Relationship between effects on time-to-disease progression and overall survival in studies of metastatic breast cancer

Abstract

The relationship between overall survival (OS) and disease progression end points has been demonstrated in colorectal, colon, and non-small cell lung cancers. We assessed the association between OS and time-to-progression (TTP) or progression-free survival (PFS) in metastatic breast cancer (MBC) studies. A literature search retrieved all randomised controlled trials since 1994 in patients with MBC in which OS and either TTP or PFS were reported. Summary data on trial and patient characteristics were abstracted. Study effect sizes were derived as the ratio of median progression (or survival) times, which approximates the hazard ratio. Effects were centred at zero for regression analyses weighted by sample size. Numerous treatments were represented in 67 studies (17 081 patients). Modeling showed a positive association between outcomes for progression and survival (R(2)=0.30) with a slope of 0.32 (P<0.001) and a non-significant intercept. Thus, a treatment effect on TTP/PFS translated into a concordant effect on OS, but with attenuated effect size. Similar results were found in models of subsets and sensitivity analyses. These results demonstrate that treatment effects on progression end points in MBC trials are expected to result in treatment differences on OS that are smaller yet consistently in the same direction.

Figure 1

Flow diagram of literature search.

Figure 2

Plot of HR for survival vs HR for progression by study size regression line: Effect_OS=0.32 × Effect_TTP where Effect is the HR centred at unity. Bubbles show relative sample sizes from each study.