Dr. Eser worked as the Australian consultant for Stratec Medizintechnik, Pforzheim, Germany. All other authors state that they have no conflicts of interest.
Skeletal Benefits After Long-Term Retirement in Former Elite Female Gymnasts†
Article first published online: 21 DEC 2009
Copyright © 2009 American Society for Bone and Mineral Research
Journal of Bone and Mineral Research
Volume 24, Issue 12, pages 1981–1988, December 2009
How to Cite
Eser, P., Hill, B., Ducher, G. and Bass, S. (2009), Skeletal Benefits After Long-Term Retirement in Former Elite Female Gymnasts. J Bone Miner Res, 24: 1981–1988. doi: 10.1359/jbmr.090521
Published online on May 18, 2009;
- Issue published online: 21 DEC 2009
- Article first published online: 21 DEC 2009
- Manuscript Accepted: 15 MAY 2009
- Manuscript Revised: 12 JAN 2009
- Manuscript Received: 27 JUL 2008
- retired athletes;
- bone geometry
Bone strength benefits after long-term retirement from elite gymnastics in terms of bone geometry and volumetric BMD were studied by comparing retired female gymnasts to moderately active age-matched women. In a cross-sectional study, 30 retired female gymnasts were compared with 30 age-matched moderately active controls. Bone geometric and densitometric parameters were measured by pQCT at the distal epiphyses and shafts of the tibia, femur, radius, and humerus. Muscle cross-sectional areas were assessed from the shaft scans. Independent t-tests were conducted on bone and muscle variables to detect differences between the two groups. The gymnasts had retired for a mean of 6.1 ± 0.4 yr and were engaged in ≤2 h of exercise per week since retirement. At the radial and humeral shafts, cortical cross-sectional area (CSA), total CSA, BMC, and strength strain index (SSIpol) were significantly greater (13–38%, p ≤ 0.01) in the retired gymnasts; likewise, BMC and total CSA were significantly greater at the distal radius (22–25%, p ≤ 0.0001). In the lower limbs, total CSA and BMC at the femur and tibia shaft were greater by 8–11%, and trabecular BMD and BMC were only greater at the tibia (7–8%). Muscle CSA at the forearm and upper arm was greater by 15–17.6% (p ≤ 0.001) but was not different at the upper and lower leg. Past gymnastics training is associated with greater bone mass and bone size in women 6 yr after retirement. Skeletal benefits were site specific, with greater geometric adaptations (greater bone size) in the upper compared with the lower limbs.
It has been well established that exercise during growth is associated with site-specific greater bone strength.(1–4) However, only if skeletal benefits can be maintained into adulthood and older age can exercise during growth be postulated to prevent or alleviate osteoporosis. A number of studies have investigated the effects of retirement from gymnastics in particular(5,6) and other sports in general(7–9) on bone strength and have found that some bone loss may occur but that retired athletes still have greater bone strength than sedentary control subjects even after many years of retirement from sport. Whereas all of the above studies used a cross-sectional study design (with the exception of Reference 5, where some subjects were followed up), similar results were found in longitudinal studies. Two studies by Kontulainen et al.(10,11) showed that almost the entire gain achieved at the end of the subjects' active racquet sport career was maintained 4–5 yr after reduced training. Furthermore, a study using a 7-mo jumping intervention in (pre)pubertal children found some residual bone strength benefits at the hip (1.4%) still present 8 yr after termination of intervention.(12) Most of the bone mass gain achieved through ice hockey playing in young males was lost after cessation of training; however, bone mass at the humerus was still higher in formerly active players than in controls.(13)
In most of the existing studies that have investigated potential bone strength surpluses in retired athletes, bone properties have been quantified using DXA (except for the study by Haapasalo et al.(9)). DXA results (BMC and aBMD) are affected by several systematic and nonsystematic errors: (1) body size of the subject(14); (2) rotation of the measured bone; (3) fat/lean ratio of the surrounding soft tissue(15); and (4) inhomogeneities of the surrounding soft tissue.(16) When athletes are compared with sedentary individuals or when they are followed up beyond retirement, differences in soft tissue composition are the rule rather than the exception. Furthermore, QCT and pQCT studies have shown that the skeletal benefits from exercise during growth are of geometric rather than densitometric nature at the bone shafts,(3,9) meaning that bone girth and cortical cross-sectional area (CSA) are increased, leading to an increase in compressive, bending, and torsional strength of the bone. At the epiphyses, increases were found to be densitometric (e.g., trabecular BMD) rather than geometric.(17) The 2D nature of DXA does not allow an accurate separate assessment of bone geometry and true volumetric BMD.
The aims of this study were to determine (1) whether estimated bone strength would be greater in former elite gymnasts compared with normally active controls several years after their retirement; (2) the skeletal sites that benefited most (diaphyses or epiphyses, arms, or legs); (3) whether benefits at the epiphyses were mainly geometric or densitometric; and (4) whether the benefits are negatively related to time of retirement.
MATERIALS AND METHODS
A total of 60 premenopausal female subjects ≥18 yr of age were recruited for this study, consisting of 30 retired gymnasts and 30 age-matched controls. Inclusion criteria for the retired gymnasts were (1) participation in high-level, competitive gymnastics during growth (childhood and adolescence) for at least 4 yr, training for a minimum of 6 h/wk (2) retirement from the sport for at least 3 yr; and (3) participation in no more than 2 h/wk of regular physical activity since retirement. Inclusion criteria for the controls were participation in no more than 2 h/wk of regular physical activity during growth and adulthood. Furthermore, both groups were required to have no history of disease known to affect bone health and no recent long-term periods of bed rest or limb immobilization.
Subjects were recruited through Gymnastics Australia, staff and students of Deakin University, and word of mouth. The Deakin University Human Research Ethics Committee for the Faculty of Health, Medicine, Nursing and Behavioural Sciences approved the study, and written consent was obtained from all participants.
Body weight was measured on balance scales to the nearest 0.05 kg in light clothing and without footwear. Standing height was measured using a stadiometer to the nearest 0.1 cm. Participants were asked about dominance of hand, and consequently, the limbs to be measured were determined as the nondominant arm and the contralateral leg (except if a fracture had occurred within the last 5 yr; then the contralateral limb was used). Bone length of the tibia and ulna were measured using an anthropometric tape measure to the nearest 0.5 cm. Ulnar bone length was measured from the ulnar styloid process to the olecranon. Ulnar length was used for both radial and humeral length, because these two bones are of proportional length in the same individual.(18) Tibial length was measured from the distal edge of the medial malleolus to the medial joint cleft of the knee. Tibial length was also used in place of femoral length because of poor accessibility of the femur and high correlation between the length of these two bones.(18)
Data on menstrual history including age at menarche and any menstrual irregularities, use of contraceptives, fracture history, and past and present activity status were assessed by questionnaire. In gymnasts, further questions included age of onset of gymnastics training, interruptions to their gymnastics career, intensity and duration of gymnastics training (number of sessions/hours of training completed per week, and level of competition), and age of retirement from gymnastics training.
All measurements were performed with a Stratec XCT 3000 scanner (Stratec Medical, Pforzheim, Germany). This pQCT apparatus measures attenuation of X-rays that are linearly transformed into hydroxylapatit (HA) densities. Unlike some other pQCT scanners, the Stratec XCT 3000 is calibrated with respect to water that is set at 60 mg hydroxylapatit (HA), so that fat results in 0 mg HA.(19) HA equivalent densities are automatically calculated from the attenuation coefficients by using the manufacturer's phantom, which itself is calibrated with respect to the European Forearm Phantom (EFP; QRM, Erlangen, Germany).(19) Based on the radiation dose measurements performed at the Charles Gardiner Hospital by R. Prince, we calculated the effective dose to be 0.2 μSv per scan and per scout view.
Scout views of the distal end of the tibia, femur, radius, and humerus were performed. At the distal tibia, the reference line was placed on the distal end of the lateral half of the tibia, at the distal femur on the distal end of the lateral condyle. At the radius, the automatic placement provided by the manufacturer was used, which places the reference line at the medial distal endplate, and at the humerus, the reference line was placed at the distal end of the medial epicondyle.
At the tibia and the radius, scans were performed at 4% and 66% of the bone's total length measured from the reference line. At the femur, scans were placed at 4% and 25%, and at the humerus at 25% only. Slice thickness was 2.3 mm, and voxel size was set at 0.5 mm with a scanning speed of 20 mm/s. For the femur, voxel size was set at 0.3 mm with a scanning speed of 10 mm/s.
Bone parameters measured by pQCT
Epiphyseal scans (4%):
The periosteal surface of each bone's epiphysis was found by a contour algorithm based on thresholding at 180 mg/cm3. BMC per centimeter of slice thickness, CSA, and total BMD (BMDtot) was determined. Concentric pixel layers were peeled off from the bone's perimeter until a central area covering 45% of the total bone CSA was left. From this central area, trabecular BMD (BMDtrab) was determined.
Diaphyseal scans (25% and 66%):
The periosteal surface of the bone's diaphysis was found by a contour algorithm based on a threshold of 280 mg/cm3. BMC, total CSA (CSAtot), and the polar bone strength strain index (SSIpol)(20) were calculated. Cortical bone was selected with an inner and outer threshold of 710 mg/cm3. Of the selected area, cortical CSA (CSAcort) and cortical BMD (BMDcort) were calculated. Cortical thickness (THIcort) was calculated based on the assumption that bone shaft is cylindrical from CSAtot, which included the bone marrow, and CSAcort of the diaphyseal scans. In the same scans, medullary CSA was calculated by subtracting CSAcort from CSAtot. Subcutaneous fat CSA was determined by selecting the area with thresholds −40 to +40 mg/cm3 HA density (contour mode 3, peel mode 1), and muscle CSA was determined by subtracting the total bone CSA (threshold, 280 mg/cm3; contour mode 1, peel mode 2) and subcutaneous fat CSA from the total limb CSA (threshold, −40 mg/cm3; contour mode 3, peel mode1).
Total body fat mass and lean mass were determined by DXA (Prodigy; Lunar, Madison, WI, USA), with analysis software version enCORE Multimedia version 8.10.027. Positioning of patients for each scan was made according to standard procedures.
Independent t-tests were performed between retired gymnasts and controls for age, height, weight, muscle and fat CSA, and bone length. Independent t-tests were also performed for BMC, CSA, BMDtot, and BMDtrab at the epiphyses, as well as for BMC, CSAtot, CSAcort, BMDcort, THIcort, and SSIpol of the diaphyses, to detect site specific differences between retired gymnasts and controls. The effect size was calculated for each parameter by calculating the mean ± SD of both groups and dividing the mean difference between the two groups by the mean SD. An effect size of 1 means that the difference between the two groups is 1 SD. This allows comparing parameters with differrent distributions with each other. Fisher's exact test was used to compare oral contraceptive use between the retired gymnasts and controls, and a Mann-Whitney U-test was used to compare the sum of all periods of oral contraceptive pill use. Linear regression was used to determine the relationship between retirement duration and trabecular BMD and cortical CSA. These two bone parameters were chosen because they are likely to be quickly affected by decreased training or immobilization.(21) Linear regression was also used to explore the relationship between limb muscle CSA and epiphyseal BMC and diaphyseal cortical CSA of the corresponding bones for both groups separately. All statistical analyses were completed using SPSS for Windows (version 14.0). Statistical significance was set at 0.05.
Characteristics of the subjects
Retired gymnasts and controls were comparable with regard to age, height, weight, limb length, and total body lean and fat mass (Table 1). The age range of retired gymnasts was 18–36 yr, whereas for the controls it was 18–44 yr. The gymnasts had 15% and 18% greater muscle CSA at the lower and upper arm, respectively (p ≤ 0.01), and nonsignificant differences of 3% and 1% at the lower or upper leg, respectively (Table 1). Menarche occurred 2.9 years later in the retired gymnasts compared with controls (14.7 versus 11.8 respectively, p ≤ 0.001; Table 1). In addition to the prevailing primary amenorrhea in the retired gymnast group (eight versus only one in the control group), this group also included seven subjects with at least one episode of secondary amenorrhea. Total duration of this secondary amenorrhea was up to 0.5 yr in three subjects and between 0.5 and 1 yr in four subjects. In the control group, only two subjects had secondary amenorrhea of 1 and 3 yr. The prevalence and duration of oral contraceptive pill (OCP) use tended to be less in the gymnasts than controls (33%, 4.6 yr versus 63%, 6.0 yr, respectively, p ≤ 0.07; Table 2). The number of recent sessions of vigorous and moderate intensity exercise (i.e., in the past 2 wk) was similar between gymnasts and controls; however, the controls participated in a higher duration of moderate activity than the retired gymnasts (Table 2). The gymnasts began training at 6.1 yr (range, 3.0–9.6 yr), and they had trained for an average of 10.5 yr (range, 6–16 yr). At the highest level reached, the gymnasts trained an average of 23.0 h/wk (range, 9.0–40.0 h/wk). At the time of this study, they had been retired for an average of 6.1 yr (range, 3–18 yr).
pQCT scans were completed in all subjects at the distal epiphyses and shafts of the radius, humerus, tibia, and femur. The following pQCT scans were excluded because of movement artefact: 4% radius in one gymnast and one control, 66% radius in five gymnasts and five controls, 4% femur in four gymnasts, and 25% femur in four gymnasts and three controls. The following scans were excluded because of other reasons: 25% humerus in one control and 25% femur in one control. At the radial shaft, because of exclusion of shaft thicknesses <2 mm (to avoid partial volume effect), cortical BMD of only 13 retired gymnasts and 18 controls was used for data analysis. The DXA measurements were completed in 55 of the subjects (1 gymnast and 4 controls did not have a DXA measurement).
At the radial distal epiphysis, total CSA and BMC were 25% and 22% greater in the retired gymnasts than the controls, respectively (p ≤ 0.0001). There was a trend for greater trabecular BMD (9%) in the retired gymnasts (p ≤ 0.056; Table 3). Total BMD was not different between groups.
At the radial diaphysis, total, cortical and medullary CSA, BMC, and SSIpol were greater in the gymnasts than controls (13–56%, p ≤ 0.01; Table 3). Cortical BMD was slightly smaller in the retired gymnasts (p ≤ 0.002; Table 3; Fig. 1A).
At the tibial distal epiphysis, total CSA and total BMD were not significantly different between the two groups. BMC and trabecular BMD were 8% and 7% greater in the gymnasts than controls, respectively (p ≤ 0.05; Table 4; Fig. 1C).
At the tibial diaphysis, BMC, total CSA, cortical CSA, cortical thickness, and SSIpol were greater in the retired gymnasts compared with the controls (7–12%, p ≤ 0.05). In contrast, no difference was detected for medullary CSA or cortical BMD (Table 4; Fig. 1C).
At the femoral diaphysis, BMC, total CSA and medullary CSA, and SSIpol were all greater in the retired gymnasts than the controls (10–14%, p ≤ 0.05). There were no differences between groups for cortical CSA, cortical wall thickness, or cortical BMD.
At the epiphyses, the osteogenic effect of gymnastics on total CSA in the gymnasts was 1.1 SD greater at the radius than the tibia (p ≤ 0.0001). There was no difference between the radius and tibia for trabecular vBMD (Fig. 1).
At the diaphyses, the osteogenic effect on total CSA in the gymnasts was 0.3 and 0.5 SD greater at the arms than the legs (radius versus tibia and humerus versus femur, respectively, p ≤ 0.01). In contrast, the osteogenic benefit of gymnastics in cortical CSA did not differ between the radius and tibia (Fig. 1); however, cortical CSA was 0.6 SD greater at the humerus than the femur (p ≤ 0.0001).
Linear regressions between muscle and bone parameters are shown for the upper limb in Fig. 2. Significant linear relationships existed for both groups between lower arm muscle CSA and distal radial BMC (gymnasts: r2 = 0.401, p ≤ 0.001; controls: r2 = 0.582, p ≤ 0.0001; Fig. 2A) and radial shaft BMC (gymnasts: r2 = 0.421, p ≤ 0.0001; controls: r2 = 0.533, p ≤ 0.0001; Fig. 2B), as well as between upper arm muscle CSA and humeral shaft BMC (gymnasts: r2 = 0.563, p ≤ 0.0001; controls: r2 = 0.682, p ≤ 0.0001; Fig. 2C). Linear regression coefficients for the lower limb were generally lower. Here, significant relationships existed between lower leg muscle CSA and distal tibial BMC (gymnasts: r2 = 0.252, p ≤ 0.005; controls: r2 = 0.318, p ≤ 0.001) and tibial shaft BMC (gymnasts: r2 = 0.264, p ≤ 0.004; controls: r2 = 0.598, p ≤ 0.0001). Further significant relationships between upper leg muscle CSA and distal femoral BMC existed (gymnasts: r2 = 0.297, p ≤ 0.005; controls: r2 = 0.552, p ≤ 0.0001) and femoral shaft BMC (gymnasts: r2 = 0.243, p ≤ 0.01; controls: r2 = 0.359, p ≤ 0.001).
No significant linear relationships were found between duration of retirement and any of the bone or muscle parameters.
This study found significant bone geometric and densitometric benefits (in the order of 20%) at the arms of retired female gymnasts compared with controls. The benefits at the radius and humerus were of geometric nature, whereas densitometric properties did not differ. The distal radial epiphysis was enlarged to a similar degree (25%) than the radial and humeral shafts (32% and 20%, respectively). Relative differences between the two groups were much smaller in the lower limbs compared with the upper limbs (tibia and femur parameters were 5–12% greater in the retired gymnasts). Estimated bone strength surpluses were paralleled by surpluses in muscle CSA, which was found to be ∼15% and 18% greater at the lower and upper arms, respectively, of retired gymnasts but not different between groups at the legs. At the legs, the between-group difference was smaller in terms of muscle size and bone strength, possibly because physical activities that the controls pursued involved the legs more than the arms or because the osteogenic benefit of gymnastics was smaller at the legs than the arms even when the gymnasts were active.
There are no other existing studies that have assessed bone properties in retired gymnasts by means of pQCT. Some studies have investigated sustained estimated bone strength benefits in retired gymnasts by means of DXA. For instance, Kudlac et al.(22) found aBMD of proximal femur and total body but not lumbar spine significantly higher in female gymnasts who had retired for a mean of 4 yr compared with a control group. Female gymnasts who had been retired for 4 yr but remained physically active reported 7–17% greater aBMD than controls for total body and hip.(23) Uzunka et al.(8) found 6–15% greater aBMD than controls at all measured body sites in former female gymnasts who had retired 8 yr earlier. Similarly, Bass et al.(5) found sustained osteogenic benefits between 6% and 26% at the arms, legs, and spine in gymnasts who had retired for a mean duration of 8 yr, and this difference did not diminish with increasing duration since retirement. In former gymnasts who had been retired for 17 yr, Zanker et al.(6) reported 10–17% greater aBMD at the spine, hip, and total body compared with controls. Pollock et al.(24) studied former gymnasts at a mean retirement duration of 24 yr and found residual osteogenic benefits in aBMD only at the legs. However, aBMD measured by DXA is systematically affected by soft tissue composition, and the former gymnast group in the study of Pollock et al. had significantly less body fat than the control group, meaning that their aBMDs were prone to be underestimated.(15)
To date, only two studies have assessed estimated bone strength in active gymnasts by pQCT.(17,25) Similar to our results, Ward et al.(17) found the gymnasts' surpluses to be greater at the radius than the tibia, where most differences did not reach statistical significance. At the tibial and radial distal epiphyses, they only found a difference in BMD and not in CSA. Dyson et al.(25) found CSA at the distal radius to be greater in gymnasts than controls, although the difference was not statistically significant. In contrast, our study of retired gymnasts showed a 25% greater CSA of the radial distal epiphysis, supporting the postulation by Frost(26) that, at the epiphyses, increased mechanical loading has to be accommodated by an increase of joint surfaces, to protect the cartilage from damage caused by overload. It is possible that the gymnasts in the study of Ward et al. had not been training for a sufficient duration of time to show this difference. Furthermore, a confounding factor may have been the 11% lighter body weight (which was not adjusted for in the analysis) of the female gymnasts, which suggests a generally slimmer build of the gymnasts. Also the gymnasts in the study of Dyson et al. were shorter in stature and lighter than controls.(17,25) The results of our study suggest that the gymnasts had greater CSA at the distal radius at least by the end of their active sports career. However, detraining may have resulted in some decline in trabecular and total BMD but not below control levels. The reason for this may be the maintenance of a greater muscle mass at the arms in our retired gymnasts, despite the fact that they were no longer involved in any sports that require large arm muscle forces. It is noteworthy, however, that the gymnasts seem to have maintained more bone mass at the arms than what would be expected from their muscle volume (intercept for linear regressions in Fig. 2 are greater for gymnasts than controls). At the distal tibia and femur, our results support those by Ward et al.(17) in that epiphyseal CSA was not increased in retired gymnasts. Low body weight in active elite gymnasts may limit the muscle mass surplus over their normally active peers at the legs.
Our results agree with studies using pQCT in currently active athletes other than gymnasts who started their training during adolescence. They have unequivocally found bone adaptation to increased mechanical loading at the extremity bone shafts in the form of geometrical increases in cortical CSA and outer bone diameter, resulting in greater bone compressive, bending, and torsional strength.(9,27,28) Cortical BMD has mostly been found unchanged or slightly reduced(9,28) because of increased remodeling caused by microdamage.(26) Medullary CSA was found increased, decreased, or unchanged; these results are possibly dependent on the starting age of athletes.(2) Results on bone adaptations to exercise at the epiphyses are more controversial: often an increase in trabecular and total BMD was found(17) or an increase in CSA of the epiphysis.(3) Again, differences are likely to depend on starting age of athletes, with those starting before puberty being more likely to show adaptations in epiphyseal CSA,(26) and those starting after puberty being more likely to show adaptations in BMD.(3)
The strength of this study is that stringent selection criteria were applied in terms of the degree of retirement. We only included former elite female gymnasts who, after their retirement, participated in a maximum of 2 h of physical activity per week. None of the subjects in either the retired gymnast or control group participated in competitive sports at the time of the study, and there was little difference in present activity between the two groups. However, the slightly higher level of moderate activity in the control group may have reduced the relative differences between the two groups at the legs. Furthermore, the selection of several measuring sites at the arms and the legs gives a differentiated picture of the skeletal adaptations to high impact exercise and their preservation.
The limitations of the study are the cross-sectional rather than longitudinal study design, so that the reduction in bone strength since the time of retirement is not known. Bone geometry is likely to have changed at least at the shafts because former gymnasts often experience a period of catch-up growth after retirement.(29,30) Furthermore, the skewed distribution of duration of retirement with 20 subjects who had retired for a duration of 3–6 yr and only 10 subjects who had retired for a duration of 6–18 yr means that slow losses in estimated bone strength over long time periods may have been masked in this study. Last but not least, gymnasts differed from controls with regard to age at menarche, primary and secondary amenorrhea, recent moderate activity, arm muscle CSA, and leg fat CSA and tended to differ with regard to OCP use. However, these between-group differences would have reduced the difference in bone parameters between the two groups because they were of generally greater adverse effect on bone in the retired gymnast group. Performed analyses of covariance adjusting for these differences (results not shown) showed very similar results to the presented ANOVAs. Effects of OCP use are controversial(31,32); however, the adverse effects on bone by primary and secondary amenorrhea as well as late menarche, of which the retired gymnasts suffered more severely, have been well documented.(33–35)
In summary, at a mean of 6 yr after retirement from the sport, estimated bone strength in female former elite gymnasts was greater than in normally active controls, with the difference greater at the arms than at the legs. The skeletal adaptations to gymnastics were of geometric rather than densitometric nature. At the radius and humerus, bone cross-section was 20–32% greater in retired gymnasts compared with controls. Similarly, muscle cross-section at the lower and upper arm was 15% and 18% greater, respectively, in retired gymnasts compared with controls. Estimated bone strength benefits at the tibia and femur were relatively small 6 yr after retirement from gymnastics. No association in bone strength was found with increasing duration of retirement.
The authors thank all the retired gymnasts and control subjects who participated in this study. We greatly appreciated the help of Gymnastics Australia who provided names and addresses of retired gymnasts. The authors thank Hans Schiessl from Stratec Medizintechnik who contributed with many stimulating comments at the outset of the study.
- 32003 Effect of long-term impact-loading on mass, size, and estimated strength of humerus and radius of female racquet-sports players: A peripheral quantitative computed tomography study between young and old starters and controls. J Bone Miner Res 18:352–359., , , ,
- 112001 Good maintenance of exercise-induced bone gain with decreased training of female tennis and squash players: A prospective 5-year follow-up study of young and old starters and controls. J Bone Miner Res 16:195–201., , , , , , ,
- 201996 Noninvasive bone strength index as analyzed by peripheral quantitative computed tomography (pQCT). In: SchönauE (ed.) Paediatric Osteology: New Developments in Diagnostics and Therapy. Elsevier Science, Amsterdam, The Netherlands, 141–146., , ,
- 262004 The Utah Paradigm of Skeletal Physiology, 1st ed., vol. 1, Bone and Bones and Associated Problems. International Society of Musculoskeletal and Neuronal Interactions, Athens, Greece.