Commentary on “Aggressive course of disease in dementia” Jason Brandt* The Johns Hopkins University School of Medicine, Baltimore, MD, and The Copper Ridge Institute, Sykesville, MD, USA Gauthier and colleagues (this issue) provide a valuable service to Alzheimer’s disease (AD) clinicians and investi- gators by reviewing the sizable literature on rapidly pro- gressive dementia and the variables that contribute to it. The authors are entirely correct in declaring that there is great variability in rates of cognitive and functional decline in AD, but that there are no uniform methods for defining or measuring such decline. They are also correct in identifying several risk factors for rapid decline that emerge relatively consistently from the methodologically diverse studies per- formed to date: younger age at onset, early presence of psychotic features (hallucinations and delusions), early Par- kinsonism, and specific genetic factors. This is an area of investigation fraught with conceptual, methodologic, and data analytic problems [1], many, but not all, of which are acknowledged by the authors. First, diag- nostic criteria need to be sufficiently operational and precise to allow easy replication, and research samples should be as diagnostically homogeneous as possible. To discuss factors predictive of rapid course of disease in dementia, rather than in a specific dementia-causing disease, is to invite trouble. In fact, many of the studies described by Gauthier et al. are of patients with Parkinson’s disease (PD) or dementia with Lewy bodies (DLB), rather than AD. Although these disor- ders often coexist in the same patient, they undoubtedly have unique pathophysiologies and should therefore be kept distinct, especially when discussing genetic and other bio- logical influences on rate of decline and response to phar- macologic treatment. Second, it is not clear whether patients who have an “aggressive course” or are rapid decliners constitute a dis- tinct subtype. This is implied to be the case when one attempts to describe the prevalence of “the condition.” But are patients with an aggressive course categorically differ- ent from patients with a more gradual decline? Or is it instead the case that rate of progression is a continuously distributed variable, and that patients described as having an aggressive course are simply at one end of the distribution? There appears to be greater support for this latter position. There is a certain circularity to much of the research in this area. Many of the so-called predictors of rapid decline (eg, low neuropsychological test scores, poor nutrition, pa- tient apathy) may, in fact, be consequences of rapid decline. To conduct a study and conclude that patients who have deteriorated quickly before first being assessed are likely to continue to progress rapidly, or that those who decline rapidly have a poorer prognosis (ie, will reach clinical milestones sooner), seems to be discovering the obvious. The authors point out that there is no uniformity among studies in the choice of outcome variables. Some investiga- tors examine rate of change of scores (typically points per year) on 1 or more cognitive tests or rating scales. The Mini-Mental State Exam (MMSE)— either the original, 30- point form or any of a number of expanded modifica- tions— has been used often for this purpose. Other investi- gators use survival approaches, modeling time to reach critical endpoints (eg, a particular MMSE score, requiring nursing home care, death). Neither of these approaches is inherently superior to the other, and both have limitations. It is impossible to study the natural history of a disease like AD, including its rate of progression, apart from the instruments and metrics used to assess it. Therefore, atten- tion to the psychometric properties of measurement instru- ments is critical in this area of dementia research. When carefully examined, the MMSE may be a poor choice for the detection of patients with an aggressive course, primarily because of its insensitivity to change relative to its test– retest reliability. The average annual decline in MMSE score among AD patients followed up prospectively in the CERAD study was 3.4 points, only marginally greater than the standard deviation of the measurement (2.8 points) [2]. As a result, even after 4 years of follow-up, approximately 16% of the CERAD AD patients had no clinically mean- ingful decline (defined as 3 points). Multiple research groups, studying very diverse samples, have concluded that *Corresponding author. Tel.: 410-955-2619; Fax: 410-955-0504. E-mail address: jbrandt@jhmi.edu Alzheimer’s & Dementia 2 (2006) 218 –219 1552-5260/06/$ – see front matter © 2006 The Alzheimer’s Association. All rights reserved. doi:10.1016/j.jalz.2006.03.012