Study F1K-MC-EVAD: Relative Risk under Original and Amended Protocols

1. Study F1K-MC-EVAD: Relative Risk under Original and Amended Protocols Relative Risk Patient Population Original Protocol Amended Protocol Interaction P-Value Entire Population 0.94 0.71 0.08 Sites Enrolling under Both Protocols 0.87 0.77 0.50 6586.01

2. Relative Risk Site Enrollment No. of Sites (no. of patients) Original Protocol (no. of patients) Amended Protocol (no. of patients) Interaction P-Value  25 11 (457) 0.91 (202) 0.83 (255) 0.75  20 20 (655) 0.84 (291) 0.67 (364) 0.36  15 38 (956) 0.79 (417) 0.69 (539) 0.52  10 62 (1255) 0.86 (537) 0.72 (718) 0.40  5 105 (1551) 0.89 (664) 0.72 (887) 0.24 All Sites 164 (1690) 0.94 (720) 0.71 (970) 0.08 Study F1K-MC-EVAD: Relative Risk by Site Enrollment 6587.01

3. Study F1K-MC-EVAD: Mortality by Type of Clinical Trial Material 6936.01

4. Study F1K-MC-EVAD: Cox Regression Analysis by Baseline and Any Heparin Exposure Placebo Patients 9270.01

5. Study F1K-MC-EVAD: Patient Populations within First APACHE II Quartile – IL-6 Level 6940.01

6. Study F1K-MC-EVAD: Location and Functional Status – Original vs Amended Protocol 6351.02

7. Study F1K-MC-EVAD: Safeguards to Maintain Study Blind during Conduct of Interim Analyses Unblinded pharmacist provides drotrecogin alfa (activated) or placebo to investigator Patient entered at investigative site Patient receives study drug and up to 28 days of follow-up External CRO provides randomized treatment code to unblinded pharmacist at site Data management functions performed by Lilly blinded to patient treatment codes Blinded interim analysis data sets prepared by Lilly and sent to external statistical services organization (SSO) SSO merged patient treatment codes provided by CRO and prepared interim analysis reports SSO provided interim analysis reports to DSMB DSMB provides recommendation to Lilly 6452.01

8. Study F1K-MC-EVAD: Mortality by Adequate Use of Antibiotics (CEC) 6718.01

9. Study F1K-MC-EVAD: Mortality by Culture or Blood Culture (CEC) 6714.01

10. Study F1K-MC-EVAD: Stepwise Multiple Logistic Regression • Common approach to model selection • Model provides probability of death estimates based on important covariates and treatment • Intuitively, covariates are included based on their explanatory "value" • Assess evidence of potential differential treatment effects with drotrecogin alfa (activated) accounting for the condition of the patient from multiple perspectives 6737.02

11. Study F1K-MC-EVAD: Stepwise Multiple Logistic Regression 21 variables considered for inclusion • Demographics: age, gender, origin, geographic region • Patient location prior to hospitalization, co-morbidity status, functional dependency status • Infection site and type, surgical status • Clinical markers: APACHE II, number of organ failures, renal SOFA score, respiratory SOFA score, cardiovascular SOFA score, ventilation status, shock status • Biochemical markers: Protein C activity, prothrombin time, APTT, log IL-6 6738.02

12. Study F1K-MC-EVAD: Stepwise Multiple Logistic Regression • Treatment required in model • All two-factor interactions (e.g., age by treatment, age by gender) included in stepwise procedure • Schwartz criterion chosen as method to select terms • Forward and backward steps • Goodness-of-fit assessed using the Hosmer-Lemeshow chi-square statistic 6739.01

13. Study F1K-MC-EVAD: Stepwise Multiple Logistic Regression Results • Following covariates retained in model • Age • APACHE II score • PT, log IL-6 • Dependency status, urosepsis or not • Goodness-of-fit statistic supports model's adequacy (p=0.50) • Estimated constant 40.2% increase in the odds of survival with drotrecogin alfa (activated) across the population • No interaction terms • no treatment-by-covariate interactions 6740.01

14. Study F1K-MC-EVAD: Data Safety Monitoring Board Members • Steven Opal, M.D. (Chairman)Professor of Medicine, Brown University • Edward Abraham, M.D.Professor of Medicine, Division of Pulmonary Sciences and Critical Care Medicine, University of Colorado Health Science Center • Steve Lowry, M.D.Chairman and Professor, University of Medicine and Dentistry of New Jersey. • Janet Wittes, Ph.D.Statistician, Statistics Collaborative Incorporation, Washington, DC 20036 • Statistician resource- Pat O'Meara, PhD Pat O'Meara Associates, Inc, a Statistical Services Organization (SSO) 6722.01

15. First patient enrolled - amended protocol Last patient enrolled - original protocol 0.40 Number of Sites = 164 Number of Patients = 1690 0.35 Placebo 28-Day Mortality Rate 0.30 0.25 Drotrecogin Alfa (activated) 0.20 1 Aug 1 Nov 1 Feb 1 May 1 Aug 1 Nov 1 May 1 Jul 1 Feb 1998 1998 1999 1999 1999 1999 2000 2000 2000 Over Time (by Covance Randomization Date) Study F1K-MC-EVAD: Cumulative Mortality 6316.02

16. First patient enrolled - amended protocol Last patient enrolled - original protocol 0.40 Number of Sites = 99 Number of Patients = 1463 0.35 Placebo 28-Day Mortality Rate 0.30 0.25 Drotrecogin Alfa (activated) 0.20 1 Aug 1 Nov 1 Feb 1 May 1 Aug 1 Nov 1 Feb 1 May 1 Jul 1998 1998 1999 1999 1999 1999 2000 2000 2000 Over Time (by Covance Randomization Date) Study F1K-MC-EVAD: Cumulative MortalitySites Enrolling under Original and Amended Protocols 6404.02

17. Why not a randomized controlled trial in kids? • Assuming a 12.5% placebo mortality rate and 10% drotrecogin alfa (activated) mortality rate • Using a 2-sided chi-square test for equality of proportions with an alpha level of 5% • With 80% power need 5172 patients • Largest pediatric trial ever was 396 patients 9040.01

18. Controlled Trial Sample Size Estimates Using a 2-sided chi-square test for equality of proportions with an alpha level of 5% and 80% power to detect a treatment effect 9191.01

19. Studies F1K-MC-EVAO/EVAD: Number of Organ Failures at Baseline – Pediatric versus Adult Patients *Wilkinson JD et al, 1987 J. Pediatr. 111(3):325-328 9303.01

20. Studies F1K-MC-EVAO/EVAD: Type of Organ Failure at Baseline – Pediatric versus Adult Patients *Wilkinson JD et al, 1987 J. Pediatr. 111(3):325-328 9304.01

21. 40 35 30 Pediatric Patients (N=83) 25 Adult Patients (N=1690) 20 Percent of Patients 15 10 5 0 Gram Gram Gram Fungal Viral Positive Negative Mixed Studies F1K-MC-EVAO/EVAD: Percent of Patients with Positive Cultures at Baseline 9140.01

22. Reporting of Adverse Events – Study F1K-MC-EVAO versus Study F1K-MC-EVAD • Study F1K-MC-EVAO: All clinical manifestations of severe sepsis were collected as adverse events • Study F1K-MC-EVAD: Clinical manifestations of severe sepsis were not collected as adverse events unless considered drug related 9137.01

23. Study F1K-MC-EVAO: Adverse Events Occurring in 5% of Patients – Part 2 9031.01

24. Study F1K-MC-EVAO: Baseline Organ Failures for Selected Safety Categories 9193.03

25. Study F1K-MC-EVAO: Intracerebral Hemorrhage – Patient Scenario 1 • 14 year-old enrolled with N. meningitidis in blood and cerebral spinal fluid, severe multi-organ dysfunction, shock requiring multiple inotropes, and coagulopathy. Anisocoria diagnosed and drotrecogin alfa (activated) stopped after 10.5 hours (platelet count 26x103/μL). Continuous veno-venous hemofiltration with heparin began about 10 hours post study drug and continued until the patient died. Clinical brain death diagnosed on Study Day 13. Postmortem CT scan revealed a right frontal parachymal hematoma, severe cerebral edema, and subarachnoid hemorrhage. The severe intracranial hemorrhage was considered possibly related to study drug infusion by the investigator. 6646.02

26. Study F1K-MC-EVAD: Location Prior to Hospitalization – First APACHE II Quartile 6497.02