Acute effects of marathon running on the heart: a systematic review and meta-analysis

Abstract

Background

Marathon running has numerous cardiovascular effects. Studies report transient physiological changes, while others identify possible harmful cardiovascular changes. This systematic review and meta-analysis aimed to characterise acute cardiac effects of marathon running.

Methods

A systematic search was conducted in PubMed, Embase, and SPORTDiscus up to 11 April 2025. Studies were included if they reported (i) primary data (ii) in English (iii) healthy adults completing a road marathon, (iv) quantitative pre- and post-marathon cardiac effects. After data extraction (myocardial biomarkers, morphology, and function), random-effect meta-analyses were performed to determine mean differences and relative changes between pre- and post-marathon. Meta-regression analyses were performed in metrics with ≥7 studies.

Results

Of 9080 articles, 69 were eligible for systematic review (3274 participants: age 30–55 years, >80% men) and 49 for meta-analysis on three biomarkers, 18 echocardiography metrics, and five cardiac MRI metrics. Marathon running-induced increases in circulating troponin T (+42 (32.6, 52.3) ng/L; p<0.001), troponin I (+77 (55.1, 98.3) ng/L; p<0.001) and N-terminal pro B-type natriuretic peptide (NT-proBNP) (+114 (89.0, 138.6) ng/L; p<0.001). On echocardiography, left ventricular (LV) volume and diastolic function, and right ventricular (RV) systolic function decreased post-marathon, while RV end-systolic and diastolic volumes increased. Cardiac MRI revealed an increase in LV ejection fraction. On meta-regression analysis, these responses varied according to age, sex, training status and marathon performance.

Conclusion

Marathon running induces acute changes in biomarkers of myocardial injury, cardiac structure and function. The clinical significance and potential for harm of these changes are yet to be established.

PROSPERO registration number

CRD42023399120

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Endurance exercise induces short-term elevations in cardiac biomarkers and changes in cardiac function.

• Individual studies have reported post-marathon cardiac changes across small cohorts, but findings have been inconsistent.

• Whether such acute changes are physiological responses or early markers of adverse cardiac remodelling is unclear.

WHAT THIS STUDY ADDS

• Marathon running increases circulating levels of cardiac troponin T (cTnT), cardiac troponin I (cTnI) and N-terminal pro B-type natriuretic peptide (NT-proBNP).

• Chamber volumes and cardiac function are moderately altered after completing a marathon run.

• These responses vary according to age, sex, training status, and marathon performance, highlighting considerable inter-individual variability.

• Changes may represent a normal physiological response to endurance exercise.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Future research should prioritise sex-stratified analyses, standardised biomarker sampling and comprehensive reporting of training status. Moreover, longitudinal follow-up studies are needed to ascertain whether these acute changes translate to long-term cardiac (mal)adaptations.

Introduction

Regular endurance running is associated with numerous health benefits, including reduced risk of fatal and non-fatal cardiovascular disease (CVD) and improved physical fitness. Yet, high levels of lifetime running may be associated with increased CVD risk, as suggested by the debated U-shaped relationship between lifetime exercise and all-cause mortality, where both low and extremely high levels of exercise may increase mortality risk.¹

In recent years, the number of individuals participating in endurance running events – such as marathons – has risen, increasing the relevance and clinical need to identify potential upper limits of physical exercise on cardiovascular health.² For example, increased rates of coronary artery calcification, increased incidence of atrial fibrillation and the presence of myocardial scars have been reported to be associated with long-term endurance exercise.³ Several studies have investigated the effect of marathon running on surrogate markers of myocardial injury, demonstrating increased post-marathon levels of circulating biomarkers (eg, troponins). Elevated levels of such biomarkers correlate with myocardial damage, haemodynamic stress or cardiac inflammation in patients,⁴ but their significance in endurance athletes remains unclear. Furthermore, studies linking such immediate changes to detrimental morphological alterations or cardiovascular events are lacking,^{5 6} and studies that have synthesised findings across populations to quantify acute effects are scarce.⁷

Further challenging the interpretation of post-marathon acute cardiac changes is the fact that studies report heterogeneous findings.8,11 This variability is likely due to differences in methodology across studies, sample sizes and the use of non-standardised imaging techniques and investigative modalities. Consensus from large populations on the cardiac effects of marathon running is lacking, making it challenging to provide evidence-based recommendations for clinical care and follow-up research. Given that such large, comprehensively phenotyped study populations are lacking, a meta-analysis pooling data from smaller studies may address this gap. We therefore performed a systematic review and meta-analysis to comprehensively summarise and quantitatively analyse the acute effects of marathon running on the heart.

Methods

Data sources and searches

We performed a literature search guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines.¹² To identify all relevant publications, we conducted systematic searches in PubMed, Embase.com and SPORTDiscus (EBSCO) up to 11 April 2025. The following terms were used: “Long-distance Running”, “Marathon”, “Cardiovascular”, “Adults, 20 years and over”, “Men”, and “Women”. We also searched for relevant publications in the reference lists of the identified studies. The full search strategies are provided in the online supplemental material. The protocol for this systematic review was registered at PROSPERO (registration ID CRD42023399120; submitted 13 February 2023).

Study selection and analysis

We included studies that (i) were published in English language in a peer-reviewed journal, (ii) were presented as an original article that described the results of primary research data, (iii) included healthy adults with no known cardiovascular disease who completed a road marathon (42.195 km), and (iv) quantitatively presented comparisons between pre- and post-marathon cardiac parameters. We defined the pre-marathon period as 2 weeks or less before the marathon. The post-marathon period was defined as directly after the marathon and within 24 hours post-marathon for biomarkers (because circulating levels of most cardiac biomarkers are highest during the first 24 hours)¹³ and echocardiography, and within 3 days for cardiac MRI.

Two reviewers (IL and PR) independently screened relevant titles and abstracts for eligibility. If necessary, the full text was checked to establish eligibility. The two reviewers discussed discrepancies in article selections. In case of disagreement, a third reviewer (EV) assessed the article in question.

Risk of bias assessment

Two authors (IL and PR) independently assessed the risk of bias of all included studies using the National Heart, Lung, and Blood Institute (NHLBI) Quality Assessment Tool for Before-After (Pre-Post) Studies With No Control Group (2007).¹⁴ Each included article was evaluated according to the 12 evaluation criteria. We answered each evaluation criterion with either ‘yes’, ‘no’, or ‘other’, which included ‘cannot determine (CD)’, ‘not applicable (NA)’, and ‘not reported (NR)’. After the assessment, we classified each study as having either a low risk of bias (most assessments answered with ‘yes’, an unclear risk of bias (1 ‘CD’ or ‘no’), or a high risk of bias (more than 1 ‘no’ and/or ‘CD’).¹⁵ Regardless of their risk of bias assessment results, all included studies were incorporated for further analysis.

Data synthesis and meta-analysis

Two reviewers (IL and LDL) independently extracted the following data from each article: marathon event, number of study participants, biological sex, age, finish time, participants’ running experience, the timing of pre- and post-marathon data collection, and specific cardiac assessments. Our primary metrics of interest were (1) myocardial biomarkers, (2) echocardiography findings, and (3) cardiac MRI findings. We included the following biomarkers in the meta-analysis: cardiac troponin T (cTnT), cardiac troponin I (cTnI), and N-terminal pro B-type natriuretic peptide (NT-proBNP), which are widely used indicators of myocardial injury or ischaemia and cardiac wall stress.¹⁶ Echocardiographic metrics were categorised into measures of cardiac morphology, systolic function, and diastolic function. The cardiac metrics were categorised into measures of cardiac morphology, systolic function and diastolic function. Morphological metrics included interventricular septal thickness (IVS), left atrial diameter (LAD), left atrial volume index (LAVi), left ventricular end-diastolic diameter (LVEDd), left ventricular end-diastolic volume (LVEDV), left ventricular end-diastolic volume index (LVEDVi), left ventricular end-systolic diameter (LVESd), left ventricular end-systolic volume (LVESV), left ventricular end-systolic volume index (LVESVi) and right ventricular end-diastolic diameter (RVEDd). Systolic function metrics included left ventricular ejection fraction (LVEF), fractional shortening (FS) and tricuspid annular plane systolic excursion (TAPSE). Diastolic function metrics included early ventricular filling velocity (E), late ventricular filling velocity (A) and the E/A ratio.^{17 18}

If a study reported multiple measurements within the pre- and post-marathon periods, we selected those closest to before the start and after the finish of the marathon. Any disagreement was resolved by consensus, with a third reviewer (EV) rendering a final decision.

The meta-analysis did not include studies that provided insufficient information (eg, lacking the number of participants or uncertainty measures). Studies that duplicated data were considered only once for each outcome measure, with a preference for the earliest publication. When unsure if there was sample replication between studies, we contacted the original authors for confirmation.19,21

We conducted meta-analyses using RStudio²² (1+748) integrated with R (Version 4.4.0). Meta-analyses were performed only on parameters reported in at least two studies. We used reported means and SD of pre- and post-marathon measurements to describe acute cardiac effects of running a marathon. Reported medians were treated as means, and the corresponding SD was approximated by assuming a normal distribution, where the IQR approximates 1.34 times the SD.²³ Mean differences were calculated as the absolute difference between pre- and post-marathon measurements. Reported p-values from the studies were used to determine the mean difference’s SE. This SE was used to estimate the correlation coefficient between pre- and post-marathon measurements for all MRI and transthoracic echocardiography metrics. If the estimated correlation coefficient fell outside the valid range (−1 to 1), it was excluded and replaced with the mean imputed correlation of the remaining studies for that metric. A random-effects meta-analysis was then conducted using the Restricted Maximum Likelihood method to estimate the pooled marathon effects. Effects were expressed as mean differences (95% CI). Summaries of forest plots are presented in percentage mean change of each parameter, by dividing the mean difference by the pre-intervention mean and multiplying by 100. The SE of the relative mean change was determined by dividing the adjusted SE of the mean difference by the pre-intervention mean and multiplying by 100. The mean differences (95% CI), followed by relative mean change in percentage (95% CI), were only reported in echocardiography and cardiac MRI metrics. The weight to each estimate was calculated as the inverse of its squared standard error (SE²), divided by the sum of all estimates’ inverse squared standard errors. The resulting value was multiplied by 100 to express the weight as a percentage. Heterogeneity between studies was quantified using the I² statistic, and between-study variance was calculated using the τ². A p-value of ≤0.05 was considered statistically significant.

Meta-regression analysis

To explore potential sources of between-study heterogeneity, random-effects meta-regression analyses were performed using restricted maximum likelihood.²⁴ Hartung–Knapp adjustment was applied to obtain more reliable CIs, particularly if the number of studies was small.^{25 26} Meta-regression analyses were conducted on metrics reported in ≥7 studies to ensure a more stable regression.²² Therefore, the analyses were limited to biomarker and selected echocardiographic metrics. Prespecified moderators included marathon finish time, mean participant age, percentage of male participants, and training status (amateur/recreational, mixed, or trained). Continuous moderators were scaled to facilitate interpretation of regression coefficients (finish time per 10 min intervals, age per 10 year period, and proportion of male participants per 10%).

Sensitivity analysis

Sensitivity analyses were performed by excluding studies at high risk of bias to assess the robustness of the results. Findings were compared with the primary analyses to evaluate the impact of study quality.

Results

Study selection

Our systematic search identified 9080 references: 2426 in PubMed, 5502 in Embase and 1152 in SPORTDiscus. After removing duplicates, we reviewed 7377 manuscripts for inclusion and identified 69 eligible articles for systematic review (figure 1 and online supplemental 1).

Figure 1. PRISMA flow diagram. n reports the number of articles. Abbreviations: A, late ventricular filling velocity; cTnI, cardiac troponin I; cTnT, cardiac troponin T; E, early ventricular filling velocity; FS, fractional shortening; IVS, inter-ventricular septum; LAD, left atrial diameter; LAVi, left atrial volume index; LVEDd, left ventricular end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEDVi, left ventricular end-diastolic volume index; LVEF, left ventricular ejection fraction; LVESd, left ventricular end-systolic diameter; LVESV, left ventricular systolic volume; LVESVi, left ventricular end-systolic volume index; MRI, magnetic resonance imaging; NT-proBNP, N-terminal pro B-type natriuretic peptide; RVEDd, right ventricular end-diastolic diameter; RVEDVi, right ventricular end-diastolic volume index; RVEF, right ventricular ejection fraction; RVESVi, right ventricular end-systolic volume index; TAPSE, tricuspid annular plane systolic excursion.

Risk of bias assessment

Seven articles were classified as having a high risk of bias, and 17 as having an unclear risk of bias. Of the assessed articles, only six studies reported an a priori sample size calculation. The risk of bias assessment of all included articles is shown in online supplemental table 1.

Description of included articles

The 69 articles included for systematic review were published between 1978²⁷ and 2025,²⁸ comprising 3274 participants. This total number of participants excluded three studies^{19 28 29} that reported different results from the same participants.^{21 30 31} Most studies were conducted after 2000, with seven studies performed before 2000.2732,37 Sample sizes ranged from 4³⁸ to 482³⁹ participants. Not all studies reported biological sex distribution, but overall, participants were predominantly men (73%) aged 27 to 63 years-old. Athletic levels varied from amateur groups with heterogeneous experience levels to well-trained athletes. Pre-marathon measurements were performed from 2 weeks before the marathon to up to 1 hour before the start. Post-marathon measurements were taken immediately after the marathon and up to 3 days later. In the 69 included articles (online supplemental table 2), cardiac biomarkers were reported in 36, echocardiography in 30, and cardiac MRI in six. Twenty articles reported on ≥2 of the abovementioned assessments.58,10 19 27 28 33 40

Meta-analysis

In total, 49 articles were eligible for meta-analysis. Articles by Oxborough et al,²⁰ and Wilson et al,¹⁹ Lewicka-Potocka et al,^{29 30} Lasocka et al,³¹ and Lasocka-Koriat et al,²⁸ were found to report on the same study participants but on different outcome measures. We only included data from the earliest articles in case of duplicate data. Twenty articles could not be included due to incomplete data,536 38 39 52,54 no paired data for meta-analysis,2743 55,61 or non-relevant metrics for meta-analysis.^{32 35 62 63} Ultimately, our meta-analyses incorporated 25 outcome measures: three biomarkers, 18 echocardiography metrics, and five cardiac MRI metrics.

Biomarkers

A total of 36 studies reported pre- and post-marathon biomarkers. We excluded 12 articles due to incomplete data, data only presented in figures or as fractions, or reporting on non-cardiac-specific biomarkers, leaving 24 articles eligible for meta-analysis. Other biomarkers identified during systematic review (online supplemental table 2), were not meta-analysed due to an insufficient number of studies. Of the 15 studies in the meta-analysis reporting on cTnT, four reported high-sensitivity assays,4664,66 while for cTnI, two reported using a high-sensitivity assay.^{28 67} All biomarkers analysed in the meta-analysis (online supplemental figures 1–3) showed consistent post-marathon elevations. The cTnT increased by +42.4 (32.6, 52.3) ng/L (p<0.001), cTnI+74.5 (52.4, 96.7) ng/L (15 studies; p<0.001) and NT-proBNP+113.8 (89.0, 138.6) ng/L (16 studies; p<0.001), after marathon running. These values exceeded commonly used clinical thresholds for myocardial injury or ischaemia, and heart failure.^{16 68}

Echocardiography metrics

Of the 29 articles with systematically reviewed echocardiography findings, three were not eligible for meta-analysis (no paired data^{43 55 59}), leaving 26 articles for meta-analysis. We extracted 18 metrics for meta-analysis. A summary of all echocardiography parameters is presented in figures24, . Forest plots of meta-analyses are in online supplemental figures 4–21.

Figure 2. Summary plot of transthoracic echocardiography (TTE) morphology metrics.

Figure 4. Summary plot of transthoracic echocardiography (TTE) diastolic function metrics.

Figure 3. Summary plot of transthoracic echocardiography (TTE) systolic function metrics.

Morphological changes of the left ventricle (LV) included reductions in end-diastolic volume (EDV) by −6.5 (-10.4, −2.5) mL or −5 (−9, −2) % (p=0.001), LVEDV index (LVEDVi) −5.8 (-8.5, −3.1) mL/m² or −9 (−13, −5) % (p<0.001), LVESV index (LVESVi) −2.6 (-3.8, −1.5) mL/m² or −10 (−15, −6) % (p<0.001), and LV diastolic diameters (LVEDd) −1.7 (-2.7, −0.7) mm or −3 (−5, −1) % (p<0.001). The right ventricular (RV) EDV index (RVEDVi) and RV end-systolic volume index (RVESVi) increased post-marathon: +8.3 (2.8, 13.7) mL/m² or +16^{6 27} % (p=0.003) and+6.9 (4.3, 9.5) mL/m² or+28 (17.4, 38.7) % (p<0.001), respectively. The LV end-systolic volume (LVESV), LV end-systolic diameter (LVESD), left atrial diameter (LAD), RV end-diastolic diameter (RVEDd) and inter-ventricular septum (IVS) thickness remained unaltered. The magnitude of the morphological changes was modest and fell outside the range of what would commonly be interpreted as clinically meaningful structural cardiac alterations.⁶⁹

Systolic parameters included an increase in LV fractional shortening (FS) of +3.1 (1.0, 5.2) % or +8^{3 13} % (p=0.003). In comparison, RV ejection fraction (RVEF) and tricuspid annular plane systolic excursion (TAPSE) decreased by −3.5 (-6.2, −0.8) % or −6 (−10, −1) % (p=0.010), and by −1 mm (−1.6, −0.4 mm) or −4 (−6, −1) % (p=0.002), respectively. The LV ejection fraction (LVEF) remained unchanged. The changes in systolic parameters were modest in magnitude and fell outside of ranges commonly accepted to indicate clinically relevant ventricular function impairment in healthy individuals.⁷⁰

Diastolic function in terms of early ventricular filling velocity (E) decreased by −12.7 (-16.7, −8.8) cm/s or −16 (−21, −11) % (p<0.001), and late ventricular filling velocity (A) increased by+13.1 (10.3, 15.8) cm/s or 26^{21 31} % (p<0.001), translating to a decrease of mean E/A ratio by −0.48 (-0.57, −0.39) or −33 (−39, −27) % (p<0.001). The changes in diastolic indices were consistent across studies, and within physiological ranges.⁴⁰

Cardiac MRI metrics

Of the six systematically reviewed articles with cardiac MRI findings, four were eligible for meta-analysis (two studies^{5 42} were not included due to only 1 metric available for analysis). Five cardiac metrics of interest were analysed: LVEDVi, LVESVi, LVEF, LA volume index (LAVi) and RVEF (online supplemental figures 22–26). LVEF increased by 1.9 (0.9, 3.0) % or 3^{2 5} % (p<0.001), while we found no marathon-induced changes in the other metrics (figure 5). The small increase in LVEF and absence of changes in other MRI-derived metrics suggests that marathon running does not induce any signs of clinically relevant structural myocardial injury as seen on MRI.^{49 71}

Figure 5. Summary plot of cardiac magnetic resonance metrics.

Meta-regression analysis

Meta-regression analyses demonstrated that higher post-race NT-proBNP concentrations were associated with longer marathon finish times (+20.1 (14.9, 25.2) ng/L per 10 min; p<0.001), older age (+37.9 (9.0, 66.7) ng/L per 10 years; p=0.013), and a higher proportion of female participants (+7.4 (1.8, 13.0) ng/L per 10%; p=0.013). Troponin I concentrations were associated with training status, with higher values observed in less trained cohorts (+91.4 (10.3, 172.6) ng/L per training status; p=0.03). Echocardiographic RVEDd was positively associated with training status, with higher values observed in more trained participants (+17.8 (10.6, 24.9) mm for trained individuals; p=0.001), while lower A velocity was associated with more trained participants (−15.3 (-28.9, −1.6) cm/s for trained individuals; p=0.031). A higher E/A ratio was associated with older age (+0.2 (0.1, 0.3) per 10 years; p<0.001). Detailed results are presented in online supplemental table 3.

Discussion

This systematic review and meta-analysis comprehensively synthesised measurements of acute cardiovascular responses to marathon running. Our pooled findings confirm and quantify previously reported post-marathon elevations in cardiac biomarkers and alterations in ventricular function. However, the clinical relevance of these changes remains unclear. A considerable amount of the studies included in our review scored high on the risk of bias assessment, indicating room for methodological improvement. Well-designed studies in clearly defined populations are therefore needed to investigate whether these effects represent physiological responses to extreme levels of exercise or reflect early markers of pathological cardiac remodelling.

Our meta-analysis revealed a marked increase in biomarkers measured within the first hour following marathon completion. Post-marathon cTnI reached mean values of ≥42 ng/L, surpassing thresholds indicative of myocardial infarction in rule-in and rule-out algorithms commonly used in clinical practice.⁷² Biomarker elevations demonstrated considerable variability across studies, eg, for cTnI, a range from+20 ± 10 ng/L³⁷ up to+200 ± 300 ng/L.⁴⁹ Our study highlights the variability of troponin release post-marathon. As such, there is potential for misdiagnoses when interpreting these biomarkers in clinical contexts, which could lead to unnecessary investigations or treatments in individuals who have recently exercised intensively.

Multiple hypotheses have been proposed to explain the mechanisms underlying post-exercise troponin elevations.⁷ These include exercise-induced increased cardiomyocyte membrane permeability, and secretion of cytoplasmic-free cardiac troponin.⁷ While exercise-induced cTn elevation is frequently considered a benign phenomenon in otherwise healthy individuals, it may also unmask subclinical cardiac disease such as coronary atherosclerosis, especially in older individuals.⁷³ However, the prevalence and magnitude of coronary atherosclerosis has not been shown to correlate with cTn levels in middle-aged recreational athletes.⁷⁴ Potential mechanisms of reversible myocardial injury include exercise-induced myocardial stress, which could lead to reversible injury through cell wounds, cytoplasmic blebbing, or the release of extracellular vesicles.⁷⁵ Apoptosis or ischaemia-induced necrosis may be due to cardiomyocyte pressure overload, mechanical overstretching and metabolic stress, causing irreversible myocardial injury.⁷⁵ Further complicating the interpretation of troponin release is the notion that multiple mechanisms may contribute to elevated circulating troponin levels, where the dominant mechanism may differ from one individual to another.¹³ Finally, individuals’ training status, exercise duration and intensity have also been associated with the extent of increase in levels of circulating troponins.¹³

Eight studies reported a mean post-marathon NT-proBNP level exceeding the 99^th percentile (≥125 pg/mL), a threshold indicative of congestive heart failure in clinical cardiology.⁶⁸ NT-proBNP is synthesised and secreted by cardiomyocytes in response to myocardial stretch caused by pressure overload or volume expansion of LV cardiomyocytes.⁷⁶ During a marathon, cardiac stress is elevated for a prolonged time, potentially leading to excess cardiomyocyte stretch, endocrine activation, and myocardial hypoxia, which could predispose NT-proBNP secretion.⁷⁷ While our meta-analysis provides a reference range of commonly found NT-proBNP levels post-marathon, a potential (dose-dependent) long-term association with cardiac adverse outcomes remains unclear.

The metrics of cardiac morphology in our meta-analysis were mainly investigated using echocardiography and, in a minority of studies, cardiac MRI. Acute echocardiographic morphological changes comprised mainly reductions in LV dimensions and increases in RV volume and diameter. Post-marathon, both RVEDVi and RVESVi exhibited increases of +8 mL/m² and+7 mL/m², respectively, combined with relative reductions of 6% in RVEF and 4% in TAPSE, indicative of slightly impaired RV contractility. Potential mechanisms for this functional reduction include compensation to volume overload or the so-called ‘cardiac fatigue’ phenomenon.⁷⁸ Decreases in RV function have previously been reported to correlate with increases in cTnI and B-type natriuretic peptide.⁷⁰ This RV enlargement may be part of the mechanism where repeated extreme RV stress, characteristic of marathon running, may contribute to long-term cardiac alterations, including, in rare cases, exercise-induced arrhythmogenic RV cardiomyopathy (ARVC) development.⁷⁹ These changes typically resolve after detraining.⁸⁰ Our findings underscore the need for longitudinal studies to determine whether acute adaptations observed immediately post-exercise are benign or serve as precursors to chronic structural damage and arrhythmias in susceptible populations.⁸¹

The LVEDV reductions highlight the importance of assessing its relationship with systolic function parameters, particularly FS, that increased by+3.1% in our meta-analysis. An increase in FS paired with changes in LV volume may indicate either enhanced LV contractility or modifications in loading conditions during or after marathon running. Even though LV volumes and diameters somewhat decreased, LVEF remained unchanged. The LV volume reduction could be partly attributed to dehydration.⁸²

Furthermore, the E/A ratio post-marathon consistently decreased across all studies, driven by a reduction in E (early filling) and an increase in A (atrial contraction) velocities. This could indicate a decline in LV diastolic function post-marathon. The complex interplay between hydration state, subtle changes in FS and decreased diastolic function after endurance exercise are incompletely understood and merit further elucidation.⁸³

Notably, the cardiac MRI studies in our meta-analysis did not mirror the LV and RV changes observed in the echocardiography studies. One reason for this may be the timing of the imaging modalities. Overall, MRI examinations were performed~7 hours after marathon completion, while the echocardiograms were performed immediately after completion. This 6 hour difference may have influenced factors driving some of the observed changes, such as hydration state and transient exercise-induced increases in pulmonary pressures, explaining some of the disparities in cardiac volume measurements between the two modalities. Our study revealed substantial heterogeneity across the included studies. This heterogeneity likely stems from several factors: study design, participant characteristics, and specific cardiac metrics assessed. Meta-regression analyses revealed sources of between-study heterogeneity, demonstrating that variability in metrics was influenced by participant characteristics, including age, sex, training status and marathon performance. In particular, higher NT-proBNP concentrations were associated with older age, longer finish times and a higher proportion of female participants, suggesting that elevated levels of circulating biomarkers reflect both exercise-induced haemodynamic stress and known biological variation, including higher natriuretic peptide levels in women.⁸⁴ Similarly, the association between cTnI and training status indicates that less trained individuals may experience greater myocardial stress during prolonged endurance exercise, supporting the interpretation that biomarker elevations are modulated by physiological adaptation rather than representing clear myocardial injury.^{85 86} Associations between RV dimensions and training status further suggest that structural and functional responses of the RV reflect a combination of acute loading conditions and chronic exercise adaptation.⁷⁰ Finally, associations of diastolic indices with age and training status indicate that post-exercise alterations in ventricular filling are influenced by baseline physiological characteristics. These changes were modest, and therefore unlikely to represent clinically meaningful signs of cardiac dysfunction in healthy athletes.⁴⁰ Overall, our findings show that the magnitude of acute cardiac responses to marathon running is not uniform, but is largely determined by individual physiological characteristics and exercise load.

Notably, most studies included male-dominant samples, limiting the generalisability of our findings – in spite of our meta-regression analysis – to female marathon runners, who may exhibit different cardiovascular responses, both acute and chronic.⁸⁷ Additionally, training status, a potential confounder, was not consistently reported across studies. While we elected to include all eligible studies regardless of their risk of bias status, this decision could have contributed to the observed heterogeneity. To investigate this, we performed a sensitivity analysis excluding studies classified with a high risk of bias. As this did not change the pooled estimates, this suggests that study quality did not substantially influence the overall findings. Future research should address these limitations by prioritising sex-stratified analyses, standardised biomarker sampling, and comprehensive reporting of training status. Moreover, longitudinal follow-up studies are needed to ascertain whether these acute changes translate to long-term cardiac (mal)adaptations.

The availability of cardiac MRI studies focusing on marathon runners remains limited compared with echocardiography studies. Four of the six MRI studies in our systematic review and meta-analysis primarily investigated cardiac chamber volumes and function. Only one study used advanced quantitative MRI techniques, assessing myocardial extracellular volume fraction and mean water diffusivity.⁵ These techniques, unique to cardiac MRI, could provide important insights into myocardial tissue properties post-marathon.

Strengths and limitations

Our study has several strengths. First, we consolidated data from multiple studies, quantifying acute changes in cardiac biomarkers, morphology and function. Second, our study combines results from marathon events at various geographic locations and populations, offering a diverse and comprehensive sample across different settings. Finally, we restricted our study to a narrow pre- and post-marathon time window, allowing us to evaluate any short-term cardiac effects of a marathon run.

Some aspects of our study warrant consideration. First, ethnicity was rarely adequately reported, limiting generalisability. Second, other factors that were included in the meta-regression analysis, such as age, running time and training duration, were not reported consistently for all studies.

Conclusion

Marathon running induces acute changes in circulating levels of myocardial biomarkers and moderately alters chamber volumes and cardiac function. Elevations of cTnT, cTnI and NT-proBNP were consistently observed post-marathon. LA diameter, LV diastolic function and RV systolic function decreased, while RV end-systolic and end-diastolic volumes increased after marathon completion (figure 6). These acute alterations likely represent predominantly physiological responses to the intense cardiovascular demands of marathon running. Their magnitude varies according to individual characteristics – including age, sex, training status and performance – highlighting substantial inter-individual variability. However, the potential for pathological consequences in susceptible individuals or with repeated participation in extreme endurance events remains. Longitudinal studies, including state-of-the-art imaging investigations for comprehensive phenotyping directly after the completion of a marathon and after recovery from marathon running are warranted to improve our understanding of the cardiac effects of extreme endurance exercise.

Figure 6. Summary of the relative mean change of cardiac morphological and functional parameters after marathon running: left ventricular (LV) volume and diastolic function and right ventricular (RV) systolic function decreased, while RV end-systolic and diastolic volumes increased. Abbreviations: E/A, early/late ventricular filling velocity; FS, fractional shortening; IVS, inter-ventricular septum; LAD, left atrial diameter; LAVi, left atrial volume index; LVEDd, left ventricle end-diastolic diameter; LVEDV, left ventricular end-diastolic volume; LVEDVi, left ventricular end-diastolic volume index; LVESd, left ventricular end-systolic diameter; LVESV, left ventricular systolic volume; LVESVi, left ventricular end-systolic volume index, MRI, magnetic resonance imaging; RVEDd, right ventricular end-diastolic diameter; RVEDVi, right ventricular end-diastolic volume index; RVEF, right ventricular ejection fraction; RVESVi, right ventricular end-systolic volume index; TAPSE, tricuspid annular plane systolic excursion.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.