Females have a blunted cardiovascular response to one year of intensive supervised endurance training

Abstract

Cross-sectional studies in athletes suggest that endurance training augments cardiovascular structure and function with apparently different phenotypes in athletic males and females. It is unclear whether the longitudinal response to endurance training leads to similar cardiovascular adaptations between sexes. We sought to determine whether males and females demonstrate similar cardiovascular adaptations to 1 yr of endurance training, matched for training volume and intensity. Twelve previously sedentary males (26 ± 7, n = 7) and females (31 ± 6, n = 5) completed 1 yr of progressive endurance training. All participants underwent a battery of tests every 3 mo to determine maximal oxygen uptake (V̇o_2max) and left ventricle (LV) function and morphology (cardiac magnetic resonance imaging). Pulmonary artery catheterization was performed before and after 1 yr of training, and pressure-volume and Starling curves were constructed during decreases (lower-body negative pressure) and increases (saline infusion) in cardiac volume. Males progressively increased V̇o_2max, LV mass, and mean wall thickness, before reaching a plateau from month 9 to 12 of training. In contrast, despite exactly the same training, the response in females was markedly blunted, with V̇o_2max, LV mass, and mean wall thickness plateauing after only 3 mo of training. The response of LV end-diastolic volume was not influenced by sex (males +20% and females +18%). After training Starling curves were shifted upward and left, but the effect was greatest in males (interaction P = 0.06). We demonstrate for the first time clear sex differences in response to 1 yr of matched endurance training, such that the development of ventricular hypertrophy and increase in V̇o_2max in females is markedly blunted compared with males.

regular exercise training leads to changes in the structure and function of cardiac and skeletal muscle, which is fundamentally influenced by the type of exercise performed (endurance vs. strength training) (36). Physiological skeletal muscle hypertrophy is associated with strength training, whereas endurance training is a more potent stimulus to induce physiological cardiac hypertrophy. Clear sex differences are observed between strength-trained athletes, such that males demonstrate a greater magnitude of skeletal muscle mass compared with females (1), primarily due to the profound anabolic effect of testosterone on protein synthesis within the skeletal muscle (9).

It is less clear whether the cardiovascular response to endurance training is also influenced by sex. Cross-sectional studies of endurance athletes suggest that females have a lower maximal oxygen uptake (V̇o_2max) and cardiac output (Qc) and smaller left ventricular (LV) dimensions and wall thickness compared with similar male athletes (12, 34, 40). Confounding the interpretation of sex differences is the effect that body size has on cardiovascular anatomic and physiological variables (11). Typically studies have scaled cardiovascular variables to body size or lean body mass, which eliminates some, but not all, of the differences between sexes.

Cross-sectional studies are also limited by the different sports played by male and female athletes and different training performed, which may alter the magnitude of the effect on the cardiovascular adaptations observed. Longitudinal studies comparing the cardiovascular response in males and females with a comparable endurance training program is lacking.

We have recently demonstrated in previously sedentary young subjects that 1 yr of intensive endurance training increases cardiac mass to levels similar to elite endurance athletes, results in favorable cardiac remodeling, and improves LV compliance (3). In this secondary analysis of the same study, we sought to determine whether males and females demonstrate similar cardiovascular adaptations to 1 yr of endurance training, matched for training volume and intensity. We hypothesized that the response to training would be similar in males and females, evidenced by a similar change in V̇o_2max, cardiac morphology, and function in response to training. To test this hypothesis, we prescribed a progressive training plan, identical between males and females, based on a strategy derived from optimal training by competitive athletes; this study design allowed us to collect time-dependent measures of metabolic and LV structure parameters in response to 1 yr of training.

METHODS

Subjects

This study is a secondary analysis of 12 healthy previously sedentary males (n = 7) and females (n = 5), who completed 1 yr of intensive endurance training. The primary findings from this study have been published recently (3). One subject completed all of the initial testing but became pregnant in the second quarter of training and was excluded from the study. Participants were not eligible for the study if they had engaged in any regular endurance training defined as >30 min/day more than three times per week. All participants were carefully screened and underwent a physical examination, electrocardiogram (ECG), and echocardiogram. None of the subjects smoked, used recreational drugs, or had significant chronic medical problems. All female subjects were eumenorrheic when they commenced training and remained so for the duration of the study.

Ethical Approval

All subjects signed an informed consent, which was approved by the Institutional Review Boards of the University of Texas Southwestern Medical Center at Dallas and Texas Health Resources Presbyterian Hospital of Dallas.

Exercise Training

Endurance training was designed to enable subjects to complete a marathon at the end of the 1-yr period. All subjects performed an incremental treadmill test for determination of V̇o_2max at baseline and every 3 mo to document exercise performance and for optimal training prescription. A detailed description of the exercise intervention has been previously described (3, 18). Training zones for optimal training prescription were individualized by heart rate (HR) and lactate response to incremental treadmill exercise. “Maximal steady state” was first estimated from the ventilatory and lactate threshold according to standard criteria (18). The HR at maximal steady state and maximum HR were then used to calculate training zones for each subject; five training zones were determined for individualized training prescription as follows: zone 1, recovery, zone 2, base pace, zone 3, maximal steady state or “threshold,” zone 4, race pace/critical power, and zone 5, intervals. During the early phase of the training program, the subjects trained three to four times per week for 30-45 min/session at base pace by brisk walking, slow jogging, swimming, or cycling. As subjects became fitter, the duration of the base training sessions was prolonged, including the addition of one long run per week, which was performed at the lower end of base pace HR range. In addition, during the second and third quarters of the training program, sessions of increased intensity (maximal steady-state and interval sessions) were added first one time and then two times and occasionally three times per week. Interval sessions were followed the next day by a recovery session to maximize performance gains. By the end of the year-long training program, subjects were exercising for 7–9 h/wk, including long runs of up to 3 h, plus regular interval sessions on the track and races. The purpose of this template was to maximize training efficiency and to provide a periodization of the training program. This strategy of varying intensity and duration of training session within any given week (microcycles) applying periods of increasing stress followed by recovery from month to month (mesocycle) with an ultimate goal of a specific competition (macrocycle) is a classic training routine used by competitive athletes and is widely considered the optimal approach to training (37). Although the majority of subjects chose to complete a marathon at completion of the intervention, one female subject completed a 100-mile endurance cycling race, and one male subject completed an Olympic distance triathlon. We followed the principles of training specificity; thus, the subjects completing the marathon performed predominantly running and the cyclist cycled while the triathlete performed a combination of swimming, cycling, and running.

HR was recorded during each training session using a HR monitor provided to each participant. This approach was an important aspect of the study and gave us the ability to match and quantify training load throughout the intervention. To quantify the training stimulus, we calculated the training impulse (“TRIMP”) (5). This method multiplies the duration of a training session by the average HR achieved during that session, weighted for exercise intensity as a function of HR reserve. With the use of this method, exercise sessions of longer duration and/or greater intensity such as interval workouts are assigned relatively higher TRIMP values (higher HR and higher weighting factor) than sessions of shorter duration and/or lower intensity. The average TRIMP for the male and female subjects is presented in Fig. 1. Note the oscillations in TRIMP score throughout the training period, i.e., micro- and macrocycles of training in Fig. 1 and the remarkably similar training load between sexes.

Fig. 1.

Average training impulse (TRIMP) scores per month in males and females. Individual sex values are presented as means ± SD.

Study Protocol

All measurements were performed before training (baseline) and repeated after 3, 6, 9, and 12 mo of training. Pressure/volume studies were completed pre- and posttraining.

Exercise Testing

A modified Astrand-Saltin incremental treadmill protocol (individualized treadmill speed, with changes in grade of 2% every 2 min) was used to determine maximal exercise capacity. Measures of ventilatory gas exchange were made by use of the Douglas bag technique. Gas fractions were analyzed by mass spectrometry (Marquette MGA 1100), and ventilatory volume was measured by a dry-gas meter (Collins). V̇o_2max was defined as the highest oxygen uptake measured from ≥40 s Douglas bag. In nearly all cases, a plateau in V̇o₂ was observed with increasing work rate, confirming the achievement of V̇o_2max. Qc was measured with the acetylene rebreathing method, which has been previously validated in this laboratory (19). HR was monitored continuously via ECG (Polar). Systemic arteriovenous oxygen difference (a-vO_{2 diff}) was calculated from the Fick equation (a-vO_{2 diff} = V̇o₂/Qc) while stroke volume (SV) was calculated as Qc/HR. SV and Qc were scaled relative to baseline body surface area [stroke index (SI) and cardiac index (QI)] by clinical convention to reduce the confounding effect of body size and composition (11).

Body Composition

Body density and composition were determined by underwater weighing with correction for residual lung volume (41). Each participant performed at least three adequate measurements defined as a definite plateau in underwater weight, and the mean value was calculated.

Plasma, Blood Volume, and Hematocrit

At baseline and at each quarter testing session, all subjects underwent measurement of plasma volume using Evans Blue dye indicator dilution technique. The methods from this technique have been published (14). Briefly, individuals rested in the supine position for at least 30 min after which a known quantity of Evans Blue dye was injected through a peripheral intravenous catheter, and venous blood was drawn at 10, 20, and 30 min after injection for the measurement of absorbance at 620 and 740 nm via spectrophotometry (DU 600; Beckman). Hematocrit was measured via microcapillary centrifuge, and blood volume was estimated by dividing plasma volume by 1 − hematocrit using appropriate corrections for trapped plasma and peripheral sampling. To reduce the confounding effect of body size and composition on blood volume (11), absolute values were scaled relative to total body mass (ml/kg) and fat-free mass (FFM; ml/kg FFM).

Cardiac Magnetic Resonance Imaging

Cardiac morphometric parameters were assessed by a cardiac magnetic resonance imaging (cMRI) 1.5-tesla Phillips NT MRI Scanner (Phillips Medical Systems, Best, The Netherlands). Short-axis, gradient echo, and cine magnetic resonance imaging (MRI) sequences with a temporal resolution of 39 ms were obtained to calculate ventricular volumes as previously described (17). Ventricular mass was computed as the difference between epicardial and endocardial areas multiplied by the density of heart muscle, 1.05 g/ml (21). The Simpson's rule technique was used to measure LV mass, which has been demonstrated to be accurate and reproducible in our laboratory (21) and by others (15).

For LV volumes, the endocardial border of each slice was identified manually at end diastole and end systole, and volumes were counted by summation (31). Mean wall thickness (MWT) for the entire LV, including the papillary muscle, was calculated as previously described (30). For each short-axis slice, the epicardial area (LV chamber plus myocardial wall) and endocardial area (chamber area) were determined using software on the imaging device. The “average” radius for each area was calculated by approximating the cross section as a circle and using the equation for the area of a circle (area = πr²). To reduce the confounding effects of body size and lean mass on LV and right ventricle (RV) mass, measures were scaled to FFM (7).

Cardiac Catheterization and Experimental Protocol

Cardiac catheterization was performed at baseline and after the completion of the training period. Participants were studied in the resting supine position. A 6-Fr balloon-tipped fluid-filled catheter (Swan-Ganz; Baxter) was placed using fluoroscopic guidance through an antecubital vein in the pulmonary artery. All intracardiac pressures were referenced to atmospheric pressure with the pressure transducer (Transpac IV; Abbott) zero reading set at 5 cm below the sternal angle. The wedge position of the catheter tip was confirmed using fluoroscopy, as well as the presence of an appropriate pulmonary capillary wedge pressure (PCWP) waveform. The mean PCWP was determined visually at end expiration and was used as an estimation of LV end-diastolic pressure.

Qc was measured, and SV was calculated from Qc/HR measured during rebreathing. Left ventricular end-diastolic volume (LVEDV) was measured with two-dimensional echocardiography using standard views and formulas as described by the American Society of Echocardiography (27). Images were obtained with an annular phased-array transducer using a frequency of 2.5–3.5 MHZ (Interspec Apogee CX) and stored on VCR tapes for off-line analysis by a skilled technician. For calculation of LVEDV for each subject, either the modified Simpson's rule method, the area length method, or the bullet model (cylinder hemiellipsoid) was chosen on the basis of which views provided the most optimal endocardial definition (39). The same formula was used for each individual subject throughout the study.

Testing protocol.

Resting supine measures (baseline 1) were collected after ensuring hemodynamic stability (∼20 min of quiet rest), and then cardiac filling was decreased using lower-body negative pressure (LBNP) as previously described (26). Measurements of PCWP, Qc (and therefore SV), LVEDV, HR, and blood pressure were made after 5 min each of −15 and −30 mmHg LBNP. The LBNP was then released. After repeat baseline measurements to confirm a return to hemodynamic steady state (usually 20–30 min), cardiac filling was increased by rapid (200 ml/min) infusion of warm (37°C), isotonic saline. Measurements were repeated after 15 and 30 ml/kg had been infused.

Data were used to construct Frank-Starling (SV/PCWP) and pressure-volume (PCWP/LVEDV) curves. For the purpose of the present study, we characterized two explicitly different but related mechanical properties of the heart during diastole: 1) static stiffness or overall chamber stiffness referred to as the stiffness constant, S, of the logarithmic equation describing the pressure-volume curve (see below); and 2) distensibility, which is used to mean the absolute LVEDV at a given distending pressure, independent of dP/dV, or S.

To characterize the LV pressure-volume relation, we modeled the data in the present experiment according to the equation described by Nikolic et al. (28):

$$\text{P}=-S\hspace{0.17em}\ln \left[\left({\text{V}}_{\text{m}}-\text{V}\right)/({\text{V}}_{\text{m}}-{\text{V}}_o)\right]$$

where P is PCWP, V is LVEDV, V₀ is equilibrium volume or the volume at which P = 0, V_m is the maximal volume obtained by this chamber, and S is a stiffness constant that describes the shape of the curve. In addition, pressure-volume curves were also calculated using the difference between PCWP and right atrial (RA) pressure (PCWP − RA) as an index of transmural filling pressure (8) to assess the contribution of pericardial constraint.

Statistics

Continuous data are expressed as means ± SD except for in Figs. 2–7 in which SE is used. Baseline sex differences were compared with Student's t-test. Continuous variables measured over the 12-mo study duration were analyzed longitudinally using linear mixed-effects model repeated-measures analysis. The repeated-measures model had five repeated measurements (time points at baseline and 3, 6, 9, and 12 mo), and the study participant was modeled as a random effect. The covariance structure was selected based on Akaike's Information Criteria and model parsimony. Our analysis was unadjusted, and the analysis was by available data (last observation was not varied forward). To test our primary hypothesis whether previously sedentary healthy young males and females differ in response to 1 yr of endurance training, the sex × time interaction was used, with a P value of <0.10 declared significant. Similar regression models were used to assess pressure-volume curves with the addition of a fixed effect to test for the difference and response interactions between the baseline and training curves and sex. To express the dose-response relationship between the exercise training stimulus and changes in LV mass, the relationship between the quarterly training impulse (monthly TRIMP) and cardiac adaptations at baseline and 3, 6, 9, and 12 mo was estimated with quadratic polynomial regression models. Statistical analysis was performed using commercially available software (IBM SPSS, SAS version 9.3; SAS Institute, Cary, NC). Statistical significance was declared at a P value of <0.05.

Fig. 2.

A and B: effect of 1 yr of endurance training on maximal oxygen uptake (V̇o_2max) indexed to baseline body mass in males and females (left), significant sex × time interaction P = 0.084 and changes in left ventricle (LV) mass measured by magnetic resonance imaging (MRI) scaled to baseline fat-free mass, significant sex × time interaction P = 0.031 (right). Both measured every 3 mo during the training program. FFM, fat-free mass. Post hoc comparison with baseline (*), with month 3 (†), and with month 6 (‡) for P < 0.05 from linear mixed model.

Fig. 7.

A and B: group mean Frank-Starling curves representing pulmonary capillary wedge pressure (PCWP) as an index of LV end-diastolic pressure vs. stroke volume index (SI), over a range of LV filling produced by lower-body negative pressure (two lowest levels of PCWP), quiet baseline (two middle values of PCWP), and rapid saline infusion (two highest values of PCWP) as described in text. Each data point represents the mean ± SE of males or females, pre vs. post (sex × time P = 0.06).

RESULTS

At 1 yr, all subjects completed either a marathon (n = 10, 4 females and 6 males), Olympic distance triathlon (n = 1 male), or 100-mile endurance cycling race (n = 1 female). One male subject had a metal implant and was unable to undergo MRI, but completed all other testing. The baseline characteristics and effect of training on body size, composition, and blood volume are summarized in Table 1. Comparison between the sexes showed that females had smaller body size, less FFM and higher percent body fat (P ≤ 0.031), and lower hematocrit levels (P = 0.009) before training than males. After training there was no difference between sexes for hematocrit, body fat, and indexed blood volume. There was a significant sex × time interaction (P = 0.081) for percent body fat. Females demonstrated a reduction in body fat after 6 mo of training (22.8 ± 4.5 vs. 18.1 ± 2.1%, P = 0.048); a further 6 mo of training sustained the reduction in body fat (P = 0.006, month 12 compared with baseline, 16.2 ± 4.2%). In males, percent body fat did not decrease significantly until month 12 of training (19.3 ± 6.1 vs. 15.0 ± 3.4%, P = 0.026). The effect of the intervention on lean body mass was similar between sexes (sex × time P = 0.38), with training resulting in a similar significant increase (main effect of time P = 0.019).

Table 1.

Baseline characteristics, body size and composition, and total blood volume after endurance training

	Males	Females	P Value (sex difference at baseline)
Age, yr
Baseline	26 ± 7	31 ± 6	0.21
Height, cm
Baseline	177 ± 2	169 ± 6	0.031
Weight, kg
Baseline	78 ± 6	60 ± 3	<0.0001
12 Months	77 ± 6	60 ± 4
BSA, m2
Baseline	1.96 ± 0.07	1.67 ± 0.07	0.0006
12 Months	1.93 ± 0.07	1.69 ± 0.08
FFM, kg
Baseline	62 ± 5	47 ± 5	0.0007
12 Months	65 ± 4	50 ± 3**
Body fat, %
Baseline	19 ± 6	23 ± 4	0.270
12 Months	15 ± 3*	16 ± 4*
Hematocrit, %
Baseline	44 ± 3	39 ± 3	0.0089
12 Months	42 ± 2**	41 ± 4
TBV, ml/kg
Baseline	66 ± 9	63 ± 4	0.493
12 Months	67 ± 12	67 ± 4*
TBV, ml/kg FFM
Baseline	82 ± 7	82 ± 8	0.947
12 Months	79 ± 15	80 ± 5
PV, ml/kg
Baseline	40 ± 5	42 ± 2	0.513
12 Months	42 ± 8	43 ± 3

Values are means ± SD.

BSA, body surface area; FFM, fat-free mass; TBV, total blood volume; PV, plasma volume.

^*P < 0.05 and

^**P < 0.01 vs. baseline.

Training Impulse

The amount of training performed was virtually identical between sexes (males total TRIMP = 19,919 ± 8,687 vs. females total TRIMP = 19,989 ± 2,695, P = 0.98; Fig. 1).

Maximal Exercise Response

The effect of the 1 yr of training on V̇o_2max indexed to body mass is shown in Fig. 2A. Endurance training increased V̇o_2max by 22% in males and 15% in females. There was a significant sex × time interaction for absolute (P = 0.018; our primary outcome variable) and V̇o_2max scaled to body mass (P = 0.084) and FFM (P = 0.094), which resulted in the males demonstrating a progressive increase in V̇o_2max over the first 9 mo of endurance training and then plateauing from months 9 to 12 (Fig. 2A and Table 2). The response of V̇o_2max in the females was blunted compared with the males, with the majority of the increase in V̇o_2max occurring during the initial 3 mo of training (13% increase from baseline) plateauing thereafter despite the further increases in training load. In both sexes, training decreased maximal HR and increased QI, which resulted in a significant increase in SI after completion of 1 yr of training (Table 2). There was a significant sex × time interaction for a-vO_{2 diff} (P = 0.026; Table 2), where males increased a-vO₂ after 9 mo of training while females increased after 6 mo of training before decreasing back to baseline levels.

Table 2.

Effect of 1 yr of endurance training on maximal exercise response

		Month
	Baseline	3	6	9	12	P Value (main effect of time)	P Value (sex × time)
V̇o2, l/min
Males	3.36 ± 0.44	3.75 ± 0.43*	3.85 ± 0.35*	4.07 ± 0.43*†‡	4.09 ± 0.47*†‡	<0.001	0.013
Females	2.19 ± 0.14	2.48 ± 0.16*	2.57 ± 0.18*	2.48 ± 0.21*†	2.51 ± 0.12*†
V̇o2, ml/kg FFM
Males	53.6 ± 5.0	59.9 ± 4.9*	61.4 ± 3.6*	64.9 ± 4.9*†‡	65.3 ± 5.6*†	<0.001	0.094
Females	47.0 ± 3.5	53.3 ± 5.8*	55.2 ± 5.4*	53.1 ± 4.5*	54.2 ± 7.1*
Heart rate, beats/min
Males	200 ± 12	189 ± 7	189 ± 10	190 ± 12	188 ± 12	<0.001	0.56
Females	192 ± 7	185 ± 9	185 ± 6	183 ± 6	186 ± 8
Cardiac index, l·m−2·min−1
Males	12.2 ± 1.3	13.6 ± 1.3	12.5 ± 1.3	12.0 ± 1.9	12.9 ± 1.6	0.029	0.32
Females	8.9 ± 1.1	9.9 ± 1.5	8.9 ± 0.8	10.3 ± 2.1	10.4 ± 1.4
Stroke index, ml/m2
Males	60.8 ± 5.2	72.5 ± 6.8	66.6 ± 9.3	63.7 ± 12.6	69.1 ± 12.0	0.002	0.37
Females	46.2 ± 5.8	53.9 ± 9.8	48.1 ± 3.9	56.3 ± 10.9	55.7 ± 5.9
a-vO2 diff, ml/100 ml
Males	14.2 ± 1.6	14.2 ± 1.4	15.9 ± 2.0	17.6 ± 2.0*†	16.4 ± 1.8	0.021	0.009
Females	14.8 ± 1.5	15.1 ± 2.7	17.2 ± 1.3*	14.6 ± 2.4‡	14.6 ± 1.8‡
Lactate, mmol/l
Males	10.4 ± 1.6	10.6 ± 3.2	10.6 ± 1.4	9.3 ± 1.0	9.2 ± 1.7	0.55	0.59
Females	9.9 ± 3.4	9.0 ± 1.9	10.6 ± 2.9	10.1 ± 1.7	9.2 ± 2.7

Values are means ± SD. a-vO_{2 diff}, systemic arteriovenous oxygen difference. The P value within the table represents the main effect of time. Post hoc comparisons are indicated from the linear mixed-effects model where there was a statistically significant, P < 0.10 sex × time interaction; P < 0.05 vs. baseline (*), 3 mo (†), and 6 mo (‡).

Effect of Sex on Left Ventricular Adaptations to Training

The adaptations in LV structure to the intervention are presented in Figs. 2–5 and Table 3. Similar to the response of V̇o_2max, we detected a significant sex × time interaction for LV mass indexed to FFM (P = 0.031). The pattern in LV mass relative to FFM remodeling is shown in Fig. 2B. In males, LV mass increased by 11 ± 2% (P < 0.05) during the first 3 mo of training and 9 ± 2 (P < 0.01) between 3 and 6 mo, with the total increase during the first 6 mo being 20 ± 2% (P < 0.01). In females, the greatest increase in LV mass indexed to FFM (13 ± 2%, P < 0.01) occurred during the first 3 mo of training. In contrast to the males, there was no further statistically significant increase in LV mass indexed to FFM in the females (Fig. 2B). When the mean quarterly values for LV mass were compared with the average TRIMP values obtained for each quarter in males and females, we observed a strong positive relationship in the males such that as training impulse increased LV mass was markedly augmented, whereas in the females the effect of increased training impulse was markedly attenuated with a small gradual increase in mass with increasing training stimulus (Fig. 3, A and B).

Fig. 5.

A and B: group mean pressure-volume curves for male (A) and female (B) subjects with data points derived from baseline, lower-body negative pressure (LBNP), and rapid saline infusion, similar to Fig. 6. Each data point represents the mean ± SE of males or females, pre vs. post (sex × time P = 0.18).

Table 3.

Effect of 1 yr of endurance training on left and right ventricle morphology and function measured by cardiac MRI

	Baseline	Month 3	Month 6	Month 9	Month 12	P Value (main effect of time)	P Value (sex × time)
MWT, cm
Males	1.00 ± 0.13	1.19 ± 0.13*	1.27 ± 0.11*†	1.23 ± 0.12*	1.23 ± 0.06*	<0.001	<0.001
Females	1.00 ± 0.05	1.11 ± 0.05*	1.12 ± 0.02*	1.05 ± 0.05†‡	1.07 ± 0.07*
LV mass, g
Males	192 ± 31	212 ± 27*	230 ± 24*†	234 ± 23*†	238 ± 26*†	<0.001	0.001
Females	140 ± 15	158 ± 13*	160 ± 12*	157 ± 13*‡	164 ± 12*
LV EDV, ml
Males	132 ± 10	136 ± 11	143 ± 11	151 ± 14	158 ± 13	<0.001	0.39
Females	98 ± 10	99 ± 9	104 ± 10	118 ± 14	116 ± 15
LV ESV, ml
Males	43 ± 8	40 ± 6	41 ± 7	44 ± 8	44 ± 8	<0.001	0.79
Females	30 ± 6	28 ± 4	31 ± 3	32 ± 5	34 ± 6
LV SV, ml
Males	89 ± 3	96 ± 5*	102 ± 5*†	107 ± 7*†‡	113 ± 7*†‡§	<0.001	0.035
Females	68 ± 5	71 ± 6	74 ± 6*	85 ± 10*†‡	81 ± 9*†‡
LV ejection fraction, %
Males	68 ± 4	71 ± 3	71 ± 3	71 ± 3	72 ± 3	0.001	0.28
Females	69 ± 4	72 ± 2	71 ± 1	72 ± 2	70 ± 2
LV mass/volume
Males	1.45 ± 0.19	1.56 ± 0.16*	1.61 ± 0.13*	1.55 ± 0.11*	1.51 ± 0.08‡	<0.001	0.008
Females	1.42 ± 0.04	1.60 ± 0.03*	1.54 ± 0.14*	1.34 ± 0.11†	1.34 ± 0.11†‡
RV EDV, ml
Males	156 ± 9	180 ± 8	188 ± 11	193 ± 10	198 ± 17	<0.001	0.20
Females	112 ± 13	126 ± 9	128 ± 10	146 ± 16	144 ± 14
RV ESV, ml
Males	70 ± 6	82 ± 5	86 ± 6	86 ± 6	88 ± 12	<0.001	0.26
Females	46 ± 7	53 ± 5	54 ± 5	63 ± 11	61 ± 9
RV SV, ml
Males	87 ± 4	98 ± 5	103 ± 5	107 ± 7	110 ± 7	<0.001	0.31
Females	66 ± 7	73 ± 4	74 ± 7	85 ± 9	82 ± 7
RV mass, g
Males	69 ± 7	78 ± 7*	83 ± 7*	86 ± 8*†	91 ± 9*†‡	<0.001	0.066
Females	56 ± 6	62 ± 5	61 ± 5	63 ± 14	69 ± 11*‡
RV ejection fraction, %
Males	56 ± 2	55 ± 1	55 ± 1	55 ± 2	55 ± 2	0.70	0.82
Females	59 ± 2	58 ± 2	58 ± 2	58 ± 2	58 ± 2
RV mass/vol
Males	0.44 ± 0.05	0.44 ± 0.04	0.45 ± 0.04	0.45 ± 0.02	0.46 ± 0.05	0.037	0.019
Females	0.51 ± 0.03	0.48 ± 0.03	0.46 ± 0.05*	0.42 ± 0.06*†‡	0.49 ± 0.06§

Values are means ± SD; n = 6 male and 5 female subjects for all parameters. All variables were measured by cardiac magnetic resonance imaging (MRI).

MWT, mean wall thickness; RV, right ventricle; LV, left ventricle; EDV, end-diastolic volume; ESV, end-systolic volume; SV, stoke volume.

The P value within the table represents the main effect of time. Post hoc comparisons are indicated from the linear mixed-effects model where there was a statistically significant, P < 0.10 sex × time interaction; P < 0.05 vs. baseline (*), 3 mo (†), 6 mo (‡), and 9 mo (§).

Fig. 3.

A and B: quadratic regression analysis between average quarterly TRIMP values as measure of training stimulus and LV mass. Light gray lines, individual curves; solid black line, curves for males and females.

MWT was increased by training and paralleled the increase in LV mass in both sexes (Table 3). In males, MWT increased by 27 ± 3% during the first 6 mo of training. In females the increase in MWT was 11 ± 2% during the first 3 mo.

The effect of endurance training on LVEDVI was similar between sexes (sex × time P = 0.34; Fig. 4A). The overall increase in LVEDVI was 20% for males and 18% for females.

Fig. 4.

A and B: effect of 1 yr of endurance training on left ventricle end-diastolic volume (LVEDV, A) and right ventricle end-diastolic volume (RVEDV, B) scaled to baseline total body surface area, measured by MRI every 3 mo during the 1-yr training program. For left ventricular end-diastolic volume index (LVEDVI) and right ventricular end-diastolic volume index (RVEDVI), males and females responded in a similar manner to the training (LVEDVI sex × time P = 0.48 and RVEDVI sex × time P = 0.22).

The pattern of concentric and eccentric modelling was overall similar between males and females; however, the time course of change was different (sex × time P = 0.008, Table 3). Both sexes demonstrated an initial increase in mass and wall thickness, consistent with concentric remodeling. The LV mass-to-volume ratio returned to near baseline levels in the males at completion of the study, whereas in the females the mass-to-volume ratio decreased to below baseline values, suggesting even greater eccentric remodeling.

Effect of Sex on Right Ventricular Adaptations to Training

The RV volumetric adaptations that occurred during the 1 yr of training were similar between sexes (Fig. 4B and Table 3). There was a sex × time (P = 0.066) interaction for RV mass (Table 3), demonstrating again that males have a more pronounced hypertrophic response to training compared with females. There was a significant sex × time interaction for RV mass-to-volume ratio (Table 3). Interestingly, males maintained a similar mass-to-volume ratio throughout the intervention. However, the ratio decreased in females as the training progressed.

Pressure-Volume Curves

The LV pressure-volumes curves constructed from group mean data in males and females are shown in Fig. 5, A and B, respectively. Both V_m and V₀ increased in the males and females consistent with significant physiological remodeling. The response to endurance training for the pressure-volume curves was similar between sexes (sex × time P = 0.18), with both males (P = 0.039) and females (P < 0.001) demonstrating a significant rightward shift. LV pressure-volume curves (Fig. 6, A and B) constructed using transmural pressure demonstrated a significant interaction (sex × time P = 0.07), with females demonstrating a significant shift in the pressure-volume curve posttraining (P = 0.02), but not males (P = 0.75).

Fig. 6.

A and B: group mean transmural pressure (TMP)-volume curves for male (A) and female (B) subjects with data points derived from baseline, LBNP, and rapid saline infusion. Each data point represents the mean ± SE of males or females, pre vs. post (sex × time P = 0.07).

Frank-Starling Curves

Frank-Starling curves are shown in Fig. 7, A and B, for males and females, respectively. There was a significant sex × time interaction (P = 0.06). The training effect in females was ∼7.1 ml less than in the males.

DISCUSSION

The main new findings from the present study were clear differences in the time course and magnitude of cardiovascular adaptation between previously sedentary males and females in response to identical endurance training loads. Males had more pronounced LV hypertrophy throughout the 1 yr of training, whereas females reached the maximal LV mass after only 3 mo of training. The LV hypertrophic response closely paralleled the increase in V̇o_2max in both sexes, which was blunted in females. We found that ventricular remodeling and increased compliance were similar between sexes, but endurance training resulted in greater augmentation of the Frank-Starling mechanism in the males compared with females. The greater effect of endurance training on the Frank-Starling mechanism (relationship between LV filling pressures and SV) in the males means that males had a greater diastolic reserve after 1 yr of training. This adaptation to training is extremely beneficial for performing exercise, allowing individuals to generate a higher SV and cardiac output in response to an increase in filling pressure. Indeed, an enhanced Frank-Starling mechanism has been shown to be a hallmark characteristic of endurance athletes (26).

Effect of One Year of Endurance Training on Exercise Performance and Cardiac Morphology

This is the first study to demonstrate that healthy previously sedentary males have a markedly greater increase in LV mass and V̇o_2max compared with the response in females to 1 yr of identical endurance training. To that end, we recorded each heart beat during exercise to carefully quantify the amount of training performed using the TRIMP method (6). As can be seen from Fig. 1, males and females received a nearly identical training stimulus. However, there was a divergent response in the observed adaptations to training. Surprisingly V̇o_2max plateaued after only 3 mo of training in females, but continued to increase progressively in males until month 9, resulting in an ∼7% greater increase in V̇o_2max at month 12. Indeed, as training intensity increased in the males, we observed a progressive increase in LV mass that was not evident in the females (Fig. 2B). To our knowledge no endurance training studies have compared sex differences in healthy young subjects beyond 3 mo of training. Therefore, until now, important differences in the cardiovascular response to training may have been missed. Indeed 90 days of intense endurance training in female and male rowers resulted in no significant differences between sexes for change in any cardiovascular variables (4). Thus it seems likely that cardiovascular sex differences have been underestimated due to the duration of prior training studies.

Consistent with the sex difference observed in the LV, the RV also increased in mass in response to training. The response in males was again more prominent than the females, who did not demonstrate a significant increase in RV mass until completion of the study. Few prospective studies have assessed the degree of RV hypertrophy with endurance training, due to the difficulty in reliably assessing morphology using echocardiography. Recent evidence from cross-sectional studies with cMRI suggests that the mass of RV is increased to match the increased work of the LV and maintain a balanced ratio to enhance function in male and female athletes (32, 33). Transient increases in pulmonary artery pressure during exercise likely contribute to the enlargement of the RV. Endurance athletes have greater RV enlargement and wall thickening, which is thought to be due to the disproportionate increase in hemodynamic load placed on the RV during endurance exercise (25). Given these prior findings, our results raise the possibility that the male participants may have been under a greater hemodynamic stress during training, which is reflected by the greater increase in RV mass. In the present studies, the increase in RV and LV mass in females was attenuated compared with the changes observed in the LV and in the males, but still reached levels comparable with female athletes (29).

Potential Underlying Mechanisms Regulating Sex Differences in Response to Training

Sex differences in the endogenous androgens play a key role in the development of both skeletal muscle and cardiac hypertrophy. In male rats who have undergone orchiectomy ventricular mass is decreased, while testosterone administration induces myocellular hypertrophy (22). Regular exercise training is a strong stimulus to increase circulating endogenous testosterone levels, which promotes muscle growth, through protein synthesis and inhibits protein degradation (38). Circulating estrogen levels influence cardiac hypertrophy by preventing the development of pathological cardiac hypertrophy in premenopausal females, with the benefit reduced in the postmenopausal state (16). It remains unclear whether estrogen also influences exercise-induced cardiac hypertrophy. Indeed, rodent studies suggest an enhanced cardiac hypertrophic response to training in females compared with males (24). However, key differences in animal and human female reproductive physiology, namely estrous vs. menstrual cycle, limit the interpretation of the influence of sex hormones on cardiac adaptations in mice and must be considered when interpreting animal studies. Moreover, diet also influences cardiac adaptation in mice and must be considered when using this model to assess sex differences (23). The present study suggests that endurance training has an early anabolic effect on female hearts while the increase in male heart size is more progressive possibly due to small elevations in androgens.

It should be noted that we observed a marked reduction in body fat in female participants from month 6 of training onward. This reduction in fat mass coincided with an increase in training impulse (Fig. 1), since both intensity and duration of training were dramatically increased. We did not observe similar changes in body fat in the males in response to the increase in training load. Our intervention provided only general dietary advice. Thus a possible explanation for the lack of effect of endurance training on V̇o_2max and cardiac mass in females may be partially explained by a suboptimal energy intake. The American College of Sports Medicine recognizes that female athletes are plagued by low energy intake (2). A reduced energy intake will result in the utilization of fat and lean tissue stores to fuel the body during exercise and may impair muscle protein synthesis (35).

Alternatively, other aspects of the subject's lives may have influenced the lack of adaptation observed in the female subjects, including lack of appropriate recovery or other life-related stressors (e.g., sleep, work, and emotional upheavals). Indeed all subjects showed some signs of overtraining toward the end of the study. We have previously reported that prolonged intense training does not necessarily result in greater enhancement of dynamic regulation of HR or blood pressure (18). In our prior study, we demonstrated a bell-shaped relation with late changes in autonomic control of the circulation, possibly due to the effects of overtraining. Unfortunately, we did not collect information on the subject's quality of life throughout the study or indicators of overtraining. Future studies should plan to include specific dietary advice as would be recommended to male and female athletes performing such a high volume of endurance training, and include assessments of quality of life.

Another possible explanation for the observed sex differences in cardiac hypertrophy may relate to differences in signaling events that mediate hypertrophy. Cardiac hypertrophy due to exercise is mediated by peptide growth factors and signaling through the phosphatidylinositol 3-kinase/Akt pathway (13). Premenopausal females have a fivefold increase in Akt when compared with age-matched males (10). We speculate that young females may be somewhat protected from marked increases in ventricular hypertrophy by the increased levels of Akt. Evidence from animal studies also provides some support for this hypothesis. Endurance-trained female mice have been shown to have higher levels of phosphorylated glycogen synthase kinase-3beta (GSK) in isolated cardiac tissue compared with male mice (24). GSK is a downstream target of Akt kinase and negative regulator of cardiac hypertrophy. Thus the hypertrophic response to endurance training in young females may be blunted by a combination of increased levels of Akt and GSK.

Ventricular Remodeling

The superior athletic ability of the elite athlete is partially due to a large compliant LV and enhanced ability to utilize the Frank-Starling mechanism to increase SV during exercise (26). Endurance athletes have a greater augmentation in SV for any given filling pressure and increased diastolic reserve, meaning their LVs are more compliant and distensible compared with sedentary subjects (26). We recently demonstrated that 1 yr of training resulted in a significant improvement in ventricular distensibility and possible reduction in pericardial constraint in healthy young individuals (3). In the present study, we extend these findings to examine whether the response to training differs in males and females. Despite clear sex differences in the hypertrophic response to endurance training, both males and females had a similar augmentation in end-diastolic volume and LV compliance and distensibility. The Frank-Starling mechanism was augmented to a greater extent in males compared with females, which is likely due to the greater amount of hypertrophy in the males and therefore larger SV (26). An increase in wall thickness of the ventricle results in a reduction in wall stress and afterload. This in turn allows for a smaller end-systolic volume and thus larger SV. In addition, ventricular remodeling occurs in response to endurance training, which increases muscle mass and results in greater ventricular compliance. Therefore, for a given filling pressure, the males were able to generate a greater end-diastolic volume. We speculate that even more prolonged training (e.g., years), or the commencement of training during developmental years, may be required to improve LV compliance to levels similar to endurance athletes (26). Long-term follow-up is also required to determine whether maintenance of this training frequency through middle age and beyond will indeed maintain LV compliance.

Limitations

The primary limitation with this study is its small sample size, a direct reflection of the intensive and invasive nature of the training and assessment tools. Few studies have reported the effects of sex on cardiac adaptations to exercise training. Thus, while our sample size is small, these results are highly novel in the field of exercise physiology. The small sample size is also offset by our observations of significant sex × time interactions, and our within-subject design, which minimized important sources of error. It is also notable that MRI is substantially more precise in terms of cardiac morphological assessment than echocardiography (20). LV pressure-volume curves were evaluated by use of mean PCWP as a surrogate for LV end-diastolic pressure. Unfortunately we do not have direct measures of sex hormones, although all females experienced normal menstrual cycles at all times during the study. Future studies should include these measures. Although we observed a divergent response between sexes for a-vO_{2 diff}, this finding seems to be driven by the response of one female, a further limitation of our small sample size.

Conclusion

In summary, the cardiovascular adaptation to 1 yr of endurance training has a different time course and magnitude of effect in males and females, even when performing identical endurance training. Males had a more prominent increase in LV mass throughout the study, whereas females reached maximal LV hypertrophy in 3 mo, although the intensity of training was similar in both sexes. Aerobic power was clearly related to LV hypertrophy. Ventricular compliance and distensibility improved similarly in males and females, but males demonstrate greater enhancement in the Frank-Starling mechanism compared with the females, emphasizing the divergent sex response to endurance training.

GRANTS

This study was supported by National Aeronautic and Space Administration Specialized Center for Research and Training Grant NGW-3582, the S. Finley Ewing Chair for Wellness at Texas Health Presbyterian Hospital, and the Harry S. Moss Heart Chair for Cardiovascular Research.

DISCLOSURES

The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

Author contributions: E.J.H., M.P., R.M.P., A.A.-Z., B.A.-H., and B.D.L. analyzed data; E.J.H., M.P., R.M.P., R.Z., A.A.-Z., B.A.-H., and B.D.L. interpreted results of experiments; E.J.H. prepared figures; E.J.H. and M.P. drafted manuscript; E.J.H. and B.D.L. edited and revised manuscript; E.J.H., M.P., R.M.P., R.Z., A.A.-Z., B.A.-H., and B.D.L. approved final version of manuscript; M.P., R.M.P., R.Z., A.A.-Z., and B.D.L. performed experiments; R.Z. and B.D.L. conception and design of research.