Running: The Risk of Coronary Events: Prevalence and Prognostic Relevance of Coronary Atherosclerosis in Marathon Runners

Stefan Möhlenkamp; Nils Lehmann; Frank Breuckmann; Martina Bröcker-Preuss; Kai Nassenstein; Martin Halle; Thomas Budde; Klaus Mann; Jörg Barkhausen; Gerd Heusch; Karl-Heinz Jöckel; Raimund Erbel


Eur Heart J. 2008;29(15):1903-1910. 

In This Article


Details on the Marathon study design have been reported previously.[15] Participants were recruited in three ways: 1) an advertisement in a German marathon journal ('Runners World'), 2) a press conference during inauguration of the study, and 3) inclusion of colleagues and friends of participants, if inclusion criteria were met. Matched control groups were selected from the Heinz Nixdorf Recall Study (HNRS).[16] Both studies were approved by the local ethics committee and by the National Institute of Radiation Protection (Bundesamt für Strahlenschutz, Munich, Germany). All participants gave written informed consent prior to participation in both studies including informed consent for clinical follow-up and evaluation of hospital records.

Inclusion and Exclusion Criteria

Males ≥50 years were eligible, if they had completed at least five full-distance marathons (42.195 km) during the preceding three years. Exclusion criteria comprised history of established heart disease, diabetes mellitus, angina pectoris, and renal failure, musculo-skeletal disease at inclusion preventing future regular marathon running, psychiatric disease, and unwillingness to give informed consent.[15] Two males were excluded from the study because of prior unreported myocardial infarction in one and severe renal failure because of untreated prostate disease in another.

Cardiovascular Risk Factors

Details on cardiovascular risk factor quantification and laboratory measurements have been described elsewhere.[15,16] Blood pressure was measured with an automated oscillometric blood pressure device (Omron 705-CP, Omron, Germany). Current smoking was defined as a history of cigarette smoking during the past year. Participants were defined diabetic if they either reported a physician's diagnosis of diabetes, if they were taking anti-diabetic medication, or if their fasting glucose level was >126 mg/dL. All questionnaires, including those to quantify weekly exercise,[17] interviews, and test protocols were identically used in both studies. The Framingham risk score (FRS) was computed as previously described.[16]

Electron-beam Computed Tomography

Non-enhanced electron-beam computed tomography (EBCT) scans were performed on C-150 scanners (GE Imatron, South San Francisco, USA). EBCT scans for the studies were obtained at three sites with identical scanning protocols as previously described.[18] The Agatston CAC score was quantified and percentile CAC values were calculated based on data from the HNRS.[18] The CAC score was not given to participants or to general practitioners in both studies.

Cardiac Magnetic Resonance Imaging

cMRI scans were performed in the marathon study but not in the HNRS. All examinations were performed on a 1.5 T MR scanner equipped with high-performance gradients (Magnetom Avanto, Siemens, Erlangen, Germany). An inversion recovery fast low angle shot sequence (IR-turboFLASH: TR 8.0 ms, TE 4.0 ms, TI 180-240 ms, FA 20°) was acquired in short- and long-axis views 10-15 min after injection of 0.2 mmol/kg body weight of gadolinium-diethylene triamine pentaacetic acid (DTPA) (Magnevist, Schering AG, Berlin, Germany) to identify LGE. Pattern and extent of LGE were assessed using short- and long-axis views[19] and were defined as present only if detectable in two orthogonal planes. A repeat cMRI study was performed when LGE was detected.

Follow-up and Definition of Coronary Events

Follow-up information was obtained from annual questionnaires and personal communication. Events were confirmed from hospital records. Coronary events were defined as sudden coronary death, myocardial infarction, and coronary revascularization.

Statistical Analysis

For comparison with the general population, two groups matching the marathon runners (group I) were drawn from the HNRS cohort, restricted to males without CAD, aged ≥50 (n = 1842). These were group II: 1:8 matching in four-year age classes, group III: 1:2 matching within ±3 years of age, within ±3 kg/m2 body mass index (BMI), within ±4% Framingham risk per 10 years, and by smoking status (present/former/never smokers). Furthermore, the matched cohort group III was restricted to HNRS participants without a history of stroke or diabetes (n = 1597). Matching was performed using PROC SURVEYSELECT of SAS (SAS Institute Inc., Cary, NC, USA) to generate group II and the algorithm described in Schröder et al.[20] to generate group III.

Data were presented as mean ± SD, median -- 25th and 75th percentiles (Q1 and Q3), or proportions, where appropriate. Correlations involving CAC score were calculated according to Spearman, associations with physical activity parameters were also analysed with linear regression analyses for log-transformed (CAC + 1). To compare matching factors between groups, the Mann-Whitney U test, χ2 test, or Fisher's exact test were employed. To evaluate group effects controlling for matching factors, general linear or logistic models (PROC GLM or PROC LOGISTIC of SAS) were used. Because of the strongly skew CAC distribution, models for CAC were based on ranks. Predictive models for CAC were also calculated using general linear models, with the Agatston score transformed as log2(CAC + 1). All linear regression models were inspected by analysis of residuals and checked for nonlinear dependencies. Except models for log2(CAC + 1) there were no abnormalities.

Presence of LGE was modelled by logistic regression in one and two variables. Because of the low number of LGE, P-values from a two-variable model, comprising number of marathons (logarithmized), and CAC percentile values, were recalculated with exact logistic regression using LogXact (Cytel Software Corporation, Cambridge, MA, USA).

Event-free survival rates were estimated following the Kaplan-Meier method and overall group differences were evaluated by log-rank statistics. In addition, a Cox regression model with log-transformed CAC -- log2(CAC + 1) -- as independent variable was calculated.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: