Left ventricular hypertrophy is a notable cardiovascular disease risk factor (13) associated with excess cardiovascular mortality and morbidity (16). If left ventricular mass is to be measured routinely and is to inform clinical decision making, care must be taken to report appropriate indices (7). Left ventricular mass is determined principally by body size and is associated in particular with lean body mass (1,4). Therefore, absolute measures of left ventricular mass must be normalized appropriately to provide a size-independent outcome for single evaluations of individuals, between-group comparisons (cross-sectional or randomized trials), or tracking individual patients or groups over time (2). Moreover, correct scaling for body size informs the development of reference norms to help differentiate physiologic from pathologic hypertrophy (7). Recently, using magnetic resonance imaging (MRI), we found that left ventricular mass was related to lean body mass with an allometric (power function) exponent of 0.90 ± 0.15 (11). Hence, for a pragmatic clinical index, left ventricular mass could be expressed as a simple ratio to lean body mass, as we have reported previously using less precise echocardiographic and anthropometric methods (1,10).
Cross-sectional comparisons typically demonstrate greater left ventricular mass in athletes (20). It is unclear from such designs, however, whether the relative left ventricular hypertrophy is due to physiological training adaptations per se or is simply related to the increased body size in athlete groups. In a recent study (28), endurance-trained male athletes had substantially greater left ventricular mass than sedentary controls. However, after ratiometric normalization by fat-free mass, the between-group variability was reduced, with the differences no longer statistically significant. The authors concluded that the apparent "athlete's heart" could simply be a consequence of training-induced increases in fat-free mass.
The suggestion that increased left ventricular mass with exercise training occurs merely in line with increases in lean body mass (or fat-free mass) requires interrogation, however, in a longitudinal design, with proper evaluation of whether these two measures increase in direct proportion. The limited evidence from longitudinal studies suggests that exercise training results in both greater left ventricular mass and lean body mass (20). Furthermore, Myerson et al. (18) reported an increase in left ventricular mass that remained even after ratio adjustment for lean body mass in participants with the DD angiotensin-converting enzyme genotype undergoing a 10-wk physical training program. However, this normalization method relies on two factors. First, the data must fit a ratio model, and second, the precise relationship between left ventricular mass and lean body mass must remain constant from baseline to follow-up. Teasing out the degree of left ventricular mass growth accounted for by increases in lean body mass necessitates a repeated-measures allometric model to adjust properly for the relationship between the two measures over the course of the study.
It has been argued that a failure to consider body size appropriately in evaluating cardiovascular structure when screening adults might lead to errors in the recognition and treatment of cardiovascular disease states (7). Importantly, in this regard, if the magnitude of growth in left ventricular mass with exercise training is dissociated from that in lean body mass, then reference norms based on indexing simply for body size could be confounded in highly active or physically trained populations. In a recent editorial, Sedehi and Ashley (25) highlighted the scant attention given currently to the influence of body size (lean body mass) and training adaptations in attempting to tackle the key challenge in cardiovascular screening of distinguishing "normal" from "abnormal." There are no published studies addressing the relative growth of left ventricular mass and lean body mass in response to exercise training using correct statistical modeling and the most precise measurement of outcomes. The purpose of this study, therefore, was to model the longitudinal relationship between changes in left ventricular mass and lean body mass (derived from MRI) to ascertain whether left ventricular hypertrophy and lean body mass increases occur at the same magnitude in response to exercise training.
Our sample consisted of 116 male Caucasian British Army recruits (age 17-28 yr) undergoing both cardiac and whole-body MRI as part of a previous study (18). Ethics approval was obtained from the Royal Brompton Hospital Ethics Committee and the Army Medical Services Research Executive. All recruits provided written informed consent, and the investigation conforms to the principles outlined in the Declaration of Helsinki.
All participants were normotensive and apparently healthy. Height and body mass were recorded using standard procedures. Participants were scanned using a custom-built mobile MRI scanner situated at the Army regiment (0.5 T, imaging software; Surrey Medical Systems, Surrey, United Kingdom), and image analysis was performed by one observer during a single period using a commercially available software (CMRtools©; Imperial College, London, United Kingdom). MRI is the criterion method for both left ventricular mass measurement (19) and body fat analysis (8) and does not rely on assumptions of cardiac geometry or fat distribution throughout the body, with superior accuracy and reproducibility for both measurements (8,19). Participants were measured at baseline and after 10 wk of standard British Army basic training. This included five weekly 2-h sessions of intensive mixed endurance and strength training, concentrating on upper body strength and lower body endurance training (principally running, occasionally with external load carriage). In addition, substantial amounts of the remaining training time included physical training, for example, carrying equipment, lengthy treks, team ball games, swimming, agility sessions, and so forth. We were not permitted access to the specific details of the physical training sessions conducted; however, Williams (29) described the range and modes of activity typical of a standard British Army basic training program, and this is representative of the training conducted in the current study. Access to military fitness test data on pretraining and posttraining 1.5-mile run time (s) was obtained, providing a practical indicator of the efficacy of the training program.
Measurement of left ventricular mass.
All images were obtained using a standard protocol (17) with prospective ECG gating and a single breath-hold cine sequence for each slice. Briefly, a stack of left ventricular short-axis cine images was obtained (fast low-angle shot sequence; field of view = 400 × 400 mm2, temporal resolution = 100 ms) covering the length of the left ventricle in contiguous 10-mm slices. Manual tracing of end-diastolic epicardial and endocardial borders was performed. Myocardial volume was determined by summing the epicardial area from each short-axis slice and then removing the summation of endocardial areas. Left ventricular mass was estimated by multiplying myocardial volume by the specific density of myocardial tissue (1.05 g·cm−3).
Measurement of body composition.
Whole-body MRI was performed according to an established protocol (23). Forty transaxial spin-echo images (echo time = 40 ms, repetition time = 500 ms, field of view = 450 × 450 mm2, slice thickness = 10 mm) were obtained, covering the whole body in participants lying prone with arms outstretched. The image slices were noncontiguous, with a 40-mm gap between slices such that the leading edge of each slice was 50 mm from the next. Signal averaging was performed to reduce respiratory motion artifact, with four averages for the abdominal region and two for the pelvic region (intervertebral disc L4/5 and below). Adipose tissue (which appears as a high signal on spin-echo MRI) was measured in each image using a semiautomated threshold technique (30) with manual inspection to ensure correct segmentation. Separation into visceral and subcutaneous adipose tissue was performed using a manual tracer. Areas of bone were excluded from the analysis because of the high signal intensity from marrow lipid. Total adipose tissue volumes were calculated using the Cavalieri technique (5) of averaging the area of adipose tissue between two images using a truncated pyramid formula. This volume was multiplied by the density of adipose tissue (0.95 g·cm−3) to obtain the fat mass. Lean body mass was calculated by subtracting the total fat mass from total body mass (weight).
Data analysis and statistics.
We adopted a within-subjects allometric model to determine the change in left ventricular mass from before to after exercise training, adjusted for the change in lean body mass (SAS® PROC MIXED, SAS® Version 9.1; SAS Institute, Inc., Cary, NC). First, the pre- and posttraining left ventricular mass and lean body mass values were naturally log-transformed to linearize Huxley's general allometric equation (Y = aXb) (24). The change in left ventricular mass with training was then modeled as the dependent variable, with change in lean body mass as the independent variable, with the change in left ventricular mass adjusted allometrically for change in lean body mass. The within-subjects power function exponent (b) in the longitudinal relationship left ventricular mass = a × (lean body mass)b was also calculated to determine the precise nature of the relationship. Finally, the unadjusted change in left ventricular mass was calculated to examine the degree of change that can be accounted for by changes in lean body mass. The effect was derived as a percent change via back-transformation of the log-log estimate. In accordance with recent recommendations, uncertainty surrounding point estimates of effects was expressed using 90% confidence intervals (14,27).
In the absence of a robust clinical anchor for the minimum important physiological change in left ventricular mass, we adopted a distribution-based approach in predefining a small standardized mean change (Cohen's d) of 0.2 between-subject SD (6) as the minimum important difference (the smallest increase in left ventricular mass worth detecting from a clinical or practical standpoint). We then based our inference on the disposition of the confidence interval for the effect to this threshold; given the observed effect, the probability that the true population change in left ventricular mass (unadjusted and adjusted models) is greater than or equal to the predefined minimum important change was calculated and presented (3,9,26). Qualitative probabilistic terms were assigned using the following scale (14): <0.005, most unlikely (to be important), almost certainly not; 0.005-0.05, very unlikely; 0.05-0.25, unlikely, probably not; 0.25-0.75, possibly; 0.75-0.95, likely, probably; 0.95-0.995, very likely; >0.995, most likely, almost certainly.
Measures of centrality and dispersion are mean ± SD. For variables that were log-transformed before modeling, the mean shown is the back-transformed mean of the log transform, and the dispersion is the SD expressed as a coefficient of variation (%) (14). Improvement in the 1.5-mile run time was quantified using the mean difference between the pre- and posttraining scores, together with its 90% confidence interval (paired t-statistic). For this analysis, the minimum important change was again defined as a standardized mean difference of 0.2 between-subject SD, with a qualitative probabilistic term assigned as described above.
Summary data for the sample are presented in Table 1. The left ventricular mass increased after training by 4.8% (90% confidence interval = 3.5%-6%) or approximately 9 g. The probability that the true population change in left ventricular mass is greater than or equal to the minimum important change is >0.999-almost certainly important. This unadjusted change in left ventricular mass is equivalent to an increase of 0.34 SD (0.25-0.43). Lean body mass increased by 2.6% (2.1%-3.0%) or approximately 1.5 kg. The change in left ventricular mass adjusted for the change in lean body mass using the within-subjects allometric model was 3.5% (1.9%-5.1%). The probability that the true population change in the adjusted left ventricular mass is greater than or equal to the minimum important change is 0.78-likely to be important. (There is a probability of 0.22 that the true change is "trivial," that is, within plus or minus the minimum important change and a probability of <0.001 of a meaningful decrease in left ventricular mass.) The adjusted change in left ventricular mass is equivalent to an increase of 0.25 SD (0.14-0.37). The within-subjects allometric exponent (b) for lean body mass was 0.49 (0.11-0.88). The 1.5-mile run time improved after training by 29 s (24-34 s). The probability that the true population change in 1.5-mile run time is greater than or equal to the minimum important change is >0.999-almost certainly important.
This study reveals that, after correcting for allometric growth rates in response to exercise training, the magnitudes of the increases in left ventricular mass and lean body mass are dissociated. The size of the unadjusted increase in left ventricular mass observed in this study (4.8%) has been reported previously (18) and is comparable to other samples undergoing similar training. After accounting allometrically for the change in lean body mass, the increase in left ventricular mass after training was less but remained a substantial effect. Indeed, only about one-third of the observed change in left ventricular mass is accounted for by the change in lean body mass; this effect has not been properly evaluated in previous cross-sectional or longitudinal studies. Our finding runs counter to the conclusion of Whalley et al. (28) on the basis of cross-sectional data that physiological left ventricular hypertrophy merely reflects a response to increases in lean body mass. This finding highlights the problems with cross-sectional studies of athletes and untrained controls, where selection bias can be a threat to validity, and underlines the need for full and proper evaluation in longitudinal studies. Our finding that left ventricular hypertrophy and lean body mass changes do not occur at the same magnitude in response to exercise training supports that of Myerson et al. (18), who used ratio normalization before and after training in participants with angiotensin-converting enzyme gene deletion polymorphism. Previously postulated links between skeletal and cardiac muscularity (12) imply a role for testosterone as a primary signal messenger; however, the fact that in the current study the relative increases in left ventricular mass and lean body mass were dissociated might point to additional mechanisms for exercise-induced increases in left ventricular mass. These additional mechanisms could include extra hemodynamic load (volume and pressure) resulting in concentric and eccentric left ventricular growth (20). This potential mechanism is consistent with the exercise stimuli in the current study, involving a mix of both aerobic endurance and static and dynamic resistance exercise.
The within-subjects allometric lean body mass exponent for the scaling of left ventricular mass was 0.49. This value seems lower than the between-subjects lean body mass exponents observed typically in cross-sectional studies. For example, our group recently reported a lean body mass exponent of 0.90 in the scaling of MRI-derived left ventricular mass in army recruits (11), a similar magnitude to that we reported previously in recreationally active men and women (1). Whereas, in principle, the between-subject allometric relationship applies to serial changes in the same participants, the repeated-measures allometric exponent could be attenuated more by noise in the measurement of changes in lean body mass within subjects. (Importantly, there is √2 more noise in the changes in lean body mass than in the single measures of lean body mass.) In short, intersubject differences in the change in lean body mass are small compared with the absolute differences in lean body mass between participants, and the lean body mass exponent is therefore likely to be attenuated by measurement error. It is noteworthy, however, that the confidence interval for the within-subjects lean body mass exponent is relatively wide and overlaps substantially with those reported in cross-sectional between-subject scaling studies (1,11).
It is important to acknowledge some limitations to our study. First, the changes in left ventricular mass were modest, although in line with previously reported exercise-induced changes, and this might limit the examination of any moderators of left ventricular growth. The sample, undergoing almost identical exercise training, was relatively large, however, and identifying large enough groups of participants undergoing higher levels of the same standardized exercise training program is difficult. Indeed, although the specific physical training undertaken was necessarily not as precisely defined as is usually the case in the exercise physiology laboratory, the recruits all underwent the same group-based training sessions, and the sample was homogeneous for age, sex, diet, and physical environment. This homogeneity is a key strength of the study and helps counter any lack of precision in the detail of physical training sessions. Second, this study involves a cohort of military recruits, and there is no contemporaneous control group. Hence, it could be argued that one cannot confidently attribute the observed changes in left ventricular mass (or indeed in lean body mass) to the exercise training per se rather than to the maturation in left ventricular size. However, exercise has been used as a physiological model of left ventricular hypertrophy to understand pathological hypertrophy because it results in much more rapid changes (left ventricular growth) (22), and this would seem implausibly fast to result from maturation. Notwithstanding the lack of a comparison group (which was not possible given the military context), we feel that the observed effect most likely results from the 10-wk intensive exercise training program. Furthermore, the substantial improvement in the 1.5-mile run time observed is a robust practical indicator of the efficacy of the training program in leading to beneficial physiological adaptations. (We appreciate that there is the same history and maturation threat to the validity of this apparent fitness improvement; however, in the context of institutionalized military basic training, this threat is largely theoretical, and we are confident that the effect is due to the training program.) Third, our finding applies only to young male Caucasian participants, and examination of its robustness in older, female, other ethnic group, and overweight and obese samples is warranted. Fourth, outcome measurement occurred only at baseline and after the 10-wk training program. Additional time points could have yielded important information about the nature of the longitudinal development of the left ventricular and lean muscle hypertrophy and provided insight as to whether these changes occur in parallel or whether the increase in lean body mass precedes that in left ventricular mass. However, this would have been disturbing to the military training process and would not have been permitted. Finally, if the repeated-measures lean body mass exponent is attenuated, as described above, then the effect size for the increase in left ventricular mass not accounted for by the change in lean body mass might be overestimated. However, the magnitude of the adjusted increase in the current study is such that we are confident that a left ventricular hypertrophy independent of increases in lean body mass was indeed present.
We believe that it is an ethical requirement to define, a priori, what one considers to be the minimum important change in the outcome variable because poststudy inferences can only be drawn properly in relation to this value. However, other researchers and clinicians might take issue with our choice of the smallest worthwhile change in left ventricular mass. We adopted a distribution-based approach linked to a small standardized mean difference of 0.2 between-subject SD (6,14). This approach is one of three accepted formal methods for defining the targeted difference (15). The two other approaches are an anchor-based method (the change in left ventricular mass required to result in a step change on an anchor measure that has already been shown to be clinically or mechanistically relevant) and opinion seeking. In the absence of an accepted and robust anchor for a physiological increase in left ventricular mass, we considered the distribution-based approach defining a small standardized effect size as the targeted change to be the best option. Moreover, it has been shown that distribution- and anchor-based approaches to predefining the targeted difference provide essentially equivalent information (21). However, those wishing to use a different threshold for the minimum important change in left ventricular mass can use the disposition of our presented confidence intervals to their own choice of targeted change to draw inference.
In conclusion, this is the first study to demonstrate conclusively that, in response to exercise training, hypertrophy occurs to a greater extent in the left ventricle than in total lean body mass after correcting for uneven growth rates. Critically, we used a longitudinal design, rigorous precise measurement methods for both dependent and independent variables, and appropriate within-subjects allometric modeling to adjust the increase in left ventricular mass for the influence of change in lean body mass. Our study begins to lay the foundation for the creation of a method to examine left ventricular hypertrophy in contrast to lean body mass changes in response to exercise training. Indeed, although left ventricular mass is determined principally by body size, we have shown that it is influenced beyond this by physical training; establishment of a method to create standardized norms that reflect this relationship could result in a potentially useful clinical diagnostic tool. Failure to account properly for the relationship between lean body mass and left ventricular mass consequent to chronic exercise training could lead to many false positives in screening for cardiovascular disease states. In addition, longitudinal allometric modeling of this type in future studies might help to tease out the specific mechanisms underpinning differential growth rates of various tissues and organs. Further research is required to examine the robustness of our findings in diverse populations and to explore the mechanisms for this independent exercise-induced left ventricular hypertrophy.
This work was supported by the British Heart Foundation (FS 97030 to S.G.M.). S.G.M. is supported by the Oxford Partnership Comprehensive Biomedical Research Centre with funding from the Department of Health's National Institute for Health Research Biomedical Research Centre's funding scheme.
The authors have no conflicts of interest to declare.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Batterham AM, George KP, Mullineaux DR. Allometric scaling of left ventricular mass by body dimensions in males and females. Med Sci Sports Exerc
2. Batterham AM, George KP, Whyte G, Sharma S, McKenna W. Scaling cardiac structural data by body dimensions: a review of theory, practice, and problems. Int J Sports Med
3. Batterham AM, Hopkins WG. Making meaningful inferences about magnitudes. Int J Sports Physiol Perform
4. Bella JN, Devereux RB, Roman MJ, et al. Relations of left ventricular mass to fat-free and adipose body mass: the strong heart study. Circulation
5. Cavalieri B. Geometria Indivisibilibus Continuorum. Typis Clementis Feronij, Bononiae, 1635
. Reprinted as Geometria degli Indivisibili
. Torino (Italy): Unione Tipografico-Editrice Torinese; 1966. p. 20. Italian.
6. Cohen J. Statistical Power Analysis for the Behavioral Sciences
. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum Associates; 1988. p. 567.
7. Dewey FE, Rosenthal D, Murphy DJ Jr, Froelicher VF, Ashley EA. Does size matter? Clinical applications of scaling cardiac size and function for body size. Circulation
8. Fowler PA, Fuller MF, Glasbey CA, Cameron GG, Foster MA. Validation of the in vivo
measurement of adipose tissue by magnetic resonance imaging of lean and obese pigs. Am J Clin Nutr
9. Froehlich GW. What is the chance that this study is clinically significant? A proposal for Q
values. Eff Clin Pract
10. George KP, Batterham AM, Jones B. The impact of scalar variable and process on athlete-control comparisons of cardiac dimensions. Med Sci Sports Exerc
11. George KP, Birch KM, Pennell DJ, Myerson SG. Magnetic resonance imaging-derived indices for the normalization of left ventricular morphology by body size. Magn Reson Imaging
12. George KP, Wolfe LA, Burggraf GW. The 'athletic heart syndrome.' A critical review. Sports Med
13. Ghali JK, Liao Y, Simmons B, Castaner A, Cao G, Cooper RS. The prognostic role of left ventricular hypertrophy in patients with or without coronary artery disease. Ann Intern Med
14. Hopkins WG, Marshall SW, Batterham AM, Hanin Y. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc
15. Lassere MN, van der Heijde D, Johnson KR. Foundations of the minimal clinically important difference for imaging. J Rheumatol
16. Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Heart Study. N Engl J Med
17. Lorenz CH, Walker ES, Morgan VL, Klein SS, Graham TP Jr. Normal human right and left ventricular mass, systolic function, and gender differences by cine magnetic resonance imaging. J Cardiovasc Magn Reson
18. Myerson SG, Montgomery HE, Whittingham M, et al. Left ventricular hypertrophy with exercise and ACE gene insertion/deletion polymorphism: a randomized controlled trial with losartan. Circulation
19. Myerson SG, Montgomery HE, World MJ, Pennell DJ. Left ventricular mass measurement: reliability of M-mode and 2-dimensional echocardiographic formulas. Hypertension
20. Naylor LH, George K, O'Driscoll G, Green DJ. The athlete's heart: a contemporary appraisal of the 'Morganroth hypothesis.' Sports Med
21. Norman GR, Sridhar FG, Guyatt GH, Walter SD. Relation of distribution- and anchor-based approaches in interpretation of changes in health-related quality of life. Med Care
22. Payne JR, Eleftheriou KI, James LE, et al. Left ventricular growth response to exercise and cigarette smoking: data from LARGE Heart. Heart
23. Ross R, Shaw KD, Martel Y, de Guise J, Avruch L. Adipose tissue distribution measured by magnetic resonance imaging in obese women. Am J Clin Nutr
24. Schmidt-Nielsen K. Scaling: Why Is Animal Size So Important?
New York (NY): Cambridge University Press; 1984. p. 14-7.
25. Sedehi D, Ashley EA. Defining the limits of athlete's heart: implications for screening in diverse populations. Circulation
26. Shakespeare TP, Gebski VJ, Veness MJ, Simes J. Improving interpretation of clinical studies by use of confidence levels, clinical significance curves, and risk-benefit contours. Lancet
27. Sterne JA, Davey Smith G. Sifting the evidence-what's wrong with significance tests? BMJ
28. Whalley GA, Doughty RN, Gamble GD, et al. Association of fat-free mass and training status with left ventricular size and mass in endurance-trained athletes. J Am Coll Cardiol
29. Williams AG. Effects of basic training in the British Army on regular and reserve army personnel. J Strength Cond Res
30. Yang GZ, Myerson S, Chabat F, Pennell DJ, Firmin DN. Automatic MRI adipose tissue mapping using overlapping mosaics. MAGMA