This prospective 5-year follow-up study of 64 adult female racquet sports players and 27 controls assessed the changes in the playing-to-nonplaying arm bone mineral content (BMC) differences to answer three questions: (1) Are training-induced bone gains lost with decreased training? (2) Is the bone response to decreased training different if the playing career has been started before or at puberty rather than after it? (3) Are the possible bone changes related to the changes in training? The players were divided into two groups according to the starting age of their tennis or squash playing. The mean starting age was 10.5 years (SD, 2.2) among the players who had started training before or at menarche (young starters; n = 36) while 26.4 years (SD, 8.0) among those players who had begun training a minimum of 1 year after menarche (old starters; n = 28). At baseline of the 5-year follow-up, the mean age of the young starters was 21.6 years (SD, 7.6) and that of old starters was 39.4 years (SD, 10.5). During the follow-up, the young starters had reduced the average training frequency from 4.7 times a week (2.7) to 1.4 times a week (1.3) and the old starters from 4.0 times a week (1.4) to 2.0 times a week (1.4), respectively. The 5-year follow-up revealed that despite reduced training the exercise-induced bone gain was well maintained in both groups of players regardless of their clearly different starting age of activity and different amount of exercise-induced bone gain. The gain was still 1.3–2.2 times greater in favor of the young starters (at the follow-up, the dominant-to-nondominant arm BMC difference was 22% [8.4] in the humeral shaft of the young starters versus 10% [3.8] in the old starters, and 3.5% [2.4] in controls). In the players, changes in training were only weakly related to changes in the side-to-side BMC difference (rs = 0.05–0.34, all NS), and this was true even among the players who had stopped training completely a minimum 1 year before the follow-up. In conclusion, if controlled interventions will confirm our findings that an exercise-induced bone gain can be well maintained with decreased activity and that the maintenance of the bone gain is independent of the starting age of activity, exercise can be recommended for preventing osteoporosis and related fractures.
Tennis and squash players are an interesting target group when studying the long-term influences of physical activity on bone. A study design based on a playing-to-nonplaying arm comparison eliminates the confounding effects of genetic, hormonal, and nutritional factors, which are encountered in the cross-sectional comparisons between athletes and their controls. The previous studies on tennis players have given evidence that bone tissue on the playing arm clearly benefits from mechanical loading, the playing-to-nonplaying arm differences being over 20% in favor of the playing arm, as compared with less than 5% dominant-to-nondominant arm differences in nonplayers.(1–10) Controlled racquet sports studies also have given the most convincing clinical evidence that the majority of the playing-induced bone gain is obtained already before the end of the longitudinal growth, and that if the playing activity is started after the pubescent years, the bone gain is clearly smaller.(1,2,4)
Although the effects of mechanical loading on the pubertal skeleton can be considerable, little is known whether this additional bone mass is maintained into adulthood despite decreased activity. Retrospective cross-sectional studies on former athletes and their controls have given preliminary evidence that at least some residual benefits appear to be maintained into adulthood.(11–17) However, these studies have many confounding factors that may obscure the conclusion. First, in retrospective cross-sectional studies the amount of bone gained by exercise before cessation of the training is not known. Thus, the likelihood of pure selection bias when comparing former athletes and their controls cannot be excluded. Second, the results of the cross-sectional studies can be biased because the quantity and quality of the physical activity of the former athletes and their controls during childhood or adolescence are difficult to evaluate retrospectively-not to speak about their other bones-affecting living habits in that time. Third, although the athletes of these studies have ceased their active career, it is possible that they are still more active than the sedentary controls, and this obscures the conclusion further.
Our preliminary prospective study (a 4-year follow-up study of male tennis players) evaluating the effects of decreased training on bone indicated rather good maintenance of exercise-induced bone gain despite decreased playing activity.(18) To confirm these results and especially to evaluate whether the bone response to decreased training is different when the exercise-induced bone gain has been acquired during the growing years rather than in adulthood, a prospective 5-year follow-up of female tennis and squash players was conducted. The purpose of this study was to compare the changes in the playing-to-nonplaying arm difference in the bone mineral content (BMC) of the players and their controls to answer three questions:
(1)Are training-induced bone gains lost with decreased training?
(2)Is the bone response to decreased training different if the playing career has been started before or at puberty rather than after it?
(3)Are the possible changes in side-to-side BMC difference related to the changes in training?
MATERIALS AND METHODS
In 1993 we examined the bones of both upper extremities of nationally ranked adult female tennis and squash players (n = 105) and their age-, height-, and weight-matched controls (n = 50) and showed that the BMC of the playing arm was about two times greater if the playing career had been started before or at menarche rather than after it.(1) The 5-year follow-up measurements could be done on 64 of these players and 27 controls. The nonattending persons were either living too far or could not find enough time to attend. Of note, their baseline BMC values did not differ from those of the attending persons. The attending players were divided into two groups according to the starting age of their tennis or squash training (either before or after menarche) to examine the possible difference in the maintenance of the loading-induced BMC benefit.
The mean starting age of tennis or squash was 10.5 years (SD, 2.2) among those players who had started their training before or at menarche (young starters, n = 36) and 26.4 years (SD, 8.0) among those players who had begun training a minimum 1 year after menarche (old starters, n = 28). During the follow-up, the young starters had reduced the average training frequency from 4.7 times a week (2.7) to 1.4 times a week (1.3) and the old starters from 4.0 times a week (1.4) to 2.0 times a week (1.4), respectively (Table 1).
Table Table 1.. Characteristics of the Subjects (Mean ± SD)
The study groups were clinically healthy, and they did not have past upper extremity fractures. None of the controls had been involved in physical activity or work affecting the dominant extremity only. All subjects were informed of the study procedure, purposes, and known risks, and all gave their informed consent.
A detailed training and medical history of each subject was obtained in 1993 and 1998, the latter interview concentrating on the possible changes between the initial and follow-up measurements. The interview included data on starting age and years of active playing, training sessions per week, duration of each session, other physical activities, possible injuries, menstrual cycle, medication, diet, consumption of alcohol, smoking, and known diseases. The onset of menses was asked in 1993 and the possible irregularities of menstrual pattern during the measurement years were examined at the follow-up.
Anthropometric and isometric strength measurements
Height (cm) and weight (kg), and circumference of upper extremities, elbow extension and flexion forces, and grip strength were measured using the procedure described in detail elsewhere.(1,19)
Bone mineral measurements
BMC (g) was measured with a dual-energy X-ray absorptiometric (DXA) scanner (XR-26; Norland, Inc., Fort Atkinson, WI, U.S.A) at the proximal humerus, humeral shaft, and distal radius.
All baseline and follow-up DXA measurements were performed using the standard bone measurement procedure of our laboratory. The in vivo precision of the BMC measurements has been shown to be 1–2%.(20) The scanner performance was monitored by our quality assurance program(21) and no scanner drift was observed during the follow-up.
The data were analyzed using the SPSS statistical package (SPSS, Inc., Chicago, IL, USA). The relative side-to-side difference was calculated by dividing the dominant-to-nondominant side difference by the nondominant side value and then multiplying the absolute value of the outcome by 100.
In players and controls, the changes in side-to-side difference in arm circumference, muscle strength, and BMC between 1993 and 1998 were determined with the matched, paired t-test. The change in the relative side-to-side BMC difference across the three study groups (young starters, old starters, and controls) was analyzed using the analysis of variance (ANOVA) with repeated measurements (the within-subject factors were limb and time, and the between-group factor was the study group). The post hoc group comparisons with corresponding the 95% CIs (adjusted for multiple comparisons) were done by the Scheffè's method. The associations between the changes in the training variables of the players and the changes in their relative side-to-side BMC differences were determined with the nonparametric Spearman rank correlation coefficients.
The results are expressed as the mean ± SD and the 95% CI. The given significance levels refer to two-tailed tests. An α of less than 5% (p < 0.05) was considered statistically significant.
Changes in lifestyle and medical history
Diet, Ca-intake, consumption of alcohol, and smoking at baseline have been reported previously.(1) In these variables, no major changes occurred during the 5-year follow-up. Number of sessions per week in physical activities other than racquet sports had slightly increased in young starters and controls and slightly decreased among the old starters (Table 1).
There were no severe injuries during the follow-up period, and no one used bone-affecting medication. No menstrual irregularities were reported in the player groups but two controls reported oligomenorrhea in both interviews. In the player groups, the number of oral contraceptive users increased from 16 in 1993 to 28 in 1998. In the control group, there were 11 users in 1993 and 12 users in 1998. Two players and three controls had given birth during the follow-up. At the follow-up, five players (old starters) and three controls reported that menopausal symptoms had started, and all of them except one player used estrogen replacement therapy.
Changes in arm circumference and muscle strength
The average arm circumferences and the results of the strength and bone measurements in 1993 and 1998 are listed in Table 2, the significance of the change by time given as the 95% CI.
Table Table 2.. Dominant Versus Nondominant Arm Comparisons (Mean ± SD)
Among the young starters, a statistically significant decrease by time was seen in elbow flexion (−7%) and grip strength (−9%) while in the old starters the change in the elbow extension was the only statistically significant change (−7%) (Table 2).
Changes in bone
Change in the side-to-side BMC difference from 1993 to 1998
In the young starters, a 2.7% decrease (95% CI, −4.6% to −0.7%, and p = 0.009) in the relative side-to-side BMC difference was seen at the proximal humerus (Table 2; Fig. 1). At the humeral shaft, there was no significant change, whereas at the distal radius a 2.1% decrease (95% CI, −4.2% to −0.1%, and p = 0.040) was seen. Among the old starters, in turn, the relative side-to-side BMC difference increased 2.3% (95% CI, 0.1% to 4.5%, p = 0.044) at the proximal humerus and 1.3% (95% CI, 0.3%-02.3%, and p = 0.016) at the humeral shaft whereas no change was seen at their distal radius (Fig. 1). There were no statistically significant side-to-side BMC changes in the control group.
The noted decrease in the young starters' relative side-to-side BMC difference at the proximal humerus was caused by a slight increase (1.1%) in the absolute BMC value of the nonplaying arm and simultaneous decrease (−1.1%) in that of the playing arm, while among the old starters the absolute BMC of this nonplaying arm site decreased (−1.7%) and that of the playing arm remained at the 1993 level (0.1%) leading thus to the previously noted increase in the relative side-to-side BMC difference (Table 2; Fig. 1).
Change in the relative side-to-side BMC difference across the study groups
In the ANOVA, the only statistically significant between-group difference in the change of the relative BMC was seen at the proximal humerus (p = 0.001; Fig. 1). At this site, the post hoc analysis revealed a 4.1% (95% CI, 1.1–7.2%, and p < 0.001) decrease in the relative side-to-side BMC difference between the young and old starters, and this was caused by the previously noted opposite changes in the relative BMC differences between the young and old starters.
In controls, the changes in the absolute BMC values of the proximal humerus were similar to those of the old starters, resulting, in the posthoc analysis, in a 3.8% (95% CI, 0.8–7.0%; p = 0.002) decrease in the relative side-to-side BMC difference between the young starters and controls, and this explained the unchanged relative BMC difference between the old starters and controls (0.3%; 95% CI, 2.8–3.5%).
The ANOVA showed no statistically significant between-group differences in the change of the relative side-to-side BMC difference at the humeral shaft (p = 0.47) or distal radius (p = 0.26; Fig. 1).
Players' side-to-side BMC difference in relation to change in training
In both player groups, the Spearman's rank correlation coefficients between the change in the relative side-to-side BMC difference and the change in the training variables (number of training session per week, and the number of the training hours during the year preceding the measurement) varied from 0.05 to 0.36, showing thus a weak, but an insignificant relationship in every comparison.
The individual side-to-side BMC changes also were analyzed separately among the 10 players who had completely stopped their tennis or squash training at least 1 year before the follow-up measurement; however, no clear and systematically decreasing trend was found in these players' relative side-to-side BMC difference and this concerned both the young starters and the old starters.
Retrospective cross-sectional studies on former athletes and their controls have given preliminary evidence that at least a part of the exercise-induced bone gain that is obtained during the years of growth may persist despite decreased physical activity.(11–17,22,23) This 5-year prospective follow-up study of female racquet sports players showed a good maintenance of the side-to-side BMC difference between the playing and nonplaying extremity during the follow-up, although the mean training frequency and the mean hours of training were decreased clearly, thus supporting the results of the previously noted cross-sectional studies. This result was also in line with that of our recent investigation in male tennis players.(18)
Our initial investigation of female players provided strong evidence that the effect of unilateral loading on the playing arm BMC is about two times greater if the playing is started before or at puberty rather than after it.(1) The current study, in turn, indicated that the exercise-induced bone gain was well maintained in both groups of players regardless of the clearly different starting age of activity and the magnitude of the exercise-induced bone gain. Thus, our results did not give direct support to the notion that the exercise-induced bone gain that is obtained during the growth may better withstand the effects of decreased training than the bone gain obtained in adulthood.(24) At the proximal humerus, the old starters even slightly increased their relative side-to-side BMC difference compared with the reverse change among the young starters. These changes were caused by the opposite changes in the absolute BMC values of the dominant and nondominant sides of these two groups. However, because all these absolute BMC changes were generally small and within the 1–2% precision error of the measurement,(20) we should not emphasize these findings too much.
One limitation of our study is that a longer than 5-year follow-up period may be needed to see greater changes (reductions) in players' side-to-side BMC difference, at least to get more players with complete cessation of the activity. In other words, the lack of negative bone response to reduced training in this study may have been partly because of the fact that most of the players were still active enough to maintain their exercise-induced bone gain. It may well be that the decreased but still regularly done tennis or squash training was able to produce a sufficient stimulus for the maintenance of the training-induced bone gain.
On the other hand, our study allowed us to gather some information about the effects of completely stopped tennis or squash training on the exercise-induced bone gain via the players who had not played for at least a year before the follow-up measurement. These players' individual changes in side-to-side BMC difference showed that players with the highest side-to-side difference seemed to have lost some of their exercise-induced bone gain although there also were such players who had well maintained the side-to-side BMC difference in the upper and forearm bones despite ceased tennis activity. Interestingly, both the individual changes of the detrained players and the mean changes in the young and old starters' side-to-side BMC difference seemed to be greater at the bone sites that contained more trabecular bone (proximal humerus and distal radius) than cortical bone (humeral shaft). This was especially clear among the young starters whose side-to-side difference decreased at the proximal humerus and distal radius but remained unchanged at the humeral shaft (Fig. 1). It may well be that decreased loading has a greater influence on trabecular bone sites where a high surface-to-volume ratio makes the bone more susceptible to rapid bone mineral turnover.(25)
Some animal studies support the notion of maintenance of exercise-induced bone gain during at least a short detraining period.(26–28) In contrast, some human studies(29–31) have shown that even a short detraining period reduces the exercise-induced bone gain at the trabecular regions to pretraining values thus suggesting that long-term benefits are only retained with continuing exercise. As evident in the current study, the amount of training needed to maintain the exercise-induced bone alterations clearly can be less than that needed to achieve the additional bone. To support this, aerobic classes approximately twice a week turned out to be an efficient way to maintain premenopausal women's exercise-induced bone gain after cessation of an intense jumping training.(32)
It has been assumed that growing bone responds to exercise through significant additions of new bone at both the cortical site (periosteal expansion and endosteal enlargement) and the trabecular site while there is no exercise-induced endosteal bone accrual in a mature bone.(24,25) Ashizawa et al.(33) suggested that the mature bone might have an alternative mechanism to respond to exercise by increasing it's volumetric trabecular density. Their measurements, carried out with peripheral quantitative computed tomography (pQCT), revealed that all three of the tennis players who had not started playing until the age of 16 years seemed to have increased the trabecular density, not the cross-sectional area, of the distal radius, while the opposite was the case in players who had started playing when younger.
Because the majority of the benefits of training on bone seems to be achieved already during the pubertal years, including the adolescent growth spurt when the effect of physical activity on bone has been shown to be large,(2,34–38) the exercise-induced changes in bone mass and geometry(5,33,35) also might be more permanent if the benefit is obtained during the growing years rather than later on. However, as noted before, the current study could not show more permanent bone preservation with decreased loading in those with whom the bone gain was achieved before or at menarche than in those who had obtained the bone gain later in adulthood. On the other hand, in both groups of players only a few subjects had completely stopped the racquet sports activity and the average one to two times a week of playing may well have been enough for the maintenance of the loading-induced bone gain. Consequently, there is a clear need to follow our players for a longer time, until most of them have ceased the playing completely, and then analyze again the side-to-side BMC difference in both groups. In addition, prospective, randomized exercise trials with children, adolescents, and adults, using precise measurements of the geometry, composition, and volumetric density of the bone compartments, are needed, and these study subjects should be followed long enough after the decrease or cessation of the activity.
Overall, if controlled interventions will confirm our findings that an exercise-induced bone gain can be well maintained with decreased activity and that the maintenance of the bone gain is independent of the starting age of the activity, exercise can be recommended for preventing osteoporosis and related fractures.
We are grateful to Ulla Hakala, Kirsi Martinsen, and Virpi Koskue for their skillful bone measurements. This study was financially supported by the Ministry of Education, Helsinki, the Emil Aaltonen Foundation, Tampere, and the Medical Research Fund of the Tampere University Hospital, Tampere, Finland.