ICARUS is a monocentric, randomized, controlled, three-arm, parallel-group trial designed to determine whether exercise training load (moderate versus high) modulates the human gut ecosystem, with a primary focus on gut microbial short-chain fatty acid (SCFA) production, and to test whether interindividual gut responses relate to physiological adaptations and potential maladaptations (i.e., overreaching).
STUDY DESIGN: 45 healthy, recreationally active men will be randomized in a 1:1:1 allocation ratio to: (i) a non-training control group, (ii) a moderate-load cycling training group, or (iii) a high-load cycling training group. The intervention lasts 8 weeks and is followed by a 10-day taper period. Outcomes are assessed at four time points: baseline (T1), mid-intervention after 4 weeks (T2), post-intervention after 8 weeks (T3), and post-taper 10 days later (T4).
RANDOMIZATION: Randomization will be stratified by baseline habitual training volume (>3.5 hours versus ≤3.5 hours of exercise per week) using permuted blocks of varying size and a computer-generated allocation sequence. Allocation will be implemented using a secure REDCap web application to support allocation concealment from personnel involved in data collection and analysis.
SETTING AND RECRUITMENT: The trial is conducted at the Exercise and Muscle Physiology Research Group, Department of Movement Sciences, KU Leuven (Leuven, Belgium). Recruitment uses multiple channels, including email invitations, classroom announcements, posters/flyers, and social media.
FAMILIARIZATION AND STANDARDIZATION PROCEDURES: Approximately one week before the intervention starts, participants attend one familiarization session to minimize learning effects and standardize procedures across subsequent assessments. Familiarization includes key performance and physiological testing procedures and instruction on study devices and diaries. A personalized, standardized dietary plan is developed for the three days preceding each assessment period and the assessment days to improve within-participant dietary consistency across time points and to align with specific test requirements (e.g., fiber intake, carbohydrate loading, lactose/mannitol restriction where applicable).
INTERVENTION ARMS - CONTROL GROUP: Participants do not perform a structured training program. To control for expectancy effects, they receive a supervised weekly placebo capsule (1 g granulated sugar) presented as a 'novel plant extract' purported to support gut health and performance. Participants are asked to maintain their habitual exercise patterns and to monitor relevant behaviors as instructed.
INTERVENTION ARMS - MODERATE-LOAD TRAINING GROUP: Participants complete 8 weeks of supervised indoor cycling followed by a 10-day taper. Training is performed on an electromagnetically braked cycle trainer, supervised at the study center. The weekly structure comprises four sessions per week with a fixed sequence of session types (moderate-intensity continuous training, high-intensity interval training, moderate-intensity continuous training, and sprint interval training), and progressive load increases according to a standardized plan.
INTERVENTION ARMS - HIGH-LOAD TRAINING GROUP: Participants complete 4 weeks identical to moderate-load training group, followed by 4 weeks of high-load training in which training volume is doubled relative to MOD by increasing session frequency while maintaining the same intensity distribution.
EXERCISE INTENSITY PRESCRIPTION: Training intensity is individualized using physiological thresholds derived from an incremental cardiopulmonary exercise test with lactate measurements, including power at the first lactate threshold, second lactate threshold, and maximal power output. Training is prescribed based on cycling power output.
ASSESSMENT PERIODS: Each of the four assessment periods spans three consecutive days of home-based and study center-based procedures under standardized conditions and at the same time of day for each participant to limit circadian variability. Assessments integrate biospecimen collection and multi-system phenotyping, including gut-related measures (e.g., SCFAs, gut barrier integrity testing), exercise performance tests, skeletal muscle phenotyping, cardiometabolic testing, autonomic measures, and self-report instruments relevant to recovery and psychological well-being. Stool samples are collected at home and stored frozen prior to transfer to the study center for biobanked storage.
BIOSPECIMEN HANDLING AND STORAGE: Biological samples including blood, stool, skeletal muscle, urine, and saliva are labeled using a study-specific identifier and stored at -20°C of -80°C until analysis and/or future use in accordance with institutional biobank requirements and applicable regulations.
BLINDING AND BIAS MITIGATION: Because of the behavioral nature of the intervention, participants cannot be blinded to group assignment. To reduce expectancy effects, the control group receives placebo supplementation presented as performance-relevant. Where feasible, outcome assessors and laboratory personnel remain blinded to group allocation during data collection and analyses, although some partial unblinding may occur due to staffing constraints.
SAMPLE SIZE AND STATISTICAL PRINCIPLES: The planned sample size is 45 participants (15 per arm), accounting for attrition, and is based on power considerations derived from prior evidence on fecal butyrate responsiveness to endurance training, given the absence of robust human data for the primary circulating SCFA endpoint at the time of planning. Analyses will follow intention-to-treat principles. Group differences over time will be estimated using constrained longitudinal data analysis models that jointly model baseline and follow-up measures while constraining baseline means to be equal across groups, thereby providing ANCOVA-equivalent estimands and accommodating missing data under a missing-at-random assumption via maximum likelihood. A hierarchical testing strategy is planned for the primary and key secondary endpoints to control the type I error rate. Sensitivity analyses may include per-protocol analyses defined prior to database lock.