Rate of cognitive decline and mortality in Alzheimer’s disease.
JS Hui, RS Wilson, DA Bennett, JL Bienias, DW Gilley and DA Evans;
Commented by , 17 Dec 2003
Aim of the study
To assess the relationship between rate of cognitive decline and mortality in Alzheimer’s disease (AD).
Method
410 community-living patients with mild to moderate probable or possible AD, MMSE 11 or more, were followed annually over 4 years with structured medical and neuropsychological evaluations. Eighteen cognitive function tests were performed and regrouped for analysis into episodic memory, visuoconstruction, repetition, naming.
Composite z scores were calculated. Rate of change in cognitive function was assessed using a growth curve approach. Cox proportional hazard models were used to examine the association of baseline level of cognition and rate of cognitive decline with risk of death.
Results
354 patients were eligible for analysis, 112 dying during the 4 years of follow-up. Baseline level of cognition was not related to mortality (p=0.12). There was substantial heterogeneity of decline in global cognition (0.56 unit/year, SD 0.41).
Compared to patients with the least decline, the risk of death was increased more than threefold in the subgroup with mild decline, more than fivefold with moderately rapid decline, more than eightfold with the most rapid decline. Significant differences (p<0.01) were found between baseline characteristics of persons who survived or died during follow-up for age (older), sex (men), visuoconstruction deficits, comorbid medical conditions, impaired basic activities of daily living.
Discussion
This study is important in many respects. First the combination of patients with a diagnosis of probable and possible AD, which has been supported by another longitudinal study in terms of rate of dementia progression, nursing home admission and death (1) and the excellent autopsy confirmation rate in the current study (52/54), will facilitate enrollment in future etiologic and therapeutic research.
Second the use of a growth curve approach to changes in cognitive function is a logical extension of the concept of retrogenesis (2). Third, although the heterogeneity in the progression of AD adds difficulties to randomized clinical trials (RCT) aiming at disease modification, a number of factors at baseline were identified in this study, that could be used for stratification (for example equal numbers of men in each treatment arms), enrollment criteria or planned analysis (for example number and control of co-morbid medical conditions or amount of disability at baseline).
It is reassuring that most of the deaths occurred after year 3 of follow-up, since current RCT with the add-on design (adding a novel agent or a placebo to a cholinesterase inhibitor) are of one year duration. The age at baseline has been shown to be a very significant factor in progression in this and other studies (3,4), and current enrollment in RCT is usually between 50 and 90, avoiding the two extremes of rapid progression through disease milestones for early onset familial AD and very late onset AD with major co-morbities.
References
1. Villareal et al, Neurology 2003; 61: 661-667
(Back up to text)
2. Reisberg et al, Int Psychogeriatrics 1999; 11: 7-23
(Back up to text)
3. Brookmeyer et al, Arch Neurol 2002; 59: 1764-1767
4. Wolfson et al, N Engl J Med 2001; 344: 1111-1116
(Back up to text)