Accurate ADC thresholds for the diagnosis of middle ear cholesteatoma: Recurrence prediction, pathological correlation, and subgroup comparison

Objective: This study aims to differentiate cholesteatoma from other inflammatory lesions using quantitative data derived from ADC maps and to improve diagnostic accuracy by determining cutoff values. Additionally, it seeks to predict the pathological structure and recurrence risk of the lesions based on ADC values. Materials and Methods: This retrospective study was conducted between 2018 and 2025 in the Radiology and Otorhinolaryngology departments of Çukurova University Faculty of Medicine. Patients with preoperative DWI imaging and pathologically confirmed diagnoses were included. ADC measurements were performed blindly using ROIs placed on hyperintense lesions in non-EPI DWI sequences. For each patient, mADC values were obtained from both the cholesteatoma and corresponding white matter regions, and normalized ADC (nADC) values were calculated. Based on pathological findings, patients were classified into three groups: cholesteatoma, non-cholesteatomatous inflammatory lesions, and mixed lesions. Patients who developed recurrence during postoperative follow-up were also identified. Sensitivity, specificity, and cutoff values of ADC were determined using ROC curve analysis to differentiate between cholesteatoma and other lesions. Results: Of the 134 patients included, 57.5% were male and 42.5% were female, with a mean age of 33.3±9.2 years. Cholesteatoma was pathologically confirmed in 83.6% of cases; 50.7% had cholesteatoma with inflammation, 35.1% had pure cholesteatoma, and 14.2% had only inflammation. The mean mADC and nADC values in the cholesteatoma group were 869.2±322.5 and 1.07±0.39, respectively, whereas in the non-cholesteatoma group, they were 1338.1±366.1 and 1.66±0.43 (p<0.001). According to the ROC analysis, an nADC cut-off value of <1.208 demonstrated the highest diagnostic accuracy for predicting cholesteatoma, with a sensitivity of 85.11% and a specificity of 100% (AUC: 93.4%). Additionally, an mADC cut-off value of <887×10⁻⁶ mm²/s showed a sensitivity of 80.8% and a specificity of 100%. Patients who developed recurrence had significantly lower mADC and nADC values compared to those without recurrence (p=0.032 and p=0.017, respectively). Moreover, pure cholesteatoma could be statistically distinguished from inflammatory groups (p<0.001). Conclusion: mADC and nADC values obtained from DWI-MRI were found to be statistically significant in the differential diagnosis of cholesteatoma and in predicting recurrence. These values also proved useful in assessing the inflammatory nature and prognosis of the lesion, achieving high diagnostic accuracy. Notably, the nADC value demonstrated outstanding specificity. Keywords: Cholesteatoma, Diffusion-Weighted MRI, ADC, Normalized ADC, Recurrence, Diagnosis, Prognosis.