1. Surrogate Endpoints: The Challenges are Greater than they Seem March 7, 2005 Thomas R. Fleming, Ph.D. Professor and Chair of Biostatistics University of Washington NIDDK Workshop:

2. Surrogate Endpoints Criteria for Study Endpoints A Correlate does not a Surrogate Make Validation of Surrogates Controversial Issues with AA

3. Criteria for Study Endpoints in Clinical Trials Measurable/Interpretable Sensitive Clinically relevant ~ Retinopathy, Nephropathy ~ Major hypoglycemic events: Coma/Seizure

4. Use of Surrogate Endpoints Treatment Effects on Surrogate Endpoints eg: ~ Oncology: Tumor Burden Outcomes ~ HIV/AIDS: CD4, Viral Load ~ Cardiovascular Dis: B.P., Cholesterol ~ Type 1 Diabetes: HbA 1c, C-Peptide Establishes Biological Activity But Not Necessarily Clinical Efficacy

5. Surrogate Endpoints Criteria for Study Endpoints A Correlate does not a Surrogate Make Validation of Surrogates Controversial Issues with AA

6. Surrogate Endpoint: Not in Causal Pathway of Disease Process Disease Surrogate True Clinical Endpoint Endpoint Causal Pathway

7. The Surrogate Endpoint is not in the Causal Pathway of the Disease Process. Diseas e Biomarker Mother-to-Child e.g., CD4 Trans of HIV HIV Viral Load Anti-Islet End-Organ Autoantibodies Diabetic Complications β-Cell Function “Correlates”: Useful for Disease Diagnosis, or Assessing Prognosis and Effect Modification “Valid Surrogates”: Replacement Endpoints

8. Surrogate True Clinical Endpoint Intervention Disease True Clinical Endpoint Surrogate Endpoint Disease Intervention Multiple Pathways of the Disease Process

9. Surrogate True Clinical Endpoint Intervention Disease End-Organ Diabetic Complications HbA 1c Glycemic Control Disease Intervention Multiple Pathways of the Disease Process

10. CD4 Cell AIDS Events Count & Death IL-2 Disease IL-2: known > 200 CD4 cell count increase Unknown whether IL-2 is increasing the level of functional CD4 cells NIH is sponsoring the evaluation of 6000 patients, followed for >5 years, in SILCAAT and ESPRIT Time

11. Surrogate True Clinical Endpoint Endpoint Disease Intervention Interventions having Mechanisms of Action Independent of the Disease Process

12. Arrhythmia Overall Suppression Survival Disease Intervention Interventions having Mechanisms of Action Independent of the Disease Process

13. Surrogate Endpoints Criteria for Study Endpoints A Correlate does not a Surrogate Make Validation of Surrogates Controversial Issues with AA

14. End Stage Renal Disease Goal: Normalize Hematocrit Values and reduce Death and MI

15. Patient Distribution & Percent Deaths by Hematocrit % 27-30 30-33 33-36 36-39 39-42 STANDARD DOSE EPOGEN 60% 45% 30% 15% 0%

16. End Stage Renal Disease Goal: Normalize Hematocrit Values and reduce Death and MI

17. End Stage Renal Disease High Dose Epogen Standard Dose Epogen R Goal: Normalize Hematocrit Values and reduce Death and MI

18. Patient Distribution & Percent Deaths by Hematocrit % 27-30 30-33 33-36 36-39 39-42 STANDARD DOSE EPOGEN 60% 45% 30% 15% 0%

19. Patient Distribution & Percent Deaths by Hematocrit % 27-30 30-33 33-36 36-39 39-42 STANDARD DOSE EPOGEN HIGH DOSE EPOGEN 30%  death RR for 10 pt  in hem. 60% 45% 30% 15% 0% 60% 45% 30% 15% 0% 27-30 30-33 33-36 36-39 39-42

20. Patient Distribution & Percent Deaths by Hematocrit % 27-30 30-33 33-36 36-39 39-42 STANDARD DOSE EPOGEN HIGH DOSE EPOGEN 30%  death RR for 10 pt  in hem. 60% 45% 30% 15% 0% 60% 45% 30% 15% 0% 27-30 30-33 33-36 36-39 39-42

21. Patient Distribution & Percent Deaths by Hematocrit % 27-30 30-33 33-36 36-39 39-42 STANDARD DOSE EPOGEN HIGH DOSE EPOGEN 30%  death RR for 10 pt  in hem.  in hematocrit 60% 45% 30% 15% 0% 60% 45% 30% 15% 0% 27-30 30-33 33-36 36-39 39-42

22. Patient Distribution & Percent Deaths by Hematocrit % 27-30 30-33 33-36 36-39 39-42 STANDARD DOSE EPOGEN HIGH DOSE EPOGEN 30%  death RR for 10 pt  in hem.  in hematocrit 30%  in death RR 60% 45% 30% 15% 0% 60% 45% 30% 15% 0% 27-30 30-33 33-36 36-39 39-42

23. End Stage Renal Disease High Dose Epogen Standard Dose Epogen R Goal: Normalize Hematocrit Values and reduce Death and MI Besarab et al, NEJM 339:584-590, 1998: “  in incidence of thrombosis of vascular access sites”

24. How does one validate a surrogate endpoint?

25. Validation of Surrogate Endpoints Property of a Valid Surrogate  Effect of the Intervention on the Clinical Endpoint is reliably predicted by the Effect of the Intervention on the Surrogate Endpoint

26. Prentice’s Sufficient Conditions 1.The surrogate endpoint must be correlated with the clinical outcome 2.The surrogate endpoint must fully capture the net effect of the intervention on the clinical outcome

27. Z = 1 : Control ; Z = 0 : Intervention S(t) : Surrogate Endpoint at t (t | Z) = 0 (t) e    (t | Z,S(t) ) = 0 (t) e  Z +  S(t)  Proportion of net intervention effect explained by the surrogate endpoint: DeGruttola et al, J Infectious Diseases 175:237-246, 1997 p = 1 -  

28. Meta-analyses are required to explore the validity of surrogate endpoints

29. Z = 1 : Control ; Z = 0 : Intervention S(t) : Surrogate Endpoint at t (t | Z) = 0 (t) e    (t | Z,S(t) ) = 0 (t) e  Z +  S(t)  Proportion of net intervention effect explained by the surrogate endpoint: DeGruttola et al, J Infectious Diseases 175:237-246, 1997 p = 1 -  

30. HbA 1c Major Clinical Glycemic Control Events Unintended negative effects Alternative beneficial effects Disease Intervention Time

31. Validation of Surrogate Endpoints Statistical  Meta-analyses of clinical trials data Clinical  Comprehensive understanding of the ~ Causal pathways of the disease process ~ Intervention’s intended and unintended mechanisms of action

32. Hazard Ratios for DFS vs Overall Survival

33. Endpoint Hierarchy True Clinical Efficacy Measure Validated Surrogate Endpoint (Rare) Non-validated Surrogate Endpoint that is “reasonably likely to predict clinical benefit” Correlate that is solely a measure of Biological Activity

34. Illustrations of Valid Surrogates Preventing Mother-to-Child Transmission of HIV when using short course antiretrovirals ~ Prevention of AIDS and Death often occurring within two years Substantial Sustained Reduction in Blood Pressure when using β-blockers or low dose diuretics ~ Prevention of Fatal and Non-fatal Stroke

35. Hierarchy for Outcome Measures True Clinical Efficacy Measure Validated Surrogate Endpoint (Rare) Non-validated Surrogate Endpoint that is “reasonably likely to predict clinical benefit” Correlate that is solely a measure of Biological Activity

36. Establishing a Level #3 Outcome Measure Accurately representing the treatment’s effect on the predominant mechanism through which the disease process induces clinical risks Lack of large adverse effects on clinical endpoint not captured by the outcome measure Net effect on the clinical endpoint is consistent with what would be predicted by level of effect on the outcome measure Targeted effect on outcome measure sufficiently strong and durable to predict meaningful benefit

37. Hierarchy for Outcome Measures True Clinical Efficacy Measure Validated Surrogate Endpoint (Rare) Non-validated Surrogate Endpoint that is “reasonably likely to predict clinical benefit” Correlate that is solely a measure of Biological Activity

38. Surrogate Endpoints Criteria for Study Endpoints A Correlate does not a Surrogate Make Validation of Surrogates Controversial Issues with AA

39. FDA Oncology Drugs AC: 3/12-13/03 ’95-’00: 12 Accelerated Approvals Facts presented to ODAC: Of 12 AA, 8 remain unresolved: Average time from AA to Completion of Validation Trial projected to be 10 years In one case, sponsor enrolled 8 pts/year In 3 cases, Validation Trial indicated minimal treatment benefit

40. FDA Oncology Drugs AC: 3/12-13/03 ’95-’00: 12 Accelerated Approvals Disturbing Issues re Validation Trials: Enrollment difficulties into validation trials Cross-ins on the control arm Loss of “sense of urgency” by sponsor Lack of clear vision for proper process when the validation trial is not conclusively positive

41. FDA Oncology Drugs AC: 3/12-13/03 ’95-’00: 12 Accelerated Approvals Facts presented to ODAC: Of 12 AA, 8 remain unresolved: Average time from AA to Completion of Validation Trial projected to be 10 years In one case, sponsor enrolled 8 pts/year In 3 cases, Validation Trial indicated minimal treatment benefit

42. FDA Oncology Drugs AC: 3/12-13/03 ’95-’00: 12 Accelerated Approvals Disturbing Issues re Validation Trials: Enrollment difficulties into validation trials Cross-ins on the control arm Loss of “sense of urgency” by sponsor Lack of clear vision for proper process when the validation trial is not conclusively positive

43. Hierarchy for Outcome Measures True Clinical Efficacy Measure Validated Surrogate Endpoint (Rare) Surrogate Endpoint that is “reasonably likely to predict clinical benefit” None of the Above: A Correlate that is solely a measure of Biological Activity

44. Use of Biological Markers As “Correlates”… Disease Diagnosis, or assessing Prognosis or Effect Modification In Screening or Proof of Concept Trials… Primary Endpoint In Definitive Trials… Supportive Data on Mechanism of Action

45. NIDDK Workshop Surrogate Endpoints The Next Step after the Phase 1 Trial

46. Development Strategies After Phase 1: What should be the next step? ~ Phase 2 ~ Phase 2B (Intermediate Trial) ~ Phase 3

47. Why Conduct a Phase 2 Trial? Obtain improved insights: Biological Activity: Proof of Concept Refinements in dose/schedule Safety Improving adherence to interventions Improving quality of trial conduct - Timely accrual - High quality study implementation - High quality data, including retention

48. Development Strategies After Phase 1: What should be the next step? ~ Phase 2 ~ Phase 2B (Screening Trial) ~ Phase 3

49. The Randomized Phase 2B “Screening Trial” Illustration: Type 1 Diabetes Primary Endpoint: Time to Hypoglycemic Events or End-Organ Diabetic Complications Targeted Treatment Effect: 33% reduction in progression rate

50. Screening Trial Design Phase 3 Trial Design -33% 0% 33% 44% 67% Further Studies Positive -17% 0% 17% 33% 50% Positive

51. An Illustration of the Use of an Intermediate Trial Before a Definitive Trial Surgical Adjuvant Therapy of Colorectal Cancer 5-FU + Levamisole Levamisole Control R

52. SURGICAL ADJUVANT THERAPY OF COLORECTAL CANCER Surviving, % 0 1 2 3 4 5 6 100 - 80 - 60 - 40 - 20 - 0 Years from randomization NCCTG Trial 5-FU+LEV n=91 Levamisole n=85 Control n=86

53. Screening Trial Design Phase 3 Trial Design -33% 0% 33% 44% 67% Further Studies Positive -17% 0% 17% 33% 50% Positive

54. SURGICAL ADJUVANT THERAPY OF COLORECTAL CANCER Surviving, % 0 1 2 3 4 5 6 100 - 80 - 60 - 40 - 20 - 0 Years from randomization NCCTG Trial 5-FU+LEV n=91 Levamisole n=85 Control n=86

55. SURGICAL ADJUVANT THERAPY OF COLORECTAL CANCER Surviving, % 0 1 2 3 4 5 6 100 - 80 - 60 - 40 - 20 - 0 Years from randomization NCCTG TrialCancer Intergroup Trial 0 1 2 3 4 5 6 7 8 9 100 - 80 - 60 - 40 - 20 - 0 Years from randomization 5-FU+LEV n=91 Levamisole n=85 Control n=86 5-FU+LEV n=304 Levamisole n=310 Control n=315

56. Important Observations Confirmatory trials of promising results from Intermediate Trials can be performed successfully Confirmatory trials - can reveal true positives (eg, 5-FU+Lev) - can reveal true negatives (eg, Levamisole)

57. SURGICAL ADJUVANT THERAPY OF COLORECTAL CANCER Surviving, % 0 1 2 3 4 5 6 100 - 80 - 60 - 40 - 20 - 0 Years from randomization NCCTG TrialCancer Intergroup Trial 0 1 2 3 4 5 6 7 8 9 100 - 80 - 60 - 40 - 20 - 0 Years from randomization 5-FU+LEV n=91 Levamisole n=85 Control n=86 5-FU+LEV n=304 Levamisole n=310 Control n=315

58. R AZTLabor/Delivery/1 wk to I NVPSingle doses to M/I Illustration of a Screening Trial with “Compelling” Results: HIVNET 012 8/99 ResultsLancet 1999; 354: 795-802 MCT of HIV N 6-8 wks 14-16 wks AZT 30259 (21.3%) 65 (25.1%) NVP 30735 (11.9%) 37 (13.1%) 1p = 0.0014 1p = 0.0003

59. Screening Trial Design Phase 3 Trial Design -33% 0% 33% 44% 67% Further Studies Positive -17% 0% 17% 33% 50% Positive

60. Goals in Development of Diabetes Drugs Using Biomarkers to achieve a cost-effective research strategy Achieving rapid availability of drugs providing improved benefit-to-risk profile Achieving reliable as well as timely evaluation of both efficacy and safety of new interventions