A magnetic resonance imaging-based decision-making tool for predicting complex anal fistulas healing in the early postoperative period

Abstract

Abstract Background Magnetic resonance imaging (MRI) has excellent accuracy in diagnosing preoperative lesions before anal fistula surgery. However, MRI is not good in identifying early recurrent lesions and effective methods for quantitative assessment of fistula healing are still warranted. This retrospective study aimed to develop and validate a specific MRI-based nomogram model to predict fistula healing during the early postoperative period. Methods Patients with complex cryptoglandular anal fistulas who underwent surgery between January 2017 and October 2020 were included in this study. MRI features and clinical parameters were analyzed using univariate and multivariate logistic regression analysis. A nomogram for predicting fistula healing was constructed and validated. Results In total, 200 patients were included, of whom 186 (93%) were male, with a median age of 36 (18–65) years. Of the fistulas, 58.5% were classified as transsphincteric and 19.5% as suprasphincteric. The data were randomly divided into the training cohort and testing cohort at a ratio of 7:3. Logistic analysis revealed that CNR, ADC, alcohol intake history, and suprasphincteric fistula were significantly correlated with fistula healing. These four predictors were used to construct a predictive nomogram model in the training cohort. AUC was 0.880 and 0.847 for the training and testing cohorts, respectively. Moreover, the decision and calibration curves showed high coherence between the predicted and actual probabilities of fistula healing. Conclusions We developed a predictive model and constructed a nomogram to predict fistula healing during the early postoperative period. This model showed good performance and may be clinically utilized for the management of anal fistulas.

Keywords

• Anal fistulas
• Magnetic resonance imaging
• Nomogram
• Prediction
• Fistula healing