Metabolic Predictors of Cardiorespiratory Fitness Responsiveness to Continuous Endurance and High-Intensity Interval Training Programs: The TIMES Study-A Randomized Controlled Trial

Abstract

Background/Objectives: Cardiorespiratory fitness (CRF) levels significantly modulate the risk of cardiometabolic diseases, aging, and mortality. Nevertheless, there is a substantial interindividual variability in CRF responsiveness to a given standardized exercise dose despite the type of training. Predicting the responsiveness to regular exercise has the potential to contribute to personalized exercise medicine applications. This study aimed to identify predictive biomarkers for the classification of CRF responsiveness based on serum and intramuscular metabolic levels before continuous endurance training (ET) or high-intensity interval training (HIIT) programs using a randomized controlled trial. Methods: Forty-three serum and seventy intramuscular (vastus lateralis) metabolites were characterized and quantified via proton nuclear magnetic resonance (¹H NMR), and CRF levels (expressed in METs) were measured in 70 sedentary young men (age: 23.7 ± 3.0 years; BMI: 24.8 ± 2.5 kg·m^-2), at baseline and post 8 weeks of the ET, HIIT, and control (CO) periods. A multivariate binary logistic regression model was used to classify individuals at baseline as Responders or Non-responders to CRF gains after the training programs. Results: CRF responses ranged from 0.9 to 3.9 METs for ET, 1.1 to 4.7 METs for HIIT, and -0.9 to 0.2 METs for CO. The frequency of Responder/Non-responder individuals between ET (76.7%/23.3%) and HIIT (90.0%/10.0%) programs was similar (p = 0.166). The model based on serum O-acetylcarnitine levels [OR (odds ratio) = 4.72, p = 0.012] classified Responder/Non-responders individuals to changes in CRF regardless of the training program with 78.0% accuracy (p = 0.006), while the intramuscular model based on creatinine levels (OR = 4.53, p = 0.0137) presented 72.3% accuracy (p = 0.028). Conclusions: These results highlight the potential value of serum and intramuscular metabolites as biomarkers for the classification of CRF responsiveness previous to different aerobic training programs.

Keywords: NMR; endurance training; high-intensity interval training; metabolomics; trainability.

Figure 1

Summary of the experimental design. Participants underwent a clinical anamnesis and consumed a standardized meal. Blood and muscle tissue samples were collected 12 h later in a fasting state. After 72 h, body composition was measured using a full body plethysmograph, followed by measurements of cardiorespiratory fitness (CRF). Participants were randomly distributed into three groups: ET, HIIT, and Control (CO). CRF was reassessed after 4 weeks to adjust the training intensity and five days after the last training session to evaluate the chronic adaptations. Then participants were categorized as Responders or Non-responders to the interventions and their baseline metabolic profile was associated with CRF responses.

Figure 2

Distribution of the CRF individual responses ± 95% confidence intervals (95% CIs) in relation to the minimal clinically important difference (MCID) for continuous endurance training (ET), high-intensity interval training (HIIT), and the control group (CO) in TIMES. The 95% CIs are calculated as the observed response ± 1.96 $\times$ (technical error). The dashed line represents the MCID [1 multiple of the resting metabolic rate (MET)].

Figure 3

Odds ratio for the classification of CRF responsiveness based on baseline serum (A) and intramuscular (B) metabolite levels. The x-axis is represented on a logarithmic scale. * p < 0.05.

Figure 4

Violin plots of baseline metabolite levels for the compounds significantly retained in the models. Data are individual values (log-transformed data) and horizontal lines representing the median (solid center line) and first and third quartiles (dashed lines); * p < 0.05 when compared to Non-responders, adjusted for the training programs. At the baseline, Responders showed higher levels of (A) serum O-acetylcarnitine and (B) intramuscular Creatinine compared to Non-responders (p = 0.012 and p = 0.017).

Figure 5

Baseline metabolomic signature between Responders and Non-responders regardless of the training group [n = 59 for serum (A) and n = 57 for muscle (B)]. Cells represent the mean standardized concentration levels (z-score) for each metabolite (lines) between groups (columns). Yellow and blue colors represent higher and lower concentration levels, respectively.