A novel proteomic-based screening method for ovarian cancer using cervicovaginal fluids: A window into the abdomen.

OBJECTIVE

Our objective is to develop a site-specific proteomic-based screening test for ovarian cancer(OC) using the mucus of the cervix and vagina and evaluate a potential means for home testing.

METHODS

Cervicovaginal fluid samples were obtained from ovarian cancer and normal control patients for LC-mass spectrometry(MS) proteomic evaluation. Statistical modeling determined the protein panel with the highest penetrance across ovarian cancer samples. A subcohort of patients consented to provide self-collected vaginal samples at home with questionnaire on feasibility. Cohen's kappa methodology was utilized to determine agreement between physician-collected and patient-collected samples.

RESULTS

A total of 83 consecutive patient samples were collected prospectively (33 ovarian cancer & 50 controls). Thirty patients consented for self-collection. Using LC-MS, 30 peptides demonstrated independent statistical significance for detecting ovarian cancer. Using statistical modeling, the protein panel that determined the best predictor for detecting OC formed a "fingerprint" consisting of 5 proteins: serine proteinase inhibitor A1; periplakin; profilin1; apolipoprotein A1; and thymosin beta4-like protein. These peptides demonstrated a significant increase probability of detecting ovarian cancer with the ROC curve having an AUC of 0.86 (p = 0.00001). Physician-collected and patient-collected specimens demonstrated moderate agreement with kappa average of 0.6 with upper bound of 0.75.

CONCLUSIONS

Using novel site-specific collection methods, we identified an OC "fingerprint" with adequate sensitivity and specificity to warrant further evaluation in a larger cohort. Agreement of physician-collected and patient-collected samples were encouraging and could improve access to screening with a home self-collection if this screening test is validated in future studies.