Abstract B31: Endocervical microRNA profiling for detection of ovarian cancer

Though the necessity for ovarian cancer detection at earlier stages has long been acknowledged, biomarker-based discoveries have yielded underwhelming outcomes. The failures of CA-125 and other approaches have raised concern over the sensitivity and robustness of serum biomarkers. However, the discovery that most ovarian cancers originate in the fallopian tube may offer a unique opportunity; since the natural washdown of fallopian tube cells and debris travels through the uterus and into the cervix and vagina, we hypothesized that biomarkers may be concentrated in these proximal fluids. Though we previously demonstrated that tumor DNA is present in the vaginal tract of ovarian cancer patients, the deep sequencing required and low sensitivity (60%) render it insufficient for clinical use. Therefore, we examined other potential biomarkers with high stability that are detectable with methods that are cost effective and noninvasive, such as endocervical sampling. We performed mass spectroscopy on 41 endocervical cytobrush (ECC) samples and found the greatest discriminating pathway between ovarian cancer and benign patients was exosomal composition and trafficking (p=1.3 × 10−21). MicroRNAs (miRNAs) are common exosomal cargo, boast high extracellular stability in bodily fluids, and govern a multitude of cellular processes that can influence ovarian cancer initiation and progression. We therefore used NanoString to perform comprehensive miRNA profiling in matched ECC and plasma samples from high-grade serous ovarian cancer and normal patients. We discovered 13 significant differentially expressed ECC miRNAs and 7 plasma miRNAs individually capable of distinguishing cancer from normal. Not only did ECC samples yield a higher number of significant hits than plasma samples, but these hits also showed better sensitivity and specificity as assessed through their concordance index (c-index) scores generated by area under receiver operator curve analysis (c-index averages of 0.9141 in ECC vs. 0.8796 in plasma). 3 miRNAs overlapped between ECC and plasma, and in each case higher c-index scores were present in ECC samples than in plasma. ECC samples also showed more dramatic differences in expression between cancer and benign (1.4 log2 fold change in ECC vs. 0.8 log2 fold change in plasma), suggesting ECC may yield more consistent and detectable differences in miRNA expression than its plasma counterpart. The miRNAs discovered in our ECC samples showed individual sensiti