A pitfall of MRI for assessing mandibular invasion of squamous cell carcinoma

Abstract

Although a lot of studies have reported the usefulness of MRI and CT for evaluating patients with squamous cell carcinoma in the oral cavity, it seems that further investigation is necessary to define which modality is superior in predicting tumor invasion into the mandible. To compare the diagnostic accuracy of MRI and that of CT for assessing mandibular invasion by squamous cell carcinoma in the oral cavity. Fifty patients with squamous cell carcinoma of the oral cavity with clinical suspicion of mandibular invasion, who were examined by both MRI and CT prior to surgery, were included in this study. All patients underwent marginal or segmental mandibulectomy. The primary site of the carcinoma was the gingival mucosa in 20, the floor of the mouth in 24 and the buccal mucosa in 6 patients. MR and CT images were evaluated for the presence or absence of mandibular invasion by tumor, and the results were correlated with histopathologic findings of the surgical specimens. The images used for the evaluation were axial and coronal T1-weighted images for MRI and bone algorithm axial and dental CT reformatted images for CT. The criterion for mandibular invasion was defect of the cortex of the mandible adjacent to the tumor mass. Twenty-four out of the 50 tumors had histopathologic evidence of mandibular invasion, whereas the remaining 26 had no evidence of invasion. The sensitivity and specificity for MRI in detecting mandibular invasion were 96% and 63%, respectively, and those for CT were 100% and 93%, respectively. The specificity of CT was significantly better than that of MRI (McNemar test, P < .05). Most of the false-positive cases in MRI were those with a carcinoma of the anterior floor of the mouth. In such cases, the lingual cortex was suspected to be involved by tumor on MRI, whereas it was intact on CT. Our study showed that the specificity of MRI for detecting mandibular invasion by squamous cell carcinoma was significantly lower than that of CT. Chemical shift artifacts induced by bone marrow fat, a potential pitfall of MRI, were considered to account for most of the false-positive results.