Abstract P4-04-17: Evaluation of a prognostic immune gene expression signature in different breast cancer molecular subtypes

Abstract

Listen

Translate article

Abstract Background: Cancer immunotherapy is increasingly important in many malignancies. Breast cancers with greater numbers of tumor-infiltrating lymphocytes (TILs) exhibit better responses to neoadjuvant therapy. Additionally, increased CD8+ TILs in the breast tumor microenvironment have better outcomes. The Total Cancer Care (TCC) biorepository at Moffitt Cancer Center allows for the collection of all clinical patient data and tissue for analysis of novel biomarkers with prognostic and predictive abilities. Using TCC consented patients, we sought to determine if an immune based signature score derived from gene expression profile data would have a significant prognostic ability, and determine if this signature is differentially expressed within different PAM50 molecular subtypes. Methods: Tumors were arrayed on Affymetrix HuRSTA-2a520709 GeneChips (Affymetrix, Santa Clara, CA). The chips contain ∼60,000 probe sets representing ∼25,037 unique genes (Affymetrix HuRSTA-2a520709). Gene expression data was normalized using iterative rank-order normalization and de-batched using Principal Component Analysis (PCA). The immune score for each tumor was derived using the first component from a PCA model based on all immune related genes in breast primary tumor samples. Mean immune score between molecular subtypes was compared using one way ANOVA. Results: A total sample of 310 non-metastatic patients with full clinical and gene expression data were available for the analysis. Mean age was 55. Staging breakdown was Stage I: 26%, stage II: 46%, and stage III: 17%. For tumor grades, 10% were well differentiated, 45% poorly or undifferentiated, 41% moderately differentiated, and 3% missing. Mean follow up for overall survival (OS) was 6.84 years and 6.10 years for recurrence free survival (RFS). For IHC subtypes 62% were hormone receptor positive, 18% triple negative, and 20% HER2 positive. The mean immune score across the whole population was 0.05 (range -2.22 to 2.27). Mean scores were higher in PAM50 basal (.379, 95% CI .143 to .615) and HER2 enriched (.316, 95% CI .143 to .6155) vs. luminal A (-.215, 95% CI -.34 to -.09)/B (-.168, 95% CI -.332 to -.004) subtypes. Immune score was not prognostic for OS hazard ratio (HR) 1.11 (95% CI: 0.91-1.36) p=.30, but was for RFS HR=0.71 (95% CI: 0.53-0.96) p=.02 in a multivariate analysis controlling for age, race, stage, grade, adjuvant treatment, and molecular subtype. Conclusions: The mean immune score is higher in basal and HER2 enriched tumors, suggesting we can potentially identify immune infiltrate enriched patients for immunotherapy trials using existing gene expression profiling data routinely obtained on breast cancer primary tumors. Correlation between score and lymphocyte infiltrate on histology is ongoing and will be presented. A higher score also appears to be an independent predictor of RFS which will need to be confirmed in additional independent well annotated datasets. Citation Format: Sheen MA, Connor CG, Berglund AE, Prabhakaran S, Rizk VT, Soliman HH. Evaluation of a prognostic immune gene expression signature in different breast cancer molecular subtypes. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P4-04-17.