Clinical overdiagnosis of vascular dementia versus necropsy confirmed series.

A. Roses,A. Saunders

DOI: https://doi.org/10.1136/jnnp.62.6.677

1997-06-01

Journal of Neurology Neurosurgery & Psychiatry

Abstract:are meant as a "guideline" and "await testing and validation."2' It is common to overdiagnose vascular dementia in many clinical series. Until and unless Frisoni et al have neuropathological confirmation of their diagnostic accuracy, their conclusions must be seriously questioned. In fact, as questionable as the diagnostic considerations for a clinical diagnosis of vascular dementia may be, other investigators have failed to replicate the increased £4 allele frequency previously reported by Frisoni et al in clinical vascular dementia.4-6 Four series have independently measured the specificity, sensitivity, and positive predictive value of apoE genotyping in prospectively ascertained patients with neuropathologically confirmed dementia (table 1).7 " (Welsh-Bohmer et al, unpublished data). As often found in very large neuropathology series, vascular dementia was uncommon in each of these series. Moreover, the positive predictive value of having a single £4 allele was greater than 95% in all series of symptomatic patients, despite differences in ascertainment protocols. For example, subjects in the CERAD series were enrolled early in their disease from more than 20 different centers. The positive predictive value in the interim report of this large series with 162 subjects was 97% (table 2) (Welsh-Bohmer et al, unpublished data). Other authors besides Frisoni et al have implied that their clinical diagnoses are without error and should be accepted as the gold standard. Recently, Slooter et al reported that "apoE test characteristics for a diagnosis of Alzheimer's disease were calculated, with patients having another type of dementia as reference, and our diagnostic work up as the gold standard."' The real question for the evaluation of these data is whether one accepts the clinical diagnosis of vascular dementia by Frisoni et al as the "gold standard" or whether one looks to necropsy confirmed series for formal measurement of the positive predictive value. The definitive diagnoses of Alzheimer's disease, vascular dementia, and other dementias are based on neuropathological criteria.2' The missing data for Frisoni et al,' Slooter et al," and others are the necropsy confirmed accuracy of their diagnoses. It is highly likely that many vascular dementia patients of Frisoni et al will meet the neuropathological criteria for Alzheimer's disease. ALLEN D ROSES ANN M SAUNDERS J7oseph anid Kathleeni Bryanz Alzheimner's Disease Research Center, Divisioni ofNeurology Duke University Medical Ceniter, Durham, ANC 27710-2900, USA