Effects of intensive exercise on hypothalamus–pituitary–testis (HPT) axis remain controversial. Our aim was to determine the effects of intensive, long-term treadmill running on reproductive hormones, HPT axis, and semen quality. A total of 286 subjects were randomly assigned to moderate-intensity exercise (∼60% maximal oxygen uptake (VO2max); group 1, n=143) and high-intensity exercise (∼80% VO2max; group 2, n=143) groups. The two groups exercised for 60 weeks in five sessions per week, each session lasting 120 min. This was followed by a 36-week low-intensity exercise recovery period. All subjects underwent routine semen analysis. Blood samples were drawn for the determination of the levels of the following hormones: LH, FSH, prolactin, testosterone (T), free testosterone (fT), inhibin B, and sex hormone-binding globulin (SHBG). The HPT axis was assessed using GnRH and human chorionic gonadotropin tests. After 24 weeks of exercise, the subjects exercising with high intensity demonstrated significantly declined semen parameters compared with those exercising with moderate intensity (P=0.03). Serum T and fT began to decrease, and serum SHBG began to increase at the end of 12 weeks with both moderate- and high-intensity exercises. The serum LH and FSH concentrations decreased below the baseline level at 12 weeks in both groups (P=0.07 in group 1 and 0.03 in group 2). Both groups had blunted LH and FSH responses to GnRH. These parameters improved to their pre-exercise level during the recovery period. Long-term strenuous treadmill exercises (overtraining syndrome) have a deleterious effect on reproduction.
There is little argument that physical activity is beneficial for one's health. Indeed, increasing the amount of physical activity is associated with extensive health improvements. However, strenuous exercise represents a physical stress that challenges homeostasis. In response to this stressor, the endocrine and nervous systems are known to react and participate in the maintenance of homeostasis and the development of physical fitness. Regardless of the many known health benefits of exercise, there is a body of evidence suggesting that endurance exercise is associated with some health problems (Mastaloudis et al. 2004, Scharhag et al. 2005). Despite the many advances in ‘andrology’ in the last decade, a considerable proportion of ‘male factor’ infertilities still remain unknown or idiopathic. In recent years, prolonged strenuous exercise has been proposed as a possible reason for male factor infertility (Arce & De Souza 1993). A subset of endurance-trained men, particularly runners, present with profound changes in their GH (Kilic 2007), thyroid hormone and testosterone, LH, dehydroepiandrosterone sulfate, cortisol (Tremblay et al. 2005), and prolactin (PRL; Bridge et al. 2003). It has been reported that both hypothalamic and testicular endocrine functions are suppressed during acute prolonged physical exercise. The exercise-induced suppression of serum testosterone is associated with suppressed endogenous GnRH stimulation of gonadotropin release during exercise (Kujala et al. 1990). Qualitatively and quantitatively normal spermatogenesis is critically dependent on an intact hypothalamus–pituitary–testis (HPT) axis. Androgens are essential for the maintenance of normal spermatogenesis.
Reactive oxygen species (ROS) produced in metabolic and physiological processes and harmful oxidative reactions may lead to oxidation of biological molecules such as lipids, proteins, and DNA (Halliwell & Gutteridge 1984). There is a body of evidence suggesting that endurance exercise is associated with oxidative stress. During the resting state, the human body produces ROS continuously, but in healthy individuals these ROS are produced at levels well within the capacity of the body's antioxidant defense system. During endurance exercise, there is a 10- to 20-fold increase in whole body oxygen (O2) consumption (Astrand & Rodahl 1986), and O2 uptake in the active skeletal muscle increases 100- to 200-fold (Halliwell & Gutteridge 1999). This increase in O2 utilization may result in the production of ROS at the rates that exceed the body's capacity to detoxify them (Alessio 1993). However, some studies have failed to observe exercise-induced oxidative stress (Poulsen et al. 1996, Oztasan et al. 2004). An increase in the formation of ROS decreases fertility, as the ROS will attack the membranes of the spermatozoa, decreasing their viability (Irvine 1996). Altogether these can impair spermatogenesis and fertility capacity in men. Less attention has been directed toward identifying changes in spermatogenesis and fertility capacity as a result of endurance training.
Treadmills offer many distinct advantages. Speed, slope, and environmental factors can be easily controlled and data from repeated running cycles can be collected. Treadmills have therefore been used in many experimental studies. In this study, we addressed the effects of prolonged high-intensity treadmill running on semen quality, reproductive hormones, and HPT axis.
Materials and Methods
A total of 362 male volunteers between February 2002 and July 2006 were recruited from sport facilities for this study. Volunteers for the study were aged 20–40 years and recruited through local advertisement. The trial requirements were carefully explained to participants, who were given the opportunity to withdraw at any time they desired. They were habitual aerobic exercisers and reported training for an average of 1.8 h per day, 5 days per week. Participation was voluntary throughout the study period, and exercise equipment and supplies were free of charge. Eligible participants were invited to an orientation session. Individuals who remained interested were assessed for eligibility and signed a written informed consent form. The investigation was approved by the local medical ethical committee, which was conducted in accordance with the International Conference on Harmonisation-Good Clinical Practice (ICH-GCP) guidelines and the principles of the Declaration of Helsinki.
Inclusion and exclusion criteria
Subjects were included in the study after fulfilling the following criteria: no known medical condition that could interfere with their fertility and total testicular volume (measured by ultrasound) ≥12 ml. All participants were required to have ceased all medical therapies at least 12 weeks before the start of the study. Participants were excluded if they had abnormal semen analysis according to the World Health Organization (WHO) criteria (1992), abnormal reproductive endocrine profiles, or positive antisperm antibody assay (ASA) test; present use of ergogenic aids such as creatine monohydrate, herbal stimulants, or any anabolic agents such as steroids; presence of genital diseases (cryptorchidism, present genital inflammations, or varicocele); a diagnosis of alcohol and/or drug dependence; Y chromosome microdeletions or karyotype abnormalities; and sexually transmitted disease or relevant genitourinary infection. Participants were also excluded if they had a body mass index (BMI) ≥30; occupational and environmental exposures to potential reproductive toxins; participation in reduced caloric intake diets that may alter basal hormonal levels; and were participating in another investigational study.
All subjects were screened for nutritional habits by a registered dietician. Subjects, who were unable to intake adequate energy, including 30–40% energy from fat and 15–20% energy from protein, were excluded from the study. The participants underwent a thorough physical examination, anthropometric measurements, urine analysis and serum chemicals, hematological laboratory, and thyroid function tests. Body weight was determined using a Detecto beam scale (Brooklyn, NY, USA) accurate to 0.1 kg, and height was determined using a Seca stadiometer accurate to 0.1 cm. Body composition was estimated with Harpenden skinfold calipers using the standard methods outlined in the Anthropometric Standardization Reference Manual (Harrison et al. 1988).
Testicular volumes were measured with scrotal ultrasound by a sonologist using electronic calipers. A volume of less than 12 ml was considered small. The presence of varicocele was determined by Doppler ultrasonography of scrotum with the Valsalva maneuver. Genetic testing included karyotype analysis and Y chromosome microdeletion analysis.
All participants had at least two baseline semen analysis following an abstinence period of 3–4 days. An average of two for each subject and each parameter was taken for comparison with the post-trial values. When the values differed by more than 20%, a third test was done. All procedures and interpretations used were in accordance with the established WHO (1992) criteria, besides morphology, which was established according to the Kruger parameters. The normal WHO values included sperm concentration of 20×106 spermatozoa/ml or more and motility of 50% or more with forward progression. Using the Kruger strict criteria, males with greater than 14% normal forms were considered normal. The total motile sperm count was determined by the formula: (semen volume×sperm density×motility percentage)/100. Total sperm per ejaculate was calculated as the product of ejaculate volume and sperm concentration. Quality control and proficiency criteria were set for <15% variation for certifying the technicians trained to do semen analysis for the study. In every subject, the immunobead test for ASA binding was also performed.
Two early morning blood samples were drawn from each subject after overnight fasting at 20-min intervals for the determination of the resting levels of the following hormones: FSH, LH, inhibin B, PRL, total testosterone (T), free testosterone (fT), and sex hormone-binding globulin (SHBG).
The GnRH stimulation test was performed as follows: 100 μg GnRH analog (GnRHa; Decapeptyl, Ferring, Germany) was administered by i.v. bolus. Blood was drawn (for serum LH and FSH determination) at −15, 0, 20, 40, 60, and 120 min after injection. We defined a more than 3-fold increase in FSH and a more than 4-fold increase in LH in response to Decapeptyl as GnRH positive, and increases of less than 3-fold and 4-fold respectively as GnRH negative.
Three days after GnRH test, all participants received an injection of human chorionic gonadotropin (hCG; Pregnyl, 50 IU/kg i.m., Organon, Oss, The Netherlands). Serum T was determined before the hCG injection and on the third day afterward. After the hCG injection, the serum T level should rise in 3 days by a factor of 2–4 compared with the basal value (positive test).
Single-antibody, solid-phase RIA was used to determine the immunoreactivity levels indicative of blood concentrations of T and fT (Diagnostic Products, Los Angeles, CA, USA). Variances for the RIAs performed in this investigation were as follows: total testosterone (intra-assay ≤2.5%, interassay=3.0%) and free testosterone (intra-assay ≤2.5%, interassay=9.6%). Serum LH and FSH were measured using time-resolved immunofluorometric assay kits (DELFIA hLH for LH and DELFIA hFSH for FSH; Wallac Co., Turku, Finland). The intra- and interassay coefficients of variation of the individual immunofluorometric assay method was below 9% within the reference ranges. The reference ranges were as follows: T, 9–38 nmol/l; fT, 90–310 pmol/l; LH, 1.0–8.4 IU/l; and FSH, l.0–10.5 IU/l.
Serum SHBG was determined using 1235 AutoDELFIA automatic system based on a time-resolved fluoroimmunoassay (AutoDELFIA SHBG, Wallac Co). The between-assay coefficient of variation is 2.3–3.0%. Non-SHBG-bound fT was obtained using the following formula: fT (pmol/l)=T (nmol/l)/(K×SHBG (nmol/l)+1)×1000, where K is the equilibrium constant for T binding to SHBG (1.6×109 l/mol; Glüer et al. 1995). Serum levels of PRL were assayed by commercial RIA kit. This commercial kit has been used previously with inter- and intra-assay variations of less than 10%. The reference ranges for PRL was 92–697 pmol/l. Serum Inhibin B was determined by ELISA method using kit reagents and inhibin B standard (Oxford Bio-innovation Ltd, Oxon, UK). The assay sensitivity was 4 pg/ml and the between-assay variation was 15%. The antibodies bound to the surface of the sperm were detected with the ‘immunobead-binding test’. ASA was performed using specific beads that bind separately to IgG and IgA. The ASA test was considered positive when there was more than 20% sperm-bound immunobeads as recommended by the WHO (1992). Out of the 362 screened subjects, 304 met inclusion/exclusion criteria and entered into the screening phase of the study.
The study consisted of a 12-week screening and familiarization phase, a 60-week treadmill running phase (intervention), and a 36-week recovery period (Fig. 1).
All subjects were required to complete a Physical Activities Readiness Questionnaire (Thomas et al. 1992) and the American College of Sports Medicine risk stratification for coronary artery disease risk factors (Franklin et al. 2000) to ensure that all were free from any known cardiovascular, pulmonary, or musculoskeletal conditions.
All exercises were done on a motorized treadmill (HP Cosmos Venus, Traunstein, Germany). The driving system provided a range of constant speed from 1 to 22 km/h (by minimum increments of 0.1 km/h). The maximal belt speed error reached 7% with a subject of 100 kg walking at 6 km/h.
The subjects visited the laboratory on two occasions and completed a continuous, inclined treadmill test using the modified Åstrand protocol (Pollock et al. 1978) to volitional exhaustion for the determination of maximum oxygen consumption (VO2max). Oxygen consumption was measured using indirect calorimetry. Metabolic measures were collected using standard open-circuit spirometric techniques (metabolic cart 2700Z; SensorMedics, Yorba Linda, CA, USA). Maximum VO2 (VO2max) was considered to be attained when the subjects met at least two of the following criteria: 1) a plateau in VO2 with an increase in workload, 2) respiratory exchange ratio greater than or equal to 1.15, and 3) maximum age-predicted heart rate (220 – age) is attained. The highest VO2 attained during the tests was registered as the VO2max value.
For the purposes of this study, low-, moderate-, and high-intensity exercises (HIE) were defined as 30, 60, and 80% VO2max respectively. Exercise at 60% VO2max will be sufficient to elicit clinically significant physiological changes associated with exercise training; 80% VO2max will allow those who wish to work at a higher intensity but not above most individuals’ anaerobic threshold (NIH Consensus Development Panel on Physical Activity and Cardiovascular Health 1996, Snyder et al. 1997).
At the end of week 4, the subjects were recommended to complete a ‘multi-stage’ incremental treadmill exercise in which the running speed, incline, and duration were increased gradually to the subject's limit of tolerance. The starting level and the rate of increments were voluntary. However, the participants were required to increase the exercise level to their pre-determined 80% VO2max for 2 h, 5 sessions per week, maximum within the 12-week screening period. Out of the 304 participants who entered the screening phase, 286 met the study protocol and continued participation with the intervention or running phase.
Participants were randomly assigned to moderate-intensity exercise (MIE; ∼60% VO2max; group 1, n=143) and high-intensity exercise (HIE; ∼80% VO2max; group 2, n=143) groups. Assignment to exercise group was performed using an interactive voice response system, which followed a randomization table generated by the method of random permuted blocks. Participant randomization numbers were allocated to each site in an ascending sequence in blocks. Randomization was not stratified by age. The two groups exercised for 60 weeks in five sessions per week, each session lasting 120 min. All subjects were asked to exercise to the limit of the study. Other types of exercises such as swimming, cycling, and sauna bathing were not allowed during the whole study protocol. To avoid the confounding factors associated with a training effect, the subjects agreed to maintain their exercise regimen for the duration of the study period. Throughout the running phase, the subjects were made aware of the potential risks of exercises on reproductive capacity.
Out of the 286 subjects who participated in the running phase, 264 (134 out of 143 in group 1, and 130 out of 143 in group 2) were eligible and consented to continue with the recovery phase of the study (Fig. 1). Subjects with any violation of the study protocol, which could affect the study results, were excluded according to the investigator's discretion.
The last 36 weeks consisted of the post-intervention phase during which the long-term effects of the intervention were assessed every 12 weeks. In this phase, all subjects were asked to continue exercising only at low-intensity exercise (LIE) level (∼30% VO2max) if they so desired. They were required to to exercise 120 min per day, 5 days per week, for 36 weeks. The subjects were free to choose further participation with study protocol or discontinue any kind of participation.
Evaluations during the whole study protocol
During each exercise, surface skin thermistor probes were attached to the skin surface at four locations (chest, triceps, thigh, and calf) to determine the weighted mean skin temperature (Ramanathan 1964). Excellent adherence to both intervention and measurement is necessary. Each week, a tracking report was generated to monitor adherence to exercise prescription. In order to maintain high adherence rates, flexibility is required. If a participant failed to meet the study requirements for 1 week, the study requirement for the following week (s) could be increased up to 1.5 kcal/kg per week (KKW) to compensate. For the planned exercise sessions, exercise intensity was quantified using the data from the Polar XL HR monitor (Polar Heart Rate Monitor, Kempele, Finland) worn by the participants. The appropriate heart rate range for the prescribed intensity (30, 60, and 80% VO2max) was calculated following completion of the exercise. If a participant's HR falls or rises out of range, the speed and/or grade of the treadmill were being adjusted to maintain the prescribed intensity. For monitoring daily activity outside the assigned exercise session, each participant was given a step counter (ACCUSPLIT Eagle, Japan) at baseline and instructed to wear it at all times during waking hours. The participants were asked to remove the step counter only during planned exercise sessions, or for purposes of bathing, sleeping, or dressing. At the end of each day, the participants recorded the number of steps taken and reset the step counter for the following day. The participants were also asked to make note of any events resulting in significant changes in activity, for example, illness or injury. In addition, physical activity questionnaires were completed by the participants during all three phases of the trial.
During the monthly visit, a medical history, a physical examination, and a resting 12 lead electrocardiogram were obtained, and a comprehensive health status questionnaire was administered. Actual duration and frequency of exercises was recorded in the diaries that were monitored for exercise compliance during monthly visits. At follow-up visits (every 12 weeks, Fig. 2), two semen samples were collected within a 1- to 2-week period of each other and endocrinological studies were done as described. Hormonal profiles were assessed on a separate occasion from exercise intervention, but at the same time of day, to establish a reference baseline.
Statistical analysis of outcomes will be based on the intention-to-treat principle. Sample size was projected using PASS 2000 (NCSS statistical software 2000), and was powered for a difference of ∼1 s.d. between the HIE and LIE groups. Given these assumptions, and with an α risk of 5% (two-tailed) and a β risk of 15%, we calculated that 115 subjects per arm were required to achieve sufficient power to allow comparison of exercises. Taking into account dropout rates during the screening period, additional participants were included to ensure sufficient participants in each arm. Data are presented as means±s.d., unless otherwise stated. Differences in sociodemographic variables between groups were determined with the paired t-test/one-way ANOVA for dichotomous and normally distributed continuous variables respectively. All data were log transformed before analysis and the variances analyzed for homogeneity. Nonparametric statistics were used when equal variance tests failed. The significance of the difference in mean values between the study groups was tested with the unpaired t-test. FSH and LH responses to GnRHa were compared by repeated-measures ANOVA with post hoc pairwise comparison (Scheffe's test). The nonparametric Mann–Whitney test was also used to compare the LH and FSH responses with that of GnRHa at various time points with Bonferroni correction. The response of the various hormones to GnRHa and hCG was evaluated as area under curve (AUC), calculated as the difference between the area under the curve (calculated by trapezoidal analysis) during the 120-min test and the area corresponding to the mean of the two basal concentrations. Linear regression was used to analyze the correlation between the peak and area of LH response to GnRHa with serum T level. Correlations were assessed by Spearman's rank correlation test. P<0.05 was considered statistically significant. Statistical analysis was performed using SPSS Base version 10.0 (SPSS, Chicago, IL, USA) and SigmaStat software SPSS.
Characterization of the subjects
The participant flow diagram is shown in Fig. 2. The baseline characteristics of the study subjects are provided in Table 1. Anthropometric and descriptive characteristics of the subjects for age, height, weight, and body fat were 27.5±9 years, 174.4±6.6 cm, 72.6±4.5 kg, and 28.2±6.2% respectively (Table 2). At the end of the 60-week running period, all anthropometric variables differed significantly between groups, especially the amounts of percent body fat and the waist/hip ratio. Out of the 286 recruited subjects, 216 (75.5%) had fathered a pregnancy, previously. The actual intensities measured during the three stages of the exercise trial were 31, 61, and 79% (±1%) of VO2max in the group as a whole, with no differences among the three intensities. At the beginning of running phase, semen parameters and endocrinological studies did not differ significantly from baseline values.
Characteristics of the study subjects (mean±s.d.) at the randomization in the exercise
|HIE group (n=143)||MIE group (n=143)||P value|
|Free testosterone (pmol/l)||195±58||198±61||0.384|
|Inhibin B (ng/l)||105±26||107±25||0.536|
|Acrosome reaction (%)||33±8||34±8||0.654|
|Ejaculate volume (ml)||2.7±1.4||2.8±1.3||0.642|
|Total sperm/ejaculate (×106)||196±32.6||197±32.4||0.358|
|Sperm concentration (×106/ml)||66.2±14.6||64.4±14.4||0.247|
|Motility (% motile)||58.2±4.4||57.7±4.6||0.652|
|Strict morphology (% normal)||19.2±2.6||19.3±2.7||0.662|
HIE, high-intensity exercise; MIE, moderate-intensity exercise; BMI, body mass index; PRL, prolactin; SHBG, sex hormone-binding globulin.
Summary of anthropometric data at baseline, and at the end of running phase
|High-intensity exercise||Moderate-intensity exercise|
|Variable||Baseline||Running phase||P value*||Baseline||P value †||Running phase||P value ‡||P value §|
|Body mass (kg)||73.6±3.5||65.7±11.2||0.03||71.6±5.5||NS||67.1±11.4||0.04||0.03|
|Body fat (%)||28.3±6.2||24.3±6.8||0.02||28.1±6.4||NS||26.8±6.7||0.04||0.04|
Values are expressed as means±s.d. BMI, body mass index; HIE, high-intensity exercise; MIE, moderate-intensity exercise. *P value, 60-week HIE versus baseline HIE. †P value, baseline MIE versus baseline HIE. ‡P value, 60-week MIE versus baseline MIE. §P value, 60-week MIE versus 60-week HIE.
Mean testicular volumes remained unchanged during the running phase. At the end of this phase, testicular volume in the MIE group (23.3±2.6 ml) did not differ significantly from that in the HIE group (22.6±2.6 ml; P=0.07).
Table 3 shows post-exercise semen parameters. The most impaired post-exercise semen analysis was used for data analysis. After 24 weeks of exercise, the subjects exercising with high intensity demonstrated significantly declined semen parameters (sperm density, motility, and morphology) compared with those exercising with moderate intensity (P=0.03). At 36 (P=0.02) and 48 weeks (P=0.01), these differences were more significant. Continuing HIE worsened impairment in semen parameters. At the end of the running phase, the subjects in the MIE group had also decreased semen parameters compared with baseline (P=0.044), which did not reach statistical significance after the Bonferroni correction. In the overall analyzed subjects, mean percentage changes in sperm parameters from baseline at 24 weeks, 36 weeks, and 48 weeks were different from each other in both groups (Fig. 3). A significant correlation was found between HIE and sperm count (r=−0.62, P=0.01), as well as mean sperm motility (r=−0.65, P=0.01) and sperm morphology (r=−0.54, P=0.01). Also, there was a significant correlation between HIE duration and sperm count (r=−0.68, P=0.01), sperm motility (r=−0.68, P=0.01), and sperm morphology (r=−0.66, P=0.01). By the end of the running phase, percentage of acrosome reaction decreased from 34±8, and 33±8, to 30±10, and 22±9, in groups 1, and 2 respectively (P=0.07 and 0.01). HIE duration had significant correlation with acrosome reaction (r=−0.52, P=0.02).
Summary of semen parameters and reproductive hormones at the end of running and recovery phases
|High-intensity exercise||Moderate-intensity exercise|
|Parameters||Running phase||P value*||Recovery phase||P value†||Running phase||P value‡||P value§||Recovery phase||P value†||P value∥|
|Ejaculate volume (ml)||2.7±1.6||0.1||2.7±1.7||0.1||2.8±1.1||0.1||0.1||2.7±1.4||0.1||0.1|
|Sperm count (106)||106±20.8||0.01||188±31.2||0.92||161±31.4||0.044||0.03||191±30.4||0.72||0.64|
|Sperm concentration (106/ml)||35.4±4.2||0.01||64.8±4.2||0.84||56.8±3.6||0.042||0.02||62.8±3.4||0.86||0.23|
|Sperm motility (% motile)||47.6±3.4||0.02||56.8±4.1||0.78||54.2±3.2||0.09||0.02||56.8±4.2||0.09||0.74|
|Strict morphology (% normal)||14.6±2.4||0.02||18.8±2.1||0.07||16.2±2.1||0.07||0.06||18.4±1.8||0.08||0.07|
|Free testosterone (pmol/l)||105±22||0.02||200±62||0.09||164±60||0.048||0.02||202±64||0.08||0.1|
|Inhibin B (ng/l)||71±22||0.03||98±25||0.08||94±22||0.08||0.02||109±20||0.1||0.08|
|Acrosome reaction (%)||22±9||0.01||31±9||0.09||30±10||0.043||0.02||34±11||0.1||0.08|
Values are expressed as means±s.d. HIE, high-intensity exercise; MIE, moderate-intensity exercise; PRL, prolactin; SHBG, sex hormone-binding globulin. *P value, 60-week HIE versus baseline. †P value, 36-week recovery phase versus baseline. ‡P value, 60-week MIE versus baseline. §P value, 60-week MIE versus 60-week HIE. ∥P value, 36-week recovery phase after MIE versus 36-week recovery phase after HIE.
Mean baseline serum LH, FSH, PRL, inhibin B, T, fT, and SHBG concentrations were not significantly different among groups at baseline (all P=0.1; Table 1).
Serum T and fT began to decrease, and serum SHBG began to increase at the end of 12 weeks with both moderate- and HIE. These alterations from baseline reached statistical significance with HIE at 24 weeks. Decreases in T and fT and an increase in SHBG were significant (P=0.01, 0.02, and 0.03 respectively) during HIE throughout the running phase. A correlation analysis between the duration of exercises and the serum levels of T, fT, and SHBG demonstrated a significant correlation between the duration of HIE and the serum levels of T (r=−0.64, P=0.01), fT (r=−0.58, P=0.02), and SHBG (r=0.52, P=0.03).
The serum LH and FSH concentrations decreased below the baseline level at 12 weeks in both groups (P=0.07 in group 1 and 0.03 in group 2). These are followed by more decreases during the running phase. At the end of the 60-week running phase, serum FSH and LH levels in the HIE group were lower than those in the MIE group (P=0.01; Fig. 4). Significant correlation coefficients were obtained between the HIE and serum levels of LH (r=−0.67, P=0.01) and FSH (r=−0.68, P=0.01). Also, a high negative correlation existed between the HIE duration and serum levels of LH (r=−0.74, P=0.01) and FSH (r=−0.78, P=0.01).
PRL concentration rose progressively in the HIE group and was significantly greater than those in the MIE group. The differences in plasma PRL release were also reflected in the calculated AUC, which was greater after 12 weeks of exercise in the high-intensity condition (P=0.01). We found a significant positive correlation between the serum levels of PRL and the high-intensity type of exercise (r=0.66, P=0.02) and exercise duration (r=0.68, P=0.01).
By the end of the 60-week running phase, serum inhibin B significantly decreased in the HIE group compared with the LIE group (P=0.03). A significant correlation exists between HIE duration and serum inhibin B concentration (r=−0.56, P=0.02).
At the end of the running phase, the injection of GnRHa yielded a mild increase in the FSH and LH levels of the HIE group compared with the subjects in the MIE group (P=0.01). Both groups had blunted LH and FSH responses to GnRHa (Fig. 5).
In the HIE group, FSH increased less than twofold, from an average of 6.9 IU/l before the injection to 11.4 IU/l (P=0.001), and LH increased less than twofold, from an average of 6.4 to 10.5 IU/l (P=0.001). In the MIE group, FSH rose from 6.4 to 15.8 IU/l (less than threefold), and LH increased from 6.6 to 19.4 IU/l (less than fourfold). Thus, all subjects were negative for both parameters. There were significant correlation coefficients between moderate and HIE and LH (r=−0.45, P=0.02 and r=−0.79, P=0.001 respectively) and FSH (r=−0.48, P=0.02 and r=−0.78, P=0.001 respectively) responses to GnRHa. There was a significant correlation between the serum levels of LH and FSH and their responses to GnRHa (P=0.01). Log-transformed LH and FSH AUC were significantly correlated with progressive duration of exercises (r=−0.56, P=0.01; r=−0.54, P=0.01 respectively). Mean integrated LH AUC demonstrated a difference between exercises (mean LH AUC: 415±143 vs 619±146 IUmin/l for HIE vs MIE; P=0.01). There was also a trend for a lower FSH AUC in the HIE group (mean FSH AUC: 365±104 IUmin/l) compared with the MIE group (mean FSH AUC: 545±112 IUmin/l; P=0.01).
Human chorionic gonadotropin test
On the third day after hCG administration, the serum levels of T increased by 74% in the HIE group, and 126% in the MIE group. Serum T responses to hCG were significantly correlated with the high-intensity type (r=−0.75, P=0.01) and duration of exercise (r=−0.76, P=0.01; Fig. 5).
At the beginning of this phase, semen parameters and endocrine profiles were abnormal in all subjects (Table 3).
These parameters improved to their pre-exercise level during the recovery period (Table 3). There were positive significant correlations between the recovery time length and sperm concentration (r=0.74, P=0.01) and sperm motility (r=0.64, P=0.02). The same was found regarding the normal sperm morphology (r=0.59, P=0.02). The recovery time length showed a positive significant association with sperm concentration (β=10.0×106/ml per 12 weeks), total sperm motility (β=5% per 12 weeks), and normal sperm morphology (β=2% per 12 weeks).
The results of the serum hormones for the two groups are given in Table 3. Both the moderate- and HIE groups showed marked and significant increases in serum LH, FSH, T, fT, and inhibin B concentrations and significant decreases in the serum levels of PRL and SHBG, at each visit up to week 36. During the recovery phase, serum hormone concentrations returned to baseline in both the moderate- and HIE groups, although the recovery was somewhat more rapid in the MIE group (P=0.07).
GnRH and hCG tests
The GnRH and hCG tests were repeated in all participants at 12, 24, and 36 weeks. GnRH test was positive in 54%, 87%, and 94% of subjects in 12, 24, and 36 weeks of recovery phase respectively. Recovery was faster in the MIE group (P=0.03). Of the participants in the recovery phase, 49%, 78%, and 92% had positive hCG tests, in 12, 24, and 36 weeks of the recovery phase respectively. Recovery of GnRH and hCG tests had significant correlation with recovery phase length (r=0.75, P=0.01 and r=0.78, P=0.01 respectively).
Weighted mean skin temperature
Room temperature ranged between 22 and 25 °C during the treadmill exercises. Mean skin temperature was increased progressively during exercise in all three groups (P<0.001), reaching 39.2±0.7 °C in the HIE group, 39.1±0.6 °C in the MIE group, and 39.0±0.67 °C in the LIE group.
The present investigation is a longitudinal study of reproductive endocrine profile and HPT axis during long-term HIE in men. The results of this study demonstrate that strenuous long-term treadmill exercise caused a significant decrease in plasma LH, FSH, and T concentrations. Our aim was to include subjects with no known medical condition (such as Y chromosome microdeletion) that could interfere for their fertility. Therefore, Y chromosome microdeletion analysis was carried out in all subjects. Yq microdeletions have been previously reported in some studies of fertile males (Kobayashi et al. 1994, Vogt et al. 1996). These microdeletions may be associated with suboptimal sperm count (Andersen et al. 2000). Hence, deterioration of semen parameters in these individuals cannot be attributed solely to intensive treadmill running. There are some studies showing data of the quality of semen in highly endurance-trained groups.
Gebreegziabher et al. (2004) investigated the influence of cycling on sperm characteristics. They concluded that endurance cycling appears to be associated with a significant alteration in sperm morphology. In another study, gonadal hormones and semen quality were evaluated in male runners (De Souza et al. 1994). In this study, the volume of training was significantly correlated with sperm motility, density, and number of round cells. Hormonal milieu was also assessed in endurance-trained male athletes. Low serum FSH, LH, and T values indicate hypothalamo-pituitary disease, i.e., hypogonadotropic hypogonadism. Lower basal levels of T or fT have also been reported in endurance-trained individuals in previous studies (Cooper et al. 1998, Hackney et al. 1998). In another study, 20 male marathon athletes were evaluated by hormonal profiles, psychological testing, anthropomorphic indices, and semen evaluations by Ayers et al. (1985). In this study, vigorous endurance training was associated with significantly decreased T values. The male reproductive system is regulated by the HPT axis. The principal regulator of the HPT axis is GnRH. GnRH stimulates pituitary FSH and LH secretion and, subsequently induces spermatogenesis by seminiferous tubules and testosterone secretion by the Leydig cells. Exercise presents a serious challenge to the homeostasis of the body. The pituitary hormone response to such exercise is often described as a ‘stress response’. The stress system consists of the hypothalamic–pituitary–adrenal (HPA) axis along with the arousal and autonomic nervous systems. The stress response in the neuroendocrine system includes release of corticotropin-releasing hormone (CRH) from the hypothalamic paraventricular nucleus and the secretion of pituitary ACTH, leading to the secretion of glucocorticoid hormones by the adrenal cortex (Carrasco & Van de Kar 2003). In response to strenuous exercise, there is an enhanced secretion of cortisol, stimulated by increased release of ACTH, which serves to mobilize fuel stores (Deuster et al. 1989) and CRH-induced proopiomelanocortin peptides, such as β-endorphin, and inhibits hypothalamic GnRH secretion (Chen et al. 1992). In addition, glucocorticoids suppress gonadal axis function at the hypothalamic–pituitary level (Sakakura et al. 1975). CRH and its receptors have also been identified in the Leydig cells of the testis, where CRH exerts inhibitory actions on testosterone biosynthesis (Fabri et al. 1990). In fact, CRH hyperactivity is inferred to be associated with the overtraining syndrome (Keizer 1998). An inhibitory effect of the HPA axis on the female reproductive system has also been demonstrated (Chrousos et al. 1998).
Despite the strong evidence for the health benefits of exercise, serum markers do not necessarily support this (Cooper et al. 1998). The exercise protocol used in this study proved to be an effective means of examining the effects of stress on reproductive function. Furthermore, the quality of semen was negatively influenced by repeated HIE sessions over a period of 60 weeks. The evaluation of fresh semen revealed that the percentage of major defects, particularly abnormal acrosome reaction started to rise 12 weeks after the beginning of the intensive exercise. The most impaired post-exercise semen analysis was used for data analysis. We also used the least impaired semen analysis for statistical analysis. Findings were similar between exercise groups. In addition, the changes in semen parameters were all within the fertile range. In the present study, we used a long bout of high-intensity treadmill exercise to test the integrity of the HPT axis, and we tested the hypothesis that long-duration HIE is associated with the hyporesponsiveness of the HPT axis as reflected by low serum levels of LH, and FSH after the HPT axis activation. The changes in gonadotropins and testosterone are clinically relevant with hyporesponsive HPT axis. Kujala et al. (1990) have reported that the exercise-induced suppression of serum T is associated with two effects: suppressed endogenous GnRH stimulation of gonadotropin release during exercise, and decreased testicular capacity to secrete T during recovery period. Assessment of sex hormone status in runners revealed diminished fT values attributable to strikingly increased SHBG concentrations. In addition, direct RIA of fT concentrations showed significant decreased values in the HIE group, suggesting that the runners were, in fact, hypogonadal. Our findings are consistent with the report of Remes et al. (2004). They reported that long-term low- to MIE had influence on serum estradiol, T, fT, or SHBG levels in middle-aged men. SHBG concentrations have been observed to increase acutely with aerobic physical activity exercises such as cycle ergometry and longer term endurance training (Gray et al. 1993). Several other hormones appear to be correlated with SHBG levels. For instance, it has been reported that estradiol significantly increases SHBG, and by contrast, insulin and testosterone decrease SHBG (Haffner 1996). In the present study, T levels (r=−0.37, P=0.01) were negatively associated with SHBG levels. In this study, baseline T and fT concentrations, as expected, were not normally distributed, and thus were log transformed prior to regression and analysis. Obesity is associated with alterations in the pituitary–adrenal and pituitary–gonadal functions. (Glass, 1989) Visceral fat accumulation is associated with decreases in testosterone concentrations in males and reduces SHBG concentrations in both sexes (Lindstedt et al. 1991, Haffner 2000). Significant negative Pearson correlations between SHBG concentrations and fat body phenotypes at baseline were found (r=−0.36, P=0.01). For the training response, significant negative correlations were observed for SHBG with fat mass (r=−0.16, P=0.04) and BMI (r=−0.14, P=0.03).
Elevated PRL levels were also noted by HIE. Since multiple neural pathways that influence PRL secretion converge on the hypothalamus from other parts of the brain, the effect of exercise on the secretion of PRL may also reflect the action of different neural inputs on the activity of the hypothalamic–pituitary axis.
It has been reported that endurance exercise is associated with oxidative stress (Mastaloudis et al. 2001). F2-isoprostanes is a sensitive biomarker to assess lipid peroxidation in trained subjects. Plasma F2-isoprostane levels increase significantly during the 50 km ultramarathon. As we mentioned previously, exercise may increase free radicals and ROS, which may interact with lipids, DNA, and proteins (Mickelborough et al. 2000, Hreljac & Stergiou 2002). Left unchecked, these ROS may cause protein, lipid, and/or DNA damage. However, some studies have suggested that exercise training enhances antioxidant capacity (Child et al. 1998, Clarkson & Thompson 2000). Indeed, the machinery eliminating ROS adapts after regular exercise and actually lowers the amount of ROS that is produced, especially in the major organs (muscles) of oxygen consumption and ROS production. Exercise training tends to decrease ROS also in body fluids, although no data concerning seminal fluid seem to be available. The possibility of ROS affecting sperm quality remains, but participation of other, maybe unknown factors seems more probable. Regardless of the exercise protocol studied, increases in DNA damage in peripheral human white cells have been reported, generating the consensus that exercise does indeed induce DNA damage (Hartmann & Niess 2000). After an exercise bout, DNA damage persists for up to 7 days (Tsai et al. 2001). The presence of high ROS levels has been reported in the semen of between 25 and 40% of infertile men (Padron et al. 1997). This is because ROS, at high levels, are potentially toxic to sperm quality and function (Saleh & Agarwal 2002). Therefore, persistent ROS formation during continuous strenuous exercise might be harmful for normal spermatogenesis. In order to confirm this link it is important to assess the oxidative stress and antioxidant levels in the seminal plasma at baseline and after therapy. We did not do this and this is one of the study limitations.
The intratesticular temperature should be lower than the core temperature for normal sperm production (Hjollund et al. 2002). External heating of the testes for short periods of time results in a dramatic but temporary decrease in sperm count after a delay of 6–8 weeks (Procopé 1965). The adverse effect of physical stress might be the result of increased body temperature compromising testicular thermoregulation during strenuous exercise.
After 1 year of exercise intervention, serum levels of T, fT, LH, and FSH activity were significantly lower in the HIE than MIE group. Testicular heating, ROS formation, and gonadotropin suppression are all contributing factors to semen changes. Another study limitation is that subfertile men were not included and hence extrapolation of the falls in the semen parameters to a subfertile population should be made cautiously.
Because the changes in hormone levels returned to baseline values in recovery phase, our intervention trial with LIE had no influence on serum LH, FSH, T, fT, or SHBG levels in middle-aged men. Our exercise program at low to high intensity progressed following a 12-week retraining program and began under close supervision. In addition, a strong point of our study is the excellent adherence and compliance to the intervention, as indicated by the low dropout rate.
However, we cannot generalize the present results to other age groups and medically frail men. Furthermore, the influence of the specific type of exercise training (e.g., bicycling versus jogging) could not be evaluated.
Despite strong evidence for the health benefits of continued aerobic exercise into later life, our results demonstrate that long-term strenuous exercise (overtraining syndrome) impairs reproductive capacity and the HPT axis in middle-aged men. After cessation of HIE, impairment of the reproductive system is fully recovered.
Declaration of interest
We declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector.
We thank all of the subjects for participation in the study. We would like to thank the two referees for their helpful comments on an earlier draft of this article. We acknowledge the invaluable contributions of the following individuals to the present project: Dr H Ziraksaz for his continuous support and valuable contribution, Ms Shiva Safarinejad for critical reading of the manuscript, Mrs Nayyer Shafiei for her laboratory assistance, Ms Saba Safarinejad for her invaluable help during the data collection, Mr A R Mohammadian for excellent technical assistance, and Dr S Y Hosseini for body composition measurements in this study.
FranklinBAWhaleyMHHowleyET. Philadelphia: Lippincott Williams & Wilkins.
HalliwellBGutteridgeJMC. New York: Oxford University Press Inc..
PollockMLWilmoreJHFoxSMIII. New York: John Wiley & Sons, Inc.
SnyderKADonnellyJEJacobsenDJHertnerGJakicicJM1997The effects of long-term, moderate intensity, intermittent exercise on aerobic capacity, body composition, blood lipids, insulin and glucose in overweight females. International Journal of Obesity and Related Metabolic Disorders211180–1189.