Abstract Aim: To identify pretreatment biomarker predictors of Ki67 response to lapatinib in women with HER2− primary breast cancer. Background: Lapatinib is an EGFR/HER2 inhibitor. Its clinical use is restricted to HER2 overexpressing disease. The MAPLE (Molecular Antiproliferative Predictors of Lapatinib's Effects) presurgical window of opportunity study of lapatinib vs placebo was conducted in women with HER2-amplified (HER2+) or HER2− primary disease. Ki67 (primary end-point) was reduced by a geomean 46% (95%CI 23–63%, p = 0.002) and 27% (95%CI 8–42%, p = 0.008) in HER2+ and HER2− disease, respectively (Leary et al, AACR 2012). We have now assessed whether predictive biomarkers of the antiproliferative response in HER2− disease could be identified. Methods: 121 primary breast cancer patients were randomized (3:1) to 14 days of 1500mg/d lapatinib or placebo before surgery. Biopsies were taken before treatment and at surgery. Ki67 responders were defined as having a >/=50% reduction in Ki67 compared to baseline (Ellis, P et al, Breast Cancer Res Treat 1998, 48, 107). ER, PgR, HER2, EGFR, pAKT, pERK1/2 (nuclear and cytoplasmic), stathmin and apoptosis (TUNEL) were assessed by IHC (+FISH for HER2[all cases]) and scored visually by continuous methods. HER2, HER3, epiregulin (epir), amphiregulin (amphir) and neuregulin (neur) were assessed by qrtPCR. Results: Three of the 121 patients were excluded because of inadequate biopsy material. Ninety-one of the remaining 118 patients received lapatinib: 7/19 (37%) HER2+ cases and 10/72 (14%) HER2− cases were Ki67 responders. Thus while the proportion of Ki67 responders was higher for HER2+ disease there was a similar or higher absolute number of responders with HER2− disease. All of the following relates to patients with HER2− disease. None of the pretreatment levels of ER, PgR, pAKT, pERK1/2, EGFR, epir, amphir or neur were associated with Ki67 response (p > 0.20). However, HER3 (p = 0.01) and HER2 (p = 0.06) mRNA levels were associated with greater Ki67 response. There was a tendency for Ki67 response to be greater with lower baseline Ki67 (p = 0.07). Multivariate analysis showed only HER3 mRNA levels to be independently significant. HER2 and HER3 mRNA levels were highly correlated (rho = 0.67, p < 0.001), a relationship confirmed in 2 other datasets (Wang et al, Breast Cancer Res, 2011, 13, R92; Dunbier et al, submitted). All Ki67 responders were above the median for both HER3 and HER2 expression. Conclusions: Lapatinib is antiproliferative in a subgroup of HER2− tumours. This exploratory analysis indicates that they are characterized by high HER3 expression. The possible importance of high HER2:HER3 heterodimers in predicting this response is supported by the relationship between HER2 and HER3 expression. Further exploration of lapatinib is merited in HER2− cases with high HER3 expression. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr PD07-07.