Atrasentan For Men with Metastatic Hormone Refractory Prostate Cancer ODAC September 13 th, 2005 Clinical-Statistical Presentation Amna Ibrahim MD Shenghui Tang PhD DDOP, FDA

Outline of Presentation 1- Past approvals 2- Phase III study –Study design –Results of protocol-specified endpoints –Subgroup analyses –Reliability of results –Clinical relevance of results 3- Phase II study Protocol, conduct and results 4- Major safety concerns- Phase III and II

Past HRPC Approvals 1.Mitoxantrone + Prednisone (approved 1996) - Primary endpoint: confirmed improvement in pain - Interpretable pain severity scale - Pre-specified analysis plan for pain evaluation - Improved time to progression 2.Taxotere + Prednisone (approved 2004) - Primary endpoint: Overall survival For Both Approvals: Primary endpoint: Pre-specified Results: Persuasive- statistically & clinically

Design of the Atrasentan Study Phase III Study Double blinded randomized study Central independent review blinded to PSA QoL was tertiary endpoint without detailed analysis plan No mention of a specific measurement of pain submitted

Disease Progression Definition Phase III Study Protocol-specified Primary Endpoint: Time to Disease Progression- ITT Events constituting Disease Progression (DP) Radiographic Progression Pain requiring intervention Prostate Cancer complications requiring interventions Skeletal Related Events (SREs)

Events defining DP Phase III Study- applicant analyses For both treatments arms combined (all progression events) A- Radiographic Progression74% B- Pain20% C- Interventions3% D- SREs2% ABCD % patients with any event Placebo N=401 57%17%2% 77% Atrasentan N=408 55%14%2% 73%

Results of Protocol-Specified Endpoints Phase III Study Failed primary endpoint TDP Failed 4 of 5 secondary endpoints (-) OS, Time to PSA progression, change in bone scan index, & PFS (+) Mean change in ALP but difference of only 20 ng/ml, and missing data Failed several tertiary endpoints (-) QoL adjusted TDP (QATDP), KPS and mean change in PSA.

Phase II Study Randomized 3-arm, phase II study Pretreated population –different from Ph III study – Any therapy (except hormones) 75% on placebo & 66% atrasentan 10 mg –prior chemo: 25% on placebo & 18% on Atrasentan 10 mg Pts with prostate ca pain not excluded Primary endpoint: TDP (defined differently from Ph III)

Disease Progression Phase II Study Disease Progression Definition intervention for cancer-related event new symptoms related to tumor growth New pain requiring opioids (evidence of disease and duration not required) Other investigator defined measures not well-defined (incl. PSA rise, weakness, steroids use and deterioration)

Conduct of Studies (Applicant analyses for Phase III & II Studies) Phase III study protocol violations: Atrasentan 10 mg: 12% Placebo: 18% Phase II study protocol violations: Atrasentan 10 mg:58% Placebo:38% Phase II missing data for imaging studies: < 50% paired bone scans and CT scans available

FDA Statistical Review Shenghui Tang, PhD Mathematical Statistician FDA/CDER/Division of Biometrics I

Outline: Major Statistical Problems Early closure of the phase III study Failed primary efficacy (Phase II & III) Submitted analyses are post hoc: –subgroup & pooled analyses –QoL and pain analyses No adjustment for multiple comparisons

Early Closure of Phase III Study Independent Data Monitoring Committee recommended closure of the phase III study due to lack of efficacy The study closed on March 19, patients, 343 TDP events

Failed primary efficacy Phase III & II ITT Analyses Phase IIIPhase II N Median Atrasentan 10mg Placebo 91 days 86 days 183 days 137 days P-value, Log-rank Hazard Ratio 95% CI 0.88 (0.75, 1.03) 0.77 (0.55,1.09)

Failed primary efficacy Phase II & III No alpha left for further testing Any further analysis inflates false positive rate

Secondary Efficacy Analyses Time to onset of PSA progression unadjusted nominal p-value=0.344 Mean rate of change from baseline to final value in total bone scan index unadjusted nominal p-value=0.051 Overall survival unadjusted nominal p-value=0.791, HR=0.982 Mean change from baseline to final value in bone alkaline phosphatase (ALP) unadjusted nominal p-value=0.001

Secondary and Tertiary Efficacy Analyses are Exploratory Per applicant’s specified protocol: “……. If the primary efficacy analysis is not statistically significant at the α=0.05 level, then statistical significance will not be declared for any of these secondary analyses, regardless of the observed p- values.”

Subgroup Analyses Two main subgroup analyses in phase III study were submitted: Per-protocol patient population (12/04) 83% overall patient population Patients with bone metastases at baseline (BM) (12/04) - Clinical Disease Progression analysis in BM subgroup (First reported in the briefing package)

Subgroup TDP Analyses Phase III Per-Protocol & Bone Mets at Baseline Per-protocolBone Mets N Medians Atrasentan Placebo 90 days 85 days 92 days 85 days Hazard Ratio 95% CI 0.80 (0.68, 0.96) 0.80 (0.67, 0.95)

Subgroup TDP Analyses Phase III Per-Protocol & Bone Mets at Baseline Protocol stated that significance will not be declared for the per-protocol analysis Bone metastatic subgroup analysis is a post-hoc analysis

Clinical Disease Progression Analysis Not pre-specified Not adjusted for multiple comparisons Informative censoring of radiological progressions >75% censored for progression

Guidelines for Statistical Principles for Clinical Trials (ICH E9) ‘ …, adjustment should always be considered and the details of any adjustment procedure… should be set out in the analysis plan.’ ‘ In most cases, however, subgroup and interaction analyses are exploratory and should be clearly identified as such; they should explore the uniformity of any treatment effects found overall.’

Guidelines for Structure and Contents of Clinical Study Reports (ICH E3) Examination of Subgroups: ‘These analyses are not intended to "salvage" an otherwise non-supportive study but may suggest hypotheses worth examining in other studies or be helpful in refining labelling information, patient selection, dose selection etc.’

Problems with Subgroup Analyses High false positive or false negative rates. False positive finding increases with number of significance tests Not pre-specified = Post-hoc analysis Primary failed → P-value not interpretable Subgroup Analyses are Exploratory

Pooled Analysis of Phase II & Phase III Studies Pooled analysis not acceptable: 1.Neither trial individually shows a statistically significant difference 2. Different Definitions of TDP 3. Different Patient populations 4. Atrasentan formulations not bioequivalent Continued

Pooled Analysis Phase II & Phase III Studies (cont’d) 5. Post-hoc analysis 6. No independent review of progression evaluation conducted in Phase II study 7. Pooling data causes imbalance in randomization 8. Type I error not controlled

Quality of Life Analysis QoL tertiary endpoint Measured using FACT-P and EORTC-C30 No hypothesis for QoL analysis Compare differences in mean scores No adjustment for multiple comparisons

Quality of Life Analysis Mean Change from Baseline to Final Assessment for FACT-P and Subscores: ITT Subject Population

Prostate Cancer Subscore (PCS) I am losing weight I have a good appetite I have aches and pain that bother me I have certain areas of my body where I experience significant pain My pain keeps me from doing things I want to do I am satisfied with my present comfort level I am able to feel like a man I have trouble moving my bowels I have difficulty urinating I urinate more frequently than usual My problems with urinating limit my activities I am able to have and maintain an erection

Problems with QoL Analysis Clinical significance of the PCS mean change of 1.02 on a scale of 0-48 Recall Bias PCS score does not capture all the patient perceived impact of atrasentan treatment Missing data

Pain Analyses Change in pain-related scores (12/04) Time to 50% deterioration (First reported in briefing package, 8/05)

Pain-related Questions in PCS I have aches and pain that bother me I have certain areas of my body where I experience significant pain My pain keeps me from doing things I want to do I am satisfied with my present comfort level Not Specific to Bone or Prostate Cancer Pain

Mean Change in Pain-related Scores in PCS Not designed or validated for such use. Clinical significance of the PCS pain score mean change of 0.7 on a scale of 0 to 16 Each pain item measures a different attribute of pain 7-day recall period. Questions not specific to bone pain. Do not adequately measure pain

Time to 50% Deterioration in FACT-P Pain-Related QoL Scores Analysis in BM AtrasentanPlacebo Bone Mets, N= Time 0, N=321 (90%)305 (91%) Time 100 days, N=173 (49%)131 (39%) Time 500 days, N=23 Not Adjusted for Multiple Comparisons

Clinical Review Continued Outline Clinical relevance of results (TDP) Reliability of results (TDP) Post hoc analyses (clinical aspects) Safety Conclusion

Hazards of the Hazard Ratio Regarding Ratios (e.g. HR, proportions and odds ratios) Meaningful only if also clinically relevant Units of time not represented (improvement of 3 days to 6 days same as 3 years to 6 years) For primary endpoint in the 3 populations, HR <1, but not clinically meaningful.

Clinical Relevance of TDP Results Phase III Study – Applicant Analyses ArmNEventsMedianDifference ITT Placebo40178%86 days 5 days Atrasentan40873%91 days Per protocol Placebo32982%85 days 4 days Atrasentan34275%89 days Bone mets at baseline Placebo33279%85 days 7 days Atrasentan35274%92 days

Reliability of difference in rTDP hypothetical situation Actual Progression Arm A Imaging Study Months 3 Actual Progression Arm B TDP is 3 months on both arms Time Zero

Reliability of TDP Results Phase III Study Time to radiographic progression is time to imaging Radiographic progressions drove study results (74% of DP events) Imaging scheduled q 12 weeks per protocol (84 d) Median TDP is ~ 89 days No reliable effect of atrasentan was observed

Mean time to bone scan Phase III study Pre-specified time to bone scan Atrasentan Days SD Placebo Days SD Week Week Week Week Week Week

Subgroup Analyses Phase III Study - Applicant Table from CSR Population/SubpopulationHRHR “p” ITT Per-protocol Bone Mets at baseline No Bone mets at baseline Soft-tissue mets at baseline No Soft-tissue mets at baseline Bone & Soft-tissue mets at baseline Bone but no Soft-tissue mets at baseline Soft-tissue but no Bone mets at baseline No mets at baseline

Subgroup Analyses Phase III Study - Applicant Table from CSR Population/SubpopulationHRHR “p” ITT Per-protocol Bone Mets at baseline No Bone mets at baseline Soft-tissue mets at baseline No Soft-tissue mets at baseline Bone & Soft-tissue mets at baseline Bone but no Soft-tissue mets at baseline Soft-tissue but no Bone mets at baseline No mets at baseline

Pain Pain not primary endpoint for atrasentan studies unlike mitoxantrone study. Mitoxantrone & Prednisone study - 2 point improvement in 6 point scale of at least 6 weeks of duration (29% vs 12%) - Improved median time to progression (4.4 mo vs. 2.3 mo.) Atrasentan study - ~1 in 0-48 point scale (FACT-PCS) in a 2-8 point scale (EORTC QLQ-C30, 2 items only) & 0.7 in a 0-16 point scale (4 of 12 items from PCS) Magnitude of effect on pain small and duration not considered in atrasentan study

Retrospective Pain Analysis Time to 1 st AE of bone pain - No requirement for routine assessment of bone pain in AE reporting - for e.g. prostate cancer AE reporting was 12% on atrasentan and 16% on placebo arm AE reporting not meant to be used as an endpoint

Changing Efficacy Analyses 03/03 - DSMB closed phase III study for futility - NDA submitted. Many subgroup analyses and pooled analyses submitted as basis of efficacy. - FDA requests statistical adjustments for multiple post-hoc analyses and primary basis for efficacy - “The efficacy analysis … (to) support of the clinical benefit of atrasentan is the analysis of the population of patients with baseline metastatic disease to bone in Study M00-211” - New indication proposed based on a retrospective analysis - Briefing document with new analyses submitted 12/04 4/05 6/05 7/05 8/05

Efficacy Summary Phase II study is not acceptable. Phase III Study: Primary endpoint failed Secondary endpoints failed or not clinically meaningful (includes bone markers, & QoL) Bone markers & QoL endpoints pre-specified in the secondary and tertiary analyses failed Marginal improvement of questionable clinical relevance and reliability in all 3 populations presented Multiple post hoc analyses warrant further study

Safety Phase III & II studies

Safety Phase III Study Adverse EventAtrasentan N= 404 Placebo N=397 Deaths- any cause166 (41%)158 (39%) Deaths- CV causes8 (2%)2 (0.5%) CAD grade 3/4 events8 (2%)2 (0.5%) CHF grade 3/4 events11 (3%)3 (0.75%) Arrhythmias20 (5%)5 (1%) Pneumonia grade 3/46 (1%)0

Coronary Artery Disease Phase III Study Costart TermAtrasentan N=404 Placebo N=397 MI92 Angina pectoris53 Coronary artery disorder20 No. of pts- all CAD events135 Pts w grade 3 or 4 events82 Deaths21

Coronary Artery Disease Phase II Study Costart Term Atrasentan 10 mg (N=89) Atrasentan 2.5 mg (N=95) Placebo (N=104) Angina pectoris231 CA disorder300 Coronary occlusion010 MI100 # of Patients with CAD 6 (7%)4 (4%)1 (1%) Deaths110

Arrhythmias Phase III study Arrhythmia Events: Arrhythmia, Atrial fibrillation, Atrial flutter, Bradycardia, Extrasystoles, Palpitation, Supraventricular extrasystoles, Supraventricular tachycardia, Tachycardia, Ventricular extrasystoles Number of patients: Atrasentan (n=404):20 (5%) Placebo (n=397): 5 (1%)

Arrhythmias Phase II study Arrhythmia Events: Arrhythmia, atrial fibrillation, AV block, bigeminy, electrocardiogram abnormal, palpitations, PR interval prolonged, sinus bradycardia, tachycardia, ventricular extrasystoles Number of patients: Atrasentan 10 mg (n=89): 9 patients (10%) Atrasentan 2.5 mg (n=95): 8 patients (8%) Placebo (n=104): 7 patients (7%)

Heart Failure Phase III Study Costart termAtrasentanPlacebo Peripheral edema16250 CHF Excl. peripheral edema 173 Grade 3/4 CHF Excl. peripheral edem 113 Deaths61

Heart Failure Phase II Study Costart TermAtrasentan 10 mg Atrasentan 2.5 mg Placebo Peripheral edema # of patients with CHF 441

Conclusions Primary endpoint failed Pre-specified stats plan: If Primary endpoint fails, study fails Most secondary endpoints failed (including time to PSA change, PFS & OS) Many tertiary endpoints failed (including bone markers and QoL-related endpoints) Pain-related endpoints not prespecified Safety concerns (CAD, CHF, Arrhythmias) Atrasentan compared to placebo and not active control

Considerations Phase III study results not robust Closed early by DSMB for futility Failed pre-specified endpoints Not persuasive- statistically or clinically Signals for serious cardiac toxicity noted Additionally Failed study in earlier stage of prostate cancer Further studies in planning stage Should this drug be studied further before consideration of widespread use?