Prediction of postoperative residual primary ovarian neoplasm or metastatic lesion close to rectum of serous ovarian carcinoma based on clinical and MR T1-DEI features.

Abstract

Background: The optimal primary debulking surgery outcome of serous ovarian carcinoma (SOC) is greatly affected by primary ovarian neoplasm or metastatic lesion close to the rectum.

Purpose: To study the risk factors affecting postoperative residual primary ovarian neoplasm or metastatic lesion close to the rectum of SOC.

Material and methods: The clinical and MRI data of 164 patients with SOC eligible from institution A (training and test groups) and 36 patients with SOC eligible from institution B (external validation group) were collected and retrospectively analyzed. The clinical data included age, serum carbohydrate antigen 125 (CA-125), human epididymis protein 4, and neutrophil-to-lymphocyte ratio (NLR). Magnetic resonance imaging (MRI) data included ovarian mass distribution, maximum diameter of ovarian mass, ovarian mass features, degree of rectal invasion of the primary ovarian neoplasm or metastatic lesion, and amount of ascites. A model was established using multivariate logistic regression.

Results: By univariate and multivariate logistic regressions, CA-125 (P = 0.024, odds ratio [OR] = 3.798, 95% confidence interval [CI] = 1.24-13.32), NLR (P = 0.037, OR = 3.543, 95% CI = 1.13-12.72), and degree of rectal invasion of the primary ovarian neoplasm or metastatic lesion (P < 0.001, OR = 37.723, 95% CI = 7.46-266.88) were screened as independent predictors. The area under the curve values of the model in the training, test, and external validation groups were 0.860, 0.764, and 0.778, respectively.

Conclusion: The clinical-radiological model based on T1-weighted dual-echo MRI can be used non-invasively to predict postoperative residual ovarian neoplasm or metastasis close to SOC in the rectum.