Author + information
- Received March 21, 2016
- Revision received August 5, 2016
- Accepted August 12, 2016
- Published online September 4, 2017.
- Gherardo Finocchiaro, MD,
- Harshil Dhutia, MBBS,
- Andrew D’Silva, MBBS,
- Aneil Malhotra, MBBChir,
- Alexandros Steriotis, MD, PhD,
- Lynne Millar, MBBS,
- Keerthi Prakash, MBBS,
- Rajay Narain, MBBS,
- Michael Papadakis, MD, MBBS,
- Rajan Sharma, MD, MBBS and
- Sanjay Sharma, MD, MBChB∗ ()
- Cardiology Clinical Academic Group, St. George's, University of London, St. George's University Hospitals NHS Foundation Trust, London, United Kingdom
- ↵∗Address for correspondence:
Dr. Sanjay Sharma, St. George’s University of London, Cardiovascular Sciences, Cranmer Terrace, London SW17 0RE, United Kingdom.
Objectives This study sought to investigate the effect of different types of exercise on left ventricular (LV) geometry in a large group of female and male athletes.
Background Studies assessing cardiac adaptation in female and male athletes indicate that female athletes reveal smaller increases in LV wall thickness and cavity size compared with male athletes. However, data on sex-specific changes in LV geometry in athletes are scarce.
Methods A total of 1,083 healthy, elite, white athletes (41% female; mean age 21.8 ± 5.7 years) assessed with electrocardiogram and echocardiogram were considered. LV geometry was classified into 4 groups according to relative wall thickness (RWT) and left ventricular mass (LVM) as per European and American Society of Echocardiography guidelines: normal (normal LVM/normal RWT), concentric hypertrophy (increased LVM/increased RWT), eccentric hypertrophy (increased LVM/normal RWT), and concentric remodeling (normal LVM/increased RWT).
Results Athletes were engaged in 40 different sporting disciplines with similar participation rates with respect to the type of exercise between females and males. Females exhibited lower LVM (83 ± 17 g/m2 vs. 101 ± 21 g/m2; p < 0.001) and RWT (0.35 ± 0.05 vs. 0.36 ± 0.05; p < 0.001) compared with male athletes. Females also demonstrated lower absolute LV dimensions (49 ± 4 mm vs. 54 ± 5 mm; p < 0.001) but following correction for body surface area, the indexed LV dimensions were greater in females (28.6 ± 2.7 mm/m2 vs. 27.2 ± 2.7 mm/m2; p < 0.001). Most athletes showed normal LV geometry. A greater proportion of females competing in dynamic sport exhibited eccentric hypertrophy compared with males (22% vs. 14%; p < 0.001). In this subgroup only 4% of females compared with 15% of males demonstrated concentric hypertrophy/remodeling (p < 0.001).
Conclusions Highly trained athletes generally show normal LV geometry; however, female athletes participating in dynamic sport often exhibit eccentric hypertrophy. Although concentric remodeling or hypertrophy in male athletes engaged in dynamic sport is relatively common, it is rare in female athletes and may be a marker of disease in a symptomatic athlete.
Long-term athletic training is associated with a series of alterations in cardiac structure, function, and electrical activity that are collectively termed athlete’s heart (1–3). The ability to accurately diagnose cardiovascular diseases and, specifically, to differentiate physiological cardiac adaptation caused by exercise from cardiac pathology constitutes one of the most fundamental aspects of sports cardiology (4–6).
Although numerous studies have evaluated the cardiac response to regular physical training in male athletes, there are limited data on female athletes, who constitute an increasing number of elite athletes worldwide. A former large study of female Italian Olympian athletes revealed that none exhibited an absolute left ventricular (LV) wall thickness exceeding predicted upper limits for the general population and the LV cavity was considered enlarged in only 8% (7). These results indicate that the quantitative alterations in absolute cardiac dimensions in females rarely overlap with the primary cardiomyopathies, which are recognized causes of exercise-related sudden cardiac death in young adults. Absolute values for cardiac dimensions are dependent of sex, size, and type of sport. In this regard, the assessment of LV geometry using left ventricular mass (LVM) index and relative wall thickness (RWT) is an increasingly important component in differentiating athlete’s heart from pathological left ventricular hypertrophy (LVH), such as hypertrophic cardiomyopathy (8–10). However there are few reports on sex-specific LV geometry alterations in athletes. This study compared LV geometry in a large cohort of highly trained male and female athletes.
The United Kingdom does not support a state-sponsored cardiac screening program in athletes. However, the charitable organization Cardiac Risk in the Young (www.c-r-y.org.uk) has an established cardiac screening program for young individuals and also serves many professional sporting organizations in the United Kingdom. Up to 1,000 athletes from numerous regional or national sporting squads are assessed annually. Most preliminary evaluations, including electrocardiogram (ECG) and echocardiography, are performed at training centers by experienced cardiologists through a mobile investigations unit and supervised by the principal investigator (S.S.). Between 2010 and 2013, a total of 1,364 elite athletes age 14 to 35 years were assessed with a health questionnaire, ECG, and echocardiogram. Of these, black females comprised only 40 (3%) athletes. Given that the aim of this study was to specifically assess sex differences in LV geometry in a large cohort we confined our analysis only to white athletes. The final study group consisted of 1,083 consecutive elite athletes of which 40% were female. The mean age of the cohort was 22 ± 6 years of age.
Standard 12-lead ECGs were performed as described elsewhere (11). Sokolow-Lyon voltage criteria for LVH were defined as the sum of S in V1 + R in V5 or V6 ≥35 mm. ST-segment shift was considered significant if ≥0.1 mV in ≥2 contiguous leads. Biphasic T-wave inversion (TWI) was considered abnormal if the negative deflection of the T-wave exceeded ≥−0.1 mV. TWI ≥0.1 mV in ≥2 contiguous leads was considered abnormal. Deep TWI was defined as a T-wave deflection ≥−0.2 mV. An abnormal Q-wave was defined as a Q-wave with duration ≥40 ms or a Q/R ratio >0.25. The normal frontal cardiac axis was considered to be >−30° but <120°. Left atrial enlargement was defined by P-wave duration ≥0.12 s in the frontal plane associated with a terminal P negativity in lead V1 of duration ≥0.04 s and depth ≥1 mm. The ECG was interpreted according to the “refined” criteria (12). TWI between V1 and V3 was considered as a normal juvenile ECG pattern in asymptomatic athletes <16 years of age.
Two-dimensional echocardiography was performed using either a GE Vivid I (GE Healthcare, Tirat, Israel), Philips Sonos 7500, Philips iE33, or Philips CPX50 (Philips Healthcare, Bothell, Washington). Standard views were obtained as previously described (13). Assessment of diastolic function included traditional pulsed-wave Doppler across the mitral valve and tissue Doppler velocity imaging of the septal and lateral mitral valve annulus (14). Digitized images of 2 beats were stored and analyzed by cardiologists and expert sonographers blinded to the clinical characteristics offline in accordance with the European Society of Echocardiography guidelines.
Left ventricular (LV internal diameter, septal wall thickness, and posterior wall thickness were measured from 2-dimensional images in the parasternal long-axis view in end-diastole. When measuring septal thickness, care was taken to exclude right ventricular septal bands. In measuring the LV posterior wall thickness, care was taken to exclude posterior wall chordae. Indexed LV cavity size was considered increased if >31 mm/m2. RWT was defined as the ratio of the sum of the interventricular septum and posterior wall thickness in end-diastole to the left ventricular end-diastolic diameter (LVEDD). A RWT was considered to be abnormal if >0.42 (13). LVM calculation was based on a prolate ellipse model of the left ventricle in accordance with American Society of Echocardiography formula: LVM = [0.8 × 1.04[(LVEDD + interventricular septum + posterior wall thickness)3 − (LVEDD)3]] + 0.6 g. An abnormal LVM was defined as >95 g/m2 in women and >115 g/m2 in men (13). LV volumes and LV ejection fraction were assessed from the apical views, using the biplane method of discs. LV systolic function was considered to be reduced if the ejection fraction was <50% (15).
Based on the guidelines from the European and American Society of Echocardiography (13,15), LV geometry was classified into 4 groups based on the RWT and LVM as represented in Figure 1: 1) normal (normal LVM/normal RWT); 2) concentric hypertrophy (increased LVM/increased RWT); 3) eccentric hypertrophy (increased LVM/normal RWT); and 4) concentric remodeling (normal LVM/increased RWT).
Ethical approval was granted by the National Research Ethics Service, Essex 2 Research Ethics Committee in the United Kingdom. Written consent was obtained from individuals ≥16 years of age and from a parent/guardian for those <16 years of age.
Athletes with abnormalities in the history, 12-lead ECG, and echocardiography underwent further assessment including exercise ECG, ambulatory monitoring, signal-averaged ECG, and cardiac magnetic resonance. Specific triggers for additional evaluation included: 1) symptoms suggestive of cardiac disease; 2) family history of hereditary cardiac disease or of premature (≤50 years of age) sudden cardiac death; 3) ECG abnormalities according to “refined” criteria; and 4) structural and functional abnormalities on the echocardiogram, including a LV wall thickness >11 mm in females and >12 mm in males, dilated LV with ejection fraction <50%, and abnormal diastolic function (e.g., average septal/lateral E′ <10 cm/s or E/E′ ratio >15).
Statistical analysis was performed using the PASW software (PASW 18.0 Inc., Chicago, Illinois). Results are expressed as mean ± SD for continuous variables or as number of cases and percentage for categorical variables. Comparison between groups was performed using Student t tests for continuous variables with adjustment for unequal variance if needed and chi-square tests or Fisher exact tests for categorical variables. Intraobservation and interobservation variability was assessed by selecting 80 random studies that were blindly reanalyzed by a separate investigator. Intrareader and inter-reader variability was quantified using mean differences and Pearson correlation and intraclass correlation coefficients.
The mean age of the athletes was 22 ± 6 years and 996 (92%) athletes were >16 years of age. The average hours of exercise were 21 ± 8 per week and were similar between males and females. Males had a greater body surface area (BSA) than females (2.0 ± 0.2 m2 vs. 1.7 ± 0.2 m2; p < 0.001). Athletes engaged in 40 different sporting disciplines, which were subdivided into static, dynamic, or mixed as per the Mitchell classification (16) (62% mixed, 28% pure dynamic, 10% pure static), with no significant differences according to sex. The top 5 sports represented were swimming (n = 123; 11%), cricket (n = 95; 9%), football (n = 89; 8%), rowing (n = 89; 8%), and rugby (n = 72; 6%).
The echocardiographic characteristics of the study population are shown in Table 1. Most athletes showed normal LV geometry (69% males vs. 71% in females; p = 0.54). Females exhibited a lower absolute LVM and RWT compared with males. A RWT >0.42 was observed in 8% of females and 12% of males (p = 0.04). None of the female athletes showed a RWT >0.48 compared with 1.3% of males (p = 0.04) (Figure 2).
Average LVM indexed for BSA was higher in males and almost a quarter of males and females had an increased indexed LVM. None of the females showed a maximal wall thickness >12 mm compared with 2.5% of males (p < 0.001). Females demonstrated a lower absolute LVEDD compared with males. An LVEDD >54 mm was present in only 7% of females versus 47% of males (p < 0.001), but females showed a higher LVEDD indexed for BSA. A total of 18% of females showed an increased indexed LVEDD (>31 mm/m2) compared with 10% of males (p < 0.001).
There was a modest relationship between RWT and LVM in females and males (r = 0.30, p < 0.001 and r = 0.24, p < 0.001, respectively). An important minority of athletes showed concentric hypertrophy/remodeling, which was less common in females (7% vs. 12%; p = 0.009).
There were no significant sex differences relating to the LV geometry in athletes competing in static or mixed sport. In contrast, females competing in dynamic sports exhibited a higher prevalence of eccentric hypertrophy compared with males (22% vs. 16%; p < 0.001). Only 4% of females competing in dynamic sports showed concentric hypertrophy/remodeling compared with 15% of males (p < 0.001) (Figure 3). None of the athletes with concentric LVH/remodeling showed abnormal indices of diastolic function.
Ninety-seven (9%) athletes had an abnormal ECG according to the refined criteria. ECG abnormalities were predominantly driven by abnormal TWI (anterior, n = 68, 6%; inferior, n = 40, 4%). Although there were quantitative differences in the percentage of otherwise normal ECGs between females and males, there were no significant differences in the prevalence of ECGs characterized as distinctly abnormal between females and males (n = 48, 11% vs. n = 49, 8%; p = 0.11). The QRS duration was similar in the 2 sexes (82 ± 12 ms vs. 87 ± 13 ms in females and males respectively; p = 0.156), but females very rarely exhibited a QRS >100 ms (3% vs. 33%; p < 0.001). Sokolow-Lyon voltage criteria for LVH were fulfilled more frequently in male athletes compared with female athletes (14% in females vs. 42% in males; p < 0.001). Left axis deviation was more common in males (0.4% in females vs. 4.0% in males; p < 0.001). The presence of left atrial enlargement was slightly more frequent in males but did not reach statistical significance (5.0% in males vs. 2.5% in females; p = 0.06). Females had a higher prevalence of anterior TWI (n = 39, 9% in females vs. n = 29, 4% in males; p = 0.05), whereas males had higher prevalence of inferior TWI (n = 31, 5% in males vs. n = 9, 2% in females; p = 0.02).
There were no significant differences in the prevalence of abnormal ECG between athletes with normal LV geometry compared with those with abnormal geometry (8% vs. 10%; p = 0.33). Among athletes with abnormal geometry, there were no significant differences in the prevalence of abnormal ECG between athletes who exhibited concentric hypertrophy/remodeling and those with eccentric hypertrophy (11% vs. 10%; p = 0.63).
In relation to a potential diagnosis of pathological LVH, none of the athletes with concentric LVH/remodeling showed TWI in the lateral leads, ST-segment depression, or pathological Q waves. Five athletes had inferior TWI, 6 had anterior TWI that was confined to V1-V2, and to V1-V3 in 3 cases. Five athletes showed left atrial enlargement in isolation and 8 showed left axis deviation in isolation.
Diagnosis and subsequent investigations
Thirty-four athletes (3.1%) revealed minor congenital/valvular abnormalities at echocardiography (bicuspid aortic valve, n = 6 [0.6%]; mitral valve prolapse, n = 2 [0.2%]; patent foramen ovale, n = 11 [1%]; possible patent foramen ovale, n = 2 [0.2%]; mild aortic regurgitation, n = 8 [0.7%]; mild mitral regurgitation, n = 2 [0.2%]; possible cor triatriatum, n = 2 [0.2%]; and mild pulmonary stenosis, n = 1 [0.1%]). One athlete (0.1%) was diagnosed with the Wolff-Parkinson-White ECG pattern.
A further 64 athletes (6%) revealed features at initial evaluation that, after ECG and echocardiogram, warranted further investigation to exclude cardiac disease. These included 1 athlete with a family history of cardiomyopathy in a first-degree relative and 8 athletes with possible cardiac symptoms. Forty-two athletes (3.8%) had an abnormal ECG, including 3 with a prolonged QT interval, 31 with TWI, and 8 with ventricular extrasystoles. Thirteen athletes revealed structural abnormalities at the echocardiogram that overlapped with cardiomyopathy, including 2 athletes with LV enlargement and mildly reduced LV systolic function (ejection fraction 48% in both cases), 2 mild LVH (>13 mm), and 9 athletes with significant right ventricular enlargement and mild systolic dysfunction. Athletes with abnormalities at the echocardiogram and TWI, multiple VEs, family history, and symptoms were investigated comprehensively to exclude pathology. All 64 athletes underwent an exercise stress test and a 24-h Holter and 45 athletes (1 with family history of cardiomyopathy, 31 with abnormal TWI, and 13 that showed structural features overlapping with cardiomyopathy) were subject to cardiac magnetic resonance. None of the athletes showed any features of cardiomyopathy on further assessments.
The average difference between 2 independent readers (interobserver variability based on 80 echocardiograms) was 1.8 ± 0.4 mm for LVEDD (Kappa interobserver coefficient of 0.86), 0.6 ± 0.2 mm for interventricular septum (Kappa interobserver coefficient of 0.79), and 0.5 ± 0.2 mm for posterior wall (Kappa interobserver coefficient of 0.79).
Data relating to cardiac adaptation in female athletes are relatively limited particularly with respect to changes in LV geometry associated with exercise. As previously reported in a smaller cohort of exclusively male athletes (8), this study reinforces the concept that LV geometry is also normal in most female athletes. Importantly, this study shows the effect of sex on LV geometry according to sporting discipline. Whereas 15% of male athletes engaged in dynamic exercise exhibit concentric LVH or remodeling, females predominantly develop eccentric LVH.
LV geometry in athletes
In 1975, Morganroth et al. (17) described LV adaptation in athletes using echocardiography and concluded that athletes involved in dynamic exercise had a greater LVM because of a greater LV end-diastolic volume, whereas athletes involved in static exercise were found to have a greater LV wall thickness than nonathletes. These findings have been challenged by at least 2 studies in male athletes. Spence et al. (18) studied 23 male subjects randomly assigned to dynamic or static training using cardiac magnetic resonance and speckle tracking echocardiography and showed that eccentric cardiac hypertrophy was frequently observed in dynamic sports, whereas static training was not associated with any substantive LV remodeling. Utomi et al. (8) recently showed that normal LV geometry was predominant in a cohort of athletes engaging in dynamic (n = 18) and static (n = 19) sports, with 30% of the athletes participating in dynamic sports demonstrating eccentric hypertrophy.
In contrast we observed that a significant proportion of athletes involved in dynamic, static, or mixed training showed concentric remodeling/hypertrophy and the discrepancy with previous observations may be related to the fact that in our athletes the training hours per week were significantly higher. Athletes with concentric remodeling/hypertrophy showed normal systolic and diastolic function. Moreover, none revealed ECG abnormalities characteristic of HCM including TWI in the lateral leads, ST-segment depression, or pathological Q waves.
Female athlete’s heart
Most studies evaluating cardiac adaptation in athletes have focused on males, although the past 3 decades have observed an exponential increase in the number of women participating in competitive sport (2,3,19,20). The physiological and morphological differences between males and females are likely to affect cardiac adaptation to exercise, but few studies have specifically investigated this issue.
In the largest study of female athletes to date, Pelliccia et al. (7) assessed 600 elite female athletes and showed that women did not exhibit an LV wall thickness >12 mm and rarely revealed significant LV cavity enlargement (LVEDD >54 mm). These results have important clinical implications especially in terms of screening, because they define the physiological upper limits of exercise cardiac adaptation in females. However, absolute dimensions may not be particularly sensitive in distinguishing physiology from pathology, whereas LV geometry may be a more useful tool in this setting. Our study confirmed previous findings and showed that none of the females revealed a LV wall thickness >12 mm and only 7% exhibited a LVEDD >54 mm. Although absolute LVEDD in females was lower than in males, when the LV measurements were indexed for BSA, women exhibited a higher LV cavity size. Assessment of LV geometry in athletes engaged in dynamic sports showed that females exhibit predominantly eccentric LVH, whereas a significant proportion of males (15%) show concentric LV remodeling/LVH. Possible mechanisms underlying the quantitative and geometric differences between males and females include higher circulating concentration of testosterone and a higher density of myocardial testosterone receptors in males (21). Recent data suggest that testosterone and its highly active metabolite dihydrotestosterone has a prohypertrophic effect on murine cardiac myocytes (22,23). Additionally antiandrogenic therapy has been shown to reverse pathological cardiac hypertrophy in murine models (24). Higher peak exercise-related systolic blood pressure may also play an important role in the development of LVH in males (25). Finally, it is also possible that there are quantitative and qualitative differences to the pattern of training among men that may promote increased hypertrophy.
Although the absolute prevalence of abnormal ECGs was similar in male and female athletes, we observed significant differences in the ECG patterns, suggesting the need of a “sex-based” approach for interpreting the athlete’s ECG.
At odds with recent reports, this study revealed that some athletes do develop concentric remodeling/hypertrophy, especially males. Concentric LVH/remodeling is observed in only 4% of female athletes, involved in purely dynamic exercise. This study also demonstrated that none of the female athletes exhibit an RWT >0.48 or an LVM >145 g/m2. Although none of the female athletes with concentric LVH/remodeling in this study showed other features of pathological LVH, these cutoff values may be an important starting point for the differential diagnosis with hypertrophic cardiomyopathy in a female athlete with cardiac symptoms or abnormal ECG.
We indexed cardiac dimensions for BSA, but this may not be the most accurate method of scaling LV size in athletes. Other studies have considered height, lean body mass, and allometric scaling (26,27). However, most American and European guidelines suggest reference values normalized per BSA. Our study was conducted on a cohort of highly trained elite athletes (21 ± 8 h/week of exercise) and these results may not be applicable to recreational athletes. Finally, because of our inability to access a substantial number of black female athletes during the study period, we focused solely on white athletes and the results should not be extrapolated to black athletes.
Sex has an important effect on cardiac adaptation to exercise. Most elite athletes manifest normal LV geometry. A small proportion shows concentric hypertrophy/remodeling, with a higher prevalence in males (12% vs. 7%). Conversely, a significant subset (21%) of females adapt by developing eccentric hypertrophy, particularly those engaged in dynamic sports.
COMPETENCY IN MEDICAL KNOWLEDGE: Left ventricular geometry is often normal in elite athletes, but generally females tend to commonly develop eccentric hypertrophy, whereas males exhibit frequently concentric remodeling/hypertrophy. None of the female athletes exhibited a relative wall thickness >0.48 or a LV mass >145 g/m2.
TRANSLATIONAL OUTLOOK: In an era where the number of females participating in endurance events is increasing, a better understanding of cardiac adaptation in female athletes is needed. Further studies are required to validate these findings and to explore the intrinsic mechanisms underlying the sex effect on physiological responses to exercise.
The study is funded by the charitable organization CRY. All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- body surface area
- left ventricle
- left ventricular end-diastolic diameter
- left ventricular hypertrophy
- left ventricular mass
- relative wall thickness
- T-wave inversion
- Received March 21, 2016.
- Revision received August 5, 2016.
- Accepted August 12, 2016.
- 2017 American College of Cardiology Foundation
- George K.P.,
- Warburton D.E.R.,
- Oxborough D.,
- et al.
- Arbab-Zadeh A.,
- Perhonen M.,
- Howden E.,
- et al.
- Pelliccia A.,
- Maron B.J.,
- De Luca R.,
- Di Paolo F.M.,
- Spataro A.,
- Culasso F.
- Drezner J.A.,
- Ashley E.,
- Baggish A.L.,
- et al.
- Chandra N.,
- Bastiaenen R.,
- Papadakis M.,
- Sharma S.
- Utomi V.,
- Oxborough D.,
- Ashley E.,
- et al.
- Sheikh N.,
- Papadakis M.,
- Ghani S.,
- et al.
- Lang R.M.,
- Badano L.P.,
- Mor-Avi V.,
- et al.
- Mitchell J.H.,
- Haskell W.,
- Snell P.,
- Van Camp S.P.
- Bhella P.S.,
- Hastings J.L.,
- Fujimoto N.,
- et al.
- Pelliccia A.,
- Kinoshita N.,
- Pisicchio C.,
- et al.
- Svartberg J.,
- von Mühlen D.,
- Schirmer H.,
- Barrett-Connor E.,
- Sundfjord J.,
- Jorde R.
- Marsh J.D.,
- Lehmann M.H.,
- Ritchie R.H.,
- Gwathmey J.K.,
- Green G.E.,
- Schiebinger R.J.
- Cavasin M.A.,
- Sankey S.S.,
- Yu A.-L.,
- Menon S.,
- Yang X.-P.
- Zwadlo C.,
- Schmidtmann E.,
- Szaroszyk M.,
- et al.
- Bella J.N.,
- Devereux R.B.,
- Roman M.J.,
- et al.