Potential insights from population kinetic assessment of progression-free survival curves

Abstract

Progression-free survival (PFS) curves follow first order kinetics on exponential decay nonlinear regression analysis (EDNLRA). Some exhibit 1-phase-decay, some have 2-phase-decay, some are convex. We digitized, performed EDNLRA and generated log-linear plots for 887 published PFS curves for incurable solid tumors treated with various systemic therapies. Proportion of curves fitting 2-phase-decay varied by therapy (p < 0.0001). For 13 therapies, >64 % of PFS curves had 2-phase-decay. This included epidermal growth factor receptor inhibitors in unselected lung cancer patients (some with, some without mutations), immune checkpoint inhibitors, interferon, breast cancer hormonal therapies, and selected others, suggesting 2 distinct, potentially identifiable subpopulations with differing progression rates. For 22 other therapies, <25 % of PFS curves had 2-phase-decay. Only 1 therapy was in the mid-range. Small cell lung and colon carcinomas were particularly likely to yield highly convex curves (p < 0.006), probably from discontinuation of effective therapies. PFS curve shape may yield biological and clinical insights.

Keywords: Population kinetics; Progression-free survival curve shape.