Key Discussions on Progression-Free Survival in Clinical Research

1. “Understanding Progression-free Survival”Thoughts around some clinical and methodological issues and possible regulatory consequences EFSPI, Basel, June 2010

2. Clinical issues • Is delaying PFS per se of value? • Yes, tumour progression heralds symptomatic progression. Patients rarely die of non-progressing tumours. • If you compare AB and A, is an event of progression on AB the same as an event on A? • Probably not, at least progression on AB signifies resistance to both A and B BJ, June 2010

3. Methodological Issues • Independent review (IR) vs. Investigator’s assessment (IA) • Poor concordance • Investigator events are censored • Informative censoring • Time to event is prolonged based on IR data BJ, June 2010

4. Methodological Issues • Patients withdrawn prior to event, if not followed off therapy until event • Irrespective of cause • In many cases informative censoring • Direction might be hard to understand • Only “non-progressing” and tolerability problems • Open label BJ, June 2010

5. Regulatory Considerations • AB vs. B and the progressing tumour • Focused on introduction of maintenance therapy and compounds designed to delay premalignant disorder to progress to malignant disorders. • (At least) ORR on next-line therapy • How to compare? • If we could assess all patients enrolled, no problems • But a sub-set? (e.g. 40% in the exp. arm vs. 55% in the control) and if the ORR tend to be lower in the exp. arm? BJ, June 2010

6. Tumour Progression • Angiogenesis inhibitors (AI)? • If the PFS benefit is e.g. 3 months, survival benefit tends to be similar • Does this hold true for AI? • Non-clinical data indicate that tumour characteristics may change towards a more “metastatic behaviour” • PFS is a composite • At least to look into proportion of events • new lesions, increased size of existing lesions, deaths BJ, June 2010

7. PFS as an informative composite • Targeted drugs and “new toxicities” • Hard to differentiate deaths due to cancer and complications related to therapy. • Events of death in the PFS composite BJ, June 2010

8. Primary analyses? • IR vs. IA • IR mainly to confirm absence of bias. • IA as primary analysis. • Should we recommend a confirmatory round of imaging after investigator assessed progression? BJ, June 2010

9. ”ITT” • Main concern, informative censoring. • Follow patients until PFS event on or off study therapy or next line therapy. • Primary or secondary/sensitivity analysis? • To me no major issue BJ, June 2010

10. Summary • ”PFS” has served us reasonably well • But • The definition of ”progression” is rather arbitrary (WHO, RECIST) • The meaning of ”progression” is likely to differ between treatment arms • It is a composite where the components are rarely analysed and discussed • The otherwise standard ITT approach has been ”forgotten” • We have accepted a method leading to informative censoring as ”gold standard” BJ, June 2010

11. Back-up topics • Interim analyses • NfG – to be done only when data are ”mature” • Treatment effect might differ btweeen those with good and bad prognosis. • ”Mature” when there is a ”sufficient” number of events after the median • Has never been challenged from a statistical perspective • Still we see this in many ScA procedures BJ, June 2010

12. Back-up topics • When OS and when PFS? • NfG – OS when the expected OS after progression is short and no known active next-line therapies • Haematological Appendix – focuses on B/R in the planning – if major difference in toxicity is expected favouring the control – OS • Class of compounds of importance??? BJ, June 2010

13. Revision of the NfG foreseen • Concept paper • Official comment on CP to EMA • Personal comments on this presentation/topics discussed today and on the current NfG and appendices may be directly sent to me • Bertil.Jonsson@mpa.se BJ, June 2010