Lead Investigator: Sheng Luo, Duke University
Title of Proposal Research: Development of novel statistical methods to detect treatment effect in Alzheimer’s disease clinical trials over time
Vivli Data Request: 8438
Funding Source: None
Potential Conflicts of Interest: None
Summary of the Proposed Research:
Background
Alzheimer’s disease (AD) is a chronic progressive neurodegenerative disorder that affects 10.7% of people older than 65 in the United States alone. An estimated 6.5 million Americans aged 65 and older are living with Alzheimer’s dementia in 2022.
AD causes impairment in multiple domains (e.g., cognitive, and behavioral). The disease progresses heterogeneously in time and across domains and individuals: decline may be observed in some, but not all health outcomes at any given time interval and the trajectory of progression may vary between different domains, both within and across AD patients. Therefore, no single health outcome reliably reflects the full spectrum of the disease severity and progression. Clinical trials of AD repeatedly collect multiple health outcomes to obtain an overview of disease progression of AD patients. To objectively assess the treatment efficacy in clinical trials, one needs to account for the trajectory of the treatment effect in terms of multiple outcomes overtime. There are three limitations of current research in assessing the treatment efficacy: (1) the test procedures are usually applied to the change outcome data from baseline to the last observation to test the treatment efficacy, ignoring the treatment effect and observed data over the whole treatment period; (2) most available tests are not directional. Those tests focus on whether there are differences between two groups and cannot tell if treatment is efficacious as compared to the placebo; and (3) the traditional test procedures are based on normal distribution assumptions on outcome variables. Hence, these tests are not appropriate for outcomes that are of different types (e.g., ordinal variable such as pain scale from 0 to 10, binary yes/no questions).
Therefore, we aim to develop novel statistical methods to detect treatment efficacy in AD clinical trials with multiple outcomes over time. The proposed test will compare the sums of ranking of each outcome variable at each time point for all individuals in either treatment or control groups. This test procedure does not require any distribution assumptions (e.g., normal distribution) on the outcome variables and it objectively evaluates the effect of a new therapy overtime by fully utilizing the whole trajectories of the multiple outcomes. Furthermore, we aim to develop novel methodologies to address relevant issues in AD clinical trial designs under the proposed test framework.
Statistical Analysis Plan:
We select (or search for) a specific study based on considerations from 3 aspect: (1) the patients are in early stage of AD, since the treatment is most likely to be efficacious for patients in early stage of AD; (2) in each study, the multiple outcomes had been repeatedly measured at different time points; (3) there are more than one treatment group, because we also want to assess the treatment effect from multiple groups simultaneously. For this criterion, we selected these four completed AD clinical trials. For example, in the study of ABBA-8E12 (ID: NCT02880956), multiple outcomes (CDR-SB Score, Cmax, ADAS-Cog-14 and so on) were measured at baseline, Week 24, Week 48, Week 72 and Week96. There are three treatment groups: ABBA-8E12 300 mg, ABBA-8E12 1000 mg and ABBA-8E12 2000 mg. We will analyze the data from the four studies separately to maintain the structure and the independence of the studies.
Given the multivariate outcome data across time in each study, we will develop a novel rank-based test procedure. Specifically, we will rank the measure of each outcome for all the individuals from both control and treatment groups. After obtaining the mid-ranks of the outcomes at each time point, a test statistic, containing the information of ranking difference between the two compared groups, will be proposed to test whether treatment is efficacious, as compared to the control group. We will investigate the theoretical properties of the proposed test statistics and resolve multiple design issues including sample size and power calculation, missing data handling, interim analysis for efficacy and futility. The test statistics will be applied to the four AD clinical trials.
Requested Studies:
A Phase 2 Multiple Dose, Multicenter, Randomized, Double-Blind, Placebo-Controlled Study to Evaluate the Efficacy and Safety of ABBV-8E12 in Subjects With Early Alzheimer’s Disease
Data Contributor: AbbVie
Study ID: NCT02880956
Sponsor ID: M15-566
An Extension Study of ABBV-8E12 in Early Alzheimer’s Disease
Data Contributor: AbbVie
Study ID: NCT03712787
Sponsor ID: M15-570
Public Disclosures:
Zhou, X., Zou, H., Lutz, M.W., Arbeev, K., Akushevich, I., Yashin, A., Welsh‐Bohmer, K.A. and Luo, S., 2024. Assessing tilavonemab efficacy in early Alzheimer’s disease via longitudinal item response theory modeling. Alzheimer’s & Dementia: Translational Research & Clinical Interventions, 10(2), p.e12471. Doi : 10.1002/trc2.12471