Background: The lapatinib-taxane combination led to shorter PFS than trastuzumab-taxane in HER2+ metastatic breast cancer. We investigated the prognostic and predictive effects of pretreatment serum HER2, CA IX, and TIMP-1. Methods: MA.31 accrued 652 patients; 537 (82%) were centrally-confirmed HER2+. Biomarkers were categorized for univariate and multivariate predictive investigations with a median cut-point, ULN cut-points (15 ng/ml for HER2; 506 pg/ml for CAIX; 454 pg/ml for TIMP-1), and custom cut-points (30 and 100 ng/ml for HER2). Stratified step-wise forward Cox multivariate analysis used continuous and categorical biomarkers for PFS in the ITT and central HER2+ populations; central HER2+ biomarker results are shown. Results: Serum was banked for 472 (72%) of 652 patients. Higher serum HER2 (>median; >15; >30; or >100 ng/ml; p=0.05-0.002); higher CAIX (>median; >506 pg/ml; p=0.02; p=0.001); and higher TIMP-1 (>median; >454 pg/ml; p=0.001; p=0.02) had worse univariate PFS. In multivariate analysis, higher continuous TIMP-1 was associated with significantly worse PFS: HR=1.001 (95% CI=1.000-1.002; p=0.004). Continuous serum HER2 and CAIX were not significantly associated with PFS. HER2 of 15 ng/ml or higher had shorter PFS (p=0.02); higher categorical CAIX had worse PFS (p=0.01-0.08). The interaction terms of HER2, CAIX, and TIMP-1 with treatment were not significant. Multivariate PFS categorical serum results (Table). Conclusions: Higher levels of serum TIMP-1, CAIX, and HER2 were significant prognostic biomarkers of shorter PFS. No serum biomarker was predictive of differential response to lapatinib vs. trastuzumab. Evaluation of TIMP-1 and CAIX targeted therapy in addition to HER2 targeted therapy is warranted in patients with elevated serum levels of these biomarkers.

*DH and JH contributed equally.