Rivaroxaban (XARELTO )

Transcription

1 Rivaroxaban (XARELTO ) Cardiovascular and Renal Drugs Advisory Committee January 16, 2014 CC-1

2 Introduction Paul Burton, MD, PhD, FACC Vice President, Janssen Research & Development, L.L.C. CC-2

3 Rivaroxaban (XARELTO ) is Approved for 6 Indications in the US Over 70,000 patients participated in clinical trials leading to the approval of Xarelto for: Reduction of Risk of Stroke and Systemic Embolism in Nonvalvular Atrial Fibrillation Treatment of Deep Vein Thrombosis Treatment of Pulmonary Embolism Reduction in the Risk of Recurrence of Deep Vein Thrombosis and of Pulmonary Embolism Prophylaxis of Deep Vein Thrombosis Following Hip or Knee Replacement Surgery Global use of XARELTO at the currently approved doses exceeds 1.4 million patient years since 2008 CC-3

4 Regulatory History of the Rivaroxaban ACS snda in the United States June 2009 ATLAS ACS program granted Fast Track status December 2011 snda submitted, and application granted Priority Review May 2012 Advisory Committee meeting June 2012 Complete Response Letter (CRL) issued September 2012 Janssen submits response to CRL 1 March 2013 CRL 2 issued May 2013 Janssen submits an application for Dispute Resolution August 2013 Janssen submits response to CRL 2 CC-4

5 Proposal Under Consideration Today Shortened Duration of Therapy in ACS Patients Rivaroxaban 2.5 mg BID is indicated to: Reduce the risk of thrombotic cardiovascular events in patients with ACS (STEMI, NSTEMI or UA) in combination with ASA alone or with ASA plus a thienopyridine (clopidogrel or ticlopidine). Rivaroxaban has been shown to reduce the risk of a combined endpoint of CV death, MI or stroke. The difference between treatments was driven by CV death and MI. CC-5

6 Key Recommendations from the May 2012 Advisory Committee Meeting Follow up and retrieve missing vital status data to assess if differential informative censoring had biased the study results Conduct sensitivity analyses to assess the robustness of the study results Conduct additional analyses to assess the balance of benefit and risk CC-6

7 What Has Been Done Since the May 2012 Advisory Committee Meeting? Vital status retrieved for an additional 930 subjects The sponsor performed the analyses recommended by the advisory committee as well as those raised in CRL1 and CRL2 These analyses demonstrate the consistency of the original study finding and the robustness of the mortality results CC-7

8 Outcome of the Dispute Resolution Process The Office of Drug Evaluation stated that 4 out of 5 topics identified in the Complete Response had been addressed and did not represent a barrier to approval Impact of missing data A lack of supporting data with other anticoagulants A lack of internal consistency between the two rivaroxaban doses Inability to confirm concomitant thienopyridine therapy during the study Strength of evidence could represent a potential barrier to approval Identified a pathway forward based on a shortened duration of therapy a limitation in rivaroxaban s use for one month following ACS would provide a better benefit-risk relation, because the efficacy is more evident in the early period after administration, whereas the overall risk of bleeding would be smaller. If you would like to pursue this path, the Division will be glad to work with you. CC-8

9 Rationale Supporting a 90 Day Duration of Rivaroxaban Therapy in ACS Patients (1 of 2) Risk of recurrent thrombotic events is high during the first 90 days 49% of cardiovascular death, MI, and stroke events occurred within the first 90 days Rivaroxaban (2.5 mg BID) substantially reduced Primary endpoint: CV death, MI or stroke; 0.78 (0.65, 0.93) CV death: 0.65 (0.44, 0.97) Few TIMI life threatening bleeding events in the first 90 days CC-9

10 Rationale Supporting a 90 Day Duration of Rivaroxaban Therapy in ACS Patients (2 of 2) 98% of subjects had known vital status 94% of stratum 2 subjects were receiving thienopyridine Evidence of effectiveness is consistent across Stratum 1 and 2 Subgroups Both doses Components of the composite endpoint Net clinical benefit (mitt, Stratum 2, 2.5 mg BID) In a hypothetical population of 10,000 patients treated, 81 fewer cases of ischemic CVD/MI/stroke events and no excess fatal bleeding or ICH events would be predicted CC-10

11 EU Approved Rivaroxaban for ACS Based on a Refined Patient Population Rivaroxaban 2.5 mg BID indicated for cardiac biomarker positive ACS in patients with no prior stroke/tia mitt/all Strata/2.5 mg Sub group Primary endpoint HR (95% CI) CV death HR (95% CI) Overall study population 0.84 (0.72, 0.97) 0.66 (0.51, 0.86) Exclude prior stroke/tia 0.81 (0.69, 0.94) 0.63 (0.48, 0.82) Biomarker positive and exclude prior stroke/tia 0.80 (0.68, 0.94) 0.55 (0.41, 0.74) CC-11

12 Sponsor Presentation Introduction Paul Burton, MD, PhD Vice President, Clinical Development, Janssen ATLAS ACS Program Overview and Results C. Michael Gibson, MS, MD Principal Investigator ATLAS ACS 2 TIMI 51 Trial Professor of Medicine, Harvard Medical School Roderick A Little, PhD Richard D. Remington Distinguished University Professor of Biostatistics, University of Michigan Evidence of Effectiveness Jay P. Siegel, MD Head of Scientific Strategy and Policy, Johnson & Johnson Balancing Benefit and Risk Marvin A Konstam, MD Chief Physician Executive, The CardioVascular Center at Tufts Medical Center Professor of Medicine at Tufts University School of Medicine CC-12

13 Additional Experts Available for Questions Keith A AFox, BSc, MBChB, FRCP, FMed Sci Professor of Cardiology Edinburgh Christopher Hammett, MD Interventional Cardiologist, Royal Brisbane Hospital, Australia Janet Wittes, PhD President, Statistics Collaborative Washington DC CC-13

14 ATLAS ACS Program: Rationale C. Michael Gibson, MS, MD Principal Investigator ATLAS ACS 2 TIMI 51 Trial Professor of Medicine, Harvard Medical School CC-14

15 ACS Represents a Large Unmet Medical Need Each year, an estimated Americans have a new coronary attack and have a recurrent attack Approximately every 34 seconds, 1 American has a coronary event, and approximately every 1 minute, an American will die of one. Executive Summary: Heart Disease and Stroke Statistics 2013 Update: A Report From the American Heart Association CC-15

16 Despite New Therapies ACS Continues to Represent an Unmet Medical Need TRILOGY Study: Prasugrel vs Clopidogrel in medically managed patients 20 Clopidogrel Prasugrel CV death/mi/stroke 20.3% 18.7% Endpoint (%) The risk of CV death, MI or stroke is very high early after ACS Days Adapted from Roe MT et al., N Engl J Med 2012: 367(14): Within 3 years of an ACS event 1/5 patients will have died, had another MI or a stroke TIMI Major Bleeding % 1.8% CC-16

17 Persistent Elevation of Thrombin Generation in Post ACS Patients 1.4 Admission 6 Months 1.2 F 1.2 nmol/l Controls Stable Angina Unstable Angina MI Slide by C. Michael Gibson, M.S., M.D. Merlini et al. Circ 1994;90:61 68 CC-17

18 Anticoagulation is a Therapeutic Strategy for ACS Patients with a Proven Track Record Based on the WARIS (n=1,214) and WARIS II (n=3,630) studies warfarin is indicated for chronic use in ACS patients INDICATIONS AND USAGE COUMADIN is a vitamin K antagonist indicated for: Prophylaxis and treatment of venous thrombosis and its extension, pulmonary embolism (1) Prophylaxis and treatment of thromboembolic complications associated with atrial fibrillation and/or cardiac valve replacement Reduction in the risk of death, recurrent myocardial infarction, and thromboembolic events such as stroke or systemic embolization after myocardial infarction (1) CC-18

19 Anticoagulation is a Therapeutic Strategy for ACS Patients with a Proven Track Record Anticoagulation (e.g. enoxaparin) is part of the foundation of the acute management of ACS patients INDICATIONS AND USAGE Lovenox is a low molecular weight heparin [LMWH] for: Prophylaxis of ischemic complications of unstable angina and non Q wave myocardial infarction [MI] (1.3) Treatment of acute ST segment elevation myocardial infarction [STEMI] managed medically or with subsequent percutaneous coronary intervention [PCI] (1.4) CC-19

20 RECORD: Rivaroxaban Significantly Reduces Thrombosis Risk Compared to Enoxaparin 30 Enoxaparin Rivaroxaban Event Rate (%) DVT/PE/Death Note: mitt analysis RRR = 70% 3.7% 40 mg QD 1.1% 10 mg QD RECORD 1 RRR = 79% 9.3% 40 mg QD 2.0% 10 mg QD RRR = 49% 18.9% 40 mg QD 9.6% 10 mg QD RECORD 2 RECORD 3 CC-20

21 Dose Ranging and Finding Studies with Anticoagulants in ACS 8% U Shaped Dose Response Curve: APTT After Thrombolytic Therapy for Acute MI Probability of Death 6% 4% 2% Therapeutic Window 0% APTT (seconds) at 12 Hours Slide by C. Michael Gibson, M.S., M.D. Granger, CB et al. Circulation 1996 vol. 93, no. 5, p CC-21

22 Fondaparinux: Lowest Dose Associated with Best Efficacy In ACS Patients Phase 2 PENTUA Study Death/MI/Recurrent Ischemia n=1138 Phase 3 OASIS 5 Study Death n=20,078 Enoxaparin Incidence 40% 30% 20% 10% 0% Enoxaparin Fondaparinux Days 4.0 mg 8.0 mg 12.0 mg 2.5 mg Cumulative Hazard Fondaparinux 2.5 mg HR (95% CI) 0.89 (0.80, 1.00) p= Days Simoons et al. J Am Coll Cardiol. 2004; The Fifth Organization to Assess Strategies in Acute Ischemic Syndromes Investigators N Engl J Med 2006; 354: CC-22

23 U Shaped Dose Response Curve: Warfarin Dose in Atrial Fibrillation 20 Therapeutic Window 15 Odds Ratio 10 Intracranial Bleeding 5 Ischemic Stroke INR ICH is the most lethal form of stroke. 30-day mortality rates with ICH estimated at 30% to 55% 1,2 Abbreviation: INR = International Normalized Ratio. Adapted from Fuster V et al. J Am Coll Cardiol. 2011;57(11):e101-e198. Modified with permission from Hylek EM, Singer DE. Ann Intern Med. 1994;120: Data from Odén A, Fahlén M, Hart RG. Thromb Res. 2006;117: Freeman WD, Aguilar MI. Expert Rev Neurother. 2008;8(2): Aguilar MI et al. Mayo Clin Proc. 2007;82(1): Slide by C. Michael Gibson, M.S., M.D. CC-23

24 Lower Dose Edoxaban has Greater Mortality Benefit in AF Patients Edoxaban 30 mg od %/year (N=7034) Edoxaban 60 mg od %/year (N=7035) Warfarin %/year (N=7036) Myocardial infarction Major bleeding Fatal Edoxaban 30 mg od vs warfarin HR, p value* 1.19 ( ) 0.47 ( ) 0.35 ( ) Edoxaban 60 mg od vs warfarin HR, p value* 0.94 ( ) 0.80 ( ) 0.55 ( ) Mortality ( ) p= ( ) p=0.08 Edoxaban 30 mg reduces bleeding events and is associated with a lower mortality than 60 mg Offset in mortality not completely explained by fatal bleeding deaths C. Michael Gibson, M.S., M.D.; Giugliano et al., 2013 CC-24

25 Lower Edoxaban Doses Reduce Mortality 60 mg Edoxaban (%/yr) 30 mg Edoxaban (%/yr) Edoxaban 60 mg reduces the primary efficacy endpoint more than Edoxaban 30 mg 60 mg 30 mg Edoxaban (%/yr) Efficacy Minor Bleeding Major Bleeding Fatal Bleeding Mortality However, Edoxaban 30 mg reduces bleeding events and is associated with a lower mortality than 60 mg Offset in mortality not completely explained by fatal bleeding deaths C. Michael Gibson, M.S., M.D.; Giugliano et al., 2013 CC-25

26 ATLAS ACS Program: Efficacy Results CC-26

27 ATLAS Phase 2 Study Enrolled 3491 subjects Collected >200 primary endpoint events (death, myocardial infarction, stroke, or severe recurrent ischemia requiring revascularization) Suggested that rivaroxaban may reduce the risk of important clinical events in ACS patients Lowest rivaroxaban twice daily doses seemed to have the promise of efficacy with the best bleeding profile Paved the way for the ATLAS ACS 2 TIMI 51 Phase 3 study CC-27

28 ATLAS Phase 2: Higher Rivaroxaban Total Daily Doses Increased the Rate of Bleeding Events Cumulative Event Rate (%) Clinical Significant Bleeding All Strata P < Rivaroxaban 20 mg Rivaroxaban 15 mg Rivaroxaban 10 mg Rivaroxaban 5 mg Placebo Days After Start of Treatment CC-28

29 ATLAS Phase 2: Higher Rivaroxaban Doses Did Not Result in Improved Efficacy Stratum 2: ASA + Thienopyridine 6 Month Treatment 6 Event Rate (%) Death, MI, Stroke % 2.6% 3.0% 5.9% 3.1% Placebo Total Dose (mg) CC-29

30 ATLAS ACS TIMI 46 Phase 2: Rivaroxaban Showed Promise In Reducing Important Clinical Events HR (95% CI) 6 All Doses 0.69 (0.50, 0.96) p=0.03 Placebo 5.5% Death, MI, or Stroke (%) N =3, % Rivaroxaban Days after Randomization CC-30

31 ATLAS Phase 2: Low Dose (2.5 mg and 5 mg) BID Rivaroxaban Doses Selected for Phase 3 5 Stratum 2: ASA + thienopyridine 6 month treatment BID dosing Cumulative Event Rate (%) Gibson CM, AHA Placebo 3.8% Riva 2.0% Riva 1.2% Placebo 0.2% Days Death, MI, Stroke HR=0.55 ( ) p=0.09 TIMI Major Bleed p=0.03 CC-31

32 ATLAS ACS 2 TIMI 51: Phase 3 Trial The ATLAS ACS 2 TIMI 51 Trial (The second trial of Anti-Xa Therapy to Lower cardiovascular events in Addition to standard therapy in Subjects with Acute Coronary Syndrome, Thrombolysis In Myocardial Infarction 51 Trial) Randomized, Double Blind, Placebo Controlled, Event Driven, Multicenter Phase III 766 Study Centers 44 Countries CC-32

33 ATLAS ACS 2 TIMI 51 Primary Efficacy and Safety Objectives To determine if rivaroxaban when added to a foundation of standard antiplatelet therapy: Is effective at reducing the risk of the primary composite of cardiovascular (CV) death, MI, or stroke compared with placebo in subjects with a recent ACS If rivaroxaban is effective, what is the preferred dose? Assess the occurrence of TIMI major bleeding events not associated with CABG surgery (primary safety endpoint) Assess the overall safety profile of rivaroxaban in subjects with a recent ACS CC-33

34 ATLAS ACS 2 TIMI 51 Study Design N=up to Recent ACS Patients Stabilized 1 7 Days Post Index Event MD Decision to Treat with a Thienopyridine NO STRATUM 1 Stratum 1: ASA Only Stratum 2: ASA + Thienopyridine YES STRATUM 2 Placebo RIVA 2.5mg BID RIVA 5mg BID Placebo RIVA 2.5mg BID RIVA 5mg BID Target 983 Primary Efficacy Endpoint Events Primary Efficacy Endpoint: CV Death, MI or All Cause Stroke CC-34

35 Principal Efficacy Analyses mitt: 1,002 events All randomized subjects All data included up to the earlier of global treatment end date, and 30 days after early discontinuation of study drug ITT: 1,101 events All randomized subjects All data included up to global treatment end date mitt and ITT analyses exclude 184 subjects from three sites (091001, and ) due to potential trial misconduct. CC-35

36 The Primary Analysis Population for ATLAS is The mitt Population 09/12/2008 CC-36

37 The Pre Planned Statistical Testing Strategy on the mitt Population was Agreed with the FDA Endpoint Test 1: Primary Efficacy Endpoint Rivaroxaban Group vs Placebo All Strata All Riva vs Placebo Test 2: Primary Efficacy Endpoint Stratum 2 All Riva vs Placebo Test 3: Primary Efficacy Endpoint Stratum mg Riva Stratum 2 5 mg Riva CC-37

38 Test 1, in The mitt Population is the Correct Test to Determine if ATLAS Succeeded or Failed Endpoint Test 1: Primary Efficacy Endpoint Rivaroxaban Group vs Placebo All Strata All Riva vs Placebo CC-38

39 TEST 1: Combined Doses vs Placebo mitt/all Strata/Primary Efficacy Endpoint 15 HR (95% CI) mitt ITT Combined Doses 0.84 (0.74, 0.96) p=0.008 p=0.002 Cumulative Event Rate (%) 10 5 Placebo Rivaroxaban 0 2 Year Cumulative Event Rate: 8.9% vs 10.7% Days from Randomization CC-39

40 Test 1: The Effect of Rivaroxaban is Consistent Across All Analysis Populations All Strata/Combined Doses p value mitt ITT ITT Total Safety TE 30 days Per Protocol mitt Per Investigator Hazard Ratio (95% CI) Favors Rivaroxaban Favors Placebo CC-40

41 TEST 2:Combined Doses vs Placebo mitt/stratum 2/ Primary Efficacy Endpoint Cumulative Event Rate (%) HR (95% CI) mitt ITT Combined Doses 0.86 (0.75, 0.98) p=0.024 p=0.004 Placebo Rivaroxaban 2 Year Cumulative Event Rate: 8.8% vs 10.4% Days from Randomization CC-41

42 TEST 3: Each Dose vs Placebo mitt/stratum 2/Primary Efficacy Endpoint Cumulative Event Rate (%) HR (95% CI) mitt ITT 2.5 mg BID 0.85 (0.72, 0.99) p=0.039 p= mg BID 0.87 (0.74, 1.01) p=0.075 p=0.020 Placebo Riva 2.5 mg BID Riva 5 mg BID 2.5 mg, 2 Year Cumulative Event Rate: 10.3% vs 8.9% 5 mg, 2 Year Cumulative Event Rate: 10.3% vs 8.7% Days from Randomization CC-42

43 Effect of Rivaroxaban on Components of the Primary Efficacy Endpoint mitt/stratum 2/Each Dose Parameter Primary CV Death MI Stroke 2.5 mg BID vs Placebo 5 mg BID vs Placebo HR (95% CI) 0.85 (0.72, 0.99) 0.62 (0.47, 0.82) 0.92 (0.75, 1.12) 1.31 (0.84, 2.05) mitt p value ITT p value <0.001 < HR (95% CI) 0.87 (0.74, 1.01) 0.95 (0.74, 1.21) 0.83 (0.68, 1.02) 1.39 (0.89, 2.16) mitt p value ITT p value Tes ng the primary endpoint for each dose in the mitt population was pre defined in the statistical analysis plan and adjusted by multiplicity. All other p values are nominal and not adjusted for multiplicity. CC-43

44 Rivaroxaban Substantially Reduces the Risk of Cardiovascular Death mitt/stratum 2/2.5 mg BID Cumulative Event Rate (%) HR (95% CI) mitt ITT 2.5 mg BID 0.62 (0.47, 0.82) p= p= Placebo Riva 2.5 mg BID Year Cumulative Event Rate: 2.5% vs 4.2% Days from Randomization CC-44

45 Primary Efficacy Endpoint Results are Consistent Across Subgroups mitt/stratum 2/2.5 mg BID Overall Age [years] (<55 75) Sex (Male; Female) Race (White, Other) Weight [kg] (<60; 90 kg) CrCl [ml/min] (<30; >80) Index Event (STEMI; NSTEMI; Unstable angina) Prior MI (yes, no) PCI for Index Event (yes, no) Elevated Cardiac Biomarker (yes, no) Congestive Heart Failure (yes, no) Prior Ischemic Stroke/TIA (yes, no) Hypertension (yes, no) Diabetes (yes, no) Hazard Ratio and 95%CI Favors Rivaroxaban Favors Placebo Interaction p value: p=0.023 p=0.082 CC-45

46 Stent Thrombosis CC-46

47 Rivaroxaban Plus Aspirin and Clopidogrel Reduces in Stent Thrombosis ex vivo 12 Thrombus Mass (mg) * p<0.01 vs control # p<0.01 vs ASA + clopidogrel N=7 N=6 N=7 Adapted from Becker et al., J. Thrombosis and Haemostasis, 2012, 10: * # Rivaroxaban + ASA + Clopidogrel CC-47

48 Assessment of Stent Thrombosis in ATLAS Cardiovascular events were assessed for stent thrombosis using the Academic Research Consortium (ARC) criteria At the request of the FDA, a sample of angiographic films for cases of definite or probable stent thrombosis underwent Blinded angiographic core lab assessment at the Cardiac Research Foundation (Dr. G. W. Stone) Complete, independent clinical re adjudication at Duke Clinical Research Institute (Dr. K. Mahaffey) CC-48

49 Stent Thrombosis: Definite or Probable ITT Total/Stratum 2/Each Dose 3.0 HR (95% CI) 2.5 mg BID 0.61 (0.39, 0.93) p=0.022 Cumulative Event Rate (%) mg BID 0.74 (0.49, 1.12) p=0.15 Placebo Riva 5 mg BID Riva 2.5 mg BID Relative Days From Randomization CC-49

50 Stratum 1 Results CC-50

51 Ideal World: All ACS Patients Get a Thienopyridine Real World: They Don t 72,352 patients 251 Get With The Guidelines US Centers 26.8% of patients go home without clopidogrel after an ACS event Clopidogrel at Discharge (%) Overall n=72,352 Somma KA et al., Circ Cardiovasc Qual Outcomes % 85.6 % 67.0% STEMI n=23,386 NSTEMI n=48,966 CC-51

52 Each Dose vs Placebo: Primary Endpoint mitt/stratum 1/Each Dose Cumulative Event Rate (%) HR (95% CI) 2.5 mg BID 0.74 (0.45, 1.22) p= mg BID 0.64 (0.38, 1.07) p=0.089 Placebo Riva 5 mg BID Riva 2.5 mg BID Days from Randomization CC-52

53 Multiple Studies Within Each Study: Consistent Results Across Phase 2 and 3 Studies Stratum 1: Phase 2 Phase 3 Phase Stratum 2: Phase 2 Phase 3 Phase Combined Strata: Phase mitt Population Phase 2: 2.5 and 5 mg BID Hazard Rate (95% CI) Nominal p value not adjusted for multiplicity Favors Rivaroxaban Favors Placebo p value CC-53

54 Summary: Efficacy The ATLAS study demonstrated that Rivaroxaban 2.5 mg BID when added to antiplatelet therapy reduced the rate of the primary endpoint (CV Death/MI/ Stroke) Result driven by a 38% reduction in CV death Consistent benefit in both Stratum 1 (ASA alone) and in Stratum 2 (ASA plus a thienopyridine) Benefit consistent across subgroups 37% reduction in stent thrombosis CC-54

55 ATLAS ACS Program: Safety Results CC-55

56 Non CABG TIMI Major Bleeding: Primary Safety Endpoint TE/Stratum 2/Each Dose Cumulative Event Rate (%) HR (95% CI) 2.5 mg BID 3.35 (2.01, 5.60) p< mg BID 4.26 (2.58, 7.03) p<0.001 Riva 5 mg BID Riva 2.5 mg BID Placebo Relative Days From the First Dose CC-56

57 TIMI Life Threatening Bleeding Events TE/Stratum 2/2.5 mg Riva 2.5 mg BID N=4772 n (%) Placebo N=4773 n (%) TIMI life threatening bleeding 40 (0.8) 18 (0.4) Fatal 5 (0.1) 8 (0.2) Symptomatic intracranial hemorrhage 13 (0.3) 5 (0.1) Hypotension requiring inotropic agents 3 (0.1) 2 (<0.1) Surgical intervention for ongoing bleeding 7 (0.1) 8 (0.2) Blood transfusion of 4 units over 48 h 19 (0.4) 5 (0.1) CC-57

58 Adverse Event Profile of Rivaroxaban TE/Stratum 2/2.5 mg Riva 2.5 mg BID N=4772 n(%) Placebo N=4773 n(%) Adverse events 2740 (57.4) 2650 (55.5) Serious adverse events 1078 (22.6) 1065 (22.3) TE AE resulting in death 89 (1.9) 143 (3.0) TE bleeding AE 948 (19.9) 616 (12.9) Bleeding resulting in study drug D/C 179 (3.8) 91 (1.9) TE bleeding resulting in death 5 (0.1) 8 (0.2) AE resulting in study drug D/C 436 (9.1) 386 (8.1) Post baseline AEs occurring within 30 days prior to study drug D/C 308/1280 (24.1) 292/2697 (23.3) CC-58

59 Summary: Safety 2.5 mg BID Rivaroxaban increased the risk of bleeding in ACS patients The rate of non bleeding adverse events and serious adverse events were similar between the rivaroxaban and placebo groups Bleeding events leading to irreversible harm were low in number ICH and hemorrhagic stroke were more common in the rivaroxaban group compared to the placebo group The rates of fatal ICH and fatal bleeding events were balanced between treatment groups CC-59

60 Evaluation of Shorter Duration of Therapy CC-60

61 Shorter Duration of Rivaroxaban Therapy Makes Good Clinical Sense 30 Days of therapy: A profound reduction in the risk of the primary efficacy endpoint A reduction in the risk of all components of the primary endpoint All of this achieved in the setting of >99% of patients have confirmed vital status, >97% of patients receiving a thienopyridine mitt, 30 Days HR (95% CI) p value TEST 1 (S1+S2, Combined doses) 0.66 (0.52, 0.84) <0.001 TEST 2 (S2, Combined doses) 0.73 (0.57, 0.94) TEST 3 (S2, 2.5mg BID) 0.71 (0.53, 0.97) CV death 0.81 (0.49, 1.31) NS MI 0.71 (0.48, 1.06) NS Stroke 0.67 (0.27, 1.63) NS CC-61

62 Shorter Duration of Rivaroxaban Therapy Makes Good Clinical Sense 90 Days of therapy: Maintains profound benefit seen with 30 days of treatment Stronger CV mortality reduction >98% of patients have confirmed vital status, >94% of patients receiving a thienopyridine mitt, 90 Days HR (95% CI) p value TEST 1 (S1+S2, Combined doses) 0.78 (0.65, 0.93) TEST 2 (S2, Combined doses) 0.82 (0.68, 0.99) TEST 3 (S2, 2.5mg BID) 0.76 (0.61, 0.96) CV death 0.65 (0.44, 0.97) MI 0.79 (0.59, 1.05) NS Stroke 0.94 (0.46, 1.90) NS CC-62

63 Effect of Rivaroxaban Therapy for 90 Days on the Risk of the Primary Endpoint mitt/ Stratum 2 / 2.5 mg / 90 Days Cumulative Event Rate (%) HR (95% CI) 2.5 mg BID 0.76 (0.61, 0.96) p=0.020 Placebo Riva 2.5 mg BID Days from Randomization CC-63

64 Effect of Rivaroxaban Therapy for 90 Days on the Risk of Cardiovascular Death mitt/ Stratum 2 / 2.5 mg / 90 Days Cumulative Event Rate (%) HR (95% CI) 2.5 mg BID 0.65 (0.44, 0.97) p=0.031 Placebo Riva 2.5 mg BID Days from Randomization CC-64

65 Stent Thrombosis: Definite or Probable (Any Stent Placement at Baseline) ITT Total/Stratum 2/Each Dose Cumulative Event Rate (%) HR (95% CI) Overall HR (95% CI) 90 Days 2.5 mg BID 0.61 (0.39, 0.93) 0.69 (0.41,1.14) 5 mg BID 0.74 (0.49, 1.12) 0.64 (0.38,1.07) Placebo Riva 5 mg BID Riva 2.5 mg BID Relative Days From Randomization CC-65

66 Effect Of Shorter Duration Rivaroxaban Therapy in Patients Only Receiving Aspirin mitt/stratum 1/Combined Dose Cumulative Event Rate (%) HR (95% CI) Combined Doses 0.49 (0.28, 0.87) p=0.012 Placebo Rivaroxaban Days from Randomization CC-66

67 Primary Efficacy Endpoints After Last Dose Off Treatment/All Strata/Combined Doses From Last Dose 1 to 2 days 3 to 30 days Rivaroxaban Combined N=9864 n/n (%) Placebo N=4910 n/n (%) CV Death/MI/Stroke 133/9864 (1.3) 63/4910 (1.3) Cardiovascular Death 82/9864 (0.8) 46/4910 (0.9) MI 43/9864 (0.4) 11/4910 (0.2) Stroke 22/9864 (0.2) 10/4910 (0.2) CV Death/MI/Stroke 82/9643 (0.9) 39/4810 (0.8) Cardiovascular Death 57/9693 (0.6) 26/4827 (0.5) MI 32/9658 (0.3) 13/4817 (0.3) Stroke 11/9675 (0.1) 7/4820 (0.1) All safety subjects who had at least 1 day of follow up after last dose of study drug administration. CC-67

68 TIMI Life Threatening Bleeding Over the 1 st 90 Days TE/Stratum 2/2.5 mg Riva 2.5 mg BID N=4772 n (%) Placebo N=4773 n (%) Non CABG TIMI major 21 (0.4) 7 (0.1) TIMI life threatening bleeding 15 (0.3) 10 (0.2) Fatal 2 (<0.1) 5 (0.1) Symptomatic intracranial hemorrhage 3 (0.1) 2 (< 0.1) Hypotension requiring inotropic agents 2 (< 0.1) 2 (<0.1) Surgical intervention for ongoing bleeding 4 (0.1) 5 ( 0.1) Blood transfusion of 4 units over 48 h 6 (0.1) 2 (<0.1) CC-68

69 Rationale Supporting a 90 Day Duration of Rivaroxaban Therapy in ACS Patients (1 of 2) Risk of recurrent thrombotic events is high during the first 90 days 49% of cardiovascular death, MI, and stroke events occurred within the first 90 days Rivaroxaban (2.5 mg BID) substantially reduced Primary endpoint: CV death, MI or stroke; 0.78 (0.65, 0.93) CV death: 0.65 (0.44, 0.97) Few TIMI life threatening bleeding events in the first 90 days CC-69

70 Rationale Supporting a 90 Day Duration of Rivaroxaban Therapy in ACS Patients (2 of 2) 98% of subjects had known vital status 94% of stratum 2 subjects were receiving thienopyridine Evidence of effectiveness is consistent across Stratum 1 and 2 Subgroups Both doses Components of the composite endpoint Net clinical benefit (mitt, Stratum 2, 2.5 mg BID) In a hypothetical population of 10,000 patients treated, 81 fewer cases of ischemic CVD/MI/stroke events and no excess fatal bleeding or ICH events would be predicted CC-70

71 Analyses Addressing Issues Raised In Complete Response 1 And 2 CC-71

72 A Lack of Supporting Data with Other Anticoagulants CC-72

73 In ATLAS ACS 2 TIMI 51 Patients With a Prior Stroke or TIA Were Specifically Excluded TRITON Prasugrel ATLAS2 TIMI 51 Rivaroxaban 5 mg BID 15 month APPRAISE 2 Apixaban 5 mg BID 15 month TIMI data on file History of Stroke Prasugrel Better Rivaroxaban Better Clopidogrel Better Placebo Better Apixaban Better Placebo Better Hazard Ratio P Value (95% CI) Interaction 0.79 (0.71, 0.88) (0.89, 2.13) 0.88 ( ) 1.61 ( ) 0.89 (0.74,1.06) 1.32 (0.88, 1.99) CC-73

74 A Lack of Internal Consistency Between the Two Rivaroxaban Doses CC-74

75 All Cause Mortality in Subjects Without Non CABG TIMI Major Bleeding TE/Stratum 2/5 mg Exclusion of 97 patients with non CABG TIMI major bleeding 3.0 HR = 0.75 (0.56, 0.99) All Cause Mortality (%) % 84 events Rivaroxaban 5 mg BID (n=4631) 2.5% 117 events Placebo (n=4693) CC-75

76 What Types of MI Would Rivaroxaban be Expected to Reduce? Type 1: Spontaneous Type 2: Secondary Type 3: SCD due to Suspected MI Type 4A: Peri PCI Type 4B: Stent Thrombosis Type 5: Peri CABG SCD: sudden cardiac death. CC-76

77 Rivaroxaban Reduces the Risk of Spontaneous MI: Both Doses Have a Similar Effect Incidence of Spontaneous MI (%) Cavender, JACC mg BID HR 0.84 ( ) 6 p= Placebo 5.7% Rivaroxaban 2.5 mg BID 4.7% 5 mg BID HR 0.77 ( ) 6 p=0.01 Placebo 5.7% Rivaroxaban 5 mg BID 4.1% Time (Months) Time (Months) 2 Year KM Estimates HR (95% CI) CC

78 Rivaroxaban Reduces the Risk of Large MIs: Both Doses Have a Similar Effect Incidence of Spontaneous MI (%) > 10 X ULN 2.5 mg BID HR 0.75 ( ) 3 p= Placebo 2.4% Rivaroxaban 2.5 mg BID 1.8% 5 mg BID HR 0.71 ( ) 3 p=0.04 Placebo 2.4% Rivaroxaban 5 mg BID 1.5% Time (Months) Time (Months) 2 Year KM Estimates HR (95% CI) CC-78 Cavender, JACC (2013 Presented by CM Gibson) 2 1

79 Inability to Confirm Concomitant Thienopyridine Therapy During the Study CC-79

80 The Effect of Rivaroxaban is Present in Patients Receiving Active Thienopyridine A sensitivity analysis was performed, censoring patients prior to the initiation of a PPI (omeprazole or esomeprazole) or cessation of their thienopyridine therapy The effect of rivaroxaban is consistent with the overall study result in subjects receiving active thienopyridine therapy The overall study result: Stratum 2, mitt CV death, MI, stroke: HR=0.85 (0.73, 1.00) The 90 day study result: Stratum 2, mitt CV death, MI, stroke: HR=0.84 (0.68, 1.05) CC-80

81 Missing Data Roderick A Little, PhD Richard D. Remington Distinguished University Professor of Biostatistics University of Michigan CC-81

82 Key Questions Raised by the Advisory Committee in 2012 Is it possible to collect more vital status data? Does including the new vital status information change the conclusion? Is there evidence of bias from differential informative censoring? Is the study result robust to a variety of assumptions of missingness? CC-82

83 Key Questions Raised by the Advisory Committee in 2012 Is it possible to collect more vital status data? Does including the new vital status information change the conclusion? Is there evidence of bias from differential informative censoring? Is the study result robust to a variety of assumptions of missingness? CC-83

84 Subjects and Person Years Missing Data on Vital Status Missing vital status at the time of the last AC, May 23 rd 2012 Missing Vital Status mitt ITT Subjects Person Years 698/15,526 (4.5%) 38.2/16,470 (0.2%) 1338/15,526 (8.6%) 1,184/19,962 (5.9%) CC-84

85 Subjects and Person Years Missing Data on Vital Status Missing vital status at the time of the last AC, May 23 rd 2012 Including the new vital status (VS) data Missing Vital Status mitt mitt new VS ITT ITT new VS Subjects 698/15,526 (4.5%) 278/15,526 (1.8%) 1338/15,526 (8.6%) 495/15,526 (3.2%) Person Years 38.2/16,470 (0.2%) 15/16,470 (0.09%) 1,184/19,962 (5.9%) 486/19,934 (2.4%) CC-85

86 Missing Data Have Been Substantially Reduced Number of subjects: mitt N=15,526 ITT N=15,526 Missing vital status at the last Ad Com 698 1,338 Sponsor was not allowed to contact Sponsor was allowed to contact 516 1,025 Vital status confirmed* Still missing vital status * Includes 49 and 87 subjects in the mitt and ITT populations respectively whose vital status was known, but where subject consent was not obtained to use subject level data. CC-86

87 Why did the Sponsor Follow up on Vital Status Rather than the Primary Outcome? The May 2012 Advisory Committee recommended following up of vital status data The sponsor discussed the approach with the FDA Vital status is the most objective clinical endpoint, not subject to recall and ascertainment bias Following up vital status rather then the primary outcome substantially increases the proportion of subjects in which information can be obtained A very high proportion (87%) of all cause deaths reported in ATLAS are cardiovascular (CV) deaths and the reduction in the primary efficacy endpoint is driven by CV death CC-87

88 Follow Up: Distribution of Cases by Treatment Group for All Randomized Subjects/ITT Number of subjects: Rivaroxaban 2.5 mg BID N=457 5 mg BID N=471 Placebo N=410 Vital status confirmed* Alive Died Still missing vital status Allowed to be contacted 52 (34%) 49 (35%) 41(36%) Not allowed to be contacted 102 (66%) 93 (65%) 71 (64%) *Includes 87 subjects whose vital status was known, but where subject consent was not obtained to use subject level data. CC-88

89 Follow Up: Distribution of Cases by Treatment Group for All Randomized Subjects/mITT Number of subjects: Rivaroxaban 2.5 mg BID N=245 5 mg BID N=246 Placebo N=207 Vital status confirmed* Alive Died Still missing vital status Allowed to be contacted 23 (26%) 29 (35%) 16 (28%) Not allowed to be contacted 65 (74%) 55 (65%) 41 (72%) *Includes 49 subjects whose vital status was known, but where subject consent was not obtained to use subject level data. CC-89

90 Key Questions Raised by the Advisory Committee in 2012 Is it possible to collect more vital status data? Does including the new vital status information change the conclusion? Is there evidence of bias from differential informative censoring? Is the study result robust to a variety of assumptions of missingness? CC-90

91 Implications of the New Vital Status Data for All Cause Mortality mitt/stratum 2/2.5 mg BID 7.0 HR (95% CI) mitt 2.5 mg BID 0.64 (0.49, 0.83) p<0.001 Cumulative Event Rate (%) Riva 2.5 mg Placebo Days from Randomization CC-91

92 Implications of the New Vital Status Data for All Cause Mortality mitt/stratum 2/2.5 mg BID 7.0 HR (95% CI) mitt 2.5 mg BID 0.64 (0.49, 0.83) p<0.001 Including new VS 0.65 (0.50, 0.85) p<0.001 Cumulative Event Rate (%) Riva 2.5 mg Placebo Riva 2.5 mg including new VS data Placebo including new VS data Days from Randomization CC-92

93 All Cause Mortality Results are Unchanged Following Vital Status Ascertainment All Strata 2.5 mg BID vs Placebo HR (95% CI) 5 mg BID vs Placebo HR (95% CI) Originally reported mitt 0.68 (0.53, 0.87) 0.95 (0.76, 1.19) Originally reported ITT 0.72 (0.57, 0.90) 0.99 (0.80, 1.21) CC-93

94 All Cause Mortality Results are Unchanged Following Vital Status Ascertainment All Strata 2.5 mg BID vs Placebo HR (95% CI) 5 mg BID vs Placebo HR (95% CI) Originally reported mitt 0.68 (0.53, 0.87) 0.95 (0.76, 1.19) Following VS ascertainment mitt 0.69 (0.54, 0.88) 0.95 (0.76, 1.19) Originally reported ITT 0.72 (0.57, 0.90) 0.99 (0.80, 1.21) Following VS ascertainment ITT 0.74 (0.60, 0.92) 0.98 (0.81, 1.20) CC-94

95 Key Questions Raised by the Advisory Committee in 2012 Is it possible to collect more vital status data? Does including the new vital status information change the conclusion? Is there evidence of bias from differential informative censoring? Is the study result robust to a variety of assumptions of missingness? CC-95

96 What is Informative Censoring? Censoring is informative if subjects with missing follow up have different hazards than those with complete follow up, after adjusting for data up to time of loss of follow up. Hypothetical scenario of interest If subjects with missing follow up had a high rate of bleeding before dropping out from study, and High rate of bleeding leads to a higher chance of subsequent CV outcome events CC-96

97 Differential Informative Censoring: The Key Question Informative censoring is differential if it leads to bias in the comparison of rivaroxaban and placebo groups, that is, the differences in the hazard due to informative censoring in the treatment groups do not cancel out CC-97

98 Subjects with Missing Vital Status Resemble Those Who Did Not Have a Primary Endpoint ITT Population/ All Strata Primary efficacy endpoint Missing primary EP data Missing VS after follow up Baseline characteristics Death Yes No N 538 1,111 14, Median risk score * * Risk score is based on the GRACE score (Fox et al., BMJ : ) CC-98

99 Subjects with Missing Vital Status Resemble Those Who Did Not Have a Primary Endpoint ITT Population/ All Strata Primary efficacy Missing Missing endpoint primary VS after Baseline characteristics Death Yes No EP data follow up N 538 1,111 14, Median risk score * Components of Risk Score Prior CHF CrCl (ml/min) < Age > STEMI or NSTEMI No PCI for index event * Risk score is based on the GRACE score (Fox et al., BMJ : ) CC-99

100 Baseline Characteristics in Subjects with Missing Vital Status by Treatment Group ITT/All Strata Baseline characteristics Missing VS after follow up Riva 2.5 mg Riva 5 mg Placebo N Median risk score * * Risk score is based on the GRACE score (Fox et al., BMJ : ) CC-100

101 Adverse Events Within 90 Days Prior to Withdrawal/Last Contact ITT/All Strata Events 90 days prior to withdrawal/ last contact Death Primary efficacy endpoint Yes No Missing primary EP Missing VS after followup N 538 1,111 14, Number (%) TIMI major bleeding 33 (6.1) 45 (4.1) 20 (0.1) 5 (0.3) 4 (0.8) TIMI major/ minor bleeding 43 (8.0) 55 (5.0) 38 (0.3) 14 (0.9) 7 (1.4) Adverse events 511 (95) 608 (55) 1505 (10) 326 (22.0) 117 (24) CC-101

102 Adverse Events in Patients with Missing Vital Status by Treatment Group ITT Population/ All Strata Events 90 days prior to withdrawal/last contact Missing VS after follow up Riva 2.5 mg Riva 5 mg Placebo N Number (%) TIMI major bleeding 4 (0.8) 1 (0.5) 2 (1.1) 1 (0.7) TIMI major/minor bleeding 7 (1.4) 2 (1.1) 4 (2.3) 1 (0.7) Adverse events 117 (24) 47 (25) 38 (22) 32 (24) CC-102

103 Key Questions Raised by the Advisory Committee in 2012 Is it possible to collect more vital status data? Does including the new vital status information change the conclusion? Is there evidence of bias from differential informative censoring? Is the study result robust to a variety of assumptions of missingness? CC-103

104 Sensitivity Analysis Via Event Addition Commonly used but flawed for a number of reasons: Not specifically adjusted for subject characteristics Does not factor in time from withdrawal to end of study Does not take into account expected hazards over that time NRC panel recommended varying interpretable parameters (e.g., hazards) within a statistical model (e.g., Cox model) CC-104

105 NRC Endorsed Approach: Increase Hazard Rate in the Treatment Group to Determine Tipping Point Estimate hazard at time of withdrawal, adjusting for observed covariates and prior bleeding. Inflate the hazard only in the Rivaroxaban group by a known factor; assumes non informative censoring in the control group Impute events to the end of the study 1,000 times, assuming Weibull Distribution; combine using standard multiple imputation combining rules Increase the inflation factor until upper limit of 95% confidence interval crosses 1.0 the tipping point To assess the primary outcome, analysis is based on data prior to follow up on missing vital status CC-105

106 Primary Efficacy Results are Robust to Inflation of the Hazard in Subjects with Missing Data mitt/all Strata/Combined Doses TEST 1: Tipping Point = 2200 to 2300% inflation 1.1 HR (95% CI) Percent Inflation of the Hazard in Rivaroxaban Treated Subjects with Missing Data CC-106

107 Primary Efficacy Results are Robust to Inflation of the Hazard in Subjects with Missing Data mitt Stratum Rivaroxaban test group vs placebo Tipping Point % (fold) inflation of hazard in the Riva group Test 1 All Strata Combined Test 2 Stratum 2 Combined Test 3 Stratum mg BID % (23 24 fold) % (11 12 fold) % (4 5 fold) CC-107

108 CV Death Results are Robust to Inflation of the Hazard in Subjects with Missing Data mitt/stratum 2/2.5 mg BID 1.2 Tipping Point 9300% inflation 1.0 HR (95% CI) Percent Inflation of the Hazard in Rivaroxaban Treated Subjects with Missing Data CC-108

109 Primary Efficacy Results are Robust to Inflation of the Hazard in Subjects with Missing Data ITT Stratum Rivaroxaban test group vs placebo Tipping Point % (fold) inflation of hazard in the Riva group Test 1 All Strata Combined Test 2 Stratum 2 Combined Test 3 Stratum mg BID % ( fold) % ( fold) 60 70% ( fold) CC-109

110 Time to All Cause Death from Original Vital FU Date to New Vital FU Date for the 843 Patients who had Additional Follow up on Vital Status ITT/All Strata/Each Dose Cumulative Event Rate (%) HR (95% CI) 2.5 mg BID 1.12 (0.51, 2.47) p= mg BID 0.99 (0.44, 2.22) p=0.976 Placebo Riva 5 mg BID Riva 2.5 mg BID Relative Days from Original Last Contact on Vital Status CC-110

111 Treatment Effect is Not Overstated Because it is Larger in Regions with Low Missing Data mitt/all Strata/Combined 1.0 Primary Endpoint Hazard Ratio SA O WE EE A 0.6 NA Percent of Subjects With Missing Data SA = South America NA= North America WE = Western Europe EE = Eastern Europe A = Asia O = Other CC-111

112 Summary: ATLAS ACS 2 TIMI 51 Study Result is Robust to Missing Data Sponsor followed up about 2/3rds of the cases missing vital status data, and about 90% of cases they were allowed to follow up Inclusion of these data did not change the conclusions on survival at all; I expect similar results for the primary outcome Subjects still missing vital status after follow up resembled missing cases who were followed up, and subjects who completed the study without any clinical event Bleeding events prior to withdrawal in subjects with incomplete follow up were rare and quite balanced across treatment groups Study result withstands wide variety of sensitivity analyses CC-112

113 Dispute Resolution Position on Missing Data As noted in the March 4, 2013, complete response letter, the Division found the additional follow up to be reassuring with respect to informative censoring. Missing data do not present a barrier to approval of this supplement (page 5 of the FDA s dispute resolution response) Office of Drug Evaluation I Center for Drug Evaluation and Research Response to Janssen Research and Development Dispute Resolution CC-113

114 Strength Of Evidence Of The ATLAS ACS Program Results Jay P. Siegel MD Head of Scientific Strategy and Policy Chief Biotechnology Officer Johnson & Johnson CC-114

115 Legal Standard for Evidential Effectiveness Source of data to meet legal standard for substantial evidence clarified: Original language: Substantial evidence consists of evidence from: Adequate and well controlled trials Plural had been interpreted to mean at least 2 trials CC-115

116 Legal Standard for Evidential Effectiveness Source of data to meet legal standard for substantial evidence clarified: Original language: Substantial evidence consists of evidence from: Adequate and well controlled trials Plural had been interpreted to mean at least 2 trials New, additional language added in 1997 in FDAMA: The secretary may determine, based on relevant science, that data from: one adequate and well controlled clinical investigation and confirmatory evidence are sufficient to establish effectiveness and constitute substantial evidence CC-116

117 FDA Guidance Provided 1998 FDA provided guidance as to how it would interpret the law Focus on standards for one trial with confirmatory evidence CC-117

118 Key Concepts in the Guidance Four concerns requiring confirmatory evidence Biases, generalizability, fraud, chance Two major changes in clinical research since 1960s that allowed these concerns to be addressed without a 2 nd trial Multicenter studies of rigorous design and execution Narrower indications and uses Data from related uses/indications could support a single trial CC-118

119 Implication of these Changes The standard no longer requires two positive trials at p<0.05 Current approach Replication not required, may not be optimal Confirmatory evidence can come from a broad variety of sources CC-119

120 Guidance Section II.C.3: Evidence of Effectiveness from a Single Study characteristics of a single adequate and well controlled study that could make the study adequate support for an effectiveness claim. Large multicenter study (section II.C.3.a) Consistency across study subsets (section II.C.3.b) Multiple studies in a single study (section II.C.3.c) Multiple endpoints involving different events (section II. C. 3.d) Statistically very persuasive findings (section II.C.3.e) the presence of one or more in a study can contribute to a conclusion that the study would be adequate to support an effectiveness claim. FDA Guidance for Industry: Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products. May CC-120

121 Guidance Section II.C.3: Evidence of Effectiveness from a Single Study characteristics of a single adequate and well controlled study that could make the study adequate support for an effectiveness claim. Large multicenter study (section II.C.3.a) Consistency across study subsets (section II.C.3.b) Multiple studies in a single study (section II.C.3.c) Multiple endpoints involving different events (section II. C. 3.d) Statistically very persuasive findings (section II.C.3.e) the presence of one or more in a study can contribute to a conclusion that the study would be adequate to support an effectiveness claim. FDA Guidance for Industry: Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products. May CC-121

122 Guidance Section II.C.3: Evidence of Effectiveness from a Single Study characteristics of a single adequate and well controlled study that could make the study adequate support for an effectiveness claim. Large multicenter study (section II.C.3.a) Consistency across study subsets (section II.C.3.b) Multiple studies in a single study (section II.C.3.c) Multiple endpoints involving different events (section II. C. 3.d) Statistically very persuasive findings (section II.C.3.e) the presence of one or more in a study can contribute to a conclusion that the study would be adequate to support an effectiveness claim. FDA Guidance for Industry: Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products. May CC-122

123 Statistically Very Persuasive Finding Protocol pre specified primary analysis: All Strata, Combined Doses, mitt, HR=0.84; p=0.008; ITT p=0.002; CV Death made the greatest contribution; HR=0.80 mitt, 90 day; HR=0.78, p=0.006 Study subset of greatest interest: Stratum 2, Combined Doses, mitt: HR=0.86; p=0.024; ITT p=0.004 CV Death made the greatest contribution; HR=0.78 mitt, 90 day; HR=0.82, p=0.041 Statistically very persuasive finding on primary endpoint with positive results in subset of greatest interest If one views stratum 2 finding as primary, finding remains positive and is supported by the confirmatory evidence CC-123

124 Guidance Section II.C.3: Evidence of Effectiveness from a Single Study characteristics of a single adequate and well controlled study that could make the study adequate support for an effectiveness claim. Large multicenter study (section II.C.3.a) Consistency across study subsets (section II.C.3.b) Multiple studies in a single study (section II.C.3.c) Multiple endpoints involving different events (section II. C. 3.d) Statistically persuasive findings (section II.C.3.e) the presence of one or more in a study can contribute to a conclusion that the study would be adequate to support an effectiveness claim. FDA Guidance for Industry: Providing Clinical Evidence of Effectiveness for Human Drug and Biological Products. May CC-124

125 Stratum 1 Data Provide Highly Relevant Support for Efficacy Finding in Stratum 2 Findings in stratum 1, separate test of closely related hypothesis: HR= 0.69; p = (ITT=0.22) 90 days, p value=0.012 Phase 2 trial p value=0.034 (All doses, death/mi/stroke) Guidance specifically recognizes the confirmatory value of studies in combination and monotherapy Section II.C.2.d: Studies in combination or in monotherapy Section II.C.3.c Multiple studies in a single study ISIS 2 example in the guidance CC-125

126 Additional Confirmatory Evidence from Within ATLAS 2 Efficacy at 2.5 mg and 5 mg daily doses 2.5 mg: p=0.039 (ITT p=0.011) 5 mg: p=0.075 (ITT p=0.020) Evidence of efficacy on stent thrombosis Stent thrombosis p=0.022 (Stratum 2, 2.5 mg, definite or probable) Independent evidence of efficacy in medically managed patients: Primary efficacy endpoint p=0.03 (mitt, Stratum 2, Combined doses) CC-126

127 A Second Trial ATLAS ACS TIMI Study Centers 27 Countries N=3, Study Centers 44 Countries N=15,526 CC-127

128 Section 2.C.2.f: Studies in less closely related diseases, but where the general purpose of therapy is similar Rivaroxaban has FDA approval for several indications with the general purpose of decreasing thrombosis and its sequelae: Reduction of Risk of Stroke and Systemic Embolism in Nonvalvular Atrial Fibrillation Treatment of Deep Vein Thrombosis Treatment of Pulmonary Embolism Reduction in the Risk of Recurrence of Deep Vein Thrombosis and of Pulmonary Embolism Prophylaxis of Deep Vein Thrombosis Following Hip or Knee Replacement Surgery CC-128

129 Substantial Confirmatory Evidence for Rivaroxaban in Acute Coronary Syndrome ATLAS ACS 2 TIMI 51 Robust 766 center trial, 15,526 subjects Strongly positive results overall mitt or ITT, positive in Stratum 2, positive for CV mortality (overall and stratum 2) Abundant confirmatory evidence establishing stratum 2 results did not arise by chance Consistency across subgroups Benefit on 2.5 mg and 5 mg Benefit in stratum 1 as well as stratum 2 Benefit on stent thrombosis Independent benefit in medically managed patients Consistent findings in phase 2 Rivaroxaban s proven benefit on thrombotic complications in several uses CC-129