Objective To establish the minimal clinically relevant change (MCRC) on the Alzheimer's Disease Assessment Scale-Cognitive (ADAS-Cog) for patients with mild Alzheimer's disease (AD).
Design Cohort study.
Setting 59 recruiting sites for the Alzheimer's Disease Neuroimaging Initiative.
Patients Outpatients with AD in the Alzheimer's Disease Neuroimaging Initiative.
Main outcome measures The authors applied anchor-based MCRC methodology comparing ADAS-Cog change against clinicians' judgement of clinically relevant worsening between baseline and 6 months in four domains: memory and non-memory cognitive performance; Clinical Dementia Rating Scale; and Functional Assessment Questionnaire. The analysis was repeated for the 6–12-month interval. To support these findings, the authors calculated distribution-based measures including half-baseline SD (1/2 SD) and SEM.
Results 181 patients (baseline ADAS-Cog score 18.5±6.4) had ADAS-Cog data at 0 and 6 months. Those undergoing clinically significant worsening on any of the four anchor questions (n=41–47) had an average ADAS-Cog change of 3.1–3.8 points. Similar results were found for the 177 patients with 6–12-month data. The average 1/2 SD for the baseline ADAS-Cog score was 3.2, and the SEM was 3.7.
Conclusions 3 points decline on the ADAS-Cog may be an appropriate MCRC for clinical trials of patients with early AD. However, further studies assessing the MCRC for improvement on the ADAS-Cog, using patient-based judgement as an anchor, and determining the minimal clinically relevant difference between change on two treatments are required.
- Alzheimer's disease
- outcome measure
- minimal clinically relevant change
- parkinson s disease
- clinical neurology
Statistics from Altmetric.com
- Alzheimer's disease
- outcome measure
- minimal clinically relevant change
- parkinson s disease
- clinical neurology
The Alzheimer's Disease Assessment Scale-Cognitive (ADAS-Cog) is a principal-outcome measure for clinical trials in established Alzheimer's disease (AD).1 However, statistically significant changes in ADAS-Cog scores may not equate to clinically relevant changes.2 The minimal clinically relevant change (MCRC) on the ADAS-Cog often used in clinical studies to define responders varies between 3 and 5 points, with a change of ≥4 being recommended by a consensus committee of the FDA.3
An alternative to relying on consensus opinion is to determine the MCRC empirically. A number of methods have been proposed, including: (1) anchor-based techniques, which compare change against independent, clinically relevant indices,4 and (2) distribution-based methods, based solely on the statistical characteristics of the outcome measure. The latter includes half baseline SD (1/2 SD)5 and standard error of the mean (SEM).6 These methodologies have been used to assess the MCRC for outcome measures in several disorders, including pain, osteoarthritis, Parkinson's disease7 and some outcomes measures for AD.8 However to our knowledge these methods have not been applied to the ADAS-Cog. We aimed to determine the MCRC of the ADAS-Cog in patients with mild AD using anchor-based methodology, calculating distribution-based results for comparison.
All subjects were drawn from the Alzheimer's disease Neuroimaging Initiative, a multicentre public-/private-funded longitudinal study investigating adult subjects with AD, mild cognitive impairment and normal cognition (see supplementary material for further details). We included all patients with AD at baseline and ADAS-Cog data at 0 and 6 months; and for confirmation those with AD at 6 months with ADAS-Cog data at 6 and 12 months. Written informed consent was obtained, as approved by the relevant Institutional Review Boards.
All participants underwent baseline and periodic clinical and neuropsychometric assessments (http://www.adni-info.org). Ratings included the ADAS-Cog, a structured scale scored from 0 to 70, with higher scores indicating worse performance1; the Functional Activities Questionnaire (FAQ),9 an interview-based disability measure scored from 0 to 20, with higher scores indicating more disability; the Clinical Dementia Rating Scale (CDR),10 comprising a global staging of dementia with 5 stages: 0, 0.5 (mildly impaired), 1, 2, 3 (most impaired) and a severity score—the CDR sum of boxes—from 0 to 18; and a comprehensive neuropsychological test battery including tests of memory, naming and executive function.
At both the 6- and 12-month follow-up, clinicians were provided with the current and prior test results and asked to judge whether, compared with the prior assessment, there had been any clinically relevant worsening (yes/no) in a number of domains including the following four which were used as anchors: (a) memory, (b) non-memory cognitive function, (c) FAQ and (d) CDR.
We analysed data at screening/baseline and 6 months for all patients with AD at baseline; and 6–12 months for patients with a diagnosis of AD at 6 months for confirmation. Those with missing or incomplete data were excluded. Statistical analyses were carried out in STATA V.10.
Calculation of the minimal clinically relevant ADAS-Cog score change
We examined the mean ADAS-Cog score change and 95% CIs in patients with and without a clinically relevant change on each anchor question. As those judged to have clinically relevant change also included those with more than minimal clinically relevant change, the MCRC was judged to lie above the mean of the unchanged group and below the mean of the clinically changed group. Changes in all groups were compared with the baseline using t tests. We also calculated the ADAS-Cog change score in those who had deteriorated by one stage (0.5 to 1) on the CDR-global score and those who were unchanged (in addition to the clinical judgement of clinically relevant worsening on the CDR as one of the four anchor questions). We also calculated the effect size (ES=mean difference/baseline SD) and standardised response mean (SRM=mean difference/SD of change) of ADAS-Cog change. An ES or SRM of 0.5 has been proposed as appropriate for a MCRC, with an ES or SRM of 0.2 considered a small change and 0.8 a large change.11
To complement these anchor-based methods, we calculated two additional distribution-based measures: 1/2 SD for baseline ADAS-Cog scores; and SEM (SEM =SD*√(1 –Cronbach's α coefficient)) based on the baseline ADAS-Cog scores.
Patient characteristics are shown in table 1. One hundred and eighty-one patients (baseline ADAS-Cog score=18.5±6.4) were included in the 0–6-month analysis; and 177 (baseline ADAS-Cog score=17.8±6.2) in the 6–12-month analysis. Changes in ADAS-Cog score over these intervals are shown in table 2.
For all four anchor questions for the 0-6 month interval, the mean ADAS-Cog change in those with clinically relevant change was 3.1–3.8 points. The mean changes in the clinically unchanged group were 1.9–2.0 points, with the upper 95% CI not exceeding 3. Very similar results were seen over 6–12 months. For both groups and time intervals, for all four anchor questions the change in ADAS-Cog was statistically significantly greater than zero. The ES for the clinically relevant change group was higher (0.4–0.6) than for the unchanged group (0.3), as was the SRM (0.6–0.7 vs 0.3). Those who deteriorated by one stage on the CDR-global scale had a mean change score of 3.98. Compared with the unchanged group, those changing stage on the CDR-global scale had substantially larger ES (0.70 vs 0.26) and SRM (0.83 vs 0.27) than those not changing, consistent with this change being more than an MCRC (table 3).
The SEM for the ADAS-Cog at baseline was 3.7. The 1/2 SD for baseline ADAS-Cog score was 3.0–3.6, with an average of 3.2. One-fifth of the SD at baseline (a small change) was 1.3, and four-fifths of the SD (a large change) was 5.3. Similar results were seen between 6 and 12 months (table 2).
Comparing individuals with/without a clinically relevant change in any of the domains, there were no reliable differences in baseline Mini-Mental State Examination, ADAS-Cog, FAQ score, gender or AChEI use. There was a consistent trend for individuals undergoing a clinically relevant change to have more years of education than those who did not. However, although this reached nominal statistical significance (at the p<0.05 level), this was no longer the case when a correction was made for multiple comparisons (supplementary table).
The ADAS-Cog is a validated and robust scale for measuring change in AD, and remains the regulatory standard outcome for AD trials. To date, however, decisions on the magnitude of change required to show a clinically meaningful change have principally been based on expert consensus. In this study, using several MCRC methodologies, we demonstrate that a 3-point change on the ADAS-Cog score is an appropriate MCRC for mild AD.
Anchor-based MCRC techniques based on memory, non-memory, disability and global function produced very consistent results: the mean ADAS-Cog change in those judged to have clinically relevant change was just over 3 points, with the remainder changing by approximately 2 points. The ES/SRM scores of the ADAS-Cog for those undergoing change were in the ranges typically considered to be appropriate for an MCRC (ie, 0.4–0.7), with values for those not undergoing a clinically significant change being smaller (0.2–0.4). Change by one stage on the CDR-global score was associated with an ADAS-Cog change of 4 points. However, this change is likely to represent more than just a minimal clinically relevant change, not only on clinical grounds but also as reflected by the relatively large ES and SRM scores (0.7–0.8). Distribution-based methods, which we undertook for comparison, provided broadly similar results, with the SEM of the ADAS-Cog being 3.7 and 1/2 SD 3.2.
Determining a cut-off for an MCRC is inevitably a balance: too low a level risks licensing a drug with borderline clinically relevant change, while an overly high MCRC risks not licensing a drug which may provide at least a minimally significant change to patients. On the basis of this study, we suggest that the current FDA requirement for a 4-point change may be too severe and that 3 points is likely to be the most appropriate whole number for an MCRC for patients with early AD.
There are a number of potential limitations of this study. These results are drawn from a relatively small natural-history study and reflect clinically relevant deterioration rather than stabilisation or improvement expected in a treatment trial. The MCRC may well differ in patients with different disease severities.12 Importantly, MCRC calculations reflect average group changes and may not be appropriate to judge changes in individual patients. The MCRC is an important concept for planning and interpreting trials, providing reassurance that the effects of a drug are not only statistically, but also clinically, relevant. Future studies should examine the MCRC score based on patients' rather than clinicians' judgement. Analysis of data from positive clinical trials would also allow both for the MCRC for improvement rather than deterioration, and the related but different concept of minimal clinically relevant difference between changes in two treatment arms to be assessed.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
Funding This work was undertaken at UCLH/UCL, which received a proportion of funding from the Department of Health's NIHR Biomedical Research Centres funding scheme. JMS receives research funding from Alzheimer's Research UK. Data collection and sharing for this project were funded by the Alzheimer's Disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904). ADNI is funded by the National Institute on Ageing, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: Abbott, AstraZeneca AB, Bayer Schering Pharma AG, Bristol-Myers Squibb, Eisai Global Clinical Development, Elan Corporation, Genentech, GE Healthcare, GlaxoSmithKline, Innogenetics, Johnson & Johnson, Eli Lilly and Co, Medpace, Merck and Co, Novartis AG, Pfizer, F. Hoffman-La Roche, Schering-Plough, Synarc, as well as non-profit partners the Alzheimer's Association and Alzheimer's Drug Discovery Foundation, with participation from the US Food and Drug Administration. Private-sector contributions to ADNI are facilitated by the Foundation for the National Institutes of Health (http://www.fnih.org). The grantee organisation is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer's Disease Cooperative Study at the University of California, San Diego. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of California, Los Angeles. This research was also supported by NIH grants P30 AG010129, K01 AG030514, and the Dana Foundation.
Competing interests None.
Ethics approval Provided by Local Review Boards of the ADNI participating centres.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.