Diagnostic accuracy of AIDS-related CNS lesions.

Because of the high incidence of neurological complications seen in patients with acquired immunodeficiency syndrome (AIDS), an increasing number of these cases are being referred to neurosurgeons for consideration of intracranial biopsy. To better determine the need for biopsy in these patients we evaluated the accuracy of non-tissue-based neurological diagnoses in AIDS patients who subsequently had a final diagnosis on the basis of biopsy or postmortem brain examinations. The records of 56 AIDS patients who had undergone either autopsy or brain biopsy were retrospectively reviewed. Of the ten patients who underwent biopsy, three were found to have a lesion that was different from the suspected diagnosis and that resulted in a change in treatment. Thirty patients with neurological symptoms had postmortem brain examinations. In the case of the 12 patients who had carried specific diagnoses and received treatments based on those diagnoses, only six diagnoses (50%) were proven correct at autopsy. Of the 18 cases that did not record a specific antemortem diagnosis, in only 5 were normal brains reported, while the others reported a variety of nonspecific or infectious findings. Twelve patients without neurological symptoms had postmortem brain examinations and only six of these (50%) had normal or slightly atrophic brains. Case reports of the others noted nonspecific findings most of which were suggestive of subacute HIV encephalitis. The poor rate of diagnostic accuracy in this series suggests that biopsy should be considered for atypical lesions or those that do not respond to empiric therapy. The use and relative sensitivities of various diagnostic studies are also discussed.