Abstract P4-04-10: Molecular features of dormancy in ER+ breast cancers

Background: Late recurrence (emergence from dormancy) is characteristic of ER+ breast cancers. Despite adjuvant endocrine therapy, many breast cancers recur decades after their initial diagnosis and treatment. Why this occurs is poorly understood. Methods: We studied 2 independent datasets of endocrine treated, ER+ breast cancers with up to 20 years follow-up. The 1st comprised matched samples from the primary tumor pretreatment at diagnosis and the first recurrence after or during adjuvant endocrine therapy (all FFPE). The 2nd dataset comprised pretreatment biopsies only (all snap frozen). For both datasets, high quality RNA was amplified, labelled, and subjected to transcriptome analysis using the Affymetrix technology (U133 Plus 2.0). Low quality data were identified using 9simpleaffy9 and 9ffpe9, and removed; all tools were from the R package unless otherwise noted. Remaining data were normalized using 9frma9. Genes differentially expressed between early (≤3 years) and late (≥ 5 years) were selected using limma. Unsupervised hierarchical clustering and PCA explored the structure of the data. A similar molecular analysis was done on the 2 nd dataset. A classification scheme that robustly separated early from late recurrences was validated in an independent public dataset of comparable patients, array platform, and frozen tissues. We also explored features in pretreatment samples that predetermined response duration. Results: Genes that separated pretreatment specimens by recurrence time did not separate posttreatment specimens. Specimens did not cluster in patient pairs or by site of recurrence. 8245 genes were differentially expressed between early and late recurrences in the FFPE samples, while 2400 genes were significantly different in the same comparison in the frozen samples. Initial pathway analysis was done on each dataset independently using IPA (Ingenuity® Systems, www.ingenuity.com). 70 canonical pathways were identified in common between the two datasets (pretreatment). We then looked for genes regulated in both datasets (ignores FFPE and frozen tissue as source). There were 279 genes in common that differentially regulated in the same direction (upregulated; downregulated). IPA analysis of these genes identified 49 canonical pathways. We also explored the differentially expressed gene sets using 9GSEA9 (www.software.broadinstitute.org/gsea/index.jsp). Pathways consistently associated with early vs. late recurrence include integrin signaling, the unfolded protein response, endoplasmic reticulum stress, actin-based motility, and estrogen biosynthesis. Conclusion: Analysis of pretreatment tumors can predict early recurrences from those that will remain dormant and recur much later. Recurrent tumors exhibit a remodeled molecular landscape that likely reflects the effects of treatments and/or a recreation of a niche with potentially common features at the site of recurrence. Changes in molecular signaling associated with duration of recurrence are consistent with our experimental model studies in vitro implicating UPR signaling as a major integrator of response to endocrine therapy and duration of survival. Additional data sets are being arrayed and more detailed molecular signaling studies are in progress. Citation Format: Clarke R, Dixon MJ, Jin L, Turnbull A, Hu R, Zwart A, Wang Y, Xuan J, Sengupta S, Renshaw L, Sims A, Liu MC. Molecular features of dormancy in ER+ breast cancers [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P4-04-10.