MEMORANDUM DEPARTMENT OF HEALTH AND HUMAN SERVICES PUBLIC HEALTH SERVICE FOOD AND DRUG ADMINISTRATION CENTER FOR DRUG EVALUATION AND ESEARCH __________________________________________________________________________ CLINICAL INSPECTION SUMMARY DATE: May 18, 2012 TO: Alison Blaus, Regulatory Health Project Manager Martin Rose, M.D., J.D., Clinical Medical Officer Nhi Beasley, PharmD, Safety Medical Officer Division of Cardiovascular and Renal Drug Products FROM: Sharon Gershon, Pharm.D Good Clinical Practice Assessment Branch Division of Good Clinical Practice Compliance Office of Scientific Investigations THROUGH: Susan Thompson, M.D. Acting Team Leader, Good Clinical Practice Assessment Branch Division of Good Clinical Practice Compliance Office of Scientific Investigations THROUGH: Lauren Iacono-Connors, Ph.D. Acting Branch Chief, Good Clinical Practice Assessment Branch Division of Good Clinical Practice Compliance Office of Scientific Investigations SUBJECT: Evaluation of Clinical Inspections NDA: NDA 202155 (IND 68,598) APPLICANT: Bristol Myers Squibb Company Princeton, New Jersey DRUG: Eliquis® (apixiban) tablets NME: Yes THERAPEUTIC CLASSIFICATION: Priority Review Reference ID: 3132882 Page 2 Clinical Inspection Summary NDA 202155 [apixiban] (4) INDICATION: Prevention of stroke, systemic embolism, in patients with non-valvular atrial ?brillation (AF) RELEVANT DATE Clinical Site/ Sponsor Audit Request November 17, 201 1 Inspection Surmnary Goal Date (Original) February 28, 2012 Division PDUFA Date (Original) March 28, 2012 Major Amendment Letter February 28, 2012 Sponsor/CRO Consult Request (Second) March 7, 2012 New Action Goal Date May 15, 2012 New PDUFA Date J1me 28, 2012 I. BACKGROUND: Apixaban, an inhibitor of the coagulation Factor Xa is being developed as an antithrombotic/ anticoagulant agent for multiple indications. Under NDA 202155, apixaban is proposed for the reduction of stroke and systemic embolism (SE) in subjects with non- valvular atrial ?brillation (AF). The drug is formulated as an oral dosage form (2.5 mg and 5 mg tablets), to be administered twice a day. AF is a cardiac which is present in approximately 2.6 million patients in the United States and is associated with a ?ve-fold increase in the risk of stroke. Crurent guideline-recommended agents for prevention of stroke in patients with AF include anticoagulants like warfarin or other vitamin antagonists (VKAs). Other therapies for stroke prevention include aspirin (ASA), and most recently, dabigatran. Antiplatelet therapy with ASA is recommended in patients with AF at low risk for stroke. ASA is also recommended for some patients at moderate risk for stroke, with a higher estimated risk of bleeding if anticoagulated, or with dif?culty accessing high quality anticoagulation monitoring. Whereas warfarin provides effective protection against stroke and systemic embolism, warfarin and other VKAs have a number of limitations: variable pharmacokinetic and phannacodynamic pro?les, numerous drug interactions, need for laboratory monitoring and dosage titrations, and bleeding. All of these factors lead to under use in patients with AF. Warfarin must be carefully dosed and continually monitored by means of International Normalized Ratio (INR) testing. As the INR drops below 2.0, the risk of ischemic stroke rises; as the INR rises above 4.0, the risk of hemorrhagic stroke increases. For patients to be safely anticoagulated with warfarin, the INR should be maintained in a range of between 2.0 and 3.0. In clinical trials, this desirable result is achieved about 65% of the time. To overcome some of the above limitations, newer antithrombotic agents are being developed that are highly selective for speci?c coagulation factors blocking the of thrombin. In early January 2012, BMS noti?ed OSI concerning allegations of suspected GC misconduct at the clinical investigator site of Dr. Shiyao Wu (Site 1200) in Shanghai, China. A PPD monitor (m6) present at the site in preparation for FDA inspection observed (based on her earlier visits to the site) that handwritten comments on subject records were Reference ID: 3132882 Page 3 Clinical Inspection Summary NDA 202155 [apixiban] no longer present. Preinspectional site visits occurred between November 14 and December (b) (6) (b) (6) 7, 2011. In addition, Mr. , a second PPD monitor, shared with Ms. that he had changed documents and changed patient data on a USB drive at the direction of (b) (6) Ms , BMS Senior Clinical Site Manager. BMS became aware of these allegations on December 7, 2011. In follow-up, BMS senior management conducted an investigation into these reported allegations and their investigation took place over three weeks between December 20, 2011 - January 12, 2012. The investigation included visits by senior BMS management to China Site 1200, and included a visit to China Site 1178 where a second FDA inspection was to be conducted. As part of their investigation BMS conducted interviews with key study personnel and performed a comprehensive review of study documentation including subject diaries, outpatient visit records, and hospital records. They also performed a comparison between the electronic files for nine subjects on the USB drive to their outpatient visit records. BMS concluded that inappropriate activities had in fact (b) (6) (b) (6) occurred at Site 1200 that involved Ms. Mr. , and other BMS staff in China who had knowledge of the GCP deficiencies at Site 1200; that records of subjects on the USB drive had been modified; and that the integrity of the data from Site 1200 had been compromised. The GCP deficiencies identified by BMS included drug accountability issues (missing warfarin bottles), subject diaries that were missing year, month and subject number, informed consent documents signed by different people, four potential unreported SAEs, late reporting of three SAEs, inconsistencies between the SAE report versus eCRF and Chinese versus English versions, possible unreported endpoint events (two bleeds and one stroke), unreported adverse events, source documents missing patient name or number, physical examinations not done at end of treatment visits, ECGs not done at screening, incorrect concomitant medications and stop date. In their response to an OSI Information Request issued on February 15, 2012, BMS stated (b) (6) that Ms. worked as Site Manager for the BMS ARISTOTLE study and that she visited 18 of the 36 Chinese sites while the study was underway. Her duties included conduct of co-monitoring visits if needed; site recruitment/enrollment; resource needs; and following up on quality issues noted by the Site Monitor. She reviewed site visit reports written by the site monitor. During pre-inspection visits, she reviewed study files and subject medical records and discussed issues with study personnel. The sites in China where (b) (6) Ms. worked were: 1180, 1182, 1199, 1206, 1207, 1221, 1223, 1246, 1247, 1266, 1547, 1548, 1549, 1550, 1552, 1555, 1556, and 1558. She worked on no sites outside (b) (6) China. To prepare for the upcoming FDA inspections, Ms. worked at Sites: 1200 (b) (6) and 1178. PPD employee performed site monitoring responsibilities for the (b) (6) following Sites: 1200, 1198, 1168, 1244, and 1287. Mr. did not work at non-China (b) (6) (b) (6) sites. BMS terminated Ms. and Mr. , as well as a third senior BMS (b) (6) management employee who had communicated with Ms. regarding potential GCP violations at the site. A brief summary of the pivotal protocol and the study results are given below. STUDY PROTOCOL CV185030: A Phase 3, Active (Warfarin) Controlled, Randomized, Double-Blind, Parallel Arm Study to Evaluate Efficacy and Safety of Apixaban in Preventing Stroke and Systemic Embolism in Subjects with Non-valvular Atrial Fibrillation Reference ID: 3132882 Page 4 Clinical Inspection Summary NDA 202155 [apixiban] (ARISTOTLE: Apixaban for Reduction In STroke and Other ThromboemboLic Events in Atrial Fibrillation) In this current Phase 3 study, subjects with AF and at least one additional risk factor for stroke were evaluated for study eligibility. Eligible subjects were randomized in a 1:1 ratio to either apixaban or warfarin titrated to a target INR range of 2.0 to 3.0. Subjects who were on warfarin or another VKA prior to randomization had their VKA discontinued prior to randomization. The Intended Treatment Period started on the day of randomization and ended at the efficacy cut-off date. The efficacy cut-off date was the date on which it was expected that the target number of primary efficacy events (448) would have occurred. The date was set (arbitrarily) as January 30, 2011. Primary efficacy events were defined as all suspected efficacy events as adjudicated by the Adjudication Committee (also referred as Clinical Event Committee [CEC]); these included death, stroke, SE, and MI, as defined by the Adjudication Committee charter. The primary efficacy endpoint was the number of days from randomization to first occurrence of confirmed stroke (hemorrhagic, ischemic or unspecified type) or SE during the Intended Treatment Period. The key secondary efficacy endpoints were defined as some composite of stroke, SE or major bleeding and all-cause death during the Intended Treatment Period, as defined by the protocol. Brief Summary of Results The sponsor claims that apixaban was superior to warfarin for reduction of stroke (hemorrhagic or ischemic), SE (two-sided p=0.0114, HR=0.79), and all-cause death (twosided p=0.0465, HR=0.89). The incidence of each individual efficacy endpoint including hemorrhagic stroke, ischemic or unspecified stroke, SE, and MI was lower in the apixaban arm than in the warfarin arm. The non-inferiority (NI) of apixaban versus warfarin for prevention of stroke or SE, using both NI criteria described in the protocol was demonstrated (one-sided p-value < 0.0001). Superiority of apixaban versus warfarin for prevention of stroke or SE was then assessed and also demonstrated (two-sided p-value = 0.0114). The frequency of bleeding-related adverse events (AEs) with onset during the Treatment Period was lower in the apixaban group compared with the warfarin group (25.2% vs. 32.7%, respectively). The most common bleeding-related AEs by System Organ Class (SOC) were respiratory, thoracic, and mediastinal disorders (apixaban 6.9%, warfarin 8.7%), and gastrointestinal disorders (apixaban 5.9%, warfarin 7.6%). The most common bleeding-related AEs (reported for 5% of more subjects in either treatment group) were epistaxis (apixaban 6.2%, warfarin 7.5%), and contusion (apixaban 3.3%, warfarin 5.3%). Study CV185030 was initiated on December 19, 2006 (first subject visit date) and completed on May 25, 2011 (last subject visit date). The study enrolled 18,201 subjects at 1053 sites in 40 countries. Approximately 40% of the randomized subjects were in Europe, Reference ID: 3132882 Page 5 Clinical Inspection Summary NDA 202155 [apixibanCanada, 19% in Latin America, and 16% in Asia/Paci?c. The comitries with the greatest proportion of randomized subjects (25% of the total) were USA Russia Argentina Canada and Ukraine A total of 8,043 subjects in the apixaban group and 7,933 subjects in the warfarin group completed the study. II. RESULTS (by Site) For approval of this NDA, an initial OSI Consult (dated November 17, 2011) requested the inspection of seven foreign sites, one domestic site, and the Sponsor (BMS). A second OSI Consult (dated March 7, 2012), requested a re-inspection of the Sponsor (BMS), and an inspection of the R0 PPD who had responsibility for most of site monitoring for ARISTOTLE. The original PDUFA date was March 28, 2012. After the Maj or Amendment Letter (issued to BMS on February 28, 2012), the PDUFA action date was extended to June 28, 2012. The Maj or Amendment Letter was issued to give the Review Division more time to review various issues and questions that arose during review of the application. Speci?cally, these issues related to the following: 1) a large munber of medication dispensation errors (disproportionate munber of errors in the apixiban versus the placebo am); 2) potential to unblind by scratching off coverings to bottle labels; 3) potential to unblind because of dissimilar appearance between apixiban and matching placebo; 4) concerns of inadequate sponsor monitoring: and 5) implications of potential fraud at Site 1200 and what it meant for acceptability of data from other sites, in China and elsewhere. Name of Clinical No. of Inspection Dates Final Classi?cation Investigator Subjects Enrolled Site 701 Feb 2 Feb 8, Janos Takacs 37 2012 VAI Mosonmagyarovar. Hlmgary Site 1301 Jan 16 Jan 27, Daniel Raul Vogel 12 5 2012 NAI Buenos Aires, Argentina Site 1742 Jan 30 Feb 4, Cesar Javier Zaidman 73 2012 NAI BUENOS AIRES, Argentina Site 1606 Jan 9 Jan 19, Sandeep Gupta 87 2012 VAI Lucknow. Uttar Pradesh. India Site 0019 Jan 30 Feb 9, Pending Louis Yao 1 1 3 2012 (Preliminaiy Westm- ON Classi?cation VAI) Canada Site 1200 Jan 30 Feb 3, Pending Shiyao W11 35 2012 (Preliminary Shanghai China Classi?cation OAI) Reference ID: 3132882 Page 6 Site 1178 Shulin Wu Guangzhou, Guangdong Clinical Inspection Summary NDA 202155 [apixiban] 44 Feb 6 – Feb 10, 2012 56 Feb 24 – Feb 9, 2012 China Site 796 William Kufs Saratoga, NY 12866 Bristol-MyersSquibb Princeton, New Jersey Sponsor Inspection Dec 28, 2011 – Jan 27, 2012 Bristol-Myers Squibb Princeton, New Jersey Sponsor Re-inspection Mar 14 – April 5, 2012 PPD Development Morrisville, NC CRO/Monitor Inspection Mar 12 – Mar 16, 2012 Pending (Preliminary Classification NAI) VAI Pending (Preliminary ClassificationVAI) Pending (Preliminary ClassificationVAI) NAI Key to Classifications NAI = No deviation from regulations. VAI = Deviation(s) from regulations. OAI = Significant deviations from regulations. Data unreliable. Pending = Preliminary classification based on information in 483 or preliminary communication with the field; EIR has not been received from the field, and complete review of EIR is pending. 1. Janos Takacs Site 701 Karolina Korhaz-Rendelointezet Belgy Tipusu Matrix Egyseg-Altalanos Kardio Regi Vamhaz Ter 2-4 Mosonmagyarovar, Hungary 9200 a. What was inspected: The inspection was conducted in accordance with (b) (4) Compliance Program (CP) 7348.811. Dr. Takacs has IND studies in CDER’s COMIS database and no prior inspections. At this site, 41 subjects were screened, and 37 subjects randomized. Three subjects withdrew consent, and one subject did not meet eligibility criteria. There were three deaths (endpoint events) during the study. An audit of seven subjects’ records was conducted. Subject records were reviewed for adverse events, study endpoints, and reported INR values. The data listings were compared to source documentation and to the eCRFs. Drug accountability records were audited, and financial disclosure statements reviewed. Autopsy reports and death certificates were verified for the three deaths (Subjects #7266, #5159, and #18853). These three deaths were the only reported endpoints at the site. The field investigator reviewed the pull-off drug labels attached to the CRF 800 to verify if unblinding had occurred. Reference ID: 3132882 Page 7 Clinical Inspection Summary NDA 202155 [apixiban] b. General observations/commentary: The field investigator reported that Dr. Takacs and his staff were very good about keeping records. The site used electronic charting in this hospital and the entries were locked after the date of entry. The inspection conducted a full audit of seven subject records. No major discrepancies were noted with respect to data listings, source documents, and e-CRFs. There were no discrepancies noted with respect to endpoint events or drug accountability records. A two-observational Form FDA-483 was issued for the following observations: 1) failure to prepare and maintain accurate case histories with respect to observations and data [21 CFR 312.62(b)]; and 2) an investigation was not conducted according to the investigational plan (21 CFR 312.60). With respect to Observation 1, minor inconsistencies were noted between source records and corresponding CRFs. For example: • • • Medical records document that Subject #3179 began taking metoclopramide on May 19, 2008, whereas the corresponding CRF documents the date as May 15, 2008; Progress notes document that Subject #13194 began taking Rantudil on August 11, 2010, whereas the corresponding study worksheet and CRF document a date of August 7, 2012; Medical records document eight that units of Actrapid insulin was administered to Patient #5159 on September 17, 2008, whereas the corresponding CRF does not document any concomitant medication. With respect to Observation 2, the protocol required that a urinalysis be performed at the final treatment visit. For Subject #900, the field investigator noted that according to the laboratory report, no sample was provided for urinalysis at this subject’s final treatment visit. Dr. Takacs provided an adequate response to the FDA-483 inspectional observations in a letter dated February 20, 2012. The letter included a corrective action plan. c. Assessment of data integrity: Although a few minor regulatory violations were noted during the inspection, they were isolated in nature and unlikely to impact efficacy or subject safety. The study appears to have been conducted adequately at this site, and the data generated by this site may be used in support of the respective indication. 2. Daniel Raul Vogel Site 1301 Instituto De Investigaciones Clin.as Bahia Corinaldesi, Av. Colon 305 Bahia Blanca Buenos Aires, Argentina Reference ID: 3132882 Page 8 Clinical Inspection Summary NDA 202155 [apixiban] B8000FTD a. What was inspected: The inspection was conducted in accordance with (b) (4) Compliance Program (CP) 7348.811. Dr. Vogel has IND studies in CDER’s COMIS database and no prior inspectional history. At this site, 136 subjects were screened, 125 subjects randomized, and 90 subjects completed the study. A total of 35 subjects were discontinued, including 17 deaths (7 in the apixiban arm; 10 in the warfarin arm), and 111 Serious Adverse Events (SAEs) were reported. An audit of 56 (45% of total number of subjects enrolled) subjects’ records was conducted. All study notes were written in Spanish and required English translation by a BMS translator. The field investigator reviewed hand-written source documentation and compared it to entries in the electronic-case report forms. He reviewed inclusion and exclusion criteria for the 56 subjects to ensure eligibility, and he reviewed laboratory assessments, ECGs, vital signs, and physical examinations for accuracy and completeness. b. General observations/commentary: The field investigator reported that for all study records reviewed, the primary and secondary efficacy endpoint data were supported and corroborated with data listings from the sponsor. There were no discrepancies between source documents, CRFs; and data listings. The field investigator found all subjects were appropriately enrolled, and that test article accountability records were accurately documented. INR data was observed as accurate and complete. He reported that the adverse events were reported in a timely manner, and that clinically significant bleeding events were accurately reported. There were a few minor deficiencies observed and discussed with Dr. Vogel at the conclusion of the inspection. Subject #08417 was given the wrong study medication bottle number. This error was found the next day by the monitor and reported as a protocol deviation. The subject was contacted immediately, and they returned to the clinic to receive the correct medication. The field investigator also found a few unreported non-serious adverse events for three subjects (fainting sensation for Subject #14192; dizziness for Subject #10440; and hematuria for Subject #9950). The inspection also found that an ECG for Subject #3402 was not performed at the Month 12 Visit, as per the protocol. No FDA-483 was issued. c. Assessment of data integrity: In general, no significant regulatory violations were noted and no Form FDA-483 was issued. The errors noted are isolated in nature, and not likely to impact data integrity or subject safety. The study appears to have been conducted adequately, and the data generated by this site appear acceptable in support of the respective indication. 3. Cesar Javier Zaidman Site 1742 Ciprec Av. Pueyrredon 1746, 2 A Buenos Aires, Argentina 1119 Reference ID: 3132882 Page 9 Clinical Inspection Summary NDA 202155 [apixiban] a. What was inspected: The inspection was conducted in accordance with (b) (4) Compliance Program (CP) 7348.811. Dr. Zaidman has IND studies in CDER’s database and no prior inspectional history. At this site, 88 subjects were screened, 73 subjects were randomized, and 65 subjects completed the study. There were eight deaths (3 in the apixiban arm, 5 in the warfarin arm) and 35 reported SAEs. An audit of 34 subjects’ records was conducted. All study notes were written in Spanish and required English translation by a BMS translator. The field investigator reviewed hand-written source documentation and compared them to entries in the electronic-case report forms. The field investigator reviewed inclusion and exclusion criteria, test article accountability records, and primary and secondary endpoint events. He reviewed INR measurements and dosage adjustments; adverse events, clinically significant bleeding events, deaths, laboratory assessments, ECGs, vital signs, and physical examinations. b. General observations/commentary: The field investigator reported that all subjects reviewed met inclusion and exclusion criteria. He confirmed that clinically significant bleeding events matched those in the sponsor’s data listings. The eight deaths were located in source documentation and corroborated with the data listings. The drug accountability records were found to be accurate. The field investigator reported that for all study records reviewed, the primary and secondary efficacy endpoint information and data (within the subject’s source records) corroborate the data listings. With respect to INR management, Dr. Zaidman told the field investigator that he adjusted the warfarin dosage to maintain the INR within the targeted therapeutic range of 2.0 – 3.0. One subject was unblinded during the study. Subject #18531 died due to a cerebral stroke (hemorrhage), and the family wanted to know what medication she was assigned. The site contacted the sponsor, and the sponsor allowed the subject to be unblinded. This incident was documented in the progress notes. The field investigator found a few minor violations. He found two subjects with unreported adverse events. Subject #11903 had unreported clinically significant laboratory results of low white blood cells (2.9 x 10 3 gm/dL) and low platelet counts (80,000 platelets per μl), as per laboratory report dated July 1, 2010. Subject #14333 had an unreported right wrist fracture due to a fall, according to progress notes dated November 12, 2009. No FDA-483 was issued. c. Assessment of data integrity: In general, no significant regulatory violations were noted and no Form FDA-483 was issued. The study appears to have been conducted adequately at this site, and the data generated by this site appear acceptable in support of the respective indication. Reference ID: 3132882 Page 10 Clinical Inspection Summary NDA 202155 [apixiban] 4. Sandeep Gupta Site 1606 Mv Hosp. & Res. Centre 314/30 Mirza Mandi Chowk Lucknow, Uttar Pradesh, India 226003 a. What was inspected: The inspection was conducted in accordance with (b) (4) Compliance Program (CP) 7348.811. Dr. Sandeep Gupta had in CDER’s COMIS database and no prior inspectional history. At this site, a total of 104 subjects were screened – this represented 99 unique individuals, as five subjects were rescreened (the protocol allowed subjects to be rescreened for certain laboratory abnormalities such as low platelet counts or low hemoglobin). A total of 87 subjects were randomized. There were 12 screen failures. A total of 76 subjects completed the study. Eleven subjects terminated early including 4 subjects who died, 3 subjects who withdrew consent, 3 subjects lost to follow-up, and 1 subject who was terminated from the study following a SAE. The field investigator conducted a full audit of 30 subject records (~30% of total number of enrolled subjects), including a review of signed and dated informed consent documents (all versions); inclusion and exclusion criteria; subject screening, enrollment and randomization; eCRFs; source documents; treatment/study visit timeframes; documentation of diagnosis and historical treatment of atrial fibrillation and other risk factors for stroke; IVRS confirmation faxes regarding INR values and assignment of investigational product (IP) containers; investigational product accountability logs; prior medication history; concomitant medications; laboratory reports; adverse events (AEs) and SAEs with special attention to bleeding events, stroke and death; documented protocol deviations, and verification of data listings. The field investigator conducted a 100% verification of the data listings for the 30 audited subjects. He corroborated the source documentation with the CRFs and the data listings for: discontinued subjects, including dates and reasons; deaths and narratives; non-serious AEs and SAEs; bleeding events. The field investigator reported that for all study records reviewed, the primary and secondary efficacy endpoint information and data (as documented in the subject’s source records) were supported and corroborated with the data listings from the sponsor; and protocol deviations. In addition, the field investigator randomly selected ten subject records to spot check, focusing on AEs/SAEs, INR values, and IP accountability. b. General observations/commentary: The field investigator issued a 3observational Form FDA-483, for the following deficiencies: Observation 1) Failure to obtain informed consent in accordance with 21 CFR Part 50 prior to conducting study related activities (21 CFR 50.20). Specifically, the inspection found that subjects did not always sign the newest version of the Informed Consent document at their subsequent visit. This was noted Reference ID: 3132882 Page 11 Clinical Inspection Summary NDA 202155 [apixiban] to be a systemic issue, and occurred specifically with ICD Version 3.0 (at least 15 subjects), Version 3.1 (at least 8 subjects), and Version 4 (at least 32 subjects). The field investigator observed that this observation had been noted by the monitor and documented by the monitoring reports. He also observed that Dr. Gupta had submitted protocol deviation letters to the Ethics Committee (EC) about this issue. After re-training at the site, study subjects were re-consented in a timely manner at their next scheduled visit. At the time the site began screening subjects, IC Version 3, dated July 16, 2008 was in use. There were subsequent updated, translated Version 3’s for Hindi and Urdu. There were four subsequent versions to Version 3 used: Version 3.1, Version 4, Version 5 and Version 6. The changes made to Version 3.1 (dated April 29, 2009) were administrative in nature, pertaining to an address change for the site, and a number change for one of the EC members. The changes to Version 4 (dated September 3, 2009) reflected a change in the number of study centers from 800 to 1000 sites and an increase in the number of participants from 15,000 to 18,000. The changes also pertained to compensation for participants. These changes are not likely to impact subject safety. In his February 3, 2012 response letter to the Form FDA-483, Dr. Gupta outlined a corrective action plan to prevent recurrence of this observation. Observation 2) Failure to prepare or maintain accurate case histories with respect to observations and data pertinent to the investigation [21 CRF 312.62(b)]. The following inconsistencies were noted in source documents: a) Subject 08624 was seen in the ICU with an SAE of congestive heart failure at the 8 Month Visit on Sept 2, 2009. The source notes documented no (new) AE or SAE. However, this SAE was documented on other source records maintained in the subject’s binder and reported via CRF to the Sponsor. b) Subject 14891 had an SAE of recurrent vomiting and weakness occurring on September 14, 2009 but the source notes of October 5, 2009 report no AE or SAE. This SAE was documented on other source records within the subject’s binder and was reported via CRF to the Sponsor. c) Subject 20516 experienced pain in the wrist and joints at the Week 3 Visit occurring on February 8, 2010, although the source record visit notes documented that no AE or SAE was reported at this visit. This AE was documented in other source records within the subject’s binder and was reported via a CRF to the Sponsor. d) Subject 11688 had two IVRS FAX INR confirmation forms for 2 separate INR values at the 5 month Visit on September 1, 2009 - one taken at 10:46 and another at 14:46. Study records did not explain which value was ultimately accepted by the IVRS Manager. Observation 3): An investigation was not conducted in accordance with the investigational plan (21 CFR 312.60). Reference ID: 3132882 Page 12 Clinical Inspection Summary NDA 202155 [apixiban] Specifically, the field investigator identified more than 46 subjects who had multiple (at least 3) study visits performed > 7 days outside the protocol-defined window (range was >7 to 25 days outof-window). Note: The protocol allowed subjects to be seven (7) days outside the window visit. The field investigator discussed this topic at length during the inspection. Dr. Gupta explained that in India, a patient never goes to the clinic alone, and is always accompanied by a friend, relative or neighbor. Dr. Gupta also explained that the study site is only open during the day, during times that most relatives are at work. The field investigator observed that Dr. Gupta had submitted protocol deviation notification letters to the Ethics Committee to inform them of these out-of-window visits. This action had been recommended by the monitor. The field investigator collected documentation of conversations during the consent process whereby Dr. Gupta told subjects he would send a vehicle for them. The field investigator noted that most of the out-of-window visits occurred towards the middle to end of the study, versus at the beginning. The inspector noted that the PI knowingly maintained subjects at a low time in therapeutic range (TTR). When subjects in the warfarin arm are insufficiently anticoagulated, it may result in the finding that the study drug (apixiban) is noninferior to placebo since a higher number of events would be expected to occur in the warfarin arm. Dr. Gupta had many discussions with the monitor, and additional training was provided about ways to improve INR control. Dr. Gupta stated that it was his decision to increase or decrease dosages based on patient care and safety. Dr. Gupta explained that if a patient appeared to have no difficulties with the current dosage, he did not like to make adjustments – because, if bleeding occurred by increasing the dose, it might take longer for the subject to get back to the clinic and obtain medical care. With respect to bleeding events, the field investigator reported that subject records adequately documented if bleeding occurred. If a bleeding event occurred, the study medication would be stopped until a full evaluation was conducted. The field investigator observed that the Sponsor was notified of any bleeding event within 24 hours, and the documentation was then faxed to PPD. The field investigator also reported that adverse events appeared to be well documented within the source records and on the CRFs. The field investigator did not observe any discrepancies between source documents, data entered on the CRF, and the Sponsor data listings. The field investigator observed that the majority of SAEs were bleeding events, stroke, and death. For the four subjects who died while participating in this study, the cause of death was not known for two (Subjects 8385 and 16832). The field investigator reported that Dr. Gupta and the site staff were provided additional training on the above issues. The site hired additional staff to respond to the numerous queries and enter information into the eCRFs. This action had been advised by the monitor. Dr. Gupta provided a response to the Form FDA-483 observational items, in a letter Reference ID: 3132882 Page 13 Clinical Inspection Summary NDA 202155 [apixiban] dated February 3, 2012. In his response letter, he stated that he is committed to conducting clinical research to comply with GC and all regulatory requirements. He outlined a comprehensive corrective action plan to correct the above de?ciencies. OSI considers his response adequate. c. Assessment of data integrity: Regulatory Violations were observed at this site, but these are not considered likely to importantly impact data integrity. In general, the source documents were noted to be in good order and study procedures, adverse events, and other source information were well documented. There were few discrepancies between the source records, the and the background materials. As was previously discussed with the review division, OSI will defer to the review division on the issue of assessment of the potential effect of low TTR on overall study outcome. Otherwise, the study appears to have been conducted adequately, and the data generated by this site may be used in support of the respective indication. 5. Louis Yao Site 0019 230-1920 Weston Road Weston, ON Canada M9N 1W4 a. What was inspected: The inspection was conducted in accordance with Compliance Program (C P) 7348.811. At this site, 122 subjects were screened, 113 randomized, 30 subjects were listed as ?early discontinuations: (17 were deaths), and 83 subjects completed the study. The ?eld investigator completed a data audit of 44 subject records that included: review of inclusion and exclusion criteria; ef?cacy assessments; corroboration of electronic case report forms with source documents and data listings; adverse event reporting; drug accountability records, and protocol deviations. b. General observations/commentary: In discussions with the ?eld investigator, she did not report any evidence of miderreporting of ABS and did not provide information about her corroboration of the ef?cacy endpoint data listings with the source data at the site. If any issues are observed after the review of the EIR, 031 will provide an addendum to this C13. The ?eld investigator observed that Dr. Yao co-mingled the patient study records with the patient?s non-study medical records. We, helped Dr. Yao with the conduct of this study, including the completion of documents. According to her CV, (m6) . She was a Cardiology Research Coordinator in other studies. W6) She is also listed as study coordinator for several studies. The ?eld investigator reported that Dr. Yao signed every Informed Consent Document that was reviewed. According to signatures on documents, speci?c responsibilities were as Reference ID: 3132882 Page 14 Clinical Inspection Summary NDA 202155 [apixiban] We) signed documents documenting the reporting of SAEs, and (m6) faxed the INR values. When directly asked about the roles of (m6) in the study, Dr. Yao stated that they pretty much did "secretarial wor and were never directly involved with screening, INR readings, dosage adjustments, or identifying SAEs. He also stated that both (m6) completed several BMS online-training modules prior to the beginning of the study. told the ?eld investigator that she has been helping Dr. Yao, as a Study Coordinator, since she left her job working? with At the present time, Dr. Yao had only (6) helping him with the ARISTOTLE study. A 2-observational DA-483 was issued the inspection for the following de?ciencies: 1) an investigation was not conducted in accordance with the investigational plan (21 CF 312.60); and 2) failure to prepare or maintain adequate and accurate case histories with respect to observations and data pertinent to the investigation. follows: With respect to Observation 1, the following was noted: on February 7, 2009, Subject 09747 experienced a SAE and Dr. Yao unblinded the treatment for that subject only after the subject?s daughter demanded to know what her father was taking. This was contrary to the protocol?s requirement which states, in part, that the blind should be broken only for a medical emergency. 081 Reviewer Comments: While unblinding the treatment for Subject 09747 was a protocol violation, it was an isolated event and should not impact the integrity of data generated by the site. With respect to Observation 2, the following was noted: 0 For Subject 04187, the progress notes for all visits between June 11, 2008 and December 15, 2009 are signed and dated March 6, 2010; 0 For Subject 00259, progress notes for visits between June 22, 2007 and August 1, 2007 are signed and dated October 11, 2007; 0 For Subject 02086, the progress notes for visits between February 13, 2009 and October 13, 2009 are signed and dated March 6, 2010. OSI Reviewer Comments: Reportedly, Dr. Yao did not review and sign all entries to progress notes for all visits contemporaneously - he would often go back months later to sign them. 0 ECG printouts for Subject 00199 and 02086 were not signed by Dr. Yao. 0 Visit Date ?elds in the eC RF were not always completed. For example, the Screening Visit date ?elds for Subjects 16108, 15551, and 09747 were not completed. 0 Electronic signatures in the eC RF were post-dated, between one month to over one year from the visit date for some subjects. Based on this fmding, it appears that the PI did not electronically sign off on until much later than the subject visit date. c. Assessment of data integrity: This site had very high enrolhnent. In general, the regulatory violations noted are isolated in nature based on available information and would Reference ID: 3132882 Page 15 Clinical Inspection Summary NDA 202155 [apixiban] not be expected to impact data integrity. The data generated by this site may be used in support of the respective indication. PLEASE NOTE: The EIR was not available at the time this CIS was written. The observations noted are based on preliminary communications with the field investigator, and the Form FDA 483. An inspection summary addendum will be generated if conclusions change upon receipt and review of the EIR. 6. Shiyao Wu Site 1200 Shanghai 9th Peoples Hosp. Affiliated To Kan Chen M.D. S. J.T. U. S. M. No. 639, Zhi Zao Ju Road Shanghai, China 200011 a. What was inspected: The inspection was conducted in accordance with Compliance (b) (4) Program (CP) 7348.811. Dr. Shiyao Wu has IND studies in CDER’s COMIS database, and no prior inspections. For this study, a total of 37 subjects were screened at the site and 35 subjects enrolled. There were two screen failures; five deaths, two withdrawals, and one subject lost to follow-up. Two field investigators were assigned to cover this inspection they divided their time between covering the routine Compliance Program and addressing the "for cause" addendum to the assignment issued January 20, 2012. b. General observations/commentary: A multi-part, one item FDA-483 was issued during the inspection for the following: The study was not conducted in accordance with the study protocol (21 CFR 312.60). Specifically, • The principal investigator as team leader failed to properly supervise the sub-investigators. The principal investigator failed to ensure prompt reporting of study progress in the CRF as reported by the study monitors. Reference ID: 3132882 • User accounts to gain access to computers were shared between two sub(b) (4) investigators (Dr. ), which was prohibited by Protocol Section 9.1.3, which states “user accounts are not to be shared or reassigned to other individuals.” Further, the principal investigator (b) (4) allowed Dr. to use his username to enter data into the e-CRF. • The principal investigator failed to ensure an adequate number of qualified staff and adequate facilities for the duration of a trial to conduct the trial properly. The internet connection at the study site was too slow to allow staff to enter data expeditiously, and some staff took data home on an unsecure drive to complete the task. One sub-investigator stated that it took him at least three hours to completely enter the data for one screening visit. Page 16 Clinical Inspection Summary NDA 202155 [apixiban] Additional Information on BMS Findings at Site 1200: A for-cause inspection assignment addendum was issued in order to confirm allegations of suspected misconduct including alteration of subject records during a pre-inspection audit at Site 1200 (Shiyao Wu) in China. (Please see Background for further details.) (b) (6) The observations of alleged misconduct came from a PPD monitor ( who was at Site 1200 helping to prepare the site for the upcoming FDA inspection. Specifically, (b) (6) Ms. noted that while reviewing a chart which contained records of outpatient visits, she recalled that during an earlier monitoring visit, these same outpatient records (b) (6) contained handwritten comments. During the pre-audit visit, Ms. noted that the (b) (6) printed records had no handwritten comments. In addition, a PPD employee ( ) told her that he had altered data containing subject records on a USB drive at the (b) (6) direction of Ms. BMS site manager. The records were altered in order to cover-up GCP violations which had occurred at the site. In their response to an February 15, 2012 OSI Information Request (IR) BMS provided (b) (6) information stating that Ms. worked as Site Manager for the BMS ARISTOTLE study and that she visited 18 of the 36 Chinese sites while the study was underway. Her duties included conduct of co-monitoring visits if needed; site recruitment/enrollment; resource needs; and following up on quality issues noted by the Site Monitor. She reviewed site visit reports written by the site monitor. During preinspection visits, she reviewed study files and subject medical records and discussed (b) (6) issues with study personnel. The sites in China where Ms. worked were: 1180, 1182, 1199, 1206, 1207, 1221, 1223, 1246, 1247, 1266, 1547, 1548, 1549, 1550, 1552, 1555, 1556, and 1558. She worked on no sites outside China. To prepare for the (b) (6) upcoming FDA inspections, Ms. worked at Sites: 1200 and 1178. (b) (6) PPD employee performed site monitoring responsibilities for the following Sites: 1200, 1198, 1168, 1244, and 1287. He worked at no non-China sites. OSI Reviewer Comment: FDA inspection at this clinical revealed that multiple individuals at this site were using the principal investigator’s PIN number to access an electronic data system at the site used for eCRF entry. In addition, the principal investigator allowed one of the subinvestigators to use his username to enter data in eCRFs. The PI rarely evaluated subjects personally. These violations represent serious compliance violations, since anyone who entered data under another’s identification is committing a fraudulent act. The integrity of the study records are compromised, as is the data contained in those records. In addition to the FDA inspectional findings, BMS submitted evidence based on their own investigation that fraudulent activity occurred at this site. The BMS (b) (6) Site Manager assigned to this site to prepare for the FDA inspection (b) (6) altered subject records and directed a PPD employee Mr. to alter subject data on a USB drive. These fraudulent activities were reported to occur in an effort to conceal GCP violations at Site 1200. BMS was queried as to the duties of these two individuals during the conduct of the trial as well as which sites they worked at during ARISTOTLE. As detailed above, these two Reference ID: 3132882 Page 17 Clinical Inspection Summary NDA 202155 [apixiban] individuals worked at 23 of the 36 ARISTOTLE sites located in China and one additional site (1178) in preparation for the FDA inspection. c. Assessment of data integrity: The inspection at this site found that Dr. Wu failed to maintain adequate oversight of the study. This was evidenced by sub-investigators using Dr. Wu’s PIN to make entries to records (shared user accounts were prohibited by the protocol). (b) (4) Dr. , a sub-investigator, continued using Dr. Wu's username and password even after he had his own account. In addition, due to slow internet speed, study coordinators frequently took records home on an unsecured USB drive to complete entering data into CRFs. These activities (shared user accounts and unsecure USB drive) could potentially compromise the integrity of the data generated by this site. Dr. Wu admitted that he did not regularly see study subjects – only if there was a special case. In addition to these serious inspectional observations, BMS informed that employees of BMS and PPD had manipulated study records at this site. The extent of data manipulation at this site remains unclear. For these reasons, OSI finds that the reliability of data from this site could not be verified. OSI recommends that data generated by this site be considered unreliable and not be used in support of the NDA. PLEASE NOTE: A final review of the EIR was not copmleted at the time this CIS was written. The observations noted are based on preliminary review of the EIR, and communications with the field investigator, and the Form FDA 483. An inspection summary addendum will be generated if conclusions change upon final review of the EIR. 7. Shulin Wu Site #1178 Guangdong General Hosp. No. 1, The 2nd Zhongshan Road Guangzhou, Guangdong China 510080 a. What was inspected: The inspection was conducted in accordance with Compliance Program (CP) 7348.811. At this site, there were 54 subjects screened and 44 subjects enrolled. A total of 31 subjects completed the study. The field investigator completed a review of Informed Consent Documents for all 44 subjects. Comparison of source documents from study visits was made with the eCRFs for seven subjects selected from the beginning, middle and end of the study. This included a review of all the inclusion and exclusion criteria, efficacy assessments, corroborating the source documentations with eCRFs and data listings, drug accountability records, and adverse event reporting. The CRFs were reviewed for completeness and accuracy. Five (5) additional subjects who had been hospitalized during the study were reviewed for serious adverse event reporting and concomitant medication documentation. b. General observations/commentary: All study records were well organized. There was adequate documentation to ensure all subjects were alive and available for the duration of their stated study participation. Hospital inpatient records were available electronically. A Reference ID: 3132882 Page 18 Clinical Inspection Summary NDA 202155 [apixiban] study nurse had been designated for this trial and was responsible for all electronic data entry. Sub-investigators reviewed and signed (approved) the entries. Data was usually entered the same day or within 24 hours. Safety reports were entered immediately. A few minor discrepancies were noted: • Subject 10957 was taking Digoxin 0.125 mg QD as noted in the source documents but this medication was not included in the screening medication log on the eCRF. • Subject 14672 was hospitalized July 6, 2009 – July 21, 2009, for cerebral infarction. During this time, the subject was administered one dose of clopidogrel, which is a restricted but not a prohibited medication per the protocol. No other problems were noted and no unreported serious adverse events were observed. c. Assessment of data integrity: Although minor discrepancies were observed, no FDA483 was issued. However, because of the alleged misconduct and other inappropriate activities that occurred at Site 1200, and that this site (1178) was visited by BMS Senior (b) (6) Clinical Site Manager who reportedly inappropriately influenced the study records at Site 1200 in China, OSI recommends that the data from Site 1178 (as well as all (b) (6) sites in China who were conducting the ARISTOTLE study where either or (b) (6) worked) be excluded from the primary efficacy and safety analysis. PLEASE NOTE: The EIR was not available at the time this CIS was written. The observations noted are based on preliminary communications with the field investigator, An inspection summary addendum will be generated if conclusions change upon receipt and review of the EIR. 8. William Kufs Site 796 Saratoga Cardio. Assoc., PC 6 Care Lane Saratoga, NY 12866 a. What was inspected: This domestic site inspection was conducted in accordance (b) (4) with Compliance Program (CP) 7348.811. Dr. William Kufs has listed in CDER’s database, and no prior inspectional history. At this site, 65 subjects were screened, 56 subjects were randomized, and 44 subjects completed the study. There were five deaths and seven subject discontinuations. Most discontinuation occurred because of subject request. One subject was discontinued due to poor compliance. An audit of 25 subjects’ records was conducted during the inspection. The field investigator reported that he reviewed subject clinic records and subject files as part of source documentation. There were no paper CRFs, so the field investigator corroborated data entered directly onto the eCRF with the source documentation and the sponsor’s data listings. Reference ID: 3132882 Page 19 Clinical Inspection Summary NDA 202155 [apixiban] The following items were audited: adherence to inclusion and exclusion criteria; validation of adverse events; bleeding events; visit schedules and INR testing; protocol violations; investigational product (IP) labeling; drug accountability for eight subjects; potential unblinding (as per assignment request); validation of primary and secondary endpoints, discontinuations; deaths; and informed consent documents for all randomized subjects. b. General observations/commentary: The following items were cited in the FDA Form-483: 1) Investigation not conducted according to the investigational plan (21 CFR 312.60). Specifically, two subjects who were enrolled should have been excluded according to the protocol specified exclusion criteria. Subject 04331 was taking 325 mg aspirin/day (protocol specifies exclusion if the subject is taking more than165 mg aspirin per day); and Subject 15718 was not screened for platelet count (protocol excluded subjects with platelet count >100,000/mm3) OSI Reviewer Comment: The CI’s response letter dated February 29, 2012 states that aspirin was stopped at study entry for Subject 04331; therefore, this citation appears to be inaccurate. With respect to Subject 15718, the CI’s response letter states that Subject 15718 had ‘platelet clumping’ and therefore, the core laboratory elected not to report a platelet count, and not to report it as abnormal. This was an oversight by the CI. The protocol prohibited subjects with a platelet count < 100,000 μL (later amended to < 90,000 μL) due to a higher risk of bleeding. This was an isolated finding, and unlikely to affect data integrity. 2) Investigational drug disposition records were not adequate with respect to quantity [21 CFR 312.62(a)]. The field investigator reviewed drug accountability records of returned medication for eight subjects, and found discrepancies for five of those subjects. For example, for Subject 00104, and Bottle #424040, the drug accountability log documents 52 tablets were returned on September 25, 2009 whereas the subject’s personal record documents that 58 tablets were returned. For Subject 00104 and Bottle #103585, the drug accountability log documents 87 tablets returned, whereas the subject’s personal records documents 0 tablets returned. OSI Reviewer Comment: The field investigator found five of eight subjects with a discrepancy between the amount of study drug returned during one of the subject’s scheduled visits (discrepancies were between the ‘Record of Study Medication – Warfarin/Placebo record’ and the ‘Individual Subject Clinical Supplies Inventory Warfarin’ record). In his response letter, Dr. Kufs commented that at the beginning of the study, subject diaries were encouraged but not required. These errors appear to be discrepancies between the Subject diaries and the drug accountability log, but not discrepancies between the drug accountability log and the number of pills documented in the eCRF. This finding is not likely to importantly impact study drug compliance and data integrity. Reference ID: 3132882 Page 20 Clinical Inspection Summary NDA 202155 [apixiban] 3) Failure to report to the sponsor adverse events that may be regarded as caused by the investigational drug [21 FR Speci?cally, the ?eld investigator f01md 12 of 36 serious adverse events (SAEs) that were reported to the Sponsor later than the 24?hour timeframe required by the protocol. For example, Subject 131 experienced a myocardial infarction and was hospitalized on (b This event was not reported to the sponsor Imtil (4 days later). Subject 167 was hospitalized with anemia on mm) This event was not reported to the sponsor until (we) (10 weeks later). Subject 557 had gastroenteritis on August 19, 2009 and was hospitalized on mm. This SAE was not reported to the sponsor until We) (3 weeks later). 081 Reviewer Comments: In his response letter dated February 29, 2012, Dr. Kufs stated that he failed to ?le some SAEs within 24 hours because he was not noti?ed of the SAE (in some cases) until the subject returned to the clinic for a scheduled visit. For other cases, the subjects had hospital admissions for the event, and the hospital physicians would not notify him 1mtil the subject was discharged. Dr. Kufs stated that he would then report the SAE within 24 hours of being noti?ed. The hard copy SAE reports that Dr. Kufs submitted with his response letter substantiates that although the onset date was earlier, Dr. Kufs did in fact report the events within 24- hours of being noti?ed. OSI considers his response acceptable. c. Assessment of data integrity: Minor regulatory Violations were found during the inspection at Dr. Kufs? site. These violations were isolated and adequately addressed by Dr. Kuf in his response letter. The study at Site #796 appears to have been conducted adequately, and the data may be used in support of the respective indication. 9. Bristol Myers Squibb Research Development PO. Box 4000 (Mail Stop: D22-05) Princeton, NJ 08543-4000 Sponsor Inspection (December 28, 2011 January 27, 2012) a. What was inspected: This was a high priority inspection in accordance with Compliance Program 7348.810. Two ?eld investigators were assigned to cover this inspection. They reviewed monitoring reports for the eight clinical investigator sites and documented systemic issues at the sites; they identi?ed sites that had poor INR control and how improvements were being made; they looked at the adjudication process and commented on how well communication was conducted between the site, PPD, and the sponsor; they veri?ed how well adverse events, especially bleeding events, were reported to the sponsor; they also veri?ed how and if endpoint events were reported to the sponsor. Reference ID: 3132882 Page 21 Clinical Inspection Summary NDA 202155 [apixiban] The field investigators asked the sponsor for information about the change in their decision to not collect the CRF 800’s from the site. They asked the sponsor questions about the potential for the site to break the blind by scratching off the label covering that contained the product name. This was the portion of the label affixed to the CRF 800 page. b. General observations/commentary: At the end of the inspection, a 2-observational Form FDA-483 was issued for: 1) failure to ensure proper monitoring of the study and ensure the study is conducted in accordance with the investigational plan (21 CFR 312.50); and 2) an investigator who did not comply with the general investigational plan was not promptly brought into compliance [21 CFR 312.56(b)] With respect to Observational Item #1, the field investigators reported inadequate monitoring at Site 1200 (China) and Site 1606 (India), as evidenced by review of site monitoring reports (MVR). As reported, Site 1200 had 4 different monitors, and there was considerable variation in the quality of the MVRs from the various monitors. Since PPD did not have a presence in China at the time of the study, BMS hired contractors paid by PPD but who were managed by BMS in China to conduct monitoring at Site 1200. The monitors used BMS forms for completing monitoring visits. The monitoring reports revealed that Site 1200 did not complete documents (CRFs and query resolutions) in a timely manner. For 18 of 25 monitoring visits, the field investigators found that the monitors failed to complete those sections of the MVR that indicated the names of individuals from BMS or PPD at the site during the monitoring visits. For 4 of 18 visits, the field investigators reported that the Source Data Verification Documentation section which verified if source data corroborated with the CRF entries was not completed. While reviewing monitoring reports for Site 1606, the field investigators observed documentation that Site 1606 consistently had difficulty with entering data from source records onto the eCRF in a timely manner. The field investigators also observed the following: Site 1606 did not respond to data queries in a timely manner; Site 1606 did not always inform the IRB of SAEs in a timely manner; Site 1606 had numerous out of window visits; Site 1606 showed problems with providing the correct versions of the ICD for the patient to sign; Site 1606 had numerous investigational products (IPs) issues, including bottles not returned by the subjects and bottles lost by the site. The field investigators observed that despite the apparent multitude of IP issues described by the MVR, that the monitor only documented one time (in an MVR towards the end of the study) that there were many IP bottles not returned by the subjects. Despite these problems, the site was approved by the sponsor for an increase in enrollment and was granted two extra study coordinators. The sponsor also provided more staff at the site to help keep up with data entry. The field investigators reported that because Site #1606 had so much difficulty maintaining subjects within the targeted therapeutic range, that BMS provided refresher training on the importance of INR control at the site. Reference ID: 3132882 Page 22 Clinical Inspection Summary NDA 202155 [apixiban] With respect to their decision to maintain CRF 800’s on site, BMS explained to field investigators during the inspection, and in their written response, dated February 17, 2012, that in July 2009 a decision was made by BMS to stop the process of collecting CRF pages with labels affixed and forwarding them for processing. Instead, those pages were to be kept with other study documents and made available for monitoring. They did this in order to retire the process of collecting CRF label pages for new studies. BMS stated that regardless of where the labels resided the monitoring of key aspects of the study, including drug accountability, and verification of protection of the blind, did not change. BMS reported that all monitors were instructed to review the panel stickers on the CRF 800 and identify and report if any unblinding occurred. The field investigators also reported that monitors from 2 of the 8 sites reviewed failed to ensure that sites recorded the amount of IP returned and final IP disposition. At Site 1200, the field investigators reported that 43 kits (11 apixaban kits and 32 warfarin kits) were documented as lost by subjects. This amounts to 17 of 35 subjects enrolled losing at least one kit during the course of the study. The loss of these kits was rarely documented in the MVR. At Site 1606, the field investigators noted illegible and missing records of the final disposition of IP. With respect to Observation 2, at Site 1606, the field investigators found that the PI was consistently late in entering data into the eCRF, was late in answering data queries, and was late in reviewing and approving eCRF data. The sponsor provided the site with extra study coordinators to assist the site with efforts of increased enrollment. c. Assessment of data integrity: Although regulatory violations were noted during the inspection specific to several problematic sites, overall the study appears to have been conducted adequately by the sponsor, and the data may be used in support of this NDA. PLEASE NOTE: The EIR was not available at the time this CIS was written. The observations noted are based on preliminary communications with the field investigator, and the Form FDA 483. An inspection summary addendum will be generated if conclusions change upon receipt and review of the EIR. Sponsor Re-Inspection (March 14 – April 5, 2012): During the period of March 14 – April 5, 2012, FDA field investigators re-inspected the sponsor at the BMS facility in Lawrenceville, N.J. On March 14 and 15, the investigators were accompanied by Martin Rose, Medical Officer and Alison Blaus, Senior Regulatory Project Manager in the Division of Cardiovascular and Renal Drug Products. BMS responded to the Form FDA-483 observations in a letter dated April 26, 2012. a. What was inspected: The sponsor re-inspection was conducted to more fully address the adequacy of sponsor oversight and monitoring during the conduct of the ARISTOTLE trial. The adequacy of oversight and monitoring was questioned based Reference ID: 3132882 Page 23 Clinical Inspection Summary NDA 202155 [apixiban] on the findings of alleged fraud at Site 1200, as well as the information noted by the review division that some subjects received the wrong drug (medication dispensation errors), based on mismatches between container codes entered in the eCRF at the site and the IVRS assignment. The sponsor re-inspection was also conducted to address questions about potential unblinding and the high rate of medication dispensation errors noted by the review division. The field investigators were asked to review monitoring reports for the following five sites selected randomly (high enrollment), and summarize ongoing issues found at those sites: • • • • • Site #310 (Kharman, Germany, 192 subjects) Site #1438 (Horacek, Germany, 130 subjects) Site #872 (Chzhov, Russia, 110 subjects) Site #903 (Nierop, Netherlands, 98 subjects) Site #463 (Hong, South Korea, 105 subjects) b. General observations/commentary: A 2-observational FDA-483 was issued at the conclusion of the inspection for the following: Observation 1: An investigator who did not comply with the general investigational plan and regulatory requirements was not promptly brought into compliance. Specifically, the Sponsor failed to ensure compliance of study conduct in two out of five sites reviewed through clinical site monitoring when the sites were continually found to conduct the study inadequately (according to the protocol, applicable regulatory requirements and applicable laws of the country in which the research was conducted). For example: 1. The Principal Investigator for Site #903 failed to (a) ensure IP compliance and accountability for at least one time point in 38 out of 67 subjects; (b) perform all protocol required procedures; (c) sign the medication prescriptions as required by Dutch law; (d) obtain informed consent forms prior to the required procedures; (e) delegate and approve a sub-investigator prior to study participation; (e) ensure the required protocol visits were performed within the allowed window; and (f) store the PI under appropriate conditions. 2. The PI for Site #463 failed to: (a) ensure IP compliance in 33 of 72 subjects; (b) perform all protocol required procedures; (c) delegate and approve a Study Coordinator to participate in the study; (d) obtain updated versions of the ICD; and (e) ensure the required protocol visits were performed within the allowed window. BMS Response: In their response letter, dated April 26, 2012, BMS states that despite efforts to monitor, there were areas where site performance was not as expected, but they do not believe these issues affected the validity of the data in this trial. BMS audited Site #903 during the trial when 35 of 65 subjects had been randomized. Improvement in site performance was noted following the BMS audit. Concerning the finding of instances where an ECG or laboratory test was not performed as required (Site #463), BMS states these Reference ID: 3132882 Page 24 Clinical Inspection Summary NDA 202155 [apixiban] missed procedures were sporadic across subjects and visits. There were no adverse events temporally associated with the missed procedures. OSI Reviewer Comment: OSI considers the BMS response acceptable. Observation 2a: failure to ensure proper monitoring of the study. Specifically, the field investigator observed that the Site Monitoring Plan (SMP) for this study was approved January 3, 2008. However site initiation visits (SIV) and interim monitoring visits occurred prior to the approval date for the SMP. Examples are as follows: • Site initiation visits (SIV) occurred on July 26, 2007 (for Site #310), August 10, 2007 (for Site #463), and September 11, 2007 (for Site #872). These monitoring visits occurred before approval of the SMP (January 3, 2008) • Six interim monitoring visits occurred between September 20, 2007 and December 25, 2007 for Site #310 and Site #872, respectively. These monitoring visits occurred before approval of the SMP (January 3, 2008). BMS Response: In their response letter, BMS states that these site visits to the above sites were made using a working draft of the SMP (Version 6, dated July 11, 2007). The working draft included all information necessary to conduct quality SIVs and interim site monitoring visits. In addition, BMS explained that the SOP for developing the SMP allowed for the SMP to be developed in four stages to accommodate the study progress milestones. OSI Reviewer Comment: OSI considers the BMS response acceptable. Observation 2b): According to the IP Storage Accountability and Reconciliation section of the SMP, the Clinical Research Associate (CRA) should confirm that the accountability log and the inventory agree. The site personnel did not count the study drug during each scheduled visit, and monitors (a/k/a CRAs) at the site failed to review the drug accountability logs during their monitoring visits to ensure the number of pills documented as returned corroborated with the number of pills on hand at the site. BMS Response: In their response, BMS acknowledged a systematic issue at one site (#903), where pill counts were performed at 81% of visits. However, at Site #463, drug accountability was performed at 98% of visits, and instances where it was not performed were isolated and sporadic. The frequency of missed pill counts was noted to improve throughout the study, as noted by compliance checks documented in monitoring visit reports. OSI Reviewer Comment: OSI considers the BMS response acceptable. Observation 2c): According to the SMP, the IP labels which were affixed to CRF 800 pages should be kept at the clinical sites effective as of July 2009. This was not followed. For example, approximately 72 out of 1,034 sites returned the IP label CRF 800 pages to the Sponsor between August 14, 2009 to September 19, 2011. Reference ID: 3132882 Page 25 Clinical Inspection Summary NDA 202155 [apixiban] BMS Response: In their response letter, BMS states that there was a change regarding the CRF label pages while the study was ongoing, and confusion existed as to whether these pages were to be returned or not. OSI Reviewer Comment: OSI agrees that there was a communication problem and confusion arose on how to handle the CRF 800 pages (containing the affixed labels for unblinding) throughout the conduct of the study When the CRF 800 pages were maintained at the site, either in the pharmacy or elsewhere, the labels could theoretically be scratched off and unblended. However, according to the field investigators, the site monitors did not identify unblinding as an issue during monitoring. OSI considers the sponsor’s response adequate. In addition to the above observational items listed on the FDA-483, several relevant verbal observations were presented to the sponsor at the end of the inspection. In brief, they were the following: One item pertained to medication dispensation errors. In late January 2012, the review division raised questions about the description in Section 4.3 of the ARISTOTLE Clinical Study Report that 7.3% of apixaban subjects and 1.2% of warfarin subjects had received, at some point in the study, a study medication container of the incorrect type. It was possible that these medication dispensation errors, especially when a subject receives either double active or double placebo can lead to safety events. The review division asked the sponsor in an Information Request (IR) to describe, if any, the impact of these errors (whereby subjects received the wrong medication) on the integrity of data. During a February 2012 meeting, the review division asked the sponsor for a further quantification of medication errors to assess whether the proportion of previously reported medication errors could have been under-estimated (0.38%), and to assess the impact of medication errors on the interpretability of the study results. BMS had examined all container numbers entered into the web-based eCRF, and calculated that 0.38% of all labels did not match the labels assigned in the IVRS. In their response letter dated April 26, 2012, BMS stated that upon further evaluation of these types of errors the rate of study medication treatment errors in this trial was low (” 0.1% of study medication dispensing) and balanced across treatment arms. The true proportion of errors in medication dispensations was likely even lower than the 0.38% that was derived earlier, so that the containers dispensed to the subjects very closely matched what was intended by IVRS. Further, BMS states that the actual impact on patient safety does not appear to have been significant. Only four primary efficacy or safety outcome events occurred during or within 90 days after a period of treatment with double active or double placebo. Further, the low rate of study medication treatment errors and their random nature along with the results of sensitivity analyses suggest that the study conclusions are robust. A second item pertained to reevaluating the monitoring report template to ensure that enough information was captured. The template used by PPD was very detailed, whereas the template used by BMS used a Yes/No question format. BMS stated that critical GCP principles were captured within the BMS monitoring report, but that they would undertake a Reference ID: 3132882 Page 26 Clinical Inspection Summary NDA 202155 [apixiban] re-evaluation of the content of the report as part of ongoing improvements. It is noted that the PPD (detailed) template was used for the majority of sites in the ARISTOTLE trial. c. Assessment of Data Integrity: Through review of monitoring reports, the field investigators observed ongoing issues relating to IP compliance and accountability at two of five sites. For the study as a whole, pill counts were performed at 92% of all study visits, and subject compliance was noted to be equally as high. These IP issues are unlikely to directly impact data integrity. Though other minor regulatory violations were found during the reinspection at the Sponsor site, the issues were adequately addressed by the BMS response letter. OSI considers the data acceptable in support of the respective indication. PLEASE NOTE: The EIR was not available at the time this CIS was written. The observations noted are based on preliminary communications with the field investigator, and the Form FDA 483. An inspection summary addendum will be generated if conclusions change upon receipt and review of the EIR. 10. PPD Development 3900 Paramount Pkwy Morrisville, North Carolina 27560 a. What was Inspected: This was a high priority inspection in accordance with Compliance Program 7348.810. There were 1053 clinical investigator sites in 40 countries for the ARISTOTLE trial. The inspection focused on PPD monitoring activities at the eight clinical investigator sites referenced in the original assignment. BMS had responsibility for managing two of the eight sites selected to review. This included the two China sites, Site #1178 and Site #1200. During the inspection, the field investigator reviewed selected monitoring reports from six of the eight sites referenced in the assignment. b. General observations/commentary: The field investigator reported that the monitoring reports appeared to be thorough and completed in accordance with the monitoring plan. Throughout the trial, weekly, biweekly, bimonthly, and quarterly meetings were held among team members to discuss the progress and issues regarding the study. The Core team consisted of PPD, BMS, IVRS, Lab and other individuals. The field investigator reported that in China, BMS contracted with PPD to provide temporary site monitor staff to supplement BMS-China site monitor employees. PPD provided BMS with monitors at Site 1178 and Site 1200 in China, and these monitors reported directly to BMS management. According to BMS procedures, PPD was contracted to perform clinical monitoring, project management, data management, pharmacovigilance, and medical monitoring for BMS during the ARITOTLE trial. An electronic filing system called ESF was used to upload all trial related documents directly to BMS, and all official records were stored at the BMS offices. The inspection collected copies of documents that contained operational and site information used by PPD during the conduct of the ARISTOTLE study. In addition, they collected the following documents relevant to the site inspections: Reference ID: 3132882 Page 27 Clinical Inspection Summary NDA 202155 [apixiban] • List of significant deviations for the eight inspected sites, including issues with informed consent and reporting SAEs on time. • Summary of issues at Site 1606 (Gupta, India), such as delays in data entry, delays in responding to queries, and not reconsenting subjects in a timely manner; and a timeline for how PPD handled the listed issues. • List of all Investigational Product issues noted during the monitoring visits. A description of these issues included things such as a wrong warfarin container given to the subject, or incorrect container number dispensed. At some sites, appropriate training was performed to stie staff, and the investigators were instructed to contact the patient and ask the subject to stop taking IP from the wrongly assigned bottle and to return to the site to be given a new correct IP bottle. AEs were appropriately assessed. OSI Reviewer Comment: For ARISTOTLE, study monitoring was provided by PPD in all countries except Russia, Ukraine and Israel. In China, BMS contracted with PPD to provide temporary site monitor staff to supplement BMS China Site Monitor employees. The monitors performed their study-related job responsibilities under the direction of BMSChina Site Managers. In China, the Senior Clinical Site Manager also provided support to the Clinical Site Managers to oversee the activities of the site monitors (PPD employees). In (b) (6) China, Ms. (as Site Manager) was the primary BMS contact with the site. She (b) (6) communicated with team members regarding study-related issues. Ms. reviewed site visit reports across all 36 sites in China between November 2008 and July 2011. Because she was a BMS employee, there was potential conflict of interest in her role as Site Manager in China. c. Assessment of Data Integrity: No Form FDA-483 was issued during the inspection. In general, it appears that clinical monitoring was adequate for the ARISTOTLE trial overall. Records appeared complete and organized; the SOPs were adequate to ensure quality assurance throughout the trial. The data for this study may be considered acceptable in support of the respective indication. III. OVERALL ASSESSMENT OF FINDINGS AND RECOMMENDATIONS Eight clinical investigator sites were inspected in support of this application (seven foreign and one domestic). In addition, a sponsor (BMS) inspection, a sponsor (BMS) re-inspection and a CRO (PPD) inspection were conducted. Site1301 (Vogel) and at Site 1742 (Zaidman) in South America, the inspections found that in general, they adhered to the applicable regulations and good clinical practices governing the conduct of clinical investigations. A few instances were identified of failing to report non-serious AEs at the both sites, but the events were not significant, and not likely to impact overall safety efficacy. Reference ID: 3132882 Page 28 Clinical Inspection Summary NDA 202155 [apixiban] Site 701 (Takacs) in Hungary had minor recordkeeping inaccuracies, and was classified as VAI, but in general the data was acceptable in support of the NDA. The inspection at Site #796 (Kufs) in New York identified minor regulatory violations including enrollment of two subjects who did not meet inclusion criteria; and minor inconsistencies in IP accountability records. The inspection also cited Dr. Kufs for failing to report SAEs in a timely manner. However, in most cases, it appears that the subjects were hospitalized for their events, and Dr. Kufs was not directly notified. Once informed of the event, Dr. Kufs reported the event on that date. These findings are not likely to importantly impact data integrity. OSI was notified by BMS on January 27, 2012, concerning allegations of suspected misconduct at Dr. Shiyao Wu’s Site 1200 in Shanghai, China. Specifically, a PPD monitor (b) (6) ( alleged that documents may have been altered (under the direction of a BMS site (b) (6) manager in preparation for the upcoming FDA inspection at that site. It was also (b) (6) alleged that this BMS site manager ( intentionally withheld information of findings in order to influence the outcome of the inspection. In their Allegations of Suspected Misconduct Investigations Report (January 26, 2012), BMS states that the misconduct at Site 1200 was an isolated site-specific event which resulted in compromise of the data from Site 1200. BMS stated that appropriate disciplinary action was taken including termination of employment of three employees. Lastly, BMS would provide revised statistical analysis of the ARISTOTLE study to exclude data from Site 1200. The FDA inspection at Site 1200 (Shiyao Wu, China) found that the PI did not exercise adequate control over the conduct of the study. This finding was evidenced by shared user accounts between subinvestigators, which might potentially compromise the integrity of the data. The FDA inspection also found that study coordinators frequently took home unsecured USB drives that contained patient data to enter into eCRFs because of the slow internet speed at the site. OSI considers that these two activities could have potentially compromised the integrity of the data at the site, and recommends the data not be used in the final analysis. (b) (6) In addition, OSI received information from BMS that Ms. worked as Site Manager for the BMS ARISTOTLE study and she visited 18 of the 36 Chinese sites while the study was underway; that her duties included conduct of co-monitoring visits if needed, site recruitment/enrollment; resource needs, and following up on quality issues noted by the Site Monitor; and she reviewed site visit reports written by the site monitor. During preinspection visits, she reviewed study files and subject medical records and discussed issues (b) (6) with study personnel. The sites in China where Ms. worked were: 1180, 1182, 1199, 1206, 1207, 1221, 1223, 1246, 1247, 1266, 1547, 1548, 1549, 1550, 1552, 1555, 1556, and 1558. She worked on no sites outside China. As stated above, to prepare for the (b) (6) upcoming FDA inspections, Ms. worked at Sites 1200 and 1178. In addition, (b) (6) PPD employee (the PPD employee who admitted that he had altered data (b) (6) containing subject records on a USB drive at the direction of Ms. BMS site Reference ID: 3132882 Page 29 Clinical Inspection Summary NDA 202155 [apixiban] manager) performed site monitoring responsibilities for the following Sites: 1200, 1198, 1168, 1244, and 1287. The inspection of Site 0019 (Dr. Yao, Canada) was classified as VAI. This site had very high enrollment. The field investigators observed that Dr. Yao did not adhere to good GCP practices because he co-mingled his study records with his private practice records, and he post-dated entries in progress notes for some visits, often by several months. This means that Dr. Yao did not always review entries written in progress notes until several months later, and would then sign and date the progress notes on the date he read them. Though this is considered as poor GCP practice, it is not likely to have impacted data integrity. In general, the regulatory violations noted were isolated and not likely to impact data integrity. The inspection at Site 1606 (Dr. Gupta, India) was classified as VAI. This site had very high enrollment. The main findings at this site were that the site knowingly maintained subjects at a low TTR. Once the sponsor became aware of the issues at this site, they provided additional training to Dr. Gupta and his staff on ways to improve INR management. OSI considers the data acceptable at this site. The CRO PPD was inspected and no regulatory violations were found. The sponsor (Bristol-Myers Squibb) was inspected twice. The first inspection was a routine sponsor inspection, and minor regulatory violations were found. The second sponsor inspection was conducted to review issues of oversight and monitoring, medication dispensation errors, investigational product issues and potential for unblinding. Minor regulatory violations were found, but these violations did not appear to compromise the overall integrity of the data. The inspectional findings found during inspections at one U.S. site (Site #796), and five non-U.S. sites (Sites 701, 1301, 1742, 1606, 0019) were minor, and OSI does not believe them likely to influence data integrity, study outcome or subject safety. However, OSI recommends that the data from Site 1200 (Shiyao Wu, China) and Site 1178 (Shulin Wu, China) not be used in the final analysis. Although there were no significant inspectional findings at Site 1178, the possibility of fraud cannot be excluded. In addition, (b) (6) (b) (6) OSI recommends that data from the Chinese sites where either or Mr. worked be excluded from the study analysis. These are Sites 1168, 1178, 1180, 1182, 1198, 1199, 1200, 1206, 1207, 1221, 1223, 1244, 1246, 1247, 1266, 1287, 1547, 1548, 1549, 1550, 1552, 1555, 1556, and 1558. This is because of the allegations of fraud documented to occur at Site 1200 and the potential for misconduct at other sites within (b) (6) (b) (6) China where Ms. and or Mr. had monitoring oversight or conducted monitoring activities on site. Overall, the study appears to have been conducted and monitored adequately, based on OSI inspectional findings. Although fraud at Site 1200 in China was well documented, there is no evidence that fraudulent activity occurred elsewhere. OSI has recommended that data from the above 24 sites in China be excluded because we cannot provide inspectional (b) (6) (b) (6) evidence to support data integrity and subject safety, given that the Ms. and Mr. worked at these sites. The remaining inspections of clinical investigators, CRO, and Reference ID: 3132882 Page 30 Clinical Inspection Summary NDA 202155 [apixiban] sponsor did not reveal systemic evidence of inadequate monitoring or data integrity issues. No regulatory violations were identified during the PPD inspection; and minor regulatory violations found during the Sponsor inspections. OSI recommends that the data submitted by Bristol Myers Squibb may be used in support of the respective indication. {See appended electronic signature page} Sharon Gershon, Pharm.D. Good Clinical Practice Assessment Branch Division of Good Clinical Practice Compliance Office of Scientific Investigations CONCURRENCE: {See appended electronic signature page} Susan Thompson, M.D. Team Leader, Good Clinical Practice Assessment Branch Division of Good Clinical Practice Compliance Office of Scientific Investigations CONCURRENCE: {See appended electronic signature page} Lauren Iacono-Connors, Ph.D. Acting Branch Chief Good Clinical Practice Assessment Branch Division of Good Clinical Practice Compliance Office of Scientific Investigations Reference ID: 3132882 --------------------------------------------------------------------------------------------------------This is a representation of an electronic record that was signed electronically and this page is the manifestation of the electronic signature. --------------------------------------------------------------------------------------------------------/s/ ---------------------------------------------------SHARON K GERSHON 05/18/2012 SUSAN D THOMPSON 05/18/2012 LAUREN C IACONO-CONNORS 05/18/2012 Reference ID: 3132882