Skeletal Effects of Exercise in Men – Exercise and Bone
It is generally accepted that bone adapts to changes in habitual mechanical loading in order to best withstand future loads of the same nature. This phenomenon, loosely referred to as Wolff’s Law, honors the 19th century scientist who attempted a mathematical description of the process. Evidence of Wolff’s Law emerges in the findings of both animal and human studies. In the human realm, a much larger body of evidence exists for the skeletal effects of exercise loading on females than on males. Consequently, no systematic discussion of the relationship of exercise to bone mass specific to men has been presented.
To initiate such a discussion, this post will address: characteristic features of male bone health, fundamental aspects of the response of bone to loading, and findings of cross-sectional and intervention trials investigating the effect of exercise on the skeleton of male subjects. Specifically, the relationships of body mass, exercise type, intensity and history, site specificity, and muscle strength to male bone density and geometry will be examined.
The influence of age on the response of bone to loading as well as the impact of exercise on male hormone status will be considered. Finally, we will briefly address the comparative effects of exercise on the bones of men and women.
A number of terms are commonly employed to describe bone mass and density. Even though actual bone mass is a difficult element to quantify, it is a relatively simple process to measure the amount of mineral in bone. Bone mass and mineral are highly correlated quantities (given normal matrix min-eralization); thus, bone mineral can be used to estimate reliably the mass of normal skeletal tissue. Bone mineral content (BMC) is a measure of the amount of mineral in a defined region of bone (in grams). To account for differences in bone size among individuals, bone is commonly evaluated in terms of tissue density or porosity. Areal bone mineral density (BMD) is derived by dividing BMC by the area of bone measured and is expressed in grams per square centimeter. Bone mineral apparent density (BMAD) is an approximation of volumetric BMD (Katzman et al., 1991; Carter et al., 1992) introduced to minimize the effect of bone size on two-dimensional scans by considering the dimension of bone depth. It is calculated from densitometry-derived bone area and other skeletal length dimensions and is expressed in grams per cubic centimeter.
III. The Male Skeleton—Why Care about It?
A. Bone Density: Gender Comparison
A clear gender difference exists between the mass of male and female skeletons. Men attain greater values of BMC and BMD than women (Hannan et al., 1992), primarily by virtue of having larger bones. Peak BMD in men compared with women is approximately 1.033 versus 0.942 gm/cm2 at the hip, 1.115 versus 1.079 gm/cm2 at the spine (L2-L4), and 0.687 versus 0.579 gm/cm2 at the forearm (radius). When bone size is fully taken into account, however, these differences essentially disappear; that is, volumetric bone density is approximately equal between men and women.
B. Fracture Risk
Attainment of greater peak bone mass in men is associated with superior bone integrity throughout life and a lower ultimate risk of osteoporotic fracture than for women. In 1990, male hip fractures accounted for 30% of the 1.7 million hip fractures which occurred worldwide (Cooper et al., 1992).
Even though a comparison between genders clearly indicates a problem of greater magnitude for the female population, a closer look at real numbers lends perspective to the situation. Thirty percent of 1.7 million amounts to a total of 510,000 male hip fractures, clearly a non-trivial figure. Perhaps even more noteworthy is the fact that men over 75 years of age have been observed to suffer a 21% mortality rate following hip fracture compared to 8% in women (Poor et al., 1995). With current trends of increasing life expectancy, the prevalence of men who suffer from osteoporosis and related fractures in the future is also likely to rise. Thus, efforts to determine methods of preserving bone in the male population are required.
C. Acquisition and Loss of Bone
Because a comprehensive discussion of this topic appears elsewhere in this volume, only a brief summation will be presented here. In adults, the amount of bone in the skeleton at any time represents that which was formed during growth, minus that which has been subsequently lost. Although considerable information is now available regarding the trajectory of bone acquisition in girls, understanding is less complete for boys. In both sexes, greatest bone acquisition occurs during pubertal growth (Boot et al., 1997). The rate of BMD gain in 11-year-old boys may be 2.5 times that of younger children (Gunnes and Lehmann, 1996), whereas rates of gain increase fourfold to sixfold in the 4 years encompassing ages 13-17 (Theintz et al., 1992; Bonjour et al., 1994). During this period, changes in long bone diaphyses are less marked than in the spine and hip and largely reflect increases in cortical width. Following puberty, the rate of gain in males declines but remains significant at the spine and mid- femoral shaft between the ages of 17 and 20 (Theintz etal., 1992). In females, by contrast, the rate of BMC and BMD increment is greatest between 11 and 14 years, falling dramatically after the age of 16 (Theintz et al., 1992). The rate of change in cortical BMD is thought to peak around 16 ± 0.3 years in boys and 14 ± 0.3 years in girls (Gunnes and Lehmann, 1996).
Thus, even though it has commonly been accepted that both sexes attain peak bone mass during the mid thirties, 95% of peak bone mass is actually attained by age 20. Peak bone mass in both sexes is characterized by substantial interindividual variation (Theintz et al., 1992). The standard deviation of BMD values in the population varies from one skeletal region to the next but is generally about 10-12% of the mean value. Thus, variation in peak bone mass greatly exceeds the variation associated with rates of bone loss later in life. Childhood and adolescence therefore represent crucial periods during which diet, physical activity, and other factors may exert long-term influence on skeletal integrity.
After the acquisition of peak bone mass, men and women experience similar rates of decline in bone mass across most of the lifespan, the obvious exception being the first 5 to 8 years after menopause, when women lose bone at an accelerated rate.
A study of adult human cadavera revealed the following age-related changes in male long bones (femur and humerus): cortical areas increase until approximately 60 to 75 years of age and then decline, cortical porosity increases throughout life, the number of Haversian canals increases until approximately 80 years of age and then declines, and osteon areas decrease gradually with age (Martin et al., 1980). The contribution of bone size and shape to bone strength, and the effect of aging on this relationship, is addressed in Section IVC.
IV. The Response of Bone to Loading— Fundamental Aspects
Repeated observations of relatively high bone mass in athletes have led many to conclude that physical exercise is beneficial to bone. To understand why this is so, it is necessary to achieve a better understanding of the funda-mental mechanisms by which bone responds to mechanical stimulation. The skeleton is routinely exposed to the forces of gravity and muscle contraction. To optimize strength without unduly increasing weight, bones accommodate the loads that are imposed upon them by undergoing alterations in mass, external geometry, and internal microarchitecture. In recent years, considerable energy has been directed toward elucidating mechanical load parameters which optimally stimulate a response from bone.
A. Characteristics of Effective Mechanical Loading
Mechanical loads may be characterized by several independent parameters, including type of load, load magnitude, number of load cycles, and rate at which strain is induced. Bone loads are generally expressed in terms of stress and strain. Stress is the force applied to an object, expressed per unit area. Stress in a bone is calculated by dividing the load on the bone by its cross-sectional area. Strain is a measure of bone deformation in response to the application of stress (i.e., loading) and can be calculated by dividing the change in bone length by its original length.
I. Dynamic versus Static Loading
To be an effective initiator of remodeling, mechanical stimulation must be dynamic. Hert and colleagues (1971) and Lanyon and Rubin (1984) found that simple application of a static load produced no adaptive bone remodeling nor did it protect bone from atrophy. Application of the same load in a cyclical manner, however, induced bone deposition and increased diaphyseal cross-sectional area.
2. Load Intensity versus Cycle Number
From a model comparing the relative effects of load intensity and cycle number on bone mass, Whalen etal. (1988) concluded that load intensity is a more important contributor than cycle number. This conclusion is substantiated by indications from clinical literature, in which highest bone density values have been observed in athletes whose activities include lifting of heavy loads and application of high-impact forces (Block et al., 1989; Heinonen et al., 1995; Robinson et al., 1995). It is also consonant with animal data indicating that the number of load cycles necessary to maintain bone mass is relatively small (Rubin and Lanyon, 1984); that although modest running activity is associated with higher bone mass in rats, running 3 or 18 km per day has the same effect on bone mineral content (Newhall et al., 1991); and that increasing the magnitude of loads with weighted backpacks is a more effective stimulus to increase bone mass than increasing the duration of treadmill running (van der Wiel et al., 1995).
3. Rate of Strain
Peak load magnitude per se does not describe the intensity of loading nor does it determine skeletal response. Rate of strain is a term used to describe the time over which strain develops after load application and is roughly comparable to the term impact. In several experimental models, rate of strain has been shown to be of critical importance to skeletal response, a principle that applies even at large peak strains (O’Connor et al., 1982; Turner et al., 1995a). Turner and associates (1995a) applied loads of 54 N1 at 2 cycles per second for 18 seconds each day to rat tibiae and measured the effect of varying the rate of strain on bone formation and mineral apposition. Results showed a marked linear elevation in both variables as rate of strain increased.
B. The Curvilinear Nature of Skeletal Response
Complete immobilization, as seen with high-level spinal cord injury, leads rapidly to devastating bone loss. By contrast, imposition of even substantial training regimens on normally ambulatory people or animals increases bone mass by only a few percent over a similar period. As an individual goes from immobility to full ambulation, duration of time spent walking becomes a progressively less efficient stimulus for increasing bone mass. A person who habitually walks 6 hours each day might require another 4-6 hours just to add a few more percent BMD. On the other hand, adding a more rigorous stimulus, such as high- impact loading, for even a few cycles would increase the response slope.
C. The Role of Bone Geometry
In addition to bone density, geometric features make a substantial con-tribution to bone strength. Such features include overall size, shape, and distribution of mass, as well as internal microscopic architecture, such as trabecular connectivity. Exercise loading is known to exert an influence on bone geometry.
1. Bone Deformation with Loading
Because long bones are curved, compressive loads applied at joint surfaces rarely act through the center of the bone; hence, bending occurs. In 1964, Frost proposed a Flexural Neutralization Theory of bone remodeling, by which he suggested that bone seeks the shape, size, and location that equalizes and minimizes the amount of tissue deformation incurred by normal usage. Although more recent evidence suggests that in reality this condition is neither achieved nor desirable (Rubin, 1984), it is indeed likely that bone models and remodels to maintain functional stiffness with optimal resistance to injurious bending (Schaffler, 1985). Diaphyseal width contributes significantly to the ability of bone to resist bending loads. Martin and Burr (1989) stated that “If 100 mm2 is removed from the inner cortex of. . . bone … bending strength can be maintained by putting only approximately 30 mm2 back onto the outside surface” (p. 231). Cross-sectional moment of inertia (CSMI) is a measure of bone geometry which determines the resistance of bone to bending at a particular site. It is a function of cross-sectional area and the distribution of bone in that area relative to the point about which the bone bends (axis of rotation). The further the bone is distributed from the axis of rotation, the wider the bone and the more resistance it will have to bending.
2. Age-Related Geometric Adaptation
The diameters of long bone diaphyses tend to expand with age. Expansion is achieved via the concomitant effects of periosteal bone deposition and increased endosteal resorption, such that net thickness of the bone cortex is reduced. The effect is seen in both weight-bearing (Ruff and Hayes, 1988) and non-weight-bearing (Burr and Martin, 1983) bone. Increased diaphyseal width acts to maintain the CSMI of long bones and, correspondingly, to maintain bone strength in the face of cortical thinning and increased porosity. Remains of pre-industrial bones indicate no gender-specific differences in adaptive ability in this respect (Ruff and Hayes, 1983); however, in modern times, males appear to have maintained the ability to expand diaphyseal widths to a greater extent than females (Martin and Atkinson, 1977; Burr and Martin, 1983; Ruff and Hayes, 1988). Although forearm bones in older women do reflect increased cross-sectional moments of inertia, the magnitude of change may not be enough to compensate for excessive endosteal loss of bone (Bouxsein et al., 1994). Cultural and behavioral changes, particularly in physical activity type and intensity, between pre- and post-industrial times likely contribute to gender differences in age-related geometric adaptation today. It is also likely that the superior resistance to osteoporosis in men is at least partly attributable to this more effective geometric compensation for the weakening effect of age-related increases in porosity and cortical thinning.
V. Translating Theory into Practice—Exercise and Bone
A. Limitations of the Literature
I. Implications of Study Design
It is routinely recommended that regular lifelong physical exercise is important for the prevention of osteoporosis (Jackson and Kleerekoper, 1990; Seeman, 1997); however, very few exercise trials that actually confirm or clarify this presumed relationship have been reported. The recommendation to exercise has been based primarily on data from cross-sectional studies comparing BMD between already-exercising and nonexercising groups. Unfortunately, cross-sectional studies contain inherent limitations of selection bias and, as such, may not accurately represent the general population. It is conceivable that individuals who choose to exercise have certain predisposing skeletal characteristics which influence their choice and ability to initiate and maintain regular physical activity. For example, Bennell and associates (1997) reported greater BMD in power athletes than controls, but they also found that the athletes had engaged in more physical activity during childhood than had the controls. As discussed presently, childhood activity is likely to exert a strong influence on bone mass across the lifespan.
2. Methodological Concerns
Exercise-loading bone research is rife with methodological problems which complicate data interpretation. Important examples include reliance on questionnaires and recall for accounts of previous activity, under-representation of ethnic minorities, inability to control for the effects of anabolic steroid use in athletes, variations in bone mineral measurement tools among studies, variations in the precision error of similar measurement tools, and use of “inactive” control groups which actually participate in nontrivial amounts of physical activity. In addition, many early studies utilized BMC to estimate bone mass, a value which, by failing to account for bone size, is less valid for the purposes of comparison between individuals than BMD or BMAD.
Perhaps the most worrisome methodological concern is the fact that instruments used to assess physical activity vary widely, and few data exist to establish their validity or reliability. Many instruments currently in use, designed originally to assess aerobic work or energy expenditure, may simply fail to reflect the loads actually experienced by the skeleton. Even devices that quantify the number of steps taken in a 24-hour period generally do not distinguish the intensity of impact and, therefore, do not fully describe skeletal loading.
A review of cross-sectional exercise studies is further complicated by the variety of subject groupings utilized: exercise versus sedentary, low-intensity activity versus high-intensity activity, sport versus sport, dominant-side limb versus non-dominant-side limb, and sport versus retired from sport. Given an awareness of these inherent limitations, a review of the literature can be presented for interpretation with appropriate circumspection. In spite of methodological shortcomings in many studies, the relative uniformity of findings suggests that some generalizations about the effect of exercise on bone in men can nevertheless be made with confidence.
B. Relationship of Body Mass to BMD
Many report a strong positive relationship of body mass with bone mineral density or content (Hamdy et al., 1994; Suominen and Rahkila, 1991; Snow-Harter et al., 1992; Smith and Rutherford, 1993; Welten etal., 1994; Karlsson etal., 1995; Glynn etal., 1995; Sone etal., 1996; Boot etal., 1997), although a minority of investigators have found otherwise (Nilsson and Westlin, 1971; Bevier et al., 1989). This relationship is in keeping with the tenets of Wolff’s Law in that increased body mass effectively increases the magnitude of daily gravitational load on the skeleton. Of course, a similar
gene may influence both lean body mass and bone mass, a feature which would naturally establish a fundamental relationship between them. In addition, the BMD measurement is itself influenced by bone size (positive correlation) so that genes determining body size can further influence bone density measures.
C. Relationship of Muscle Strength to BMD
The positive effects of exercise on BMD are likely to be due in part to the beneficial effects of exercise on muscle strength. That exercisers have greater muscle strength in most muscle groups than nonexercisers has been repeatedly shown (Snow-Harter et al., 1992). It is also known that muscle mass is highly correlated with muscle strength and that lean body mass is positively correlated with bone mass and cross-sectional properties (Moro et al., 1996; Doyle et al., 1970). In addition to the influence of common genetic determinants, increased muscle mass may exert an effect on bone mass in two ways: (1) by increasing total body mass and the consequent magnitude of gravitational load on the skeleton and (2) by enhancing local bone strains by virtue of an enhanced ability to apply contractile forces at sites of origin and insertion. The observation that power athletes have greater BMD than endurance athletes or controls (Bennell et ai, 1997) somewhat illustrates this relationship.
Back, biceps, quadriceps, and grip strength have all been positively cor-related with hip, spine, whole body, and tibial BMD. Back extensor muscle mass and strength in particular have been found to be the strongest, most robust predictors of BMD at many sites, particular the spine and hip (Bevier et al., 1989; Snow-Harter et ai, 1992). Grip and biceps strength correlate positively with forearm BMC (Myburgh et al., 1993; Bevier et al., 1989), quadriceps strength is a positive determinant of hip BMD (Glynn et ai, 1995) , and leg strength is similarly positively correlated with hip BMD (Block etal., 1989), illustrating a site specificity of bone response to loading.
Some authors, however, have reported that quadriceps strength is not an independent predictor of BMD (Duppe et al., 1997) and is not related to distal femoral BMD (Nilsson and Westlin, 1971). In order to elucidate the nature of the BMD-muscle strength relationship fully and account for the influence of genetic commonality between the two factors, intervention trials designed to primarily address this issue must be completed.
D. Exercise Effects—Cross-Sectional Study Findings
Animal studies have indicated that the skeletal response to mechanical loading may be tempered with age (Rubin etal., 1992; Turner etal., 1995b). Adequacy of skeletal response reflects bone cell numbers and vigor as well as hormonal and cytokine milieu. Cell populations, circulating growth factors, and production of bone matrix proteins all decline with age (Benedict et al., 1994; Termine, 1990) and may all contribute to an age-related deficit in skeletal response to loading. To account for this effect, subsequent discussion will be grouped according to the age of subjects. Data for children and adolescents (<20 years) will be discussed initially, followed by those for adult (20-65 years) and finally, older men (>65 years).
I. Young Males
Investigations of exercise effects on bone have not typically targeted the pediatric population. As a consequence, only a small amount of data is available for analysis.
a. Current Activity Even in very young children, exercise appears to pro-mote the acquisition of bone. In a study of premature infants, Moyer-Mileur and associates (1995) found that five repetitions of range of motion, gentle compression, flexion, and extension exercises five times a week resulted in greater acquisition of BMD at 4 weeks in exercised babies than in controls.
Slemenda and associates (1994) studied factors influencing the rate of skeletal mineralization in male and female children and adolescents. Even though a combined-gender analysis limits male-specific inferences which can be drawn from their report, the authors noted that physical activity was a significant predictor of BMD at the radius, spine, and hip in prepubertal but not peripubertal children. These findings suggest that exercise exerts an influence on BMD before puberty, but during puberty other factors become more influential on bone acquisition.
Researchers have repeatedly found significant associations between physi-cal activity and forearm, total body, and spine BMD in adolescent boys of various races (Duppe et al., 1997; Boot et al., 1997; Welten et al., 1994; Gunnes and Lehmann, 1996; Tsai et al., 1996). VandenBergh and associates (1995) , however, found that after adjustments for body weight, height, skeletal age, and chronological age were made, no correlation between physical fitness (measured by V02nux) and middle-phalanx BMC existed in 7- to 11-year-old boys. Upon closer analysis, greater BMC was actually found in boys categorized as highly fit versus low-fit boys older than 10, but not younger. The latter study illustrates the methodological limitations inherent in the search for a relationship between exercise per se and bone density. Maximal oxygen consumption is not an adequate surrogate for more specific measures of skeletal loading. The finding of greater BMC in highly fit boys compared to low-fit boys is likely to reflect the increased chance that more active boys will participate in activities that beneficially load the skeleton than less-active boys, but it cannot be confirmed.
b. Site Specificity Considerable evidence suggests that the adaptive bone response is site-specific; that is, only bones or regions of bone that are loaded undergo significant load-related change. Nordstrom and associates (1996) compared tibial tuberosity BMD between highly active and less-active teenage boys (average age 15). They observed significantly greater tuberosity BMD in the more active group (when Osgood Schlatters sufferers were excluded from the analysis) and found that the difference was related to quadricep strength. As the quadricep muscles insert at the tibial tuberosity, these findings support a localized muscle-loading effect on bone.
Adolescent Chinese athletes were found to exhibit sport-specific differences in BMD. Judo athletes had greater spine BMD than baseball, swimming, and track athletes, whereas baseball players had greater hip BMD than swimmers, judo, and track athletes (Tsai et al., 1996). Variations in BMD response to different sports reflect the different loading patterns of each sport.
2. Adult Males
a. Current Activity In adult men (approximately 20 to 65 years old), in-dividuals exercising at relatively high loads have consistently greater BMD than nonexercisers or those exercising at low loads. These differences have been found in the whole body (Snow-Harter et al., 1992; Karlsson et al., 1996; Bennell etal., 1997), spine, and/or proximal femur (Block etal., 1986, 1989; Colletti et al., 1989; Snow-Harter et al., 1992; Karlsson et al., 1993, 1996; Need etal., 1995; Sone etal., 1996; Duppe etal., 1997), distal femur (Nilsson and Westlin, 1971), tibia (Leichter et al., 1989; MacDougall et al., 1992; Snow-Harter et al., 1992; Karlsson et al., 1993), calcaneus (Hutchinson et ai, 1995), and distal forearm (Karlsson et al., 1993).
Even though such differences are easily shown between athletes and controls, differences between athletes participating in different sports (e.g., water polo and weight training; Block et al., 1989) or at different intensities of the same sport (MacDougall et ai, 1992) may not be as evident. This observation is predictable from the previously described curvilinear nature of the skeletal response to loading.
Moderate exercise may not apply a sufficient stimulus to the skeleton to precipitate an adaptive response. Myburgh and associates (1993) observed no differences in ulna BMC between moderately active subjects and controls, whereas differences between highly active versus moderately active subjects and controls were significant.
Some have stated that individuals who participate in reduced-weight-bearing activities such as swimming and bicycling have BMD similar to non-exercisers (Nilsson and Westlin, 1971; Taaffe et al., 1995). Discrepancies exist, however, concerning the effect of non-weight-bearing activity on bone density because some have found that male swimmers have greater radial and spine BMD than controls (Orwoll et al., 1989). Given the location of the BMD increment in the swimmers in Orwoll’s study, disparities may be based in the site specificity of bone response to loading. During swimming, contraction of muscles acting on the upper extremity may place substantial loads on the spine (latissimus dorsi) and radius (biceps brachii, brachioradi- alis). Alternatively, degree of swimming participation may strongly influence the effect of the activity on bone density. Elite swimmers who train intensively effectively unload their skeletons by spending extended periods of time in a reduced-gravity environment. Exercise loads placed on the skeleton during swimming may be of insufficient magnitude to overcome the negative impact of substantially reduced daily weight bearing.
High-intensity impact activities such as running, jumping, and power lifting are thought to be more effective bone stimulators than low-intensity or non-weight-bearing activities. Need and colleagues (1995) found that spine and hip BMD are positively correlated with activity levels in men 20 to 83 years old and that femoral neck BMD was significantly greater in joggers than sedentary subjects. Dalen and Olsson (1974) found that 50- to 59-year-old men with 25 years running experience had significantly greater BMC at the distal forearm, humeral head, femoral shaft, and calcaneus than controls. Bennell and associates (1997) reported baseline cross-sectional data on power and endurance athletes versus controls (19 to 20 years olds) and again after 12 months in order to observe longitudinal sport-specific BMD changes. They found that total body and femoral BMD increased in all groups, a generalized effect attributed to continued growth of the study subjects; however, power athletes gained more BMD than endurance athletes or controls.
Interestingly, some have observed lower skull (Karlsson etal., 1996) and rib (Smith and Rutherford, 1993) BMD in athletes than controls, raising the question of whether nonloaded bone actually suffers at the expense of BMD enhancement in loaded bone.
b. Previous Activity A previous history of substantial sports participation is likely to be beneficial to bone, although study findings are often confounded by continued activity participation. Sone and associates (1996) found men who had exercised “often” or “sometimes” in the past had greater spine and femoral neck BMD than those who had exercised “not at all.” (Those who exercised “often” in the present also had greater spine BMD than nonexercisers.)
Karlsson and associates (1996) noted that ex-weight lifters who had retired from their sport 25 ± 13 years previously maintained a significant difference in total body and hip BMD from controls. (Because 72% of exlifters continued to exercise for 5 ± 3 hours per week after retirement, the contributions of historical versus current levels of exercise are difficult to discern.) Ex-weight lifters in the age range of 50-65 were observed to maintain significantly higher total body (Karlsson et al., 1996) and spine BMD (Karlsson et al., 1995) than controls, but not after age 65 (Karlsson et al., 1996) . Significant positive correlations were also found between current ex-ercise frequency, exercise intensity, and BMD.
c. Site Specificity As with children and adolescents, the effects of exercise loading on BMD of adult men are likely to be site-specific. BMD differences in athletes are often observed only at certain skeletal locations. Hamdy and associates (1994) found BMD differences between weight lifters, runners, cross trainers, and recreational sport participants to be present only in the upper limbs. Weight lifters had greater upper limb BMD than runners and recreational athletes, and cross-trained athletes also exhibited greater upper limb BMD than runners. Because weight lifting typically loads the upper limbs to a much greater extent than running, these findings make intuitive sense. In an illustration of the vagaries of cross-sectional data, however, other studies have shown power and weight lifters to have greater spine and hip but not upper or lower limb BMD than endurance athletes or controls (Bennell et al., 1997; Colletti etal., 1989). Aloia and associates (1978) found that even though total body calcium in marathon runners exceeds that of nonrunners, no such difference was observed in radial BMC and width. Such a finding is likely to reflect the fact that the radius is not substantially loaded during running and is thus unlikely to undergo running-related adaptation. Although not specifically measured, the increments in total body calcium are likely to have resided in the weight-bearing limbs.
Limb domination (“handedness” or “footedness”) provides an elegant model of the site specificity of skeletal adaptation. Commonly, the dominant arm exhibits greater total and cortical bone mass than the nondominant arm (Rico et al., 1994). Even greater differences between right and left side limb bone masses become evident when the dominant limb is chronically overloaded. Up to a 40% difference between the playing and nonplaying arm humeral BMD of professional tennis players has been observed, compared to a 3% difference between right and left arms of non-tennis-playing controls (Haapasalo et al., 1996; Dalen et al., 1985). Dominant leg BMD has also been observed to be greater than nondominant leg BMD in players of a variety of weight-bearing sports (Nilsson and Westlin, 1971). Rowers and triathletes appear to have no such BMD sidedness (Smith and Rutherford, 1993) , an observation which is predictable given that rowing, running, swimming and biking load bilateral limbs essentially equally.
d. Does an “Exercise Intensity Benefit Ceiling” Exist? A number of investigations have reported seemingly anachronistic results describing the effect of exercise on BMD. MacDougall and associates (1992) observed a generalized increase in BMD with running mileage up to 15-20 miles per week; thereafter, the trend reversed, suggesting a possible detrimental effect of over-training. The reduction in bone density, however, was accompanied by an increase in bone area which was significantly different between controls and runners covering 40-50 miles per week. Bilanin and associates (1989) found 9% less vertebral BMD in runners who ran an average of 92 km per week for several years than in controls. Similarly, Hetland and colleagues (1993) reported a negative correlation between weekly running distance and BMC at the spine, total body, hip, and forearm. Even though the use of BMC limits the utility of these results, other findings support a negative effect of ever- increasing exercise loads. These findings suggest the existence of an “exercise intensity ceiling” beyond which bone mass declines. However, no direct evidence for such a ceiling has ever been presented, and it is equally plausible that low BMD in successful ultra-distance runners reflects a self-selection effect based on pretraining characteristics or other effects such as low body mass or nutritional inadequacy.
3. Older Males
a. Current Activity Studies of older men (>65 years) report responses of bone to exercise loading which are similar to those of younger men. Endurance and speed athletes aged 70-81 were found to have greater calcaneal BMD than controls (Suominen and Rahkila, 1991). In a study of men aged 50 to 72, Michel and associates (1989) found a positive association between weight-bearing exercise and lumbar spine BMD but no significant correlation between BMD and non-weight-bearing exercise.
By contrast, Need and associates (1995) found significant relationships of activity to BMD in younger men to dissipate after the age of 50. It is likely that the lack of relationship of BMD to current level of activity in their study reflected the low intensity of activities pursued. Pollock and associates (1997) measured whole body, spine, and hip BMD of current and exathlete men aged 60 to 90+ who continued to participate in low-, moderate-, and high- intensity forms of exercise. Even though investigators did not stratify BMD data by exercise intensity, they found that bone density was generally maintained in all men observed. This finding argues against a critical role of exercise intensity for bone density maintenance in older men.
b. Previous Activity Glynn and associates (1995) found that historical physical activity was a positive determinant of hip BMD in men aged 50 to 88, but that current leisure or occupational activities were not influential. Greendale and colleagues (1995) also reported a significant linear trend in older men between both lifetime and current exercise and hip BMD, although no significant relationship was found at other sites. Neither was there a relationship between osteoporotic fracture rate and exercise profile, a reminder that maintenance of bone mass is arguably not a primary clinical or functional goal in and of itself. In fact, BMD maintenance is largely a method of achieving the more practical goal of minimizing risk of fracture.
4. Exercise-Related Geometric Adaptation
Expanded diaphyseal diameters are frequently seen in dominant side limbs of athletes who preferentially load them. Krahl and colleagues (1994) observed significant differences in diameter and length of playing arm ulnae of tennis players compared to their contralateral arms. The second metacar- pals of playing hands were also wider and longer than in contralateral hands. No differences were observed between limbs of controls. Dalen and associates (1985) observed a 27% difference in cortical cross-sectional area between left and right humeri of tennis players compared to a nonsignificant 5% difference in controls. Significant differences between playing and nonplaying arm humeral cortical wall thickness, length, width, and cross-sectional moment of inertia were likewise observed by Haapasalo and associates (1996).
E. Exercise Effects—Intervention Trial Findings
Prospective studies designed to expose randomly selected, previously untrained subjects to exercise are a more rigorous and valid method of ob
serving the effect of exercise on BMD than cross-sectional designs. The inherent difficulties of subject recruitment and compliance associated with exercise intervention trials are, however, reflected in the substantially reduced volume of reports in the literature.
I. Adult Men
One prospective trial concluded that there were no significant differences between calcaneal BMC of consistent runners, inconsistent runners, and nonexercising controls after 9 months of marathon training (Williams et al., 1984). A closer analysis of the data, however, revealed that consistent runners indeed increased calcaneal BMC and that the amount of change in BMC was significantly different from controls. The correlation between average distance run and percent change in BMC also indicated a strong positive relationship.
Army recruits completing 14 weeks of intensive physical training have been observed to increase right and left leg BMC by 8.3 and 12.4%, respectively (Margulies et al., 1986). In a similar trial, a 7.5% increase in tibial bone density was observed in Army recruits after 14 weeks of basic training (Leichter et al., 1989). Recruits who began the training period with the lowest bone density gained the greatest amount. Those who temporarily ceased training due to stress fracture also gained bone density, but to a lesser degree (5%). It is an interesting question to consider whether reduced bone density contributed to the incidence of stress fracture in the injured group, or if the cessation of training reduced the opportunity to accrue a similar amount of bone as those who completed training. Also notable is the fact that 10% of recruits actually lost bone density. This effect was likely to stem from resorption-related remodeling porosity which had not yet been matched by replacement formation owing to the short time frame of the study. The discrepancy in results across recruits highlights the phenomenon of substantial individual variation in bone adaptation response to exercise loading.
The effects of 4 months of high-intensity resistance training three times per week on the bone metabolism of 23- to 31-year-old Asian males were investigated by Fujimura and associates (1997). They found that indicators of bone formation (serum osteocalcin concentration and serum bone-specific alkaline phosphatase activity) were increased within a month of initiating training and remained elevated throughout the training period. Markers of bone resorption (plasma procollagen type I and urinary deoxypyridinoline) were never significantly elevated. Although the findings led the authors to conclude that resistance training stimulated bone formation but not bone resorption, no significant changes in BMD were evident following training to confirm the assertion.
The influence of training intensity on bone response becomes apparent when the results of army trials are compared with those of Dalen and Olsson (1974). In the latter trial, subjects aged 25 to 52 failed to gain bone mass at the forearm, spine, humerus, femur, or calcaneus after 3 months of either walking (3 km, 5 days a week), or running (5 km, 3 days a week). In comparison, the bone mineral gains observed in the aforementioned army recruits reflect the considerably more intense nature of exercise loading during basic army training.
2. Older Men
Results from exercise intervention trials with older men are conflicting, with some suggestion that exercise does not strongly stimulate bone accretion in this population. McCartney and associates (1996) reported that 42 weeks of weight training in 60 to 80 year olds increased muscle strength and functional ability but caused no changes in whole body and spine BMD. Conversely, Welsh and Rutherford (1996) found that trochanteric BMD increased significantly in 50- to 73-year-old men performing step and jumping exercises. The effect is likely to be related to the enhancement of the strength of hip extensor muscles inserting at that site. Although gluteal muscle strength was not measured, quadriceps strength was increased after 12 months of exercise.
Sixteen weeks of progressive resistance training of 64- to 75-year-old men, with or without growth hormone supplementation, had no significant effect on BMD of either growth hormone treated or placebo groups. (A minor BMD increase only at Ward’s triangle in the placebo group, and minor decreases in whole body and femoral neck BMD in the growth hormone treatment group are of questionable significance) (Yarasheski et al., 1997). Similarly, 24 weeks of exercise intervention, with or without growth hormone, increased muscle strength (independent of growth hormone) but effected no change in BMD of older men (Taaffe et al., 1994, 1996, personal communication, 1998).
VI. Unloading Bone—In Brief
Although a thorough review of the subject is beyond the scope of the present post, it is appropriate to mention the effects of skeletal unloading— the opposite extreme of the loading continuum to chronic exercise. In keeping with Wolff’s Law, unloading bone provokes the converse reaction to that of loading it. A substantial body of evidence exists to support this claim, particularly regarding the effects of spinal cord injury, prolonged bed rest, and limb immobilization on BMD. Paraplegic and quadriplegic patients may lose more than 2% of lower extremity bone mass for the first 4-6 months after injury, thereafter losing approximately 1% a month for the remainder of the first year (Kiratli, 1996). Krolner and Toft (1983) observed patients who were hospitalized at bed rest for an average of 27 days. The average decrease in BMD during bed rest was 3.6%, equivalent to about 0.9% bone loss per week. After an average of 15 weeks of reambulation, an average gain of 4.4% BMD was observed. Even more dramatic descriptions of bone loss have been reported in astronauts exposed to microgravity.
VII. Hormonal Factors
A. Acute Exercise Response
Acute hormonal perturbation has been observed in individuals under exercise stress. Significant increases in serum concentrations of testosterone, lutenizing hormone (LH), dehydroepiandrosterone (DHEA), follicle stimulating hormone (FSH), cortisol, prolactin and androstenedione have been observed during maximum endurance exercise (Cumming et al., 1986; MacConnie etal., 1986).
Markers of bone turnover and parathyroid hormone (PTH) are also acutely stimulated by exercise (Brahm et al., 1997). Rong and associates (1997) found that strength exercise increased acute levels of PTH and that both strength and prolonged endurance exercise caused a pronounced decrease in type I collagen telopeptide (marker of bone resorption).
B. Chronic Exercise Response
The ability of chronic exercise to substantially modify hormone balance has not been shown. Investigators have found that male athletes exercising at a range of intensities appear to have serum concentrations of testosterone which lie within the normal range (MacConnie et al., 1986; Suominen and Rahkila, 1991; MacDougall et al., 1992; Hetland et al., 1993; Smith and Rutherford, 1993), including adolescents (Rowland et al., 1987). Others factors, such as LH, FSH, prolactin, cortisol, and estradiol (Rogol et al., 1984; Wheeler et al., 1984; MacConnie et al., 1986; Hackney et al., 1988) have also been found to circulate at normal levels in male athletes.
Increased concentrations of parathyroid hormone have been observed in response to a maximal exercise test before and after 6 weeks of endurance training (Zerath et al., 1997). Before training, the exercise test effected increased circulating osteocalcin (a marker of bone formation) but not after. In fact, osteocalcin concentrations decreased significantly after training. This finding may further illustrate the characteristic of diminishing returns in terms of exercising for bone mass. That is, greatest changes are seen in bone which has not previously undergone load-related bone adaptation. These responses clearly require further investigation, however, because Rong and associates (1997) found that prolonged endurance exercise increased osteo calcin levels. Long-term exercise has not been shown to preserve the decline in activity of the growth hormone-IGF-I axis (Cooper et al., 1998).
These generalizations notwithstanding, a degree of subtle hormonal per-turbation may be evident in some athletes. Smith and Rutherford (1993) found that, while in the normal range, serum total testosterone was significantly lower in triathletes than controls, but not rowers. Further, Wheeler and associates (1984) found that total serum testosterone, non-sex hormone binding globulin (SHBG)-bound testosterone, and “free” testosterone concentrations in men running more than 64 km per week to be 83, 69.5, and 68.1% that of controls, respectively. Prolactin concentrations were also significantly lower in runners than controls. In contrast, Suominen and Rahkila (1991) found that endurance athletes had significantly greater levels of SHBG than strength athletes and controls, and Cooper and associates (1998) showed substantially increased circulating SHBG in elderly long-term runners than in age-matched nonexercising men. Hackney and associates (1988) also found resting and free testosterone concentrations of trained athletes to be 68.8 and 72.6% that of controls and LH to be slightly higher in athletes than controls.
The implications of exercise-related hormonal perturbation to bone mass is somewhat unclear. Suominen and Rahkila (1991) detected a negative correlation between BMD and SHBG in endurance athletes but no relationship of BMD with testosterone. Interestingly, Fiore and associates (1991) found that intense body-building training and self-administered anabolic steroids (testosterone: 193.75 ± 147.82 mg/week) did not stimulate greater osteoblastic activity or bone formation than exercise alone. Given sparse data from long-term intervention trials, a connection between exercise, hormone status, and bone metabolism remains difficult to make.
VIII. Sex Comparison of Exercise Effect on Bone
Once again, there is only a modicum of data comparing male and female responses to exercise intervention, and information regarding an interaction of age with these responses is essentially nonexistent. Welsh and Rutherford (1996) observed the effect of 12 months of high-impact aerobics two to three times a week on hip, spine, and total body BMD of elderly men and women. They found that both men and women increased BMD a similar amount (men, approximately 1.2%; women, approximately 1.3%) at all sites, whereas controls lost BMD.
Even though evidence suggests that exercise training improves BMD in women of all ages, and it is likely that these results are indicative of the male response to exercise, the known influence of hormones on bone mass precludes premature inferences from gender-specific findings.
The common prescription of walking three to four times per week for bone health has not altered over the past 20 years (Sidney et al., 1977; Katz and Sherman, 1998). Even though a recommendation to walk is certainly appropriate for the purposes of general health maintenance, available data specific to bone gain or maintenance do not necessarily support this belief (Dalen and Olsson, 1974; Hutchinson et al., 1995). Low-impact activities may, in fact, be relatively ineffective for the purposes of increasing or maintaining bone mass in any age group, with the exception of the very young. While Michel and associates (1989) reported that up to 300 minutes of weight-bearing exercise, including walking, was linearly related to BMD in older men, a causal relationship was not established. The only other evidence for walking as a bone stimulus comes from research involving postmenopausal women, where walking more than 7.5 miles per week was associated with higher whole body, leg, and trunk BMD than walking less than 1 mile per week (Krall and Dawson-Hughes, 1994).
High-intensity resistance training or relatively high-impact weight-bearing exercise appears to impose the greatest stimulus on bone. It is something of a conundrum to design an exercise program which sufficiently loads the skeleton without placing it at risk of impact-related fracture, particularly in an osteoporotic population. In addition, running and other higher-impact activities may increase the risk of falls and possible fracture. Such injury is to be avoided at all costs, given the negative repercussions of prolonged immobilization and/or bed rest on BMD and health in general. Because exercise is known to enhance balance and neuromuscular function, activities of higher impact than walking should be given at least passing consideration as a viable option for some individuals. Examples of the controlled circumstances under which a running regimen may be effectively implemented include: establishing an adequately graduated program of increasing intensity (beginning with walking in most cases), wearing appropriate footwear and exercise clothing, obtaining clearance for other medical conditions for which high-intensity exercise is contraindicated, and monitoring exercise bouts by family members or friends. Running and other high-impact activities are not recommended for those suffering from grossly compromised skeletal components such as advanced vertebral osteoporosis.
Resistance training is also a viable option for bone health maintenance. The benefit of weight lifting resides primarily in the opportunity to load non-weight-bearing bones in a controlled exercise environment that offers minimal risk for falling. It could also be argued that, even in the absence of bone gain with resistance training, the benefits of muscle strength gains for the purposes of reducing fracture risk are sufficient in and of themselves to warrant such a recommendation. It has been recommended that resistance-training loads be chosen for their ability to induce muscle fatigue after 10 to 15 repetitions and should be increased gradually. A resistance workout should be performed approximately every 3 days to allow time for muscle recovery. Rowing machines should be avoided by individuals suffering from osteo-porosis owing to the risk of vertebral fracture from deep forward bending. Resistance exercise is not optimally recommended for individuals suffering from hypertension as transient increases in blood pressure associated with forceful large muscle group contractions may be hazardous.
There is no such thing as a “one-size-fits-all” exercise prescription for the purposes of bone gain or, indeed, any other physiological function. It is, however, safe to say that individuals who currently exist in primarily sedentary lifestyles have much to gain by increasing their levels of activity by any degree. For those who fit such a description, moderate-intensity walking of increasing frequency and duration to approximately 30 to 60 minutes, four times per week, in combination with a generalized increase in incidental activity (for example, leaf raking instead of blowing), is likely to confer substantial benefits on bone health. An emphasis on activities which improve general muscle strength, flexibility, and coordination is highly recommended, given the attendant reduced risk of falling associated with these attributes.
For those who are already somewhat active, higher-intensity, impact ex-ercises such as running up and down hills and stairs, aerobics, and jump rope are “bone-friendly” activities. Graduated increments in exercise training intensity are recommended to allow adequate time for bone remodeling and avoid bone stress injury. Resistance training of muscle groups in all regions of the body (triceps surae, quadriceps, hamstrings, gluteals, iliopsoas, erector spinae, abdominals, chest and upper back, biceps brachii/brachialis, triceps, etc.) is also recommended. Novel incidental or recreational exercises (e.g., carrying shopping bags, yard work, arm wrestling, tug-of-war) which load bones in an unusual manner or target non-weight-bearing portions of the skeleton may additionally assist in the maintenance of skeletal mass in moderately active individuals. The essential factor in the prescription of exercise for bone mass maintenance is the recommendation that it be pursued throughout the lifespan and that long periods of immobilization or inactivity are avoided.
We have reviewed the conceptual basis for understanding the relationship between mechanical loading and skeletal adaptation, along with the specific effect of exercise on bone mass in men. Complete understanding of the relationship of exercise to bone mass in humans must await development and validation of accurate, quantitative estimates of mechanical loading history. Although this has not yet been accomplished, sufficient information is available to permit general conclusions, as well as speculation about the kinds of exercises that are likely to prove most osteogenic.
Cross-sectional studies, in general, provide support for the notion that habitual athletic endeavor promotes superior bone density in men compared with that of a sedentary life-style. The magnitude of this difference is likely to depend on the nature of the activity, the age at which it was initiated, and the number of years spent in training. Physical activity may enhance peak bone mass attained if initiated before the age of 20. Muscle mass and strength is likely to contribute positively to BMD at this age, as well as during later years. Bone displays a clear site specificity for mechanical load-induced adaptation at all ages.
Exercise intensity may be an important factor in the stimulation of bone adaptation. Moderate- to high-impact weight-bearing activities such as running, jumping, and power lifting appear to be the most bone stimulatory. Walking and long-duration swimming are less effective, with the exception that swimming may have a positive impact on bones not normally loaded during weight bearing.
Although there are inadequate data upon which to make decisive con-clusions, the ability of exercise to stimulate significant gains in bone mass in older males appears to be reduced. The strong positive effect of exercise on muscle strength at all ages, however, suggests that exercise indirectly benefits bone, even in older males, as a function of improved or maintained balance, coordination, and related reduction in the risk of falling.
The response of bone to exercise loading is curvilinear in nature. The greatest increments in bone mass are generally observed in individuals with the lowest initial values. It is undoubtedly true that unloading the skeleton for extended periods is detrimental to bone. The addition of even modest activity to an immobilized subject will increase BMD to a much greater extent than will a substantial increase in training for a highly active person. Therefore, one might best consider physical activity to be an effective prevention against bone loss, rather than a means to achieving major increases in bone mass.