General Information
Conflict of Interest
Request Clinical Trials
Associated Trial(s):- NCT00679627 - A Randomized, Double-Blind, Placebo-controlled Trial of Long-term (2-year) Treatment of Galantamine in Mild to Moderately-Severe Alzheimer's Disease
- Placebo-controlled evaluation of galantamine in the treatment of Alzheimer's disease: Evaluation of safety and efficacy under a slow titration regimen
- 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
- NCT00253214 - Placebo-Controlled Evaluation of Galantamine in the Treatment of Alzheimer's Disease: Safety and Efficacy of a Controlled-Release Formulation
- NCT00253188 - Efficacy, Tolerability and Safety of Galantamine in the Treatment of Alzheimer's Disease
- NCT00261573 - The Safety and Efficacy of Galantamine in the Treatment of Vascular and Mixed Dementia
- NCT00253201 - Efficacy, Tolerability and Safety of Galantamine in the Treatment of Alzheimer's Disease
Request Clinical Trials
Data Request Status
Status: OngoingResearch Proposal
Project Title: Innovative precision medicine methods in subgroup identification for Alzheimer's disease clinical trials
Scientific Abstract:
Background: Alzheimer's disease (AD) trials often show modest average effects, potentially masking meaningful benefits in specific patient subgroups. Phase 3 trials of galantamine demonstrated biological target engagement but no population-level clinical benefit, suggesting substantial heterogeneity in treatment response.
Objective: To apply data-driven subgroup identification methods to uncover treatment-responsive subpopulations in galantamine trials and advance precision-medicine strategies in AD research.
Study Design: Secondary analysis of pooled, multi-center, randomized, double-blind, placebo-controlled trials of oral galantamine versus placebo with longitudinal follow-up. Mixed models for repeated measures (MMRM) are integrated with nonparametric interaction tree methods to detect treatment-by-covariate interactions and identify treatment-responsive subgroups.
Participants: Adults with probable or possible AD (mild to moderate) or mild cognitive impairment meeting NINCDS--ADRDA and DSM criteria, with baseline and at least one postbaseline assessment (intention-to-treat).
Primary and Secondary Outcome Measure(s): Primary outcome is longitudinal Clinical Dementia Rating--Sum of Boxes (CDR-SB). Secondary outcomes include ADAS-cog11 and Disability Assessment for Dementia (DAD).
Statistical Analysis: Outcomes are modeled using MMRM with fixed effects for treatment, visit, and their interaction, adjusting for baseline severity. Subgroup identification uses an interaction tree--MMRM framework with internal validation to ensure robustness.
Brief Project Background and Statement of Project Significance:
Alzheimer's disease (AD) is a growing public and financial healthcare crisis, affecting approximately 7.2 million individuals in the United States and projected to increase to 13.8 million by 2060 (Alzheimer's Association, 2025).
Despite extensive drug development efforts, effective therapies remain limited, and most AD phase 2/3 clinical trials have demonstrated only modest average treatment effects, constraining their perceived clinical impact. A key contributor to these failures is the substantial heterogeneity in AD disease progression (Davidson et al., 2010). Patients exhibit widely varying rates of cognitive decline and clinical trajectories, which inevitably leads to heterogeneous treatment effects. As a result, population-average analyses based on a traditional "one-size-fits-all" paradigm may obscure clinically meaningful benefits for specific subgroups of patients. This highlights the need for personalized treatment strategies that account for individual characteristics such as genetic variation and disease stage.
Evidence from galantamine trials illustrates this challenge. A recent comprehensive systematic review and meta-analysis reported statistically significant but modest improvements in cognition, global clinical status, activities of daily living, and behavior among patients with mild to moderate AD, with average effect sizes often near or below conventional thresholds for clinical meaningfulness (Lim et al., 2024). Importantly, these trials also documented substantial variability in disease progression and treatment response, with many patients showing little cognitive decline or limited benefit over typical trial durations.
Together, these findings suggest that galantamine trial data provide an ideal setting for subgroup identification. In this project, we will leverage data from multiple large, multi-center randomized galantamine trials to develop and apply rigorous subgroup identification methods. Specifically, we will employ nonparametric interaction tree--based approaches integrated with mixed models for repeated measures, along with complementary regression-based methods, to identify treatment-responsive subpopulations. By uncovering clinically meaningful heterogeneity within trials traditionally viewed as having modest effects, this work aims to repurpose existing AD trial data, advance precision-medicine strategies (Hamburg & Collins, 2010), and inform the design and interpretation of future AD clinical studies.
Specific Aims of the Project:
We propose to apply precision medicine methods to identify subpopulations that may have benefited from investigational treatments despite negative or inconclusive primary outcomes. Unlike conventional confirmatory subgroup analyses that examine a small number of prespecified subsets, exploratory subgroup identification is conducted post hoc to uncover previously unrecognized heterogeneity in treatment effects. Although many methods have been proposed in the machine learning and data mining literature (Lipkovich et al., 2017; Huber et al., 2019), their application in Alzheimer's disease (AD) clinical trials remains limited, especially for complex longitudinal outcomes.
Galantamine trials provide a compelling test case. While galantamine shows statistically significant but modest average benefits in mild to moderate AD, substantial variability in cognitive and functional responses suggests that clinically meaningful effects may be concentrated in specific subpopulations. Identifying these subgroups can clarify for whom galantamine is most beneficial and demonstrate how precision medicine can extract actionable insights from trials with modest overall effects.
Specifically, we aim to: (1) develop novel tree-based methods to identify patient subgroups that benefit from treatment when the overall effect is not clinically meaningful, and (2) apply these methods to galantamine clinical trial data to identify treatment-responsive subpopulations and illustrate their utility in AD research.
Study Design: Methodological research
What is the purpose of the analysis being proposed? Please select all that apply.: New research question to examine treatment effectiveness on secondary endpoints and/or within subgroup populations Develop or refine statistical methods Research on clinical trial methods
Software Used: R
Data Source and Inclusion/Exclusion Criteria to be used to define the patient sample for your study:
Inclusion Criteria:
- Adults meeting clinical criteria for probable or possible AD or MCI based on NINCDS--ADRDA and DSM-III-R/DSM-IV diagnostic standards.
- Enrollment in a randomized, double-blind, placebo-controlled, parallel-group trial comparing oral galantamine with placebo.
- Treatment duration >4 weeks.
- Availability of baseline and postbaseline cognitive or clinical outcome assessments (intention-to-treat or observed-case data).
Exclusion Criteria:
- Studies that were not randomized, not double-blind, not placebo-controlled, or not parallel-group in design.
- Trials with treatment duration <=4 weeks.
- Participants without a diagnosis of AD or MCI or with non-AD primary neurological conditions.
- Data from open-label phases before or after the randomized trial period.
- Participants lacking usable baseline or follow-up outcome data.
Primary and Secondary Outcome Measure(s) and how they will be categorized/defined for your study:
Primary Outcome Measure
Clinical Dementia Rating -- Sum of Boxes (CDR-SB)
Domain: Global clinical severity (cognition + function)
Scale: 0--18 (higher scores indicate greater impairment)
Description:
CDR-SB is a composite clinical measure assessing six domains: memory, orientation, judgment/problem solving, community affairs, home/hobbies, and personal care. The sum of domain scores provides a continuous measure of overall disease severity.
Outcome Definition for Analysis:
- Change in CDR-SB from baseline through Week 78, modeled longitudinally across all scheduled study visits
- Treatment effect defined as the difference in trajectories (or slope of change) between bapineuzumab and placebo within identified subgroups
Assessment Schedule:
- Baseline
- All treatment visits
- Final visit at Week 78
________________________________________
Secondary Outcome Measures
1. Alzheimer's Disease Assessment Scale -- Cognitive Subscale (ADAS-cog11)
Domain: Cognition
Scale: 0--70 (higher scores indicate worse cognitive impairment)
Description:
A clinician-administered cognitive battery assessing memory, language, attention, and praxis.
Outcome Definition:
- Change in ADAS-cog11 from baseline to Week 78, analyzed longitudinally across visits
- Used to validate whether identified subgroups show consistent cognitive treatment effects
________________________________________
2. Disability Assessment for Dementia (DAD)
Domain: Functional ability
Scale: 0--100 (higher scores indicate less functional impairment)
Description:
Assesses basic and instrumental activities of daily living, reflecting patients' level of independence in everyday functioning.
Outcome Definition:
- Change in DAD from baseline to Week 78, analyzed longitudinally across visits
- Used to assess functional relevance of subgroup-specific treatment effects
Main Predictor/Independent Variable and how it will be categorized/defined for your study:
Main Predictor / Independent Variable:
The primary independent variable is treatment assignment, defined as randomized exposure to oral galantamine versus placebo.
Definition and Categorization:
- Binary indicator: Galantamine (active treatment) vs. placebo (control).
- In dose--response and sensitivity analyses, galantamine exposure is additionally categorized by daily dose regimens (e.g., 8 mg/day, 16--24 mg/day, and higher-dose groups, where available) and treatment duration/time point (e.g., 3, 6, 12, and 24 months).
- For subgroup identification, treatment is modeled through treatment-by-covariate interaction terms within a Mixed Model for Repeated Measures (MMRM) and recursively partitioned using interaction tree methods to identify covariate-defined subgroups exhibiting heterogeneous treatment effects.
Other Variables of Interest that will be used in your analysis and how they will be categorized/defined for your study:
Other Variables of Interest:
- Baseline cognitive severity:
o MMSE (0--30; lower scores indicate greater impairment).
o ADAS-cog baseline score (continuous).
o CDR-SB baseline score (continuous; categorical staging: 0.5 = MCI/very mild, 1 = mild, >=2 = moderate/severe).
- Global clinical status:
o CIBIC-plus (ordinal, 1--7; <=4 = no change/improvement vs >4 = worsening).
- Functional status / Activities of Daily Living:
o ADCS-ADL / ADCS-ADL-MCI (continuous).
o Disability Assessment for Dementia (DAD) (0--100; higher = better function).
- Behavioral symptoms:
o Neuropsychiatric Inventory (NPI) total score (0--120; higher = worse symptoms).
- Disease progression:
o MCI to dementia conversion, defined as CDR-SB progression from 0.5 to >=1.0 (binary event).
- Demographics:
o Age (continuous; categorized in sensitivity analyses, e.g., <75 vs >=75).
o Sex (male/female).
- Treatment exposure / follow-up:
o Visit/time point (categorical: baseline, 3, 6, 12, 24 months).
o Dose regimen (e.g., 16--24 mg/day vs other regimens, where available).
- Safety and tolerability:
o All-cause discontinuation (binary).
o Adverse events (binary indicators for nausea, vomiting, dizziness, diarrhea, anorexia, weight loss, abdominal pain, tremor, agitation, headache).
o Mortality (binary).
Statistical Analysis Plan:
Below is our planned analytic workflow for the requested AD clinical trial data, moving from descriptive summaries to bivariate and multivariable modeling, and then to advanced subgroup/precision-medicine analyses consistent with our proposal.
1) Data preparation and cohort definition
- Harmonize variable definitions across trials (baseline covariates, visit windows, endpoint scales, treatment exposure, follow-up).
- Define analysis populations: ITT (primary), safety set (as available), and per-protocol sensitivity set.
- Construct longitudinal outcomes as change from baseline at each visit (primary for MMRM), plus derived summaries (e.g., individual slope) for sensitivity analyses.
- Missing data assessment (patterns by arm/visit); primary analyses rely on likelihood-based methods under MAR; add sensitivity checks.
2) Descriptive analyses
- Participant flow: enrollment, randomization, follow-up completion, discontinuation reasons.
- Baseline characteristics overall and by arm: demographics (age, sex, race/ethnicity), disease severity/stage, baseline cognitive/functional scores (e.g., CDR-SB, ADAS-Cog, ADL), genotype (e.g., APOE4), key biomarkers if available.
- Outcome summaries by visit and arm: means/SDs (and medians/IQR if skewed), trajectories over time, distribution of change scores, and proportion with minimal/no progression.
3) Bivariate (unadjusted) analyses
- Baseline balance checks: standardized mean differences; hypothesis tests as secondary.
- Arm comparisons at key time points: difference in mean change from baseline; effect sizes with CIs.
- Exploratory associations: baseline covariates vs outcomes (correlations, trend tests, stratified trajectories).
- Safety/tolerability (if available): AE rates and discontinuation by arm and by baseline strata.
4) Multivariable (adjusted) analyses
- Primary longitudinal model: Mixed Model for Repeated Measures (MMRM)
o Outcome: change from baseline across visits.
o Fixed effects: treatment, visit, treatmentxvisit; adjust for baseline outcome and key prognostic covariates; unstructured (or appropriate) within-subject covariance.
o Estimate treatment effects at clinically relevant time points via contrasts; report adjusted means, differences, CIs, and p-values.
5) Planned advanced analyses (precision medicine / subgroup identification)
- Interaction Tree integrated with MMRM (IT--MMRM):
o Recursive partitioning driven by treatmentxcovariate interaction signals to identify subgroups with heterogeneous longitudinal treatment effects (Su et al. 2011).
o Internal validation via sample-splitting/CV/bootstrapping; control false discoveries through pruning/validation rules.
6) Reporting
- Pre-specify primary endpoint/time point(s) and key moderators; provide reproducible code, model diagnostics, and sensitivity analyses (missingness, covariance structure, influential sites).
Narrative Summary: Alzheimer's disease (AD) is a rapidly escalating public health crisis. Despite its growing burden, effective treatments remain limited, and substantial heterogeneity in disease progression and treatment response often leads to modest average effects in clinical trials, potentially obscuring meaningful benefits for specific patient subgroups. This project will leverage data from multiple large, multi-center randomized trials evaluating the efficacy and safety of galantamine in patients with mild to moderately severe AD. Using rich longitudinal cognitive and clinical data, we will develop and apply rigorous subgroup identification methods, including interaction tree--based approaches integrated with mixed models for repeated measures to identify treatment-responsive subpopulations and support precision-medicine strategies in AD.
Project Timeline: We plan to complete this study within 12 months. During the first three months following data receipt, we will focus on data familiarization, cleaning, and preliminary descriptive analyses. The proposed methods will then be applied to the data over the subsequent nine months, including model development, validation, and interpretation of results. We anticipate submitting the resulting manuscript within one year of receiving the data.
Dissemination Plan: We plan to submit the manuscript to a leading Alzheimer's disease clinical journal, such as Alzheimer's & Dementia, Alzheimer's Research & Therapy, or the Journal of Alzheimer's Disease. The manuscript is targeted to biostatisticians and clinicians with interests in AD research, clinical trials, and precision medicine.
Bibliography:
Alzheimer’s Association. (2025). Alzheimer’s disease facts and figures. Alzheimer's & Dementia. DOI: 10.1002/alz.70235
Davidson J, Irizarry M, Bray B, et al. (2010). An exploration of cognitive subgroups in Alzheimer's disease. Journal of the International Neuropsychological Society, 16(2), 233-243.
Hamburg MA, Collins FS (2010). The path to personalized medicine. New England Journal of Medicine. 363, 301-4.
Huber C, Benda N, Friede T. (2019). A comparison of subgroup identification methods in clinical drug development: Simulation study and regulatory considerations. Pharm Stat. 18(5), 600-626.
Lim AW, Schneider L, Loy C. (2024). Galantamine for dementia due to Alzheimer’s disease and mild cognitive impairment. Cochrane Database of Systematic Reviews, Issue 11. Art. No.: CD001747.
Lipkovich I, Dmitrienko A, Ralph B. (2017). Tutorial in biostatistics: data-driven subgroup identification and analysis in clinical trials. Statistics in Medicine. 36: 136--196.
Su XG, Meneses K, McNees P, Johnson WO (2011). Interaction trees: exploring the differential effects of an intervention programme for breast cancer survivors. Journal of the Royal Statistical Society: Series C (Applied Statistics). 60, 457-74.