Exercise during growth results in biologically important increases in bone mineral content (BMC). The aim of this study was to determine whether the effects of loading were site specific and depended on the maturational stage of the region. BMC and humeral dimensions were determined using DXA and magnetic resonance imaging (MRI) of the loaded and nonloaded arms in 47 competitive female tennis players aged 8–17 years. Periosteal (external) cross-sectional area (CSA), cortical area, medullary area, and the polar second moments of area (IP, mm4) were calculated at the mid and distal sites in the loaded and nonloaded arms. BMC and IP of the humerus were 11–14% greater in the loaded arm than in the nonloaded arm in prepubertal players and did not increase further in peri- or postpubertal players despite longer duration of loading (both, p < 0.01). The higher BMC was the result of a 7–11% greater cortical area in the prepubertal players due to greater periosteal than medullary expansion at the midhumerus and a greater periosteal expansion alone at the distal humerus. Loading late in puberty resulted in medullary contraction. Growth and the effects of loading are region and surface specific, with periosteal apposition before puberty accounting for the increase in the bone's resistance to torsion and endocortical contraction contributing late in puberty conferring little increase in resistance to torsion. Increasing the bone's resistance to torsion is achieved by modifying bone shape and mass, not necessarily bone density.
EXERCISE DURING GROWTH results in biologically important increases in bone mass. Growth in bone width and cortical thickness before puberty occurs by greater periosteal (outer surface) apposition than by endocortical (inner surface) resorption. During puberty, estrogen production inhibits periosteal apposition but stimulates the acquisition of bone on the endocortical surface.(1)
It has been proposed that exercise will enhance formation at the surfaces of bone undergoing bone apposition.(2) Because apposition of bone on the periosteal surface is a more effective means of increasing the bending and torsional strength of bone than acquisition of bone on the inner surface,(3) exercise regimens may be more effective when undertaken at a time when the growth of bone is dominated by periosteal rather than endocortical growth.
Exercise has been reported to enhance periosteal expansion in young animals and endocortical contraction in mature animals. Thus, to determine whether the effects of exercise depend on the maturational stage of the region exposed to loading as well as the intensity and duration of the loading, we tested the following hypotheses: (i) loading of bone during tennis playing will result in increased cortical area of the playing humerus because of periosteal expansion with no endocortical apposition in the pre- and peripubertal years; (ii) during the postpubertal years, loading will increase cortical area by endocortical apposition with less contribution from periosteal apposition; and (iii) the exercise-induced increase in cortical area and bending strength of the humerus will be caused by greater periosteal apposition with little or no contribution from endocortical bone acquisition.
MATERIALS AND METHODS
Forty-seven pre-, peri-, and postpubertal competitive female tennis players aged 8–17 years were recruited from tennis clubs in Melbourne, Australia. Players were included if they had been playing competitive tennis for a minimum of 2 years and were currently playing at least 3 h/week (Table 1). Forty girls were right-handed, and 41 girls used a double-handed backhand. Longitudinal data were collected in 37 subjects after 1.1 ± 0.01 years (range, 0.8–1.5 years); 6 subjects remained prepubertal, 6 subjects became peripubertal, 9 subjects remained peripubertal, and 16 subjects remained postpubertal during the observation period. Ten subjects were not included because they were either no longer playing (n = 2), not willing to participate (n = 4), or relocated (n = 4).
Table Table 1.. Age, Age of Menarche, and Training History of Pre-, Peri-, and Postpubertal Female Tennis Players (Mean ± SEM)
All girls were healthy and received no medication known to affect the skeleton. The Deakin University and Alfred Hospital ethics committees approved the study, and written consent was obtained from all participants and their parents. Sexual maturation was self-assessed with parental guidance using the standard five-scale Tanner stages for breast development. Subjects were classified as prepubertal (Tanner stage 1), peripubertal (Tanner stage 2–4), or postpubertal (postmenarche).
Bone geometry, mass, and strength
Magnetic resonance imaging (MRI) was used to determine bone dimensions (1.5 T whole-body unit; with a commercial transit-receive torso coil; Signa Advantage GE Medical Systems, Milwaukee, WI, USA). T1-weighted spin-echo images at a repetition time (TR) of 600 ms and an echo time (TE) of 14 ms were acquired in the axial plane. Field of view was 200 mm2 and the matrix size was 512 × 192. The region of interest (ROI) was 30–60% from the distal end of the humerus and was divided into thirds. Areas of the proximal third of the ROI representing the midportion of the humerus were compared with the distal third. Five-millimeter slices (with 5-mm gaps between slices) were scanned along the ROI. Each axial image was analyzed using the OSIRIS imaging software program (Digital Imaging Unit, Center of Medical Informatics, University Hospital of Geneva, Geneva, Switzerland). Periosteal area was the external size of the bone (i.e., periosteal border) and cortical area was periosteal minus the medullary area (Fig. 1).
Summing the cross-sectional areas of each slice in the ROI divided by the total number of slices in the ROI determined average periosteal, cortical, and medullary areas. The short-term precision (CV) was 1.02% and 0.21% for periosteal and cortical bone areas, respectively. In vivo studies using bovine bones have shown that MRI provides accurate estimates of bone cross-sectional areas, and these data correlate well with quantitative computed tomography (QCT) measurements of the same bone (r2 = 0.98).(4) Bone mineral content (BMC) of the playing and nonplaying arm was measured using DXA (CV for BMC was 3.6%; Lunar DPX-L, version 1.3b; Lunar Corp., Madison, WI, USA).
To assess the bones resistance to bending (rigidity), each image was imported into Scion Image 4.0.2 (Scion Corp., Frederick, MD, USA). The maximum (IMAX, mm4) minimum (IMIN, mm4), and polar (IP, mm4) second moments of area were calculated using a custom macro. The second moment of area (I) reflects a structure's resistance to bending and is calculated by dividing the section into small areas (pixels), and multiplying each (dA) by its squared distance from the neutral plane. This procedure is integrated over the entire cross-section. The macro calculates I about all possible neutral planes and reports the largest value as IMAX and the smallest value as IMIN, which are perpendicular to one another. The polar second moment of area (IP) reflects a long bones resistance to torsion and equals the sum of the maximum and minimum moments of area (IP = IMAX + IMIN).
Data were expressed in absolute terms and as a percentage of the nonplaying arm. Within each pubertal group, side-to-side differences were assessed using paired t-tests. ANOVA, with Tukey post hoc comparisons, was used to detect differences between pubertal groups. In the longitudinal analysis, subjects were divided according to pubertal status: prepuberty to prepuberty (n = 16), peripuberty to peripuberty (n = 15, includes prepuberty to peripuberty and peripuberty to peripuberty), and postpuberty to postpuberty (n = 16). Repeated measures ANOVA and analysis of covariance (ANCOVA) were used to determine changes over time in bone strength adjusted for bone size. Significance is reported as p < 0.05; borderline significances are reported at p < 0.1. All data are reported as mean ± SE unless otherwise stated.
Growth itself, as reflected by structural changes in the nonloaded arm, resulted in a 14% increase in cortical area of the mid- and distal humerus from the pre- to peripubertal years because of greater periosteal expansion than medullary expansion (Table 2 and Fig. 2). Cortical area of the mid- and distal humerus were both ∼20% greater in the postpubertal players than in the peripubertal players (Table 2 and Fig. 2). At the midhumerus, this was the result of periosteal expansion alone, whereas at the distal humerus, medullary contraction contributed to the larger cortical area.
Table Table 2.. Cross-Sectional Analyses of the Change in Cortical, Periosteal, and Medullary Bone Areas and the Second Polar Moment of Area (Ip) in the Nonloaded Humerus With Advancing Maturation (Mean ± SEM)
The effect of loading was reflected in the side-to-side trait differences. BMC and resistance to torsion (IP) of the humerus were 11–14% greater in the loaded arm than in the nonloaded arm in the prepubertal players (both p < 0.01) and did not increase further in peri- or postpubertal players despite longer duration of loading (Tables 1 and 3). The higher BMC was the result of a 7–11% greater cortical area in the prepubertal players, which was the result of greater periosteal expansion than medullary expansion at the midhumerus but greater periosteal expansion alone at the distal humerus (Table 3 and Fig. 3). Loading during the peri- to postpubertal years resulted in medullary contraction at both sites; however, this did not lead to a significant increase in the side-to-side difference in cortical area (Table 3 and Fig. 3).
Table Table 3.. Average Bone Areas of the Mid- and Distal Regions of the Humeral Shaft in the Loaded and Nonloaded Humerus of Pre-, Peri-, and Postpubertal Female Tennis Players (Mean ± SEM)
Similar observations were made in the 37 girls followed during the 12 months of follow-up. In particular, cortical area at the distal site increased 4% more in the loaded arm than in the nonloaded arm in the postpubertal players because of contraction of medullary area (2%, p < 0.05) and increased periosteal expansion (2%, NS).
Growth in the external size of a long bone, its cortical thickness, and the distribution of cortical bone about the neutral axis are determined by the absolute and relative behavior of the periosteal and endocortical bone surfaces along the length of the bone.(5, 6) Before puberty, periosteal apposition accounts for most of the increase in cortical area. Endocortical resorption creates an enlarging marrow cavity and partly offsets the increase in cortical area produced by periosteal apposition. The net result is an enlarged cortical area located further from the neutral axis, which has the effect of increasing its resistance to bending.(3) Late in puberty, periosteal apposition continues with a contribution from endocortical apposition(7); particularly at the distal humerus, where, in this study, the increase in cortical area was the result of equal contributions from periosteal and endocortical apposition.
In addition to surface specificity, growth is also region specific with more rapid maturation of distal regions than proximal regions. Distal segments of the appendicular skeleton mature before the proximal segments.(1,7,8) The longitudinal data indicate that when the subjects were older, endocortical contraction was detected at the midhumerus but not at the distal humerus.
Loading magnifies the structural changes produced during growth and this was detected by comparing the trait differences in the loaded and nonloaded arms. The data suggest that during growth the effect of exercise, like the effect of risk factors, is determined not only by the intensity of exercise or severity of illness, but also by the timing of exposure.(7, 9) Before puberty, loading magnified periosteal apposition. During the postpubertal period, loading magnified the effect of endocortical apposition, which makes an important contribution to cortical thickness in females. Indeed, endocortical apposition accounted for most of the greater side-to-side difference attained in the postpubertal years.
Most of the structural changes occurred early in the prepubertal years because adaptive changes in response to loading were sufficient to reduce the strains in bone that may lead to microdamage if not decreased.(10, 11) The only additional benefit achieved from tennis training later in puberty was contraction of the medullary cavity, which did not confer any additional increase in the structural rigidity of the bone. Similar effects have been reported in soccer players in whom increased duration of training beyond 6 h/week had no benefit on bone mass.(12) To further modify bone mass or architecture, other components of loading other than duration (i.e., magnitude or strain patterns) would have to increase, as reported in elite gymnasts.(13)
Heterogeneity in the response to loading has been reported in several studies.(14–17) The relative contributions of periosteal and endocortical modeling and remodeling varies along the whole length of a limb.(18) Local loading will modify each part of the geometry of the bone in accordance with the imposed load. In racquet sports, the greater humeral cortical area of the loaded versus the nonloaded arm is the result of both greater periosteal expansion and greater endocortical contraction; for instance, the relative contributions of periosteal expansion and endocortical contraction to the greater cortical thickness in the loaded arm than in the nonloaded arm in a study by Haapasalo et al. were 75:25 at the proximal humerus and 10:90 at both mid- and distal humerus.(15) In the study by Jones et al., the respective relative contributions of greater periosteal expansion and greater endosteal contraction to the greater cortical thickness were 60:40 in the anteroposterior dimension and 80:20 in the mediolateral dimension in male and female tennis players.(16)
Thus, loading affects both the periosteal and the endocortical surfaces but the magnitude of the effects vary according to whether the surface is anterior, posterior, medial, or lateral and according to whether the region is proximal, central, or distal along the bone's length. Measurements of bone geometry in two dimensions using densitometry or X-rays cannot adequately describe this heterogeneity. Bone is not a cylinder with a circular perimeter and the assumption that loading will produce homogenous changes is flawed.
In conclusion, loading before puberty increases bone size and its resistance to bending. After puberty, loading increases the acquisition of bone on the endocortical surface with little benefit in the bone's resistance to bending. Growth and the effects of loading were surface specific and varied along the length of the bone depending on the maturation of the region as well as the intensity and direction of loading. Increasing the bone's resistance to bending and torsion is achieved by modifying the shape and mass of bone but not necessarily its density.
The authors thank radiographers Amanda Hunt and Glenn Rush for their technical assistance. They also thank the players and their parents for their time given to this study. This study was funded by grants from the Australian Research Council Grant and the School of Health Sciences, Deakin University.