CELL INTELLIGENCE · 肿瘤细胞治疗研究

Background/Objectives : Tumor-infiltrating lymphocytes (TILs) are recognized predictors of response to neoadjuvant chemotherapy (NACT) and prognosis in breast cancer, particularly in triple-negative (TN) and HER2-positive subtypes.

However, the additional predictive value of morphological features of the inflammatory infiltrate beyond TIL quantification is not fully established.

We aimed to assess the predictive value of TILs for response to NACT in breast cancer and to determine whether morphological characteristics of the inflammatory infiltrate enhance predictive accuracy. Methods: We analyzed 477 patients with stage II-III breast cancer treated with NACT between 2009 and 2016. Diagnostic core needle biopsies were prospectively re-evaluated. TILs were quantified according to International TILs Working Group recommendations. Morphological features of the infiltrate, including cell composition (lymphocytic vs. plasma cell-rich), heterogeneity, and localization, were evaluated using standardized criteria. Pathologic complete response (pCR) was defined as absence of invasive tumor in the breast and in the axillary lymph nodes (ypT0/Tis ypN0). Univariate and multivariate logistic regression analyses were performed to assess the predictive value of TILs (quantitative and morphological assessment) to achieve pCR for the entire cohort and by surrogate molecular subtype.

Results: A TIL cutoff of >20% was identified as optimal for predicting pCR. High TILs were significantly associated with high-grade tumors, elevatedKi67, HER2-positive and TN subtypes, presence of plasma cells, and intraepithelial and heterogeneous infiltrates. In the overall cohort, TILs > 20% significantly increased the likelihood of pCR (OR 3. 9, 95%IC 2. 5-6. 0, p < 0. 001) and was an independent predictor of pCR.

A combined variable incorporating TIL level and homogeneity improved predictive performance, with homogeneously high TILs emerging as a strong predictor of pCR (OR 5. 521, 95%CI 3. 174-9. 603, p < 0. 01). Plasma cell-rich and intraepithelial infiltrates were also associated with higher pCR rates (respectively, OR 2. 7, 95%CI 1. 5-5. 0, p = 0. 001 and OR 2. 8, 95%CI 1. 6-5. 0, p < 0. 001). Subtype-specific analyses confirmed the predictive value of TILs in TN tumors, but not in HER2-positive ones.

Notably, in luminal B-like tumors, high TILs were the only independent predictor of response (OR 17. 982, 95%CI 3. 115-103. 815, p = 0. 001). Conclusions: TIL assessment on routine H&E-stained biopsies is a robust predictor of response to NACT in breast cancer that is readily available, cost-neutral and does not require additional techniques. Integration of simple morphological features significantly enhances predictive accuracy and may refine treatment stratification, particularly in luminal B-like tumors.