Sex Differences in Cardiac Adaptation to Distinct Modalities of Exercise: A Cardiac Magnetic Resonance Study : Medicine & Science in Sports & Exercise

Journal Logo


Sex Differences in Cardiac Adaptation to Distinct Modalities of Exercise: A Cardiac Magnetic Resonance Study


Author Information
Medicine & Science in Sports & Exercise 53(12):p 2543-2552, December 2021. | DOI: 10.1249/MSS.0000000000002729


Differences between males and females in terms of the magnitude of response to exercise training have been reported for many physiological variables, including fitness (1), strength, and vascular adaptation (2). However, sex differences in cardiac adaptation to training have not been comprehensively reported in humans, despite the effects of ventricular loading during exercise on chronic structural adaptation (3) and the relevance of sex hormones to myocardial hypertrophy and adaptation (4). Furthermore, animal studies on sex differences in cardiac adaptation (5–7) are of limited relevance to humans, given that bipedal posture affects myocardial loading, training loads do not translate between animals and humans (especially for resistance training), and fundamental differences exist in reproductive physiology between small animals and humans.

Sex differences in the effect of endurance exercise training in terms of changes in the size and structure of the ventricles were recently reviewed by Diaz-Canestro and Montero (8), who indicated a surprising lack of longitudinal training studies, which have directly compared males and females who completed the same training stimulus. They found only two longitudinal studies that reported on sex differences in cardiac outcomes: the first used acetylene rebreathing to measure stroke volume (SV) in response to 9–12 months of supervised endurance training (END; 15 females, 16 males; 64 ± 3 yr of age) (9), and the second used cardiac magnetic resonance imaging (CMR) to determine left ventricular (LV) volume and mass in response to 12 months END (10). Although CMR is considered the contemporary gold standard for quantifying cardiac morphology in humans (11–13), the latter study (10) involved comparison of only 4 females and 7 males.

Resistance training (RES) has been reported to induce distinct cardiac adaptation to that associated with END, a concept expressed in the so-called Morganroth hypothesis (3,14,15). However, sex differences in adaptation to END and RES modalities have not previously been directly studied or reported. In addition, recent studies have emphasized that a large degree of interindividual difference exists in the responsiveness to exercise training in multiple physiological outcomes (16,17), but little is known regarding sex-specific differences in the magnitude of change or the “responder” rate for cardiac adaptation.

The current study aimed to provide insight into sex-based responsiveness to adaptation resulting from distinct modes of training. To this end, we used a randomized crossover design to assess sex-specific cardiac adaptation to distinct exercise modalities (RES and END) in the same individuals. Our hypothesis was that males would exhibit larger adaptations to END and RES training than females.


Full details of the study design and experimental procedures can be found in our extensive protocol article (18) and in the study registration (ACTRN12616001095459), which was published before subject recruitment and randomization. A summary is provided below. One aim of the experiment was to compare responses in monozygotic and dizygotic twins. However, this manuscript focuses on sex differences in CMR outcomes and was underpowered for twin analysis in this regard. Nonetheless, we have adjusted all analyses for twin correlations (see Statistical Analysis section).


Initial cross-sectional testing included 72 subjects (individuals: 46 females and 26 males; see Figure, Supplemental Digital Content 1, Consort diagram of participant recruitment and inclusion in the study, After this, 60 individuals (36 females and 24 males) participated in the training interventions with 58 individuals completing both modalities. Recruitment strategies included advertising in Perth-based newspapers, online and via social media, university e-mail lists, word-of-mouth referral, etc. As described in our comprehensive study design article (18), participants were required to be apparently healthy, having no signs or symptoms of congenital heart disease, ischemic heart disease, atrial fibrillation, stroke, kidney disease, liver disease, diabetes, epilepsy, respiratory disease, and exercise-induced asthma. Resting systolic and diastolic blood pressure had to be <160 and <100 mm Hg, respectively, with body mass index <35 kg·m−2. Subjects were excluded if they were smokers, or had been a smoker within the last 12 months, had severe mental illness or joint, muscular, or spinal injuries/disease that would prohibit intense exercise. Any serious illness that would compromise survival (e.g., cancer) was an exclusion. Subjects were also required to be untrained, defined as doing less than the minimum Australian guidelines for physical activity recommendations (<150 min·wk−1 of organized exercise). They were excluded if taking any prescribed medications for previously mentioned illness/disease that would affect outcome measures of the study. Subsequent to a successful phone interview and screening, participants attended separate testing sessions (ranging from 45 min to 2 h duration), which made up the cross-sectional portion of the study and were used as additional screening for inclusion in the subsequent longitudinal training study. This study was approved by the University of Western Australia Human Research Ethics Committee (reference number RA/4/7031). Oral and written consent was obtained from each participant before participation in the study, which conformed with the Declaration of Helsinki. Baseline demographic data are displayed in Table 1.

TABLE 1 - Baseline characteristics of participants enrolled in the study.
Female Male P
n = 46 n = 26
Age (yr) 24.15 ± 4.59 27.50 ± 5.97 0.374
Height (cm) 170.95 ± 8.06 174.05 ± 6.61 <0.001
Weight (kg) 61.84 ± 10.48 76.00 ± 19.11 0.005
BSA (m2) 1.71 ± 0.17 1.90 ± 0.26 <0.001
Lean mass (kg) 41.25 ± 5.96 56.69 ± 6.56 <0.001
BMI (kg·m−2) 21.08 ± 2.67 24.93 ± 5.50 0.218
Data are presented as mean ± SD.
BSA, body surface area; FFM, fat-free mass; BMI, body mass index.

Study design

After baseline testing, subjects were randomized to participate in either the 3-month RES or the END training intervention (study design in Marsh et al. (18) and Fig. 1). Participants then underwent a 3-month washout period, during which they were instructed to maintain their usual level of activities and usual diet, before crossing over to complete 3 months of the alternate exercise intervention (RES or END). We chose a 3-month washout based on recent CMR evidence of regression of athletic hypertrophy within 1 month of detraining (19). Both interventions consisted of 3 × 1-h sessions per week, for 12 wk with full details provided in our protocol article (18). Briefly, the programs were center based, supervised by an Accredited Exercise Physiologist/Scientist and progressively overloaded: END training involved 2× running and 1× cycling session per week progressing from 60% to 90% HR maximum according to their V˙O2max test; RES training involved alternating between upper/lower major muscle groups each session and progressed from 60% (targeting muscular endurance and learning technique) to 90% (targeting muscular strength) one-repetition maximum (1RM).

Study design. Subjects were randomized together to complete 3 months of an exercise intervention (resistance [RES] or endurance [END]), followed by 3 months washout period and crossover to complete 3 months of the alternate exercise modality. Outcome measures, including CMR, cardiorespiratory fitness, and strength, were collected in response to training.

Comprehensive details of our training approaches, including the ongoing progression and overload of the training intensities, are provided in our Methods article (18). Briefly, the 12-wk END program followed a periodized progressive macrocycle plan consisting of three mesocycles of 4 wk each. The first mesocycle, the “general preparatory” phase (weeks 1–4), consisted of low training intensity/volume (walking/jogging at 60% HR, 1–2.5 km running and/or 15–25 min cycling/running), including a long warm up focusing on running drills, technique, and dynamic stretching. The second mesocycle, the high-intensity phase (weeks 5–8), consisted of higher-intensity work with higher HR (up to 90%) and distance/duration (2.5–5 km running and/or 25–40 min cycling/running). The final mesocycle, the “maintenance and distance” phase (weeks 9–12), consisted mostly of maintaining subthreshold HR (70%–85%) for longer distance/duration (5–7 km running and/or 40 min in week 9 to 60 min by week 12 cycling/running). In the mesocycles, a 1:1 structure loading existed with a “hard” loading week followed by an “easier” or “maintenance” week. Within each week, there existed a structured load with a “harder” run session and an “easier” run session, and the cycling session was always the longest duration session in each week, adapting many principles from previously published work. The 12-wk RES training program also followed a periodized plan consisting of four mesocycles of 3 wk each. The first mesocycle focused on muscular endurance (weeks 1–3) where intensity was low (60%–70% 1RM), reps were high (12–15 reps), and rest periods between sets were short (30–60 s), allowing participants to focus on forming good technique habits, condition muscles, and assist recovery between sessions. The second mesocycle was muscular hypertrophy (weeks 4–6) where intensity increased (70%–75%), reps decreased (10–12 reps), but rest periods remained short between sets (30–90 s). Weeks 7–9 were a progressive step between muscular hypertrophy and strength where intensity increased (75%–80%), reps decreased (8–10 reps), and rest periods between sets increased (1–2 min). The last mesocycle focused on improving muscular strength where intensity increased (80%–90%), reps decreased (5 reps), and rest periods between sets increased (3–5 min). Each session focused on one of the five main exercises (two upper body—bench press and standing military press; three lower body—squats, deadlift and leg press) that were rotated alternating upper and lower body on separate days. There were secondary exercises performed during each session that used muscle groups that were similar to, or would assist in performing, the main exercise of the session (i.e., staggered feet leg press, seated row, lat pulldown). To guide participants’ progressions, 1RM assessments were repeated halfway through their 12-wk program.

Primary outcome measures

Briefly, 1.5 T CMR (Magnetom Espree; Siemens, Erlangen, Germany) was completed at week 0 and immediately after each exercise modality (weeks 12 and 36) to assess LV outcomes, and analyzed with specialized software (ARGUS, Siemens; see Figure, Supplemental Digital Content 2, Analysis of the left ventricle in end-systole and end-diastole in the short- and long-axis views, The subjects were scanned supine with a posterior phased array spine coil and an anterior flexible phased array body surface coil. Multiplane breath-hold (BH) trueFISP localizers were acquired to obtain the standard cardiac imaging planes. For all sequences, the BH times were between 5 and 20 s, dependent on the subject’s HR. To evaluate functional parameters, BH trueFISP cine images were acquired using a retrogated ECG trigger covering the whole R-R interval. Images of the LV were done in short-axis plane, perpendicular to the ventricular septum. Between 10 and 12 slices were acquired (6 mm slices/4 mm gap, FoV = 320–350 mm, TR = 37.68, TE = 1.29 flip angle 70°–80°, resolution 256 × 166, BW 930). Cine images of the four-chamber and LV outflow tract were also acquired (6 mm slices, FoV = 300–330 mm, TR = 38.28, TE = 1.32, flip angle 70°–80°, resolution 224 × 224, BW 930).

To assess LV morphology, systemic and pulmonary outflow tracts were autosegmented in individual slices on cine images and then manually adjusted where necessary. Short-axis cine loops were inspected to define LV end-diastolic volume (EDV) and end-systolic volume (ESV) frames. The mitral valve markers were automatically displayed on two- and four-chamber cine views but could be manually manipulated if inconsistent tracking occurred throughout the slices. However, manual placement was required for markers of the tricuspid valve in RV analysis. The basal LV and RV slices were taken as the first slice below the level of the mitral valve and tricuspid valve, respectively. Therefore, volumes above the aortic valve and those surrounding the thin myocardial wall in the mitral valve and tricuspid valve plane were excluded from analysis. Endocardial and epicardial LV borders were automatically contoured (but could be manually manipulated if required), including the septum but excluding the papillary muscles, which were added to LV EDV and LV ESV in accordance with the methods described previously (15). Left ventricular mass (LVM) was calculated by summing the LV EDV within the epi- and endocardial borders of the short-axis slices, multiplying the myocardial tissue volume by its specific density (1.05 g·cm−3). The LV EDV and the LV ESV were used to ascertain SV, ejection fraction, and cardiac output. The RV endocardium was manually traced on each short-axis frame. For assessment of RV volumetrics, instructions for tracing the endocardial border were taken from the methodology described previously (20).

As described in our published body composition article (21), dual energy x-ray absorptiometry (Lunar iDXA; GE Healthcare, Chicago, IL) was used to assess lean body mass (LM) at all time points. Standard calibration and quality assurance procedures were used, in line with the equipment documentation ( Participants arrived at the laboratory at the same time for each of the measurements (before and after each exercise intervention), usually in the afternoon/evening. Participants were instructed to fast for 3 h, be normally hydrated (i.e., not hyper/hypohydrated or postexercise) and eliminated before the test. The inter- and intratester reliability results of our machine are very highly repeatable (intraclass correlation coefficient >0.99) in 52 tested participants.

Statistical analysis

Statistical analyses were performed with SPSS 20.0 (IBM Australia Ltd., New South Wales, Australia) and STATA 11 software (StataCorp, College Station, TX). Differences between sexes (Table 1) and the effect of the exercise interventions on each outcome measure (Tables 2 and 3) were assessed using a linear mixed model, which adjusted for twin correlations. A z-test was used to assess the differences in individual response rates between exercise interventions (RES vs END) and concordance versus discordance for each variable. Significance was α < 0.05. We used the highly cited formal CMR reproducibility data of Bellenger et al. (11) who used similar approaches to those in our study (1.5 T). Based on their data, and assuming two-tailed tests and α = 0.05, n = 58 provides 94% power to detect a change in LVM of 3 g and 99% power to detect a change in EDV of 3 mL.

TABLE 2 - Female and male LV baseline values, group mean absolute changes, and percentage changes with resistance training.
Units Baseline Δ RES
F M F M F vs M (P)
LVM g 85.29 ± 12.85 135.44 ± 27.22 −0.66 ± 10.11 −4.76 ± 10.56* 0.164
% −0.23 ± 11.28 −3.46 ± 7.66 0.583
LVM iBSA g·m−2 48.06 ± 4.67 67.12 ± 10.10 −0.40 ± 5.11 −2.72 ± 5.63* 0.133
% −0.32 ± 11.35 −3.97 ± 7.85* 0.461
LVM iLM g·kg−1 2.05 ± 0.23 2.35 ± 0.27 −0.07 ± 0.23 −0.13 ± 0.17*** 0.308
% −2.54 ± 11.42 −5.45 ± 7.18** 0.664
EDV mL 139.64 ± 21.60 177.67 ± 27.98 −1.87 ± 10.22 −0.14 ± 8.73 0.446
% −1.19 ± 7.38 0.02 ± 4.86 0.914
ESV mL 51.29 ± 9.63 72.07 ± 15.34 −0.94 ± 8.38 −0.20 ± 8.76 0.668
% −0.70 ± 15.65 0.49 ± 14.90 0.991
SV mL 88.35 ± 15.53 105.60 ± 17.83 −0.93 ± 8.13 0.06 ± 10.61 0.588
% −0.67 ± 9.31 0.47 ± 9.93 0.972
Data are presented as mean ± SD.
*P < 0.05.
**P < 0.01.
***P < 0.005.
RES, resistance training; F, female; M, male; iBSA, indexed to body surface area; iLM, indexed to lean body mass.

TABLE 3 - Female and male LV baseline values, group mean absolute and percentage changes with endurance training; comparisons with resistance training (see Table 2).
Units Baseline Δ END Δ RES vs Δ END Interaction
F M F M F vs M (P) F (P) M (P) Sex–Mode
LVM g 85.29 ± 12.85 135.44 ± 27.22 3.98 ± 7.98*** 5.99 ± 10.67** 0.394 0.024 <0.001 0.066
% 4.44 ± 8.93*** 4.47 ± 8.17* 1.000 0.155 0.025 0.276
LVM iBSA g·m−2 48.06 ± 4.67 67.12 ± 10.10 1.69 ± 3.91** 2.84 ± 5.32** 0.325 0.026 <0.001 0.032
% 3.66 ± 8.03* 4.42 ± 8.25* 0.989 0.261 0.013 0.134
LVM iLM g·kg−1 2.05 ± 0.23 2.35 ± 0.27 0.08 ± 0.18** 0.08 ± 0.18* 0.987 0.001 <0.001 0.357
% 4.07 ± 9.23** 3.58 ± 7.94 0.997 0.018 0.008 0.437
EDV mL 139.64 ± 21.60 177.67 ± 27.98 1.54 ± 10.49 7.48 ± 11.91*** 0.038 0.009 0.001 0.128
% 1.16 ± 7.16 4.50 ± 6.84*** 0.250 0.462 0.121 0.233
ESV mL 51.29 ± 9.63 72.07 ± 15.34 1.17 ± 7.48 0.23 ± 10.07 0.673 0.071 0.801 0.543
% 3.27 ± 14.17 0.66 ± 14.26 0.909 0.668 1.000 0.407
SV mL 88.35 ± 15.53 105.60 ± 17.83 0.36 ± 8.06 7.25 ± 10.65*** 0.004 0.294 0.003 0.031
% 0.55 ± 9.10 7.28 ± 10.56*** 0.045 0.950 0.084 0.048
Data are presented as mean ± SD.
*P < 0.05.
**P < 0.01.
***P < 0.005.
END, endurance training; RES, resistance training; F, female; M, male; iBSA, indexed to body surface area; iLM, indexed to lean body mass.


Compliance with completion of training sessions was 94% for RES and 95% for END training. The training interventions were effective with specific regard to muscular strength and cardiorespiratory fitness results (16). Briefly, strength increased after RES but not END (1RM for leg press △47.0 vs 3.0 kg, P < 0.001, and bench press △5.1 vs −0.4 kg, P < 0.001) and V˙O2max increased after END but not RES (△0.25 vs 0.04 L·min−1, P < 0.001; △3.61 vs 0.03 mL·kg−1·min−1, P < 0.001). A higher percentage of individuals responded to RES for strength and to END for V˙O2max (P < 0.001). Within-individual responses to each mode were not correlated (P > 0.05).

Group means and specificity of training

Group mean responses to interventions are presented in Figure 2 and Tables 2 and 3. LVM increased with END in both males (P = 0.005) and females (P = 0.002). Resistance training induced a reduction in LVM in males (P = 0.037) but not females (P = 0.691); however, RES LVM for males was nonsignificant when expressed as a percentage change from baseline (P = 0.076). The magnitude of change was similar between sexes (RES, P = 0.164; END, P = 0.394) but differed between modes for each sex, more so in males (RES vs END; females, P = 0.024; males, P < 0.001). Between modes was also significant for males when expressed as a percentage change from baseline (females, P = 0.155; males, P = 0.025).

Absolute changes (△) with resistance (RES) and endurance (END) training separated by sex (females, purple; males, blue) for LV (top panel) absolute mass and LV EDV (bottom panel) in terms of group responses for absolute change (left panel) and change as a percentage of baseline (right panel).

EDV increased with END in males (P = 0.002) but not females (P = 0.373) and remained unchanged with RES for females (P = 0.226) and males (P = 0.998). This was also the case when data were expressed as a percentage change from baseline. The magnitude of change differed between sexes for END (P = 0.038) but not RES (P = 0.446) and differed between modes for each sex (RES vs END; females, P = 0.009; males, P = 0.001). When expressed as a percentage change from baseline, the magnitude of change was similar between sexes for END (P = 0.250) and RES (P = 0.914) and between modes for each sex (females, P = 0.462; males, P = 0.121).

CMR responder rates: sex specificity

The percentage of individuals who exhibited a positive response (i.e., a change in response, or △>0; Fig. 3) for LVM did not differ between females and males for RES (42% vs 32%, P = 0.45) or END (64% vs 83%, P = 0.10). The percentage of positive response for EDV did not differ between females and males for RES (36% vs 45%, P = 0.48) and END (58% vs 79%, P = 0.09). The percentage of positive response (△>0) between modes differed for males (LVM 32% vs 83% P = 0.0004, EDV 45% vs 79% P = 0.018) but not females (LVM 42% vs 64% P = 0.06, EDV 36% vs 58% P = 0.06). Therefore, males were more responsive to END than RES training, but there was no difference between sexes for either mode.

Individual changes (△) with resistance (RES; blue) and endurance (END; yellow) training separated by sex (females, left panels; males, right panels) for LVM and EDV. Dotted lines indicate the mean differences for repeated measures using 1.5-T CMR based on the widely cited reproducibility study of Bellenger et al. (11). These data indicate that 95% (for LVM) and 77% (for EDV) of subjects for END training exhibit larger changes than those expected from repeated-measures variability. For RES, 95% (for LVM) and 72% (for EDV) of subjects exceeded these variability levels.

We used a simple criterion of >0< to assess positive and negative response rate, but some investigators have argued that measurement variability should be considered when defining responsiveness. To this end, we superimposed on Figure 3 the mean differences for repeated measures expected for repeated 1.5-T CMR based on the widely cited reproducibility study of Bellenger et al. (11). These data indicate that most subjects for LVM (96% of males and 94% of females) and EDV (83% of males and 72% of females) for END training exhibit larger changes than those expected from repeated-measures CMR variability. For RES, most subjects for LVM (96% of males and 95% of females) and EDV (73% of males and 72% of females) exceeded these variability levels. Including the margin of measurement error therefore did not change the pattern emerging from the data.

Response of individuals: concordance between modalities

Concordance graphs are displayed in Figure 4, which presents each individual’s response to both RES and END. Percentages for each quadrant, whether concordant (i.e., responded to both modalities, or did not respond to either modality) or discordant (i.e., responded to only one modality), for RES and END interventions are presented. For LVM, there were no significant differences between sexes for concordance where 61% (33% both responded and 28% neither responded) of females responded similarly to both modes, compared with 50% of males (P = 0.41). There was also no difference in concordance of EDV (56% of females vs 50% of males; P = 0.68). Pearson bivariate correlations between response to RES and END training within subjects indicated that correlations for both sexes were low for LVM (females, r = 0.07, P = 0.70; males, r = 0.22, P = 0.33) and moderate for EDV (females, r = 0.71, P < 0.001; males, r = 0.48, P = 0.02). Sex differences therefore did not substantively affect the concordance of response between modalities of training.

Individual subject (females = 36, left panel; males = 22, right panel) exercise intervention change score (△) data for LVM (top panel) and EDV (bottom panel) plotted against one another with response to resistance (RES) and endurance (END) training on the x- and y-axis, respectively. A figure key (top left) depicts concordance and discordance for response to RES and END with percentages of responders for each quadrant reported for each variable.


This is the first study of humans to investigate sex differences in cardiac adaptation to distinct exercise training modalities (END and RES) that are commonly prescribed for individuals embarking on guideline-based exercise training programs. Our results indicate that in response to END, increased LVM and EDV changes characteristic of an eccentric pattern of hypertrophy were evident in males, but less so in females. Although both sexes increased LVM, only males increased EDV. This was a consistent finding whether data were expressed in absolute terms or as percent change from baseline. These data suggest that females exhibit blunted eccentric hypertrophy in response to END. With respect to RES training, EDV was unaltered in either males or females, and the decrease in LVM in males in absolute terms was not apparent when data were presented as a percentage change from baseline. These data for RES reinforce our previous findings (15), indicating that RES training, undertaken in previously inactive subjects at levels consistent with guideline-based exercise prescription, is not associated with concentric hypertrophy in either males or females.

Our findings complement the previous publications about sex differences in humans (8,10); however, no previous study to our knowledge has investigated sex-based cardiac adaptations to both END and RES training in the same individuals. Howden and colleagues (10) investigated the response of a relatively small number of subjects (7 males and 4 females) to 12 months of community-based END training. This study revealed that although both males and females increased LVM from 0 to 3 months, and this persisted over 12 months, males plateaued from month 9 to month 12 whereas females plateaued after month 3 of training. In addition, there were no changes in EDV within the first 3 months of training. In this context, our END training program was likely to be long enough for females to gain adequate adaptation and for differential adaptations between the sexes to emerge. The sex-dependent difference in EDV we observed is novel. Although our study cannot provide mechanistic reasons why such a difference was apparent, one possibility may relate to sex differences in the effect of training on blood volume. Previous studies suggest that expanded plasma and blood volumes contribute to higher EDV after endurance training in both men and women (22,23) and that women have lower volumes (24). A difference in training-induced blood volume expansion between men and women may therefore contribute to the results we observed, but future studies could address the role of cardiac loading on the volume differences we observed.

We observed substantial baseline sex differences in cardiac morphology. Females have smaller LVM, wall, and internal dimensions (25,26) compared with their male counterparts. In an attempt to account for sex-specific baseline cardiac dimensions, we scaled LVM to body surface area and lean body mass and additionally analyzed results as a percentage change from individual baselines. In line with earlier cross-sectional studies (25,27), sex differences persisted after scaling our LVM results, but we found that expressing changes as a percentage from baseline affected our results for LVM. For RES training, differences in males in absolute terms were no longer apparent when expressed as percentage change from baseline. For END training, the difference between males and females in EDV in absolute terms was less significant when these data were analyzed as percentage changes from baseline. Although males were more responsive to END training than females, our study nonetheless emphasizes the importance of taking baseline differences into account when considering the magnitude of sex differences.

The current study is the first to use a within-subject crossover design to characterize individual cardiac responses to distinct modes of exercise training. By presenting the response of each individual to both modes, we highlight the spread of response for each sex (Fig. 3). These data indicate that a larger proportion of males than females were responsive to END for LVM and EDV, and that a higher proportion of females responded to END than RES (P = 0.06). Our crossover design also allowed us to determine whether sex affected concordance in response between modes (Fig. 4). For LVM, positive concordance to training between modes was the same between sexes (32%–33%). Low responders to RES had a better than even chance of being a higher responder to END, especially in males, whereas low responders to END had a relatively low chance of responding to RES in either sex. This approach can provide invaluable information to guide exercise prescription; persisting with a modality that is not inducing positive adaptation compromises physiological and health benefit as well as subject compliance.

Our study was not designed to assess mechanisms responsible for sex differences in cardiac adaptation, but some previous animal studies provide insight. Konhilas and colleagues (5) reported that the proportional increase in Ca2+/calmodulin-dependent protein kinase activity was higher in females compared with males after endurance-type exercise involving voluntary cage wheel exposure, and that increased phosphorylation of glycogen synthase kinase-3β after 7 d of cage wheel exposure remained elevated in females but not males by 21 d of exercise. The authors concluded that females have increased exercise capacity and increased cardiac hypertrophic response to exercise. Foryst-Ludwig and colleagues (6) undertook cardiac and metabolic phenotyping in mice via echocardiography, small-animal PET, periexercise indirect calorimetry, and analysis of adipose tissue lipolysis and cardiac gene expression. Female mice increased cardiac hypertrophy compared with males, and this was associated with increased plasma free fatty acid levels and augmented lipolysis after training. In parallel, myocardial glucose uptake was reduced in female mice after exercise, analyzed by PET, whereas cardiac glucose uptake was unaltered after exercise in males. The authors concluded that sex differences in exercise-induced cardiac hypertrophy are associated with changes in cardiac substrate availability and utilization. Regarding CMR studies in animals, a double real-time gated study of 10 mice who underwent 12 wk of wheel running by Aufradet et al. (7) reported increased LVM, EDV, ESV, and SV, but no data were provided regarding sex differences in this study. However, it is important to bear in mind, when considering the effect of sex differences on cardiac adaptation, that factors including behavioral differences in activity, bipedal posture that affects myocardial loading, difficulties in comparing exercise intensities and modalities (particularly RES), and differences in reproductive physiology severely limit the translational relevance of animal studies.

A potential limitation of the current study was the relatively brief training period (3 months). However, the data of Howden et al. (10) indicate that substantial adaptation to training occurs in the initial 3 months. To complete a 12-month intervention of RES versus END training was not feasible in the current study, mainly because of the extended time it would take for participants to crossover and complete such a study (12 months of each intervention plus a lengthy washout period) and the inevitable confounding (i.e., diet changes, life changes, pregnancies, dropout rates, work changes, family changes, etc.) that would occur over that length of time. The magnitude of response for a large proportion of our data (95% of subjects for LVM in response to END and RES, and for EDV, 77% of subjects for END and 72% for RES) exceeded the measurement error of CMR based on the widely cited reproducibility study of Bellenger et al. (11). A further caveat is that although our training programs accorded with prescriptive guidelines (including those of the ACSM) and were therefore ecologically valid, it remains possible that different interventions may have induced greater degrees of cardiac adaptation. Nonetheless, all subjects in our study (males and females) were administered a matching stimulus, and sex differences were observed to these equivalent supervised programs. The lack of cardiac remodeling to resistance training has now been observed in a number of studies using CMR, and it has been proposed that early data suggestive of resistance training-induced adaptation (i.e., Morganroth-based “concentric hypertrophy”) may have been affected by cross-sectional comparisons between athletes that were inadequately scaled and failed to control for selection biases and/or the presence of anabolic steroid use (3).

A major strength of the current study, in contrast to previous experiments, was that the exercise interventions were center-based and individuals were trained at the university under the supervision of an Accredited Exercise Physiologist/Scientist, ensuring that individuals were completing the same program and at identical exercise intensity, which was well matched between individuals and monitored closely throughout each session. Our study was well powered (36 females for both modes and 22 males for RES and 24 for END) compared with previous articles on cardiac sex differences in response to END training that had considerably smaller sample sizes of 15 females and 16 males (9) or 4 females and 7 males (10).

The current study is the first, to our knowledge, to use CMR to assess sex-based group and individual responses to distinct exercise training modalities. Our interventions were ecologically valid, based on physical activity guidelines, and translatable to exercise prescriptions commonly used in the fitness industry. The major implication of our findings is that females may take longer to adapt to END training than males and/or may not adapt to the same extent. In addition, because eccentric hypertrophy is linked to increases in V˙O2max, our cardiac sex difference findings have implications for fitness benefits from END training. For cardiologists and clinicians, the length of time a patient has undergone END training should be taken into consideration when subjects present with eccentric remodeling, as this may have implications for a diagnosis of pathological hypertrophy rather than Athletes heart. RES training was not associated with adverse cardiac remodeling in either sex, a finding which indicates that concentric hypertrophy previously reported in athletes may relate to the much longer and more intense exposure to training, or to the possible effect of anabolic steroids (3).

D. J. G. was supported by an NHMRC Principal Research Fellowship (APP1080914) and an Exercise and Sport Science Australia (ESSA) Clinical Exercise Physiology Research Grant.

The authors thank Phil Watson and Warrick Briggs at Envision Medical Imaging for analysis support.

The authors declare that they have no conflicts of interest. The results of the present study do not constitute endorsement by ACSM. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.

D. J. G., L. H. N., C. E. M., and H. J. T. contributed to the design of the study. C. E. M. and H. J. T. conducted the study (recruitment, data collection, and implemented and supervised training interventions). C. E. M. conducted cardiac analysis with the oversight and expertise of L. G. D. and L. H. N. All authors contributed to the interpretation of results. C. E. M., D. J. G., and L. H. N. wrote the article. L. G. D. and H. J. T. reviewed and edited the article. All authors approved the final version of the manuscript.


1. Diaz-Canestro C, Pentz B, Sehgal A, et al. Sex differences in cardiorespiratory fitness are explained by blood volume and oxygen carrying capacity. Cardiovasc Res. 2021;cvab028.
2. Green DJ, Hopkins ND, Jones H, Thijssen DHJ, Eijsvogels TMH, Yeap BB. Sex differences in vascular endothelial function and health in humans: impacts of exercise. Exp Physiol. 2016;101(2):230–42.
3. Naylor LH, George K, O’Driscoll G, Green DJ. The athlete’s heart: a contemporary appraisal of the ‘Morganroth hypothesis’. Sports Med. 2008;38(1):69–90.
4. Angell P, Chester N, Green D, Somauroo J, Whyte G, George K. Anabolic steroids and cardiovascular risk. Sports Med. 2012;42(2):119–34.
5. Konhilas JP, Maass AH, Luckey SW, Stauffer BL, Olson EN, Leinwand LA. Sex modifies exercise and cardiac adaptation in mice. Am J Physiol Heart Circ Physiol. 2004;287(6):H2768–76.
6. Foryst-Ludwig A, Kreissl MC, Sprang C, et al. Sex differences in physiological cardiac hypertrophy are associated with exercise-mediated changes in energy substrate availability. Am J Physiol Heart Circ Physiol. 2011;301(1):H115–22.
7. Aufradet E, Bessaad A, Alsaid H, et al. In vivo cardiac anatomical and functional effects of wheel running in mice by magnetic resonance imaging. Exp Biol Med. 2012;237(3):263–70.
8. Diaz-Canestro C, Montero D. The impact of sex on left ventricular cardiac adaptations to endurance training: a systematic review and meta-analysis. Sports Med. 2020;50(8):1501–13.
9. Spina RJ, Ogawa T, Kohrt WM, Martin WH 3rd, Holloszy JO, Ehsani AA. Differences in cardiovascular adaptations to endurance exercise training between older men and women. J Appl Physiol (1985). 1993;75(2):849–55.
10. Howden EJ, Perhonen M, Peshock RM, et al. Females have a blunted cardiovascular response to one year of intensive supervised endurance training. J Appl Physiol. 2015;119(1):37–46.
11. Bellenger NG, Davies LC, Francis JM, Coats AJ, Pennell DJ. Reduction in sample size for studies of remodeling in heart failure by the use of cardiovascular magnetic resonance. J Cardiovasc Magn Reson. 2000;2(4):271–8.
12. Spence AL, Naylor LH, Carter HH, et al. Does echocardiography accurately reflect CMR-determined changes in left ventricular parameters following exercise training? A prospective longitudinal study. J Appl Physiol. 2013;114(8):1052–7.
13. Strohm O, Schulz-Menger J, Pilz B, Osterziel KJ, Dietz R, Friedrich MG. Measurement of left ventricular dimensions and function in patients with dilated cardiomyopathy. J Magn Reson Imaging. 2001;13(3):367–71.
14. Morganroth J, Maron B, Henry W, Epstein S. Comparative left ventricular dimensions in trained athletes. Ann Intern Med. 1975;82:521–4.
15. Spence AL, Naylor LH, Carter HH, et al. A prospective randomised longitudinal MRI study of left ventricular adaptation to endurance and resistance exercise training in humans. J Physiol. 2011;589(22):5443–52.
16. Marsh CE, Thomas HJ, Naylor LH, Scurrah KJ, Green DJ. Fitness and strength responses to distinct exercise modes in twins: Studies of Twin Responses to Understand Exercise as a Therapy (STRUETH) study. J Physiol. 2020;598(18):3845–58.
17. Bouchard C, Blair SN, Church TS, et al. Adverse metabolic response to regular exercise: is it a rare or common occurrence? PLoS One. 2012;7(5):e37887.
18. Marsh CE, Thomas HJ, Naylor LH, Scurrah KJ, Green DJ. Exploring human trainability: design and rationale of Studies of Twin Responses to Understand Exercise as a Therapy (STRUETH) study. Contemp Clin Trials Commun. 2020;19:100584.
19. Swoboda PP, Garg P, Levelt E, et al. Regression of left ventricular mass in athletes undergoing complete detraining is mediated by decrease in intracellular but not extracellular compartments. Circ Cardiovasc Imaging. 2019;12(9):e009417.
20. Spence AL, Carter HH, Murray CP, et al. Magnetic resonance imaging-derived right ventricular adaptations to endurance versus resistance training. Med Sci Sports Exerc. 2013;45(3):534–41.
21. Thomas HJ, Marsh CE, Maslen BA, Scurrah KJ, Naylor LH, Green DJ. Studies of Twin Responses to Understand Exercise Therapy (STRUETH): body composition. Med Sci Sports Exerc. 2021;53(1):58–67.
22. Carrick-Ranson G, Sloane NM, Howden EJ, et al. The effect of lifelong endurance exercise on cardiovascular structure and exercise function in women. J Physiol. 2020;598(13):2589–605.
23. Hagberg JM, Goldberg AP, Lakatta L, et al. Expanded blood volumes contribute to the increased cardiovascular performance of endurance-trained older men. J Appl Physiol. 1998;85(2):484–9.
24. Best S, Okada Y, Galbreath MM, et al. The effect of gender and age on hemodynamics, blood volume and cardiac size in healthy humans. FASEB J. 2012;26:lb635.
25. Celentano A, Palmieri V, Arezzi E, et al. Gender differences in left ventricular chamber and midwall systolic function in normotensive and hypertensive adults. J Hypertens. 2003;21(7):1415–23.
26. Lang RM, Bierig M, Devereux RB, et al. Recommendations for chamber quantification. Eur J Echocardiogr. 2006;7(2):79–108.
27. Sandstede J, Lipke C, Beer M, et al. Age- and gender-specific differences in left and right ventricular cardiac function and mass determined by cine magnetic resonance imaging. Eur Radiol. 2000;10(3):438–42.


Supplemental Digital Content

Copyright © 2021 by the American College of Sports Medicine