MEMORANDUM DEPARTMENT OF HEALTH AND HUMAN SERVICES PUBLIC HEALTH SERVICE FOOD AND DRUG ADMINISTRATION CENTER FOR DRUG EVALUATION AND RESEARCH ___________________________________________________________________________________________ COMPLIANCE REVIEW DATE: May 24, 2011 TO: Tyree Newman, Regulatory Project Manager Min Lu, M.D., Medical Officer Division of Hematology Products FROM: Susan D. Thompson, M.D., Medical Officer Good Clinical Practice Branch II Division of Scientific Investigations THROUGH: Tejashri Purohit-Sheth, M.D. Branch Chief Good Clinical Practice Branch II Division of Scientific Investigations SUBJECT: Evaluation of Clinical Inspections NDA: 22-406 APPLICANT: Johnson & Johnson DRUG: Xarelto (rivaroxaban) NME: Yes THERAPEUTIC CLASSIFICATION: INDICATIONS: Standard Review 1. Prophylaxis of deep vein thrombosis and pulmonary embolism in patients undergoing hip or knee replacement surgery CR LETTER DATE: May 27, 2009 (b) (4) AUDIT SUBMISSION DATE: (b) (4) INFORMATION REQUEST RESPONSE DATE: September 26, 2010 April 19, 2010 CR SUBMISSION DATE: December 23, 2010 Reference ID: 2951276 Page 2 Compliance Review of (b) (4) NDA 22-406, Xarelto I. BACKGROUND: Rivaroxaban is a highly selective direct factor Xa (FXa) inhibitor for oral administration. Inhibition of FXa produces antithrombotic effects by decreasing the amplified generation of thrombin, thus diminishing thrombin-mediated activation of both coagulation and platelets, without affecting existing thrombin levels. The sponsor states that the remaining thrombin should be sufficient to ensure primary hemostasis, resulting in a favorable efficacy to safety (bleeding) margin for rivaroxaban. The sponsor submits this NDA to support the use of rivaroxaban for the indication of prophylaxis of deep vein thrombosis (DVT) and pulmonary embolism (PE) in patients undergoing hip or knee replacement surgery. Patients undergoing major orthopedic surgery, including total hip replacement (THR) and total knee replacement (TKR) surgeries, are a group that is at a particularly high risk for venous thromboembolism (VTE), which includes DVT and PE. Without prophylaxis, the incidence of objectively confirmed total DVT based on older studies is approximately 40 to 60% following THR or TKR, with a 10-30% incidence of proximal DVT. The most appropriate strategy to reduce the incidence of VTE is prophylaxis for all patients undergoing THR or TKR. Current therapeutic agents available for anticoagulant prophylaxis include low molecular weight heparins (LMWHs), fondaparinux, and adjusted-dose vitamin K antagonists such as warfarin. The duration of therapy is at least 10 days for both THR and TKR; for patients undergoing THR, extended prophylaxis to up to 35 days after surgery is recommended. LMWHs and fondaparinux are administered subcutaneously, which may be associated with pain and bruising as well as poor compliance. Warfarin is the only available oral anticoagulant for VTE prophylaxis after major orthopedic surgery in the U.S. However, warfarin has a narrow therapeutic window, exhibits variable dose response, has many dietary and medicinal interactions, requires dose adjustment, and has a slow onset of action. Rivaroxaban offers an alternative oral prophylactic therapy for VTE. IND 64,892 for rivaroxaban was submitted on May 29, 2002 for the treatment and secondary prophylaxis of VTE by Bayer. All of the clinical trials submitted with the current NDA were conducted by Bayer. Approximately one month prior to the submission of this NDA, Bayer sold the rights of reference for use of the investigations to Johnson and Johnson. Johnson and Johnson submitted NDA 22-406 as the applicant on July 28, 2008. Of note, both Bayer and Johnson and Johnson submitted letters to the review division that the IND is now transferred to Johnson and Johnson. The pivotal protocols in support of NDA 22-406 were: RECORD 1 Study: Regulation of Coagulation in Orthopedic Surgery to prevent DVT and PE; controlled, double-blind, randomized study of BAY 59-7939 in the extended prevention of VTE in patients undergoing elective total hip replacement (Protocol 11354) RECORD 2 Study: Regulation of Coagulation in Orthopedic Surgery to prevent DVT and PE; controlled, double-blind, randomized study of BAY- 59-7939 in the extended prevention of VTE in patients undergoing elective total hip replacement (Protocol 11357) Reference ID: 2951276 Page 3 Compliance Review NDA 22-406, Xarelto (b) (4) RECORD 3 Study: Regulation of Coagulation in Orthopedic Surgery to prevent DVT and PE; a controlled, double-blind, randomized study of BAY 59-7939 in the prevention of VTE in patients undergoing elective total knee replacement (Protocol 11356) RECORD 4 Study: Regulation of Coagulation in Orthopedic Surgery to prevent DVT and PE; a controlled, double-blind, randomized study of BAY 59-7939 (rivaroxaban) in the prevention of VTE in subjects undergoing elective total knee replacement (Protocol 11355) FDA Inspections During the conduct of the clinical studies for this NDA, complaints were received regarding two investigators enrolling subjects, Dr. Arturo Corces in RECORD 2 and Dr. David Loucks in RECORD 4. A Warning Letter was issued to Dr. Corces on May 22, 2008 who enrolled subjects in RECORD 2 for failure to personally conduct or supervise the clinical investigations, failure to meet informed consent requirements, failure to ensure that studies were conducted according to the relevant current protocol, failure to maintain adequate and accurate case histories, and failure to maintain adequate drug disposition records. The Warning Letter to Dr. Corces from DSI recommended the data contributed to RECORD 2 by Dr. Corces be considered unreliable. A NIDPOE was issued on August 18, 2009 for Dr. Loucks, who enrolled subjects in RECORD 4. The OAI-NIDPOE letter describes failure to adhere to the protocol, inadequate/inaccurate records, failure to report to the IRB risk to human subjects or others, submission of false information, and failure to supervise/personally conduct a study. On June 3, 2008, after discussion with the review division regarding the inspectional findings, Bayer notified the review division that due to falsification and systematic failures of the outpatient source data, that data from Dr. Loucks’ site should be excluded from the per protocol analysis. On the same date, Bayer notified the review division of a second RECORD 4 clinical investigator, Dr. Ricardo Esquivel in Naulcapan, Mexico, with issues impacting data integrity. These issues included inability to confirm from the source record that study medication was administered per protocol during the hospitalization periods, due to systematic discarding of medical records documenting study drug administration. On July 28, 2008, Johnson & Johnson submitted the data from the RECORD 1, 2, 3, and 4 studies to the FDA to support the approval of rivaroxaban for the indication of prophylaxis of deep vein thrombosis and pulmonary embolism in patients undergoing hip or knee replacement surgery (NDA 22-406). After receipt of the NDA, eight FDA data validation inspections of investigators who enrolled subjects in the four RECORD studies were conducted. The results of these clinical investigator inspections resulted in the identification of multiple regulatory violations from many of these sites, raising concerns with the overall integrity of the data submitted for approval of the NDA. Details of the first cycle of clinical investigator, sponsor, and applicant inspections, as well as RECORD 1-4 investigators identified as problematic prior to submission of the NDA, are summarized in the following table: Reference ID: 2951276 Page 4 Compliance Review NDA 22-406, Xarelto (b) (4) Table 1: NDA 22-406 Pre-NDA and First Cycle Clinical Investigator Data Validation Audits Name of CI or Sponsor Location Andrzej Gorecki Szpital Kliniczny Dzieciatka Jezus – Centrum Leczenia Obrazen Klinika Ortopedii I Traumatologii Narzadu Ruchu Ul. Lindleya 4 02-005 Warszawa , Poland Tadeusz Gazkzik Slaska Adademia Medyczna Katedra I Oddzial Kliniczny Ortopedii’Wojewodzki Szpital Specjalistyczny Nr 5 Im. Sw. Barbaby Pl. Medykow 1 41-200 Sosnowiec, Poland Arturo Corces 7340SW 79th Street Miami Institute for Medical Research Miami, FL 33186 Protocol # and # of Subjects Major Findings Protocol #11354, RECORD 1 Site # Poland 18006 # of subjects (Total #: 71): Xarelto: 36 Enoxaparin: 35 Qingming Yang Rui Jin Hospital, Shanghai Second Medical University Orthorpaedic Department Shanghai Ryuijin Hospital No. 197 Ruijin Second Road Shanghai, China 200025 Protocol # 11357, RECORD 2 Site # China 54005 Cesar Diaz Valverde Hospital Edgardo Rebagliati Martins Av. Edgardo Rebagliati Martins S/N Jesus Maria Lima Lima, 11 Peru Protocol # 11357, RECORD 2 Site # Peru 64005 Reference ID: 2951276 None Inspection Date 1/9–1/23/09 (complaint related) Interim Classification NAI DSI Classification NAI Protocol #11354, RECORD 1 Site # Poland 18012 # of subjects (Total #: 76): Xarelto: 38 Enoxaparin: 38 None 2/2-2/6/09 NAI NAI Protocol #11357, RECORD 2 Site # 14012 # of subjects (Total #: 19): Xarelto: 9 Enoxaparin: 10 Failure to supervise, informed consent, protocol, recordkeeping and drug disposition deficiencies AEs not reported, including abnormal liver function tests and bleeding; protocol violations, recordkeeping deficiencies 3/20 – 4/26/07 (complaint related) OAI OAI (WL) 2/9-2/13/09 OAI OAI (untitled) 1/26-1/30/09 VAI VAI # of subjects (Total# 34): Xarelto: 17 Enoxaparin: 17 # of subjects (Total#: 41): Xarelto: 20 Enoxaparin: 21 AEs (relatively minor) not reported; protocol and recordkeeping deficiencies Page 5 Compliance Review NDA 22-406, Xarelto (b) (4) Table 1: NDA 22-406 Pre-NDA and First Cycle Clinical Investigator Data Validation Audits Name of CI or Sponsor Location Protocol # and # of Subjects Major Findings Inspection Date Interim Classification DSI Classification Binfang Zeng Affiliated Sixth People’s Hospital Orthorpaedic Department No. 600 Yishan Road, Xuhui District Shanghai, China 200233 Protocol # 11356, RECORD 3 Site # China 54014 AEs not reported, including 2 SAEs; protocol violations 2/15-2/19/09 OAI VAI Jacek Kruczynski Szpital Uniwersytecki im. Antoniego Jurasze Klinika Ortopedii i Traumatogii Narzadu Ruchu Ul. M. SklodowskiejCurie 9 85-094, Bydgoszcz Poland Protocol # 11356, RECORD 3 Site # Poland 18003 Protocol violations 1/26-1/30/09 VAI VAI David Loucks 14100 E. Arapahoe Rd. Suite B370 Centennial, CO 80112 Protocol # 11355, RECORD 4 Site # 14012 Recordkeeping deficiencies and falsification, IRB reporting, protocol and informed consent deficiencies 4/15-7/08 OAI OAI (NIDPOE) Drug disposition, record deficiencies, missing records Identified by Bayer (not inspected by FDA) NA Data not usable Post-operative randomization in violation of protocol, possible unblinding 2/17-2/26/09 OAI OAI-WL Informed consent deficiencies and protocol violations 1/261/28/09, 2/22/6/09, 2/122/13/09 VAI VAI # of subjects (Total# 26): Xarelto: 13 Enoxaparin: 13 # of subjects (Total# 36): Xarelto: 18 Enoxaparin: 18 # of subjects (Total# 94): Xarelto: 46 Enoxaparin: 46 Ricardo Esquivel Naulcapan, Mexico Protocol # 11355, RECORD 4 Site # 32006 # of subjects (Total# 42): Xarelto: 22 Enoxaparin: 20 R. Michael Murray Capstone Clinical Research 2018 Brookwood Medical Center Suite 314 Birmingham, AL 35209 Protocol # 11355, RECORD 4 Site # 14005 # of subjects (Total # 152) Xarelto: 76 Enoxaprin: 76 David Fox Unlimited Research, LP 12709 Toepperwein Road Protocol #11355, Record 4 Site #14022 # of subjects Reference ID: 2951276 Page 6 Compliance Review NDA 22-406, Xarelto (b) (4) Table 1: NDA 22-406 Pre-NDA and First Cycle Clinical Investigator Data Validation Audits Name of CI or Sponsor Location Protocol # and # of Subjects Suite 101 San Antonio, TX 78233 (Total # 64) Xarelto: 32 Enoxaparin: 32 Protocol # 11354, RECORD 1 Protocol # 11357, RECORD 2 Protocol # 11356, RECORD 3 Protocol #11355, Record 4 Bayer Pharmaceutical 340 Change Bridge Rd. Pine Brook, NJ 07058 Johnson & Johnson 920 U.S. Highway 202 Raritan, NJ 088690602 Protocol # 11354, RECORD 1 Protocol # 11357, RECORD 2 Protocol # 11356, RECORD 3 Major Findings Inspection Date Interim Classification DSI Classification Monitoring deficiencies, protocol violations, failure to ensure that FDA was informed of all AEs No significant issues noted; however, inspection limited in scope 2/24-2/27, 3/3-3/6, 3/9, 3/11-3/13, 3/16-3/19, 3/26, 3/303/31/09 VAI VAI 3/24/09 NAI VAI Protocol #11355, Record 4 Key to Classifications NAI = No deviation from regulations. VAI = Deviation(s) from regulations. OAI = Significant deviations from regulations. Data unreliable. As can be seen in Table 1, there were a variety of major findings, including protocol violations, deficiencies in drug dispensation records, AE reporting, and informed consent. A major issue identified during inspections of RECORD 4 study sites was post-operative randomization of subjects, instead of randomization of subjects prior to surgery as specified in the protocol. In order to characterize more fully how frequently post-operative randomization in violation of the protocol occurred, an assignment for inspection of three additional clinical investigators in RECORD 4 was issued. Details of the second cycle of clinical investigator inspections are summarized in the following table: Reference ID: 2951276 Page 7 Compliance Review NDA 22-406, Xarelto (b) (4) Table 2: NDA 22-406 Second Cycle Clinical Investigator Data Validation Audits Name of CI/Address/contact information Protocol # and # of Subjects Major Findings Inspection Date Dr. John Ward Capstone Clinical Research 2018 Brookwood Medical Center Suite 314 Birmingham, AL 35209 Phone: (205) 877-2766 Fax: (205) 877-2990 Email: capstoneclin@aol.com Protocol # 11355 RECORD 4 Post-operative randomization, IRB approval expired Dr. Craig Buettner West Alabama Research, Inc. Black Warrior Medical Building 100 Rice Mine Road Loop Suite 104 Tuscaloosa, AL 35406 Phone: (205) 248-6160 FAX: (205) 248-6467 Email: vredding @walresearch.com (coordinator) Dr. John Schwappach Colorado Orthopedic Consultants 401 W. Hampton Place Suite 220 Englewood, CO 80110 Phone: (303) 695-6060 (research dept. extension) FAX: (303) 399-9959 Email: schwappach@cocortho.com Interim Classification Final Classifiation 5/12-5/20/09 OAI OAI-WL Post-operative randomization 5/4-5/6/09 OAI OAI-WL Protocol violations 5/5-5/19/09 VAI VAI Site # 14010 # of subjects (Total # 203) Xarelto: 101 Enoxaprin: 102 Protocol #11355 RECORD 4 Site #14004 # of subjects (Total # 61) Xarelto: 31 Enoxaprin: 30 Protocol #11355 RECORD 4 Site #14045 # of subjects (Total # 106) Xarelto: 53 Enoxaprin: 53 Key to Classifications NAI = No deviation from regulations. VAI = Deviation(s) from regulations. OAI = Significant deviations from regulations. Data unreliable. Additionally, inspection of Bayer Pharmaceuticals as the sponsor of the four RECORD 4 studies revealed that the sponsor failed to 1) ensure proper monitoring of the study, 2) to ensure the study was conducted in accordance with the protocol and/or investigational plan, and 3) to ensure that FDA and all investigators were promptly informed of significant new adverse effects or risks. The sponsor inspection of Bayer revealed that some of the minor items cited in the OAI letters for Drs. Corces and Murray were not identified in site Monitoring Visit Reports although the CRAs were aware of them (either through the company’s internal audit program or FDA inspections). The major violations at these sites were not detected by sponsor monitoring. Bayer acknowledges the failure to include the cited deficiencies in Reference ID: 2951276 Page 8 Compliance Review NDA 22-406, Xarelto (b) (4) Monitoring Visit Reports in their response letter dated April 13, 2009. The sponsor inspection of Bayer does not provide information on whether or not monitoring and/or corrective actions were inadequate at other sites classified by FDA as OAI. A limited inspection of the applicant Johnson & Johnson revealed no identifiable deviations from applicant related regulations as per 21 CFR 314. Complete Response Letter to Applicant and Subsequent Activity On May 27, 2009 FDA issued an NDA Complete Response letter to Johnson & Johnson for Xarelto NDA 22-406 that listed several deficiencies, including Clinical Deficiency 1 which stated that the reasons that data from 7 Clinical Investigator sites are considered unreliable include: • Failure to conduct the study according to the signed investigator statement and the investigational plan [21 CFR 312.60] • Failure to report to the sponsor adverse events [21 CFR 312.64] • Failure to prepare or maintain adequate and accurate case histories with respect to observations and data pertinent to the inspection [21 CFR 312.62 (b)] • Failure to obtain adequate informed consent [21 CFR 50] • Failure to maintain drug accountability records [21 CFR 312.62 (a)] • Failure to report to the IRB all unanticipated problems involving risk to human subjects [21 CFR 312.66] On the basis of these findings, FDA requested in the CR letter that the applicant: a. Provide the following information regarding their QA audit program: i. A report of the QA audit plan, including the plan for securing compliance from non-compliant clinical investigators. Included should be copies of any Standard Operating Procedures that were in place during conduct of the study to address means by which corrective actions were to be taken if or when you or the CRO identified noncompliant clinical investigators. ii. A report of the sponsor’s audit findings, including any corrective actions taken and final outcome for the Yang, Murray, Corces, Loucks, and Esquivel sites and for all other sites audited under the sponsor’s QA program. iii. A description of any clinical investigators terminated for noncompliance. The list should include sites, specific violations, and whether the data were included in the NDA submission. b. Describe Bayer’s QA program with respect to the oversight of CROs that were hired to (b) (4) monitor the clinical sites, including for RECORD 4. Describe the procedures implemented to make sure that the CROs adequately monitored the clinical sites. The response should include the following information: i. Provide the procedures by which Bayer was kept apprised by the CROs concerning monitoring of the clinical site during the course of the study. Specifically, describe what information the CROs provided to the sponsor and provide a list of noncompliant clinical study sites reported by the CROs. ii. Describe how the sponsor reviewed the information provided by the CROs during the course of the study and at the end of the study. Describe what monitoring information was kept at the end of the study. Reference ID: 2951276 Page 9 Compliance Review of NDA 22-406, Xarelto c. Independent Thirty Party Audits (b) (4) (b) (4) Provide assurance that the clinical data obtained from the RECORD 1, 2, 3, and 4 studies are reliable. Specifically, perform an additional audit and supply the results of this audit within your response to this letter. Within your response, include: i. A copy o f your audit plan, including the following information: • How many clinical sites were to be audited, how many subject records were examined, and a description of the process for selection of the audited sites. • If not all subject records at a given clinical site were to be audited, describe how subject records were sampled and how the specific data from each subject were audited. ii. The timeline for completion of your audit (plan finalization, start date, completion date, report finalization date). As per above, the CR letter stated that additional third party audits should be conducted to provide assurance that the RECORD 1-4 studies are reliable and requested that Johnson & Johnson submit a proposal for these audits. On June 8, 2009, Johnson & Johnson submitted “Clarification Questions” for the Complete Response Letter, which included a proposal that 24 new audits be conducted, together with submission of the reports of the 69 routine and 5 directed/for-cause audits. Johnson & Johnson proposed that the results of the new audits be submitted as an addendum to the Complete Response. In a written response preparatory to a face-to-face meeting between FDA and Johnson & Johnson on June 19, 2009, FDA proposed that 25% of the clinical investigator RECORD 1-4 sites be audited by an independent, third party. At the June 19, 2009 meeting, FDA proposed the following: “Selection of sites from all four RECORD studies with a total enrollment of 60 or higher results in identification of 26 sites: 9 in RECORD 1 (2 already inspected by the Agency), 4 in RECORD 2, and 13 in RECORD 4 (6 already inspected by the Agency). If 5% is the margin of error for tolerance of unreliable sites detected by the audit, then the audit of 30 sites is necessary to show with 95% confidence that the percentage of unreliable sites exceeds 5%, assuming that 25% of sites are actually unreliable. Therefore, 18 high enrolling sites not previously inspected could be included in the audit plan, which represents 11% of enrolled subjects. If 30 total sites are to be audited, an additional 12 sites could be included in the audit, which represent a random sample of sites which enrolled 40-60 subjects and sites which enrolled 10-30 subjects.” Johnson & Johnson submitted the proposed audit plan on July 8, 2009. The audit plan included audits of an additional 30 clinical sites across the entire RECORD program including all 18 high enrolling sites with > 60 randomized subjects, and 12 moderate enrolling sites with 15-59 randomized subjects. The 12 moderate enrolling sites (3 per study) were randomly selected by the Johnson & Johnson statistics group from a pool of sites which met the stated enrollment criteria. None of the sites selected for audit had previously been inspected for this NDA by the FDA. Johnson & Johnson intended to audit all subjects if there were 35 subjects or less enrolled at the site. If there were more than 35 subjects, a random selection of subjects Reference ID: 2951276 Page 10 Compliance Review NDA 22-406, Xarelto (b) (4) was to be chosen such that if no data integrity issue was found in sample subjects, there would be 95% confidence to rule out more than a 5% error rate. The resulting sample size represented a 31% to 58% sampling (35 to 43 subjects) of sites which enrolled more than 35 subjects. Audit of these 30 sites resulted in a total of 950 subjects with data audited out of 12,729 subjects, which constituted 7.5% of all subjects in the RECORD program data base. (b) (4) The results of the audits were to be submitted with the CR. The parameters to be verified during each audit were those contained in the Complete Response Letter (listed in Part II.1.c. below). The audit findings at each site were to be documented in an individual site audit report and provided to the FDA. In addition, a separate (b) (4) summary report was to be provided. Johnson & Johnson proposed that conduct the targeted audits. The criteria used by Johnson & Johnson to identify the independent third party auditor required that there were 1) no previous associations with the rivaroxaban development program, and 2) no current contracts with Johnson & Johnson or Bayer. Auditors utilized were full-time employees of the independent third party or regionally based contractors who were trained on the company’s SOP and were overseen by a full-time (b) (4) employee of the independent third party. Johnson & Johnson has previously employed as an independent third party audit team. Johnson & Johnson proposed to provide a member of the Bayer Global Clinical Operations or Quality Assurance team to escort the third-party auditor for logistical support and translation, if needed. On August 5, 2009, DSI communicated in writing that DSI was in agreement with the number of sites selected and the number of subjects to be audited at each site, submission of individual site reports as well as a separate summary report, and agreement with the proposed Data (b) (4) Verification Tool for the audits. On March 5, 2010, Johnson & Johnson submitted Meeting Background Information in preparation for a face-to-face meeting on April 7, 2010. The Background Information (b) (4) contained a summary of the results of the audits. Johnson & Johnson also submitted a proposal for data verification and sensitivity analyses for RECORD 4 in order to allay concerns regarding the FDA inspectional and third party audit findings pertaining to that study. (b) (4) According to the proposal, Johnson & Johnson would employ to revisit all RECORD 4 sites to obtain unreported adverse events as well as relevant data for the sensitivity analysis. DSI responded that a review of the complete audit reports conducted for all four RECORD studies, rather than a summary, was necessary before agreement could be reached on a path (b) (4) forward. In addition, no recommendation could be given regarding the data (b) (4) verification proposal prior to the review of the RECORD 4 audits. Subsequent to this (b) (4) meeting, Johnson & Johnson submitted on April 19, 2010 the audit reports for the audits conducted between July 27, 2009 and October 16, 2009, as well as copies of the Bayer internal company audits conducted concurrently with the clinical trials. The CR was submitted on December 23, 2010. The following sections of this review will first evaluate the Applicant’s Complete Response focusing on the adequacy of responsiveness to the items requested in the Agency’s Complete (b) (4) Response Letter. This will be followed by a description of Audit findings focusing on items considered key to evaluation of data reliability. The review will then provide DSI’s analysis of the specific audit findings and their impact on data reliability, followed by an Reference ID: 2951276 Page 11 Compliance Review of NDA 22-406, Xarelto (b) (4) assessment of data reliability for each RECORD study. The review will then conclude with DSI’s conclusions and recommendations on reliability of data for the application as a whole. II. EVALUATION OF APPLICANT’S DECEMBER 23, 2010 SUBMISSION In the FDA’s April 29, 2009 Complete Response (CR) letter, a number of requests were outlined that the applicant needed to address to resolve the Agency’s concerns with respect to data integrity issues. In the sections below, each of the items in the letter will be restated in bold font, followed by a summary of Johnson & Johnson’s response, and DSI’s assessment of the adequacy of the response. 1. a. Provide the following information regarding your clinical data quality assurance (QA) audit program that was in place for the four RECORD studies: i. A report of your QA audit plan, including your plan for securing compliance from non-compliant clinical investigators. Include copies of any standard Operating Procedures (SOPs) that were in place during conduct of the study to address the means by which corrective actions were to be taken if or when you or the applicable means by which corrective actions were to be taken if or when you or the applicable contract research organization (CRO) identified noncompliant investigators. Johnson & Johnson provided a summary of their audit plans for the RECORD 1, 2, 3, and 4 studies. They also provided a summary of their audit procedures and copies of SOPs for audit procedures. Included were SOPs which address procedures for site initiation and monitoring, study management, investigator site audits, and misconduct. DSI Assessment of Response: Johnson & Johnson has adequately responded to this request. ii. A report of your audit findings, including any corrective actions taken and final outcomes for the Yang, Murray, Corces, Loucks, and Esquivel sites and for all other sites you audited under your QA program. iii. A description of any clinical investigators terminated for non-compliance. Provide a list of these clinical investigators, their sites, the specific violations, and whether the data were included in the NDA submission. The description of the findings requested in Parts 1.a.ii. and 1.a.iii. and the DSI assessment of these findings are combined below. Response to 1a.ii. Johnson & Johnson provided a summary of audit findings, corrective action plans, and outcomes for each site for clinical investigator sites that participated in the RECORD studies. There were 74 clinical investigator site audits conducted by Bayer; 69 were routine and 5 were for cause. There were 25 audits conducted at RECORD 1 sites, 15 audits conducted at RECORD 2 sites, 15 audits conducted at RECORD 3 sites, and 19 audits conducted at RECORD 4 sites. Findings during the audit were classified into Class 1, Class 2, and Class 3. Class 1 findings are findings of confirmed misconduct which endanger subject safety and/or Reference ID: 2951276 Page 12 Compliance Review of NDA 22-406, Xarelto (b) (4) would lead to rejection of data by Regulatory Authorities, whereas Class 2 and Class 3 are less serious findings. There were 2 clinical investigator sites with Class 1 findings in RECORD 1, 1 clinical investigator site with Class 1 findings in RECORD 2, 3 clinical investigator sites with Class 1 findings in RECORD 3, and 2 clinical investigators with Class 1 findings in RECORD 4. Of these clinical investigator sites with Class 1 findings, 4 were the sites which were for cause inspections: Dr. Macaire Site 11354 in RECORD 1 and Site 16009 in RECORD 3, Dr. Mortele Site 28020 in RECORD 3, Dr. Dadi Site 60017 in RECORD 4, and Dr. Loucks Site 14029 (the third inspection); these inspections will be discussed below. The remaining sites with Class 1 findings included Dr. Jasey Site 26007 in RECORD 2 which had enrollment temporarily suspended due to limited access of the site by the auditors to source documents, poorly documented changes to source documents, suboptimal level of principal investigator involvement, and enrollment of a clearly ineligible subject. Enrollment was restarted 11 days later after these issues were addressed to the satisfaction of the sponsor. The 2 remaining sites with a Class 1 finding both routinely obtained coagulation studies at the site, which could potentially result in unblinding and is a protocol violation. Both sites (Dr. Schmelz Site 10010 in RECORD 1 and Dr. Debue Site 16001 in RECORD 3) corrected the problem immediately. The last site with a Class 1 finding is Dr. Esquivel Site 32006, also discussed below. FDA requested a report of the applicant’s audit findings, including any corrective actions taken and final outcomes for the Yang, Murray Corces, Loucks, and Esquivel sites and for all other sites that were audited under their QA program. The following provides a summary of this information. Dr. Q. Yang Site 54005 RECORD 4: This site was not included in Bayer’s audit program. The regulatory violations cited by the FDA inspector are acknowledged by the applicant in the CR and reasons given for the violations. However, no evidence is presented to refute the violations observed during FDA inspections. Dr. Michael Murray Site 14005 RECORD 4: There were no Class 1 findings at the inspections of Dr. Murray, Site 14005, and Dr. John Ward, Site 14010, who enrolled as separate sites in Birmingham, Alabama under the umbrella of an SMO. Class 2 findings included source data inadequacies, systematic data inaccuracies involving adverse event and concomitant medications, and failure to obtain protocol required venograms. The applicant presents information from Dr. Murray’s letter of response to the inspectional findings. The source document issues were addressed by source verification by the site CRA, with correction as needed. These issues differ from those identified during the FDA clinical site inspection (postoperative randomization, possible unblinding) which resulted in an OAI Warning Letter. Dr. Craig Loucks Site 14029 RECORD 4: A routine audit of this site was conducted starting on December 12, 2006. A number of Class 2 findings were identified involving problems with data quality and general GCP compliance, including lack of source documentation and lack of documentation of Principal Investigator (PI) involvement. Study activities were inappropriately delegated to unqualified study personnel, and there were extensive delays in CRF completion. Enrollment was placed on hold at the conclusion of the inspection; enrollment resumed on January 16, 2007 based on feedback from the CRA monitoring the Reference ID: 2951276 Page 13 Compliance Review of NDA 22-406, Xarelto (b) (4) study. A follow-up audit was conducted starting on May 14, 2007 to determine the effectiveness of the corrective actions taken. Persistent GCP noncompliance was noted, including evidence that the original source data worksheets completed during the outpatient phase of the study had been rewritten and the original documents not maintained. Enrollment was placed on hold, and the frequency of monitoring was increased. A third audit was conducted starting January 16, 2008. This audit was precipitated by site notification to the IRB of data falsification; the IRB communicated this information to the FDA. The January, 2008 Bayer audit confirmed falsification of the signatures of the PI (and in some cases the subinvestigator) on lab reports, ECGs, hospital orders, FDA 1572s, SAE documentation, IRB submissions, and ICF documents. At least 19 patients’ source data and nine submissions to the IRB were falsified. For the NDA submission, subjects from Dr. Loucks’ site were excluded from the per protocol analyses. Patients were included in the safety and mITT analyses when validity criteria were met, and sensitivity analyses were conducted after including subjects in the per protocol analysis and excluding subjects in the mITT which revealed that the overall results were not changed. Response to 1.a.iii. Johnson & Johnson listed the following clinical investigators who were terminated for noncompliance: Dr. Esquivel Gomez Site 32006 RECORD 4: A routine audit of the site starting on October 17, 2007 revealed the Class 1 finding that the site had failed to retain all available source records due to a hospital policy of periodically purging hospital and in-patient nursing notes. The nursing notes were considered to be source documents which verified the administration of the investigational product. The site had been placed on enrollment hold by the study team on August 9, 2007 due to delays in CRF completion and the hold remained in effect for the remainder of the study. Subjects were included in the Per Protocol analyses only when it could be confirmed that eCRF data had been verified as correct by the CRA. All subjects were included in the safety and mITT populations unless the subject did not meet validity criteria. Dr. Arturo Corces Site 14010 RECORD 2: This site was not audited by Bayer. The inspection conducted by FDA found inadequate Investigator oversight and systematic use by the site of PlexiPulse pneumatic compression, which was not allowed by the protocol. An Investigative Committee was established and follow-up activities were conducted by Bayer, including assignment of an additional CRA to the site. Retraining was conducted. No subjects randomized were valid for the per protocol analysis due to the use of pneumatic compression or inadequate assessment of thromboembolism. All subjects were included in the safety and mITT analyses, unless they did not meet validity criteria. Dr. Richard Rouhe Site 14062 RECORD 4: On August 16, 2007 Dr. Rouhe was notified by his IRB of his failure to report that his medical license was on probation for 5 years by the California Medical Board. Upon transmission to Bayer of this information, enrollment at this Reference ID: 2951276 Page 14 Compliance Review of NDA 22-406, Xarelto (b) (4) site was terminated; six subjects had been randomized and two subjects were treated. The CRA noted that Dr. Rouhe’s CV and medical license were missing at the first periodic monitoring visit on May 30, 2007; however, copies of the license was available at the August 10, 2007 CRA visit, and his CV was available at the October 15, 2007 monitoring visit. Data from this site were only included in the Safety Analysis, as the two subjects did not have an adequate assessment of venous thromboembolism. Additionally, the efficacy data from Dr. P. Macaire’s site 16009 in RECORD 1 was invalidated after a for cause inspection revealed that the CRA entered data in the eCRF and made changes outside of agreed permissible clarifications. The PI refused to confirm data entered into the eCRF. The data was considered valid for safety. DSI Assessment of Response: Johnson & Johnson has adequately responded to this request. In general, review of the audits revealed that appropriate corrective action plans were generated and implemented for those clinical investigator sites with Class 1 findings. However, there were several areas of concern identified. Although significant findings were identified at Dr. Michael Murray’s site during the Bayer audit, the issues identified by FDA inspectors which resulted in an OAI classification were not identified. Of greater significance, the initial two Bayer audits at Dr. Craig Loucks site identified significant problems at this site, resulting in a temporary hold on enrollment and increased frequency of monitoring. However, the most serious issue of forging the principal investigator’s signature was apparently not identified during the two audits; it came to attention after a CRA at the site reported this violation to the IRB. The information available from the Bayer audits confirms the FDA’s finding that data from the sties of Drs. Yang, Murray, Loucks, Corces, and Esquivel are not considered reliable. b. Describe Bayer’s QA program with respect to the oversight of CROs that were hired (b) (4) to monitor the clinical sites, including for the RECORD 4 study. Describe the procedures implemented to make sure that the CRO adequately monitored the clinical sites. In your response, include the following information: i. How was Bayer kept apprised by the CROs concerning monitoring of the clinical sites during the course of the study? Specifically, what information did the CROs provide? Provide a list of non-compliant study sites reported by the CROs. ii. How did Bayer review the information obtained from the CROs, during the course of the study and at the end of the study? What monitoring information was kept at the end of the study? iii. What actions did Bayer take based on the monitoring reports? Response to 1.b.i.-iii. Bayer provided the majority of the monitoring for the RECORD 1, RECORD 2, and RECORD 3 clinical trials. The applicant presents information on the CROs that provided monitoring for RECORD 1 in Israel, RECORD 2 in Portugal and India, and RECORD 3 in Israel due to the lack of Bayer monitoring facilities in these countries. The monitoring oversight of the CRO Reference ID: 2951276 Page 15 Compliance Review of NDA 22-406, Xarelto (b) (4) and processes for study documentation by the CROs were described for each of the non-Bayer CROs. (b) (4) provided the monitoring for most RECORD 4 sites. Monitoring of (b) (4) (b) (4) RECORD 4 sites in Pakistan was provided by (by a subcontract ) (b) (4) (b) (4) and by in Israel. The following information regarding role in RECORD 4 was presented: (b) (4) • was responsible for the monitoring and management of RECORD 4. • The Bayer Study Manager was responsible for overseeing the operational conduct of the CRO. This oversight included reviewing, tracking, analyzing, and summarizing (b) (4) the study related activities and the performance of . The Bayer Study Manager kept the Bayer Study Team and relevant member of Bayer management informed of the overall progress of the study via meetings or reports. (b) (4) • had a Project Leader responsible for the overall management of the trial, (b) (4) managed by the Director of Clinical Operations. • Processes implemented to ensure sufficient oversight of outsourced trials and to ensure the CRO adequately monitored the clinical sites. o Holding the Study Kick-off Meeting chaired by the Bayer Study Manager o The Task Definition Document (TDD) detailed the expectations of each task from initiating and conducting through closing out the clinical trial. An outline (b) (4) of expectations of Bayer and responsibilities was included in this document. The TDD detailed project management, study management, monitoring, medical management, and electronic data transfer/data management. (b) (4) o Routine meetings between Bayer and were conducted. o Generation of a Monthly Status Tracking Report to track details of the study (b) (4) • A Monitoring Plan was created by for RECORD 4, reviewed, and agreed upon by the Bayer Study Manager. The Monitoring Plan detailed roles, responsibilities, (b) (4) training plan, lines of communication (within external to sites), monitoring, and site management expectations. • All CRAs were trained on the Monitoring Plan, study documents, goals, and timelines. CRAs were the primary contact with the sites, maintained the Investigator Site Files (b) (4) (b) (4) within and informed the Project Leader of any site issues. (b) (4) • was responsible for ensuring appropriate training and supervised monitoring activities. (b) (4) • Lead CRAs or Project Leader reviewed and approved Monitoring Reports. They ensured proper follow up and resolution of issues. The Monitoring Visit Reports (b) (4) were posted on the website, and the Bayer Study Manager had (b) (4) access to this website. Bayer did not conduct routine reviews of the (b) (4) Monitoring Visit Reports as this task was assigned to the Lead CRA or Regional Project Leader. The applicant states that discussion of issues identified at (b) (4) Monitoring Visits were discussed at “frequent meetings” between and Bayer. Reference ID: 2951276 Page 16 Compliance Review of NDA 22-406, Xarelto (b) (4) DSI Assessment of Response: Johnson & Johnson has adequately responded to this request. The methodology outline for Bayer’s oversight of CROs used for the RECORD studies (b) (4) including should have been adequate. However, there are clearly monitoring inadequacies in the RECORD studies, most prominently in RECORD 4 which was (b) (4) (b) (4) monitored by . Although there were meetings between and Bayer, there was no routine exchange of problematic information regarding audit findings (nor was this required by (b) (4) the agreements between & Bayer). In addition, as discussed above, critical issues identified by other means (FDA inspections, third party audits) were not routinely identified by (b) (4) site monitoring. This raises concerns, particularly, as to the adequacy of monitoring of RECORD 4 studies. c. Independent Thirty Party Audits (b) (4) The following was requested in the CR letter: Provide assurance that the clinical data obtained from the RECORD 1, 2, 3, and 4 studies are reliable. Specifically, perform an additional audit and supply the results of this audit within your response to this letter. Within your response, include: i. A copy o f your audit plan, including the following information: • How many clinical sites were to be audited, how many subject records were examined, and a description of the process for selection of the audited sites. • If not all subject records at a given clinical site were to be audited, describe how subject records were sampled and how the specific data from each subject were audited. ii. The timeline for completion of your audit (plan finalization, start date, completion date, report finalization date). iii. In addition to any other information within your audit report, address the following questions or requests: • At each site audited, how many violations involved each of the following specific issues? For each specific violation, list the clinical sites involved and provide a breakdown by treatment group for each site and overall for the four RECORD studies. o Enrollment of subjects that did not meet study eligibility criteria. o Failure of the Principal Investigator to ensure that all associates and colleagues assisting in the investigation were meeting the commitments of the study protocol. o Failure to report adverse events and serious adverse events o Failure to randomize subjects preoperatively o Failure to obtain informed consent from all subjects • List all clinical sites where either Bayer or CRO monitoring is determined to be ineffective, either in identifying significant violations or in taking actions towards securing compliance (such as notifying the sponsor). Reference ID: 2951276 Page 17 Compliance Review of NDA 22-406, Xarelto (b) (4) Response to 1c.i and ii. (b) (4) Overview of Audits (b) (4) The audit program was conducted by an independent third party, . The studies included in the audit were the four pivotal Phase 3 studies of rivaroxaban 10 mg immediate-release tablets for the prophylaxis of deep vein thrombosis (DVT) and pulmonary embolism (PE) in patients undergoing hip replacement surgery or knee replacement surgery. The objective of the audit program was to provide assurance that the data obtained from the RECORD 1-4 studies are reliable. The audits assessed compliance with the protocol and appropriate Good Clinical Practice (GCP) requirements. Additionally, compliance with International Conference on Harmonization (ICH) Guidelines, the U.S. Code of Federal Regulations as set out in 21 CFR Parts 50, 54, 56, and 312, and, where applicable, local regulatory requirements was assessed. Selected documentation including protocols and monitoring visit reports was provided to the auditors by Johnson & Johnson. Bayer clinical operations representatives assisted with logistics and translations. The audit program focused on the specific areas of concern identified by the FDA in the CR letter in six categories: • • • • • • Informed Consent Investigational Product Source Data Verification and Case Report Completion Safety Study Conduct Monitoring Each audit observation was grouped by (b) (4) into one of the following categories: CRITICAL: An observation that requires prompt corrective action to ensure compliance with regulations, guidelines, company policy, or local law. These findings if unaddressed could compromise human safety, market authorizations, or the acceptability of investigational product, data, facilities, or systems intended for regulatory submission. Regulatory authority action would appear probable. MAJOR: An observation that requires improvement to ensure compliance with regulations, guidelines, company policy, or local law. These findings if unaddressed could compromise human safety, market authorizations, or the acceptability of investigational product, data, facilities, or systems intended for regulatory submission. Regulatory authority action would appear possible. MINOR: An observation where improvement is recommended for minor deviations from regulations, guidelines, company policy, or local law. There were 30 sites selected across the RECORD studies for auditing, including all 18 highenrolling sites with > 60 randomized subjects along with 12 moderate-enrolling sites with 1559 randomized subjects. Focused audits were performed on individual subject records for 100% of the subjects enrolled in each site that had up to 35 subjects. The 12 moderateenrolling sites (3 per study) were randomly selected by the Johnson & Johnson statistics group suing SAS Version 9.1 from a pool of all sites that met the stated enrollment criteria. For Reference ID: 2951276 Page 18 Compliance Review of NDA 22-406, Xarelto (b) (4) higher enrolling sites, focused audits were performed on a random sample of 35 to 43 subjects, depending on the number of subjects required to rule out a 5% error rate or higher with 95% confidence. The 30 site audits were conducted between July 27 and October 16, 2009 by teams of 2 auditors for 28 sites and by 1 auditor for 2 sites. The number of audited sites and subjects by study and overall is shown below, taken from the sponsor’s April 19, 2010 submission. Table 3: NDA 22-406: Study RECORD 1 RECORD 2 RECORD 3 RECORD 4 Overall (b) (4) RECORD Study Site Audits Audited sites/total Audited subjects/total sites subjects at audited sites (%) 11/217 (5.1% sites) 347/626 (55.4%) 7/123 (5.7%) 216/439 (49.2%) 3/147 (2.0%) 70/70 (100%) 9/130 (6.9%) 312/636 (49.1%) 30/617 (4.9%) 945/1771 (53.4%) Audited subjects/total study subjects (%) 347/4541 (7.6%) 216/2509 (8.6%) 70/2531 (2.8%) 312/3148 (9.9%) 945/12,729 (7.4%) (b) (4) Draft audit reports were reviewed by Johnson & Johnson QA personnel and comments (b) (4) relating to the consistency of reporting were provided for their consideration before the final reports were issued. The final audit reports were reviewed by Johnson & Johnson clinical and regulatory staff for consistency. Amended reports involved only the upgrading of findings. All of the audit reports were finalized by Johnson & Johnson by November 6, 2009 and all addenda by November 30, 2009. DSI Assessment of Response: Johnson & Johnson has adequately responded to this request. Note that across the 4 RECORD studies, 2.0-6.9% of sites were audited, with audits of 2.89.9% of total subjects in the studies. This will be taken in the context of audit findings as discussed below for each of the RECORD 4 studies. iii. In addition to any other information within your audit report, address the following questions or requests: • At each site audited, how many violations involved each of the following specific issues? For each specific violation, list the clinical sites involved and provide a breakdown by treatment group for each site and overall for the four RECORD studies. o Enrollment of subjects that did not meet study eligibility criteria. o Failure of the Principal Investigator to ensure that all associates and colleagues assisting in the investigation were meeting the commitments of the study protocol. o Failure to report adverse events and serious adverse events o Failure to randomize subjects preoperatively o Failure to obtain informed consent from all subjects • List all clinical sites where either Bayer or CRO monitoring is determined to be ineffective, either in identifying significant violations or in taking actions towards securing compliance (such as notifying the sponsor). Response to 1.c.iii. J&J Analysis of Reference ID: 2951276 (b) (4) Audits Page 19 Compliance Review of NDA 22-406, Xarelto (b) (4) (b) (4) Prior to submission of the complete set of audit reports, Johnson & Johnson submitted an analysis of the audits in a March 5, 2010 meeting background package. A brief summary of (b) (4) the Johnson & Johnson analysis is given here. Johnson & Johnson analyzed the audits using two approaches: 1. By six audit categories (informed consent, investigational product, SDV/CRF, safety, study conduct, monitoring). 2. By specific audit findings by classification category (critical, major, and minor). Across the 30 audited sites, there were a total of 251 findings. Nineteen of these findings were (b) (4) categorized by auditors as critical, 121 were categorized as major and 111 were categorized as minor. The number of major findings per site ranged from 1 to a maximum of 10, with 12 of the 30 sites having 5 or more major findings (RECORD 1: 3/11 [27%], RECORD 2: 2/7 [29%], RECORD 3: 1/3 [33%], RECORD 4: 6/9 [67%]). The 19 critical (b) (4) findings recorded by occurred at 13 of the 30 audited sites in the following categories: • 1 finding for Informed Consent o For one subject at one site, a signed consent form was not available. • 2 findings for Investigational Product o Documentation of the Investigational Product administration during the inpatient phase of the study was either missing or insufficient. • 6 findings for Source Data Verification and Case Report Form Completion o These critical findings can be further broken down into findings related to missing medical records (15 subjects at 4 sites), and significantly deficient and discrepant source documentation (2 sites). • 4 findings for Safety o These findings were associated with adverse events that weren’t reported and/or deficient safety reporting practices. • 5 findings for Study Conduct o These can be further broken down into findings related to eligibility (9 subjects; 1 each at 2 sites, 7 at one site), protocol violations for study drug treatment outside the protocol specified time window (19 subjects at one site), and an improperly constituted ethics committee (1 site). • 1 finding for Monitoring o The site monitor had failed to detect unreported adverse events, failed to detect late reporting for SAEs, and failed to meet with the principal investigator for 6 months, and failed to document training. Further, Johnson & Johnson noted that there were a total of 603 audit identified (AI) AEs in the 931 audited subjects from 28 of 30 sites audited. The highest proportion of subjects with AI AEs were in the RECORD 4 study. There were eight AI SAEs found, all from RECORD 4 sites; five of these were newly reported events and three were upgraded AEs. Johnson & Johnson concluded the following regarding unreported AEs: • Qualitatively, the most commonly reported AEs were similar in the audited subjects compared to those seen overall in the originally reported RECORD population. • The AI-AEs appear to be balanced between the two treatment groups and their inclusion does not substantially alter the previously reported event rates in the audited subjects. Reference ID: 2951276 Page 20 • • Compliance Review of NDA 22-406, Xarelto (b) (4) RECORD 4 was found to have the largest number of AI-AEs, and all of the unreported SAEs were identified exclusively in the RECORD 4 study. Overall, the identification of the AI-AEs and AI-SAEs did not alter the previously reported safety profile of rivaroxaban. The applicant notes that the efficacy endpoint in the RECORD studies was a hard composite endpoint of death, symptomatic VTE, or venographically detected VTE. They also note that (b) (4) the audits did not identify any evidence that would suggest that any of the venography data were not reliable; similarly, the audits did not identify any possibly missing or invalid symptomatic DVT or PE events. The sponsor concludes that the results of the RECORD studies are valid and reliable, but that the RECORD 4 study monitoring process should be specifically further addressed by a data validation plan, outlined in their submission. DSI Assessment of Response: The sponsor’s response is considered adequate to address the (b) (4) request in the CRL. In the following section, DSI will specifically analyze the Audits (b) (4) and will discuss the Audit findings considered critical to the evaluation of data reliability. (b) (4) III. DSI Analysis of Audits (b) (4) This section will provide DSI’s analysis of the Audits focusing on items deemed critical to evaluations data reliability: • Adequacy of Monitoring • Human Subject Protections and Adverse Event Reporting • Post-Operative Randomization • Drug Accountability • Eligibility This section will also briefly touch upon (b) (4) Verification of RECORD 4 Data. 1. Adequacy of Monitoring (b) (4) (b) (4) In one method used during the audits to assess adequacy of site monitoring, auditors reviewed each individual monitoring report. The Patient Data Check (PDC) Form was then completed for each subject, answering the question: “Was the monitoring effective, either in identifying significant violations or in taking actions towards securing compliance? The results are as follows (subjects inadequately monitored/subjects audited (%)) RECORD 1: RECORD 2: RECORD 3: RECORD 4: (b) (4) 96/347 (27.2%) 55/216 (25.5%) 28/70 (40.0%) 197/312 (63.1%) The audit reports note that inadequate monitoring was considered to be present at 2/11 (9%) of RECORD 1 sites, 2/7 (29%) of RECORD 2 sites, 1/3 (33%) of RECORD 3 sites, and (b) (4) 4/9 (44%) RECORD 4 sites (Table 4) according to assessment. Note that not all audit reports contained a specific statement regarding adequacy of monitoring. DSI review of the Reference ID: 2951276 Page 21 Compliance Review of NDA 22-406, Xarelto (b) (4) (b) (4) audit reports yielded an additional RECORD 4 site at which monitoring did not detect findings which would affect the primary efficacy or safety outcome (Table 4). For 3 additional audit reports (2 at RECORD 1 sites and 1 at a RECORD 2 site), it could not be determined from review of the audit report whether key primary efficacy or safety issues (b) (4) (b) (4) detected by the auditors were noted by the Bayer/ monitors. Key primary efficacy or safety issues included study drug administration inconsistencies between source documents and eCRF, drug accountability and dosing issues, identical drug dispensation times for all subjects, and inclusion of a subject with intraocular hemorrhage in (b) (4) violation of the exclusion criteria. Note that audits of many other sites demonstrated missed monitoring issues with respect to protocol deviations, adverse event reporting, source document verification, etc. However, only those instances in which monitoring omissions involved primary efficacy parameters or safety issues, which are considered critical for the evaluation of data integrity, are addressed here. In addition, instances where a single subject at a site had an issue impacting safety or efficacy are not included here since these instances would be unlikely to significantly impact overall site data reliability. Rather, review has focused on findings at sites where a substantial number of subjects were impacted at the site, such that overall data reliability of the site is in question. Please see Table 8 in Section III.5. for a listing of specific issues impacting on data reliability at individual sites. (b) (4) Table 4: Monitoring Adequacy and Issues Based on Audit Reports RECORD 1 RECORD 2 RECORD 3 RECORD 4* (b) (4) # of -audited Number sites Number sites Number sites Number sites Sites with (n/N;%) (n/N;%) (n/N; %) (n/N; %) Monitoring Deficiencies (n)/Total # of sites (b) (4) audited by (N) Per (b) (4) audit 2/11 (9%) 2/7 (29%) 1/3 (33%) 4/9 (44%) reports; DSI concurs Monitors missed 0/11 (0%) 0/7 (0%) 0/3 (0%) 1/9 (11%)** key primary efficacy or safety issue per DSI review of (b) (4) audit (b) (4) * was monitor **Sepulveda (Site 32002): Monitoring did not detect study drug administration inconsistencies between source documents and eCRF. The specific sites deemed by Johnson & Johnson analysis of the ineffective monitoring are the following: (b) (4) audits to have Garces, RECORD 1, Site 240002: Monitoring was inadequate to detect some unreported AEs, medical historical information, and protocol deviations; 90% of subjects audited had a “no” response given for the PDC question. Slappendel, RECORD 1, Site 30002: The Executive Summary of the audit states that monitoring is inadequate. 91% of subjects audited had a “no” response given for the PDC question. Reference ID: 2951276 Page 22 Compliance Review of NDA 22-406, Xarelto (b) (4) Ono, RECORD 2, Site 50005: 92% of subjects audited had a “no” response given for the PDC question. Wang, RECORD 2, Site 54001: Monitoring inadequate as judged by a significant number of eCRF versus source discrepancies; 46% of subjects audited had a “no” response given for the PDC question. Brabants, RECORD 3, Site 28015: 100% of subjects audited had a “no” response given for the PDC question. Kilgore, RECORD 4, Site 14034: 66% of subjects audited had a “no” response given for the PDC question. Reddy, RECORD 4, Site 60001: Numerous protocol/GCP deviations were unreported by the monitor; 100% of subjects audited had a “no” response given for the PDC question. V. Shah, RECORD 4, Site 60006: Numerous protocol/GCP deviations were unreported by the monitor; 100% of subjects audited had a “no” response given for the PDC question. H. Shah, RECORD 4, Site 60004: Numerous protocol/GCP deviations were unreported by the monitor; 100% of subjects audited had a “no” response given for the PDC question. Modi, RECORD 4, Site 60010: None of the issues noted in this report were noted as deviations by the monitor. 97% of subjects audited had a “no” response given for the PDC question. (b) (4) It could not be determined from DSI evaluation of the audit results from Dr. Garces and Dr. Wang’s sites whether inadequate monitoring of the site resulted in a deleterious effect on (b) (4) key primary efficacy and/or safety findings from those sites. However, review of the audit results themselves did not raise concerns as to data reliability of these sites. Johnson & Johnson also submitted reports of 74 clinical investigator site audits conducted by Bayer GCP Study Audit Management for the RECORD 1-4 studies, 69 of which were routine. (b) (4) Review of the Bayer audit reports for the clinical investigator sites for which and/or DSI considered that the data was unreliable (when available) showed that in the majority of instances, the violation considered most significant by DSI was not reported in the Bayer audit report. Significant deficiencies at the sites of Drs. Lenart (RECORD 1), Porvaneckas (b) (4) (RECORD 1), Nararrete (RECORD 2 ), and Buettner (RECORD 4), described in the audit reports were not mentioned in the Bayer audit reports. Although significant findings were identified at Dr. Michael Murray’s RECORD 4 site during the Bayer audit, the issues identified by FDA inspectors resulting in an OAI classification were not noted. The initial two Bayer audits at Dr. Craig Loucks site in RECORD 4 did not report forgery of the Principal Investigator’s signature, which was subsequently reported to the IRB by a site CRA. Failure to identify via site audits these serious regulatory violations identifies adequacy of monitoring as a problem in the RECORD trials, especially RECORD 4. Reference ID: 2951276 Page 23 Compliance Review of NDA 22-406, Xarelto (b) (4) DSI Assessment of Response: (b) (4) (b) (4) Based on DSI review of audit reports, auditors stated that overall study monitoring was deficient at 1 of 11 (9%) sites in RECORD 1, 2 of 7 (29%) sites audited in RECORD 2, 1 of 3 (33%) sites audited in RECORD 3, and 4 of 9 (44%) sites audited in RECORD 4. DSI concurs that monitoring was deficient at these sites. According to DSI (b) (4) review, monitoring failed to detect a key efficacy or safety issue in one additional instance in RECORD 4 (Dr. Sepulveda). It could not be determined from DSI (b) (4) evaluation of the audit results from Dr. Garces and Dr. Wang’s sites whether inadequate monitoring of the site resulted in a deleterious effect on key primary efficacy and/or safety findings from those sites. It should be noted that these findings impacted a substantial number of subjects at each site, such that overall data reliability of the sites is in question. Review of the audit reports for RECORD 1, 2, 3, and 4 submitted by Johnson & Johnson also (b) (4) revealed that Bayer/ audits did not always identify serious deficiencies. (b) (4) assessment of monitoring ineffectiveness by PDC forms showed that 63% of subjects in RECORD 4 audited were not monitored effectively. RECORD 1 and 2 had similar levels of unreliable monitoring, 27% and 29%, respectively. The relatively high level of ineffective monitoring (40%) noted in RECORD 3 is very likely reflective of the comparatively low number of subjects audited in RECORD 3 together with the presence of a problematic site (Brabants – see Section III.3. below) which enrolled 27 subjects. The frequency of monitoring ineffectiveness was less in RECORD 1, 2, and 3 as compared to RECORD 4; however, there was not as large a difference between monitoring ineffectiveness between RECORD 3 as compared to RECORD 4. However, as noted above, this assessment (b) (4) ineffectiveness by was based solely on PDC form checks. Note that in DSI’s assessment of monitoring adequacy of all 4 RECORD studies, assessment of monitoring effectiveness/ineffectiveness was not based solely on PDC form evaluation and respective percentages, but rather on the specific findings and their impact on data reliability. As such, perhaps from a percentage standpoint, it may be noted that monitoring ineffectiveness of 40% for RECORD 3 is not substantially different from the 63% monitoring ineffectiveness for RECORD 4 based on the PDC form check; however, taking into account not only PDC form checks, but also the extent and scope of deficiencies noted in RECORD 4, particular concerns are raised regarding data reliability of RECORD 4 based on evaluation of monitoring. Overall, monitoring deficiencies were noted for all 4 RECORD studies; however, in comparison to RECORD 1-3, the extent and scope of monitoring deficiencies noted for RECORD 4 are considered more significant and raise concerns regarding pervasiveness of monitoring deficiencies for other sites not inspected or audited, and as such undermine the confidence in reliability of the data. Reference ID: 2951276 Page 24 Compliance Review of NDA 22-406, Xarelto 2. Human Subject Projection and Adverse Event Reporting in (b) (4) (b) (4) Audit Reports Human Subject Protection In Table 5 below are presented clinical investigator sites where any instance of failure to (b) (4) protect human subject rights was noted during DSI review of the audits. Data from the sites of Drs. Brabants, Mody, and V. Shah were assessed by DSI as unreliable based on efficacy findings as given in Table 8 in Section III.5. Additionally, four women of childbearing potential were enrolled in RECORD 2 without performance of a pregnancy test; omission of the pregnancy tests were intentional, based on cultural factors. This protocol violation had the potential to significantly adversely impact any pregnancies which had been preexisting to the study. Although the events documented at the sites of Drs. Bauer, Marinoni, and Field are of substantial concern to DSI, they either involve a single individual or did not result in subject harm, and as such are unlikely to impact data reliability of these 3 specific sites. Table 5: Clinical Investigator Sites with Instances Where Subject Safety Was Not Protected Based on DSI (b) (4) Review of Audit Reports Study Clinical Investigator Detail Site number Number of subjects RECORD 1 Bauer One subject with untreated hypertension Site 44003 63 subjects RECORD1 Marinoni • Subject 4003 Subject had history of disturbed vision Site 22001 & ITP = exclusion criteria. Subject had “pre-retinal 15 subjects hemorrhage” Day 1, study medication continued. • 4 subjects had epidural catheters inserted or removed outside of protocol requirements; none of these catheters were recorded on the CRF. Two were placed too soon after study drug administration (1.5 and 2 hours) and 2 were withdrawn too soon after study drug administration (1.5 and 4 hours after dose, rather than 2X the halflife) RECORD 2 Field Subject 7989-251107 had a diagnosis of chronic renal Site 12008 insufficiency (CRI) per medical records, no screening (b) (6) 140 subjects labs reviewed prior to surgery , screening labs signed by PI 10/14/06, subject withdrawn due to elevated BUN/Cr on 10/13/06. RECORD 2 Wang Four of six women of child bearing potential did not Site 54001 have pregnancy test performed prior to enrollment in the 88 subjects trial RECORD 3 Brabants 9 of 27 subjects had screening procedures performed Site 28015 prior to signing Informed Consent 27 subjects RECORD 4 Mody • Ethics Committee (EC) impartiality could not be Site 60010 confirmed, as the EC was established at the PI’s 68 subjects request, and the members had no training or prior experience. • Clinician review of study documents (laboratory studies, ECGs) for 25/35 subjects (71%) was either not done or not done in a timely fashion. Example = ECG showing anterior wall myocardial ischemia. RECORD 4 V. Shah Language used to discuss the Informed Consent Reference ID: 2951276 Page 25 Compliance Review of NDA 22-406, Xarelto Site 60006 80 subjects (b) (4) document with all subjects was coercive, with documentation indicating that he said “that the study drug was completely safe, that it is the best treatment currently available, that risks were minimal (same as any other surgery). . .” Adverse Event Reporting (b) (4) The audit reports were reviewed in order to assess the adequacy of adverse event reporting. Only 2 of the 30 audited sites had no unreported adverse events identified during the audits. The number of unreported adverse events ranged from 1 to 54 per site. Eight unreported SAEs were identified, all at RECORD 4 sites. Unreported adverse events were assessed as “significant” by the DSI reviewer if they clearly required further expeditious medical evaluation; all events which included bleeding or elevation of liver function tests were included in this category. Anemia in itself was not considered “significant”. Table 6 below summarizes unreported AEs by clinical investigator site audited by Table 6: Unreported Adverse Events RECORD Investigator Number and study type of unreported SAEs RECORD 1 RECORD 1 RECORD 1 RECORD 1 RECORD 1 RECORD 1 RECORD 1 RECORD 1 Bauer Kruczynski Lenart Marinoni Mazurkiewicz Garces Pesola Porvaneckas 0 0 0 0 0 0 0 0 Number of unreported adverse events/number of subjects with unreported adverse events (Excludes SAEs) 4/4 1/1 1/1 5/4 4/4 8/6 2/2 22/12 RECORD 1 RECORD 1 Schwartsmann Slappendel 0 0 0 54/21 0 9 RECORD 1 Stehlik 0 12/11 1 RECORD 2 RECORD 2 Belickas Dhanjee 0 0 8/7 4/4 2 3 RECORD 2 Field 0 21/13 2 RECORD 2 Martson 0 9/5 3 RECORD 2 RECORD 2 Nafarrete Ono 0 0 34/10 37/16 4 7 Reference ID: 2951276 Number of unreported adverse events of significance – clearly required medical evaluation 1 0 0 1 0 1 0 3 (b) (4) . Examples of significant unreported adverse events GGT = 205 Abnormal ECG Disorientation Allergic skin reaction, elevated BP SOB, wound hematoma, calf red/painful, fever, low HR requiring Rx Hypotension and chest pain Fever Hypotension, calf pain, fever Leg swelling elevated GGT Thigh hematoma, fever, hypotension Infection Hypertension, nasal Page 26 Compliance Review of NDA 22-406, Xarelto Table 6: Unreported Adverse Events RECORD Investigator Number and study type of unreported SAEs RECORD 2 Wang RECORD 3 RECORD 3 RECORD 3 RECORD 4 Number of unreported adverse events/number of subjects with unreported adverse events (Excludes SAEs) Number of unreported adverse events of significance – clearly required medical evaluation (b) (4) Examples of significant unreported adverse events bleeding during surgery Hypertension, dyspnea Leg hematoma; Shaking with fever & hallucinations, drug-induced pancreatitis, elevated GGT = 275, ARI, decreased platelets, Na = 119 with K = 2.5, irregular HR, Tx 2U PRBCs, burning calf Fever, hypotension, UTI SOB, Elevated AST/ALT/GGT/alk phos Chest pain/breathing difficulties, Tx 2U PRBCs, fever, hypertension, amylase 0 18/14 3 Brabants Paulsson Synder Dessouki 0 0 0 1: cholecystitis/ cholecystectomy 36/20 1/1 0 18/15 2+* 0 0 14 RECORD 4 RECORD 4 Hollman Jove 0 0 7/6 33/13 0 7 RECORD 4 Kilgore 1: Respiratory failure 29/25 3 RECORD 4 Mody 47/21 7 RECORD 4 Reddy 38/10 10 Fever, elevated bilirubin, left bundle branch block, decreased platelets, elevated ALT RECORD 4 Sepulveda 3: Chest infection requiring hospitalization; bedsore requiring hospitalization; hypotension & SOB requiring transfer. 3: Grade II adenoCA of the prostate; pyrexia requiring hospitalization; hospitalization more than 12 hours for catheterization 0 13/25 8 Edema, hematoma, wound infection, Reference ID: 2951276 Page 27 Compliance Review of NDA 22-406, Xarelto Table 6: Unreported Adverse Events RECORD Investigator Number and study type of unreported SAEs Number of unreported adverse events/number of subjects with unreported adverse events (Excludes SAEs) Number of unreported adverse events of significance – clearly required medical evaluation RECORD 4 H. Shah 0 44/19 7 RECORD 4 V. Shah 0 36/17 5 (b) (4) Examples of significant unreported adverse events ALT/AST > 3X ULN Probable LVH, possible MI, pitting edema, neutropenia, irregular heart beat Fever, LE swelling, elevated ALT > 3X ULN *Many are unspecified abnormal hematology and chemistry values The total number of unreported adverse events for each RECORD study is as follows: RECORD 1 – 110; 16 significant* RECORD 2 – 131; 24 significant RECORD 3 – 37; 2+ significant RECORD 4 – 265; 61 significant Total RECORD studies – 543; 103+ significant *see note below in “DSI Assessment of Response” as to how “significant” was defined Although slightly more RECORD 4 subjects were audited than in other 3 studies (and RECORD 3 subjects were audited less frequently), it appears that RECORD 4 has a disproportionate number of unreported adverse events as well as unreported significant adverse events when compared with the other RECORD studies. In addition, RECORD 4 was the only study with unreported SAEs. DSI Assessment: (b) (4) audit reports of two clinical investigator sites of 11 audited for RECORD 1 (Drs. Bauer and Marinoni), 2 sites of 7 audited for RECORD 2 (Mr. Field and Dr. Wang), 1 site of 3 audited for RECORD 3 (Dr Brabants), and 2 sites of 9 audited for RECORD 4 (Drs. Mody and V. Shah) demonstrated instances where human subject rights were not protected during the conduct of the RECORD studies. However, the findings noted in Tables 5 and 6 above for Drs. Bauer, Marinoni, Field and Wang, are not considered pervasive in nature, and unlikely to impact data reliability for their respective RECORD 1-3 studies. The findings for Drs. Mody and Shah are concerning and provide further evidence for the distinction between study monitoring/conduct of RECORD 4 as compared to RECORD 1-3. Reference ID: 2951276 Page 28 Compliance Review of NDA 22-406, Xarelto (b) (4) (b) (4) audits of the majority of sites identified unreported adverse events, ranging from 0 (2 sites) to 54 per site. When adverse events considered significant by DSI (defined as adverse events which clearly required expeditious medical evaluation and all events including bleeding or elevation of LFTs), there were 16 significant unreported AEs in RECORD 1, 24 in RECORD 2, 2+ in RECORD 3 (exact number could not be determined), and 61 in RECORD 4. Unreported AEs and SAEs identified during the data verification process conducted by (b) (4) will be presented in Section III. 6. (b) (4) The finding of unreported adverse events during the audits did not alone result in a DSI determination that data from these sites were unreliable. However, the striking finding on examination of the number of unreported adverse events and SAEs per study is the (b) (4) disproportionate number of adverse events detected during the audits of RECORD 4 (more than twice the number of undetected AEs and significant AEs) when compared with the smaller numbers reported from RECORD 1, 2, and 3. Of additional concern, are the eight (b) (4) unreported SAEs noted by auditors from the RECORD 4 audits, whereas no undetected SAEs were reported from RECORD 1, 2, or 3. The disproportionate number of adverse events (b) (4) detected during the audits of RECORD 4 when compared with RECORD 1, 2, and 3, as well as the detection of unreported SAEs only in RECORD 4 brings into question the adequacy and completeness of the RECORD 4 safety data submitted to the Agency. In addition, the relatively large number of unreported adverse events raises further concern regarding the adequacy of study conduct and monitoring of RECORD 4. 3. Post-operative Randomization in (b) (4) Audit Reports As noted in Table 2, post-operative randomization was identified by FDA audits for the NDA submission at 3 clinical investigator sites (Drs. Murray, Ward, and Buettner), all enrolling in (b) (4) RECORD 4, in violation of the protocol. This is despite the fact that the CRO monitoring RECORD 4, sent an email to all sites during the clinical trial reiterating the protocol requirement that subjects be randomized prior to surgery. One FDA inspection noted that the investigator gave permission to randomize after the patient stopped oozing at the surgical wound site. As part of the CR, Johnson & Johnson determined the incidence of postoperative randomization at all RECORD sites. The results are as follows: (b) (4) Postoperative randomization was assessed in most RECORD 4 audit reports with the following specific information regarding post-operative randomization (number of subjects randomized postoperatively/total subjects enrolled in study (%)): RECORD 1: RECORD 2: RECORD 3: RECORD 4: 18/4541 (0.4%) 13/2509 (0.5%) 9/2531 (0.4%) 1227/3148 (39.0%) Dessouki - 18/35 randomized day of surgery, no time stamp on IVRS form Hollman - Two subjects randomized post-operatively Jove - No subjects randomized post-operatively Reference ID: 2951276 Page 29 Compliance Review of NDA 22-406, Xarelto (b) (4) Kilgore - “Majority” of subjects randomized day of surgery, no time stamp on IVRS form Mody - 34/35 subjects randomized day of surgery, no time stamp on IVRS form Reddy - 12/40 subjects randomized post-operatively, deviation forms on file for 11 of these 12 subjects, in 7/40 subjects the time of randomization couldn’t be determined Sepulveda - 9 subjects were randomized on the day of surgery, no time stamp on IVRS form; there was no randomization sheet available for 1 subject H. Shah - 1 post-operative randomization V. Shah – no subjects noted to be randomized post-operatively (b) (4) Based on these results, there are three RECORD 4 clinical investigator sites from the audits where subjects were randomized postoperatively (Drs. Hollman, Reddy, and H. Shah; at Dr. Reddy’s site, these events were identified as protocol violations). At four additional sites (Dr. Dessouke, Kilgore, Mody and Dr. Sepulveda), it cannot be determined from the (b) (4) information available in the audit reports what proportion of the subjects were (b) (4) randomized postoperatively. Therefore, based on the audit reports, post-operative randomization occurred at a significant number of clinical investigator sites enrolling in RECORD 4. This protocol violation occurred in 3 of 5 of the sites originally inspected for the (b) (4) NDA, and to varying degrees in 3 additional sites audited by in addition, it cannot be determined from the site records whether subjects randomized on the day of surgery were in fact randomized post-operatively. DSI Assessment of Response According to Johnson and Johnson, postoperative randomization took place for 1227 of 3148 (b) (4) (b) (4) (39%) of RECORD 4 subjects, audited and nonaudited by Based on the audit results, 3 of the 9 sites audited for RECORD 4 randomized postoperatively; at 4 additional (b) (4) sites, it cannot be determined from the information available in the audit reports what proportion of the subjects randomized on the day of surgery were in fact randomized postoperatively. Although such postoperative randomization errors would occur in both arms of the clinical trial, it has the potential to alter the patient population included in the RECORD 4 study. If sufficient sites enrolled subjects postoperatively, especially based on specific criteria, the population included in the Xarelto product label may not reflect the population actually studied. The review division will need to assess the impact of this issue on potential product labeling. The high incidence of this protocol violation again reinforces the monitoring (b) (4) deficiencies in RECORD 4. Although was aware of the occurrence of postoperative randomization, they did not effectively enforce compliance with this protocol requirement. Post operative randomization did not occur to any significant degree in RECORD 1, 2, or 3. 4. Drug Accountability Issues in (b) (4) (b) (4) Audits Review of the audit reports focused on identification of clinical investigator sites where there were documentation issues for study drug administration and/or storage. Attention was focused on identified problems with drug administration of accountability and/or administration, such that uncertainty existed as to whether subjects actually received the assigned study drug which had been stored appropriately to maintain activity. If subjects did not receive study drug as described in the data listings, the primary efficacy outcome could potentially be compromised. Reference ID: 2951276 Page 30 Compliance Review of NDA 22-406, Xarelto (b) (4) (b) (4) DSI concurs with assessment of data from Dr. Brabants site (RECORD 3) as unreliable due to inadequacies in study drug administration documentation. Based on a review (b) (4) of the audit reports, DSI identified four additional sites for RECORD 1 (Drs. Lenart, Porvaneckas, Schwartsmann, and Slappendel), two additional sites for RECORD 2 (Drs. Naraffete and Ono), and three additional sites for RECORD 4 (Drs. Mody, Sepulveda, and Shah) which have sufficient deficits in drug administration and accountability that DSI cannot verify subjects received study drug as purported. Details of drug accountability issues for each CI are given in Table 7. At each of the additional sites, source documentation for study drug administration was missing or lacking, and/or there were significant issues with documentation of drug accountability such that it does not appear possible to verify that subjects at the site received active/correct study drug therapy. (b) (4) Table 7: Clinical Investigator Sites with Drug Administration and Accountability Issues Based on Audits Clinical Investigator Study Assessment Source Major Drug Location Site Number (FDA Inspections, Accountability/Administration (b) (4) Number of Subjects Audit Reports, Issues (b) (4) DSI Review of Audit Reports) Robert Slappendel RECORD 1 DSI review of (b) (4) • No source documentation Netherlands Site 30002 site audits for date/time of the pre61 subjects operative self-administered injection of enoxaparin/placebo by the subject or the date and time of last outpatient dosing • 10 of 35 subjects audited had drug accountability records which were incomplete and/or discrepant with other subject source documentation. Endre Lenart Hungary RECORD 1 Site 46002 87 subjects DSI review of audit reports (b) (4) Narunas Porvaneckas, Lithuania RECORD 1 Site 57001 72 subjects DSI review of audit reports (b) (4) Edmundo Berumen Naraffete RECORD 2 Site 32005 25 subjects RECORD 2 Site 50005 24 subjects FDA review of site audits (b) (4) FDA review of site audits (b) (4) Keiske Ono Brazil Reference ID: 2951276 Study coordinators log used to document drug accountability and dosing for all subjects, but entries in log were not dated/initialed Study drug administration times were exactly the same for all 34 subjects audited. Exact dosing times were not documented. Study drug administration times were exactly the same for each subject for all subjects audited • Documentation of study drug administration during inpatient phase of study was missing or deficient: 8 subjects records contained Page 31 Compliance Review of NDA 22-406, Xarelto (b) (4) Karl Brabants Belgium RECORD 3 Site 28015 27 subjects Bharat Mody India RECORD 4 Site 60010 68 subjects FDA review of site audit (b) (4) Victor Sepulveda Mexico RECORD 4 Site 32002 FDA review of site audit (b) (4) Reference ID: 2951276 site audits (b) (4) very few notations that study drug had been given, and the remaining 16 records contained none. Doses documented on the SDW were not signed/initialed or dated • Large number of discrepancies between eCRF, SDW, and medical chart information (73 discrepancies for 20 subjects – e.g. surgery start/stop time, intraoperative blood loss, drain volume) • Exact time of study drug administration was rarely recorded on the inpatient medication administration records for any of the 27 subjects – only on grid with 0800, 1200, 1600, and 2000 time points • Times of study drug administration frequently do not match the times noted on the inpatient medication administration sheets • Study coordinator was unable to define a consistent primary source for many of the data points, including drug dosing, surgery start/stop times, and laboratory draw times. • Drug accountability logs provided by Bayer were not used by the study coordinator to record drug accountability and the site did not keep a log of accountability • Ambient temperatures in study drug storage room was monitored weekly, not daily Study drug not stored in permissible temperature range of 15-30oC for 19 consecutive days, dropping to 10.2oC each day • Medical records of 10 subjects were missing from Page 32 Compliance Review of NDA 22-406, Xarelto 46 subjects • V. Shah India RECORD 4 Site 60006 80 subjects DSI review of site audit (b) (4) • • (b) (4) the site. Nursing notes, which include dosing entries, were missing for an additional 7 subjects 15 of 33 subjects audited had source vs eCRF discrepancies pertaining to study drug administration noted (ranging from 1 to all doses, most = 2-3 doses) Data discrepancies exits between the eCRF and site source documentation, including for study drug administration (26 subjects, 23 instances) Missing source documentation of drug administration for 8 of 35 subjectsc (b) (4) Further information was requested by the FDA on August 2, 2010 regarding the audit findings at eight clinical sites (sites with significant drug accountability issues as identified in Table 7 above). This Information Request was intended to obtain any additional information (b) (4) which might be available at the clinical sites to clarify what the auditors considered to be inadequate drug accountability. Please see Appendix 2 for details of DSI requests, response/finding from Johnson & Johnson received on September 26, 2010, and DSI assessment of the additional information provided. Johnson & Johnson sent monitoring personnel to seven of the clinical sites in question; the entire team including Dr. Slappendel and the Study Coordinator is no longer present at his site, so the RECORD 1 study team attempted to provide additional clarification. The results of the site revisit to Dr. Schwartsmann’s site provided sufficient evidence that study drug was given appropriately. However, the data provided for the other sites was insufficient to provide such reassurance. DSI Assessment of Response In conclusion, issues in drug accountability were identified across the RECORD studies, most seriously in RECORD 4. In some instances, it appears that routine hospital practice was followed (e.g., physician notes what time medication should be given and this information is copied onto a nurse’s sheet, with initials/dates/times of drug administration recorded only if there were variations from this procedure). However, for purposes of a clinical trial, it is imperative that documentation sufficient to assure that medication was actually administered to study subjects be provided in the source documents. The absence of actual dates/times of drug administration as well as initials of the person administering the medication results in an inability to have confidence that the subject actually received the medication as specified in the protocol. (b) (4) Overall, there were some drug accountability issues identified by audits at sites from all of the RECORD studies. The statistical import of the single site in RECORD 3 identified with Reference ID: 2951276 Page 33 Compliance Review of NDA 22-406, Xarelto (b) (4) significant drug accountability issues is difficult to assess, given that only 3 sites were audited from RECORD 3. We acknowledge that RECORD 1 had 3 sites with significant drug accountability issues, and RECORD 2 had 2 sites. These findings for RECORD 1 and 2, when interpreted together with the failure to identify deficiencies in drug accountability in FDA inspection and the relatively small number of sites audited, do not allow extrapolation to the conclusion that all sites from RECORD 1 and 2 had drug accountability deficiencies sufficient to impugn data integrity from all sites in these studies. The ultimate decision regarding overall study reliability must be based on the totality of evidence pertinent to good clinical trial conduct. In contrast to RECORD 1, 2, and 3, however, RECORD 4 had 3 sites identified with (b) (4) significant drug accountability issues by audit, in addition to the 2 RECORD 4 sites (Cources and Esquivel) already identified by DSI as unreliable based on drug accountability issues, among other violations. This suggests that drug accountability deficits are more pervasive at RECORD 4 sites. Please see Section IV. Below for a further discussion of the effect of drug accountability on overall study data reliability. 5. Eligibility Criteria in the (b) (4) Audits One item of concern identified during the initial cycle of FDA inspections was enrollment of (b) (4) subjects in violation of the protocol inclusion criteria. Review of the audit reports revealed a few subjects enrolled who did not meet eligibility criteria, but this was not a frequent finding. DSI Assessment of Response Enrollment of ineligible subjects does not appear to be a systematic problem in the RECORD studies. (b) (4) 6. Verification of RECORD 4 Data (b) (4) As described in Section C.1. above, deficiencies in monitoring by and study conduct issues appeared to be more severe and widespread in the RECORD 4 study when compared with the RECORD 1, 2, and 3 studies. Therefore, the applicant proposed a data verification plan in an attempt to demonstrate the validity of the RECORD 4 data. The sponsor’s plan was presented to the Division of Hematology Products and DSI on April 7, 2010. The goal of the data verification was to identify any AEs or SAEs present in the subjects’ medical records that were not reported to Bayer before the time of finalization of the study, to assess overall site and investigator quality, to assess the impact of postoperative randomization, and to address the Agency’s areas of concern regarding study reliability. All (b) (4) sites participating in RECORD 4 were to be visited by site monitors from , an independent CRO. Please see the CR document dated December 23, 2010 for details of the data verification plan. After revisiting all RECORD 4 sites, there were 260 newly identified treatment emergent adverse events in the rivaroxaban arm and 244 in the enoxaparin arm. This resulted in an increase in the reported rate of adverse events from approximately 80% of subjects originally reported with adverse events to 97% of subjects following data verification; there was no change in the distribution of AEs between treatment groups. There were 28 newly reported SAEs in 25 subjects (15 rivaroxaban, 12 enoxaparin, and 1 never randomized). There were 2 Reference ID: 2951276 Page 34 Compliance Review of NDA 22-406, Xarelto (b) (4) newly-reported cases of ALT>3X ULN concurrent with a total bilirubin >2X ULN identified, both in subjects receiving enoxaparin. No new events of death, DVT, or PE were identified. (b) (4) monitors were instructed to answer a series of questions regarding site and investigator overall performance. Sites and investigators were then ranked according to quality. Sensitivity analyses were performed for primary efficacy and safety by high versus low quality sites/investigators. This procedure was intended to address site performance concerns raised by the agency. Based on this procedure, the sponsor concludes that the primary efficacy and safety results remained essentially unchanged in the groups of sites with performance (b) (4) considered by as acceptable or questionable, compared to that seen in the overall patient population. (b) (4) In order to address the issue of postsurgical randomization, compared outcomes in subjects treated preoperatively versus postoperatively in both treatment arms. The applicant states that their results demonstrate that when the rates of the efficacy and safety outcomes were calculated in the two treatment groups for subjects randomized preoperatively versus postoperatively, they appeared to be comparable and similar to those results seen overall. DSI Assessment of Response (b) (4) The data verification audits of the RECORD 4 study sites were conducted with the intention of reassuring the Agency of the robust nature of the RECORD 4 data. The identification of 504 new treatment emergent adverse events as well as 28 newly identified SAEs in the RECORD 4 study provokes more concern than reassurance. Although it is certainly possible that there are unreported adverse events and SAEs in clinical trials in general, the number of unreported adverse events in the RECORD 4 trial seems excessive. Additionally, these newly reported adverse events reaffirm the concern that monitoring of the RECORD 4 trial was inadequate. Similarly, it is reassuring that no difference in efficacy or safety outcome was noted in subjects randomized pre- versus postoperatively. However, the large number of subjects randomized postoperatively in violation of the protocol raises again the issue of adequacy of study monitoring. The portion of the data verification process in (b) (4) which assessed site and investigator overall performance which was then correlated with efficacy and safety outcome is interesting, but not a validated method of assessing study conduct. DSI remains concerned with the deficiencies in clinical trial conduct and monitoring of RECORD 4 with potential deleterious effects on the validity of the efficacy and safety data from the RECORD 4 study. (b) (4) IV. DSI Review of Audits – Unreliable Sites (b) (4) In order to assess whether or not the findings from the audits significantly impacted overall data reliability from each CI site, DSI reviewed the 30 audit reports in detail. At many (b) (4) sites, auditors identified issues with study conduct, unreported adverse events, drug disposition and accountability, informed consent, source document verification and case report completion, and monitoring. If findings at a site involved more than a few subjects or appeared to significantly impact key efficacy assessments for multiple subjects, then DSI considered data from the site to be unreliable. Please see Table 8 for details of sites with efficacy data considered unreliable by DSI, which is based on review of the totality of (b) (4) information available to DSI, to include Bayer audits, audits, as well as FDA Reference ID: 2951276 Page 35 Compliance Review of NDA 22-406, Xarelto (b) (4) (b) (4) inspections. The identification by monitors that adverse events had not been fully reported did not in and of itself result in site assessment as unreliable, especially if the balance of the issues rendered the site data assessment as reliable. As seen in Table 8 below, data from the following clinical investigators had been previously identified based on DSI review of FDA inspections as unreliable with the recommendation that it should not be used in support of the application: RECORD 2: Drs. Corces and Yang; and RECORD 4: Drs. Loucks, Esquivel, Murray, Ward, and Buettner. DSI concurs with Falcon’s assessment of data from Dr. Brabants site (RECORD 3) as unreliable, and this data should not (b) (4) be used in support of the application. Based on a review of the audit reports, DSI identified three additional sites for RECORD 1 (Drs. Lenart, Porvaneckas, and Slappendel), two additional sites for RECORD 2 (Drs. Naraffete and Ono), and three additional sites for RECORD 4 (Drs. Mody, Sepulveda, and Shah), which in DSI’s opinion, provided unreliable data, and this data should not be used in support of the application. Note that the Executive (b) (4) Summary contained in the audit reports for each of these investigators lists multiple issues identified at each of these 8 sites, but stops short of stating that the data are unreliable. (b) (4) Only the data from Dr. Brabants’ site was classified as unreliable by the auditors. At each of the additional sites, source documentation for key efficacy assessments was missing or lacking, and/or there were significant issues with documentation of drug accountability such that it does not appear possible to verify that subjects at the site received active/correct study drug therapy. The following table summarizes the reasons DSI recommends that data from individual CI sites be considered unreliable and not be used in support of the NDA. The source (b) (4) of the recommendation is also given as FDA inspection, audit report, or DSI review of (b) (4) audit report. Table 8: Clinical Investigator Sites with Efficacy Data Considered Unreliable by DSI Clinical Investigator Study Assessment Source Primary Reason DSI Assesses Data Location Site Number (FDA Inspections, from Site to be Unreliable (b) (4) Number of Subjects Audit Reports, DSI Review (b) (4) of Audit Reports) Endre Lenart RECORD 1 DSI review of (b) (4) Study coordinators log used to Hungary Site 46002 audit reports document drug accountability and 87 subjects dosing for all subjects, but entries in log were not dated/initialed, and as such can’t verify accuracy of subject dosing. Narunas Porvaneckas, RECORD 1 DSI review of (b) (4) Study drug administration times were Lithuania Site 57001 audit reports exactly the same for all 34 subjects 72 subjects audited. Exact dosing times were not documented, As such, can’t verify accuracy of subject dosing. RECORD 1 DSI review of (b) (4) Robert Slappendela • No source documentation for Netherlands Site 30002 site audits date/time of the pre-operative 61 subjects self-administered injection of enoxaparin/placebo by the subject or the date and time of last outpatient dosing • 10 of 35 subjects audited had Reference ID: 2951276 Page 36 Compliance Review of NDA 22-406, Xarelto (b) (4) Table 8: Clinical Investigator Sites with Efficacy Data Considered Unreliable by DSI Clinical Investigator Study Assessment Source Primary Reason DSI Assesses Data Location Site Number (FDA Inspections, from Site to be Unreliable (b) (4) Number of Subjects Audit Reports, DSI Review (b) (4) of Audit Reports) drug accountability records which were incomplete and/or discrepant with other subject source documentationb. As such, can’t verify accuracy of subject dosing. Arturo Corces RECORD 2 FDA inspection Recordkeeping and drug disposition Miami, U.S.A. Site 14012 deficiencies, considered significant 19 subjects enough to raise concerns regarding data reliability. Qingming Yang RECORD 2 FDA inspection Failure to report AEs, significant for China Site 54005 evaluation of safety data as well as 34 subjects human subject protection Edmundo Berumen RECORD 2 FDA review of (b) (4) Study drug administration times were Naraffete Site 32005 site audits exactly the same for each subject for 25 subjects all subjects audited; as such, can’t verify accuracy of subject dosing. Keiske Ono RECORD 2 FDA review of (b) (4) • Documentation of study drug Brazil Site 50005 site audits administration during inpatient 24 subjects phase of study was missing or deficient: 8 subjects records contained very few notations that study drug had been given, and the remaining 16 records contained none. Doses documented on the SDW were not signed/initialed or dated • Large number of discrepancies between eCRF, SDW, and medical chart information (73 discrepancies for 20 subjects – e.g. surgery start/stop time, intraoperative blood loss, drain volume) The findings raised significant concerns with respect subject dosing as well as adequacy and accuracy of data on CRFs, of significant concern to impact data reliability. (b) (4) Karl Brabants RECORD 3 site audits • Exact time of study drug Belgium Site 28015 administration was rarely 27 subjects recorded on the inpatient medication administration records for any of the 27 subjects – only on grid with 0800, 1200, 1600, and 2000 time points • Times of study drug administration frequently do not Reference ID: 2951276 Page 37 Compliance Review of NDA 22-406, Xarelto (b) (4) Table 8: Clinical Investigator Sites with Efficacy Data Considered Unreliable by DSI Clinical Investigator Study Assessment Source Primary Reason DSI Assesses Data Location Site Number (FDA Inspections, from Site to be Unreliable (b) (4) Number of Subjects Audit Reports, DSI Review (b) (4) of Audit Reports) match the times noted on the inpatient medication administration sheets • Study coordinator was unable to define a consistent primary source for many of the data points, including drug dosing, surgery start/stop times, and laboratory draw times. • Drug accountability logs provided by Bayer were not used by the study coordinator to record drug accountability and the site did not keep a log of accountability • Ambient temperatures in study drug storage room was monitored weekly, not daily David Loucks Colorado, U.S.A. Ricardo Esquivel Mexico R. Michael Murray Alabama, U.S.A. John Ward Alabama, U.S.A. Craig Buettner Alabama, U.S.A. Bharat Mody India Victor Sepulveda Mexico Reference ID: 2951276 • • • • RECORD 4 Site 14029 94 subjects RECORD 4 Site 32006 42 subjects RECORD 4 Site 14005 152 subjects RECORD 4 Site 14010 203 subjects RECORD 4 Site 14004 61 subjects RECORD 4 Site 60010 68 subjects FDA inspection FDA review of site audit (b) (4) RECORD 4 Site 32002 46 subjects FDA review of site audit (b) (4) Bayer monitoring • • • FDA inspection Recordkeeping deficiencies Falsification Protocol violations Drug disposition record deficiencies Missing records Post-operative randomization Possible unblinding • • FDA inspection Post-operative randomization Study continued despite lapse of IRB approval Post-operative randomization FDA inspection Study drug not stored in permissible temperature range of 15-30oC for 19 consecutive days, dropping to 10.2oC each day • Medical records of 10 subjects were missing from the site. Nursing notes, which include dosing entries, were missing for an additional 7 subjects • 15 of 33 subjects audited had source vs eCRF discrepancies pertaining to study drug administration noted (ranging from 1 to all doses, most = 2-3 Page 38 Compliance Review of NDA 22-406, Xarelto (b) (4) Table 8: Clinical Investigator Sites with Efficacy Data Considered Unreliable by DSI Clinical Investigator Study Assessment Source Primary Reason DSI Assesses Data Location Site Number (FDA Inspections, from Site to be Unreliable (b) (4) Number of Subjects Audit Reports, DSI Review (b) (4) of Audit Reports) doses) RECORD 4 V. Shahb DSI review of (b) (4) • Data discrepancies exits between India Site 60006 site audit the eCRF and site source 80 subjects documentation, including for study drug administration (26 subjects, 23 instances) • Missing source documentation of drug administration for 8 of 35 subjectsc • Use of inappropriate correction techniques in all subject records • For 3 subjects, source documentation and eCRF entries were changed months after an event, sometimes in response to a query from data management. • Language used to discuss the Informed Consent document with all subjects was coercive, with documentation indicating that he said “that the study drug was completely safe, that is the best treatment currently available, that risks were minimal (same as any other surgery). . .” a Evaluation of data submitted by Johnson & Johnson resulted in assessment of data from 4 of the 10 subjects in question at this site as acceptable; see Section III and Appendix 2; however, the data overall from this site is still considered unacceptable. b Evaluation of data submitted by Johnson & Johnson resulted in assessment of data from 1 of the 8 subjects in question at this site as acceptable; see Section III and Appendix 2; however, the data overall from this site is still considered unacceptable. DSI Assessment of Response: In addition to sites previously identified, based on DSI inspections as providing unreliable data with the recommendation that data from the sites not be used in support of the NDA (Drs. Corces and Yang for RECORD 2, and Drs. Loucks, Esquivel, Murray, Ward, and Buettner for (b) (4) RECORD 4), DSI concurs with auditors that data from Dr. Brabants’ site enrolling in RECORD 3 be considered unreliable and that it not be used in support of the NDA. This recommendation is based on deficiencies in documentation of drug administration, such that certainty regarding study drug administration is not possible. (b) (4) Based on review of the audit reports, DSI identified 3 additional sites for RECORD 1 (Drs. Lenart, Porvaneckas, and Slappendel), 2 additional sites for RECORD 2 (Drs. Naraffete and Ono), and 3 additional sites for RECORD 4 (Drs. Mody, Sepulveda, and V. Shah) from which DSI considers key study data to be unverifiable or unreliable and recommends that data Reference ID: 2951276 Page 39 Compliance Review of NDA 22-406, Xarelto (b) (4) from these sites also not be used in support of the application. At each of these additional sites, source documentation was missing and/or there were significant issues with documentation of drug accountability such that it does not appear possible to verify that subjects at the site received active/correct study drug therapy. As such data is not recommended for use from the following sites for their respective studies: RECORD 1: Drs. Lenart, Porvanceckas, and Slappendal RECORD 2: Drs. Coreces, Yang, Naraffete, and Ono RECORD 3: Brabants RECORD 4: Drs. Loucks, Esquivel, Murray, Ward, Buettner, Mody, Sepulveda, and Shah (b) (4) DSI’s assessment of how the inspectional/ audit findings impact data reliability as a whole to each individual study based on the information available to DSI for review, is discussed in the next section. V. DSI Overall Assessment of RECORD 1, 2, 3, and 4 Studies Based on and FDA inspections (b) (4) Audits Inadequacies of study conduct and monitoring identified in the RECORD 1, 2, 3, and 4 studies during the initial NDA review cycle resulted in the request by DSI for independent third party (b) (4) audits of clinical investigator sites, which were conducted by Table 9 below (b) (4) summarizes the issues identified during FDA inspections and the audits which are considerations in the assessment of the overall integrity of each RECORD study. Clearly, drug accountability issues at a significant number of sites in each RECORD study (b) (4) raises the fundamental issue of whether DSI is able (based on inspectional findings and audit results) to confirm that subjects at each site received study drug as given in the line listings submitted with the NDA. It can be seen in Table 9 that significant drug accountability issues (i.e. affecting more than a few subjects) were noted all 4 RECORD studies, ranging (b) (4) from 27 – 33% of audited sites. Since only three RECORD 3 sites were audited, the statistical significance of this finding for RECORD 3 is uncertain. In consideration of the potential impact of drug accountability issues on overall study data integrity, DSI evaluated other determinants of study reliability. A major determinant which enables DSI to generalize the results of audit or inspectional findings is adequacy of clinical trial monitoring. If monitoring is inadequate at the majority of sites examined, it becomes impossible for DSI to provide assurance that study conduct flaws (e.g., in drug accountability) did not occur at the vast majority of clinical sites which were not audited or inspected – or that other, undetected flaws impacting on safety and efficacy data did not occur. The same principle holds true for (b) (4) assessment of the number of sites assessed as unreliable after audit or FDA inspection. Given the relatively small percentage of subjects and sites examined, consideration must be given to interrelated study conduct issues (e.g., number of unreliable sites together with ineffective monitoring in a given study) – that is, the more essential elements of good study conduct that are defective in a given study, the more likely that overall data integrity for that study is unreliable. Lastly, DSI considered the relative number of unreported adverse events and serious adverse events in the assessment of overall study integrity. Although each RECORD study had flaws which had the potential to affect data integrity, DSI took a global Reference ID: 2951276 Page 40 Compliance Review of NDA 22-406, Xarelto (b) (4) approach in applying analysis of each study conduct element to overall RECORD study reliability. We are of the opinion that assessment of significant site inadequacies in a given study across all examined study conduct issues allows a more accurate assessment of the impact of these issues on data integrity. Findings of deficits in a single area of study conduct makes extrapolation of assessment of data integrity as unreliable, problematic across an entire study, given the relatively small proportion of sites assessed. It seems reasonable, however, to have a higher level of confidence in drawing a conclusion that data integrity is unreliable, based on a small audit/inspectional sample for a given study, when all study conduct elements examined are significantly flawed. Please see discussion after Table 9 for application of these concepts to each RECORD study. Reference ID: 2951276 Inadequate monitoring (b) (4) overall assessment by site assayed (sites with inadequate monitoring/si tes audited) (%) 2/11 (18%) Inadequate monitoring (b) (4) overall assessment by subject assayed #sites unreliable (Sites unreliable/total sites audited by (b) (4) +FDA inspected) Compliance Review of Reference ID: 2951276 NDA 22-406, Xarelto No Yes – except Lenart, Porvaneckas and Slappendal Yes – except Corces, Yang Naraffete, and Ono Yes – except Brabants Overall study reliability (b) (4) RECORD 1 18/4541 (0.4%) 110/16/0 NA 3/11 (27% of 96/347 3/13 (23%) (217 sites) audited sites) (27.2%) RECORD 2 13/2509 (0.5%) 131/24/0 NA 2/7* (29% of 55/216 2/7 (29%) 4/10 (40%) (123 sites) audited sites) (25.5%) RECORD 3 9/2531 (0.4%) 37/2+/0 NA 1/3 (33% of 28/70 1/3 (33%) 1/5 (20%) (147 sites) audited sites) (40.0%) RECORD 4 1227/3148 265/61/8 504/28 3/9* (33% of 197/312 6/9 (67%) 8/16 (50%) (130 sites) (39.0%) audited sites) (63.1%) *1 additional site each from RECORD 2 and 2 additional sites from RECORD 4 had critical drug accountability issues identified during FDA inspections. TABLE 9: EVALUATION OF RECORD 1, 2, 3, AND 4 DATA INTEGRITY Study Post-operative Unreported Unreported Drug Randomization Adverse Adverse accountability #subjects Events – Events – issues (b) (4) (b) (4) POR/total (critical) subjects (%) audits AEs/ audits AEs/ Sites with significant SAEs issues/sites AEs/SAEs audited by (b) (4) Page 40 Page 41 (b) (4) Compliance Review of NDA 22-406, Xarelto DSI Assessment of RECORD 1 Reliability: DSI recommends that the data from this study can be used in support of the NDA. Although there were drug accountability issues identified at (b) (4) 27% of audited sites in RECORD 1, monitoring was assessed as adequate in the majority of subjects and sites, and earlier FDA inspections did not reveal drug accountability (b) (4) issues. Based on review of audit findings, however, there were 3 sites in RECORD 1 (Lenart, Porvaneckas, and Slappendal) for which DSI cannot assure data reliability (due to drug accountability issues). DSI acknowledges that there were unreported adverse events from this trial, and suggests that the review division consider additional events identified during the audit process in their safety analysis. There were no unreported SAEs noted from RECORD 1. Postoperative randomization did not occur to any significant degree in RECORD 1. In summary, despite some identified deficits in study conduct, the deficiencies do not appear pervasive enough to cast doubt on the overall reliability of RECORD 1 study data. DSI Assessment of RECORD 2 Reliability: DSI recommends that the data from this study can be used in support of the NDA. Although there were drug accountability issues identified at (b) (4) 29% of audited sites in RECORD 2, monitoring was assessed as adequate in the majority of subjects and sites and the number of audited sites is relatively small, and earlier FDA inspections did not reveal drug accountability issues. There were 4 clinical investigator sites in RECORD 2 (Corces, Yang, Naraffete, and Ono) for which DSI cannot assure data reliability (due to drug accountability issues and/or issues with source documentation). DSI acknowledges that there were unreported adverse events from this trial, and DSI suggests that the review division consider additional events identified during the audit process in their safety analysis. There were no unreported SAEs noted from RECORD 2. Postoperative randomization did not occur to any significant degree in RECORD 2. In summary, despite some identified deficits in study conduct, the deficiencies do not appear pervasive enough to cast doubt on the overall reliability of RECORD 2 data. DSI Assessment of RECORD 3 Reliability: DSI recommends that the data from this study can be used in support of the NDA. Although there were drug accountability issues identified at (b) (4) 33% of audited sites in RECORD 3, a very small number of RECORD 3 sites were (b) (4) audited by making the statistical assessment of this finding problematic. Monitoring (b) (4) was assessed as adequate in 42 of 70 (60%) of subjects and 2 of 3 sites audited by Based on Falcon’s monitoring audit strategy of focusing on a PDC (Patient Data Check) form for evaluation of monitoring adequacy, it appears that up to 40% of subjects had inadequacies in monitoring. However, note that DSI’s assessment of adequacy of monitoring and data reliability did not solely focus on the PDC form, but rather on the specific types of issues that were missed by monitoring and their impact on assessment of key safety and efficacy parameters. There was 1 site in RECORD 3 (Brabants) for which DSI cannot assure data (b) (4) reliability (due to drug accountability/storage condition issues identified during audit). DSI acknowledges that there were unreported adverse events from this trial, and DSI suggests that the review division consider additional events identified during the audit process in their safety analysis. There were no unreported SAEs noted from RECORD 3. Postoperative randomization did not occur to any significant degree in RECORD 3. In summary, despite some identified deficits in study conduct, the deficiencies do not appear pervasive enough to cast doubt on the overall reliability of RECORD 3 data. Reference ID: 2951276 Page 42 Compliance Review of NDA 22-406, Xarelto (b) (4) (b) (4) DSI Assessment of RECORD 4 Reliability: FDA inspections, the audits, and the (b) (4) data verification process have identified serious issues with the study conduct and monitoring of the RECORD 4 study. Postoperative randomization in violation of the protocol occurred at 1227 of 3148 (39%) of RECORD 4 subjects, despite a memo from the CRO (b) (4) monitoring the study ( that postoperative randomization was not acceptable. Although this occurred equally in both study arms, the possibility exists that because of postoperative randomization, the labeled population would not be reflective of the actual study population. (b) (4) The number of unreported adverse events detected by monitors (265) was more than twice the number from any of the other RECORD trials (110, 131, and 37 for RECORD 1, 2, (b) (4) and 3, respectively), and there were 504 unreported adverse events detected during the data verification; the review division may wish to review these adverse events for safety analysis inclusion. All newly reported serious adverse events were from RECORD 4 sites: 8 (b) (4) (b) (4) from the audits and 28 from the data verification. In addition, there were (b) (4) serious drug accountability issues at 3 of 9 (33%) of audited RECORD 4 sites, in addition to 2 sites with serious drug accountability issues identified earlier by DSI (Corces and (b) (4) Esquivel). The audit finding that 197 of 312 (63%) of subjects and 6 of 9 (67%) of (b) (4) sites in RECORD 4 were monitored inadequately by is striking, and higher than the other RECORD studies. (b) (4) Eight of 16 (50%) sites of the RECORD 4 sites audited by or inspected by FDA ended with an evaluation that the data from the sites was not reliable, reflective of drug accountability deficiencies and other violations of good clinical practice, including postoperative randomization, falsification, missing records, and improper study drug storage. DSI does not (b) (4) feel that the data verification process conducted by has been validated, nor does it (b) (4) negate the findings described above. It is important to note that these sites audited by represent only 7% of total sites and 10% of total subjects in the RECORD 4 study. The additional audits were conducted with the expectation that failure to identify additional sites with serious deficiencies would provide assurance that the remaining unaudited sites provided reliable data. The pervasive nature of study conduct deficiencies, including particular inadequate monitoring, raises the possibility that there may be deficiencies affecting the primary efficacy outcome which were not detected, e.g. venography conduct. Based on serious drug accountability issues, a relatively large number of unreported adverse events and serious adverse events, a high rate of postoperative randomization in violation of the protocol, and inadequate monitoring of a majority of the RECORD 4 sites as well as the relatively small proportion of sites audited, DSI recommends that the data from RECORD 4 be considered to be unreliable. While the Applicant attempted to provide further assurance that data from this (b) (4) (b) (4) study was reliable via the data verification process, findings do not negate the (b) (4) findings described above. Recall that the audit proposed by J&J was intended to be a specific methodology for analysis of the audited data, not the performance of 3rd party audits, per se, and that FDA did not agree or review as to the usage of this methodology for this intended purpose. VI. OVERALL ASSESSMENT OF FINDINGS AND RECOMMENDATIONS Executive Summary Conclusion DSI finds that Johnson and Johnson’s response to the FDA’s May 27, 2009 Complete Reference ID: 2951276 Page 43 Compliance Review of NDA 22-406, Xarelto (b) (4) Response Letter addresses all of the DSI items requested in the CR Letter. However DSI’s review concludes that the data generated by the RECORD 4 study is unreliable, and recommends that the data not be used in support of the respective indication of prophylaxis of deep venous thrombosis and pulmonary embolism after total knee arthroplasty. Given serious drug accountability issues, a relatively large number of unreported adverse events and serious adverse events, a high rate of postoperative randomization in violation of the protocol, and inadequate monitoring of a majority of the RECORD 4 sites as well as the fact that only a subset of sites have been audited, DSI cannot provide a favorable assessment of RECORD 4 data reliability for the remaining 88% of uninspected/unaudited clinical investigator sites based (b) (4) on extrapolation of the audit findings. Although issues exist with the study conduct of RECORD 1, 2, and 3, they are not sufficiently pervasive to reflect negatively on overall study data integrity, and the data from these 3 studies are considered to be reliable, with the exception of a few sites. Summary Assessment and Recommendation On May 27, 2009 FDA issued an NDA Complete Response letter to Johnson & Johnson for the Xarelto NDA 22-406 for the indication of prophylaxis of deep vein thrombosis and pulmonary embolism in patients undergoing hip or knee replacement surgery. Prior to submission of this NDA, FDA inspections based on complaints received resulted in one Warning Letter and one NIDPOE, as well as an investigator being discontinued by the sponsor due to failure to maintain clinical trial records. Inspections conducted in support of the NDA resulted in four OAI classifications, five VAI, and two NAI. Evaluation of the inspections revealed serious deficiencies in adverse event reporting, drug accountability and administration, and adherence to the protocol especially postoperative randomization. Also of serious concern were deficiencies in monitoring noted at the inspected CI and sponsor sites noted in all four RECORD studies, but particularly pervasive in RECORD 4. The CR Letter requested, in part, evidence that the four RECORD studies are reliable, and proposed that independent third party audits be conducted at additional CI sites to provide reassurance of the reliability of the RECORD 1, 2, 3, and 4 study data. (b) (4) Johnson & Johnson submitted a CR on December 23, 2010. was selected to conduct the third party independent audits. There were 30 clinical sites audited across all four RECORD studies: all 18 high enrolling sites (previously uninspected) with > 60 randomized subjects and 12 moderately enrolling site with 15-59 randomized subjects randomly selected. All subjects at sites were audited if there were less than 35 subjects; otherwise, a random sample sufficient to provide 95% confidence to rule out a 5% error rate was chosen. Audits of these 30 sites resulted in audit of 950 subjects out of 12,729 total, which constituted 7.5% of all subjects in the 4 RECORD studies. The parameters examined during the audits were adequacy of monitoring, adverse event reporting, adherence to protocol including postoperative randomization, informed consent, investigational product, and source data verification and CRF completion. Also submitted with the CR were the reports of the Bayer audits. (b) (4) Adequacy of clinical trial monitoring was assessed in several ways. auditors stated that overall site study monitoring was deficient at 1 of 11 (9%) of RECORD 1 sites, 2 of 7 (27%) RECORD 2 sites, 1 of 3 (33%) RECORD 3 sites, and 5 of 9 (56%) RECORD 4 sites; key efficacy and safety findings were missed during monitoring of these sites. Assessment of Reference ID: 2951276 Page 44 Compliance Review of NDA 22-406, Xarelto (b) (4) monitoring by individual subjects resulted in the following assessment of inadequate monitoring: RECORD 1 96/347 (27.2%) subjects; RECORD 2 55/216 (25.5%) subjects, RECORD 3 28/70 (40.0%) subjects, and RECORD 4 197/312 (63.1%) subjects. Lastly, Johnson & Johnson submitted the results of 74 clinical investigator site audits conducted by (b) (4) Bayer; 69 were routine. Significant findings noted during the audits at the sites of Drs. Lenart (RECORD 1), Porvaneckas (RECORD 1), Nararrete (RECORD 2), and Buettner (RECORD 4) were not mentioned in the Bayer audit reports. FDA inspectional findings at the sites of Dr. Michael Murray (RECORD 4) were not described in the Bayer audit, nor did the Bayer audits detect the most serious deficiency which resulted in disqualification of Dr. Craig Loucks (RECORD 4). Inspection of Bayer as the sponsor of the NDA revealed some monitoring deficiencies as well, in that the major issues at the sites of Drs. Corces (RECORD 2) and Murray (RECORD 4) were not identified by Bayer monitoring. Monitoring for the RECORD 1, 2, and 3 studies was performed by Bayer, while the monitoring for RECORD 4 (b) (4) was conducted by the CRO Although issues with clinical trial monitoring inadequacies were present in all four RECORD trials, the deficiencies were most frequent in the RECORD 4 study. Deficiencies in clinical trial monitoring raise serious concern regarding the validity of data submitted in RECORD 4. In particular, the widespread monitoring deficiencies do not provide reassurance that study conduct deficiencies are not present at the approximately 90% of RECORD 4 sites which were not inspected by FDA or audited. (b) (4) Based on DSI’s assessment of audit reports, drug accountability deficiencies were present at 3/11 (27%) of RECORD 1 sites, 2/7 (29%) of RECORD 2 sites, 1/3 (33%) of RECORD 3 sites, and 3/9 (33%) of RECORD 4 sites. Site drug accountability was considered deficient if source documentation for key efficacy assessments was absent and/or there were significant issues with documentation of drug accountability such that it does not appear possible to verify that subjects at the site received study drug. Further information from Johnson & Johnson was requested that might provide assurance of drug administration at the problematic sites, such as pharmacy or nursing records. For the sites assessed as deficient here, no such documentation was located. Note that a very small number of RECORD 3 sites (b) (4) were audited by making the statistical assessment for this study problematic. We acknowledge the finding that 27-33% of RECORD 1-3 sites had deficiencies in drug accountability; however these findings were not replicated in FDA inspectional findings. In contrast with RECORD 4, however, audits of the RECORD 1, 2, and 3 studies did not demonstrate systematic deficiencies in multiple aspects of clinical trial conduct, such that data integrity from all study sites must be questioned. However, the findings that 33% of RECORD (b) (4) 4 sites audited by (as well as 2 additional sites, Corces and Esquivel, identified earlier by DSI) had serious drug accountability deficiencies, 67% had inadequate monitoring, and 50% of sites audited or inspected were determined to provide unreliable data, together indicate that the data from RECORD 4 cannot be considered reliable. (b) (4) Failure to report adverse events was identified at all but 2 sites audited by There were 110 unreported AEs in RECORD 1, 131 unreported AEs in RECORD 2, 37 unreported AEs in RECORD 3, and 265 unreported AEs in RECORD 4. There were 8 unreported SAEs noted in (b) (4) the audits, all in RECORD 4. When the unreported AEs were individually examined for significance as defined by the necessity for expeditious medical evaluation, or were AEs involving bleeding or hepatic events, there were 16 in RECORD 1, 24 in RECORD 2, and 265 Reference ID: 2951276 Page 45 Compliance Review of NDA 22-406, Xarelto (b) (4) in RECORD 4; RECORD 3 could not be tabulated due to failure to list individual laboratory (b) (4) abnormalities. During the data verification process of RECORD 4, 504 unreported (b) (4) AEs were noted, as were 28 previously unreported SAEs. The audits identified more than twice as many AEs in RECORD 4 than in the other RECORD studies, and all of the unreported AEs were from RECORD 4. The high number of unreported AEs and SAEs from RECORD 4 may impact labeling for safety, and is again reflective of inadequate monitoring of RECORD 4. Failure to adhere to the protocol, in particular postoperative randomization, occurred in 39% of RECORD 4 subjects. Although postoperative randomization would not be expected to affect the primary efficacy outcome since it occurred in both study arms, the concern remains that the population described in the product label may not be reflective of the actual study population if subjects are screened and enrolled by criteria other than those in the protocol. There was no other evidence of widespread failure to adhere to the inclusion criteria, and there was no significant postoperative randomization in RECORD 1, 2, or 3. Again, the failure of the CRO (b) (4) to enforce compliance with the protocol requirement for preoperative randomization is reflective of inadequate monitoring of RECORD 4. (b) (4) The audits of the RECORD 4 study sites were conducted in an attempt to provide assurance of the validity of the data from RECORD 4. There was no difference in primary efficacy or safety outcome when sensitivity analyses were conducted on high versus low quality sites or investigators or on subjects randomized preoperatively versus postoperatively. (b) (4) Although interesting, the methodology is not validated, nor does it address the effects of inadequate monitoring of RECORD 4, which may have introduced unidentified errors not accounted for in the data verification. In summary given the pervasive findings of deficient clinical trial monitoring, high number of clinical investigator sites with data assessed as unreliable, failure to follow the protocol including postoperative randomization, and deficient clinical trial conduct including failure to report significant adverse events and SAEs, DSI cannot provide a favorable assessment of RECORD 4 data reliability for the remaining unaudited sites based on extrapolation of the (b) (4) audit findings. Although some issues exist with the study conduct of RECORD 1, 2, and 3, they are not sufficiently pervasive to recommend an unfavorable assessment of data reliability. Therefore, the data from RECORD 1, 2, and 3, with exception of select sites as identified earlier, are considered reliable in support of the application. The data from RECORD 4 are not considered reliable in support of the respective indication. Reference ID: 2951276 Page 46 Compliance Review of NDA 22-406, Xarelto (b) (4) {See appended electronic signature page} Susan D. Thompson, M.D. Medical Officer Good Clinical Practice Branch II Division of Scientific Investigations CONCURRENCE: {See appended electronic signature page} Tejashri Purohit-Sheth, M.D. Branch Chief Good Clinical Practice Branch II Division of Scientific Investigations Reference ID: 2951276