Beyond TILs: Predictors of pathologic complete response (pCR) in triple-negative breast cancer (TNBC) patients with moderate tumor-infiltrating lymphocytes (TIL) receiving neoadjuvant therapy.

Abstract

572 Background: Increased TIL in TNBC is associated with higher rates of pCR. High TIL is also associated with improved disease free survival and overall survival. The aim of this study is to identify data cut-points of pre-treatment low, moderate and high TIL count based on pCR and to identify clinical and pathological predictors of pCR in patients with moderate TIL. Methods: We evaluated the relationship between pCR and TIL in 180 patients with stage I-III TNBC enrolled in the ARTEMIS trial (NCT02276443). Recursive portioning was used to identify cut-points. Clinical and pathological variables such as age at diagnosis, stage, race, histology as well as Ki-67, vimentin, and androgen receptor (AR) by immunohistochemistry, were evaluated in pts with moderate TIL. A multivariable logistic regression model identified variables independently, significantly associated with pCR. Results: Four TIL groups were identified with pCR rates of 23%, 31%, 48% and 78% respectively (p < 0.0001) (Table A). In the two combined moderate TIL groups, 90 (97%) pts were evaluable for the multivariate model. Stage I-II disease, high Ki-67 and low AR were associated with increased probability of pCR (Table B). The multivariable logistic regression model area under the ROC curve was 0.78 (95% CI=0.68-0.88; p<0.0001). A model of computed risk score [ Stage I-II (score 2)+Ki-67≥ 50% (score 1)+AR<10% (score 1)] predicted a probability of 67% for pCR when all three variables were favorable (Table). Conclusions: Four TIL groups were identified. In pts with moderate TIL levels, early stage disease, high Ki-67 and low AR were associated with increased probability of pCR with neoadjuvant therapy. [Table: see text]