Progress in understanding variability in cognitive responses to cholinesterase inhibitor treatment

Limitations on the duration of clinical trials, and the constraints of participant selection for such studies, have left many unanswered questions regarding the optimal duration of drug treatment for Alzheimer's disease patients, as well as the subgroups of patients that benefit most. Carefully designed observational studies in naturalistic settings can provide important supplementary information to aid clinical decision-making and patient counseling. A paper by Wattmo and colleagues published recently in Alzheimer's Research & Therapy has provided important new information on differential responses to cholinesterase inhibitor (ChEI) treatment in specific subgroups of patients over a 3-year follow-up period. All of the participants in their study were started on one of three ChEIs after their initial assessment, and periodic assessments of cognitive change and the dosage of ChEIs as well as concomitant medications were subsequently recorded. In addition to providing strong evidence of nondifferential effects on cognition of the three ChEIs as used in this practice, the study identified clinically significant differences in the responses of specific subgroups of patients to the initiation of ChEI treatment. Of particular interest to clinicians is the finding that older patients and those with worse cognitive functioning at baseline had a better treatment response. The notion that treatment may be futile in the oldest or the most impaired patients was thus not supported by Wattmo and colleagues' cohort. Additional well-designed naturalistic studies of this type are needed to advance our knowledge of the long-term outcomes obtained with different therapeutic agents, and of the covariates that significantly modify responses to Alzheimer's disease treatments.

Th e study by Wattmo and colleagues published recently in Alzheimer's Research & Th erapy is a welcome addition to the literature on predictors of cognitive decline in Alzheimer's disease (AD) [1]. A large number of risk factors for cognitive decline and AD in the population have been identifi ed and studied in detail [2][3][4][5][6]. In contrast, relatively little is known about factors that aff ect progression of cognitive and functional decline once the diagnosis of dementia is established. Th e median survival time after onset of AD symptoms can be as high as 11 years [7,8], and it is important for both clinical management and caregiver counseling to be able to predict the probable time course of an individual's loss of cognitive and functional abilities. A critical question for clinicians is whether long-term treatment with anti-dementia drugs, beyond the intervals documented in clinical trials, is benefi cial in altering progression rates. Clinical trials cannot supply this information because of their relatively brief duration compared with the number of years that people actually live with the disease. Naturalistic studies, on the other hand, are limited because they are not randomized with respect to drug use and often suff er from attrition and other biases.
Th e meticulous study by Wattmo and colleagues is an important contribution to our understanding of factors that aff ect the cognitive response (Mini-Mental Status Exami na tion and Alzheimer's Assessment Scalecognitive subscale) to initiation and maintenance of cholinesterase inhibitor (ChEI) treatment over 3 years in a large cohort of patients with possible or probable AD [1]. Th e main objective of the study was to compare outcomes for the three diff erent ChEIs (donepezil, rivastigmine, and galan tamine). In contrast to other observational studies in which an untreated reference group was identifi ed [9,10], this study does not address the issue of whether treat ment was helpful compared with no treatment. Outcomes in Wattmo and colleagues' study

Abstract
Limitations on the duration of clinical trials, and the constraints of participant selection for such studies, have left many unanswered questions regarding the optimal duration of drug treatment for Alzheimer's disease patients, as well as the subgroups of patients that benefi t most. Carefully designed observational studies in naturalistic settings can provide important supplementary information to aid clinical decision-making and patient counseling. A paper by Wattmo and colleagues published recently in Alzheimer' s Research & Therapy has provided important new information on diff erential responses to cholinesterase inhibitor (ChEI) treatment in specifi c subgroups of patients over a 3-year follow-up period. All of the participants in their study were started on one of three ChEIs after their initial assessment, and periodic assessments of cognitive change and the dosage of ChEIs as well as concomitant medications were subsequently recorded. In addition to providing strong evidence of nondiff erential eff ects on cognition of the three ChEIs as used in this practice, the study identifi ed clinically signifi cant diff erences in the responses of specifi c subgroups of patients to the initiation of ChEI treatment. Of particular interest to clinicians is the fi nding that older patients and those with worse cognitive functioning at baseline had a better treatment response. The notion that treatment may be futile in the oldest or the most impaired patients was thus not supported by Wattmo and colleagues' cohort. Additional well-designed naturalistic studies of this type are needed to advance our knowledge of the long-term outcomes obtained with diff erent therapeutic agents, and of the covariates that signifi cantly modify responses to Alzheimer's disease treatments. were not dependent upon the particular ChEI. Th e most interesting fi ndings in this report thus involve interactions between certain back ground variables (age, gender, education level, and baseline cognitive scores) and the longitudinal outcomes studied. In particular, older men (age 85) with worse baseline Alzheimer's Assessment Scale -cognitive sub scale performance showed the greatest improvement in cognitive scores after treatment initiation. Although higher education was associated with better baseline cognitive scores, individuals with higher educational attain ment experienced an accelerated rate of cognitive decline over time. Th e general fi nding that older individuals and more severe patients had more improve ment after initiation of treatment is important, since clinicians and family members may be tempted to view treatment as more futile or less benefi cial in these subsets of patients.
Another clinically signifi cant result reported by Wattmo and colleagues is the clear observation of an initial improvement in the responsive subgroups at 6 months, followed by nearly parallel trajectories of decline over the next 30 months. Although cognitive function continues to decline, initiation of ChEI treatment establishes a new, higher, baseline of performance for the individual. Th is fi nding adds further evidence to support the persistent use of ChEI treatment over a long time period [10]. Th e modeling results, and the graphic presentation of these results, also emphasize the point that baseline cognitive scores are highly predictive of cognitive outcomes observed over time. In Wattmo and colleagues' population, even though ChEI treatment initiation produced greater improvement in more severe patients, those improve ments, in general, did not lead to better longterm outcomes compared with patients who started out with milder dementia. Th is fi nding is consistent with the report by Doody and colleagues that a standardized clinician assessment of the annual rate of Mini-Mental Status Examination decline from onset of symptoms to diagnosis -the pre-progression rate -predicted the subse quent trajectory of cognitive and functional outcomes over as many as 7 years of follow-up [11].
In many societies, educational attainment is strongly correlated with pre-morbid IQ. Th e construct of cognitive reserve has been elaborated to explain the phenomenon that individuals with high pre-morbid cognitive functioning perform better on cognitive tests after an AD diagnosis than those with lower pre-morbid functioning. Some studies, including that of Wattmo and colleagues, fi nd that persons with higher educational attainment decline faster after diagnosis than those with lower education [1,12,13]. In our own analysis of progression rates, however, higher education was associated with slower decline [14], a result also seen in some other analyses. Furthermore, we found that education was not even a signifi cant predictor of out comes after accounting for pre-morbid IQ, suggesting that IQ may be a better surrogate for cognitive reserve. Data from the Scottish birth cohort studies indicate that higher IQ measured in childhood, before the individual's ultimate educational status is known, reduces the risk of late-onset dementia [15]. In their discussion, Wattmo and colleagues suggest that the discrepant results for education between their cohort and ours may be attribu table to diff erences in the mean educational level of the participants. Mean education across clinical cohorts varies, as does the relationship among educa tional attain ment, socio economic status, and general cognitive ability. Th is variability argues for the use of pre-morbid IQ instead of education as the predictor of interest. To advance our understanding of how pre-morbid cognitive ability aff ects the course of cognitive decline in AD, it would be highly desirable for AD research groups and clinical centers to include a measure of pre-morbid IQ, either with or without education, in their models of progression. Th is practice would greatly accelerate the development of a consensus regarding the role of cogni tive reserve in both response to treatment and the overall rate of decline after diagnosis.
In conclusion, rigorous observational studies -with careful documentation of both the drug exposure of interest and the outcomes following exposure -are a necessary supplement to data generated in clinical trials. We would argue that studies such as this one authored by Wattmo and colleagues are necessary to make informed treatment recommendations for practice guidelines, and to provide accurate prognostic information to patients and their families. We hope Wattmo and colleagues' study will serve as a model for quality naturalistic AD research, with the caveat that pre-morbid IQ should be included whenever possible as a potential explanatory variable.