Progression-free survival as a surrogate endpoint for overall survival in metastatic triple-negative breast cancer: A systematic review and meta-analysis.

Abstract

1131 Background: Overall survival (OS) remains the gold standard endpoint in metastatic triple negative breast cancer (mTNBC), but its assessment is confounded by post progression therapies and prolonged follow up. Progression-free survival (PFS) is frequently used as a primary endpoint; however, its validity as a surrogate for OS in mTNBC remains uncertain. We conducted an updated systematic review and meta analysis to evaluate the strength and reliability of PFS as a surrogate endpoint for OS across contemporary RCTs. Methods: We searched PubMed, Embase, Cochrane Library, and major oncology conference proceedings through January 2026 for randomized controlled trials evaluating systemic therapies in metastatic or unresectable TNBC. Eligible trials reported hazard ratios (HRs) for both PFS and OS. Trial level correlations between log(HR_PFS) and log(HR_OS) were assessed using weighted Spearman correlation and weighted linear regression, with sample size based weights. Certainty of evidence was assessed using a modified GRADE framework. Results: Twenty-seven treatment comparisons from 29 trials (8,527 patients) met inclusion criteria. Across all comparisons, PFS demonstrated a strong correlation with OS (weighted Spearman ρ = 0.80; 95% CI, 0.55-0.91), with a moderate surrogate relationship (R² = 0.64). The regression slope (0.66; 95% CI, 0.46-0.87) indicated attenuation of OS benefit relative to PFS. The surrogate threshold effect corresponded to an HR PFS of 0.96. Eight trials (30%) showed concordant PFS and OS benefit, while 44% demonstrated PFS improvement without significant OS gain. Surrogacy was strongest in phase III trials (R² = 0.71). Treatment class analyses revealed strong surrogacy for immune checkpoint inhibitors (R² = 0.95) and platinum based chemotherapy (R² = 0.87), but weak correlation for AKT pathway inhibitors (R² = 0.31). Overall certainty of evidence was moderate. Risk of bias assessment using the Cochrane RoB 2.0 tool showed 30% of comparisons at low risk, 56% with some concerns (primarily due to open label design), and 15% at high risk, mainly related to crossover and outcome assessment. Conclusions: PFS is a moderate and clinically acceptable surrogate for OS in metastatic TNBC, particularly for immune checkpoint inhibitors and platinum based regimens. However, surrogate validity varies substantially by treatment class, underscoring the need for confirmatory OS data especially for targeted therapies with novel mechanisms.