Little is known about the efficacy of HER2-targeted therapy in patients with breast cancer showing different HER2-pathway dependence and immune phenotypes. Herein, we report a NeoALTTO exploratory analysis evaluating the clinical value of 22 types of tumor-infiltrating immune cells by CIBERSORT and 5 immune-related metagenes in the overall patient population, and in subgroups defined by the TRAR classifier as HER2-addicted (TRAR-low) or not (TRAR-high).
Association of baseline TRAR, immune-related metagenes, and CIBERSORT data with pathologic complete response (pCR) and event-free survival (EFS) were assessed using logistic and Cox regression models. Corrections for multiple testing were performed by the Bonferroni method.
A total of 226 patients were analyzed: 80 (35%) achieved a pCR, and 64 (28%) experienced a relapse with a median follow-up of 6.7 (interquartile range 6.1–6.8) years; 108 cases were classified as TRAR-low, and 118 TRAR-high. Overall, γδ T-cell fraction [OR = 2.69; 95% confidence interval (CI), 1.40–5.18], and no immune-related metagenes were predictive of pCR. Notably, lymphocyte-specific kinase (LCK) predicted pCR to combination (OR = 2.53; 95% CI, 1.12–5.69), but not to single-agent trastuzumab or lapatinib [OR = 0.74; 95% CI, 0.45–1.22 (Pinteraction = 0.01)]. Integrating LCK with γδ T cells in a multivariate model added to the discriminatory capability of clinical and molecular variables with a shift in AUC from 0.80 (95% CI, 0.74–0.86) to 0.83 (95% CI, 0.78–0.89). In TRAR-low cases, activated mast cells, IFN and MHCII were reduced, and STAT1, HCK1, and γδ T cells were associated with pCR. STAT1 was broadly associated with improved EFS regardless of pCR, and nodal status in overall (HR = 0.68; 95% CI, 0.49–0.94) and in TRAR-low cases (HR = 0.50; 95% CI, 0.30–0.86).
Immuno-phenotyping holds the promise to complement current predictive models in HER2-positive breast cancer and to assist in new therapeutic development.