Tumor-infiltrating lymphocytes differ across MammaPrint® classifications in breast cancer

Abstract MammaPrint® refines risk stratification in early oestrogen receptor-positive, HER2-negative breast cancer, evolving from a binary to a four-tier classification (UltraLow, Low, High-Risk1, High-Risk2). The relationship between routine histopathological features, immune infiltration and genomic risk within this framework remains incompletely characterised in luminal disease. We retrospectively analysed 492 luminal breast carcinomas with available MammaPrint® results. Clinicopathological variables (including histological subtype, grade, Ki-67, hormone receptor expression, lymphovascular invasion (LVI) and HER2-low status) were recorded. Stromal tumor-infiltrating lymphocytes (TILs) were quantified according to Salgado et al. criteria and spatially categorised as immune-deserted, stromal-restricted, immune-excluded or inflamed patterns. CD4 and CD8 infiltration was assessed by immunohistochemistry on tissue microarrays. Associations with binary and four-tier MammaPrint® categories were examined using multivariable models. High-Risk tumours (41%) were enriched for increased grade, Ki-67 and LVI; and lower PR expression. In High-Risk inflamed spatial patterns were more frequent and median TILs levels were significantly higher compared with Low-Risk (15% vs. 5%, p<0.001). CD4 and CD8 infiltration increased with genomic risk; and CD4 retained a modest but statistically significant association after adjustment for conventional pathological variables. In the four-tier model, UltraLow/Low tumours showed minimal TILs and were enriched for invasive lobular carcinoma, whereas High-Risk1/High-Risk2 displayed progressively higher proliferative and immune features. No association was observed between HER2-low status and genomic risk. Immune infiltration parallels proliferative and genomic risk gradients in luminal breast cancer. These findings indicate that immune descriptors align with the genomic risk continuum, although their independent prognostic contribution beyond established genomic assays requires further evaluation.