Background: The IO score is a is a 27-gene signature developed to classify the tumor immune microenvironment derived from the 101-gene TNBCtype genomic classifications. The IO score predicts clinical outcome following immune checkpoint inhibitor therapy in NSCLC and bladder cancer, and recently was shown to predict benefit by pCR of atezolizumab plus CT over neoadjuvant CT alone in early stage TNBC (NeoTRIPaPDL1 trial). The IO score has not yet been evaluated in mTNBC or with pembro in breast cancer. Methods: We report preliminary associations of IO score with response from a phase Ib trial (NCT02734290). mTNBC subjects received 1st/2nd line pembro (200mg IV q3wk) plus investigator’s choice paclitaxel (80mg/m2 IV q1wk, n = 15) or capecitabine (2000mg PO BID x 7d, q2wk, n = 14). Baseline (n = 23) and on-treatment (at wk 6, n = 10) biopsies were analyzed for IO score and genomic subtype by RNA exome sequencing. Objective response rate (ORR, partial or complete response, 12 weeks) and survival was determined among response-evaluable subjects (n= 21). Tumor PD-L1 was assessed by IHC (combined positive score, CPS > 10%). The IO signature was analyzed as a binary classifier (IO+/IO-) and as a continuous variable (IO score). Results: 39% of evaluable subjects were IO+ (n =9/23). IO+ was associated with improved clinical outcome, including ORR (IO+ 43%, IO- 29%), median progression free survival (mPFS, IO+ 138d, IO- 79d), and median overall survival (mOS, IO+ 687d, IO- 305d). IO+/IO- classification and IO scores were stable across serial biopsies (Cohen’s kappa = 0.74, r = 0.84). IO score was not strongly correlated with PD-L1 CPS (r = 0.27) or sTILs (r =.09). PD-L1-/IO+ tumors constituted 31% (n = 5/16) of PD-L1- cases and exhibited favorable outcome (ORR 40%, mPFS 162d, mOS 556d). IO score and ORR varied across TNBCtype classifications (BL1 subtype: 50% ORR, 66% IO+; BL2 subtype: 0% ORR, 66% IO+; LAR subtype: 50% ORR, 0% IO+, MSL subtype: 33% ORR, 60% IO+). Conclusions: IO score is associated with favorable outcome following pembro + CT, and may identify PD-L1-negative cases that respond to pembro + CT. Further investigation in larger datasets is warranted to ascertain the clinical utility of IO score in this setting. Funding: Drug support and funding provided by Merck Sharpe & Dohme as part of the Merck Investigator Studies Program. Clinical trial information: NCT02734290.