Influence of nutrition pattern on exercise performance, inflammation and muscle damage biomarkers in a non-athlete healthy young cohort

Abstract

Physical exercise induces muscle damage and inflammation, particularly in non-trained individuals, leading to reduced performance. This study explores the influence of dietary patterns on exercise outcomes and systemic physiological biomarkers in this population. This is an observational study complemented by prospective short-term follow-up to assess the influence of diet on exercise-induced inflammation and muscle damage. Recreationally active volunteers (45 females and 33 males) answered a food frequency questionnaire corrected by FETA. A step-exercise was performed until exhaustion and plasma samples were obtained before (basal) and 2 h and 48 h post-exercise. Muscle damage biomarkers (creatine kinase, CK; lactate dehydrogenase, LDH activities) were evaluated through commercial kits and spectrophotometry, and 8 cytokines were assessed by multiplex ELISA. Principal components analysis (PCA) was applied to cytokine to derive systemic inflammatory scores, and muscle damage biomarkers were standardized to comparable scales. Structural equation modeling was then used to evaluate the latent nutritional pattern and its associations with inflammation, muscle damage, and exercise performance. A dietary pattern characterized by higher intake of proteins, fats, and carbohydrates positively influenced physical performance through muscle mass. PCA evidenced 2 inflammation scores (PC1 and PC2), which explained most of cytokine´s variance with opposing correlations with nutrients. PC1 had a negative correlation with proteins, unsaturated fats, folates and vitamin D, while PC2 had positive correlations with simple sugars, saturated fats, insoluble fiber and folates. Exercise volume influenced early systemic inflammation but had no effect on CK or LDH. No sex differences, other than muscle mass, were detected in the population. In healthy young non-athlete population, nutrient-rich dietary patterns appear to enhance exercise performance through muscle mass in both sexes. Healthy fat intake is a relevant factor modulating these responses and plays a role in inflammation and recovery.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-37363-8.

Introduction

Exercise is fundamental for human health, promoting physical fitness, reducing mortality and improving quality of life, demonstrated by a plethora of studies¹. However, physical exhaustion, particularly unaccustomed or strenuous exercise, especially involving eccentric contractions, damages skeletal muscle. High mechanical loads cause disruption of muscle fiber contractile elements, as well as to the cell membrane, resulting in increased membrane permeability. Consequence of this micro-trauma, extravasation of muscle proteins into the blood stream occurs, commonly creatine kinase (CK), lactate dehydrogenase (LDH), or myoglobin, which can be evaluated systemically and are often used as biomarkers reflecting muscle fiber damage². Besides, some chemotactic factors and myokines are released to extracellular space, which in turn induces the infiltration of immune cells and a transient inflammatory response is initiated³. Cytokines excreted by trauma are key biomarkers in this response and its resolution, particularly IL-6 and IL-10^4,5 and, if physical activity is vigorous, TNFα, MCP-1 and IL-1 related family are also released^6,7. A consequence of the muscle damage is delayed onset muscular soreness (DOMS), which includes pain, edema, weakness and limited muscle movements. These symptoms affect performance in the subsequent training session^8,9, which has promoted investigation on methods to reduce pain and accelerate recovery to restore functional capacity¹⁰.

Nutrition is an important external modulator of exercise performance and inflammation. Evidence demonstrates that adequate timing and proportion of macronutrients intake contributes to augment protein synthesis and muscle mass, enhancing recovery and tissue repair^11,12. Besides, the influence of diet on inflammation is well recognized¹³. In the context of sport, specific dietary components have been shown to attenuate markers of muscle damage and reduce post-exercise inflammation. Among them, omega-3 polyunsaturated fatty acids, antioxidant-rich foods or supplementation with creatine, vitamin D^14,15, hydroxy-methyl-butyrate or L-carnitine¹⁶ have been demonstrated. The evidence is limited by small trials mostly conducted in athletes, and few studies have simultaneously evaluated dietary patterns and inflammatory and muscle-damage biomarkers in non-trained population.

Despite the growing interest of the population of all ages in physical exercise and the awareness of nutrition for wellbeing, much less explored is the influence of dietary patterns on exercise performance and recovery in people not engaged in competitive sports. This gap of information prompted the present study, aiming to evaluate how nutrition patterns modulate the capacity to perform exercise and the resulting muscular damage and inflammation in a population of individuals who perform sport for leisure, analyzing the influence of sex. To achieve the objective of this study, structural equations are proposed, since they are useful for modeling complex relationships between dietary behaviors, biological factors, and health outcomes, particularly in the context of inflammation^17,18.

Methods

In this study it was hypothesized that nutritional pattern, as latent variable, has an impact on muscle mass, influencing performance, as well as on systemic inflammation and muscle damage in the early (active damage) and late (recovery) post-exercise period (Fig. 1). The study was designed and recruited the population to test this hypothesis.

Fig. 1.

Theoretical model on the modulation of the latent nutritional pattern on physical exercise performance and how this can influence muscle damage and inflammation in the early and late (recovery) post-exercise period.

Study design and population

The present study has the approval of the Research Ethics Committee of Universidad Autónoma of Madrid (Ref. CEI-136–2901, approved on February 9th, 2024) and the processing of personal data is subject to the Organic Law 3/2018 of 5 December and by the EU Regulation 2015/2283 which safeguards the security of the participant. All methods were performed in accordance with the relevant guidelines and regulations, including the Declaration of Helsinki and applicable institutional and national research ethics standards. In addition, this observational study focused on nutrition follows the STROBE-nut checklist¹⁹; additional file 1).

Participants were recruited at the Faculty of Medicine among students from several degrees of Universidad Autónoma de Madrid (Madrid, Spain). Recruitment was performed through non-probabilistic convenience sampling, and no priori sample size calculation was performed. A team member informed about the study and those interested were called to the first appointment, during which the volunteers signed an informed consent and suitability for the study was evaluated (appointment 1). The participants filled in an ad-hoc sociodemographic questionnaire (the Spanish version used in this article can be found in additional file 2), assessing age, educational level, economic situation and health, and a physical activity level questionnaire²⁰to estimate the weekly total activity by kg (in metabolic equivalents task, METs; kcal; the Spanish version can be found in additional file 2). The reliability of the physical activity level questionnaire has been reported in young adults, comparison by accelerometer data (intraclass correlation coefficients: 0.49–0.60²⁰;. Thereafter, body composition was obtained through a scale-tallimeter (Tanita WB 380-H; BioLogica, Barcelona, Spain) and impedanciometry (Bodystat 520 tetrapolar electrode configuration, 50 kHz, ± 2Ω for resistance, ± 1Ω for reactance; BioLogica, Barcelona, Spain). Previous studies show small bias comparing body composition by electrical impedance and X-ray absorptiometry²¹. From these parameters body mass index (BMI, kg/m²), lean, muscle and fat mass were recorded. In addition, the waist and hip circumference were measured, and the ratio (WHR, arbitrary units) was calculated. These circumferences were measured following the National Institutes of Health Anthropometric Measurement Guidelines²² by a trained nutritionist. The waist circumference was measured with participants standing upright, feet together, arms relaxed at the sides, and abdomen in a relaxed state. A non-stretchable tape was positioned horizontally at the midpoint between the lowest rib margin and the iliac crest along the mid-axillary line. Measurements were recorded at the end of a normal expiration, avoiding compression of the skin. Hip circumference was obtained in the same standing posture, placing the tape horizontally around the widest portion of the buttocks and ensuring it remained parallel to the floor. In both cases, the tape was applied snugly without compressing the underlying soft tissue. The absolute technical measurement error for the waist and hips was ≤1.0 cm.

Active smokers, participants with recent inflammatory processes, chronic inflammatory conditions (diabetes, or celiac disease, among others) taking sport-related supplements or body composition outside the normal range for sex and age, and those involved in competitive sport were excluded from the study.

A total of 100 participants was recruited, and from them, 78 completed the study, being 45 females and 33 males. The flow chart of recruitment and participation is shown in Fig. 2.

Fig. 2.

Recruitment flow chart. Sample size (n).

Accepted participants were informed by email and were asked to come to two additional appointments (2 and 3), refraining from performing exercise in the 24 h prior to the visit.

During appointment 2 the participants filled Food Frequency Questionnaire (FFQ)²³, validated for Spanish population²⁴. The internal consistency of the FFQ has also been reported adults, being Cronbach´s α = 0.947–0.962 and high reliability comparison by 24-hours dietary recalls data (intraclass correlation coefficients: 0.75²⁵;. This version can be found in additional file 2. The FFQ data were recorded manually in a data frame. To minimize transcription errors ensuring accuracy, two independent researchers entered all responses and random spot checks of ~ 10% of entries against the original questionnaires were performed. Thereafter, the analysis was carried out using FFQ EPIC Tool for Analysis (FETA)²⁶ to calculate daily nutrient intakes, including: carbohydrates (g) and single sugars (g), proteins (g), fats (g), saturated fatty acids (SFAs; g); Monounsaturated fatty acids (MUFAs; g), polyunsaturated fatty acids (PUFAs, g), insoluble fiber (g), total folates (µg), and vitamin D (µg) intakes. Nutritional recommendations were described based on dietary reference intakes (RDIs) of databases for age and gender²⁷.

The exercise protocol consisted of up and down stepping cycles on a plyometric box (Vevor, Madrid, Spain) with a height approximately the lower leg length, as previously described²⁸ with some modifications. In particular, the participants were instructed to perform a longer step-down, to increase the eccentric component. The exercise was guided through a metronome established at 11 cycles/min instructing the volunteer to perform the step-down phase in the double time of the step-up phase. The exercise was executed first with the dominant leg until exhaustion, followed by a 5 min recovery and then with the other leg also until exhaustion. Every 2 min a team member asked about the degree of exhaustion though a Borg scale²⁹, and the number of repetitions (1 repetition = 1 up and down step) were annotated with a counter. Another team member oversaw the correct performance of the exercise and verbally encouraged the volunteer to complete the exercise. The exercise was discontinued when participants reached self-reported exhaustion accompanied by clinical signs such as lower-limb weakness, pronounced dyspnea, or dizziness. Once these criteria were met, the protocol was terminated.

Blood extraction and plasma sample collection

Muscular damage and inflammation biomarkers were assessed in plasma, obtained from a capillary blood sample. Capillary blood samples were obtained by trained personnel under strict hygiene and safety conditions, with a TASSO + M20 device (TASSO Inc.; WA, USA) coupled to a lithium-heparin microtainer tube (BD Vacutainer^®, NJ, USA) from the deltoid region. To facilitate flow the area was previously warmed by friction and disinfected with an alcohol wipe. At the appointment 2 blood was obtained before exercise (basal) and 2 h post-exercise, instructing the volunteer not to eat or drink anything except water during this period. The volunteer was asked to come back for appointment 3, after 48 h to obtain a third blood sample.

Immediately after blood extraction, the samples were centrifuged at 4 °C, 900 g for 10 min and the plasma was stored at −80 °C until use.

Plasma biomarkers of inflammation and muscle damage

To evaluate inflammation, the following plasma cytokines: interleukin (IL)−1α, IL-1β, IL-6, IL-12, TNFα, monocyte chemoattraction protein (MCP)−1, IL-10 and IL-1ra, were assessed with an ELLA™ ultrasensitive multiplex ELISA system (Bio-techne Instruments, MN, USA), using cartridges coated with the detection antibodies against the specific cytokines. Cytokine levels were simultaneously measured in a 30 µL plasma sample. First, the sample was diluted 1:2 in the diluent to load a volume of 50 µL into the cartridge well. The ELLA™ automates simultaneous duplications via microfluidic design and micro-reactors channels. The signal detected by the instrument’s fluorescence reader was used to calculate concentrations in pg/ml. The instrument requires the low and upper limits (LLOQ; ULOQ) of quantification, being IL-1α (0.49; 1880 pg/ml), IL-1β (0.4; 1530 pg/ml), IL-6 (0.28; 2652 pg/ml), IL-12 (0.62; 5890 pg/ml), TNFα (0.3; 1160 pg/ml), MCP-1 (1.52; 5780 pg/ml), IL-10 (0.38; 1446 pg/ml) and IL-1ra (7.37; 4500 pg/ml). The overall variance of cytokines was summarized as inflammatory scores by PCA (see below).

To evaluate muscular damage, CK and LDH activity were quantified using kits, according to the manufacturer instructions (Cromakit, Granada, Spain; sensitivity = 1 U/L = 0.0001 ΔAbs/min) by a spectrophotometric method. All reactions were performed at 37 °C and reading reactions were adjusted to zero against distilled water. The absorbance was performed on a plate reader (Synergy HT Multi-Mode, BioTek, VT, USA).

To determine CK activity (detection limit = 0.878–1300 U/L), 5 µl of plasma was mixed with 250 µl of working reactive, which contain the reagents to quantify the reaction phosphocreatine + ADP to creatine + ATP, and was incubated 2 min. Then, for 3 min, the absorbance was read at 340 nm for each minute. The CK activity was calculated as the difference between absorbances and the average of these differences per minute (ΔAbs/min)×11.3.

To determine LDH activity (detection limit = 2–1500 U/L), 4.2 µl of plasma was mixed with 250 µl of working reactive, which contain the reagents to quantify the reaction pyruvate + NADH + H⁺ to L-lactate + NAD⁺, and was incubated 1 min. Then, for 3 min, the absorbance was read at 340 nm for each minute. The LDH activity was calculated as the difference between absorbances and the average of these differences per minute (ΔAbs/min)×13.6.

To evaluate CK and LDH contribution of the theoretical model, robust z-scores were calculated clustered by sex as the value of each biomarker minus its median divided by the interquartile range (Q1-Q3).

Statistical analysis

To analyze the data, the free software R (ver. 4.5.0) was used through the RStudio (ver. 2025.05.0 + 496, R Foundation for Statistical Computing, Vienna, Austria) interface, using the packages: rio, rstatix, compareGroups, dplyr, ggplot2, corrplot, ggcorrplot, ggpubr, factoextra, psych, ppcor, lavaan³⁰, DiagrammeR, and semPlots.

In plasma variables, the missing data ​​were imputed as 1/2 of the minimum value of the variable, clustered by sex and time. The data were summarized as median and interquartile range (IQR) [Q1; Q3] for quantitative variables and as relative frequency (%) and sample size (n) for categorical variables. To evaluate group differences, a non-paired Mann-Whitney´s U test was used for quantitative variables, and Fisher’s exact test for categorical variables. To determine the correlation between quantitative variables, Spearman correlations were used, and Rho coefficients were extracted.

At basal time and clustered by sexes, the cytokine variables were robust, centered and scaled by the median and IQR to avoid the effect of outliers. Then, the principal components analysis (PCA) was performed to approximate the systemic inflammatory scores as principal components (PCs). PCs with eigenvalues higher than 1.0 were extracted, explaining, at least, 60% of the total variance of cytokines. Sphericity was reported using the Bartlett test and the PCA fit model was reported by Root Mean Square Residual (RMSR: acceptable values are close to 0) index, mean item complexity (> 1 indicates that items are loading on more than one factor) and fit based upon off diagonal values (acceptable values are close to 1). The systemic inflammatory scores were interpreted as PCs with higher explained variance. From the selected PCs, the weights of the cytokines in the rotated matrix were extracted using the varimax method. Muscle damage biomarkers (CK and LDH) were also robust typed to obtain similar scores and generate similarities between plasma indicators.

To determine the veracity of theoretical model, structural equation models (SEM) were used, where the standardized coefficient estimators ± standard error (SE) were extracted. In addition, a SEM was constructed to determine the baseline latent nutritional pattern variable through the observed nutritional variables that showed significance with the systemic inflammatory scores and muscle damage biomarkers. According to the definition of SEM, the first observed variable introduced is estimated = 1.00. Furthermore, all nutritional variables, muscle mass and exercise performance (repetitions) enlisted in the confirmatory SEM were robustly typed to reduce data noise. The standardized coefficients, covariance and the regressions were extracted from this SEM. Causal relations among variables were shown by single-headed arrows and double-headed arrow assumed correlation between variables. To simplify, the residual terms were not shown. A P-value (P) < 0.05 was considered statistically significant, and P < 0.1 as quasi-significant.

Results

Contextual variables of the population

The recruited cohort was 20.0 [18.0; 23.0] years old, 94.9% of whom were Spanish and single, and 14.1% were active working. None of the social or total physical activity variables were significantly different between sexes (Table 1). Regarding body composition, as is biologically plausible for this age group, females had higher body fat and lower lean and muscle mass than males. A significantly lower BMI was also found in females, but all the anthropometric parameters analyzed were normal for age and sex (Table 1).

Table 1.

Social variables, nutrients intake and body composition of the cohort.

	All (n = 78)	Female (n = 45)	Male (n = 33)	P
Age (years)	20.0 [18.0; 23.0]	20.0 [19.0; 23.0]	20.0 [18.0; 23.0]	0.865
Monthly Income
<500 €/month	65.4% (n = 51)	62.2% (n = 28)	69.7% (n = 23)	0.185
>500 €/month	16.7% (n = 13)	13.3% (n = 6)	21.2% (n = 7)
Total activity (METs; kcal/kg/week)	30.6 [21.7; 40.5]	31.2 [23.1; 44.8]	30.0 [20.0; 35.7]	0.305
Nº of meals	4.0 [3.0; 4.0]	4.0 [3.0; 4.0]	4.0 [3.0; 4.0]	0.900
Nº supplements	0.0 [0.0; 1.0]	0.0 [0.0; 1.0]	0.0 [0.0; 1.0]	0.109
Total body water (l)	30.4 [28.0; 37.9]	28.3 [27.0; 29.4]	38.4 [35.9; 39.9]	< 0.001
Weight (kg)	59.5 [52.9; 68.4]	54.9 [50.7; 59.9]	68.5 [63.6; 73.4]	< 0.001
Waist-to-hip ratio (a.u.)	0.75 [0.72; 0.80]	0.73 [0.70; 0.75]	0.80 [0.77; 0.81]	< 0.001
Body mass index (kg/m2)	21.5 [19.8; 23.4]	21.0 [19.0; 22.4]	22.2 [20.4; 23.8]	0.028
Fat mass index (kg/m2)	4.90 [3.60; 5.88]	5.60 [4.80; 6.60]	3.50 [3.00; 4.30]	< 0.001
Lean mass index (kg/m2)	16.0 [15.2; 18.4]	15.4 [14.3; 16.0]	18.7 [17.2; 19.8]	< 0.001
Lean mass (%)	76.5 [72.4; 83.3]	73.2 [70.3; 75.2]	83.7 [81.1; 85.4]	< 0.001
Fat mass (%)	23.5 [16.7; 27.6]	26.8 [24.8; 29.7]	16.3 [14.6; 18.9]	< 0.001
Muscle mass (kg)	20.1 [18.2; 27.5]	18.5 [17.6; 19.2]	28.1 [26.5; 28.5]	< 0.001
Muscle mass (%)	35.3 [33.3; 40.4]	33.6 [31.5; 35.0]	41.6 [39.5; 43.8]	< 0.001
Muscle mass index (kg/m2)	7.40 [6.79; 8.86]	6.90 [6.59; 7.17]	9.04 [8.65; 9.43]	< 0.001
Carbohydrate intake (g/day)	194 [160; 232]	196 [169; 233]	180 [148; 231]	0.561
Protein intake (g/day)	92.3 [81.0; 112]	89.7 [76.3; 105]	93.5 [84.9; 122]	0.121
Fat intake (g/day)	67.3 [57.3; 88.8]	63.7 [58.0; 78.5]	72.9 [56.9; 94.4]	0.433
SFAs intake (g/day)	24.9 [19.2; 31.8]	22.3 [18.9; 31.2]	26.5 [19.3; 32.4]	0.360
MUFAs intake (g/day)	26.5 [22.4; 33.9]	26.3 [22.7; 32.5]	29.0 [20.8; 36.1]	0.541
PUFAs intake (g/day)	11.2 [8.78; 13.5]	10.4 [8.97; 12.9]	11.6 [8.51; 13.8]	0.750
Insoluble fiber intake (g/day)	13.6 [10.3; 16.7]	13.9 [11.2; 17.3]	11.8 [9.46; 14.9]	0.180
Single sugars intake (g/day)	93.9 [74.8; 117]	97.6 [76.3; 124]	82.3 [74.7; 114]	0.291
Folates intake (µg/day)	223 [194; 266]	228 [206; 272]	219 [182; 256]	0.495
Vitamin D intake (µg/day)	3.43 [2.73; 5.44]	3.07 [2.72; 5.28]	3.69 [2.91; 5.53]	0.324

Regarding nutritional data, there were no significant differences in any of the nutrient intakes between sexes (Table 1). Considering dietary recommendations²⁷, carbohydrate consumption in the population was slightly higher than the recommended dietary intake (RDI = 130 g/day). Protein intake was above the reference for both sexes (RDI: female = 46 g/day; male = 56 g/day), indicating that the participants followed a protein-based diet. On the other hand, intakes of folate (RDI = 400 µg/day) and vitamin D (RDI = 15 µg/day) were below the RDIs. Fiber intake was also below reference values (RDI: female = 25 g/day; male = 38 g/day); however, these values are estimates for total fiber and not just insoluble fiber. It was also found that 23.1% of the population did not consume wine or beer, while 60.3% consumed it sporadically.

The data shows the median and interquartile range [Q1; Q3] for quantitative variables and the relative frequency (%) for categorical variables. The P values ​​(P) were extracted by Mann-Whitney´s U test for quantitative variables and Fisher’s exact test for categorical variables. Arbitrary units (a.u.); Metabolic equivalent tasks (METs); saturated fatty acids (SFAs); monounsaturated fatty acids (MUFAs); polyunsaturated fatty acids (PUFAs); sample size (n).

Basal systemic inflammatory and muscle damage biomarkers

Basal plasma levels of cytokines and muscle damage biomarkers can be found in additional file 2 Table S1. Systemic inflammatory scores were estimated by PCA, using plasma IL-1α, IL-1β, IL-1ra, IL-6, IL-10, IL-12, TNFα, and MCP-1 levels. PC1 explained 27.3% of cytokines variance, while PC2 and PC3 explained 17.1% and 16.3%, respectively. The cumulative variance explained by the PCs was 60.8% (Sphericity: χ² = 3228.3; P < 0.001; RMSR = 0.12; mean item complexity = 1.3; fit based upon off diagonal values = 0.7). The rotated loading factors can be found in additional file 2 Table S2. No significant differences between sexes were detected either for PCs or for muscle damage biomarkers (Table 2).

Table 2.

Basal systemic inflammatory and muscle damage scores of the cohort by sex.

	All (n = 78)	Female (n = 45)	Male (n = 33)	P
Inflammatory PC1 (a.u.)	0.08 [−0.42; 0.60]	−0.11 [−0.44; 0.51]	0.31 [−0.24; 0.66]	0.176
Inflammatory PC2 (a.u.)	0.11 [−0.49; 0.45]	0.35 [−0.52; 0.66]	0.06 [−0.49; 0.35]	0.438
Inflammatory PC3 (a.u.)	0.00 [−0.35; 0.49]	0.16 [−0.30; 0.49]	−0.05 [−0.35; 0.35]	0.331
CK (robust Z-score)	−0.21 [−0.74; 0.55]	0.06 [−0.45; 0.73]	−0.56 [−0.84; 0.25]	0.107
LDH (robust Z-score)	−0.10 [−0.78; 0.55]	−0.06 [−1.33; 0.64]	−0.10 [−0.39; 0.55]	0.408

The data show the median and interquartile range [Q1; Q3]. The P values ​​(P) were extracted by Mann-Whitney´s U test. Arbitrary units (a.u.); sample size (n).

Relationship between nutritional variables and basal systemic biomarkers

The basal inflammatory PC1 score showed significant and negative correlations with the intake of protein, fat, MUFAs, PUFAs, total folates and vitamin D, while the inflammatory PC2 score showed significantly positive correlations with the intake of carbohydrates, fats, SFAs, MUFAs, PUFAs, insoluble fiber, simple sugars and total folates (Fig. 3). The inflammatory PC3 score showed no significant correlation with any nutrient. Regarding muscle damage biomarkers, CK correlated significantly and positively with carbohydrate and simple sugar intake. However, LDH did not show significant correlations with any nutrient (Fig. 3). Therefore, PC3 and LDH were excluded from posterior analysis. A SEM confirmed that the latent nutrition pattern was covered by carbohydrates, proteins, fats, SFAs, MUFAs, PUFAs and vitamin D (additional file 2 table S3).

Fig. 3.

Correlation between dietary nutrients with basal systemic inflammatory scores (PC1, PC2 and PC3) and biomarkers of muscle damage (CK and LDH). The color gradient indicates Spearman’s Rho coefficient scale, and significant coefficients (P < 0.05) are shown inside the cells. Principal component (PC); creatine kinase (CK); lactate dehydrogenase (LDH); Saturated fatty acids (SFAs); monounsaturated fatty acids (MUFAs); polyunsaturated fatty acids (PUFAs).

Relationship between exercise and evolution of inflammatory scores and muscle damage biomarkers

The number of repetitions were significantly lower in females (291 [179; 399]) compared to males (458 [324; 694]; P < 0.001). Repetitions correlated positively with muscle mass (ρ = 0.40; P < 0.001). Therefore, when repetitions were relativized to muscle mass, this difference was lost (females = 16.2 [9.8; 21.4] rep./kg muscle, males = 17.1 [12.2; 24.7] rep./kg muscle; P = 0.439).

IL-6, IL-10, IL-1b and CK were significantly elevated to 2 h post-exercise, returning the IL-6 and IL-10 to basal values at 48 h (additional file 2 figure S1).

No significant differences were detected between sex, either in the inflammatory scores (PC1 and PC2) or in the CK biomarker along the different times (Fig. 4A). Similarly, no significant differences were found in any of the scores along time (Fig. 4B).

Fig. 4.

Systemic inflammatory scores (PC1 and PC2) and biomarkers of muscle damage (CK) by sex (A) and time (B). The data show the median and interquartile range [Q1; Q3]. The P values (P) were extracted by Mann-Whitney´s U test. Principal component (PC); creatine kinase (CK). Female = 45, male = 33.

At 2 h post-exercise (active inflammation period), the number of repetitions negatively correlated with the inflammatory PC1 score, while it correlated positively with plasma CK levels. On the other hand, the number of repetitions correlated, with almost significant difference with inflammatory PC2 score (Table 3). However, statistical significance was lost when repetitions were relativized to muscle mass.

Table 3.

Correlations of systemic biomarkers of inflammation and muscle damage with exercise performance at 2 h and 48 h.

	Repetitions		Repetitions/muscle mass
	2 h	48 h	2 h	48 h
Systemic inflammatory PC1 score	−0.44 (P < 0.001)	−0.18 (P = 0.117)	−0.18 (P = 0.120)	−0.15 (P = 0.200)
Systemic inflammatory PC2 score	0.19 (P = 0.098)	0.10 (P = 0.404)	0.17 (P = 0.147)	0.14 (P = 0.228)
CK muscle damage	0.22 (P = 0.033)	−0.15 (P = 0.196)	0.14 (P = 0.225)	−0.19 (P = 0.091)

At 48 h post-exercise (recovery period), no significant correlations were found between repetitions and plasma biomarkers. When adjusted for muscle mass, a nearly significant and negative correlation was found for CK levels (Table 3). Biomarker levels between 2 h and 48 h showed significantly positive correlations (PC1: ρ = 0.42, P < 0.001; PC2: ρ = 0.36, P = 0.002; CK: ρ = 0.65, P < 0.001). At 2 h post-exercise, PC1 and PC2 were significantly and negatively correlated (ρ=−0.35; P = 0.002), whereas inflammatory scores PC1 and PC2 did not correlate with CK. At 48 h post-exercise, PC1 and PC2 maintained their negative correlation (ρ=−0.35; P = 0.001) and were still unrelated to CK.

The data show Spearman’s Rho correlation coefficient and its associated P-value (P).

Effect of nutrition on exercise and its repercussion on inflammation and muscle damage

The SEM model was constructed to explore the effect of nutritional pattern (latent variable) on exercise performance, covaried by muscle mass, on systemic inflammation and muscle damage at 2 h and its recovery at 48 h. Fat, SFAs, MUFAs, and PUFAs intakes showed the strongest loading on dietary pattern. The dietary pattern showed a positive association with the number of repetitions performed (β = 0.10 ± 0.02; P = 0.024). On the other hand, muscle mass had a strong covariation with repetitions (σ = 0.30 ± 0.07; P = 0.007).

The number of repetitions had effects on PC1 and CK at 2 h, and PC1 and CK at 2 h predicted PC1 and CK at 48 h (Fig. 5).

Fig. 5.

Relationship between nutrients, exercise performance and their association with systemic inflammation and muscle damage at 2 h and 48 h post-exercise. The data shows the standardized coefficient of the latent variable (η), covariance (σ), and regression (β). The latent variable is shown in a circle, and the observed variables are shown in boxes. Correlation between variables is shown as a double-headed arrow and causal relations among variables are shown by single-headed arrows. In grey line is shown non-significant value. For simplification, the residual terms are not shown. Muscle mass and nutrients are considered exogenous variables, and repetitions and plasma biomarkers (PCs-related cytokines and CK) as endogenous variables. Saturated fatty acids (SFAs); monounsaturated fatty acids (MUFAs); polyunsaturated fatty acids (PUFAs); Principal component (PC); creatine kinase (CK).

Discussion

Dietary strategies are well established in competitive sports. However, much less is known about the influence of dietary patterns in non-athletes. The present study explored in a young, healthy population, recreationally active, but not engaged in competitive sport, how nutrition patterns modulate the capacity to perform exercise and the resulting muscular damage and inflammation, analyzing the influence of sex. Using SEM, it was found that a dietary pattern characterized by higher intakes of proteins, fats, and carbohydrates positively influenced physical performance through muscle mass. Systemic inflammation, evaluated by 8 cytokines and PCA, evidenced two scores (PC1 and PC2) which explained most of cytokine´s variance and had opposing correlations with nutrients intake. The volume of exercise had a direct influence on early systemic inflammation, but not on enzymatic muscle damage biomarkers.

Relationship between nutrition, muscle performance and muscle mass

Nutrition has been widely studied in the context of sport performance, particularly in competitive sports and in clinical contexts. However, there is scarcity of information on the impact of nutrition on exercise capacity in healthy, recreationally active subjects, but without regular training. This population is highly interested in nutrition and fitness, but also exposed to misinformation, fueled by social media news³¹, sometimes without scientific background, leading to misconceptions. Our study using SEM demonstrated that dietary patterns had an influence on exercise performance with a strong covariation with muscle mass. All main macronutrient categories of the latent variable showed a significant impact on skeletal muscle, particularly fats and carbohydrates, and to a lesser extent protein intake. In general, it has been taken for granted that an increase of protein intake elevates muscle mass, and protein supplementation has gained popularity. However, although athletes may need additional protein, evidence shows that in individuals who perform sport for leisure, supplementation is not required to promote skeletal muscle gain³². SEM revealed that among macronutrients, fats -particularly PUFAs and MUFAs- had the strongest positive association with the latent dietary factor influencing performance. This suggests a synergy between nutrient intake, muscle mass and performance. Literature shows that fat intake is positively associated with muscle mass and strength, with an optimal intake threshold around 1.6–1.9 g/kg/day, beyond which benefits may plateau or reverse³³. On the other hand, muscle mass is one of the strongest predictors of maximal fat oxidation during exercise suggesting that individuals with greater muscular content are better equipped to utilize dietary fats as fuel, thereby supporting endurance and enhancing post-exercise recovery³⁴. Furthermore, some studies suggest that fat intake may be a better predictor of performance than protein when both are examined within dietary patterns. For example, in old adults a dietary pattern with moderate fat and protein, coupled with high intake of whole grains and vegetables, was the most predictive of strength and physical function³⁵. Our results in a young population mirror this observation, reinforcing the importance of examining macronutrients as part of a broader dietary pattern rather than in isolation.

Relationship between nutrition and muscle damage

One of the aims of our study was to evaluate the influence of nutrition on inflammation-induced by physical exercise. Systemic inflammatory scores were estimated by PCA, using plasma levels of 8 cytokines, revealing that two main components (PC1 and PC2), which explained the majority of cytokines variance. Our findings demonstrate an opposing influence of dietary pattern on these components. The PC1 was negatively correlated with intakes of proteins, unsaturated fats, vitamin D and folate, while the PC2 was positively associated with carbohydrates (including simple sugars), all types of fats and folates. During competition, carbohydrates intake seems to reduce inflammation³⁶. However, high glycemic-load diets, especially those low in fiber, are known to increase CRP and IL‑6 levels, whereas complex carbohydrates and fiber-rich diets may dampen inflammatory signaling and support gut-derived short-chain fatty acids production, with immunomodulatory effects³⁷. MUFAs and PUFAs, particularly long-chain omega-3, have also been demonstrated to reduce systemic markers of inflammation, such as IL‑6 and C reactive protein³⁸, while SFAs activate pro-inflammatory transcription factors like factor nuclear κB^39,40. All the above could suggest that the PC1 reflects a more anti-inflammatory profile, whereas PC2 appears to be more pro-inflammatory. However, both components represent a balance of cytokine actions, and their effects should be interpreted as part of a complex, synergistic network rather than in isolation. Thus, for recreationally active young individuals, diets rich in protein, unsaturated fats, and micronutrients, limiting refined carbohydrates and SFAs may help reduce baseline inflammation and could attenuate acute inflammatory responses post-exercise.

Under physiological conditions, the inflammatory cascade triggered by exercise is limited by the subsequent release of anti-inflammatory mediators such as IL‑10 and IL‑1ra, which facilitate recovery⁴¹. Our data supports this dynamic: PC1, potentially representing an anti-inflammatory profile, was negatively correlated with exercise performance in the acute phase but not at 48 h, suggesting a resolving pattern.

Together with inflammation, exercise is associated with muscle membrane disruption and release of proteins, such as CK, myoglobin or LDH, considering the most representative biomarkers of muscle damage^2,9. In fact, the post-exercise increases in CK observed in our study were associated with the number of repetitions, suggesting a direct link between exercise volume and skeletal muscle membrane disruption. The release of molecules related to muscle disruption and inflammation after exercise occurs concomitantly, and it has been taken for granted that they are related events. However, our results show that early CK release and inflammation are dual independent pathways. In fact, in response to exercise distinct mechanisms occur: myofibrillar damage and structural protein degradation resulting in membrane damage with alterations in excitation-contraction coupling, calcium accumulation in the cytoplasm and release of intracellular proteins, which include CK and LDH. On the other hand, the release of myokines, particularly IL-6, leads to the inflammatory cascade that follows^2,42. Besides, it has been shown that plasma CK release and immune cell activation occur at different time points². A third consequence of the immune response cascade is swelling and activation of nociceptors leading to pain and impaired exercise performance in subsequent days^8,9. In the present study, DOMS were not measured, since there is evidence of a relationship between muscle damage, inflammation and soreness⁸, it would be interesting to evaluate the role of diet in this response.

Sex differences in exercise performance and muscle damage

One of the aims of the present study was to assess the influence of sex on exercise induced muscular damage and performance. Sex is a factor usually under-explored in the context of sports science and exercise research, while awareness is growing and recently attracted considerable scientific and public attention^43,44. The population under study was homogeneous in terms of age, socioeconomical level, nutritional behavior and level of physical activity. We found a difference in body composition, with males displaying higher muscle mass, as expected⁴⁵. Performance capacity (evaluated through the number of repetitions until exhaustion) was directly related to muscle mass, since differences between sexes disappeared when adjusted for this variable, and confirming that it is a key determinant of exercise performance⁴³. No significant differences between sexes were found on other parameters, either under basal conditions or after exercise. Baseline systemic cytokines were comparable between sexes. In this respect, the literature is not conclusive, and some studies describe that female have lower levels of pro-inflammatory cytokines in association with lesser oxidative damage, probably due to the protective influence of estrogens⁴⁶, while other epidemiological studies show that females have larger plasma levels of C reactive protein, as inflammatory biomarker³⁶. Our results, evaluating 8 cytokines, shown no sex differences in a well-nourished healthy young population. However, it must be considered that ovarian cycle was or not considered, and impact of the menstrual cycle on immunologic parameters has been demonstrated⁴⁷. Similarly, we did not find sex differences in inflammation components after exercise, either in the early (active inflammatory phase) or late (recovery) periods. It would be desirable to evaluate this aspect in a larger population considering ovarian hormones, which exhibit a role in moderating the exercise-induced immune response in women⁴⁸.

Limitations and future research

Several limitations should be acknowledged when interpreting the present findings. First, the study cohort was composed for young and healthy adults, which limits the generalizability of the results to older populations, or individuals with comorbidities that may influence nutrient metabolism, exercise response, or muscle recovery. Future research should seek to replicate these findings in more diverse populations, including those with metabolic disorders, to establish broader applicability. Second, given the recruitment approach, sampling bias cannot be ruled out, and factors such as menstrual cycle hormones, which could influence inflammation, performance and perceived recovery, should be carefully accounted for in future studies. Third, dietary intake was assessed at a single point, relying on self-reported data, and only the most relevant nutrients were analyzed. Although validated methods were used, single point dietary assessments are prone to recall bias and may not reflect changes in eating behavior. Longitudinal studies assessing dietary tracking, and full dietary pattern analysis could provide complementary insights and enhance reliability. Fourth, the temporal window for biomarker collection was limited to 48 h post-exercise. While this timeframe captures early inflammatory and muscle damage responses, it may not fully encompass the delayed recovery trajectory, particularly in untrained individuals. Extending biomarker follow-up to 72 h would provide a more comprehensive understanding of the physiological processes underlying muscle repair and adaptation. In addition, the use of impedanciometry to estimate fat mass and fat-free mass could be a methodological limitation, introducing systematic bias in body-composition measurements, although the tetrapolar electrode device is the recommended system⁴⁹. However, it is practical and accessible, and the magnitude of error associated with impedance-based estimates is similar to that observed with X-ray absorptiometry⁵⁰. Besides, all participants were assessed using the same impedanciometry device and protocol, and therefore any possible bias was likely applied uniformly across the cohort. Finally, although the structural model revealed robust associations between dietary patterns, muscle mass, and exercise performance, the observational design prevents causal interpretation. To address this, future studies should incorporate randomized controlled trials manipulating dietary fat composition, alongside objective physical performance metrics and time-resolved biomarker analysis.

Conclusions

This study, conducted in healthy a young non-athlete population, suggests that fat intake, particularly healthy fats, such as PUFAs and MUFAs, is a key dietary factor modulating the complex interplay between muscle mass and exercise performance, which may also contribute to control and resolve inflammation and post-exercise recovery. These findings challenge traditional protein-centered paradigms and emphasize the need to incorporate dietary fat as a central component of performance-oriented nutritional strategies. It would be desirable to extend this research by exploring the role of nutrition in other populations with different physical conditions.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

BMI

Body Mass Index

CK

Creatine Kinase

DOMS

Delayed Onset Muscular Soreness

FETA

FFQ EPIC Tool for Analysis

FFQ

Food Frequency Questionnaire

IL

Interleukin

IQR

Interquartile Range

LDH

Lactate Dehydrogenase

LLOQ

Low Limit of Quantification

MCP-1

Monocyte Chemoattraction Protein 1

METs

Metabolic Equivalents Task

MUFAs

Monounsaturated Fatty Acids

PCA

Principal Components Analysis

PCs

Principal Components

PUFAs

Polyunsaturated Fatty Acids

RDIs

Dietary Reference Intakes

RMSR

Root Mean Square Residual

SE

Standard Error

SEM

Structural Equation Models

SFAs

Saturated Fatty Acids

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

TNF

Tumor Necrosis Factor 

ULOQ

Upper Limit of Quantification

WHR

Waist-to-Hip Ratio

Author contributions

J.M. and S.M.A. Conceived and designed research; D.R.-C., R.A.C., P.R.-R., B.S., G.B. and G.D. performed experiments; D.R.-C. and S.M.A. analyzed data, interpreted results of experiments; D.R.-C. prepared figures, drafted manuscript; D.R.-C. and S.M.A. edited and revised manuscript. All authors read and approved the final manuscript.

Funding

Funded by Ministerio de Ciencia Innovación y Universidades (Spain), Department of Physiology, Faculty of Medicine & Department of Agricultural Chemistry and Bromatology, Faculty of Sciences, Grant Number: PID2022-138440OB-I00.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files, and they are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The present study has the approval of the Research Ethics Committee of Universidad Autónoma of Madrid (CEI-136-2901, approved on February 9th, 2024) and all participants gave their consent to participate and publish these results.