Predictive value of ADC texture analysis on preoperative DWI for postoperative recurrence of epithelial ovarian cancer and association with progression free survival time

Objective To investigate the predictive value of texture analysis of appearance diffusion coefficient (ADC) in preoperative diffusion-weighted imaging (DWI) for postoperative recurrence of epithelial ovarian cancer (EOC) and its relationship with progression-free survival. Methods A total of 84 EOC patients who were surgically and pathologically diagnosed in the First Central Hospital of Baoding from January 2019 to January 2020 were enrolled and their clinical data and preoperative DWI imaging were retrospectively analyzed. According to post-operative one-year relapse, they were divided into two groups: a recurrence group (n=43) and a non-recurrence group (n=41). Both groups were compared for clinical data and ADC texture parameters (kurtosis, 10th percentile, 25th percentile, indolence, correlation, contrast, variation, and entropy). Logistic regression analysis was used to analyze the risk factors of recurrent EOC, and the ROC curve was plotted to analyze the efficacy of risk factors in predicting EOC recurrence. The follow-up was 36 months, and the survival time of EOC patients with different risk factors was compared. Results There were statistical differences in cancer antigen 125 (CA125), human epididymis protein 4 (HE4), albumin (Alb), tumor size and FIGO stage at the first diagnosis in the recurrence group (P<0.05). The recurrence group presented remarkable decreases in 10th percentile, 25th percentile and correlation (P<0.05) and increases in indolent, contrast, variation, and entropy in the non-recurrence group (P<0.05). Logistic regression analysis showed that tumor size, FIGO stage, indolence, and correlation were the independent risk factors of EOC recurrence (P<0.05). According to ROC curve, the combination of tumor size, FIGO stage, indolence, and correlation predicted the largest AUC area of recurrence, which was 0.992. Patients with tumor diameter<55.374 mm, FIGO stage<2, indolent <696.74, and correlation ≥ 0.000425 showed remarkably longer survival time than those with tumor diameter ≥ 55.374 mm, FIGO stage ≥ 2, indolent ≥ 696.74, and correlation < 0.000425 (P<0.05). Conclusions ADC texture analysis in DWI before EOC is helpful for predicting EOC recurrence and survival time.