Randal S. Weber, MD,Mario A. Luna, MD Lawrence Ginsberg, MD,
Ya-Yen Lee, MD,Patti Hankins, RN, Pamela Van Tassel, MD, Pat Wolf

Determining the disease extent for carcinomas of the larynx and hypopharynx is important for staging and treatment planning. This comparison study between computed tomography (CT) and magnetic resonance imaging (MRI) with histopathologic correlation was undertaken to determine their sensitivity, specificity, accuracy, and positive predictive value (PPV) for cancer staging. Twenty-three untreated patients with carcinoma of the larynx ( 18 patients) or hypopharynx (5 patients) were imaged and the studies were evaluated in a blinded fashion. The laryngeal specimens were fixed, serially sectioned, and examined histologically for tumor extent. Sensitivity, specificity, accuracy, and PPV were determined by comparing the clinical and imaging assessment of tumor extent to the pathologic findings. We found MR] to be more sensitive than CT for tumor involvement of the laryngeal ventricle, cartilage, anterior commissure, and preepiglottic space, whereas CT was more sensitive than MR1 for subglottic extension. The mean overall sensitivity, specificity, accuracy, and PPV for MRI versus CT were 84% Versus 74%: 59% versus 66%; 70% versus 69%; and 60% versus 61 1% While MRI is more sensitive than CT for determining tumor extension, their overall accuracy is equivalent. In the absence of superiority, decisions regarding selection of imaging should be based on cost and patient acceptance.