Abstract Background: In the phase 3 KEYNOTE-355 trial (NCT02819518), pembrolizumab (pembro) combined with chemotherapy (chemo) showed statistically significant improvements in OS and PFS compared to placebo + chemo in patients with previously untreated locally recurrent inoperable or metastatic TNBC whose tumors expressed PD-L1 with a combined positive score (CPS) ≥10. There were no statistically significant differences between treatment groups in the CPS ≥1 population, and statistical significance was not tested in the ITT population due to the prespecified testing strategy. Here, we compare outcomes in subgroups of patients by additional CPS cut-offs to the primary results of KEYNOTE-355. Methods: 847 patients with measurable disease per RECIST v1.1, ECOG PS 0-1, and ≥6 month disease-free interval were randomized 2:1 to pembro + chemo (nab-paclitaxel 100 mg/m2 days 1, 8, and 15 every 28 days; paclitaxel 90 mg/m2 days 1, 8, and 15 every 28 days; or gemcitabine 1000 mg/m2 + carboplatin AUC 2 days 1 and 8 every 21 days) or placebo + chemo for up to 35 administrations of pembro/placebo or until progression/intolerable toxicity. Patients were stratified by chemo type (taxane or gemcitabine-carboplatin), PD-L1 status (CPS ≥1 or <1), and prior neoadjuvant/adjuvant treatment with same-class chemo (yes or no). Dual primary endpoints were OS and PFS (RECIST v1.1, blinded independent central review) in patients with PD-L1-positive tumors (CPS ≥10 and ≥1) and in the ITT population. Hazard ratios (HRs) and 95% CIs in the additional CPS subgroups were based on an unstratified Cox model. Results: At the time of the final analysis (data cutoff, June 15, 2021), the median time from randomization to data cutoff was 44 months. Baseline characteristics of the CPS 1-9, 10-19, and ≥20 subgroups were generally similar to those of the ITT population. In the primary analyses, the HRs (95% CI) for OS were 0.73 (0.55-0.95) in the CPS ≥10 subgroup, 0.86 (0.72-1.04) in the CPS ≥1 subgroup, and 0.89 (0.76-1.05) in the ITT population; HRs (95% CI) for PFS were 0.66 (0.50-0.88), 0.75 (0.62-0.91), and 0.82 (0.70-0.98), respectively. Efficacy data are presented for the additional CPS subgroups in the Table. For OS, results in the CPS 1-9 subgroup showed comparable efficacy for pembro + chemo and placebo + chemo; however, results in the CPS 10-19 and CPS ≥20 subgroups showed a similar treatment benefit with the addition of pembro. Results for PFS were generally consistent with those observed for OS. Conclusions: These results provide support that CPS ≥10 is a reasonable cut-off to define the population of patients with metastatic TNBC expected to derive treatment benefit from pembro + chemo. Table. Efficacy in additional subgroups of patients by PD-L1 CPSCPS subgroupTreatmentMedian OS, mo (95% CI)OS HR (95% CI)*Median PFS, mo (95% CI)PFS HR (95% CI)*1-9†Pembro + Chemo. n = 20513.9. (12.2-16.6)1.09 (0.85-1.40)5.7. (5.4-7.4)0.85 (0.65-1.11)Placebo + Chemo n = 10815.5. (12.4-17.6)5.6. (5.3-7.6)10-19†Pembro + Chemo. n = 8020.3. (15.2-29.0)0.71 (0.46-1.09)9.9. (7.4-11.8)0.70 (0.44-1.09)Placebo + Chemo n = 3917.6. (10.3-25.8)7.6. (5.3-9.7)≥20‡Pembro + Chemo. n = 14024.0 (19.0-28.3)0.72 (0.51-1.01)9.2 (7.6-11.8)0.62 (0.44-0.88)Placebo + Chemo n = 6415.6. (12.3-20.8)5.4 (3.9-7.2)*Analyses based on an unstratified Cox model. †Post-hoc analyses. ‡Prespecified exploratory analyses. Citation Format: Javier Cortes, David W. Cescon, Hope S. Rugo, Zbigniew Nowecki, Seock-Ah Im, Mastura Md Yusof, Carlos Gallardo, Oleg Lipatov, Carlos Henrique Barrios, Jose Perez-Garcia, Hiroji Iwata, Norikazu Masuda, Marco Torregroza Otero, Erhan Gokmen, Sherene Loi, Zifang Guo, Xuan Zhou, Vassiliki Karantza, Wilbur Pan, Peter Schmid. Final results of KEYNOTE-355: Randomized, double-blind, phase 3 study of pembrolizumab + chemotherapy vs placebo + chemotherapy for previously untreated locally recurrent inoperable or metastatic triple-negative breast cancer [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr GS1-02.