Background:
The presence of TILs has been correlated with clinical outcomes and response to therapy in breast cancer. However, evaluation of TILs in breast cancer has largely been based on pathologic examination of tumor samples. Here, we report the relationship between invasive breast cancer locoregional recurrence (LRR) and the presence of TILs estimated by transcriptomic analysis with the deconvolution algorithm CIBERSORT.
Methods:
Patients were identified from an IRB-approved prospective tissue collection protocol at one academic institution and two community hospitals. 526 primary breast tumor samples were identified and gene expression profiling was assessed with high density Affymetrix microarray chips. Proportions of 22 different TIL types in samples were inferred based on the CIBERSORT algorithm, which uses gene expression data to estimate TIL presence. TIL presence was determined by dichotomization at the level of the first quartile among all samples (>Q1=TIL presence). Patient characteristics and clinical outcomes were obtained by chart review. Time to event analysis was performed using Kaplan Meier (KM) estimates and the log-rank test. Associations between patient factors, tumor factors, TIL presence, and LRR were explored with univariable (UVA) and multivariable (MVA) analyses. Factors significant on UVA (p<0.10) were included on MVA. P<0.05 was considered statistically significant on MVA.
Results:
526 women with invasive breast cancer and available genomic profiling were retrospectively identified for analysis. Median age at diagnosis was 58 years. 70% of tumors were Stage I-II. 69% were luminal subtypes and 17% were triple negative. 37% received mastectomy, 25% received mastectomy + radiation, and 32% received breast conserving therapy. 64% received chemotherapy, and 62% received hormonal therapy. Median follow-up was 74.4 months. There were 61 LRRs. We found significant differences in time to LRR when comparing presence vs. no presence of resting memory CD4+ T-cells (RMCD4+) (p=0.01), activated natural killer cells (ANK) (p=0.003), and neutrophils (PMNs) (p=0.03). On UVA, factors associated with LRR were patient age at diagnosis (p=0.009), pathologic T stage (p=0.045), Estrogen receptor status (p=0.03), biologic subtype (p=0.01), lymphovascular invasion (LVI) (p=0.018), positive margins (p<0.0001), receipt of hormonal therapy (0.014), and presence of tumor infiltrating RMCD4+ (p=0.012), ANK (p=0.0004), and PMNs (p=0.033). On MVA, factors remaining significant were LVI (HR 2.16 CI 1.13-4.13, p=0.011), positive margins (HR 4.36 CI 1.57-12.11, p=0.018), receipt of hormonal therapy (HR 0.31 CI 0.12-0.77, p=0.042), and presence of RMCD4+ (HR 0.48 CI 0.26-0.88, p=0.017), ANK (HR 0.43 CI 0.23-0.83, p=0.012), and PMNs (HR 2.15 CI 1.02-4.53, p=0.043).
Conclusion:
In this study of 526 women with invasive breast cancer, we identified that enrichment of certain TILs is associated with LRR. These results suggest genomic-based assays of TIL presence may be useful to predict LRR in invasive breast cancer.
Citation Format: Liveringhouse CL, Grass GD, Figura NB, Mills MN, Purcell JD, Rosensweig SR, Blumencranz PW, Allen KG, Ahmed KA, Harrison LB, Torres-Roca JF, Robinson TJ, Diaz R. Locoregional recurrence in invasive breast cancer and association with tumor infiltrating leukocyte (TIL) presence [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P4-08-15.
