Evaluation Of MRI Features In Differential Diagnosis Of Benign And Malignant Tumors Of Parotid Gland

Purpose:To compare MRI features that may improve differential diagnosis between benign and malignant parotid tumors. To define some typical MR findings for some frequently encountered parotid tumors.Materials and methods:165consecutive cases with parotid tumors and underwent MR imaging before surgery were retrospectively analyzed. Histology was available for all tumors. The following parameters were analyzed:patient’s gender, age, tumor location, size, shape, margins, signal intensity, growth pattern, extension, and lymphadenopathy. All the patients were divided into benign and malignant group to compare the MRI features. The most frequently encountered tumors in our group include pleomorphic adenoma, adenolymphomas, basal cell carcinoma, acinar cell carcinoma, mucoepidermoid carcinoma, squamous cell carcinoma, oncocytic carcinoma, and lymphoma. MRI findings of these tumors were compared with the group for other tumors.Results:The features highly suggestive of benign tumor include:hyperintensity on T2WI (P=0.004), homogeneous signal on T2WI (P=0.023), defined margin (precontrast P=0.001, postcontrast P=0.000), strong enhancement (P=0.000), homogeneous enhancement (P=0.003). The feature highly suggestive of malignant tumor include: tumor location in deep lobe(P=0.024, accuracy to predict malignancy=70%); moderate signal on T2WI (P=0.024, accuracy to predict malignancy=67.9%); ill-defined margins (precontrast)(P=0.000, accuracy to predict malignancy=72.7%), ill-defined margins(postcontrast)(P=0.000, accuracy to predict malignancy=70.9%); infiltration of subcutaneous tissue (P=0.000, accuracy to predict malignancy=87.2%); infiltration of parapharyngeal space/masticator space/facial nerve(P=0.01, accuracy to predict malignancy=76.3%); lymphadenopathy(P=0.000, accuracy to predict malignancy=80%). inhomogeneous signal on T2WI (P=0.023, accuracy to predict malignancy=41.8%); inhomogeneous enhancement (P=0.002, accuracy to predict malignancy=46.7%); moderate enhancement (P=0.000, accuracy to predict malignancy=69.7%);The following features have no significant difference between malignant and benign tumors:tumor size, patients age, sex, tumor location, polylobulated shape, signal on T1WI, inhomogeneous signal on T1WI, hypointensity on T2WI, moderate enhancement, diffuse growth, single/mulfocal lesions, unilateral/bilateral, cross leaves, cystic/necrotic change, infiltration of bone.Supporting MRI features of tumors include:Pleomorphic adenoma:females (p=0.000), defined margins (p=0.000), T2WI hypointensity (P=0.007), strong enhancement(P=0.023); Adenolymphoma:male(P=0.000), superficial lobe(P=0.000), parotid tail(p=0.000),bilateral(p=0.000),multifocal(P=0.000),moderateenhancement(p=0.001), and homogeneous signal on postcontrast(P=0.023), defined margins(precontrast and postcontrast)(P=0.029,0.023); Basal cell adenoma:low signal ring (P=0.002), defined margins(precontrast and postcontrast)(P=0.037,0.001); Acinar cell carcinoma: polylobulated shape (P=0.001), lymphadenopathy(P=0.031); Mucoepidermoid carcinoma:polylobulated shape (P=0.001), ill-defined margins (P=0.005), lymphadenopathy (P=0.003), cystic/necrotic (P=0.019); Squamous cell carcinoma: polylobulated shape (P=0.045), infiltration of subcutaneous tissue (P=0.003), inhomogeneous signal on TIWI (P=0.001); Oncocytic carcinoma:female (P=0.023), polylobulated shape (P=0.003), deep lobe (P=0.041); Lymphoma were the following: inhomogeneous signal on TIWI (P=0.045), moderate enhancement (P=0.027), infiltration of masticator space/parapharyngeal (P=0.004), diffuse growth (P=0.045), defined margins (P=0.004).Conclusions:Tumor location in deep lobe, ill-defined margins, low signal intensity on T2WI, moderate enhancement, infiltration of surrounding tissue and lymphadenopathy are highly suggestive of malignant parotid tumors. Pleomorphic adenoma, adenolymphoma, basal cell carcinoma, acinar cell carcinoma, mucoepidermoid carcinoma, squamous cell carcinoma, oncocytic carcinoma, lymphoma has definite supporting MRI features.