1031 Background: In MA.31 the trastuzumab-taxane combination led to longer PFS than lapatinib-taxane in HER2+ metastatic breast cancer (MBC). In MA.31 we previously reported the prognostic/predictive utility of pretreatment serum PD-L1 (SABCS 2018, PD3-10) and serum activin A (SABCS 2016, P6-07-06) separately; here we evaluate them combined. Methods: MA.31 accrued 652 centrally and/or locally-identified HER2-positive patients, and pretreatment serum was available for 382 patients (184 in trastuzumab arm, 198 in lapatinib arm). The ELLA immunoassay platform (ProteinSimple, San Jose, CA) was used to quantitate serum PD-L1, and ELISA for activin A (R&D Systems, Minneapolis, MN). Results: In correlation analysis, pretreatment serum PD-L1 was moderately correlated with serum activin A (r = 0.21, p = 0.004). In univariate analysis for OS, the combination of higher serum PD-L1 and higher serum activin A (median cutpoints) (vs. both low) was significant for shorter OS in the trastuzumab arm (HR 6.62, p=0.0005) and in the lapatinib arm (HR 3.25, p=0.0003)(table). In multivariate analysis for OS (17 covariates included), elevated serum activin A/PD-L1 combination remained the most significant independent covariate in the trastuzumab arm (HR 12.40, p=0.001), and in the lapatinib arm (HR 5.2, p=0.0001). Conclusions: In the CCTG MA.31 trial, elevated pretreatment serum activin A (TGF-B superfamily) and PD-L1 was associated with a shorter OS to HER2-targeted treatment. Multiple mechanisms, including immune evasion, may decrease the effectiveness of HER2-targeted therapy. Elevated serum activin A and PD-L1 may identify HER2-positive MBC patients who would benefit from inhibitors of the HER2, PD-1, and activin A pathways. [Table: see text]