Tumour-proximal deep serum proteomics identifies biomarkers associated with ovarian cancer.

Abstract

Ovarian cancer diagnosis is often delayed due to poorly defined clinical symptoms, underscoring the need for novel biomarkers that enable earlier detection. Protein biomarker discovery in blood samples is analytically challenging using mass spectrometry, as high-abundance proteins can obscure detection of low-abundance biomarkers. Ovarian vein (ovarian) and median cubital vein (peripheral) serum were collected from healthy women or patients with benign or malignant (majority stage I) serous ovarian tumours. Proteograph XT assay was employed as a protein enrichment methodology prior to LC-MS/MS analyses to obtain deeper proteome coverage in these samples. From an initial cohort of 53 samples, we identified 6603 proteins across all samples, including 3467 proteins uniquely identified in ovarian serum. Following exclusion of red blood cell-contaminated samples, differential abundance analysis identified 86 and 101 malignancy-associated protein candidates in ovarian and peripheral samples, respectively. Notably, ANK1, CAND1 and GMPR2 were differentially abundant across both sample types. Using logistic regression with repeated cross-validation and permutation testing, ANK1 demonstrated strong discriminatory performance [area under the curve (AUC) = 0.97] when combined with clinical CA125 measurements. These findings highlight the importance of ovarian vein sampling for enhanced access to tumour-associated proteins and support further validation of these candidate biomarkers for early ovarian cancer detection.