A two-step strategy for identification of plasma protein biomarkers for endometrial and ovarian cancer

Abstract

Background: Over 500,000 women worldwide are diagnosed with ovarian or endometrial cancer each year. We have used a two-step strategy to identify plasma proteins that could be used to improve the diagnosis of women with an indication of gynecologic tumor and in population screening.

Methods: In the discovery step we screened 441 proteins in plasma using the proximity extension assay (PEA) and five Olink Multiplex assays (CVD II, CVD III, INF I, ONC II, NEU I) in women with ovarian cancer (n = 106), endometrial cancer (n = 74), benign ovarian tumors (n = 150) and healthy population controls (n = 399). Based on the discovery analyses a set of 27 proteins were selected and two focused multiplex PEA assays were developed. In a replication step the focused assays were used to study an independent set of cases with ovarian cancer (n = 280), endometrial cancer (n = 228), women with benign ovarian tumors (n = 76) and healthy controls (n = 57).

Results: In the discovery step, 27 proteins that showed an association to cancer status were identified. In the replication analyses, the focused assays distinguished benign tumors from ovarian cancer stage III-IV with a sensitivity of 0.88 and specificity of 0.92 (AUC = 0.92). The assays had a significantly higher AUC for distinguishing benign tumors from late stage ovarian cancer than using CA125 and HE4 (p = 9.56e-22). Also, population controls could be distinguished from ovarian cancer stage III-IV with a sensitivity of 0.85 and a specificity of 0.92 (AUC = 0.89).

Conclusion: The PEA assays represent useful tools for identification of new biomarkers for gynecologic cancers. The selected protein assays could be used to distinguish benign tumors from ovarian and endometrial cancer in women diagnosed with an unknown suspicious pelvic mass. The panels could also be used in population screening, for identification of women in need of specialized gynecologic transvaginal ultrasound examination.

Funding: The Swedish Cancer Foundation, Vinnova (SWELIFE), The Foundation for Strategic Research (SSF), Assar Gabrielsson Foundation.

Keywords: Diagnostics; Endometrial cancer; Ovarian cancer; Proximity extension assay (PEA); Sensitivity; Specificity.

Fig. 1

The study design for identification of protein biomarkers using PEA

Fig. 2

Distribution of differences in NPX values in the discovery step between benign tumors and OC (a–c) and between OC and EC (d). Only protein labels for the significant differences are shown

Fig. 3

Reporter operator characteristic (ROC) for combinations of proteins in the discovery step. a Benign tumors versus Ovarian cancer stage I–II. b Benign tumors versus Ovarian cancer stage III–IV. c Benign tumors versus Ovarian cancer stage I–IV. d Population controls versus Ovarian cancer stage I–II. e Population controls versus Ovarian cancer stage III–IV. f Population controls versus Ovarian cancer stage I–IV. g Benign tumors versus Endometrial cancer. h Controls versus Endometrial cancer. i Ovarian cancer stage I-IV versus endometrial cancer versus

Fig. 4

Correlation between CA125 values from the multiplex PEA used in the discovery step and from clinical ELISA, for benign tumors (in black) and ovarian cancer patients (in red)

Fig. 5

Correlation in protein abundance between the 92-plex panels and the focused protein panels for the six proteins a CA125, b IL10, c ENRAGE, d HE4, e MK, f Dkk4

Fig. 6

Reporter operator characteristic (ROC) for combinations of proteins in replication stage. The blue line is using CA125 and HE4 only. a Benign tumors versus Ovarian cancer stage I–II. b Benign tumors versus Ovarian cancer stage III–IV. c Benign tumors versus Ovarian cancer stage I–IV. d Population controls versus Ovarian cancer stage I–II. e Population controls versus Ovarian cancer stage III–IV. f Population controls versus Ovarian cancer stage I–IV. g Benign tumors versus Endometrial cancer. h Controls versus Endometrial cancer. i Ovarian cancer stage I–IV versus endometrial cancer versus

Fig. 7

Correlation between AUC values for some of the proteins overlapping between our study and that by Boylan et al. [5] for the comparison between benign tumors and ovarian cancer stage III–IV (left panel) and I–II (right panel)