KEYNOTE-119

Background

Phase III, open-label RCT. N=622 with relapsed/refractory metastatic TNBC after 1–2 prior lines of chemotherapy for metastatic disease. PD-L1 by 22C3; hierarchical OS testing in CPS ≥10, CPS ≥1, and ITT.

Interventions and follow up

Arm A: Pembrolizumab 200mg q3wk
Arm B: Physician's choice chemotherapy (gemcitabine, capecitabine, or vinorelbine)
Primary endpoint: Overall survival (OS)
mFollow up: 19.3 months

Results

OS (CPS ≥10): 12.7 vs 11.6mo, HR 0.78 — hierarchical boundary not met
OS (CPS ≥1): 10.7 vs 10.2mo, HR 0.86 — NS
OS (ITT): 9.9 vs 10.8mo, HR 0.97 — NS
ORR (CPS ≥10): 26.3% vs 11.5%

Adverse events

Overall (grade ≥3, pembro vs chemo): 14.3% vs 36.5%
Immune-mediated (grade ≥3): 4.3% vs 0.5%
Tolerability: Pembrolizumab substantially better tolerated than chemotherapy

Conclusions

Pembrolizumab monotherapy did not improve OS vs chemotherapy in 2nd/3rd-line mTNBC in any PD-L1 subgroup. Definitively negative — pembrolizumab alone is not effective salvage monotherapy in unselected mTNBC.

Key Limitations

Monotherapy in heavily pretreated, unselected mTNBC was likely an underpowered approach; effective IO typically requires chemotherapy combination or earlier-line use. CPS ≥10 numeric ORR advantage (26.3% vs 11.5%) did not translate to OS. PD-L1 CPS enriches response but does not guarantee late-line OS benefit.

Clinical Context

Important negative trial: single-agent pembrolizumab is insufficient in 2nd/3rd-line mTNBC. Contrast with KEYNOTE-355 (1L pembrolizumab + chemo, CPS ≥10, positive OS) shows IO needs earlier use and/or chemo combination. Sacituzumab govitecan (ASCENT) remains standard 2nd/3rd-line. Pembrolizumab not approved in salvage mTNBC.

References