Precision gynecologic oncology: circulating cell free DNA epigenomic analysis, artificial intelligence and the accurate detection of ovarian cancer

www.nature.com/scientificreports OPEN Precision gynecologic oncology: circulating cell free DNA epigenomic analysis, artificial intelligence and the accurate detection of ovarian cancer Ray O. Bahado‑Singh1, Amin Ibrahim2, Zaid Al‑Wahab1, Buket Aydas3, Uppala Radhakrishna1, Ali Yilmaz1 & Sangeetha Vishweswaraiah2* Ovarian cancer (OC) is the most lethal gynecologic cancer due primarily to its asymptomatic nature and late stage at diagnosis. The development of non-invasive markers is an urgent priority. We report the accurate detection of epithelial OC using Artificial Intelligence (AI) and genome-wide epigenetic analysis of circulating cell free tumor DNA (cfTDNA). In a prospective study, we performed genomewide DNA methylation profiling of cytosine (CpG) markers. Both conventional logistic regression and six AI platforms were used for OC detection. Further, we performed Gene Set Enrichment Analysis (GSEA) analysis to elucidate the molecular pathogenesis of OC. A total of 179,238 CpGs were significantly differentially methylated (FDR p-value < 0.05) genome-wide in OC. High OC diagnostic accuracies were achieved. Conventional logistic regression achieved an area under the ROC curve (AUC) [95% CI] 0.99 [± 0.1] with 95% sensitivity and 100% specificity. Multiple AI platforms each achieved high diagnostic accuracies (AUC = 0.99–1.00). For example, for Deep Learning (DL)/ AI AUC = 1.00, sensitivity = 100% and 88% specificity. In terms of OC pathogenesis: GSEA analysis identified ‘Adipogenesis’ and ‘retinoblastoma gene in cancer’ as the top perturbed molecular pathway in OC. This finding of epigenomic dysregulation of molecular pathways that have been previously linked to cancer adds biological plausibility to our results. The age-controlled incidence of Ovarian Cancer (OC) is increasing world-wide1. While OC is the second most common gynecologic cancer it remains the most lethal. A primary cause of the lethality is the late stage of detection, due to the absence or non-specific nature of its symptoms. Early-stage disease, confined to the ovary has a 5-year survival of 90% while with distant dissemination, this plummets to 29.2% s urvival2. Unfortunately, only 15% of OC cases are diagnosed at the localized stage while 59% have already metastasized at the time of diagnosis3. There is therefore an urgent need to identify accurate screening markers for OC detection. The most widely used biomarker currently is the cancer antigen-125 (CA-125), a membrane glycoprotein. CA-125 has a low sensitivity for early-stage OC. It also suffers from a lack of specificity given that both benign and malignant non-OC disorders can result in elevated serum levels, limiting its utility as a screening agent4. Imaging studies on the other hand face challenges in resolving cancers of small size and differentiating malignant and non-cancerous l esions5. There has been a surge in scientific interest in the capture and analysis of circulating tumor cells or cell free nucleic acids (“liquid biopsy”) given its potential for minimally invasive cancer detection. Although the various mechanisms of cell free tumor DNA (cfTDNA) release remain to be definitively established, cfTDNA is released with cell necrosis, a poptosis6 along with ongoing release from intact cancer cells. Higher levels of circulating cell free DNA (cfDNA) occur in cancer patients, most of which originate from the neoplastic cells. DNA methylation generally refers to the addition of a carbon atom (aka “methyl group”) to the cytosine nucleotide in DNA. The methylation of the cytosine in the dinucleotide cytosine-phosphate-guanosine (‘CpG’) 1 Department of Obstetrics and Gynecology, Oakland University-William Beaumont School of Medicine, Royal Oak, MI, USA. 2Department of Obstetrics and Gynecology, Beaumont Research Institute, 3811 W. 13 Mile Road, Royal Oak, MI 48073, USA. 3Department of Care Management Analytics, Blue Cross Blue Shield of Michigan, Detroit, MI, USA. *email: Scientific Reports | (2022) 12:18625 | https://doi.org/10.1038/s41598-022-23149-1 1 Vol.:(0123456789) www.nature.com/scientificreports/ remains the most studied epigenetic mechanism. Changes in cytosine methylation is known to alter the 3-dimensional structure DNA and the binding of transcription factors and is thus associated with altered gene expression. This effect is most pronounced for cytosines located in the gene promotor r egion7. Promotor methylation classically results in inactivation of tumor suppressor genes and has been found to be an early event in the ultimate development of ovarian c ancer8. Further, DNA methylation patterns are reportedly significantly different between OC patients and those with benign ovarian mass or healthy c ontrols9. DNA methylation changes are known to occur in multiple locations throughout the genome and thus have lower detection limits compared to mutation or protein marker assays. Indeed, relevant methylation changes that occur in cancer tend to be rare in normal cells10. Quantitation of circulating cfTDNA has been evaluated as a diagnostic approach for OC, however systematic reviews suggest the superiority of epigenetic markers over circulating cfTDNA quantitation11. For all the reasons outlined above, DNA methylation analysis of circulating cfTDNA hold significant promise for minimally invasive OC detection. Artificial intelligence is a broad term that refers to the capability of computers to execute tasks that were previously regarded as having an intellectual basis e.g., reason and learning, and therefore the exclusive domain of humans. Machine learning (ML) is a branch of AI where computers ‘learn’ from previous exposure (input data) and based on that ‘knowledge’ can execute functions such as group classification from a new data set. Studies have reported ML to be superior to conventional statistical approaches for group classification in clinical medicine e.g., accurately differentiating cases from unaffected controls12,13. Its extraordinary capacity for handling high dimensional or big data makes AI attractive for use in omics studies. The authors have focused on combining AI and epigenomics for minimal invasive disease detection14,15. Given the enormous currently untapped potential of Artificial Intelligence in the medical sciences, the current enthusiasm for the use of AI in cancer research16 appears warranted. A limitation of current DNA methylation analysis of cfTDNA in ovarian cancer is the focus on single or a small number of target genes previously identified to be involved in cancer p athogenesis10 which limits diagnostic precision. In this preliminary study, we performed genome wide methylation analysis of cfDNA for the minimally invasive detection of OC. A significant objective of Precision Oncology is to elucidate the pathogenesis of cancers with the ultimate intent of developing targeted t herapeutics17,18. We therefore also evaluated the molecular pathogenesis including gene pathways associated with (...truncated)