Do progression-free survival benefits justify increased toxicity and treatment burden? A meta-regression analysis of phase III immune checkpoint inhibitor trials from a value-based oncology perspective.

PURPOSE: Whether PFS reliably predicts overall survival (OS) in ICI trials, and whether PFS gains justify the toxicity they impose, is unclear. We quantified both relationships across phase III ICI trials from a value-based oncology perspective.

METHODS

We analysed 80 phase III randomized ICI trials (48,415 patients). Pooled hazard ratios (HRs) were estimated by random-effects meta-analysis, and the trial-level PFS-OS association by weighted least squares meta-regression with prespecified stratification by treatment type and models for non-linearity. Treatment burden was the between-arm difference in grade ≥3 adverse events (primary measure), serious adverse events, and treatment discontinuation. Only trial-level aggregate data were available.

RESULTS

Pooled PFS HR was 0.717 (95% CI 0.676-0.762; I2=82.6%) and OS HR 0.759 (0.736-0.784; I2=25.5%). PFS benefit explained 11.7% of between-trial variance in OS benefit (bootstrap 95% CI 1.0-32.0%). In exploratory stratified analyses this differed by treatment type (Chow F=4.99, p=0.0003), being largest for ICI plus chemotherapy (R2=50.4%, 95% CI 34.7-68.4%), of uncertain strength for ICI plus TKI (R2=43.5%) and dual ICI (R2=31.3%), and weakest for monotherapy (R2=3.9%, 95% CI 0.0-61.9%), with non-linearity in the latter (p = 0.013). Grade ≥3 adverse events increased in 66% of ICI arms and larger PFS gains accompanied greater toxicity (R2=22.5%, p < 0.0001).

CONCLUSIONS

PFS showed a limited and treatment-type-dependent association with OS, and PFS gains frequently accompanied increased toxicity. Formal surrogate validation requires individual-level data and was not undertaken.