In clinical trials studying Alzheimer’s disease — a complex neurodegenerative condition that gradually impairs cognitive functions — cognitive performance and functional abilities are often assessed together. Understanding these dimensions and how they’re measured in clinical trials is essential in shaping Cytel’s statistical analyses.

Here, we discuss our experience working with a sponsor on a Phase III clinical trial evaluating the safety and efficacy of monotherapy in patients with Alzheimer’s disease and the statistical model we used to analyze the repeated measurements on two co-primary endpoints.

Alzheimer’s disease

Alzheimer’s disease is a complex neurodegenerative condition that gradually impairs cognitive functions. Its onset and progression are influenced by a range of risk factors and some of the most well-established include age, gender, family history, genetic predisposition, and underlying health conditions.

The disease unfolds in distinct stages, each reflecting a different level of cognitive and functional decline. These stages range from mild cognitive impairment to severe dementia, with symptoms worsening as the disease advances.

Evaluation of Alzheimer’s disease in clinical trials

In clinical trials, the severity of impairment is evaluated using various scales, each addressing distinct aspects of cognitive and functional decline. The most effective approach combines both cognitive and functional assessments, as functional abilities are closely tied to cognitive performance.

Understanding these dimensions and how they’re measured in clinical trials is essential in shaping the statistical analyses used. Multiple discussions between stakeholders and the sponsor need to take place to reach a consensus on the appropriate endpoints and statistical methods to be used for the analyses.

Investigating safety and efficacy of monotherapy in patients with Alzheimer’s disease

We recently collaborated with a small biotech company specializing in Alzheimer’s research on a Phase III clinical trial investigating the safety and efficacy of monotherapy in participants with Alzheimer’s disease, followed by a 12-month open-label treatment. This study has been the subject of complementary analyses exploring biomarkers (p-tau181 and p-tau217) and additional comparative effectiveness analyses with external control arms.

Two primary endpoints: ADAS-Cog₁₁ and ADCS-ADL₂₃

To evaluate treatment efficacy in the Phase III trial, we focused on two co-primary endpoints: the ADAS-Cog₁₁ and the ADCS-ADL₂₃, measured at multiple timepoints throughout the study.

ADAS-Cog₁₁: The cognitive assessment

The ADAS-Cog₁₁ is a cognitive subscale that assesses key domains such as memory, praxis, orientation, and language. Scores range from 0 to 70, with higher scores indicating greater cognitive impairment. A more refined version of the ADAS-Cog₁₁, known as the ADAS-Cog₁₃, includes two additional items that assess memory and attention. This new version provides additional sensitivity to change in cognition at earlier stages of AD.

For the primary analysis, ADAS-Cog₁₁ was retained as the primary endpoint. This decision was guided by its use in previous studies evaluating the same investigational product, ensuring consistency and comparability across trials. The added value of the ADAS-Cog₁₃ was also analyzed as an explorative efficacy variable to provide deeper insights into cognitive outcomes.

ADCS-ADL₂₃: The functional perspective

The ADCS-ADL₂₃ scale complements the ADAS-Cog₁₁ by providing a functional perspective that reflects the impact of cognitive decline. It evaluates the ability to perform daily living activities, with scores ranging from 0 to 78, where higher scores reflect better functional ability and thus less impairment.

Cytel’s approach: Analysis with Mixed Models for Repeated Measures (MMRM)

To analyze the repeated measurements on the co-primary endpoints, we employed Mixed Models for Repeated Measures (MMRM). This approach allows the comparison of cognitive and functional changes over time across treatment arms in a robust and flexible way.

In our models, several key risk factors are included to ensure a well-adjusted analysis. These include baseline disease severity, as measured by the Mini-Mental State Examination (MMSE), prior use of standard AD treatments, and geographic region, as fixed effects. Adjustment for baseline values of the ADAS-Cog₁₁ or ADCS-ADL₂₃ scores is considered to account for differences between subjects at baseline. This helps improve the precision of treatment effect estimates and correct for any imbalances between treatment groups. We also include the treatment group indicator along with its interactions with visit timing to capture if and how treatment effects evolve over time.

This method is particularly valuable for multiple reasons. First, it allows controlling for variables that could influence the observed outcomes — like known risk factors — to be able to understand the treatment effect more accurately. Additionally, by using mixed effects models, both the between and within-subject variability over time is accounted for, which is especially important in a heterogeneous condition like Alzheimer’s. Finally, one of the key strengths of MMRM is its ability to handle incomplete data, meaning it can account for missing values without requiring imputation.

The MMRM method supports the generation of individual and group profile graphs over time. These visualizations offer a clear and intuitive way to observe the evolution of treatment effect. They make it easier to compare trends across groups or subjects, and communicate findings in a straightforward manner, both to scientific audiences and to stakeholders who may not be familiar with the statistical details.

Final takeaways

Alzheimer’s disease is the most prevalent neurodegenerative disease and remains one of the most complex challenges in clinical research, requiring robust methodologies to capture both cognitive and functional decline over time. Complementary and adapted clinical scales are essential tools for assessing disease progression, and advanced statistical methods offer a robust and flexible interpretation of the treatment effect.

By leveraging adaptive models, mixed-effects approaches, and sensitivity analyses, we help sponsors generate reliable insights that drive decision-making in neurodegenerative drug research.

Share Via