RT induces biologically coherent, load-modulated serum metabolomic shifts detectable only through multivariate analysis, with training-status separation achieving AUC=1.00 and load-specific divergence captured by a multivariate signature of choline, glucose, and alanine (AUC=0.94).
Key Findings
Results
Eight weeks of resistance training significantly altered 10 serum metabolites, with five consistently modulated across both high-load and low-load protocols.
17 healthy young men completed an 8-week RT intervention (HL: 80% 1-RM, n=9; LL: 30% 1-RM, n=8)
Both protocols were performed to volitional failure
The five shared metabolites were 3-hydroxyisovalerate, 3-hydroxybutyrate, acetone, isobutyrate, and lactate
These shared metabolites reflect adaptations in amino acid turnover and ketone body metabolism
Univariate comparisons used paired t-tests and one-way ANOVA with Benjamini-Hochberg correction
Results
A Random Forest multivariate model achieved perfect discrimination between trained and untrained states with AUC=1.00.
3-hydroxyisovalerate was identified as the dominant discriminant feature
Models were validated by stratified 5×5-fold cross-validation and 1000-iteration permutation testing
Group separation was assessed by sensitivity, specificity, and AUC
Results
Load-specific metabolic differences between high-load and low-load protocols were detectable only through multivariate analysis, not univariate testing.
Load-specific divergence was captured by an exploratory multivariate signature of choline, glucose, and alanine (AUC=0.94; p=0.008)
None of the three load-discriminating metabolites (choline, glucose, alanine) was individually significant in univariate testing
This finding demonstrates that univariate approaches fail to capture pathway-level shifts defining the metabolomic response to RT in protocols differing in load
Results
Discriminant metabolites mapped onto three well-established metabolic pathways: leucine catabolism, ketone body turnover, and glycolytic-oxidative rebalancing.
Pathway integration was performed independently from the statistical discrimination models
Mapping was done to assess biochemical plausibility of the discriminant metabolites
3-hydroxyisovalerate and isobutyrate relate to leucine catabolism
3-hydroxybutyrate and acetone relate to ketone body turnover
Lactate, glucose, and alanine relate to glycolytic-oxidative rebalancing
Conclusions
Untargeted 1H-NMR spectroscopy of fasting serum was used to acquire metabolic profiles, and the findings are considered hypothesis-generating pending external validation.
Fasting serum profiles were acquired by untargeted 1H-NMR spectroscopy
The study included only 17 healthy young men, representing a small sample
Authors state findings 'are hypothesis-generating and require external validation in independent cohorts before applied implementation'
The study design was an 8-week RT intervention comparing HL (80% 1-RM) and LL (30% 1-RM) conditions
What This Means
This research suggests that resistance training—whether done with heavy weights (80% of maximum) or light weights (30% of maximum) taken to the point of failure—causes measurable and distinct changes in blood chemistry after 8 weeks. By analyzing small molecules (metabolites) in the blood using a technique called NMR spectroscopy, researchers identified 10 metabolites that changed significantly with training. Five of these changed similarly regardless of whether participants used heavy or light loads, pointing to shared biological adaptations such as changes in how the body processes amino acids and produces ketone bodies.
A key finding was that distinguishing between trained and untrained individuals was possible with near-perfect accuracy (AUC=1.00) using a machine learning approach, with one molecule called 3-hydroxyisovalerate being the strongest indicator of training status. More subtly, differences between the heavy-load and light-load training groups could only be detected when analyzing multiple metabolites together (choline, glucose, and alanine as a combined signature, AUC=0.94)—none of these three molecules showed a significant difference on their own when tested individually. This demonstrates that traditional single-variable statistical tests can miss important biological patterns that only emerge when examining combinations of molecules.
This research suggests that the metabolic 'fingerprint' left by resistance training in the blood is both real and nuanced, with some features shared across training intensities and others specific to the load used. However, the study involved only 17 young men, and the authors themselves emphasize that these findings are hypothesis-generating and need to be confirmed in larger, independent groups before any practical applications could be considered.
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Salgueiro D, Martins M, Scherrer G, Valério D, Castro A, Barroso R, et al.. (2026). Serum metabolomic profiling reveals load-specific adaptations to resistance training.. Metabolomics : Official journal of the Metabolomic Society. https://doi.org/10.1007/s11306-026-02514-5