Abstract Introduction: The use of archived benign breast samples (BBS) for breast cancer risk research is limited by the fact that menstrual luteal phase and menopausal status is either unknown or inaccurate by self-report. As a result, physiologic variation in gene and protein expression cannot be accounted for. We previously showed that morphologic features (PMID: 12481017) and progesterone (P)-related gene expression (PMID: 23512947) in breast specimens can aid in luteal phase classification. Here we report a validation of the concept, in a larger sample size, regarding gene expression, menstrual-specific morphology factors and a combination of both. Methods: BBS were obtained from the contralateral unaffected breast (CUB) of unilateral breast cancer patients and reduction mammoplasty (RM) controls of premenopausal women (n = 81) under IRB approved protocol. Gene expression analysis was performed with 100 ng by qRT-PCR of 21 target genes. Serum concentration of estradiol (E), P, and follicular stimulating hormone (FSH) were measured by enzyme immunoassay and IMMULITE. Subjects were classified using hormone levels as postmenopausal (E< 30 pg/ml and FSH> 30 IU/L) or luteal phase (E>60pg/ml and P>3ng/ml); premenopausal women who did not meet luteal phase hormonal criteria were called non-luteal. A single study pathologist evaluated samples from premenopausal subjects, masked to the menstrual phase of subjects, using histo-morphologic features to categorize subjects as luteal or non-luteal. Gene expression and morphology were evaluated individually and together, through multivariate analysis, for luteal-phase stratification. Receiver operator curve (ROC) analysis was used to assess the accuracy (AUC) for prediction of luteal phase using gene expression, morphology and a combination of both. Results: Our study population, matched by age, included 81 premenopausal subjects (CUB=51, RM=30) with mean age of 42. For luteal phase classification, we identified P-correlated genes; TNFSF11, DI02, and MYBPC1 genes were significantly correlated with serum progesterone levels. ROC analysis results using individual genes, average morphology score and multivariate model is in the table below. The results are reported with the area under the curve (AUC) value along with the confidence interval. The multivariate model consists of the three genes with the average morphology score. Conclusions: We provided preliminary validation for three P-related genes as classifiers of luteal phase status. We find that histomorphologic classification of menstrual phase is more prominent in the RM samples. These findings along with our validation data, for the first time, provide RNA-based classifiers for menstrual phase of BBS, and offer the potential of increasing the robustness and validity of biomarker research by accounting for important physiologic variation in the hormonal milieu of the donor. CUB (n = 51) AUC +/- ciRM (n = 30) AUC +/- ciTNFS11 gene expression0.84 (0.74 – 0.95)0.76 (0.54 – 0.97)DI02 gene expression0.86 (0.75 – 0.96)0.72 (0.49 – 0.96)MYBPC1 gene expression0.80 (0.68 – 0.92)0.67 (0.45 – 0.89)Average morphology score0.67 (0.54 – 0.81)0.87 (0.76 – 0.98)Multivariate model0.89 (0.79 – 0.98)0.948 (0.87 -1.00) Citation Format: Omid Hosseini, Irene Helenowski, Oukseub Lee, Hui Zhang, Jun Wang, Luis Blanco, Seema A Khan. Menstrual phase classification of benign breast tissue using hormone-regulated gene expression and morphology [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P2-11-03.
Read full abstract