Active Ratio Test (ART) as a Novel Diagnostic for Ovarian Cancer

Abstract

Background: Despite substantial effort, there remains a lack of biomarker-based, clinically relevant testing for the accurate, non-invasive diagnostic or prognostic profiling of epithelial ovarian cancers (EOC). Our previous work demonstrated that whilst the inflammatory marker C-X-C motif chemokine ligand 10 (CXCL10) has prognostic relevance in ovarian cancer, its use is complicated by the presence of multiple, N-terminally modified variants, mediated by several enzymes including Dipeptidyl Peptidase 4 (DPP4).

Methods: In this study, we provide the first evidence for the "Active Ratio Test" (ART) as a novel method to measure biologically relevant CXCL10 proteoforms in clinical samples.

Results: In a cohort of 275 patients, ART accurately differentiated patients with malignant EOCs from those with benign gynaecological conditions (AUC 0.8617) and significantly out-performed CA125 alone. Moreover, ART combined with the measurement of CA125 and DPP4 significantly increased prognostic performance (AUC 0.9511; sensitivity 90.0%; specificity 91.7%; Cohen's d > 1) for EOC detection.

Conclusion: Our data demonstrate that ART provides a useful method to accurately discriminate between patients with benign versus malignant EOC, and highlights their relevance to ovarian cancer diagnosis. This marker combination may also be applicable in broader screening applications, to identify or discriminate benign from malignant disease in asymptomatic women.

Keywords: Active Ratio Test; CXCL10; ELISA; biomarkers; early detection; ovarian cancer.

Figure 1

Representative standard curves for the detection of recombinant, full-length CXCL10 using (A) mAb-RA2 and (B) mAb-RG2 in the range 49–200,000 pg mL⁻¹. Mean +/− SD; triplicate measurements per data point.

Figure 2

(A) Comparison of total CXCL10 detection using ART or commercial ELISA in ascites fluid from ovarian cancer patients (n = 212). Data is mean +/− SD, triplicate replicates per sample; (B,C) Significant correlations were observed between measurements made using ART or commercial ELISA for both (B) benign (n = 50; p = 0.0371); and (C) malignant (n = 200; p = 0.0002) sample types.

Figure 3

ART successfully discriminated citrullinated from non-citrullinated CXCL10. (A) In vitro citrullination of recombinant, full length CXCL10 by PAD2 resulted in loss of detection by commercial anti-CXCL10 antibody ab9807 (upper left), and a substantial reduction in detection by mAb-A2 (lower left). Citrullination was confirmed using an anti-citrulline antibody (upper right). Detection of citrullinated CXCL10 by mAb-RG2 was unaffected. (B) ELISA using mAb-RA2 (left) or mAb-RG2 (right) confirmed that detection of citrullinated CXCL10 by mAb-RA2 was abrogated, whilst detection by mAb-RG2 was unaffected. n = 3 replicates per data point.

Figure 4

ART differentiates between benign and malignant disease in ascites fluid from ovarian cancer patients. (A) Total and (B) active CXCL10 concentrations measured in benign (n = 51) and malignant (n = 212) ascites fluid samples. (C,D) Calculated active ratio between (C) benign and all malignant samples; or (D) benign and cancer samples separated according to FIGO disease stage. **** p ≤ 0.0001.

Figure 5

CA125 does not correlate with active ratio. (A) Plasma CA125 was measured for patients with either benign (200.8 ± 368.7 U mL⁻¹; n = 48) or malignant ovarian tumours (1697.0 ± 3409.0 U mL⁻¹; n = 188). Mean +/− SD, **** p ≤ 0.0001. (B) No significant correlation was observed between CA125 and active ratio in either case.

Figure 6

DPP4 abundance and specific activity in clinical samples. (A) DPP4 abundance (benign n = 49; malignant n = 148) and specific activity (benign n = 49; malignant n = 50) in patient’s ascites fluid were measured by ELISA or activity assay, respectively. (B) Correlations between active ratio and DPP4 abundance (upper panel) or specific activity (lower panel) in matching ascites samples.

Figure 7

Marker combinations using ART discriminate between benign and malignant disease. Receiver operator curves (ROC) were used to evaluate the discriminatory power of ART alone compared with (A) individual measurements of total or active CXCL10; or (B) CA125 alone, DPP4 alone or a combination of CA125, DPP4 and ART in ascites samples.

Figure 8

ART performed using either plasma or cervicovaginal swabs discriminated between benign and malignant disease. CXCL10 was measured using ART in (A) cervicovaginal swabs (CVS) and (B) matched plasma samples from patients with either benign or malignant ovarian cancer. n = 50/group; **** p < 0.0001, ** p < 0.01.