This review highlights the author’s current perspective on the most prominent hypotheses that have been proposed to explain the high prevalence of menstrual disorders observed in physically active women. Readers are referred to earlier reviews (7,13,14) as introductions to more comprehensive considerations of the related literature. In athletes, most menstrual disorders result from a disturbance of the gonadotropin releasing hormone (GnRH) pulse generator in the hypothalamus of the brain. This is reflected in a disruption of the pulsatile rhythm of luteinizing hormone (LH) pulsatility in the blood, on which ovarian function critically depends. What disturbs the GnRH pulse generator in athletes has been the subject of controversy for 20 yrs. The earliest hypothesis based on anthropometric measurements attributed these disruptions to insufficient body fat stores. Later hypotheses based on other behavioral observations and biochemical measurements attributed the disruptions to the stress of exercise and to energy deficiency caused by dietary restriction or exercise energy expenditure.
FAILURE OF THE BODY COMPOSITION HYPOTHESIS
The early hypothesis that the female reproductive system is disrupted when body fat declines below a critical level has been contradicted by observations that amenorrheic and regularly menstruating athletes span a common range of body composition lower than the general population of women. More persuasive experimental evidence has been available in the animal literature for many yrs, and recently such evidence was reported in humans, as well (7,13,14). In a study of severely obese women (body weight ∼ 130 kg; BMI ∼ 47), surgical reduction of their stomach volume reduced the amount of food that these patients could eat, resulting in rapid weight loss and amenorrhea while they were still greatly overweight (body weight ∼ 97 kg; BMI ∼ 35) (2).
Interest in the body composition hypothesis was rejuvenated several yrs ago by the discovery of leptin. Because leptin is secreted by adipose tissue cells, it was originally thought to signal information about the size of body fat stores. Rapid and profound declines in leptin were soon observed in response to fasting and dietary restriction, however, and similarly extreme increases were observed in response to overfeeding and refeeding after energy restriction, all before changes in adiposity could occur. These observations led to the revised hypothesis that leptin actually signals information about dietary energy intake. Since then, however, we have shown that the level and diurnal rhythm of leptin actually depend on energy availability (defined as dietary energy intake minus exercise energy expenditure) and that exercise itself has no suppressive effect on leptin beyond the impact of its energy cost on energy availability (4).
FAILURE OF THE STRESS HYPOTHESIS
The hypothesis that stress (as a phenomenon distinct from energy availability) disrupts the reproductive system in athletes was inspired by observations of elevated cortisol levels in amenorrheic athletes like those in animals forced to exercise, and by the demonstration that the GnRH pulse generator can be disrupted by activation of the hypothalamic–pituitary–adrenal (HPA) axis (10). To be sure, the mildly elevated cortisol levels in amenorrheic and regularly menstruating athletes may be interpreted as evidence that the HPA axis was activated by stress, but they also may be interpreted as evidence that the axis was activated by energy deficiency, because cortisol is a glucoregulatory hormone. Indeed, women athletes display an entire spectrum of metabolic hormone and substrate abnormalities, as well as a low metabolic rate, all of which are signs of energy deficiency (7).
Several yrs ago, we recognized that all previous experimental investigations of the influence of the so-called “activity stress paradigm” on reproductive function in animals and humans had confounded the stress of exercise with both the stress of the means used to force animals to exercise (e.g., electroshock or threat of drowning), with the energy cost of exercise (e.g., training animals to run farther and farther for smaller and smaller food rewards), or both. The first challenge to the stress hypothesis came from the experimental demonstration that glucose supplementation suppresses the cortisol response to exercise (11). So we investigated the independent effects of exercise stress and energy availability on LH pulsatility in women (9).
We defined and controlled energy availability as dietary energy intake minus exercise energy expenditure, and independently defined and controlled exercise stress as everything associated with exercise except its energy cost (Fig. 1). We admitted only regularly menstruating women into the experiment, supervised their daily exercise treatments on a treadmill in the laboratory, and fed them a commercial liquid dietary product to control their energy intake. Because the pulsatile rhythm of LH changes during the menstrual cycle, we studied all women in the same mid-follicular phase of each woman’s menstrual cycle. Because the pituitary gland normally secretes pulses of LH approximately once hourly during this part of the follicular phase, accurately determining the frequency and amplitude of LH pulsatility requires LH to be measured in the blood at 10-min intervals for 24 h. So, after administering treatments to a subject for 4 d, we admitted her to the research unit of a hospital and drew 145 blood samples at 10-min intervals for 24 h through a venous catheter line inserted into her arm. To control for differences in LH pulsatility between women, each woman repeated this whole process twice, once with balanced and once with low energy availability, in random order, and we compared the results to determine the effect of energy availability. To determine the effect of exercise stress, we compared results in two groups of women, one receiving exercise treatments and one not.
We found that low energy availability disrupted LH pulsatility and that exercise stress did not. LH pulsatility was suppressed regardless of whether energy availability was reduced by dietary restriction alone or by exercise energy expenditure alone. Supplementing the diet to replace the energy cost of exercise prevented the disruption of LH pulsatility in exercising women.
Since then, amenorrhea has been induced in monkeys by training them to run voluntarily on a motorized treadmill for longer and longer periods while their food intake remained constant. When the diet of half of the monkeys was then supplemented, their menstrual cycles were restored without any moderation of their exercise regimen (15).
Further data undermining the stress hypothesis have also been reported in a study of young male soldiers participating in the 8-wk U.S. Army Ranger training course (3). This course is divided into four 2-wk phases in forest, desert, mountain, and swamp environments, during which trainees undergo daily military skill training, 8- to 12-km patrols carrying 32-kg rucksacks, sleep deprivation (∼3.6 h of sleep nightly), and dietary intakes during alternate wk of ∼2000 and ∼5000 kcal·d−1. During the course, trainees lost ∼12 kg of body weight. Blood sampling at the end of each wk revealed that triiodothyronine (T3), insulin-like growth factor I (IGF-I), and testosterone levels fell ∼20%, ∼50%, and ∼70%, respectively, during wk on diets of 2000 kcal·d−1 and returned to normal initial levels during alternate wk on diets of 5000 kcal·d−1, despite continued exposure to all other training stresses.
Thus, exercise appears to have no deleterious effect on reproductive function apart from the impact of its energy cost on energy availability. If the HPA axis disrupts the GnRH pulse generator in athletes, it probably does so by mediating the influence of energy availability.
SUCCESS OF THE ENERGY AVAILABILITY HYPOTHESIS
Mammals partition dietary energy among five major metabolic activities (cellular maintenance, thermoregulation, locomotion, growth, and reproduction) and the use of energy for one function, such as locomotion, makes it unavailable for others, such as reproduction (13,14). The energy availability hypothesis holds that failure to provide sufficient metabolic fuels to meet the energy requirements of the brain causes an alteration in brain function that disrupts the GnRH pulse generator, through an as yet unknown mechanism.
Considerable observational data from biological field trials support the hypothesis that mammalian reproductive function depends on energy availability, with the dependence operating principally in females. Experimental animal data also support this hypothesis. For example, investigators have induced anestrus in Syrian hamsters by restricting food intake, by pharmaceutically blocking carbohydrate and fat metabolism, by administering insulin to shunt metabolic fuels into storage, by cold exposure (which consumes metabolic fuels in shivering and nonshivering thermogenesis), and by increasing physical activity (13,14).
Of course, the energy availability hypothesis also is consistent with metabolic and endocrine signs of energy deficiency in athletes. Only one experiment to date successfully induced menstrual disorders in regularly menstruating women. That experiment (1), which imposed a high volume of aerobic exercise abruptly, caused a large prevalence of luteal suppression and anovulation in the first month and an even larger prevalence in the second. Disorders were more prevalent in a subgroup fed a controlled weight-loss diet than in another subgroup fed for weight maintenance, but even the weight maintenance subgroup may have been underfed, because body mass is an unreliable indicator of energy balance (6).
To extend our finding that the influence of exercise on LH pulsatility is via the impact of its energy cost on energy availability, we investigated the physiological dose-dependent effects of restricted energy availability on LH pulsatility in exercising women (Fig. 2) (8). We found that within 5 d, LH pulsatility is disrupted below a threshold of energy availability between 20 and 30 kcal·kg−1 lean body mass [LBM]·d−1 (Figs. 3 and 4). We also found that this disruption is substantially more extreme in women with short luteal phases (Fig. 5). If the latter finding is confirmed through further investigations, the screening of women for luteal length may be a convenient way to identify those who need to take extra care to avoid falling below the threshold of energy availability needed to maintain normal reproductive function.
The maintenance of normal LH pulsatility in this experiment despite a 33% restriction of energy availability to 30 kcal·kgLBM−1·d−1 demonstrated for the first time that LH pulsatility is not simply proportional to energy availability. Rather, the regulation of the reproductive system in women seems to be robust against decreases in energy availability as large as 33%. Because the exercise energy expenditure in this experiment was ∼840 kcal, many women may be able to maintain normal LH pulsatility while running up to 8 miles·d−1 as long as they do not simultaneously reduce their dietary energy intake below 45 kcal·kgLBM−1·d−1. If they do reduce their dietary energy intake, as many exercising women do, then they risk falling below the threshold of energy availability needed to maintain normal LH pulsatility.
More prolonged experiments are needed to confirm that such short-term effects of restricted energy availability on LH pulsatility in exercising women are predictive of chronic effects on ovarian function and to quantify the physiological dose-dependent effects of chronically low energy availability on ovarian function. Meanwhile, it is encouraging to note that monkeys maintained on a 30% restricted diet for 6 yrs showed no disruption of menstrual cycling or reproductive hormones and no reduction in bone mineral density (BMD), despite a fat mass 46% lower than that in monkeys fed a control diet (5). Furthermore, recent cross-sectional comparisons of estimated energy availability in athletes also support the notion that menstrual function is disrupted at the threshold of energy availability needed to maintain normal LH pulsatility. Amenorrheic athletes were estimated to habitually self-administer an energy availability of 16 kcal·kgLBM−1·d−1, whereas regularly menstruating athletes habitually self-administered 30 kcal·kgLBM−1·d−1 (12). Thus, although the precise location of the energy availability threshold between 20 and 30 kcal·kgLBM−1·d−1 remains to be determined, 30 kcal·kgLBM−1·d−1 appears to be sufficient energy availability to preserve normal reproductive function and skeletal health.
Evidence continues to accumulate to support the hypothesis that exercise has no suppressive effect on the reproductive system beyond the impact of its energy cost on energy availability. These results encourage the hope that physically active women may be able to prevent or reverse menstrual disorders by dietary supplementation without any moderation of their exercise regimen. More short-term experiments are needed to answer many questions about the mechanism by which low energy availability disrupts LH pulsatility. Further research also is needed to confirm that there are some women who experience more extreme disruptions of reproductive function than others. In addition, prolonged experiments are needed to confirm that short-term effects on LH pulsatility are predictive of chronic effects on ovarian function and to identify effective interventions for preventing and reversing menstrual disorders in athletes.
Supported in part by the U.S. Army Medical Research and Material Command (Defense Women’s Health & Military Medical Readiness Research Program; grant no DAMD 17–95–1–5053), and in part by the General Clinical Research Branch, Div. of Research Resources, NIH (grant no. M01 RR00034). The content of the information presented in this manuscript does not necessarily reflect the position or the policy of the government, and no official endorsement should be inferred.
1. Bullen, B. A., Skrinar, G. S. Beitins, I. Z. von Mering, G. Turnbull, B. A.and McArthur. J. W. Induction of menstrual disorders by strenuous exercise
in untrained women. N. Engl. J. Med. 312: 1349–1353, 1985.
2. Di Carlo, C., Palomba, S. De Fazio, M. Gianturco, M. Armallino, M.and Nappi. C. Hypogonadotropic hypogonadotropism in obese women after biliopancreatic diversion. Fertil. Steril. 72: 905–909, 1999.
3. Friedl, K. E., Moore, R. J. Hoyt, R. W. Marchitelli, L. J. Martinez-Lopez, L. E.and Askew. E. W. Endocrine markers of semistarvation in healthy lean men in a multistressor environment. J. Appl. Physiol. 88: 1820–1830, 2000.
4. Hilton, L. K., and Loucks. A. B. Low energy availability, not exercise stress
, suppresses the diurnal rhythm of leptin in healthy young women. Am. J. Physiol. Endocrinol. Metab. 278: E43–49, 2000.
5. Lane, M. A., Black, A. Handy, A. M. Shapses, S. A. Tilmont, E. M. Kiefer, T. L. Ingram, D. K.and Roth. G. S. Energy restriction does not alter bone mineral metabolism or reproductive cycling and hormones in female rhesus monkeys. J. Nutr. 131: 820–827, 2001.
6. Leibel, R. L., Rosenbaum, M.and Hirsch. J. Changes in energy expenditure resulting from altered body weight. N. Engl. J. Med. 332: 621–628, 1995.
7. Loucks, A. B. Physical health of the female athlete: observations, effects, and causes of reproductive disorders. Can. J. Appl. Physiol. 26( suppl): S176–S185, 2001.
8. Loucks, A. B., and Thuma. J. R. Luteinizing hormone pulsatility is disrupted at a threshold of energy availability in regularly menstruating women. J. Clin. Endocrinol. Metab. 88: 297–311, 2003.
9. Loucks, A. B., Verdun, M.and Heath. E. M. Low energy availability, not stress
, alters LH pulsatility
in exercising women. J. Appl. Physiol. 84: 37–46, 1998.
10. Rivier, C., and Rivest. S. Effect of stress
on the activity of the hypothalamic-pituitary-gonadal axis: peripheral and central mechanisms. Biol. Reprod. 45: 523–532, 1991.
11. Tabata, I., Ogita, F. Miyachi, M.and Shibayama. H. Effect of low blood glucose on plasm CRF, ACTH, and cortisol during prolonged physical exercise
. J. Appl. Physiol. 71: 1807–1812, 1991.
12. Thong, F. S., McLean, C.and Graham. T. E. Plasma leptin in female athletes: relationship with body fat, reproductive, nutritional, and endocrine factors. J. Appl. Physiol. 88: 2037–2044, 2000.
13. Wade, G. N., and Schneider. J. E. Metabolic fuels and reproduction in female mammals. Neurosci. Biobehav. Rev. 16: 235–272, 1992.
14. Wade, G. N., Schneider, J. E.and Li. H. Y. Control of fertility by metabolic cues. Am. J. Physiol. 270: ( 1 Pt1) E1–19, 1996.
15. Williams, N. I., Helmreich, D. L. Parfitt, D. B. Caston-Balderrama, A. L.and Cameron. J. L. Evidence for a causal role of low energy availability in the induction of menstrual cycle disturbances during strenuous exercise
training. J. Clin. Endocrinol. Metab. 86: 5184–5193, 2001.