Oncology and Pharmacogenomics Insights in Polycystic Ovary Syndrome: An Integrative Analysis

Polycystic ovary syndrome (PCOS) is a heterogeneous endocrine disorder characterized by hyperandrogenism, ovulatory dysfunction, and polycystic ovaries. Epidemiological findings revealed that women with PCOS are prone to develop certain cancer types due to their shared metabolic and endocrine abnormalities. However, the mechanism that relates PCOS and oncogenesis has not been addressed. Herein, in this review article the genomic status, transcriptional and protein profiles of 264 strongly PCOS related genes (PRG) were evaluated in endometrial cancer (EC), ovarian cancer (OV) and breast cancer (BC) exploring oncogenic databases. The genomic alterations of PRG were significantly higher when compared with a set of non-diseases genes in all cancer types. PTEN had the highest number of mutations in EC, TP53, in OC, and FSHR, in BC. Based on clinical data, women older than 50 years and Black or African American females carried the highest ratio of genomic alterations among all cancer types. The most altered signaling pathways were p53 in EC and OC, while Fc epsilon RI in BC. After evaluating PRG in normal and cancer tissue, downregulation of the differentially expressed genes was a common feature. Less than 30 proteins were up and downregulated in all cancer contexts. We identified 36 highly altered genes, among them 10 were shared between the three cancer types analyzed, which are involved in the cell proliferation regulation, response to hormone and to endogenous stimulus. Despite limited PCOS pharmacogenomics studies, 10 SNPs are reported to be associated with drug response. All were missense mutations, except for rs8111699, an intronic variant characterized as a regulatory element and presumably binding site for transcription factors. In conclusion, in silico analysis revealed key genes that might participate in PCOS and oncogenesis, which could aid in early cancer diagnosis. Pharmacogenomics efforts have implicated SNPs in drug response, yet still remain to be found.

Mutations codified with unknown significance were not considered. For PRG and PNRG gene set statistics normalization considered numbers of genes examined.
The clinical annotations selected were diagnosis age and race. For clinical data comparison within each cancer type normalization contemplated the number of patients in each category (ratio). The ratio and percentage of genetic alterations per age and race group were calculated, with this data the ranking of genes and categories with the greatest number of all genetic alterations were determined. Regarding age: 45, 47, 299 women aged less or 50 years and 459, 143, 695 age more than 50 years old in endometrial, ovarian and breast cancer respectively. In the race group: 4, 2, 1 individuals were American Indian or Alaska Native; 20, 7, 59 were Asian; 101, 19, 162 were Black or African American; 342, 157, 687 were White in endometrial, ovarian and breast cancer respectively. Only in endometrial cancer the 9 individuals were Native Hawaiian or Other Pacific Islander.
Kruskal Wallis-test with Bonferroni correction was performed in python to detect significant differences of frequency in all genetic alterations among gene set and race categories, while Mann-Whitney U test for age groups statistics.

KEGG Pathways enrichment analysis of associated PCOS genes: David
Bioinformatics Resources website (https://david.ncifcrf.gov/summary.jsp) was used to obtain unified data from KEGG (21,22). The enrichment analysis of signaling pathways was carried out in the 264 PRG genes, considering terms with a significant FDR < 0.01.
Then, to identify the most perturbated signaling pathways in each cancer type. The number genetic alterations of the genes in each signaling pathway were added and normalization took into account the number of genes in the pathways and the individuals in each cancer type.

Protein expression analysis
Protein profiling in normal and human tumor tissue based on immunohistochemisty using tissue microarrays is available in Human Protein Atlas (HPA, https://www.proteinatlas.org/) portal (24,25). Therefore comparisons among protein expression levels (high, medium, low and non-detected) of the 264 PRG between normal and cancer tissues were performed. Protein expression level of normal tissue were taken from endometrium glandular cells, ovarian stroma cell and breast glandular cells.

Characterization of overlapped genes
To investigate if the genes commonly altered in at least 2 of the cancer types are cataloged as oncogenes and/or tumor suppressor genes, the Network of Cancer Genes (NCG6.0) database was examined. It has a list of 711 known cancer genes with their respective annotations (26). General functions of the 31 genes obtained in this study were investigated using g:profiler with the settings previously mentioned.