A systematic review and meta-analysis of the use of ultrasound to diagnose borderline ovarian tumours. (January 2020)

• Record Type: Journal Article
• Authors: Otify, M.; Laios, A.; Elshamy, T.; D'Angelo, A.; Amso, N.N.
• Is Part Of: European journal of obstetrics, gynecology, and reproductive biology. Volume 244(2020)
• Page Start: 120
• Page End: 127
• Publication Date: 2020-01
• Subjects: Ultrasound -- Ovary -- Borderline ovarian tumour -- Meta-analysis; Obstetrics -- Periodicals; Gynecology -- Periodicals; Reproductive health -- Periodicals; Gynecology -- Periodicals; Obstetrics -- Periodicals; Reproduction -- Periodicals; Obstétrique -- Périodiques; Gynécologie -- Périodiques; Reproduction -- Périodiques; Verloskunde; Gynaecologie; Voortplanting (biologie); Gynecology; Obstetrics; Reproduction; Electronic journals; Periodicals; Electronic journals; 618.05
• DOI: 10.1016/j.ejogrb.2019.11.016
• Languages: English
• ISSNs: 0301-2115

Abstract

Abstract: Borderline ovarian tumours (BOTs) are difficult to diagnose preoperatively. The ability to distinguish between BOTs and other ovarian cancer types prior to surgery could have a profound impact on patient childbearing counselling and surgical planning. Ultrasound (US) pattern recognition by an expert examiner can be an excellent tool for the discrimination of benign and malignant ovarian masses. With respect to US features, most studies were based on well-known risk models. Nevertheless, very few studies have solely evaluated the utility of ultrasound in diagnosing BOTs. We aimed to evaluate the use of US in identifying BOTs solely from benign and malignant ovarian tumours in isolation from risk models. We performed a systematic literature review to identify publications that evaluated the use of US to differentiate between BOTs and malignant and/or benign ovarian tumours using Pubmed, Web of Science and the Cochrane Library. We performed a meta-analysis of the diagnostic sensitivity and specificity studies. We computed the summary estimates for sensitivity and specificity of US in diagnosing BOTs using the bivariate approach of Reitsma in the mada package in R. The initial search resulted in 24, 737 publications. Hundred and seven publications were screened, and five studies contained diagnostic data. Different US criteria applied to identify BOTs. Four out of five studies including 244 women with BOTs and 965 women with benign or malignant tumours were suitable for the meta-analysis. Pooling of the results from four studies showed an overall sensitivity of 0.660 (95 % CI: 0.597 - 0.718) and specificity of 0.854 (95 % CI: 0.728 - 0.927). The overall US accuracy was uniform in sensitivity and variable in specificity. A low false positive rate, 0.146 (95 % CI: 0.073 - 0.272) was observed. US correctly identified BOTs in more than six out of 10 women for potential ovarian sparing surgery, whereas it correctly identified the absence of BOTs in more than eight out of 10 symptomatic women. More carefully designed studies are needed to evaluate the use of pre-operative US for the diagnosis of BOTs.