What are the obstacles for an accurate clinical diagnosis of Pick's disease? A clinicopathologic study.

Several studies have evaluated the reliability and validity of the clinical diagnosis of Alzheimer's disease (AD) using well-defined neuropathologic criteria, but none has attempted to evaluate the diagnostic accuracy of Pick's disease. We determined the accuracy of the clinical diagnosis of Pick's by presenting 105 autopsy-confirmed cases of Pick's (n = 7) and related disorders (non-Pick's, n = 98) as clinical vignettes in randomized order to six neurologists who were unaware of the autopsy findings. The group of raters had moderate to fair agreement for the diagnosis of Pick's as measured by the kappa statistics. The sensitivity for the diagnosis of Pick's for the first visit (mean, 53 months after onset) and last visit (mean, 78 months after onset) was low (range, 0 to 71%), but specificity was near-perfect. Median positive predictive values at both visits were 83 to 85%. False-negative misdiagnoses mainly involved AD. False-positive diagnoses were rare and occurred with corticobasal degeneration (first visit) and with dementia with Lewy bodies (last visit). Pick's was also misdiagnosed by primary neurologists. The best clinical predictors for the early diagnosis of Pick's included "frontal" dementia, early "cortical" dementia with severe frontal lobe disturbances, absence of apraxia, and absence of gait disturbance at onset. However, the first neurologic evaluation in some of the Pick's cases took place in advanced stages of the disease. Our findings suggest that this disorder is underdiagnosed in clinical practice. Although the low sensitivity for the clinical diagnosis of Pick's is disappointing, our data suggest that when clinicians suspect Pick's, their diagnosis is almost always correct. Absence of awareness of the main features of this disorder and of specificity of the frontal lobe syndrome may partially explain the low detection of Pick's disease.