Original Article
European Journal of Applied Physiology
Accepted: 12 June 2001 / Published online: 25 August 2001

Influence of prolonged continuous exercise on hormone responses to subsequent exercise in humans

Atko M. Viru^1,, Anthony C. Hackney², Erli V?lja¹, Kalle Karelson¹, Tamara Janson¹ and Mehis Viru¹

(1)	Institute of Exercise Biology, University of Tartu, 18 Ylikooli, Tartu 51014, Estonia
(2)	Endocrine Section, Applied Physiology Laboratory, University of North Carolina, Chapel Hill, NC, USA

Abstract. This study examined the possibility that fatigue may modify the hormone responses to exercise. A group of 12 endurance trained athletes ran for 2 h (blood lactate concentrations of approximately 2 mmol?l^-1) in order to induce fatigue. The subjects exercised for 10 min at 70% maximal oxygen uptake before (1st test) and after (2nd test) the 2 h run to Breitling Navitimer Replica assess hormone responsiveness. A 1 min anaerobic power test was performed to assess muscle power. Cortisol, growth hormone, testosterone and insulin concentrations were determined before and after the 1st and 2nd tests. The 1st test resulted in increases in concentrations (P<0.05) of cortisol and growth hormone, a decrease in insulin concentration (P<0.01) and no change in testosterone concentration. The 2 h run caused decreases of insulin, increases of growth hormone concentration and variable responses in the concentrations of cortisol and testosterone. The 2nd test decreased insulin concentration further (P<0.05), but responses of the concentrations of testosterone, growth hormone and cortisol were variable. In 6 subjects (group A) cortisol displayed an increase [mean (SD)] from baseline concentrations [+304.0 (60.0) nmol?l^-1], while in the other 6 subjects (group B) a decrease or no change was seen [+3.1 (5.3) nmol?l^-1 ,between groups, P<0.05]. Growth hormone concentration was substantially higher in group A [+14.7 (4.8) ng?ml^-1] than group B [+6.0 (2.9) ng?ml^-1] following the 2nd test. In group A anaerobic muscle power was higher, while in group B it was lower, after the 2 h run than before the 2 h run (P<0.05). The findings suggest that fatigue from prolonged endurance activity may introduce a resetting in the pituitary-adrenocortical component of the endocrine system, expressed either by intensified or by suppressed endocrine functions.

Keywords. Cortisol - Fatigue - Growth hormone - Insulin - Testosterone

Introduction

Numerous studies report that prolonged, high intensity, exercise causes significant increases in concentrations of circulating pituitary and adrenal hormones (Dessypris et al. 1976; Fry et al. 1998; Hackney et al. 1995; Schmid et al. 1982; Sutton and Lazarus 1976; Viru et al. 1992). Yet, while reported in many articles, such findings are not universal. Rivoire et al. (1953) found prolonged (1.5-2.5 h) intense exercise decreased the excretion of urinary 17-hydroxycorticoids. These findings were later confirmed in other human studies involving prolonged exercise (an initial increase in blood corticosteroid concentration was followed by decreases; Staehelin et al. 1955; Viru et al. 1973). Animal experiments also demonstrated similar findings; that is, blood concentrations of glucocorticoids increased after short-term and decreased after prolonged exercise lasting several hours (K?rge et al. 1974; Viru and ?kke 1969). Interestingly, an administration of corticotropin (ATCH) eliminated the decrease in the activity of the adrenocortical system during the exercise (Viru and ?kke 1969). Thus, the reason for the subnormal activity of the adrenocortical system did not appear to be exhaustion of the adrenocortical cells (Viru and ?kke 1969). These above findings were dismissed as primarily aberrations of the less specific methods for hormone analysis employed at the time of the studies (i.e. colorometric and fluorometric assays, K?rge et al. 1974; Staehelin et al. 1955; Viru and ?kke 1969; Viru et al. 1973), especially after the advent of the highly specific and precise immunoassay methods.

Nevertheless, more contemporary research also supports these early studies. Dessypris et al. (1976) reported extremely low cortisol concentration in a marathon runner who collapsed after 15 km of running. Similarly, during a regatta a male rower collapsed. A day later abnormally low cortisol concentrations were found in this subject (Urhausen et al. 1987). Feldmann et al. (1992) found that during a 6 day ski race, the elevation in concentration response to daily skiing in ACTH, cortisol and bgr -endorphin disappeared on day 3; and the hormones remained unchanged or reverted to pre-race concentrations over the remaining race days. Middle-long distance runners during a 10 day stage of high volume training showed increases in cortisol concentration in response to exercise sessions in the first days, but later the hormone concentrations were suppressed (Viru et al. 1988). Additionally in an early stage of the overtraining syndrome, despite increased ACTH release after administration of corticotropin-releasing hormone, the cortisol response was blunted (Lehmann et al. 1998). Low cortisol responses to repeated standardized exercise tests were also found in overtrained athletes (Fry et al. 1998).

These lowered or suppressed hormone responses to subsequent exercise cannot be attributed to a beneficial training effect or adaptation. Some investigators have speculated these represent a mal-adaptation or pathology in the athletes (Hackney and Styers 1999; Lehmann et al. 1998). But a possibility exists that fatigue induced by the prior exercise may have modified the hormone responses to subsequent exercise. The purpose of this study was to examine this hypothesis. To test this possibility, continuous prolonged aerobic exercise of 2 h duration was used as a means of inducing fatigue. A standardized test of short-term submaximal exercise was performed before and after the 2 h prolonged exercise to assess the influence of fatigue upon hormone responsiveness. The hormones examined were cortisol, growth hormone, testosterone and insulin.

Methods

Subjects

A group of 12 endurance-trained male athletes (7 cross-country skiers, 2 rowers, a middle-distance runner, a swimmer and an orienteer) gave informed consent to participate. Investigative procedures were approved by the Ethics Committee of the Medical Faculty of the University of Tartu. The mean (SD) age of the subjects was 20.7 (1.9) years, height 184.8 (6.2) cm and body mass 76.4 (6.4) kg. The maximal oxygen uptake ( $\dot V{\rm O}_{{\rm 2max}}$ ) of the subjects ranged from 48.2 to 76.3 ml?kg^-1?min^-1. All participants trained regularly 5-6 times a week year round. The experiments, however, took place in the pre-season preparatory phase of their training. All subjects were healthy and displayed no symptoms of acute or chronic diseases; they were all nonsmokers and were not consumers of alcohol or drugs of any kind.

Experiment design

The experiments were performed in the afternoon, at the subject's usual time for their training sessions (1300-1800 hours). Subjects did not eat for at least 2.5 h before coming to the laboratory, and were asked to avoid serious physical exercises or mental efforts during the day before and the day of the experiments.

The overall design of the study involved the subjects performing an exercise test (1st test, see below), followed by a 2 h continuous run, followed by another exercise test (2nd test), all within a single day (see Fig. 1).

Fig. 1. A diagram showing the design of the experiment. The 10 min exercise tests were performed at 70% of the individual's maximal oxygen uptake while the 1 min anaerobic test was performed on a cycle ergometer at the highest power output maintainable. The 2 h run was performed outdoors or on a treadmill (at an intensity that would yield a blood lactate concentration of approximately 2 mmol?l^-1). BS Blood sampling

On the day of the experiments the subjects arrived at the laboratory and rested quietly for 30 min. After this rest the first blood sample (1BS) was obtained. Then 5 min later the 1st test session began; this consisted of: 10 min of cycling at 70% $\dot V{\rm O}_{{\rm 2max}}$ , followed by a second blood sample [2BS, immediately-post (within 0-3 min of stopping exercise)], 10 min of rest, followed by a 1 min anaerobic exercise test. The subjects were then given another 10 min rest. They next ran continuously for 2 h at a velocity which would have corresponded to a blood lactate concentration of approximately 2 mmol?l^-1. At the end of the 2 h run a third blood sample (3BS) was taken, then a 5 min rest was given. Following this rest, the 2nd test session was performed, as with the 1st test it consisted of: 10 min of cycling at 70% $\dot V{\rm O}_{{\rm 2max}}$ , followed by a fourth blood sample [4BS, immediate-post (within 0-3 min of stopping exercise)], 10 min of rest, followed by a 1 min anaerobic exercise test.

The 10 min cycling ergometry at 70% $\dot V{\rm O}_{{\rm 2max}}$ exercise in the 1st and 2nd test sessions was used as a challenge to induce hormone responses from the endocrine system. The 2 h run exercise served as a fatiguing element to invoke perturbations within the endocrine system. The 1 min anaerobic exercise test in the 1st and 2ndtest sessions was used to assess muscle capacity.

Subjects evaluated their rating of perceived exertion using the Borg's 10 point rating scale (Borg 1982) after the 10 min 70% $\dot V{\rm O}_{{\rm 2max}}$ exercise in the 1st and 2nd test sessions. In order to avoid dehydration effects, water consumption was allowed ad libitum throughout the experiment.

Of the subjects 8 performed their 2 h run outdoors in a nearby park while the other 4 chose to run indoors on a treadmill (weather conditions inside and outside were similar).

Exercise-testing equipment

In order to calculate individual percentage of $\dot V{\rm O}_{{\rm 2max}}$ for the 1st and 2nd test sessions, a $\dot V{\rm O}_{{\rm 2max}}$ test was performed approximately 1 week prior to the main experiment. This test used a standard clinical increment protocol using a cycle ergometer (Monark, Sweden). The oxygen uptake was monitored using a metabolic measurement system (ParvoMedics TrueMax 2400, USA).

The 10 min 70% $\dot V{\rm O}_{{\rm 2max}}$ exercise involved cycle ergometry (Monark) at a pedalling rate of 70 rpm. The oxygen uptake was monitored with the equipment noted above.

The anaerobic exercise test used in the 1st and 2nd test sessions was a 1 min Szogy and Cherebetiu (1974) cycle ergometry protocol. The average power output was calculated for each 15 s period and for the entire 1 min exercise period.

Biochemical analysis procedures

Blood samples were drawn from the anticubital area of the arm using a venipuncture procedure and vacutainer tubes. Serum was separated immediately using centrifugation at 4?C. The sera were frozen and stored at -20?C until later analysis. Cortisol, growth hormone and testosterone concentrations were determined using an automated immuno-chemoluminescence system, Immunolite (Babson et al. 1991). Detection limit, intra-assay variation, and inter-assay variation were for the cortisol assay 5.5 nmol?l^-1, 7.0% and 10.3%, respectively, for the growth hormone assay 0.05 ng?ml^-1, 6.0% and 5.1%, respectively, for the testosterone assay 0.5 nmol?l^-1, 6.7% and 7.7%, respectively. Insulin was determined by radioimmunoassay (ICN Micromedic system, USA).

Statistical analyses

Initially, a Mann-Whitney U-test was used to compare all responses of the subjects who performed their 2 h continuous run outdoors with those who ran inside on the treadmill. Analysis results indicated that the site of the 2 h run was not a significant factor in data variability. Therefore the responses of all the subjects were pooled and analysed collectively.

A MANOVA for repeated measurements was used to analyse hormone data for a main effect of time using the criterion of Cohen (1988). Subsequent to the MANOVA, mean hormone changes were evaluated using the paired Student's t-test (Bonnferoni correction applied; i.e. 1BS vs 2BS vs 3BS vs 4BS, etc.).

Where appropriate, Pearson product-moment correlation coefficients were computed and non-parametric statistics applied (Mann-Whitney U-test, Kruskal-Wallis test). In all the statistics analyses the accepted probability level was P0.05.

Results

Hormone changes. Fig. 2 displays mean hormone responses for all blood sampling times (1BS-4BS). The MANOVA analysis revealed significant main effect for time (P<0.05). Specifically, for the 1st test 10 min exercise (2BS vs 1BS), increases in growth hormone (P=0.05) and cortisol (P=0.01) concentrations and a decrease in insulin concentration (P=0.03) were found. Testosterone responses were variable and showed a non significant change (2BS vs 1BS, P=0.40).

Fig. 2. The pattern of the responses in cortisol, growth hormone, testosterone and insulin concentrations as influenced by the two 10 min exercise tests and the 2 h run. Mean and SD data are indicated. *Significant responses to the 1st and 2nd 10 min exercise tests (P<0.05), significant responses to the 2 h run (P<0.05)

After the 2 h run insulin concentration remained decreased (3BS vs 1BS, P<0.01), and growth hormone concentration was elevated further in all subjects (3BS vs 1BS, P<0.05, see Fig. 2). An increase in testosterone concentration occurred (3BS vs 1BS, P=0.04). The cortisol responses were variable and no further significant increase in concentration was observed [3BS vs 1BS, P=0.06 (NS)].

The 10 min exercise performed after the 2 h run produced different results compared to those of the first (i.e. 2nd test, see Fig. 2). Insulin concentration decreased to an even lower level in all subjects (4BS vs 3BS, 2BS, 1BS, P<0.01). But, no further significant changes in the cortisol, testosterone and growth hormone concentrations were detected. Analysis of individual data from the 2nd test showed that significant mean responses were not established in these three hormones because in some subjects hormone concentrations increased further and in others they decreased. An increase in cortisol concentration appeared in 6 subjects, of testosterone in 5 subjects, and of growth hormone in 5 subjects. The concentrations of cortisol decreased in 4 subjects, of testosterone in 6 subjects, and of growth hormone in 1 subject. Individual hormone changes were considered significant when they exceeded the following intra-individual variations: for cortisol ?25 nmol?l^-1, for testosterone ?1.0 nmol?l^-1, for growth hormone ?2.0 ng?ml^-1, and for insulin ?3.0 ?U?ml^-1(Viru et al. 1992). Interestingly, only 3 subjects exhibited simultaneous increases in concentrations of cortisol, testosterone and growth hormone.

Correlation analysis failed to find significant relationships within the changes among the individual hormones across sampling times. The only significant correlation between the hormones was for the change in cortisol and growth hormone concentrations; r=0.581, P=0.047).

In an attempt to explore the data further, the subjects were divided into two groups according to the cortisol responses to the 2nd 10 min exercise test (the above criteria for individual change were applied). Group A consisted of subjects with cortisol concentrations higher than after the 1st 10 min exercise test (n=6). Group B consisted of subjects (n=6) whose cortisol concentrations did not change or decreased in comparison with their value after the 1st 10 min test. In group A the mean (SD) cortisol concentration was higher after the 2nd 10 min exercise test (4BS) by 214 (46) mmol?l^-1 (P=0.005) than after the 1st 10 min exercise test (2BS) (see Fig. 3). The high cortisol concentrations seemed to be related to the increase of cortisol both in response to the prolonged 2 h run and to the 2nd 10 min exercise test. Group B failed to show a significant increase in cortisol concentration after the prolonged 2 h run (3BS vs 2BS) or after the 2nd 10 min exercise test (4BS vs 2BS, 3BS,Fig. 3). In fact, the cortisol concentration of group B after the 2nd 10 min exercise test was lower than after the 1st test [4BS vs 2BS, -97 (25) mmol?l^-1, P=0.02]. Additionally, in group A after the 2nd test 10 min exercise, the cortisol concentration was higher than the initial resting level (4BS vs 1BS) [+304 (60) nmol?l^-1, P=0.004]. Whereas in the group B this difference was not found (+3.1 (5.3) nmol?l^-1, P=0.580).

Fig. 3. The pattern of responses in cortisol concentrations during the experiment in two groups of subject distinguished by the cortisol concentration after the 2nd 10 min exercise test which followed the 2 h run (group A - cortisol level higher, group B - cortisol level lower). *Significant (P<0.05) difference between groups A and B, #significant (P<0.05) difference between cortisol concentrations after 1st and 2nd 10 min test

Growth hormone concentration increased during the prolonged 2 h run in groups A and B (Fig. 4), but the increase in group A [3BS, +11.9 (4.4) ng?ml^-1] was twice as great as in group B [+4.7 (0.7) ng?ml^-1]. The high cortisol concentrations of group A at the end of the 2 h run appeared to be associated with a more significant growth hormone response to the 2nd 10 min exercise test. However for group B, who had the smaller cortisol response, no significant growth hormone response was found (4BS vs 3BS). After the 2nd 10 min test (4BS), however, growth hormone concentrations were higher than initially (1BS) in both groups [+14.7 (4.8) ng?ml^-1, P=0.027 in group A, and +6.0 (2.9) ng?ml^-1 in group B, P=0.092 (NS)].

Fig. 4. The patterns of response in growth hormone and testosterone concentrations during the experiment in the two group of subjects distinguished by the cortisol concentration after the 2nd 10 min exercise test (group A - cortisol level higher, group B - cortisol level lower). *Significant (P<0.05) difference between cortisol concentrations after 1st and 2nd 10 min test

Most testosterone response comparisons were non-significant between groups A and B (Fig. 4), although a significant testosterone response was found within group B after the 1st 10 min exercise test (2BS vs 1BS) [+1.5 (0.5) nmol?l^-1, P=0.036]. Also, in group B, the testosterone concentration after the 2nd 10 min exercise test was higher than the initial level (4BS vs 1BS) [+4.2 1.6 nmol?l^-1, P=0.049]. Group A displayed no significant changes.

Groups A and B showed the same type of insulin responses (significant decreases in concentration) and no significant differences in magnitude of these changes existed between the groups (data are not shown).

The training backgrounds and $\dot V{\rm O}_{{\rm 2max}}$ of groups A and B were also compared. There were no significant differences between the groups for either of these parameters.

Anaerobic power

Table 1 reports the average 1 min power output during the maximal anaerobic test. In the overall sample (n=12) the power did not change significantly from the 1st to the 2nd test. However, when the data were analysed using the cortisol groupings, differences were found to exist. Power in group A was higher after the 2 h run (2nd test) than before it (1st test, P<0.05). In group B, however, the 2 h run decreased the mean power output during the 2nd test (2nd test vs 1st test, P<0.05). The highest power output in any 15 s period changed variably and no significant differences in responses were found (data not shown).

Table 1. Mean (SD) power output and ratings of perceived exertion (Borg 1982) in 1 min anaerobic test. Both Mann-Whitney test (P=0.014) and Kruskal-Wallis one-way ANOVA (P=0.033) indicate significant differences in the distribution of individual changes in average power output between group A and group B

	Average power output (W) during 1 min anaerobic test			Rating of perceived exertion (Borg scale) after 10 min test exercise
	Before 2 h run	After 2 h run	Change	Before 2 h run	After 2 h run	Change
Pooled (n=12)	465 (55)	467 (65)	+2	4.2 (0.8)	8.6 (1.0)	+4.4
Group A (n=6)	458 (56)	477 (66)	+19	4.5 (0.8)	9.0 (1.3)	+4.5
Group B (n=6)	477 (60)	453 (71)	-24	3.8 (0.8)	8.2 (0.4)	+4.4

Perceived exertion

All subjects (n=12) evaluated the 2nd 10 min exercise test with higher ratings of perceived excretion than during the 1st 10 min exercise test [8.6 (1.0) vs 4.2 (0.8) Borg scale units, P<0.001, see Table 1). Responses for groups A and B were similar.

Discussion

The results indicated that 2 h of prolonged, fatiguing exercise alters the hormone responses to short-term submaximal exercise. When examining the overall (n=12) mean hormone response, before the prolonged 2 h run the 10 min 70% $\dot V{\rm O}_{{\rm 2max}}$ exercise test resulted in pronounced increases in concentration of cortisol and growth hormone, decreases in concentration of insulin and no change in testosterone. The same 10 min exercise test after the fatiguing 2 h exercise invoked somewhat different responses. No further increases in cortisol and growth hormone concentrations appeared, the decline of insulin became more pronounced, and again no significant testosterone responses occurred.

Previous studies substantiate our hormone findings for the 1st 10 min exercise test (Fry et al. 1998; Sutton and Lazarus, 1976). Furthermore, many studies have also reported comparable changes in concentration of growth hormone (Sutton and Lazarus 1976; Viru et al. 1992), cortisol (Dessypris et al. 1976; Hackney et al. 1995; Keul et al. 1981), testosterone (Dessypris et al. 1976; Galbo et al. 1977; Kindermann and Schmitt 1985) and insulin (Pruett 1970; Viru et al. 1992) in response to prolonged exercise (2 h or more), although, these latter research findings are not universal (Gabriel et al. 1992; Hackney 1996). To our knowledge, however, the lack of significant change in the overall mean response (n=12) for cortisol, growth hormone and testosterone concentrations after the 2nd 10 min exercise test are new findings.

The lack of an overall response to the 2nd 10 min exercise test did not mean that there was a lack of changes in individuals. In some subjects the hormone concentrations increased, in others they decreased or did not change. We chose to explore these individual trends within the data. We used the criteria of Viru et al. (1992) to classify our subjects into sub-groups. One group responded (group A, increases), as might characteristically be expected, to the 2nd 10 min exercise test, while the other group was uncharacteristic in its responses (group B, decrease or no change). Cortisol was chosen as the classification hormone due to its links to various "fatigue mechanism" hypotheses (Lehmann et al. 1998; Newsholme et al. 1992; Viru 1975). The classification of the subjects into these two groupings revealed previously unnoticed elements within the data for the responses after the 2 h run and the 2nd test session (10 min and 1 min anaerobic exercises).

For group B, cortisol concentration failed to increase after the 2 h run or the 2nd 10 min exercise test. The group A response was opposite. It appeared that the elevated cortisol concentrations immediately before the 2 h run and the second 10 min exercise test in group A did not provoke a feedback suppression of the cortisol responses as demonstrated by Brandenberger et al. (1984). Growth hormone findings were remarkably similar in this respect. We feel these findings demonstrate a resetting of the regulation of pituitary-adrenocortical activity. The fatigue produced by the 2 h run revealed the existence of two different types of resetting. One involved an intensified mobilization of pituitary-adrenocortical function while the other reflected the inhibition of activity within this system.

Earlier work from our laboratory revealed analogous findings but for different hormones. Specifically, the hormone responses to 2 h of exercise displayed these two proposed different types of resetting within pituitary function (Viru et al. 1981). In that study, several subjects had increases of ACTH, follitropin and lutropin concentrations, whereas other subjects exhibited significant decreases in concentration of both gonadotropic hormones and a blunted ACTH response to the exercise (Viru et al. 1981). Re-examination of the effect of 2 h of exercise revealed other comparable findings (Viru et al. 1992). Increases of ACTH and cortisol concentrations at the end of 2 h exercise were seen in some subjects; but, cortisol levels fell below initial resting values by the end of exercise in other subjects (Viru et al. 1992). We feel studies by other investigators may have similar findings, but by comparing overall mean responses within their data a masking of individual differences has occurred (i.e. subjects who are highly responsive negate those who are not).

Interestingly, the high cortisol concentrations at the end of the 2nd 10 min exercise test by group A was associated with increased performance in the 1 min anaerobic test, whereas in group B, subjects with attenuated cortisol responses, a decreased anaerobic performance was observed. This may signify that an intensified mobilization of pituitary-adrenocortical function is related to a strong nerve influence on the neuromuscular apparatus, which is essential for muscle performance. Work by Kjaer et al. (1987) supports this concept. They showed that in partly curarized subjects the stronger central motor command was associated with an increased response of ACTH, growth hormone and catecholamines during exercise.

A serious problem is how to quantify fatigue in prolonged aerobic exercise. Although the subjects announced that they felt a level of fatigue similar to that after their usual training sessions, we were not able to score the fatigue objectively. Therefore, a question remained as to whether the hypothesized resetting of pituitary-adrenocortical regulation was related to actual fatigue in exercise performance. The decreased power output in the 2nd 1 min anaerobic test in the subjects with the suppressed pituitary-adrenocortical function would suggest that the answer is perhaps yes for muscle power. The pronounced increase in rating of perceived exertion after the 2nd 10 min exercise demonstrated that overall body fatigue also existed. However, this was for all the subjects, as no between group differences existed for the ratings of perceived exertion. Thus, it may not be completely justified to speak about "exercise performance fatigue" only when a decline in muscle power output is evident. It is also possible that a latent fatigue develops and precedes a decreased muscle work capacity. Perhaps the expression of the latent fatigue is an intensified mobilization of the body's capacities to compensate for earlier fatigue manifestations allowing maintenance of the level of exercise quality, and postponing the decrease in working capacity (our observations for group A). This speculation will need further scientific study before a definitive answer can be reached.

The mechanism for the proposed resetting of pituitary-adrenocortical function in response to fatigue from prolonged exercise is uncertain. It is plausible that serotonergic neurons of the hippocampus are involved. Probably, hippocampal serotonergic neurons inhibit the activity of the hypothalamic neurosecretory cells producing corticotropin-releasing hormone concomitantly with the development of fatigue. Thereby, during prolonged exercise the activity of pituitary-adrenocortical systems decreases. The release of serotonin in the hippocampus has been seen in exercising rats and associated with the development of exercise fatigue (Bailey et al. 1993; Meeusen et al. 1996). Thus, the proposed resetting of the regulation of pituitary-adrenocortical system could be related to the hypothesized serotonergic mechanisms of the central fatigue in humans (Newsholme et al. 1992). The inhibition resetting, resulting in this central inhibition of the pituitary-adrenocortical system, is perhaps a safety mechanism, i.e. an attempt to avoid a fatal exhaustion of the body's resources (Viru 1975). Why certain individuals can override this inhibition reset remains to be determined. However, our data indicate it is not due to the training background or the aerobic power of the individual. Further research is necessary to refute or confirm our conjecture on this serotonergic hypothesis linkage. Further research is necessary to clarify whether the changes in hormone responsiveness were related to dehydration. Although water intake was allowed ad libitum, the actual level of hydration was not controlled in our study. Psychological structure of subjects might also have influenced the results. Odagiri et al. (1996) reported that in a group of triathletes high postcompetition concentrations of corticotropin and bgr -endorphin were associated with high vigour regardless of high fatigue scores whereas in others corticotropin and bgr -endorphin concentrations as well as vigour scores were low.

In conclusion, collectively the findings suggest fatigue from prolonged endurance activity may introduce a resetting in the pituitary-adrenocortical component of the endocrine system. In some cases, the resetting results in an intensified mobilization of certain endocrine functions, and appears in conjunction with improved anaerobic muscle performance. However, in other cases, endocrine function is suppressed and anaerobic muscle performance is attenuated.

Acknowledgement. This study was supported by the Estonian Scientific Foundation, grant 3962.