INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

POSITIVE EFFECT OF MECHANICAL LOADING on bone mineral density (BMD) in humans has been well documented by measurements with dual-energy photon absorptiometry (3, 18) and dual-energy X-ray absorptiometry (DEXA) (5, 10); however, the values do not reflect a volumetric density (g/cm³) but rather an areal BMD (g/cm²), calculated as a quotient of bone mineral content (BMC; g) to two-dimensional projected corresponding area. In therapeutic study using adult bone, areal BMD is believed to be a good substitute for volumetric BMD, as there is little change in bone size (15). On the other hand, in studies examining factors that might influence both growth and bone density, the validity of using areal BMD as a substitute for volumetric BMD may be questioned. For example, during growth, there is little change in volumetric BMD but an increase in bone size resulting in a substantial increase in areal BMD (13, 23).

Because it is well known that mechanical loading markedly alters the size and shape of the skeleton (6, 7), the question remains open whether exercise increases volumetric BMD in humans. Furthermore, previous mechanical loading studies in humans have devoted less attention to evaluation of important aspects of bone strength, ignoring potential changes in bone geometry and architecture. It has been reported that biomechanical parameters derived from the cross-sectional bone area would be a better indicator of bone strength at the forearm than is BMD (16).

Peripheral quantitative computerized tomography (pQCT) was recently developed to provide simultaneous information on geometric properties and volumetric density of appendicular bone (12). To characterize an adaptation of bone structure to mechanical loading, we evaluated the effect of long-term unilateral physical activity on volumetric density and geometric properties of playing-arm radius of tennis players compared with nonplaying radius.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Subjects. The tennis players' group consisted of 10 young adult women and 6 young adult men, of whom 14 were right-handed, and the remaining 2 were left-handed. All players used only a dominant hand for forehand stroke, and all but one used both hands for backhand stroke. The mean age of male and female players was 20 yr. They were clinically healthy, and none of them had any diseases, was taking medication affecting bone metabolism, or had incidence of upper extremity fractures. All subjects had a history of competitive playing over a 3-yr period. They trained 5-6 times/wk, and the duration of each session was 180 min, ranging from 150 to 240 min.

The control group comprised age-matched seven healthy women and five men, who had not been involved in physical activity affecting the dominant or nondominant extremity only. The group characteristics are given in Table 1. Written informed consent was obtained before the study.

Bone mineral measurements. Bone parameters in two sites of radius of both arms were measured by using a pQCT scanner (XCT 960, Stratec, Pforzheim, Germany) with a single energy X-ray source, according to a method previously described (21). All computed tomography scans had a slice thickness of 2.5 mm and a voxel size of 0.59 mm. The measurement sites of radius were the proximal site (midradius), a site at 20% of the length of the bone from the distal end, and the distal site (distal radius), a site at 4% of the length of the bone from the distal end. A distal measuring site, which typically contains 70% of trabecular bone, is used especially for the examination of trabecular bone, and a diaphyseal site, which contains more than 90% of cortical bone, is used for cortical bone.

BMC was defined as the mineral content of the bone within a 1-mm slice (mg/mm). Periosteal bone area is the cross-sectional area of the bone after the soft tissue has been peeled off, cortical bone area is the region with linear attenuation coefficient >0.93 (27), and endocortical bone area was defined by the difference between periosteal bone area and cortical bone area. Cortical thickness was defined as the mean distance between the inner and outer edge of the cortical shell. From the periostal surface of distal radius, 55% of the total bone area was peeled away, and the remaining 45% of inner region was considered to be purely trabecular.

Strength strain index (SSI) lies within the theory of stability of mechanical structures against bending or torsion. The axial and polar SSI was calculated by where A is area of voxel (mm²), r is its distance from the center of gravity, CD is cortical density (mg/mm³) and is divided by normal physiological density of cortical bone (1,200 mg/mm³), and r_max is a maximum distance of a voxel from the center of gravity.

Coefficients of variation for the triplicate measurements on three human subjects after repositioning were 0.19-1.41% for BMC, 0.10-0.72% for volumetric BMD, 0.44-0.74% for bone area, 0.79% for cortical thickness, and 1.07-2.02% for SSI.

Statistical analysis. The statistical analysis was performed by using the StatView program, and data are shown as means ± SE. At first, the relative side-to-side differences between men and women were compared by using the Student's nonpaired t-test. Because there was no significant gender effects on the relative side-to-side differences for all measurements, the data for men and women were pooled. The side-to-side comparison of the dominant and nondominant radius was performed with the matched paired t-test. Comparisons of nondominant radius and relative side-to-side differences between players and controls were performed by using the Student's nonpaired t-test. Pearson's coefficient of correlation was used to determine the relationship between the bone parameters.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The characteristics of the subjects are given in Table 1. There were no significant differences in age, height, and weight between the players and the controls, except that the female tennis players had greater weight than female controls. The starting age of the unilateral training of the players was 12.8 yr for men and 11.6 yr for women, and the unilateral training history was 7.3 yr for men and 8.5 yr for women.

Midradius. Players exhibited an increase in total BMC, periosteal bone area (cross-sectional bone area), cortical BMC, and cortical bone area in the playing arm compared with the nonplaying arm. On the other hand, volumetric density of the total bone and the cortical bone was lower in playing arm than in the nonplaying arm, although the decrease reached statistical significance only at cortical bone. Cortical thickness and the endocortical bone area were also larger in the playing arm. The axial and polar SSIs of the midradius were clearly and significantly higher in the playing arm than in the nonplaying arm (Table 2).

In controls, significant side-to-side differences were also found in total BMC, cortical BMC, cortical area, and x-axis of SSI. These relative side-to-side differences in control subjects were significantly less than those observed in players. Decrease in total and cortical BMD and an increase in cortical thickness in the dominant radius was not observed in the control subjects. In the nondominant distal radius, no significant differences between controls and players were found in any measured parameters.

Distal radius. The total BMC of the playing arm was greater than that measured for the nonplaying arm in all cases for the total bone and in all but one case for the trabecular bone in tennis players. Periosteal bone area, total BMD, trabecular bone area, and trabecular BMD of the playing arm were greater than that measured for the nonplaying arm (Table 3). In controls, significant side-to-side differences were not found in any measured parameters. In the nondominant distal radius, no significant differences between controls and players were found in any measured parameters.

Correlation. Among relative side-to-side differences, midradius periosteal bone area positively correlated with total and cortical BMC but negatively correlated with total and cortical BMD at midradius. Similar correlations among relative side-to-side differences in periosteal bone area, BMC, and BMD were also observed in distal radius, where trabecular bone was assessed instead of cortical bone. The age when the subjects started to play correlated negatively with the mid and distal periosteal bone area (Table 4).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

To clarify the effect of long-term exercise on the bone in the absence of the confounding effects of genetic, hormonal, and nutritional factors, side-to-side difference of unilateral player's radius was examined by using pQCT. pQCT allows a volumetric density measurement and has the ability to separately assess cortical and trabecular bone compartments (12). In this study, distal and diaphyseal sites of the radius were used, especially for the examination of trabecular and cortical bone, respectively.

As for cortical bone in midradius, long-term, intensive tennis playing led to a slight but statistically significant decrease in the vBMD of the playing arm when compared with the contralateral arm. This finding was quite surprising, since the previous densitometric analysis using single photon absorptiometry or DEXA reported that BMD in midradius, which contains more than 90% of cortical bone, increased with exercise in most studies (1, 5, 6, 17). However, previous results may be artificial because three-dimensional property of the bone is not taken into account when BMD is expressed as areal density (g/cm²) in these technologies. Lu et al. (13) reported that cortical areal BMD increases during growth because bone size increases. Furthermore, Bhudhikanok et al. (2) reported that ethnic and gender differences in areal BMD are largely, but not entirely, explained by differences in bone size. Our findings support the suggestion of a previous study (24) that results of areal BMD by DEXA should be interpreted with caution when there is a disparity in bone size among groups being compared.

There is no clear explanation for the decreased cortical volumetric BMD with tennis playing. However, we observed a significant negative correlation between the relative side-to-side difference in periosteal bone area and cortical BMD of midradius in this study. Therefore, a possible explanation may be that midradius bone size increased at the expense of bone density. In this regard, Kleerekoper et al. (11) suggested that an increase in fractures during the peripubertal growth spurt might be correlated with a temporary decrease in bone density, because the increase in periosteal and endosteal apposition is partly offset by an increase in cortical porosity. Furthermore, similar phenomenon has been observed in insulin growth factor I-treated growing rats, in which insulin growth factor I treatment stimulated bone growth more than did mineral deposition, resulting in decreased bone density (20).

Cordey et al. (4) pointed out that in midradius, in which the proportion of the cortical shell is substantial, a consideration of the area and geometric properties of the cortical shell are of greatest importance for the prediction of fracture risk. Previous studies (6, 7) using a portable X-ray machine and DEXA showed that diaphysis of growing bones responded to mechanical loading by increases in cortical thickness. However, sufficient information concerning the role of the inner contour of the diaphysis in the gain of cortical thickness with loading was not provided by these methodologies. Measurement of bone areas at midradius using pQCT showed that average cortical thickness of the playing arm increased, accompanied by periosteal mineral apposition and endosteal bone resorption, leading the center of the cortex to be further away from the neutral axis. It is well known that the same amount of bone placed further from the long axis of bone results in a stronger bone, showing greater resistance to breaking and torsion (22). Together with an increase in cortical thickness, cortical drift toward periosteal direction resulted in a significant improvement of the mechanical characteristics (SSI) of the players' dominant radius, although the bone density was slightly lower than that of their contralateral radius. In this regard, Jorgensen et al. (8) showed that bone density may be a poor proxy for bone strength when agents that alter the size and geometry of bone are studied.

Enlargement of periosteal bone area with loading was observed in all except one player at midradius and in all but two players at distal radius in this study. Another concern in our study is that the observed increase in periosteal bone area depends on the starting age of playing tennis (Table 4). Slemenda and Johnston (25) reported that the accumulation of bone mass due to physical activity had occurred during the growth spurt, and the same phenomenon has been observed in another human study, giving evidence that, compared with their mature counterparts, growing bones were more responsive to the changing of their mass and geometry as a result of mechanical loading (5, 26). On the other hand, it has been reported that menarche (14, 28) and completion of longitudinal growth (14) are the firstsigns that bone mass development is stopping. Therefore, it seems likely that growing bone in a child who was an earlier starter of playing was more susceptible to periosteal bone formation with mechanical loading than the slowly growing bone in adolescence in a later starter. However, further study with a larger sample size is required to verify the age-related trends that we observed in the radius, because starting age of playing also correlated with the number of years of playing (r = 0.51, P < 0.05).

Mean trabecular BMD at distal radius was significantly increased by intensive tennis playing. Further analysis found that there was a significant negative correlation between the relative side-to-side difference in trabecular BMD and periosteal bone area at distal radius. Together with the fact that periosteal bone area is related to the starting age of playing tennis, as we have shown above, it could be that players who started their playing careers in late adolescence, although capable of producing mineral accumulation, did not expand periosteal bone area but, rather, thickened trabecular density by mechanical loading. Indeed, we observed in this study that the playing arm in all three players who had not started to play until the age of 16 yr, the age at which the bone has almost fully grown, had greater trabecular BMD (range: +12.9 to +16.7%) without change in periosteal bone area (range: 4.1 to +0.4%) when compared with nonplaying arm. An increase in trabecular BMD in those subjects who have no or little gain in periosteal bone area probably reflects a compensatory mechanism to increase bone strength. That is, when exercise stimulates bone formation in bones that cannot grow, bone density will increase to compensate for bone rigidity against the load imposed on it. Similar observations have been made in humerus of older tennis players, giving evidence that the side-to-side differences in the playing-arm bone widths were even somewhat smaller than those in the control subjects, despite the fact that the side-to-side differences in BMC and BMD were larger than in controls (6). In future research, it may be important to examine whether bone adaptation to mechanical loading is mediated by somewhat different mechanisms dependent on the age of the subjects.

In conclusion, our study provided simultaneous information on geometric properties of the cortical shell and volumetric BMD values for the cortical and trabecular bone through pQCT. The present study suggests that physical activity induces 1) cortical drift toward periosteal direction, resulting in a significant increase in mechanical strength despite a lower volumetric density at midradius of playing arm; and 2) an increase in the trabecular BMD of distal radius, which was inversely related to side-to-side differences in total bone area. Further studies are needed to investigate whether the same adaptation to mechanical loading takes place on anatomic regions that are characteristic sites of osteoporotic fractures, such as spine and hip.