkas and the memo of Dr.
Bastings.

Of?ce Director Decisional Memo NDA 206488 Page 33 of 42
Reference ID: 3959961

Finally, as stressed by the review team, the data from other measures of physical function, i.e.,
rise time, timed 10-meter run, and North Star Ambulatory Assessment (NSAA), show steady
decline in the eteplirsen-treated patients that does not differ substantially from the decline in the
external control group. The NSAA data are shown in Figure 14 by time on treatment (eteplirsen
patients) or time since inception (registry patients). The NSAA is thought to be a
comprehensive outcome measure, well reflecting the functional abilities of DMD patients. Of
note, the downward trajectories of the two groups are indistinguishable (the lines are virtually
parallel with equal slopes).
Figure 14: Patients in Study 202 vs. Patients in External Registries: Mean North
Star Ambulatory Assessment (NSAA) Scores by Time on Treatment

Eteplirsen
Control

Patient Testimony/Advisory Committee:
In addition to the presentations made by the applicant and the review team at the April 25, 2016,
Advisory Committee Meeting, there were testimonies from over 50 individuals and families,
including most of the patients who were participating in Study 202. (Per email communication
(b) (4)
from
, one of the applicant’s consultants, 10 of the 12 patients testified and
another patient had someone speak on his behalf.)
In addition, the applicant invited Christine McSherry, Executive Director of the Jett Foundation,
to present “Patient and Caregiver Reported Outcomes of Patients in Clinical Trials of Eteplirsen
for Treatment of Duchenne.”
Office Director Decisional Memo – NDA 206488 – Page 34 of 42
Reference ID: 3959961

 The testimonies of these patients were quite consistent and remarkably positive: all were
convinced that eteplirsen had made a substantial positive impact on their physical performance,
improving numerous aspects of their lives.
It was noteworthy that a number of individuals who were in Study 201/202 reported
improvement in physical function with eteplirsen treatment. For example, one patient stated that
he had required a wheelchair at a school he had attended in the past, whereas he no longer
needed a wheelchair at his present school. A video showed a boy who, prior to treatment, had
some difficulty climbing up into the seat of a minivan. After receiving eteplirsen for several
months, he was shown jumping up easily into the seat. In another video, a boy in the study
threw a football, a tight spiral, with ease and finesse.
Many of the Committee members seemed obviously moved and deeply affected by these
testimonies and videos, seemingly convinced that there was a treatment effect.
Importantly however, despite the claims of improvement made at the microphone at the
Advisory Committee meeting, the review team did not find any patients in Study 201/202 with
consistent improvement in physical performance as assessed by formal testing (6-minute walk,
rise time, NSAA, 10-meter run). These tests have shown moderate to extreme declines in
physical function for all patients (see NSAA data, Figure 15).
Thus, the review team and many
on the Advisory Committee
(including Benjamin Dupree, the
patient representative with DMD),
were unable to reconcile the
patient testimonies with the data
collected by the applicant: the
testimonies spoke of
improvement; the data showed
progressive worsening.

Figure 15: Study 201/202 – Individual NSAA Performance by Age

The Advisory Committee voted (7
to 3 with 3 abstentions) that the
clinical results of Study 201/202
did not provide substantial
evidence that eteplirsen is
effective for the treatment of DMD.
The 7-member majority of the
committee who voted "no" agreed
that Study 201/202 was not a wellcontrolled study. Most cited
problems with the controls. One
member explained that a
historically-controlled study could provide evidence of effectiveness, but that Study 202 did not.
Two committee members noted that the original placebo-controlled portion of the study was

Office Director Decisional Memo – NDA 206488 – Page 35 of 42
Reference ID: 3959961

 negative. One member who cited issues with the controls also noted that a single trial would be
insufficient to provide substantial evidence.
The 3 members who voted that there was substantial evidence of effectiveness explained that
the study results correlated with the testimonies presented by the public.
Commentary:
I agree with the Division, the Office of Biometrics, the Office of Clinical Pharmacology, and the
Advisory Committee with respect to the lack of substantial evidence of effectiveness for
eteplirsen. The review team elaborates on many factors that differ, or could differ, between the
treatment groups – factors that could lead to a difference in outcomes. Externally-controlled
trials are best-suited for diseases where progression is highly predictable and treatment effects
are extreme. Although there appeared to be a difference in ambulation between patients in
Study 202 and patients in the external control group, the effect size was not sufficient to be
persuasive, given the inability to control bias in an externally-controlled study. As explained in
ICH E10, “…the potential persuasiveness of findings from externally controlled trials depends on
obtaining much more extreme levels of statistical significance and much larger estimated
differences between treatments than would be considered necessary in concurrently controlled
trials.” With only 12 patients in the trial and a moderate difference in walking velocity, the study
falls short.
Finally, it is critical to note that no dose-limiting side effects were observed at either dose tested
in Study 201/202, and even the most optimistic interpretation of the data is that patients
experienced gradual decline in function – not stabilization. Even if one were to reach the
conclusion that the applicant showed substantial evidence of dystrophin production, deserving
of accelerated approval, investigation of higher doses would be imperative.
8.

Safety

As explained in the clinical review, the number of subjects exposed was too small to provide an
adequate assessment of safety. On the other hand, I also agree with the review team that the
deficiencies in safety assessments would not likely be an issue for approvability in their own
right had the drug been demonstrated to be effective. In other words, for a therapy that is
shown to be effective in a serious condition where there are no approved drugs, we would
approve a marketing application even with substantial risks, as long as we could write adequate
instructions for use. Moreover, we would not delay approval of a marketing application because
of uncertainty of risks. Instead, we would work with the applicant to obtain more extensive
safety data post-approval. Such would be the case for this application if there were substantial
evidence of effectiveness.
Of note, many patients in these studies are now receiving infusions through chronic indwelling
catheters. Although we are not aware of any serious adverse events cause by infections, with
approval of this drug there would undoubtedly be serious infections and possibly rare deaths
eventually. The risk of an indwelling IV line in patients on chronic corticosteroids should be
mentioned in labeling if the drug is approved.
Although neither immunogenicity nor allergic reactions have been reported with eteplirsen,
immunogenicity testing would be advisable in ongoing trials. Moreover, given that these

Office Director Decisional Memo – NDA 206488 – Page 36 of 42
Reference ID: 3959961

 patients may be naïve to Becker-type dystrophin, the potential for anti-dystrophin antibodies
should be studied as well.
9.

Advisory Committee Meeting

There were many important discussions at the April 25, 2016 Advisory Committee Meeting, and
they are summarized above, in context.
10.

Pediatrics

Duchenne Muscular Dystrophy is an orphan indication, not subject to the Pediatric Research
Equity Act.
11.

Other Relevant Regulatory Issues

Site Inspections:
The site at Nationwide Children’s Hospital was inspected in 2014. See description and
conclusions in Section 7, above, and, in particular, the summation and discussion in Dr.
Breder’s review.
Dr. Ashutosh Rao conducted an inspection with Young Moon Choi, Ph.D. (Office of Study
Integrity and Surveillance) and Mark Babbit (Office of Regulatory Affairs) of the facilities at
University of Iowa in Iowa City, IA and Sarepta Therapeutics Inc. in Corvallis, OR. The
inspections confirmed that the blinding procedure, handling of the sample shipment, and the
conduct of Western blot analyses of the samples from Study 301 (PROMOVI) were consistent
as predefined in the protocol.
Name Review:
The Division of Medication Error Prevention and Analysis concluded that the proposed
proprietary name, “EXONDYS 51,” is acceptable from both a promotional and safety
perspective.
12.

Labeling

I do not recommend approval, but if the drug were to be approved, the label would need to state
that no clinical benefit has been established, and explain the effect on the surrogate endpoint in
clearly understandable language (i.e., 0.3% or 3 parts in a thousand). Section 6 would need to
note that safety is not well characterized.
13.

Decision/Action

DMD is a rare genetic disease characterized by the near absence of functional dystrophin
protein, leading inexorably to myocyte degeneration, muscle dysfunction and inflammation,
severe disability, and death, robbing patients of their dignity along the way. Although steroids
are thought to slow the course of the disease and are typically considered standard of care, they
are by no means curative, and they have their own side effects.

Office Director Decisional Memo – NDA 206488 – Page 37 of 42
Reference ID: 3959961

 The cause of DMD is well established – the absence of structural dystrophin protein in
myocytes. There is wide belief in the medical/scientific community that restoration of functional
dystrophin protein has a strong potential to ameliorate the disease.
Eteplirsen is a novel PMO that is designed to lead to translation of an abnormal but functional
dystrophin protein – a protein that is produced in Becker muscular dystrophy, a far less severe
form of muscular dystrophy. The data from RT- PCR show that the drug produces the intended
Becker-type messenger RNA; we have no data on the extent of messenger RNA production.
As noted by the review team, the clinical data generated from study 201/202 do not provide
evidence of efficacy. The aim of Study 201, the only randomized placebo-controlled study
conducted by the applicant, was to assess dystrophin production in response to lower and
higher eteplirsen regimens (30 or 50 mg/kg/week) vs. placebo. Results of the original analyses
of Study 201, published in a major journal, were remarkably positive, and their publication led to
widespread enthusiasm for the drug. Unfortunately, an FDA inspection found a number of
important technical factors that rendered the data unreliable and uninterpretable: the Western
blot analyses were sub-standard; there were also critical problems with the reading of the
immunohistochemistry images. FDA recommended a blinded re-read of the images, but upon
re-read of the images by 3 blinded pathologists using FDA-recommend procedures, there was
no increase in dystrophin production.
Likewise, Study 201 did not meet its 1° clinical endpoint, 6MWT, at Week 24. Two patients in
the low-dose eteplirsen group became unable to ambulate early in the study, such that a proper
intent-to-treat analysis of the 6-minute walk data nearly showed a statistically significant
difference in favor of placebo.
The applicant switched all patients to active drug in Study 202, and has continued to follow the
patients for 6-minute walk distance, NSAA, and rise time.
Study 202 did not meet its 1° clinical endpoint, 6MWT, at 48 weeks.
The alternative analyses of Study 202 proposed by the applicant are based on comparison to an
external control group obtained from registry patients in Italy and Belgium. Questions about
comparability notwithstanding, analyses have not shown a clear separation of the disease
course between eteplirsen-treated patients and external controls. Moreover, there is not a clear
separation between eteplirsen-treated patients and patients in the CINRG registry. Thus,
neither external control group suggests there is a treatment effect.
The Western blot analyses from Week 180 of Study 201/202 showed a low quantity (0.9%) of
dystrophin; however, the study was not adequate and well controlled (the baseline level of
dystrophin was not known with certainty), and the lack of correlation between results of Western
blot and immunohistochemistry demonstrates a troubling lack of internal consistency.
Study 301, on the other hand, was an adequate and well-controlled study that provided
substantial evidence of Becker-type dystrophin production in response to eteplirsen. The mean
change in Becker-type dystrophin with treatment was 0.22% to 0.32%, depending on the
method used to impute values less than the lower limit of quantification. Although all members
of the review team believe that Becker-type dystrophin is an appropriate surrogate endpoint, the
mean quantity of dystrophin produced in Study 301 was minute by any standard. In considering

Office Director Decisional Memo – NDA 206488 – Page 38 of 42
Reference ID: 3959961

 responders, even the largest responder in Study 301 produced only 1.33% of normal
dystrophin, which is thought by many authorities to be insufficient. No other patient produced
1% dystrophin in response to treatment.
Recognizing that the threshold for the effect size needed to be ‘reasonably likely’ to predict
clinical benefit is not known, the view provided in the literature suggests that at least 3% of
normal dystrophin is inadequate, and levels perhaps much more, a minimum of 10%, would be
necessary for detectable clinical benefit. The finding in Study 301, an increase in the range of
0.22 to 0.32% of normal, is an order of magnitude below this level.
The unprecedented finding of an increase in dystrophin protein in response to eteplirsen
establishes proof-of-concept and provides great promise that this drug, or other therapies, will
be capable of ameliorating the fundamental genetic defect of DMD, but the effect size seems
insufficient at the tested doses.
Various individuals have opined that there appears to be some evidence that some patients are
producing dystrophin in response to eteplirsen; however, such optimism fails to reach the legal
threshold of ‘reasonably likely to predict clinical benefit’ required for accelerated approval.
Accelerated approval of this NDA based primarily on the change in Becker-type dystrophin in
Study 301 would be problematic for these reasons:
1. The amount of dystrophin produced in Study 301 is so meager that it could be considered to
be tantamount to any increase in dystrophin. In other words, if a statistically significant
change of 0.22% is considered adequate to support accelerated approval, then the question
arises as to whether there is any statistically significant change that would be too small to
support accelerated approval. Similarly, if a response had been defined as a treatment
effect of 1%, there would have been only one (out of 12) responders in Study 301.
If we were to adopt the concept that, for rare diseases, accelerated approval can be
supported by any statistically significant change in an appropriate surrogate (or by a
response in a single patient), we would enable accelerated approval of numerous drugs for
rare diseases. No doubt there are some who would applaud this as a regulatory advance,
but these are typically the kinds of findings that support Breakthrough Designation, not
approval. If accelerated approval based on any change in a surrogate endpoint is what is
meant by regulatory flexibility and this is the new normal, a new approval pathway is clearly
needed.
With lowering of the standard for accelerated approval, the result would be a world where
traditional clinical trials are abandoned in favor of small proof-of-concept studies designed to
show any level of production of a target protein – e.g., a statistically significant effect in a
paired pre- vs. post-treatment analysis that is clinically meaningless. There would be no
reason to pursue placebo-controlled clinical trials to support efficacy prior to accelerated
approval; in fact, the possibility of failure would provide a substantial disincentive to the
conduct of such trials. Lowering the bar to this level would be tantamount to rolling back the
1962 Kefauver-Harris Drug Amendments to the Federal Food, Drug and Cosmetic (FD&C)
Act, which have served Americans well for some 54 years.

Office Director Decisional Memo – NDA 206488 – Page 39 of 42
Reference ID: 3959961

 2. Even if the 30 mg/kg/week dose were considered to have a meaningful effect on the
surrogate endpoint, the dose is sub-therapeutic. Moreover, the short 3.5-hour half-life of
eteplirsen by no means supports a weekly dosing regimen. I question the ethics of
approving or prescribing a drug for a fatal disease at a dose that is very likely to be subtherapeutic.
Imagine that 100 years ago a promising drug called penicillin is discovered – a potential
cure for pneumococcal pneumonia – but the drug is difficult to produce and expensive. A
dose of 5 mg weekly has been shown to have statistically significant bactericidal effects on
Streptococcus pneumoniae. Would it be ethical to give the drug accelerated approval based
on this finding and allow marketing of a dose of 5 mg, absent additional information? (The
therapeutic dose is ~2 logs higher than 5 mg.) Patients who might receive a lifesaving
therapy (i.e., a higher dose) would die because the dose is too low.
Despite considerable pressure from the DMD patient community and many well-intentioned
members of the public who have lobbied on their behalf, I am unable to reach the conclusion
that the applicant has provided substantial evidence to support either conventional or
accelerated approval of eteplirsen for the treatment of DMD. This view is in agreement with the
unanimous opinions of members of the review team from the Division of Neurology Products,
the clinical pharmacology review team, and the biostatistics review team. The Advisory
Committee was under intense and near-incessant pressure from a large public audience, urging
them to believe that eteplirsen was effective, and life changing in some circumstances.
Emotions in the room ran high. In spite of this pressure, that majority of the Advisory Committee
voted against both conventional and accelerated approval.
In a June 3, 2016 letter from Dr. Janet Woodcock, the applicant was advised that “If you are
successful in showing, to FDA’s satisfaction, a meaningful increase in dystrophin by Western
blot analysis between the paired pre-and post-treatment samples, we expect to be able to grant
an accelerated approval….” It is difficult to consider production of 2 to 3 parts per thousand as
a “meaningful” change. To put this effect into perspective, if a normal amount of dystrophin
were equivalent to a $5 bill, this change would be equivalent to a penny.
With all of this information at hand, most sponsors would have concluded that exploration of
higher doses was needed; however, this applicant chose instead to trumpet the preliminary
findings from their 12-patient phase 1/2 study, convincing many in the DMD community that the
drug was highly effective, and unleashing a public media campaign (with support of many
politicians) to approve the drug. The reality is that FDA is a science-based organization. We do
not – and should not – make approval decisions based on patient anecdotes or campaigns
through social media.
I strongly agree with the decisions of Dr. Bastings, reviewer staff in the Division, the Office of
Biometrics, and the Office of Clinical Pharmacology to issue a complete response for this NDA.
I also agree that it would be desirable to provide access to this drug for DMD patients through
expanded access programs, with cost recovery, while an adequate dose-finding study is
conducted.

Office Director Decisional Memo – NDA 206488 – Page 40 of 42
Reference ID: 3959961

 Path Forward:
Based on the quantity of Becker-type dystrophin produced in Study 301 and the clinical findings
in Study 201/202, additional studies at this dose are unlikely to support any type of approval,
i.e., the data obtained for eteplirsen at a dose of 30 and 50 mg/kg/week are adequate, but they
do not support efficacy.
We remain comfortable with the concept that substantial evidence of dystrophin production from
adequate and well controlled trials could support accelerated approval, but it is clear that higher
doses are needed, and greater quantities of dystrophin would need to be produced. The path to
a conventional approval would require a double-blind, placebo-controlled (or multi-dose) study,
at least one year in duration, using some measure of physical performance as the 1° endpoint,
again, testing higher doses.
The applicant is continuing to enroll the PROMOVI study, an open-label, multi-center, 48-week
study in patients with DMD amenable to skipping exon 51. All patients are receiving eteplirsen,
30 mg/kg/week as an IV infusion.
The 1° endpoint is change in 6-minute walk test distance from baseline. A 2° endpoint is the
percentage of dystrophin-positive fibers, as assessed by immunohistochemistry. Patients
undergo muscle biopsies at baseline and various time points to assess dystrophin production.
My suggestion for a path to approval is to randomize patients in the ongoing PROMOVI study
to:
1) remain on 30 mg/kg/week; or
2) have their dose significantly increased. This could be done through use of a higher
dose, through more frequent dosing intervals (with dummy infusions), or both. Given
that many patients receive eteplirsen through indwelling IV lines and no significant
infusion reactions have occurred, perhaps these infusions could be performed at home.
For example, the study could compare 30 mg/kg weekly to 30 mg/kg daily. Patients who
do not tolerate more frequent dosing could have their doses decreased, as needed.
Based on non-clinical findings, monitoring would need to be in place to assess renal
toxicity.
Patients and investigators would be blind to treatment group. For accelerated approval, the 1°
endpoint would be dystrophin production, comparing the higher and lower doses. For standard
approval, the 1° endpoint would be a test(s) of physical performance such as rise time or the
NSAA.
Such a trial would be methodologically sound and ethical. Virtually everyone, patients and
physicians alike, want to know if higher eteplirsen doses would increase dystrophin production,
and would have equipoise for participation. Although there is concern regarding performance of
muscle biopsies in patients assigned to placebo, this concern would not exist in this study. And
if the applicant were to forego immunohistochemistry studies, needle biopsies with local
anesthesia (rather than open biopsies under more intensive anesthesia) would be sufficient.
This study design would simultaneously address another concern that I believe has been
underappreciated by many. As noted above, it would be problematic in my view to approve a
dose of 30 mg/kg/week, presumably leading to a dystrophin increase of ~0.3%, when it is

Office Director Decisional Memo – NDA 206488 – Page 41 of 42
Reference ID: 3959961

 known that this dose fails to prevent the decline in physical function and yet produces no overt
toxicity. The monkey data (Table 1) suggest that much higher doses might have a far greater
effect on exon skipping, an impact that might prevent disease progression. Thus, it seems
imperative to study higher exposures.
14.

Final

Many of us would wish to approve this drug if we could. DMD is a horrible disease and there
are no approved treatments. FDA takes seriously the patient perspective and our congressional
mandate to be flexible. But patient-focused drug development is about listening to patient
perspectives about what matters to them; it is not about basing drug approvals on anecdotal
testimony that is not corroborated by data.
FDA is charged with the responsibility of ensuring that drugs are shown to be effective prior to
marketing, based on substantial evidence. If we were to approve eteplirsen without substantial
evidence of effectiveness, or on the basis of a surrogate endpoint with a trivial treatment effect,
we would quickly find ourselves in the position of having to approve a myriad of ineffective
treatments for groups of desperate patients, in essence, allowing marketing based on
desperation, patient lobbying, and the desire and need of hope. If we were to turn the clock
back to the days prior to the 1962 Kefauver Harris Amendments to the Federal Food, Drug, and
Cosmetic Act, the damage to society and the field of evidentiary medicine would be enormous.

Office Director Decisional Memo – NDA 206488 – Page 42 of 42
Reference ID: 3959961

 --------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed
electronically and this page is the manifestation of the electronic
signature.
--------------------------------------------------------------------------------------------------------/s/
---------------------------------------------------ELLIS F UNGER
07/16/2016

Reference ID: 3959961