Individualizing Ovarian Cancer Surveillance: Discovery and Validation of Serological Personalized Biomarkers of Recurrence Using Multiplex Proteomics Technologies

In Canada, ovarian cancer is the third most common female reproductive cancer and the leading cause of deaths among gynecological cancers. Although remission is observed in most ovarian cancer patients after first-line treatment, >80% of advanced cases see recurrence with a median survival of 12-24 months from the time of recurrence. The classical ovarian cancer biomarker, CA125, is controversial for monitoring recurrence as initiating second-line therapy sooner based on CA125 does not impact survival. Furthermore, CA125 is non-elevated at diagnosis in 10-20% of advanced ovarian cancer cases in general, leaving this population with no widely used biomarkers for surveillance. Patients being monitored with CA125 also have a 10-40% chance of CA125 being non-elevated at recurrence. With increasing selection of immunotherapies and precision medicines, novel personalized ovarian cancer biomarkers could help individualize the surveillance process. Due to tumour heterogeneity, we hypothesize that quantifying the unique array of tumour-derived serological proteins with advanced proteomics methods could identify personalized marker signatures that sensitively detect relapse. We first assessed the technical potential of two multiplex proteomics technologies for detecting proteins that may correlate to tumour burden in cancer patients. For our subsequent discovery study, we employed the proximity extension assay (PEA) to simultaneously measure 1,104 proteins in 120 longitudinal serum samples (30 ovarian cancer patients). We identified 23 candidate personalized markers (plus CA125 and FDA-approved marker HE4), in which personalized combinations was informative of recurrence in more patients (92%) compared to clinical CA125 (68%) and HE4 (32%) alone. For our ensuing validation study, we used PEAs to concurrently measure 644 proteins (includes 21 previously identified candidates plus CA125 and HE4) in 234 independent, longitudinal serum samples (39 ovarian cancer patients). The 21 candidates were each informative of recurrence in 3-35% of patients. Patient-centric analysis of all 644 proteins generated a refined panel of 33 personalized tumour markers, which includes 18 validated candidates from our discovery study. Along with HE4, the 34-marker panel offered higher sensitivity (91%) by identifying personalized marker signatures of recurrence compared to clinical CA125 (59%) and HE4 (26%) alone. Our findings show that personalized tumour markers may offer the best lens into the rich heterogeneity of ovarian tumours compared to a single marker alone. Developing a panel of personalized markers for tracking custom signatures of tumour burden in each patient may offer excellent sensitivity for detecting recurrence early and aid in prompt clinical referral to imaging and subsequent treatment interventions.Ph.D