Introduction/Background The incidence of positive lymph nodes in early-stage ovarian cancer is estimated to be around 15–20%. The standard procedure for determining the stage of high-risk presumed early-stage ovarian cancer involves systematic pelvic and para-aortic lymphadenectomy. However, this surgical staging method is invasive with the risk of developing lymphedema. Our objective was to evaluate the ability of whole-body diffusion-weighted magnetic resonance imaging (WB-DWI/MRI) to accurately predict the presence of lymph node metastases in patients with early-stage ovarian cancer.

Methodology Our study, conducted from January 2012 to December 2021 in a tertiary referral hospital, was a retrospective cohort study. The participants met the following criteria: (a) All patients with a unilateral or bilateral ovarian mass, without signs of intraperitoneal or distant metastases; (b) all histological subtypes were considered; (c) WB-DWI/MRI was performed during the staging process; and (d) systematic pelvic and para-aortic lymphadenectomy was carried out, except for stage I low-grade endometrioid tumours and mucinous tumours with expansive growth.

Nodal signal intensity relative to the surrounding lymph nodes and primary tumour was used on WB-DWI/MRI as a criterion for the prediction of metastatic lymph node involvement irrespective of nodal size.

Results In the study, 49 patients were included (Table 1), out of which 10.2% had lymph node metastases (n=5). WB-DWI/MRI showed a sensitivity of 60.0% (95% confidence interval [CI] 19.9% to 91.9%), a specificity of 88.6% (95% CI 77.1% to 95.8%) and an accuracy of 85.7% (95% CI 72.8% to 94.1%). The positive predictive value was 37.5% (95% CI 11.0% to 71.0%) and the negative predictive value was 95.1 (95% CI 85.7% to 99.2%).

Conclusion WB-DWI/MRI can predict lymph node metastases in early-stage ovarian carcinoma. The high negative predictive value is particularly interesting in selecting patients with early-stage ovarian cancer for lymphadenectomy by WB-DWI/MRI.

Disclosures None.