It is well recognized that chronic endurance training induces a number of morphological and functional cardiovascular adaptations, including enlargement of the left ventricular (LV) cavity, a reduction in resting HR, and an augmented stroke volume (25). These adaptations are associated with changes in the contractile and filling properties of the heart chambers and thus confer protection against adverse cardiac events and CAD (26). Despite the evident cardioprotective effect of moderate regular aerobic exercise, the short-term consequences of performing an acute bout of strenuous exercise over a prolonged duration are less clear.
Prolonged endurance exercise requires a sustained increase in myocardial work in order to deliver sufficient oxygen to the skeletal musculature. There is a growing body of evidence to suggest that the physiological demands of maintaining a high cardiac workload for a prolonged duration may result in a transient impairment in cardiac function (8,10,11,13,21,27,32,35). This phenomenon has recently been termed exercise induced cardiac fatigue (EICF) (5).
EICF is characterized by an acute postexercise reduction in LV systolic and/or diastolic function, reflecting decreases in the contractile and filling properties of the left ventricle, respectively (5). These alterations have been observed following exercise challenges as diverse as a 20-km run lasting approximately 95 min to a 12.5-h Ironman triathlon (7,32). In contrast, some investigators have failed to demonstrate any evidence of EICF following a marathon (23) or a combined 40-km cycle and 10-km run (18). The inconsistent findings present in the literature may be attributed to variations in the methods employed, specifically, the exercise duration and the training status of participants.
Cardiovascular loading conditions directly influence cardiac function and, as such, may play an important role in the genesis of EICF. Fluid shifts during prolonged exercise are likely to induce alterations in cardiovascular loading (4), which may affect the contractile and relaxation properties of the myocardium. Despite the influence of fluid shifts, a number of researchers have reported a decrease in cardiac function under unaltered loading conditions (12,13). Further, additional studies have demonstrated evidence of EICF unrelated to changes in loading (27,32), indicating an intrinsic impairment of cardiac contractility and relaxation. The extent of the observed postexercise impairments in LV function in untrained individuals compared with well-trained individuals is unclear, although these functional alterations appear to be subclinical in both populations. Notwithstanding, the implications of this phenomenon for future health and performance in both untrained and trained cohorts remain to be elucidated.
Despite a growing literature base regarding EICF, the conflicting findings have made it difficult to establish conclusively the existence of this phenomenon and whether it is dependent on exercise duration and/or training status. Moreover, the small sample sizes that are typically used in EICF studies often result in underpowered findings. McGavock et al. (19), and Dawson et al. (5), have provided informative reviews on LV function following prolonged exercise and generally support the concept of EICF. Moreover, it is suggested that exercise duration may be an important factor in the genesis of this phenomenon. Quantification of these data via meta-analysis has yet to be completed, and following the publication of the reviews by McGavock et al. (19) and Dawson et al. (5), a number of additional studies have been conducted that have added considerably to the existing literature base. Assimilating valid data via means of specific inclusion criteria, in a meta-analysis, would yield the large overall sample size necessary to provide high-powered statistical conclusions regarding this exercise-related phenomenon. In addition, categorization of individual studies, with reference to training status and exercise duration, could offer an insight into the importance of these variables in EICF and potentially resolve the present conflict.
The aim of the present article is to report the results of a meta-analysis examining the effect of prolonged endurance exercise on LV systolic and diastolic function, in addition to indices of cardiac loading. Exercise duration and training status are also examined as potential moderators of EICF. It is hypothesized that this meta-analysis will reveal an overall postexercise reduction in LV function, mediated by exercise duration and training status.
Search strategy and criteria.
An exhaustive literature search for all peer-reviewed studies examining LV function following prolonged endurance exercise was conducted using MEDLINE and PubMed databases. The keywords and phrases employed in the online search included "left ventricular function and exercise," exercise induced cardiac fatigue," and "echocardiography and prolonged exercise." Reference lists from published papers were examined in order to identify any other relevant studies not cited in the online databases. For inclusion into the present meta-analysis, studies had to meet the following criteria: 1) available information regarding age, gender, and training status of participants; 2) two-dimensional or two-dimensional guided M-mode echocardiographic assessment of LV function prior to and immediately following aerobic exercise; 3) an exercise duration ≥ 60 min; 4) measurement of at least one of the following: ejection fraction (EF), systolic blood pressure/end-systolic volume (SBP/ESV) ratio, early-to-late diastolic filling (E/A) ratio, or parameters from which these variables could be calculated; and 5) no pharmacological interventions or infusion techniques.
The search process resulted in the identification of 34 studies that could potentially be included in the analyses. Of these studies, seven did not conform to the inclusion criteria and were subsequently removed. A further four studies were excluded because they did not provide either the before and after mean or SD data necessary for the analysis. The 23 remaining studies, three of which were subdivided into two separate studies owing to study design (28,30,35), were entered into the final analyses, resulting in an overall sample of 413 cases. Studies were categorized into one of three groups depending on exercise duration: moderate duration (i.e., < 3 h (60-151 min), 99 cases), long duration (i.e., marathon and half Ironman races (166-430 min), 217 cases), and ultralong duration (i.e., Ironman and ultraendurance events (640-1440 min), 97 cases). This categorization system differs slightly from that employed by McGavock et al. (19); however, due to a small number of studies at the ultra duration end of the spectrum, this method ensured a more even distribution of studies across exercise durations. It is pertinent to note that variable exercise durations in some studies may overlap two categories, but a general approach has been taken for the purposes of this investigation. The moderate duration (MD) group was further subdivided into two groups with regard to training status as described in each individual study: untrained (MDu, 44 cases) and trained (MDt, 55 cases). With the exception of one study (31), the long duration (LD), and ultralong duration (UD) studies involved trained athletes only; therefore, this subdivision was not necessary.
The echocardiographic parameters of EF and SBP/ESV ratio were included for the analysis of LV systolic function. Studies that did not provide these data directly, but reported appropriate echocardiographic data to enable calculation of these parameters (i.e., LV diameters or volumes in systole and diastole and SBP) were included (14) was used to calculate EF: EF (%) = [(LVIDd)3 − (LVIDs)3/(LVIDd)3] × 100, where LVIDd and LVIDs represent the LV internal diameter in diastole and systole, respectively. This formula was selected as this method was employed in the majority of the studies included in the analyses. The E/A ratio was included as an index of LV diastolic filling. To investigate the impact of hemodynamic loading conditions on cardiac function, the LVIDd was used in the analysis as a surrogate of preload, and SBP was employed as an estimate of afterload. Due to the chronotropic sensitivity of some of the functional indices, HR at the time of assessment was also included in the analysis.
A random-effects meta-analysis of the mean changes in the pre-/postexercise echocardiographic parameters was conducted. All the studies included in the analyses comprised a single-group pre/post design with no control group; therefore, the SE calculated from the between-subjects SD was meta-analyzed. This decision was based on the fact that the between-subjects SD was cited in the majority of the studies, whereas the SD of the differences was not cited in a single study. Moreover, exact P values from the appropriate repeated-measures analyses (which could have been used to calculate the SD of differences) were reported in only 16% of studies. The SD of the differences calculated from these cited P values were compared with the between-subjects SD, and little difference between these estimates was found. The fully adjusted estimates from individual studies were calculated using the inverse variance method (1). Relative weights were assigned to each study mean change on the basis of sample size and between-subjects SD. Heterogeneity was assessed using the Q-test (5), and a statistically significant Q-test was explored by subgroup analysis (1) based on duration of exercise and training status.
Correlations between postexercise changes in echocardiographic parameters and changes in the LVIDd, SBP, and HR were assessed via a Pearson's product-moment correlation coefficient analysis. Subgroup forest plots were conducted to present the results of the subgroup analyses. Differences between subgroups were examined using the 95% confidence interval (CI) of each summary mean difference, with alpha set at 0.05.
The weighted mean change and pooled variance data for each echocardiographic variable from the individual studies included in the analyses are displayed in Table 1. Overall weighted mean change in EF was significantly (P < 0.05) decreased postexercise. Heterogeneity between studies was statistically significant with a Q statistic of 50.8 (P = 0.00032); hence, a subgroup analysis was conducted. Taking into account only the trained cases, only the UD group had a significant weighted mean change in EF postexercise (mean change: −4.02%; CI: −2.26 to 5.78%), in comparison with the MDt and LD groups. There was no difference in EF mean change between the MDt and LD groups (mean change, CI: −1.11%, 1.28 to −3.5% vs −0.70%, 0.58 to −1.98%). However, MDu differed significantly (P < 0.05) from MDt, with a significant weighted mean change in EF postexercise (mean change, CI: −5.47, −3.08 to −7.85). These data are illustrated in Figure 1.
The SBP/ESV ratio also demonstrated an overall decrease postexercise (P < 0.05). Although studies were heterogeneous, with a significant Q statistic of 51.7 (P = 0.0041), there was an insufficient sample size for a subgroup analysis to be conducted due to the lack of available data.
With regard to diastolic function, there was a significant (P < 0.05) decrease in overall weighted mean change in E/A ratio. Due to significant heterogeneity between studies (Q statistic = 33.4, P = 0.000005), a subgroup analysis was performed. Considering only the trained cases, no differences between the MDt, LD, and UD subgroups were observed (Fig. 2). Differences between trained and untrained cases could not be determined due to an insufficient sample size for the untrained population.
Impact of loading and HR.
There was an overall decrease in the LVIDd of 1.3 mm (CI: -1.1 to -1.5 mm), indicating a decrease in preload postexercise. Studies were significantly heterogeneous with regard to this parameter (Q statistic = 40.4, P = 0.0011). Subgroup analysis revealed that the UD and MDt groups had a significant (P < 0.05) change in the LVIDd compared with baseline, and this reduction was greater in the UD group (mean change, CI: −3.20 mm, −1.25 to −5.16 mm vs −1.36 mm, −1.13 to −1.60 mm). There was no change in the LVIDd in the LD group (Fig. 3).
SBP demonstrated a significant (P < 0.05) overall reduction following exercise (mean change, CI: −7.0 mm Hg, −5.7 to −8.3 mm Hg), and from the subgroup analysis, this reduction was evident in both the UD and LD groups (mean change, CI: −6.6 mm Hg, −3.6 to −9.7 mm Hg and −8.2 mm Hg, −6.6 to 9.9 mm Hg, respectively). The MDu group also showed a decrease in SBP, which was not observed in the MDt group.
An overall mean increase in resting HR of 26 bpm was observed postexercise, and this increase was significantly (P < 0.05) greater in the MDt group compared with the LD and UD groups (Fig. 4). A training status subgroup analysis revealed a significantly greater increase in HR in the MDt compared with the MDu group (mean change, CI: 42 bpm, 39-46 vs 26 bpm, 22-29 bpm, respectively; Fig. 4).
The correlation coefficients between the changes in echocardiographic variables and changes in HR and LVIDd are displayed in Table 2. The EF was significantly (P < 0.05) correlated with LVIDd, and SBP/ESV ratio was significantly related to HR (P < 0.01) and SBP (P < 0.01). However, no other relationships between parameters were observed.
The results of the present meta-analysis demonstrated an overall reduction in the EF immediately following prolonged endurance exercise, suggesting a potential transient decrease in systolic function. This finding was supported by an overall decrease in the SBP/ESV ratio, although this analysis comprised a considerably smaller sample size. Downregulation of beta-adrenergic receptors as a result of prolonged exposure of circulating catecholamines may offer a potential mechanism underpinning the observed reduction in LV contractility (11,33), although it is likely that an interaction of factors are involved. Specifically, the relationships between the immediate postexercise reductions in the EF and SBP/ESV ratio, and changes in the LVIDd, SBP, and HR, respectively, may account for some of the alterations in the systolic functional parameters. Accordingly, these findings should be interpreted with caution.
With regard to diastolic function, an overall immediate postexercise reduction in the E/A ratio was observed, reflecting altered diastolic filling dynamics. The change in the E/A ratio was not correlated with the changes in the LVIDd, SBP, or HR, suggesting that alterations in diastolic filling cannot be directly attributed to these variables. Taken together, these data suggest that global LV function may be altered following an acute bout of prolonged endurance exercise and that this phenomenon is not fully explained by changes in hemodynamic loading postexercise.
Impact of Exercise Duration
Examination of subgroups with regard to exercise duration revealed a significant postexercise decrease in the EF in the UD group only, and data from this group were significantly different from both the LD and MDt groups when only the trained cases were considered. The time course of this alteration in the EF could not be quantified in the present meta-analysis due to the lack of 24-h postexercise EF data; however, the available data in the UD group indicate that the EF returns to baseline following a 24-h recovery period (35).
The present finding of a decrease in systolic function immediately following exercise durations ≥ 640 min, such as an Ironman Triathlon and other ultraendurance events, supports the theory that the onset of EICF is duration dependent (29). There are some data to suggest that a transient decrease in systolic function may only be apparent following aerobic exercise of ≥ 6 h duration (5,19,35), with alterations in diastolic filling occurring over shorter exercise durations, e.g., a half Ironman triathlon or marathon (10,27,35). Hence, the exercise durations employed with the LD and MD studies is likely insufficient to induce a reduction in LV systolic function. It was not possible to assess the potential impact of exercise intensity on EICF, as this variable has not been reported in the majority of studies. Investigation of the influence of exercise intensity would form an ideal basis for a future EICF study.
The mean weighted change in the EF was significantly related to the decrease in the LVIDd, implying that a postrace reduction in preload has a notable influence on this parameter. It is important to note that the use of the LVIDd as a surrogate of preload has limited value, as the internal diameter of the left ventricle is inherent in the calculation of the EF. It is clear that a more robust indicator of preload needs to be employed in future EICF studies.
The impact of exercise duration on the SBP/ESV ratio could not be determined due to the limited data available. Although it was possible to calculate the mean values from each study, the lack of SD data prevented further analysis. Despite the relative preload-independence of the SBP/ESV ratio, the mean weighted change in this index was closely related, not unexpectedly, to the decrease in SBP. ΔSBP/ESV ratio showed no relation to the change in the LVIDd; therefore, it is unlikely that the reduction in preload had an impact on the observed decrease in this index of systolic function. A significant relationship between ΔSBP/ESV ratio and SBP, however, suggests that a postexercise decrease in afterload may have influenced this variable. This finding should be interpreted with caution as the SBP/ESV ratio was derived from the SBP and is therefore mathematically related and spurious in nature (2). Thus, the correlation coefficient between SBP and SBP/ESV ratio would be expected to be different from zero, irrespective of any physiological relationship between these variables. The significant relationship between the alteration in the SBP/ESV ratio and change in HR is unlikely to be spurious, however, and suggests that the SBP/ESV ratio is influenced by HR despite the preload independence of this measure. Although the SBP/ESV data need to be interpreted with caution when HR discrepancies are apparent, changes in HR or loading do not fully explain the reduction in SBP/ESV. Accordingly, the inotropic state may also be implicated in the reduction in systolic function observed in the present meta-analysis.
The immediate postexercise E/A ratio was significantly different from preexercise values in all subgroups, suggesting that diastolic filling is compromised following all exercise durations > 60 min. No differences between subgroups were observed, indicating that the magnitude of the reduction in diastolic filling immediately following prolonged endurance exercise may not be related to exercise duration or training status. Despite overall decreases in the LVIDd and SBP and an overall increase in HR postexercise, these changes were not related to the change in E/A ratio, indicating that the alteration in diastolic filling may be intrinsically mediated.
Only six of the meta-analyzed studies reported 24-h postexercise LV diastolic data; therefore, they could not be included in the present investigation. From the available data, however, it is evident that the E/A ratio returns to baseline following a 24-h recovery period (7,8,15,28,30,35). The rapid return of any exercise-induced reductions in diastolic filling to baseline levels suggests that the clinical impact of this phenomenon appears to be minimal. If, however, the transient postexercise decrease in diastolic filling represents a precursor of cardiac fatigue and a subsequent reduction in functional capacity, this may have an impact on prolonged endurance exercise performance, alongside other limiting factors, such as metabolic demands and skeletal muscle fatigue. To date, no researchers have examined this issue.
Impact of Training Status
Subdivision of the MD group with regard to training status revealed a significant, immediate postexercise reduction in the EF in the MDu group, which was not observed in the MDt group. Hence, it appears that aerobic exercise of 60- to 151-min duration may be an adequate stimulus to elicit EICF in untrained, but not trained, individuals, implicating training status as an important variable in this phenomenon. The clinical ramifications of EICF in untrained individuals participating in prolonged aerobic exercise, including popular events such as 10km and half-marathon races, have not been clarified. It is evident that appropriate training preparation prior to such strenuous exertion, however, may confer protection against adverse cardiac events (26); thus, the attenuation of the onset of EICF in trained athletes over this duration may indicate a training-induced cardioprotective adaptation in this population. Further work on the impact of training status on EICF is indicated.
Limitations, Summary, and Future Directions
One limitation of the studies included in the present meta-analysis is that the echocardiography immediately following prolonged endurance exercise has been the method of choice in the majority of EICF investigations to date. Postexercise plasma volume shifts, and a drop in blood pressure may have ramifications regarding the validity of this procedure in the assessment of LV function. Notwithstanding, preload is optimized in the supine position (31). Alternative methods that may offer a more valid indication of cardiac function include invasive cardiovascular techniques and echocardiography performed during exercise, yet these methods are complex and less accessible. As such, there are presently no available data that demonstrate impaired cardiac function during submaximal prolonged endurance exercise. Future studies in this area should attempt to employ more robust techniques in the assessment of postexercise cardiac function.
In summary, the present meta-analysis revealed an overall immediate postexercise reduction in the EF, SBP/ESV ratio, and E/A ratio, suggesting a transient decrease in both systolic function and diastolic filling. Exercise duration was found to be an important factor in the onset of the impairment in systolic function, but did not appear to have the same impact on diastolic filling. Training status was also identified as a key determinant in EICF, as shown by the greater reduction in the EF in the untrained compared with the trained group. Postexercise alterations in the EF and SBP/ESV ratio were correlated with changes in the LVIDd and HR, respectively, suggesting that the observed changes may not solely reflect impaired contractility. In contrast, the alterations in loading conditions and HR postexercise appeared to have little impact on the alterations in diastolic filling, possibly indicating an intrinsically mediated decline in LV relaxation properties.
Future EICF research should attempt to examine potential mechanisms for the impairment in LV function following prolonged endurance exercise, for example, an increase in circulating catecholamines and subsequent down regulation of beta-adrenergic receptors. Data from the present meta-analysis suggest that changes in cardiac function are likely mediated by a number of factors; therefore, investigators in this area should focus on assessing the relative contribution of altered loading, HR, and inotropy in the genesis of EICF. Accordingly, it is advised that future researchers attempt to employ reliable and load-independent measures of LV function. Finally, it is recommended that authors report all available echocardiographic parameters with exact P values for each variable, in addition to providing intra-/interobserver reliability data to facilitate future meta-analysis studies in this area.
1. Alderson, P., S. Green, and J. P. Higgins (Eds). Cochrane reviewers handbook 4.2.2. In: The Cochrane Library, Issue 1
. Chichester, UK: John Wiley & Sons Ltd., 2004.
2. Atkinson, G., P. Watson, R. Maughan, S. Shirreffs, and A. Nevill. A spurious correlation. J. Appl. Physiol.
3. Boynton, M., E. Winslow, R. Ling, and D. Mehlman. Are echocardiographic estimates of left ventricular function altered by running a marathon? Med. Sci. Sports Exerc.
4. Coyle, E. F. Cardiovascular drift during prolonged exercise and the effects of dehydration. Int. J. Sports Med.
19(suppl 2):S121-S124, 1998.
5. Dawson, E., K. George, R. Shave, G. Whyte, and D. Ball. Does the human heart fatigue subsequent to prolonged exercise? Sports Med.
6. Deeks, J. J., D. G. Altman, and M. J. Bradburn. Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Systematic Reviews in Health Care: Meta-Analysis in Context
, M. Egger, G. Davey Smith, and D. G. Altman (Eds.). London: BMJ Publishing Group, 2003.
7. Douglas, P. S., M. L. O'Toole, W. D. Hiller, K. Hackney, and N. Reichek. Cardiac fatigue after prolonged exercise. Circulation
8. Douglas, P. S., M. L. O'Toole, and W. D. Hiller. Different effects of prolonged exercise on the right and left ventricles. J. Am. Coll. Cardiol.
9. Eysmann, S., E. Gervino, D. Vatner, S. Katz, L. Decker, and P. Douglas. Prolonged exercise alters beta-adrenergic responsiveness in healthy sedentary humans. J. Appl. Physiol.
10. George, K., G. Whyte, C. Stephenson, et al. Postexercise left ventricular function and cTnT in recreational marathon runners. Med. Sci. Sports Exerc.
11. Haykowsky, M., R. Welsh, D. Humen, D. Warburton, and D. Taylor. Impaired left ventricular systolic function after a half-Ironman race. Can. J. Cardiol.
12. Ketelhut, R., C. J. Losem, and F. H. Messerli. Depressed systolic and diastolic cardiac function after prolonged aerobic exercise in healthy subjects. Int. J. Sports Med.
13. Ketelhut, R., C. J. Losem, and F. H. Messerli. Is a decrease in arterial pressure during long-term aerobic exercise caused by a fall in cardiac pump function? Am. Heart J.
14. Longo, M., A. Guzman, and J. Triebwasser. Normal values and commonly used echocardiographic formulae for adults. Aviat. Space Environ. Med.
15. Lucia, A., L. Serratosa, A. Saborido, et al. Short-term effects of marathon running: no evidence of cardiac dysfunction. Med. Sci. Sports Exerc.
16. Manier, G., F. Wickers, A. M. Lomenech, G. Cazorla, and R. Roudaut. Echocardiographic assessment of myocardial performance after prolonged strenuous exercise. Eur. Heart J.
17. McGavock, J., M. Haykowsky, D. Warburton, D. Taylor, H. Quinney, and R. Welsh. Effects of prolonged strenuous exercise on left ventricular function in female athletes. Can. J. Cardiol.
18. McGavock, J., M. Haykowsky, D. Warburton, D. Taylor, A. Quinney, and R. Welsh. Left ventricular systolic performance during prolonged strenuous exercise in female triathletes. Dyn. Med.
19. McGavock, J. M., D. E. Warburton, D. Taylor, R. C. Welsh, A. Quinney, and M. J. Haykowsky. The effects of prolonged strenuous exercise on left ventricular function: a brief review. Heart Lung
20. Niemela, K., I. Palatsi, M. Ikaheimo, J. Airaksinen, and J. Takkunen. Impaired left ventricular diastolic function in athletes after utterly strenuous prolonged exercise. Int. J. Sports Med.
21. Niemela, K. O., I. J. Palatsi, M. J. Ikaheimo, J. T. Takkunen, and J. J. Vuori. Evidence of impaired left ventricular performance after an uninterrupted competitive 24 hour run. Circulation
22. Alatini, P., S. Bongiovi, F. Macor, et al. Left ventricular performance during prolonged exercise and early recovery in healthy subjects. Eur. J. Appl. Physiol. Occup. Physiol.
23. Perrault, H., F. Peronnet, R. Lebeau, and R. A. Nadeau. Echocardiographic assessment of left ventricular performance before and after marathon running. Am. Heart J.
24. Seals, D. R., M. A. Rogers, J. M. Hagberg, C. Yamamoto, E. Cryer, and A. A. Ehsani. Left ventricular dysfunction after prolonged strenuous exercise in healthy subjects. Am J Cardiol.
25. Sharma, S. Athlete's heart-effect of age, sex, ethnicity and sporting discipline. Exp. Physiol.
26. Sharma, S., G. Whyte, and W. J. McKenna. Sudden death from cardiovascular disease in young athletes: fact or fiction? Br. J. Sports Med.
27. Shave, R., E. Dawson, G. Whyte, K. George, D. Gaze, and P. Collinson. Altered cardiac function and minimal cardiac damage during prolonged exercise. Med. Sci. Sports Exerc.
28. Shave, R., E. Dawson, G. Whyte, et al. The impact of prolonged exercise in a cold environment upon cardiac function. Med. Sci. Sports Exerc.
29. Shave, R. E., E. Dawson, G. Whyte, et al. Evidence of exercise-induced cardiac dysfunction and elevated cTnT in separate cohorts competing in an ultra-endurance mountain marathon race. Int. J. Sports Med.
30. Shave, R. E., E. Dawson, G. Whyte, K. George, D. Gaze, and P. Collinson. Effect of prolonged exercise in a hypoxic environment on cardiac function and cardiac troponin T. Br. J. Sports Med.
31. Sundblad, P., and B. Wranne. Influence of posture of left ventricular long- and short-axis shortening. Am. J. Physiol. Heart.
32. Vanoverschelde, J. L., L. T. Younis, J. A. Melin, et al. Prolonged exercise induces left ventricular dysfunction in healthy subjects. J. Appl. Physiol.
33. Welsh R., D. Warburton, D. Humen, D. Taylor, J. McGavock, and M. Howkowsky. Prolonged strenuous exercise alters the cardiovascular response to dobutamine stimulation. J. Physiol.
34. Whyte, G., K. George, R. Shave, et al. Impact of marathon running on cardiac structure and function in recreational runners. Clin. Sci. (Lond.)
35. Whyte, G. P., K. George, S. Sharma, et al. Cardiac fatigue following prolonged endurance exercise of differing distances. Med. Sci. Sports Exerc.