Endocrine biomarkers in low-grade serous ovarian cancers (LGSC) and serous ovarian tumors of low malignant potential (LMP).

Abstract

e18045 Background: LGSC and serous ovarian LMP tumors are rare entities along a disease continuum. Because they are relatively chemotherapy resistant, there is interest in targeted treatment, including endocrine therapy. In studies of endocrine agents for treatment of recurrent ovarian cancer, tumor estrogen receptor (ER) status alone does not effectively predict outcomes. The objective of this study is to utilize protein and gene expression from ER and downstream targets to identify signaling components that will act as prognostic and predictive biomarkers. Methods: A single institution tumor registry was queried for patients with LGSC or serous LMP tumors between 1993-2016. Formalin-fixed, paraffin embedded (FFPE) tissue was obtained. Clinicopathologic data was collected. Immunohistochemistry (IHC) was performed and scored as dichotomous (+/-) and continuous (H-score). NanoString nCounter platform was used for gene expression of a predesigned endocrine biomarker panel. Wilcoxon Rank Sum was used to compare mRNA and protein expression. Survival analyses were performed using Cox proportional hazard (PH) method with elastic net to fit the prediction model for the gene panel. Results: Tissue was analyzed from 23 LGSC, 71 serous LMP tumors, and 2 recurrences. At diagnosis, 86% of patients with LMP tumors had stage I disease vs 78% of patients with LGSC had stage III/IV disease. There were 7 (10%) LMP and 13 (56.5%) LGSC recurrences. Median initial PFS was 76 and 37 months and median OS was 88 and 82 months for LMP and LGSC, respectively. Fourteen received endocrine therapy; nine for treatment of recurrent LGSC. Median therapy was 21 months, with 3 complete responses, 2 partial responses and 1 stable disease. All tumors were ER+ (median H-score 210) and 89% were PR+ (median H-score 130). ER H-scores and ESR1 expression levels were not significantly different between LGSC and LMP tumors (p = 0.1942, p = 0.0893). PR H-scores were higher in LMP tumors (p < 0.00001) but PGR expression was not significantly different (p = 0.6581). In the Cox PH analysis, ER H-score was a significant predictor of both PFS and OS (p < 0.001, p = 0.004). The Cox model identified independent sets of gene expression associated with PFS and OS in both the LGSC and LMP cohorts (c-indices 0.679, 0.570, 0.626, 0.689). Conclusions: Endocrine therapy is an active area of interest in the treatment of ovarian neoplasms. ER H-score and gene expression of downstream ER targets could improve upon the use of hormone receptors to determine risk of recurrence and guide use of adjuvant endocrine therapy.