Systematic review of the diagnostic accuracy of 99mTc-HMPAO-SPECT in dementia.

OBJECTIVE The authors sought to determine the diagnostic accuracy of 99mTc-HMPAO-SPECT in discriminating between Alzheimer disease (AD) and other dementias. METHODS Articles published between 1985 and 2002 were retrieved systematically from MEDLINE and EMBASE, cross-referencing with personal collections and 13 narrative reviews. Of 301 studies identified, 48 survived exclusion criteria and contained extractable data. Two authors independently assessed and graded the methodology of all included studies. Diagnostic comparison groups included vascular dementia (VD; 13 studies), fronto-temporal dementia (FTD; 7 studies), normal healthy volunteers (27 studies), and non-dementia patients (13 studies). Where statistically justified, groups were pooled in a metaanalysis; summary receiver operating curves were constructed; and heterogeneity across studies examined by regression of the diagnostic odds ratio. RESULTS The pooled weighted sensitivity of 99mTc-HMPAO-SPECT in discriminating clinically defined AD from VD was 71.3%; its specificity was 75.9%. The pooled weighted sensitivity and specificity for AD versus FTD were 71.5% and 78.2%, respectively. Variation in outcome across studies was not found to be attributable to any single factor. CONCLUSION Pathological verification studies suggest that clinical criteria may be more sensitive in detecting AD than brain SPECT (81% versus 74%). However, SPECT studies provide a higher specificity against other types of dementia than clinical criteria (91% versus 70%). SPECT may, therefore, be helpful in the differential diagnosis of AD. Clinical follow-up studies are urgently required to establish its predictive validity with regard to natural history and treatment response.