General Information
Conflict of Interest
Request Clinical Trials
Associated Trial(s):- NCT00236431 - A Randomized Double-Blind Placebo-Controlled Trial to Evaluate the Efficacy and Safety of Galantamine in Patients With Mild Cognitive Impairment (MCI) Clinically at Risk for Development of Clinically Probable Alzheimer's Disease
- NCT00236574 - A Randomized Double Blind Placebo-Controlled Trial to Evaluate the Efficacy and Safety of Galantamine in Patients With Mild Cognitive Impairment (MCI) Clinically at Risk for Development of Clinically Probable Alzheimer's Disease
Request Clinical Trials
Data Request Status
Status: Approved Pending DUA SignatureResearch Proposal
Project Title: Early Cognitive Adaptation and Progression to Dementia in Mild Cognitive Impairment: Analysis of Two Randomized Clinical Trials
Scientific Abstract:
Background: Repeated cognitive testing can produce short-term performance gains that may obscure underlying decline, but the prognostic meaning of early retest responses in mild cognitive impairment (MCI) is uncertain.
Objective: To test whether early retest response predicts progression from MCI to dementia and whether separating retest-related variation from longer-term decline improves risk stratification.
Study Design: Secondary participant-level analysis of two randomized, double-blind, placebo-controlled galantamine trials. NCT00236574 will be used for model development and NCT00236431 for external validation.
Participants: Randomized participants with MCI; primary derivation and validation will use placebo participants who remain dementia-free at the Month 3 landmark.
Primary and Secondary Outcome Measure(s): The primary outcome is time from Month 3 to dementia progression through Month 24, defined by an increase in global Clinical Dementia Rating from 0.5 to >=1. Secondary outcomes include later cognitive/functional decline and consistency across randomized treatment arms.
Statistical Analysis: Early retest response will be derived from baseline-to-Month 3 changes in repeated cognitive measures, accounting for baseline performance and prespecified covariates. Cox models will assess dementia risk. Coefficients will be frozen and externally validated in the second trial. Secondary mixed-effects analyses will separate early retest-related change from subsequent longitudinal progression.
Brief Project Background and Statement of Project Significance:
Mild cognitive impairment (MCI) is clinically heterogeneous: some individuals remain stable for years, whereas others progress to dementia. Repeated cognitive assessment is central to monitoring this transition, but interpretation of early change is complicated by retest or practice effects. Familiarity with test procedures can transiently improve performance, partially masking underlying decline. Prior work has shown that practice effects can be clinically informative: reduced benefit from repeated testing has been associated with poorer cognitive outcomes in amnestic MCI and with greater Alzheimer-related amyloid burden [2,3]. Practice effects may also influence estimated rates of decline and the design and interpretation of Alzheimer clinical trials [4,5].
Two large, parallel, randomized, double-blind, placebo-controlled galantamine trials in MCI (NCT00236574 and NCT00236431) provide a distinctive opportunity to study this problem. Both trials enrolled participants with MCI clinically at risk for probable Alzheimer disease, included repeated cognitive and functional assessments during 24 months of follow-up, and prospectively assessed progression to dementia using the Clinical Dementia Rating (CDR). Their highly similar designs allow one trial to be used for model development and the other for independent external validation rather than relying on internal data splitting. The randomized placebo and galantamine groups also allow assessment of whether the prognostic meaning of early test-performance changes is stable across treatment conditions [1].
The proposed study addresses a generalizable methodological and clinical question: when cognition changes soon after repeated testing, how much of that change is prognostically informative about underlying disease progression? We will characterize an early retest-response phenotype using prespecified repeated cognitive measures, evaluate its association with subsequent CDR-defined dementia progression, and test whether separating short-term retest-related change from longer-term decline improves prediction beyond baseline severity alone.
This work could improve interpretation of repeated cognitive assessments in MCI, reduce misclassification of apparent stability or improvement, and inform future observational studies and clinical trials that use serial cognitive testing. The goal is not to re-estimate galantamine efficacy or identify treatment-responsive subgroups, but to use the randomized trial structure to develop and independently validate a broadly applicable approach to distinguishing early test-response phenomena from clinically meaningful disease progression.
Specific Aims of the Project:
Aim 1. Determine whether early retest response provides prognostic information for subsequent progression from MCI to dementia beyond baseline cognitive and functional severity. We hypothesize that individuals showing less favorable-than-expected early retest response will have a higher subsequent risk of CDR-defined dementia.
Aim 2. Develop the early retest-response model in NCT00236574 and perform strict external validation in NCT00236431, with predictor definitions and model coefficients fixed before validation. We hypothesize that the association will reproduce with acceptable discrimination and calibration in the independent trial.
Aim 3. Evaluate whether the prognostic meaning of early retest response is robust across randomized treatment conditions and whether separating short-term retest-related variation from subsequent longitudinal decline improves risk stratification compared with baseline-only or raw-change models. We hypothesize that retest-adjusted trajectories will provide incremental prognostic information and that the prognostic direction of early retest response will be broadly consistent in placebo and galantamine groups.
Study Design: Meta-analysis (analysis of multiple trials together)
What is the purpose of the analysis being proposed? Please select all that apply.: Develop or refine statistical methods Research on clinical trial methods
Software Used: R, RStudio
Data Source and Inclusion/Exclusion Criteria to be used to define the patient sample for your study:
Data sources: Individual participant-level data from NCT00236574 (GAL-INT-11) and NCT00236431 (GAL-INT-18). No external participant-level data will be pooled with YODA data.
Inclusion criteria for both trials:
Randomized participant in the trial-defined MCI population;
Baseline global CDR of 0.5, consistent with the enrolled MCI population;
At least one postbaseline assessment relevant to cognitive/functional follow-up.
For the primary Month 3 landmark analysis, participants must additionally be alive and free of dementia at Month 3 and have sufficient baseline/Month 3 data to derive the prespecified early retest-response measures. Missing values in eligible predictors/covariates will be handled using prespecified missing-data procedures where appropriate rather than post hoc significance-based exclusion.
Exclusion criteria:
Dementia progression (global CDR >=1) before or at the Month 3 landmark for the primary prospective analysis;
No follow-up after Month 3 with which to ascertain the primary outcome;
Data inconsistencies that preclude determination of assessment timing or outcome status.
Participants randomized to placebo will form the primary derivation/validation analyses used to characterize retest response without active-treatment effects. Randomized galantamine participants will be retained for prespecified secondary analyses of transportability across treatment conditions. Sensitivity analyses will use alternative landmarks, including Month 6, when supported by data completeness.
Primary and Secondary Outcome Measure(s) and how they will be categorized/defined for your study:
Primary outcome:
Time from the Month 3 landmark to first progression to dementia through Month 24, defined according to the parent trials as an increase in global Clinical Dementia Rating (CDR) from 0.5 to >=1.0. Participants without documented progression will be censored at their last evaluable follow-up or Month 24, whichever occurs first.
Secondary outcomes:
Change in ADAS-cog/MCI after the Month 3 landmark through Month 24;
Change in ADCS-ADL/MCI after the Month 3 landmark through Month 24;
Change in Digit Symbol Substitution Test (DSST) performance after the Month 3 landmark through Month 24;
Dementia progression using an alternative Month 6 landmark as a sensitivity/prognostic analysis;
Prognostic performance of the early retest-response model within placebo and randomized galantamine groups separately.
The primary publication will clearly distinguish the prespecified primary time-to-event outcome from secondary longitudinal cognitive/functional outcomes. Any changes required because of actual data availability or harmonization will be documented and justified in the final report and manuscript.
Main Predictor/Independent Variable and how it will be categorized/defined for your study:
The main independent variable is early retest response (ERR), defined from change between baseline and the Month 3 reassessment on repeated cognitive measures available in both trials, principally ADAS-cog/MCI and the Digit Symbol Substitution Test (DSST).
For each measure, Month 3 performance will be modeled as a function of baseline performance and prespecified baseline characteristics in the placebo derivation cohort. The difference between observed and expected Month 3 performance will represent the participant's residualized early retest response, with direction harmonized so that higher values indicate a more favorable response. Standardized measure-specific ERR values will be evaluated individually and, when data completeness supports it, combined into a prespecified multidomain ERR score.
The ERR definition, scaling, and model coefficients derived in NCT00236574 will be frozen before application to NCT00236431. Raw baseline-to-Month 3 change scores will be analyzed as comparators.
For the methodological secondary aim, a retest-adjusted longitudinal trajectory will be estimated using repeated postbaseline assessments, separating an early postbaseline shift from subsequent longitudinal slope. This trajectory will be compared with conventional baseline-only and raw-change measures for association with subsequent dementia progression.
Other Variables of Interest that will be used in your analysis and how they will be categorized/defined for your study:
Prespecified variables used to characterize the cohorts and for risk adjustment will include, where available and harmonizable across both trials: age, sex, race/ethnicity, baseline ADAS-cog/MCI, baseline ADCS-ADL/MCI, baseline DSST, duration/onset of cognitive symptoms, first-degree family history of Alzheimer disease, relevant baseline comorbidity indicators, and randomized treatment assignment (galantamine versus placebo).
Baseline cognitive and functional scores will be retained as continuous variables whenever possible. Age will be modeled continuously; nonlinear effects will be assessed using restricted cubic splines if supported by sample size. Treatment assignment will be binary according to randomization. Trial will be treated as a design/stratification variable in analyses combining both studies.
Global CDR will primarily define eligibility and the dementia-progression outcome rather than serve as a component of the early retest-response predictor, reducing overlap between predictor and outcome definitions. Variables unavailable or not comparably defined in both trials will not be included in the primary externally validated model but may be summarized or explored in trial-specific sensitivity analyses. No stepwise significance-based covariate selection is planned.
Statistical Analysis Plan:
1. Data preparation and descriptive analyses
Trial documentation and data dictionaries will be reviewed before analysis. Variable definitions, assessment timing, score direction, and treatment coding will be harmonized across NCT00236574 and NCT00236431. Participant flow, missingness, baseline characteristics, and longitudinal outcome distributions will be summarized by trial and randomized arm. Continuous variables will be described using mean/SD or median/IQR as appropriate; categorical variables using counts and percentages. Standardized differences will be used to describe balance.
2. Primary early retest-response analysis
The primary analysis will use a Month 3 landmark to ensure that predictors precede the outcome. Participants who have progressed to dementia by Month 3 will not enter the primary risk set.
NCT00236574 placebo participants will serve as the derivation cohort. For each prespecified repeated cognitive measure (primarily ADAS-cog/MCI and DSST), Month 3 performance will be regressed on baseline performance and prespecified baseline covariates. Residualized observed-versus-expected Month 3 performance, standardized with a common favorable direction, will define early retest response (ERR). Measure-specific ERRs and a prespecified standardized composite will be evaluated. Raw change scores will provide comparators.
Time from Month 3 to first CDR-defined dementia progression through Month 24 will be analyzed using Cox proportional hazards models. The principal adjusted model will include ERR plus prespecified baseline demographic and cognitive/functional severity variables; covariates will not be selected by stepwise testing. Proportional-hazards assumptions and functional form will be examined. Associations will be reported as hazard ratios with 95% confidence intervals.
3. Independent external validation
The complete predictor definition, scaling parameters, and regression coefficients from NCT00236574 will be frozen before validation. They will then be applied without refitting to placebo participants in NCT00236431. Validation will assess discrimination (Harrell C-index and time-dependent AUC), overall prediction error (Brier score where feasible), and calibration (calibration-in-the-large, calibration slope, and graphical calibration at clinically relevant horizons). Performance will be compared with a baseline-only model and a conventional raw-change model.
4. Treatment-condition transportability
The frozen ERR definition will then be applied to participants randomized to galantamine. Prognostic associations will be estimated within treatment arms and, in combined analyses, by including randomized treatment and an ERR-by-treatment interaction. These analyses are intended to test whether the prognostic meaning of early retest response is stable across randomized treatment conditions, not to identify treatment-responsive subgroups or re-estimate the primary efficacy of galantamine.
5. Retest-adjusted longitudinal trajectory
As a secondary methodological analysis, repeated cognitive scores will be modeled with piecewise linear mixed-effects models that distinguish an early postbaseline shift from subsequent longitudinal slope. Participant-specific components will be used to examine whether retest-adjusted progression is more strongly associated with subsequent dementia than unadjusted longitudinal change. If informative dropout materially affects estimates, joint longitudinal-survival modeling will be used as a sensitivity analysis. This analysis will be performed at the score/domain level and is distinct from item-level latent-trait modeling.
6. Missing data and sensitivity analyses
Missing-data patterns will be described by trial, arm, and visit. For eligible participants, missing baseline covariates or early predictors will be addressed using multiple imputation when assumptions are reasonable; complete-case analyses will be reported as sensitivity analyses. Outcome censoring will occur at the last evaluable assessment or Month 24. Sensitivity analyses will include a Month 6 landmark, alternative ERR definitions, exclusion of major protocol deviations where identifiable, and competing-event analysis for death before dementia if event timing permits.
Primary inference will focus on the prespecified ERR association and external validation. Secondary and interaction analyses will be interpreted with effect sizes, confidence intervals, and attention to multiplicity rather than isolated P-value thresholds. All analyses will be conducted reproducibly in R/RStudio within the secure YODA platform.
Narrative Summary: People with mild cognitive impairment (MCI) vary widely in their risk of progressing to dementia. This study will use individual participant data from two randomized clinical trials with repeated cognitive and functional assessments over 24 months. We will examine whether early changes during repeated testing can identify individuals at higher risk of subsequent dementia and whether these early changes reflect retest-related adaptation, underlying disease progression, or treatment effects. One trial will be used for model development and the second for independent validation. The study may improve interpretation of repeated cognitive assessments and provide a practical, low-cost approach to identifying MCI patients at greatest risk of progression.
Project Timeline:
Anticipated project start: October 2026, assuming data access is granted by that time.
Months 1-2 (October-November 2026): Review trial documentation; harmonize variables and visit timing; define analysis cohorts; characterize missingness and participant flow.
Months 3-5 (December 2026-February 2027): Derive the early retest-response phenotype in NCT00236574; complete primary landmark survival models and model diagnostics.
Months 6-7 (March-April 2027): Freeze the derivation model and perform external validation in NCT00236431; conduct treatment-condition transportability analyses.
Months 8-9 (May-June 2027): Complete retest-adjusted longitudinal modeling and prespecified sensitivity analyses.
Months 10-11 (July-August 2027): Draft manuscript, internal review, and first submission to a peer-reviewed journal.
Month 12 (September 2027): Complete revisions as appropriate and report study outputs/results to the YODA Project. If access is delayed, milestones will shift relative to the date of data availability while remaining within the 12-month DUA period where feasible.
Dissemination Plan:
The primary product will be a peer-reviewed manuscript describing the prognostic value of early retest response, its external validation across two independent MCI randomized trials, and implications for interpreting repeated cognitive testing. The intended audience includes neurologists, dementia researchers, neuropsychologists, clinical trialists, and biostatisticians.
Potential journals include Alzheimer's & Dementia, eClinicalMedicine, Neurology, or Alzheimer's Research & Therapy, depending on the final scope and results. Findings may also be presented at a scientific meeting such as the Alzheimer's Association International Conference (AAIC).
Reporting will follow relevant guidance for prognostic model and secondary clinical-trial analyses. Results will be presented regardless of whether the primary hypotheses are supported. No participant-level data will be disclosed outside the secure YODA environment. Study results and publications will be reported back to the YODA Project in accordance with the Data Use Agreement.
Bibliography:
- Winblad B, Gauthier S, Scinto L, et al.; GAL-INT-11/18 Study Group. Safety and efficacy of galantamine in subjects with mild cognitive impairment. Neurology. 2008;70(22):2024-2035. doi:10.1212/01.wnl.0000303815.69777.26 IF: 8.9 Q1 B1.
- Duff K, Lyketsos CG, Beglinger LJ, et al. Practice effects predict cognitive outcome in amnestic mild cognitive impairment. Am J Geriatr Psychiatry. 2011;19(11):932-939. doi:10.1097/JGP.0b013e318209dd3a IF: 5.1 Q1 B2.
- Duff K, Hammers DB, Dalley BCA, et al. Short-Term Practice Effects and Amyloid Deposition: Providing Information Above and Beyond Baseline Cognition. J Prev Alzheimers Dis. 2017;4(2):87-92. doi:10.14283/jpad.2017.9 IF: 6.7 Q1 B3.
- Jacobs DM, Ard MC, Salmon DP, Galasko DR, Bondi MW, Edland SD. Potential implications of practice effects in Alzheimer’s disease prevention trials. Alzheimers Dement (N Y). 2017;3(4):531-535. doi:10.1016/j.trci.2017.08.010 IF: 7.4 Q1 B2.
- De Simone MS, Perri R, Rodini M, et al. A Lack of Practice Effects on Memory Tasks Predicts Conversion to Alzheimer Disease in Patients With Amnestic Mild Cognitive Impairment. J Geriatr Psychiatry Neurol. 2021;34(6):582-593. doi:10.1177/0891988720944244 IF: 2.8 Q2 B4.
- Duehring JA, Jacobs DM, Thomas ML, et al. Implications of practice effects for the design of Alzheimer clinical trials. Alzheimers Dement (N Y). 2025;11(3):e70154. doi:10.1002/trc2.70154IF: 7.4 Q1 B2.