Introduction. Comprehensive treatment of Alzheimer's disease and related dementias (ADRD) requires not only pharmacologic treatment but also management of existing medical conditions and lifestyle modifications including diet, cognitive training, and exercise. The Coaching for Cognition in Alzheimer's (COCOA) trial was a prospective randomized controlled trial (RCT) to test the hypothesis that a remotely coached multimodal lifestyle intervention would improve early-stage Alzheimer's disease (AD). AD results from the interplay of multiple interacting dysfunctional biological systems. Specific causes of AD differ between individuals. Personalized, multimodal therapies are needed to best prevent and treat AD. COCOA collected psychometric, clinical, lifestyle, genomic, proteomic, metabolomic and microbiome data at multiple timepoints across two years for each participant. These data enable systems-biology analyses. We report analyses of the first COCOA data freeze. This analysis includes an evaluation of the effect of the intervention on outcome measures. It also includes systems analyses to identify molecular mediators that convey the effect of personalized multimodal lifestyle interventions on amelioration of cognitive trajectory. Methods. A total of 55 participants with early-stage AD from Southern California were randomized into two parallel arms. Arm 1 (control; N=24) received standard of care. Arm 2 (intervention; N=31) also received telephonic personalized coaching for multiple lifestyle interventions including diet, exercise, and cognitive training. COCOA's overarching aim was to gather dense molecular data from an AD cohort to improve understanding of pathophysiology and advance treatment. For the RCT, COCOA's objective was to test the hypothesis that the Memory Performance Index (MPI) trajectory would be better in the intervention arm than in the control arm. The Functional Assessment Staging Test (FAST) was assessed for a secondary outcome. Assessments were blinded. The nature of the intervention precluded participant blinding. Results. The intervention arm ameliorated 2.6 ± 0.8 MPI points (p = 0.0007; N = 48) compared to the control arm over the two-year intervention. Top-ranked candidate mediators included: albumin, propionylcarnitine, sphingomyelin, hexadecanedioate, acetylkynurenine, tiglylcarnitine, IL18R1, palmitoyl-sphingosine-phosphoethanolamine, acetyltryptophan, and IL17D. These individual molecules implicated inflammatory and nitrogen/tryptophan metabolism pathways. No important adverse events or side effects were observed. Conclusions. Clinical trials should include frequent assessment of dense data to maximize knowledge gained. Such knowledge is useful not only in testing a primary hypothesis, but also in advancing basic biological and pathophysiological knowledge, understanding mechanisms explaining trial results, generating synergistic knowledge tangential to preconceived hypotheses, and refining interventions for clinical translation. Data from every trial should allow an intervention to be refined and then tested in future trials, driving iterative improvement. Multimodal lifestyle interventions are effective for ameliorating cognitive decline and may have an effect size larger than pharmacological interventions. Effects may be molecularly idiosyncratic; personalization of interventions is important. Dietary changes and exercise are likely to be beneficial components of multimodal interventions in many individuals. Remote coaching is an effective intervention for early stage ADRD. Remote interventions were effective during the COVID pandemic.

BackgroundMany clinical trials use performance‐based cognitive or functional test results as primary outcome measures. As few as two measurements, typically pre‐ and post‐intervention, can be used to define a cognitive trajectory. Increasing the number and frequency of assessments can increase power to test the primary hypothesis. However, increased assessment frequency may be expensive and burdensome to participants and staff. Frequent neuropsychological testing may also introduce a practice effect or learning effect. Trials should be designed with attention to the costs and benefits of more frequent assessments.MethodWe conducted two clinical trials and performed Monte Carlo simulations. The Montreal Cognitive Assessment (MoCA) and the MCI Screen (MCIS) were used to assess cognition; the Functional Assessment Staging Tool (FAST) was used to assess function. In addition to simulated data, we employed data from the Coaching for Cognition in Alzheimer’s (COCOA) and Precision Recommendations for Environmental Variables, Exercise, Nutrition and Training Interventions to Optimize Neurocognition (PREVENTION) Trials.ResultEffect sizes and measurement variability were similar between the two cognitive screens. More frequent assessments increased power to reject a null hypothesis The major factor driving this improved power was the influence of data from individuals withdrawing from or dropping out of the trial prior to trial end. More frequent testing captured more of the effect of the intervention over time in those individuals, and therefore, increased power for the entire study. The design of the MCIS with rotating word lists facilitated frequent testing without a learning effect. Hypothetically, a more frequently administered assessment might have less precision than a different assessment administered less frequently, and the unit costs of these assessments might differ. Under a wide range of parameters modeled in simulations, the advantages of more frequent testing outweighed decreases in the precision of individual assessments.ConclusionMore frequent testing should be preferred in clinical trials. Use of shorter and/or remote tests may facilitate such frequency, reduce costs, and increase safety during pandemics. In addition to gains in statistical power, dense measurements have other benefits. They may also reveal non‐linear trajectories and better enable longitudinal analyses such as time dependencies and dose‐response relationships.