Accuracy of computed tomography in predicting adrenal tumor size.

We undertook a multi-institutional, retrospective study of 51 patients with adrenal tumors (pheochromocytomas, functioning and nonfunctioning cortical adenomas, cysts and carcinomas). All patients had computed tomography (CT) with a maximum of 5 millimeter cuts the week before undergoing complete adrenalectomy. Pathologists were asked to measure the tumor to the nearest 0.1 centimeter. Tumor size obtained from pathologic reports (actual size) and CT reports (estimated size) were compared. Adrenal tumors were divided into two groups on the basis of size--tumors with actual size of 6 centimeters or greater and tumors with actual size of less than 6 centimeters. Statistical analysis was performed with Newman-Keuls analysis of variance. After controlling for tumor type and for the institution at which the measurement was made, we found that CT consistently underestimated adrenal tumor size in both tumor groups. Moreover, the average, underestimated difference for tumors 6 centimeters or greater was 32 percent, but 47 percent for tumors less than 6 centimeters (p = 0.060). CT seemed to underestimate the size of small adrenal tumors more than large tumors. Because the decision to operate on solid, nonfunctioning adrenal tumors is based on tumor size and because CT is currently the standard technique used to estimate size, our findings need to be considered before undertaking surgical treatment.