Predictive value of computed tomography-based diagnosis of intracranial tuberculomas

Abstract

The need to obtain histological diagnoses of intracranial tuberculomas, before initiating therapy, is not universally accepted, because some clinicians believe that an image-based diagnosis is fairly accurate in patients from endemic regions. To evaluate the sensitivity, specificity, and predictive value of computed tomography (CT)-based diagnosis of an intracranial tuberculoma, we prospectively compared the preoperative imaging diagnoses with histological diagnoses in 105 consecutive patients with intracranial masses. CT differential diagnoses (first or second) of tuberculomas were considered in 21 patients. Seven of them were histologically confirmed to have tuberculomas (true-positive results); 14 had other diseases (false-positive results). The 14 false-positive cases included 6 cases of astrocytomas, 5 of metastases, and 3 with miscellaneous diagnoses. All tuberculomas were correctly diagnosed on the CT scans (5 by both surgeons and 2 by one surgeon). During the study period, we encountered 11 patients who were referred by other clinicians with diagnoses of tuberculomas on the basis of their CT scans. We concurred with their CT diagnoses in 5 of them, but only 1 patient had a histologically verified tuberculoma. Astrocytomas (4 patients), metastases (3 patients), and solitary cysticercus granulomas (3 patients) were the causes of misdiagnosis in this group of patients. Although the sensitivity of CT in the diagnosis of intracranial tuberculomas is 100%, and its specificity is 85.7%, the positive predictive value is only 33% (confidence limits, 24-42%). The negative predictive value is 100%. The low positive predictive value for a diagnosis of intracranial tuberculoma on CT alone indicates the need for a confirming histological diagnosis.