Diagnostic measures comparison for ovarian malignancy risk in Epithelial ovarian cancer patients: a meta-analysis

Abstract

Epithelial ovarian cancer has become the most frequent cause of deaths among gynecologic malignancies. Our study elucidates the diagnostic performance of Risk of Ovarian Malignancy Algorithm (ROMA), Human epididymis secretory protein 4 (HE4) and cancer antigen (CA125). To compare the diagnostic accuracy of ROMA, HE-4 and CA125 in the early diagnosis and screening of Epithelial Ovarian Cancer. Literature search in electronic databases such as Medicine: MEDLINE (through PUBMED interface), EMBASE, Google Scholar, Science Direct and Cochrane library from January 2011 to August 2020. Studies that evaluated the diagnostic measures of ROMA, HE4 and CA125 by using Chemilumincence immunoassay or electrochemiluminescence immunoassay (CLIA or ECLIA) as index tests. Using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2). We included 32 studies in our meta-analysis. We calculated AUC by SROC, pooled estimated like sensitivity, specificity, likelihood ratio, diagnostic odds ratio (DOR), Tau square, Cochran Q through random effect analysis and meta-regression. Data was retrieved from 32 studies. The number of studies included for HE4, CA125 and ROMA tests was 25, 26 and 22 respectively. The patients with EOC were taken as cases, and women with benign ovarian mass were taken as control, which was 2233/5682, 2315/5875 and 2281/5068 respectively for the markers or algorithm. The pooled estimates of the markers or algorithm were sensitivity: ROMA (postmenopausal) (0.88, 95% CI 0.86-0.89) > ROMA (premenopausal) 0.80, 95% CI 0.78-0.83 > CA-125(0.84, 95% CI 0.82-0.85) > HE4 (0.73, 95% CI 0.71-0.75) specificity: HE4 (0.90, 95% CI 0.89-0.91) > ROMA (postmenopausal) (0.83, 95% CI 0.81-0.84) > ROMA (premenopausal) (0.80, 95% CI 0.79-0.82) > CA125 (0.73, 95%CI 0.72-0.74), Diagnostic odd's ratio ROMA (postmenopausal) 44.04, 95% CI 31.27-62.03, ROMA (premenopausal)-18.93, 95% CI 13.04-27.48, CA-125-13.44, 95% CI 9.97-18.13, HE4-41.03, 95% CI 27.96-60.21 AUC(SE): ROMA (postmenopausal) 0.94(0.01), ROMA (premenopausal)-0.88(0.01), HE4 0.91(0.01), CA125-0.86(0.02) through bivariate random effects model considering the heterogeneity. Our study found ROMA as the best marker to differentiate EOC from benign ovarian masses with greater diagnostic accuracy as compared to HE4 and CA125 in postmenopausal women. In premenopausal women, HE4 is a promising predictor of Epithelial ovarian cancer; however, its utilisation requires further exploration. Our study elucidates the diagnostic performance of ROMA, HE4 and CA125 in EOC. ROMA is a promising diagnostic marker of Epithelial ovarian cancers in postmenopausal women, while HE4 is the best diagnostic predictor of EOC in the premenopausal group. Our study had only EOC patients as cases and those with benign ovarian masses as controls. Further, we considered the studies estimated using the markers by the same index test: CLIA or ECLIA. The good number of studies with strict inclusion criteria reduced bias because of the pooling of studies with different analytical methods, especially for HE4. We did not consider the studies published in foreign languages. Since a few studies were available for HE4 and CA125 in the premenopausal and postmenopausal group separately, data were inadequate for sub-group analysis. Further, we did not assess these markers' diagnostic efficiency stratified by the stage and type of tumour due to insufficient studies.

Figure 1

PRISMA Flow-diagram.

Figure 2

Quality of publications (QUADAS-2) on six domains prepared by using Revman 5.3

Figure 3

Comparison of Sensitivity estimates of HE4 (A), CA125 (B), ROMA for pre-menopause (C) and ROMA for post-menopause (D).

Figure 4

Comparison of Specificity estimates of HE4 (A), CA125 (B), ROMA for pre-menopause (C) and ROMA for post-menopause (D).

Figure 5

Comparison of positive likelihood ratio estimates (using Random effect model) of HE4 (A), CA125 (B), ROMA for pre-menopause (C) and ROMA for post-menopause (D).

Figure 6

Comparison of negative likelihood ratio estimates (using Random effect model) of HE4 (A), CA125 (B), ROMA for pre-menopause (C) and ROMA for post-menopause (D).

Figure 7

Comparison of diagnostic odds ratio estimates (using Random effect model) of HE4 (A), CA125 (B), ROMA for pre-menopause (C) and ROMA for post-menopause (D).

Figure 8

Comparison of Area Under Curve (AUC) by symmetric ROC analysis for HE4 (A), CA125 (B), ROMA for pre-menopause (C) and ROMA for post-menopause (D).