Abstract P5-09-02: How Long Have I Got? Estimating Post-Progression Survival after First Line Chemotherapy for Metastatic Breast Cancer Based on Data from Randomized Trials

Abstract

Abstract Background: Two recent systematic reviews have reported that median post-progression survival (mPPS) was typically double the median progression free survival (mPFS) in contemporary randomized trials of first line chemotherapy for metastatic breast cancer (MBC). We sought to determine the ratio of PPS to PFS and the accuracy of using simple multiples of PFS to estimate PPS in an independent dataset. Methods: We analysed survival data from 323 women in a randomized trial of capecitabine versus CMF as first line chemotherapy for MBC, both for the trial overall (using the median Kaplan Meier estimates: mPFS and mPPS), and for each individual woman (using her own PFS and PPS). PPS was estimated for each woman by (i) doubling her PFS (as per the systematic reviews), and (ii) by multiplying her PFS by the median of the individual patient ratios of PPS:PFS for this trial. Estimates were deemed accurate if they were within 0.75 to 1.33 times the actual value. The primary population of 221 women excluded 27 censored for progression, 65 censored for death, 8 with PFS <2 weeks, and 2 with PFS = OS. The secondary population included all 323 patients. Distributions were summarized by their medians and percentiles reflecting the typical range (25th to 75th, IQR), worst case (90th) and best case (10th) scenarios. Results: The median age was 62, ECOG PS was 0-1 in 87%, 46% had liver metastases and 64% had hormone receptor positive disease. Kaplan Meier estimates of survival times in months for the primary population (secondary population) were mPFS 5.3 (6.3), mPPS 8.8 (11), and mOS 16 (21). The ratio of mPPS:mPFS for the trial overall was 1.7. Ratios of PPS:PFS for individual women were lower with medians of 1.5 (IQR 0.6 - 3.1) in the primary population and 1.3 (IQR 0.5 — 3.0) in the secondary population. Estimates of a woman's PPS as double her PFS were accurate in 17% of women. Estimates of a woman's PPS as 1.5 times her PFS were accurate in 20% of women. When each woman's PPS was estimated as 1.5 times her PFS, 10% of women survived for less than 0.18 times their estimate (<0.18 x 1.5PFS), 10% survived for longer than 4 times their estimate (>4 x 1.5PFS)and the middle 50% survived between 0.37 to 2.1 their estimate (0.37 x 1.5PFS to 2.1x 1.5PFS). Analysis of accuracy in the secondary population gave similar results. Conclusion: The ratio of mPFS:mPPS in this randomized trial was similar to that seen in recent systematic reviews, but higher than the individual ratios of PFS:PPS for most women in the trial. A woman's PPS was typically 1.5 times her PFS, but estimates based on this ratio were imprecise and better explained in terms of a typical range (50% living for half to double the estimate), best case scenario (10% living beyond 4 times the estimate) and worst case scenario (10% dying within one sixth of the estimate). Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P5-09-02.