Changes in Bone Turnover Induced by Aerobic and Anaerobic Exercise in Young Males


  • Part of this work was presented at the 18th Annual Meeting of the American Society for Bone and Mineral Research, Seattle, WA, U.S.A., September 1996.


Physical activity is considered an important factor in attaining bone mass. However, the mechanisms by which exercise affects bone metabolism are not completely understood. The present study was performed to investigate the effects of aerobic and anaerobic exercise on bone turnover. Twenty healthy young males (aged 20–29 years) were followed through an 8-week program of aerobic (n = 10) and anaerobic training (n = 10). Ten age-matched individuals served as controls. Serum bone-specific alkaline phosphatase (BAP), serum osteocalcin (OC), and urinary pyridinoline (Pyd) and deoxypyridinoline (Dpd) were determined as indices of bone metabolism. After 4 weeks of aerobic training, serum BAP and OC (p < 0.01), and urinary Pyd (p < 0.001) and Dpd (p < 0.01) were significantly reduced. After 8 weeks, BAP and OC levels had returned to baseline values, whereas the urinary cross-link excretion remained low. In the anaerobic training group, elevated levels of BAP (p < 0.05 vs. week 4), OC (p < 0.05 vs. week 4), and Pyd (p < 0.01 vs. week 0) were observed after 8 weeks of exercise. Changes in urinary Pyd and Dpd (week 0 vs. week 8) were positively correlated with changes in the mean power level in the Wingate test, a parameter of the anaerobic performance capacity (r = 0.50 and r = 0.55, p < 0.01, respectively). In the controls, no significant changes in biochemical markers were observed. We conclude that aerobic and anaerobic training excert different effects on bone metabolism. While aerobic training led to changes compatible with reduced bone resorption activity, anaerobic training seems to result in an overall accelerated bone turnover. Therefore, the impact of physical activity on bone turnover may depend on the kind of exercise performed.


It is well known that physical activity is a major determinant of bone mass.(1–3) However, the mechanisms by which exercise leads to changes in bone metabolism are not fully understood. In particular, little is known about the changes in bone metabolism induced by various forms of systematic exercise. Specific biochemical markers of bone turnover allow a good estimate of bone metabolic processes and have been established as useful parameters in assessing changes in bone turnover.(4) For example, accelerated bone turnover has been found to be associated with pronounced bone loss(5) and to predict fracture risk in older populations.(6) Also, markers of bone turnover have been helpful in assessing the effect of physical activity on bone turnover.

The bone-derived isoenzyme of alkaline phosphatase (BAP) represents a specific measure of osteoblastic enzyme activity and of bone formation.(7) Osteocalcin (OC) is a noncollagenous protein that is synthesized by osteoblasts and incorporated into the bone matrix.(8) Both markers are regarded as sensitive indices of bone formation.(4) The urinary excretion of the 3-hydroxy-hydroxypyridinium cross-links of type I collagen, pyridinoline (Pyd) and deoxypyridinoline (Dpd), has been established as a valid measure of bone resorption.(4) Both cross-link components are specific of mature skeletal collagens and contribute to the stability of the extracellular matrix of bone. The compounds are released upon collagen degradation (i.e., bone resorption) and are excreted into the urine where they can be measured by high-performance liquid chromatography (HPLC) or enzyme immunoassays.(9,10)

There is strong evidence that the hormonal and metabolic adaption of bone turnover to physical activity depends on age,(11,12) gender,(13) and the type of exercise performed.(12–15) However, very few studies have systemically investigated the metabolic mechanism by which bone responds to exercise. In some studies, measurements of bone turnover have been restricted to the first 48 h of exercise(16–19) or to a cross-sectional approach.(14,20) Since changes in bone mass represent the result of slow-acting metabolic processes, the present investigation was carried out using a longitudinal approach with a longer time period of observation. However, none of the published controlled studies using similar time protocols(11,12,21,22) examined the effect of different types of exercise on bone turnover.

To determine the differential effects of aerobic and anaerobic exercise on bone turnover, markers of bone formation and of bone resorption were measured in 30 healthy subjects during an 8-week course of controlled training, allowing a direct comparison of bone turnover among different types of physical activity.



Twenty-two healthy, young men between 20 and 30 years of age were recruited for the training program.(23) Also, 12 male controls matched for age, body mass index (BMI), and degree of prestudy physical exercise were included as controls. Subjects had not been on an exercise program (>1 h/week) for the previous 6 months. None of the apparantly healthy volunteers had a history of metabolic bone disease, and no drugs were taken that were known to interfere with bone turnover. All were nonsmokers and had no history of abnormal alcohol intake. Weight and height of standardized measures were taken in light clothing without shoes. Obesity was estimated calculating the BMI (kg/m2). Standardized physical examination, as well as collection of serum samples and spot urine, were performed on week 0 (before onset of the training program), as well as 4 and 8 weeks after enrollment into aerobic or anaerobic training programs. As inclusion criteria, subjects were requested to consume no alcohol and to participate in no other recreational activities (including other exercise programs) during the study period. Because of injury or noncompliance, four participants had to be excluded from the study. Thirty subjects completed the study and were included in the statistical analyses (10 subjects in each group). The study protocol was approved by the Ethics Committee of the University of Heidelberg, and written consent was obtained from all participants prior to inclusion.


A randomized, single-blinded study was performed. In a first step, participants were subdivided into three groups. During the following 8 weeks, subjects took part in either an aerobic (group I) or an anaerobic (group II) training program or served as controls (control group). The control group did not train during the study. All other participants met at the University Activity Center and exercised 3 days/week, 60 minutes/session under supervision at the same time of day. Prior to and after the 8-week exercise training, a standardized Wingate test(24) was performed to determine the individual mean (Pmean) and maximum (Pmax) anaerobic power. To evaluate the individual anaerobic threshold (IAT), the maximum oxygen uptake (VO2max), and the maximum running velocity (MRV), a stepwise increasing spiroergometry to exhaustion on a treadmill at a constant slope of 1.5% was performed.

In group I, subjects started with a standardized warm-up procedure followed by endurance running of 40–60 minutes according to a specific aerobic training concept at a heart rate corresponding to 60–85% of the VO2max. Each of the running units caused an increase in lactate ranging from 1.4 to 2.9 mM. In group II, pretraining exercise was performed according to a standardized warm-up procedure. Two of the three weekly anaerobic sessions consisted of sprints at 90–100% of maximum speed (5 × 80–300 m) with 5–8 minutes of passive recovery between each interval. At the end of each training unit, lactate measurements ranged between 14 and 18 mM. During the third weekly session, a 60-minute specific weight-lifting program for increasing leg strength and sprint ability was performed (lactate values ranged between 6 and 8 mM after each session).

Blood samples were collected after 24 h of rest between 9:00 a.m. and 12:00 a.m. on day 0 (before onset of the training program) and after 4 and 8 weeks. Following centrifugation for 10 minutes at 1000g within 2 h of sample collection, serum was separated and aliquots were kept at −80°C. A 12-h urine sample was collected during the subsequent night between 8:00 p.m. and 8:00 a.m. Urine specimens were kept at −30°C without additive until analysis.

Biochemical measurements

A solid-phase, two-site immunoradiometric assay (Tandem-R Ostase; Hybritech, San Diego, CA, U.S.A.) was employed for the quantitation of BAP. Two monoclonal mouse-anti-BAP antibodies directed toward different sites of the BAP molecule were used in a solid-phase sandwich format as described previously.(7) BAP purified from human SAOS-2 osteosarcoma cells was used as a standard. The cross-reactivity with the circulating liver isoenzyme is ∼16%. Intra- and interassay variations ranged from 3.7–6.7% and 7.0–8.1%, respectively, with a detection limit of 2.0 μg/l and normal values of 12.4 ± 4.4 μg/l in healthy males and 11.6 ± 4.1 μg/l in premenopausal females.

Table Table 1.. Population Characteristics at Baseline (Week 0)
original image

Human intact serum OC was detected with a competitive luminescence immunoassay (LUMItest® Osteocalcin; BRAHMS Diagnostica, Berlin, Germany). In this assay, OC in the sample and luminogen-marked tracer compete for the highly specific monoclonal antibody against the 37–49 sequence of human OC, fixed on the reaction tubes. Coefficients of variation were <10% for intra-assay and <15% for interassay variance. Reference values ranged from 4–12 μg/l in healthy males and premenopausal females, with a detection limit of 1.8 μg/l.

Total Pyd and Dpd were measured by HPLC as described earlier.(9) In brief, 500 μl of urine was subjected to acid hydrolysis at 107°C for 16 h to convert all cross-link components into the peptide-free form. Following partition chromatography on a CF1 cellulose column, pyridinium cross-links in samples and external standards were separated by reverse-phase ion-paired HPLC, and the eluting cross-link compounds were quantitated by fluorometry. Standard was prepared from sheep bone and was a generous gift from Dr. Simon Robins, Aberdeen, Scotland. The overall reproducibility of the assay was 8–12%. Urinary creatinine was measured in a Beckmann II creatinine analyzer utilizing the Jaffe Rate technique with alkaline picrate. Since we found no statistically significant differences between cross-link values expressed relative to the creatinine excretion compared with the total cross-link excretion in a 12-h interval, all results are given as nanomoles of cross-link excretion per millimole of creatinine (nM/mM Cr). Reference values ranges from 5.9–72.6 nM of Pyd/mM of Cr, and 2.6–18.5 nM of Dpd/mM of Cr in males and in premenopausal females.

Statistical analysis

The Statistical Analysis System (SAS) software package was used for data analysis (SAS Institute, Cary, NC, U.S.A.). Since biomarker results of the training groups were normally distributed at baseline (which was tested using the Chi-square statistic), paired Student's t-tests were performed to test for group differences. The Wilcoxon's rank-sum test was used for nonparametric variables. To correct for multiple testing, the significance level was adjusted according to the Bonferroni correction (= significance level/number of tests). Simple Pearson correlations were performed to determine the strength of association between two parameters. All statistical analyses were two-tailed, and a significance level of p < 0.05 was considered statistically significant.

Table Table 2.. Parameters of Physical Fitness
original image
Table Table 3.. Markers of Bone Turnover
original image


Study population

Anthropometric data, routine laboratory results (Table 1), and baseline parameters of fitness (Table 2) did not differ between the two training groups. All baseline laboratory values, including biochemical markers of bone turnover and of calcium homeostasis, were within the normal, age-adjusted reference range and equally distributed.

A control group was included to account for potential changes in bone turnover unrelated to physical exercise, such as seasonal effects. Age, weight, height, and BMI of the control group did not differ from the corresponding values of the two training groups (Table 1). Baseline values for BAP, OC, Pyd, and Dpd of the control subjects did not differ significantly from the values of the training groups at week 0. Biochemical markers of bone turnover did not change significantly during the 8-week follow-up (Table 3).

Parameters of physical fitness

In both training groups, the aerobic performance capacity was increased by 10% after 8 weeks of training (p < 0.05 vs. week 0), as determined by the IAT during incremental treadmill exercise. In group I, no siginificant changes were found in the anaerobic performance capacity as assessed by the mean power level (Pmean) in the Wingate test (Table 2). In contrast, in group II the anaerobic capacity was siginificantly elevated after 8 weeks of anaerobic exercise (p < 0.05) (Table 2).

Biochemical markers of bone turnover in the aerobic training group

As shown in Table 3 (absolute values) and Fig. 1 (percentage change), 4 weeks of aerobic training resulted in a significant reduction of serum BAP and OC levels (p < 0.01, respectively). After 8 weeks of aerobic training, serum values of both bone formation markers returned to baseline levels and at that point in time were significantly higher than after 4 weeks (p < 0.01). The urinary excretion of the collagen cross-links Pyd and Dpd, markers of bone resorption, was significantly lower after 4 weeks than at baseline (p < 0.001 for Pyd, p < 0.01 for Dpd), and after 8 weeks of aerobic training, both cross-link components remained at a significantly reduced level (p < 0.01 for Pyd, p < 0.05 for Dpd vs. week 0).

Biochemical markers of bone turnover in the anaerobic training group

Compared with baseline levels, no significant changes were observed after 4 weeks of training for any of the markers. In contrast, serum BAP (p < 0.05 vs. week 4), serum OC (p < 0.05 vs. week 4), and urinary Pyd (p < 0.01 vs. week 0) were all significantly elevated after 8 weeks of anaerobic training (Table 3 and Fig. 2).

Figure FIG. 1..

Markers of bone turnover in the aerobic training group. The upper panel shows the results for BAP and OC as markers of bone formation. The lower panel shows the results for Pyd and Dpd as markers of bone resorption. Results are given as percentage change from baseline values. Data are presented as box-and-whisker plots. The horizontal line in the box represents the group median and error bars the 10th and 90th percentile of individual values. Baseline values at week 0 are represented by the dotted line at 0%.ap < 0.05 vs. week 0;bp < 0.01 vs. week 0;cp < 0.001 vs. week 0;dp < 0.05 vs. week 4;ep < 0.01 vs. week 4.

Biochemical markers of bone turnover comparison of aerobic and anaerobic training

After 4 weeks, serum OC levels were significantly higher in the anaerobic than in the aerobic training group (p < 0.05). After 8 weeks, the urinary excretion of Pyd was significantly higher in the anaerobic than in the aerobic training group (p < 0.01) (Table 3).

Figure FIG. 2..

Markers of bone turnover in the anaerobic training group. The upper panel shows the results for BAP and OC as markers of bone formation. The lower panel shows the results for Pyd and Dpd as markers of bone resorption. Results are given as percentage change from baseline values. Data are presented as box-and-whisker plots. The horizontal line in the box represents the group median. Error bars represent the 10th and 90th percentile of individual values. Baseline values at week 0 are represented by the dotted line at 0%.ap < 0.05 vs. week 0;bp < 0.01 vs. week 0;cp < 0.001 vs. week 0;dp < 0.05 vs. week 4;ep < 0.01 vs. week 4.

Correlations between biochemical markers of bone turnover

Linear correlations between biomarkers of bone turnover are shown in Table 4. In both training groups, highest correlations were found between the urinary levels of Pyd and Dpd. All of the tested correlations were statistically significant, including those between markers of bone resorption and of bone formation. Overall, correlations between bone biomarkers were higher in the aerobic than in the anaerobic training group.

Table Table 4.. Pearson Correlations Between Markers of Bone Metabolism
original image

Correlations between parameters of fitness and biochemical markers of bone turnover

No significant correlations between markers of physical fitness and of bone turnover were observed at baseline (week 0). After 8 weeks of aerobic training, changes in urinary Pyd and Dpd (week 0 vs. week 8) were significantly correlated with changes in the MRV (r = −0.35, p < 0.05 and r = −0.48, p < 0.01, respectively). In the anaerobic training group, significant associations were found between changes in urinary Pyd and Dpd (week 8 vs. week 0) and changes in the mean power level (Pmean), an indicator of anaerobic performance capacity (r = 0.50, p < 0.01 and r = 0.55, p < 0.01, respectively). In both groups, changes in biochemical markers of bone formation were unrelated to changes in parameters of fitness.


The basic mechanisms carrying the effects of exercise on bone mass are not yet fully understood. Bone mineral density (BMD) is a static measure of bone turnover, and because changes in bone mass usually occur at a slow rate, measurement of BMD is inadequate to detect acute changes in bone metabolism as a result of physical activity. In recent years, the development of specific biochemical markers has enabled the detection of changes of bone turnover in response to biological (e.g., hormonal), biomechanical, and other variables. Moreover, recent studies in older populations suggest that bone turnover as determined by biochemical markers predicts future bone loss and fracture risk independently from bone mass measurements.(5,6)

We and others have shown that regular physical exercise leads to changes in bone turnover compatible with decreased bone resorption and increased bone formation.(20) These findings are in keeping with a number of reports indicating a bone conserving effect of moderate but regular physical activity.(1–3,25) As demonstrated in previous animal experiments, this effect may be mediated through a higher activity of individual osteoblasts, while the total number of bone remodeling sites is reduced.(26)

In contrast, excess exercise has been shown to potentially exert negative effects on bone metabolism and bone mass. For example, extensive long distance running leads to reduced BMD in both genders.(27,28) Premenopausal ultramarathon runners are often amenorrheoic due to the frequent development of estrogen deficiency. Because estrogen depletion is known to be one of the major pathogenetic factors for postmenopausal osteoporosis, it is conceivable that in female long distance runners the same mechanism may be responsible for the observed bone loss.(28) However, it is not clear why osteopenia also occurs in male ultramarathon runners.

In the present long-term longitudinal study, we investigated different types of physical activity in order to obtain a better understanding of the mechanisms by which exercise leads to changes in bone turnover. To our knowledge, this is the first study using a detailed protocol which allows a direct comparison between heavy aerobic and anaerobic training effects on bone metabolism. As the major study outcome, we found a clear difference in markers of bone metabolism with regard to the type of physical activity performed. Thus, an 8-week course of aerobic training led to reduced bone resorption activity, whereas anaerobic training resulted in an overall acceleration of bone turnover. Slightly unexpectedly, this relatively small amount of anaerobic training seems to be able to override the effect of continual aerobic exercise that forms a large part of everyday physical activity.

Four weeks after continuous aerobic training, we found significantly reduced serum levels of BAP and OC, indicating reduced bone formation. At the same time, a significant reduction in the urinary excretion of pyridinium cross-links was observed, a finding compatible with a state of low bone resorption. Taken together, these observations indicate that in young adult male individuals, aerobic exercise is associated with an overall suppression of bone turnover after a relatively short time period of physical activity. A somewhat different pattern was reported in a recent study by Eliakim et al.,(12) who found a reduction in the urinary excretion of N-terminal telopeptide cross-links as an index of bone resorption, but, in contrast to our findings, showed an increase of bone formation markers after 5 weeks of endurance-type training. However, their study population consisted of adolescent males, indicating that in this age group adaption to physical activity follows somewhat more distinct mechanisms than in young adults.

Interestingly, a short-term study by Welsh et al.(19) reported a significant increase in bone resorption markers (without changes in bone formation markers) 32 h after physical load. Since both study population and biomarker measurements were similar to our study, these results indicate that the acute changes in bone metabolism may well differ from the long-term effects of continous physical exercise. Because bone metabolism requires some time to adapt to changes in physical load, the changes in cross-link excretion observed after a period of 32 h may reflect effects such as renal handling, acute oxygen deficiency, or extracellular fluid shifts rather than changes in bone turnover. Nishiyama et al.(17) studied the baseline and postexercise levels of OC immediately and 1 h after 30 minutes of treadmill exercise in athletic and nonathletic young male subjects. It was shown that in both groups serum OC levels increased significantly after the exercise; however, changes were seen earlier in untrained participants. These findings again indicate that such pointed changes in markers of bone turnover may be attributed to processes of acute body adaptation rather than representing the long-term effect of physical training on bone turnover.

In the present study, markers of bone formation returned to baseline levels after 8 weeks of aerobic training, whereas markers of bone resorption remained significantly lower than baseline. In a recent study by Franck et al.,(21) who studied serum OC levels in a group of healthy young individuals performing an 8-week program comparable to the present aerobic training concept, changes in circulating OC levels were almost identical to the pattern found in the present investigation. In some studies, exercise training led to increases in serum OC levels after periods ranging from 12–20 weeks.(11,22) These observations suggest that long-term aerobic exercise is associated with decelerated bone resorption and normal to elevated bone formation. Our results are in good keeping with the findings of an earlier cross-sectional population-based study including a large number of healthy subjects aged 50–81 years. In this study, the rate of bone resorption was significantly lower in individuals on regular physical activity than in subjects with a sedentary lifestyle.(20) In fact, and as pointed out before, a number of reports have shown a significant increase in bone mass with moderate but regular physical activity.(11,29–31) Barengolts et al.(26) showed that in 9-month-old rats, regular exercise resulted in the formation of fewer bone-remodeling sites, together with a higher activity of individual osteoblasts. This observation is in good keeping with the human data and may represent a physiological mechanism for the attainment of bone mass under increased physical stress.

In contrast to aerobic exercise, 8 weeks of anaerobic training resulted in a substantial acceleration of bone metabolism, as indicated by a significant increase in markers of both bone formation and bone resorption. Moreover, changes in resorption markers were positively associated with the anaerobic performance capacity. Although there are no data available on the effect of long-term anaerobic training on bone metabolism, the changes observed in our study may be compatible with acute skeletal responses as described by Welsh et al.(19) and Nishiyama et al.(17) (see above). Thus, under the conditions of prolonged oxygen deficiency, anaerobic physical activity may lead to a subtle imbalance in bone remodeling (“uncoupling”). This notion is further supported by the fact that correlations between biochemical markers were weaker in the anaerobic than in the aerobic training group.

However, the possibility is not excluded that anaerobic training might have a net positive effect on bone formation, too. This conclusion is supported by the notion that median OC levels increased by 45.8% on week 8 compared with week 0, while at the same time the pyridinium cross-links were elevated by 14.1% (Dpd) to 23.1% (Pyd), only. However, the change in OC levels was not statistically significant versus week 0, due to the large variation of individual values.

In conclusion, aerobic and anaerobic training appear to exert different long-term effects on bone metabolism. The changes in bone turnover after aerobic training are compatible with a net increase in bone formation by virtue of decreased bone resorption. In contrast, anaerobic training seems to be associated with an overall increase in bone turnover with a less pronounced net increase in bone formation. However, results should be viewed within the framework of a small sample size and should be followed up by further studies in larger populations.


We are indebted to Mrs. Beatrice Auler for excellent technical assistance and to Hybritech, Inc. and BRAHMS Diagnostica for providing the immunoassay kits.