Detecting bone invasion of the maxilla by oral squamous cell carcinoma: diagnostic accuracy of preoperative CT versus MRI

Abstract Background For planning of the surgical resection, CT and MRI are commonly used for the preoperative assessment of bone invasion of the maxilla. The purpose of this study is to compare the diagnostic test accuracy of CT and MRI for detecting bone invasion of the maxilla in patients with squamous cell carcinoma of the maxilla (MSCC). Materials and Methods The investigators conducted a retrospective cross-sectional study and enrolled a consecutive number of patients with primary MSCC between 2000 and 2017, who had either preoperative CT scan or MRI scan. The outcome variable was the absence/presence of bone invasion, with histopathological examination as the gold standard. The predictor variable was the imaging technique (CT and MRI). The imaging results on bone invasion were compared with the histopathological results. Sensitivity and specificity were calculated, and two-sided Fisher’s exact test was used to calculate statistically significant differences between the unpaired CT and MRI results. ROC-curves were computed and Area Under the Curves (AUC) calculated. Results 72 patients were included, 29 male and 43 female patients, with mean age 72 years. 41 CT scans and 31 MRI scans were available. Histopathological examination showed bone invasion in 45 cases: in 26 of 41 patients (63%) with a CT scan and in 19 of 31 patients (61%) with an MRI scan. CT yielded 2 false positive and 2 false negative results, sensitivity 92%, specificity 87%, AUC 0.895. MRI yielded 5 false positive and 2 false negative results, sensitivity 89%, specificity 58%, AUC 0.739. No significant differences were observed for sensitivity (p=1) and specificity (p=0.185). Conclusion In the absence of metallic dental restorations, CT could detect bone invasion more accurately than MRI in this study; however, the difference was not statistically significant. The imaging method of choice may depend on other situational factors.