The purpose of the present study was to determine the optimal threshold for stromal tumor-infiltrating lymphocytes (TILs) and investigate its predictive and prognostic value in HER2-positive breast cancer treated with trastuzumab-based neoadjuvant chemotherapy (NAC). Levels of stromal TILs were evaluated using hematoxylin and eosin-stained sections of core biopsies from 116 patients. We investigated the correlation between stromal TILs and pathological response to identify its optimal threshold. Using receiver operating characteristic curve analysis, a 30 % threshold best discriminated pathological complete response (pCR) from non-pCR subgroups (P < 0.001). Lymphocyte-rich breast cancer (LRBC) was defined as having ≥30 % stromal TILs level, and was used for analysis. For analyses of predictive factors, multivariate analysis indicated that LRBC was a strong predictor of pCR with an odds ratio of 5.23 (P < 0.001). Negative hormone receptor (HR) status was also significantly associated with pCR (P = 0.028). LRBC significantly predicted pCR in both HR-positive and HR-negative tumors (P = 0.016 and 0.006, respectively). For survival analyses, LRBC was the only independent predictor of improved event-free survival (EFS) among baseline clinicopathological factors in multivariate analysis (P = 0.012). When pathological response was included, both LRBC and pCR were independent predictors of better EFS (P = 0.040 and 0.045, respectively). LRBC significantly predicted longer EFS in the non-pCR subgroup (P = 0.018), whereas LRBC was not significantly associated with EFS in the pCR subgroup (P = 0.825). A 30 % threshold for stromal TILs optimally identified response to trastuzumab-based NAC in HER2-positive breast cancer; its predictive and prognostic value was also validated in our study.
Keywords: Breast cancer; HER2; Neoadjuvant chemotherapy; Predictive factors; Prognostic factors; Tumor-infiltrating lymphocytes.