Development of a multimarker assay for early detection of ovarian cancer

Abstract

Purpose: Early detection of ovarian cancer has great promise to improve clinical outcome.

Patients and methods: Ninety-six serum biomarkers were analyzed in sera from healthy women and from patients with ovarian cancer, benign pelvic tumors, and breast, colorectal, and lung cancers, using multiplex xMAP bead-based immunoassays. A Metropolis algorithm with Monte Carlo simulation (MMC) was used for analysis of the data.

Results: A training set, including sera from 139 patients with early-stage ovarian cancer, 149 patients with late-stage ovarian cancer, and 1,102 healthy women, was analyzed with MMC algorithm and cross validation to identify an optimal biomarker panel discriminating early-stage cancer from healthy controls. The four-biomarker panel providing the highest diagnostic power of 86% sensitivity (SN) for early-stage and 93% SN for late-stage ovarian cancer at 98% specificity (SP) was comprised of CA-125, HE4, CEA, and VCAM-1. This model was applied to an independent blinded validation set consisting of sera from 44 patients with early-stage ovarian cancer, 124 patients with late-stage ovarian cancer, and 929 healthy women, providing unbiased estimates of 86% SN for stage I and II and 95% SN for stage III and IV disease at 98% SP. This panel was selective for ovarian cancer showing SN of 33% for benign pelvic disease, SN of 6% for breast cancer, SN of 0% for colorectal cancer, and SN of 36% for lung cancer.

Conclusion: A panel of CA-125, HE4, CEA, and VCAM-1, after additional validation, could serve as an initial stage in a screening strategy for epithelial ovarian cancer.

Fig 1.

Flowchart of experimental design. MMC, Metropolis algorithm with Monte Carlo simulation; SP, specificity.

Fig 2.

Cumulative receiver operating characteristic (ROC) curves in the training set and distribution of scores in training and validation sets using the four-biomarker panel with Metropolis algorithm with Monte Carlo simulation (MMC): (A) healthy controls versus stage I to IIB ovarian cancer; (B) healthy controls versus stages IIC to IV ovarian cancer. Solid line indicates four-biomarker panel; dashed line indicates CA125. (C) Distributions of scores created by MMC algorithm. P value over a group denotes statistical significance of differences between each group member and appropriate healthy control. Breast, colon, and lung cancers are compared with early-stage ovarian cancer in the validation set.