[Skip to Content]
[Skip to Content Landing]
Citations 0
February 27, 2002

Usefulness of Positron Emission Tomography in Evaluating Dementia—Reply

Author Affiliations

Stephen J.LurieMD, PhD, Senior EditorIndividualAuthor

JAMA. 2002;287(8):985-987. doi:10.1001/jama.287.8.983

In Reply: Drs Rutschmann and Matchar mention the AAN recommendations1 regarding the use of clinical criteria and the importance of early diagnosis of AD. These recommendations were made prior to our study of the accuracy of diagnosis with PET (94%-95% sensitivity, 71%-78% specificity) against the criterion standard of neuropathologic diagnosis. We agree that early diagnosis is of particular interest. That is why we not only provided data on patient composition by age and symptom severity (suggesting that subjects on average were in an early stage of cognitive decline), but also provided an analysis stratified by severity indicating that accuracy with PET was equally high for patients with only "questionable" or "mild" dementia. Although Rutschmann and Matchar assert that the sensitivity of clinical criteria "has been consistently shown to be approximately 90% compared with an autopsy criterion standard," the AAN1 committee, of which one of us (G.W.S.) was a member, found only 1 study2 on the accuracy of clinical diagnosis of early dementia having Class I quality of evidence (ie, prospective design, broad patient spectrum, blinded evaluation, and neuropathologic confirmation of diagnosis). In that study, the clinical diagnostic sensitivity (after multiple clinical examinations over an average of 3 years) ranged from 83% to 85%, compared with the sensitivity of 95% found for questionable and mild dementia after a single PET scan in our study. The AAN identified only 2 other Class I studies,3,4 and these involved patients with more advanced states of dementia. Using the recommendation of the AAN to base diagnosis on the "probable AD" criterion,1 the sensitivities measured for clinical evaluation were 49%3 and 50%.4 The AAN concluded that higher clinical sensitivity (based on the diagnosis of "possible AD") could only be achieved "at the price of specificity (average across 4 studies = 48%),"1 much lower than the specificity found using PET in our study. Nevertheless, our study did not aim to assess the incremental gain that PET provides over standard clinical criteria, but rather tried to establish as accurately as possible the sensitivity and specificity of PET for the broad spectrum of patients undergoing evaluation for dementia.