To the Editor:

Physiologic left ventricular (LV) hypertrophy is common in endurance-trained athletes (1). Controversially, a recent report from Japan suggested that a new upper limit for physiological LV hypertrophy might be present in ultramarathon runners (2). Specifically, in 291 male Japanese runners, 11% had a left ventricular internal dimension (LVIDd) >70 mm. Furthermore, maximal septal (interventricular septal thickness at end-diastole [IVSd]: 19 mm), posterior wall (posterior wall thickness at end-diastole [PWTd]: 15 mm), left atrial (left atrial dimension [LAD]: 49 mm), and aortic root (aortic root dimension [AoRt]: 50 mm) dimensions were considerably larger than previously suggested (1). These data generated considerable critical comment and concern (3) and present problems for cardiovascular screening of such athletes. Interpretation of the data of Nagashima et al. (2) is further limited by no assessment of LV function, and extrapolation to female athletes is required. Given these issues, it was important to replicate this work in a broad age range of male and female ultramarathon runners training and competing in North America. We hypothesized that LV structure and function in North American ultramarathon runners would not exhibit the extreme levels of hypertrophy observed in Japanese athletes.

Over a period of 3 years (2007 to 2009) we recruited 165 (39 women) ultramarathon runners (mean age [range]: 44 ± 9 [24 to 76] years) at the Western States Endurance Run, a 100-mile trail run in California. All runners were examined 1 to 3 days before race start in a field laboratory. The cohort self-reported no personal or early family history of cardiovascular disease, no current medication use, and training details as follows: 16 ± 11 (2 to 48) training years, 59 ± 21 (15 to 120) running miles/week before the event, and 46 ± 53 (3 to 500) completed ultramarathon runs. Initially we recorded data for: body mass: 70.9 ± 12.1 (41.0 to 100.5) kg; height: 1.75 ± 0.10 (1.47 to 1.97) m; body surface area: 1.86 ± 0.20 (1.35 to 2.19) m²; resting heart rate: 58 ± 9 (38 to 83) beats/min; and systolic: 117 ± 10 (90 to 148) mm Hg and diastolic blood pressure: 76 ± 8 (48 to 90) mm Hg.

All runners underwent an echocardiographic scan following American Society of Echocardiography guidelines with a commercially available ultrasound imaging system (Vivid I, GE Healthcare, Ltd., Horton, Norway). Conventional parasternal M-mode recordings of the LV were obtained for the assessment of LVIDd, IVSd, PWTd, AoRt, and LAD. The LV mass was calculated, and all structural data were allometrically scaled (4) to remove the influence of body surface area (BSA). Apical 2- and 4-chamber data were acquired for the assessment of LV end-diastolic and end-systolic volumes, with Simpson's biplane method, and the estimation of ejection fraction. Transmitral peak early and atrial filling velocities were recorded with pulsed-wave Doppler and peak septal systolic and early diastolic mitral annular tissue velocities were acquired with color tissue Doppler imaging. From these data the ratios peak early to atrial LV filling velocities and peak early filling velocity/early diastolic mitral annular tissue velocity were calculated. The LV structural and functional data were described for the whole cohort, and differences between male and female runners were assessed with independent t tests.

Data for absolute and body-sized scaled LV dimensions as well as LV function are contained in (Table 1). Maximal data for IVSd (14 mm), PWTd (12 mm), LVIDd (62 mm), AoRt (38 mm), and LAD (47 mm) did not approach data reported in the Japanese study (2) but were similar to multiple studies from North America or Europe (1). In all runners, including those with the largest LV dimensions, resting LV function was normal. The current data allay fears from clinicians related to the potential for high false positive rates and significant diagnostic uncertainty when performing cardiovascular screening in ultramarathon runners. Indeed, the potential for an ultramarathon runner to present for a cardiovascular screening session in North America with an LV, AoRt, or LAD within or even beyond the “grey zone” is reduced significantly.

The explanation for the substantial differences in cardiac dimensions between Japanese and North American ultramarathon runners is probably complex. It is unlikely related to mean or age range per se, because these data were similar in both groups. Interestingly, age partially predicted LV morphology in Japanese runners but not in the current study. It is unlikely that the Japanese runners were more highly trained, but comparison of training status was difficult on the basis of available data (1). Again, training data did partially predict cardiac dimensions in Japanese runners but did not in our cohort despite heterogeneous training and performance data. Conversely, although BSA was associated with cardiac dimensions in North American ultramarathon runners, this was not the case in Japan, where mean BSA was lower but the range was similar to the current study. Whether the extreme cardiac dimensions in some Japanese athletes were representative of some pathology cannot be determined, because no data for LV function were presented. Absence of pathology in the North American runners is supported by normal LV function and a lack of self-report indication of cardiovascular disease risk; however, we cannot conclusively rule out early or silent disease in the current cohort, and this could be assessed in future studies. A genetic contribution to the larger cardiac dimensions in Japan cannot be ruled out but is impossible to identify from the current studies.

Mean and maximum absolute cardiac dimensions were significantly higher in male ultramarathon runners, which mirror data from other studies of male and female endurance athletes (5). When cardiac dimensions were scaled allometrically for individual differences in BSA, most gender differences were reduced, removed, or—in the case of LV end-diastolic volume—reversed. Clearly, this suggests that a large component of any sex-based difference in cardiac morphology is due to differences in body size and composition. Furthermore, this provides more support for the need to scale all cardiac morphology data to make appropriate intra- and inter-group comparisons (4). Data for LV function were not different between sexes.

In summary, the LV, left atrial, and aortic root dimensions of male and female ultramarathon runners training and competing in North America did not approach the exaggerated limits previously reported in Japanese athletes (2). The LV function was normal, and scaling of cardiac morphology data for BSA removed most differences between male and female runners. These data have clear implications for the differentiation of pathological from physiological LV hypertrophy when screening such athletes.