Design
This blood biomarker study is built on the design of the ACT Trial, which is a single-blinded (participants are blinded to their treatment groups), 2-site (University of Minnesota [UMN]; and the University of Rochester [UR]), 2x2 factorial RCT to test the efficacy and synergistic effects of a 6-month ACT regimen on cognition and relevant mechanisms (aerobic fitness, cortical thickness, and functional connectivity in the default mode network) in older adults (age 65 years and older) with aMCI [26]. Details of the ACT Trial have been published [26]. The ACT Trial will randomize 128 participants equally to 4 arms (n=32 per arm): aerobic exercise (cycling) only; speed of processing (SOP) cognitive training only; ACT; or attention control for 6 months. All participants will then be followed for another 12 months. Cognition, including executive function and episodic memory, will be assessed at baseline, 3, 6, 12, and 18 months. MRI biomarkers (AD-signature cortical thickness and hippocampal volume) will be assessed at baseline, 6, 12, and 18 months. Because the ACT Trial does not collect blood samples, our ancillary study will enroll ACT Trial participants, perform blood collections and measure blood biomarkers at baseline, 3, 6, 12, and 18 months. The ACT Trial is registered with ClinicalTrials.gov (NCT03313895). All items from the World Health Organization Trial Registration Data Set can be found in the ACT Trial's registration.
Prior to any recruitment, the ACT Trial statistician created a randomization schedule in REDCap using a random number generator. Randomization was stratified by age (65-74, >75) and site (UMN, UR) in random blocks of 4 or 8. After completing baseline data collection, participants were randomized. The ACT Trial project manager then logged into the REDCap randomization module and revealed the participant’s group allocation. The project manager was the only person who can access the randomization module. The ACT Trial investigators and data collectors are blinded to group allocation, and participants are reminded to not discuss their assigned activity with data collectors who re-emphasize with the participants the importance of not discussing their activities during assessments. Under no circumstances, the ACT Trial will reveal allocation and unblinding is not permissible.[26] Also, this ancillary study will not have access to the intervention assignments and will not reveal allocation or unblinding.
Interventions
Participants started their assigned activities (cycling only, SOP training only, ACT, or attention control) within 2 weeks of completing baseline data collection. Each activity includes three weekly, supervised sessions for 6 months or 72 total sessions delivered over 28 weeks which accounts for missing sessions (e.g., due to illness and vacation). One interventionist supervises two to three participants a session. Cycling only includes moderate-vigorous intensity as 50 to 75% of heart rate (HR) reserve and/or 11–15 on the Borg Rating of Perceived Exertion (RPE) scale for 30–50 min per session. Cycling is set at 50–60% of HR reserve or RPE 11–12 for 30 min in session 1, and is increased by 5% of HR reserve (or 1 point on Borg) or 5-min increments as tolerated up to 65–75% of HR reserve (or RPE 13–15) for 50 min per session over time. For the SOP training, the ACT trial uses the InSight online program (Posit Science) consisted of five games (Eye for Detail, Hawk Eye, Visual Sweeps, Double Decision, and Target Tracker). These SOP training games target multiple cognitive processes, primarily in attention and processing speed and become increasingly difficult and require faster reaction times. In the ACT group, the participants cycle first and play SOP games second without time lapse. In the attention control group, participants stretch (i.e., seated movements and static stretches) [27] and engage in mental leisure activities (MLAs) (i.e., online word search, Sudoku, and solitaire games) [23, 27] that are matched to session durations for cycling and SOP training groups, respectively. These activities serve two purposes: i) as controls of the social interaction, computer, and online experience effects of the ACT; and ii) to increase participant retention.
Data entry research assistants, statisticians, the PIs and the study coordinators and the DSMB will monitor data and participant safety. In addition, the ACT Trial has a Data Safety and Monitoring Board (DSMB) to ensure participants’ safety and the integrity and validity of the collected data. Composition and reporting structure of the ACT Trial's DSMB has been published previously [26]. All study-related unanticipated adverse events and any serious AEs are reported to the IRB, DSMB, and National Institute on Aging (NIA), the ACT Trial study sponsor, within 5 business days of the event. Expected adverse events are outlined in the consent forms and are not reported to the IRB unless they persist. The DSMB in conjunction with the NIH office would make the final decision to terminate the trial.
The ACT Trial may terminate study participation temporarily or permanently based on medical circumstances as well as psychological and/or behavioral symptoms. If a participant requests to withdraw, the ACT Trial would terminate all study-related activities immediately. During the course of the trial, the ACT Trial does not limit in what the study participants can or cannot do, except for enrolling in another clinical trial which may affect cognition. There are no provisions (e.g., ancillary and post-trial care) or compensation to participants who suffer harm from trial participation.
Settings
Blood sample collections and processing will be performed using the same standardized protocols at both study sites. All samples will be stored in the -80°C freezer in PI Li's lab. We will perform biochemical analyses of blood biomarkers at the UMN under PI Li’s supervision. All personnel involved in blood sample collection, processing, storage, and biomarker biochemical assessment will be blinded to participants’ ACT Trial treatment arm assignment.
Study population
Recruitment and consent
Recruitment for this ancillary study will begin as soon as the ACT Trial initiates enrollment of its participants. Participants meeting the ACT Trial eligibility criteria [26] will need to meet 2 additional requirements to be eligible for the ancillary study: (i) agree to donate 20 mL at each blood collection (100mL total); and (ii) agree to fast for at least 8 hours (no food or drink other than water and prescribed medicines) before blood collection. The ACT Trial project managers will recruit and screen participants for eligibility and consent to the ancillary study. We will only enroll participants who have consented to be in this study. We will be blinded to participants’ treatment arm in the ACT Trial, and we will not seek to balance the participants in the blood biomarker study in terms of ACT Trial treatment arms.
Sample size and power
All 128 ACT Trial participants are expected to meet these two eligibility criteria. We expect to enroll 94% of ACT Trial’s participants to this ancillary study and have the same attrition rates as the ACT Trial: 25% at 6 months, 30% at 12 months, and 35% at 18 months. As a result, our sample sizes will be 120 at baseline, 90 at 6 months, 84 at 12 months, and 78 at 18 months. Our study is a pilot study and will provide preliminary results to estimate effect sizes for power calculations in a future large-scale study. We evaluated power for all 3 aims using a total sample size of 90 at 6 months and uneven distribution of sample sizes across 4 arms (with the difference in sample size between any 2 arms no greater than 3). For Aim 1, we conducted power calculations based on 5,000 Monte Carlo simulations for the primary hypothesis (Hypothesis 1b), which is to test the main effects of aerobic exercise and cognitive training on blood biomarker levels at 6 months. With a sample size of 90, Aim 1 has 86% power to detect a moderate main intervention effect (Cohen's d of 0.9). Mean plasma NfL levels were decreased by a Cohen's d of 0.9 in response to drug therapy in a neurological condition (multiple sclerosis)[10]. Therefore, a Cohen's d of 0.9 is realistic and achievable for neurological conditions such as MCI and AD. For Aim 2, we want to test the hypothesis of associations between blood biomarker changes and cognitive changes over time using linear regression models, which have the cognitive changes as the outcomes, the blood biomarker changes (in log scale), as well as interventions, age, sex, and a binary indicator for the presence of the APOE e4 allele as the predictors. We calculated powers using the R package ‘powerMediation’. Aim 2 has over 80% power to detect a weak correlation (r=0.3) at a significance level of 0.05. For exploratory Aim 3, we will have the same power as for Aim 2.
Retention
The ACT Trial will use 16 retention and adherence strategies: 10 at the program level to improve safety, enjoyment, comfort, and convenience, and 6 at the staff level to enhance communication, rapport, sympathy, and encouragement [26]. We further contribute to recruitment, enrollment, retention and adherence in both studies by employing 5 additional strategies to address the procedural concerns: training skilled phlebotomists for blood collection to ensure safety and comfort; conducting phone calls after blood collection to address concerns and ensure future participation; utilizing flexible scheduling of blood collection; providing transportation to blood collection; and compensating participants for blood collection.
Assessment of blood biomarkers
Blood sample collection and processing
We will collect blood samples after but within 2 weeks as the corresponding cognitive assessments at baseline, 3, 6, 12, and 18 months of the ACT Trial. Importantly, we will implement the following rules in blood collection in order to reduce pre-analytical variations that could affect biomarker levels: (i) collect blood samples after at least 8 hours of fasting (only water and medications are allowed), at the same time in the morning (between 8:30 am and 10 am) [28], and after the participant has been sitting for at least 10 mins; (ii) obtain information on medications, infection, vascular disease conditions and diets [7]; and (iii) collect blood at least 24 hours after the last intervention session [29, 30] at 3 and 6 months, in order to mitigate any effects on biomarkers (e.g., IGF-1 and BDNF) from the last bout of intervention.
The ACT Trial project managers at the UMN and UR will schedule the blood collections. Before collection, the ACT Trial project managers and staff will complete a blood-draw assessment designed to record information (e.g., infection, vascular disease conditions and diets) that may affect blood biomarkers. The day before the blood collection, the project managers remind participants by phone of their appointments and the fasting requirement. On the day of blood collection, trained phlebotomists will collect blood samples following a venous-blood collection protocol. If a participant has forgotten to fast, the phlebotomist will notify the project managers, who will reschedule the blood draw. The phlebotomist will collect a total of 20 mL of blood, half into a 10-mL plasma (EDTA-treated) tube and the other half into a 10-mL serum tube. At the UMN, blood collections will be performed by trained phlebotomists at the study participants' home. Upon collection, blood specimens will be stored immediately on wet-ice and transported to PI Li's lab at the UMN for processing. A lab technician will process and aliquot these specimens according to an established protocol. Briefly, the plasma and serum tubes will be gently mixed and centrifuged in 4°C using a temperature-controlled centrifuge with a Swing out Rotor at 1439 g for 15 mins. The tubes will be removed from the centrifuge immediately after completion. From the plasma tube, up to eight plasma aliquots of 500 μL each will be made; from the serum tube, up to six serum aliquots of 500 μL each will be made; packed cells from each plasma tube will be transferred into a 2 mL aliquot. The aliquoted samples (i.e., plasma, serum and packed cells separated from the plasma) will be stored in a -80°C freezer in PI Li's lab. At the UR, blood specimens will be performed and processed by staff using the same blood collection and processing protocols and stored temporarily a -80°C freezer locally before being shipped to UMN to PI Li's lab for long-term storage at UMN.
Biochemical Analyses
We will perform biochemical analyses of the following biomarkers: (i) plasma neuropathological and neurodegenerative biomarkers Aβ42, Aβ40, t-tau, p-tau, and NfL; (ii) serum neurotrophic biomarkers BDNF and IGF-1 (IGFBP-3 is included as part of evaluation of IGF-1) [16]; (iii) plasma neurotrophic biomarkers SCACs; (iv) APOE genotypes, at the end of the study to minimize the overall variance in biochemical analyses. The biochemical methods are described as follows:
Plasma Aβ42, Aβ40, t-tau, p-tau, and NfL
Simoa is an ultra-sensitive method coupled with the HD-1 analyzer to measure blood protein biomarkers [31] with high precision [7] and elimination of matrix interferences reported with traditional ELISAs for measurement of Aβ42 [32, 33]. Simoa assays have been recently used in epidemiological studies to measure blood neuropathological biomarkers [7, 10, 34-36]. We have used a Simoa assay to measure plasma t-tau [9]. In this study, we will use commercially available Simoa assays to measure Aβ42, Aβ40, t-tau, p-tau, NfL in plasma samples.
Serum BDNF, IGF-1, IGFBP-3
In this study, we will use commercially available ELISAs (R&D Systems, Minneapolis, MN) to measure BDNF, IGF-1, and IGFBP-3 in serum samples [37-39].
Plasma SCACs
We have used the Biocrates p180 kits to measure plasma metabolites [19, 40]. In this study, we will use the p180 kits to measure 2 SCACs C0 and C5 in plasma samples.
APOE genotypes
We will extract DNA from packed cells stored at -80°C using Puregene® reagents (Qiagen, Germantown, MD) and determine APOE genotypes (e2/e2, e2/e3, e3/e3, e3/e4, e2/e4, or e4/e4) using Taqman® SNP Genotyping assays for rs429358 and rs7412 (Life Technologies, Carlsbad, CA).
Data management
This ancillary study does not need a data monitoring committee as the venous blood collection presents minimally risk to participants' safety and the integrity and validity of the biochemical data is ensured by a Data Quality Monitoring Plan, for which PI of this ancillary study is responsible. In this plan, the PI, the ACT Trial managers, and lab study coordinator and technicians monitor and oversee all the study records, adherence to the protocols, number of blood collection performed, number of samples received, aliquoted, and stored. All paper and electronic forms are identified with a unique ID number only to protect the privacy and the safety of subjects. Paper copies of data and records at the UMN and UR are stored in a dedicated space in PI Li’s lab and CI Lin’s lab, respectively, and kept in locked file cabinets with limited access. Other data including biomarker assessment are stored electronically in Redcap, a data management software in the UMN Academic Health Center’s Information Systems (AHC-IS) servers. Both sites will have access to Redcap and use it for data storage and management. These AHC-IS servers are in a physical secure location on the UMN campus and are backed up nightly, with the backups stored in accordance with the AHC-IS retention schedule of daily, weekly, and monthly tapes retained for 1, 3, and 6 months, respectively. The AHC-IS servers provide a stable, secure, well-maintained, and high capacity data storage environment that meets the requirements for storing even HIPAA-sensitive data. Access to the AHC-IS servers by username and password are required. PI Li works with the ACT Trial managers and lab technicians at both sites to audit data quarterly to ensure accuracy and completeness of data collection and quality.
Statistical analysis
The ACT Trial will share data on cognition, AD-signature cortical thickness, and other relevant information (e.g., demographics and medications) with this ancillary study for data analysis. We will perform appropriate transformation of cognition (episodic memory and global cognition) and AD-signature cortical thickness data. We will use Holm’s approach to adjust for the hypothesis testing of the multiple neuropathological, neurodegenerative, and neurotrophic biomarkers considered here, which will control the family-wise error rate [41]. Despite randomization, it is possible that the ACT Trial intervention groups could differ in important variables (e.g., age, education, medical comorbidities, medication use [collected by the ACT Trial] and APOE genotype). We will compare groups in terms of these variables and adjust for them if significant differences are found. Missing data due to missed collection visits, loss to follow-ups, and dropouts will be recorded and reported. Missing data will not be imputed. The Inverse Probability Weighting method or likelihood-based method will be used for data analysis assuming the missing data are missing at random. Analysis of complete data or using other methods such as the mixture pattern model will be conducted to test the sensitivity of the results to the assumption of missing at random. Although our aims are not focused on age, sex, and APOE genotype, we will consider them as key covariates because they are established risk factors for AD [42]. Thus, we will adjust for these 3 variables as covariates in all analyses to test for significant associations using linear models.
Aim 1: Determine the effects of interventions on blood biomarkers over 6 months in aMCI.
To test Hypothesis 1a and 1b, we will investigate the association between interventions and changes in blood neuropathological or neurotrophic biomarker levels over time. We will use separate linear regression models for testing each hypothesis with the 6-month changes of blood biomarker levels as the outcome variables. Hypothesis 1b is the primary hypothesis. The model for testing the primary hypothesis will include 2 binary indicators, for aerobic exercise and cognitive training, respectively, as the predictors to test the main effects of aerobic exercise and cognitive training. The model for testing Hypothesis 1a will include 2 binary indicators, for aerobic exercise and cognitive training, respectively, as well as their interaction term to test for the synergistic effect of these treatments on blood biomarker level changes, respectively. We will also either include age, sex, and a binary indicator for the presence of the APOE e4 allele as covariates. The p-values from these models will be adjusted across all biomarkers using Holm’s approach.
Aim 2: Evaluate blood biomarkers as surrogate endpoints for predicting cognitive responses to interventions over 18 months in aMCI.
To test the hypothesis in this aim, we will develop longitudinal linear regression models that investigate the association between changes in blood neuropathological or neurotrophic biomarker levels and cognitive changes over time in response to interventions. These models will have the change from baseline for the cognitive responses as the response variable and will include ACT treatment arm, age, sex, presence of the APOE e4 allele, baseline cognition, and a time-varying blood biomarker (e.g., change in a blood biomarker level for a participant between baseline and 3 or 6 months]) as covariates. We will fit a model for each blood biomarker and use Holm’s method for multiple test adjustment. Any significant association between changes in cognition and in biomarker levels would be clinically useful because biochemical assays are much more sensitive to discern changes in blood biomarker levels than cognitive tests to discern changes in cognition for detecting cognitive treatment effects of interventions.
Aim 3 (exploratory): Examine the correspondence between changes in blood and MRI biomarkers in response to interventions over 18 months in aMCI.
The associations between blood and MRI biomarkers at each time point will be evaluated and 95% CIs will be reported. In addition to this analysis, we will adjust for intervention arms as well as the key biological variables. We will also examine longitudinal linear regression models similar to those used for evaluation of the association between blood biomarkers and cognitive responses in Aim 2, which will test the associations after adjusting for covariates, including ACT treatment arm, age, sex, and presence of the APOE e4 allele.
Dissemination
The study findings will be disseminated through publications and presentations. The study will follow the established publication guidelines for authorship (e.g., the International Committee of Medical Journal Editors). The study does not intend to use professional writers. Access to the final trial data set will follow the guidelines of the NIA.