220 likes | 396 Views
U.S. Food and Drug Administration. Notice: Archived Document The content in this document is provided on the FDA’s website for reference purposes only. It was current when produced, but is no longer maintained and may be outdated. .
E N D
U.S. Food and Drug Administration Notice: Archived Document The content in this document is provided on the FDA’s website for reference purposes only. It was current when produced, but is no longer maintained and may be outdated.
3 Year DFS vs. 5 Year OS as an endpoint for Adjuvant Colon Cancer Studies: Data from Randomized Trials D Sargent1, S Wieand2, R Goldberg1, M O’Connell2, A Benson3, P Catalano3, J Benedetti4, G Francini5, L Shepherd6, JF Seitz7, R Labianca8 1Mayo Clinic-NCCTG, 2NSABP, 3ECOG, 4SWOG, 5University of Siena (Italy), 6NCIC-CTG (Canada), 7University of the Mediterranean (France), 8Ospedali Riuniti (Italy)
Disclosure • Very preliminary work • Further analyses underway • Not for distribution or citation
Suggestion • 3 year disease free survival (DFS) is an appropriate endpoint to replace 5 year overall survival (OS) as an endpoint in adjuvant colon trials • Would allow more rapid completion, reporting of trials
Methods • Data from large randomized trials • Individual patient data (when available), or investigator furnished summaries. • Compare 3 year DFS, 5 year OS for each arm • Compare difference between ‘control’ and experimental arms • Weighted linear regression
*Did not include a no treatment control arm Total: 38 treatment arms
3 year DFS vs 5 year OS ~1980: C01 NCCTG-784852
3 year DFS vs 5 year OS ~1985: INT0035 CO2
3 year DFS vs 5 year OS ~1988: NCIC NCCTG-874651 Siena FFCD GIVIO
3 year DFS vs 5 year OS ~1990: INT0089 NCCTG-894651
3 year DFS vs 5 year OS ~1995: NCCTG-914653
Results • Regression equation: • 3 yr DFS=0.05+.93*5 yr OS • Coefficients not significantly different from 0 and 1 • Correlation 0.92, R2 = 0.85 • Largest difference 6%, 36 of 38 < 3% difference
Conclusion (1) • On an arm by arm basis, 3 year DFS is an excellent predictor of 5 year OS • Event rates virtually identical • No impact on sample size • Power for 3 yr DFS will adequately power 5 yr OS
Within study comparisons • Does a comparison of arms using 3 yr DFS reach the same conclusion as using 5 yr OS?
3 yr DFS > 5 yr OS 6 ‘Experimental’ 3 ‘Control’ 3 yr DFS < 5 yr OS 5 ‘Experimental’ 7 ‘Control’
Difference in 3 yr DFS vs Difference in 5 yr OS In 8 of 16 within study comparisons, difference in 3 yr DFS > difference in 5 yr OS, 4 had the same difference.
Results • Regression equation: • D 5 yr OS = 0.85* D 3 yr DFS • Correlation 0.87 • R2 = 0.75 • Of 16 total comparisons • 12 same conclusion both DFS & OS • 9 No difference, 3 significant difference • 3 significant only for DFS • 1 significant only for OS
Conclusion (2) • As an endpoint for comparison, 3 yr DFS D may slightly overestimate 5 yr OS D • May change the conclusion in a non-trivial proportion of trials • If long-term OS is what we really care about, conclusions based on 3 yr DFS must be considered ‘subject to confirmation’
Discussion • Did not have individual patient data from all trials • Most had a no treatment control • Need to study more trials with ‘active’ treatment in all arms • B’s and C’s mixed together, may be beneficial to separate • Relevance for non-cytotoxic agents
Future work • More trials • Mimic actual trial conduct • Recreate dataset that would have been available at each given time-point • Other endpoints: 2 yr DFS, 3 yr OS