The purpose of this review is (i) to describe the changes in skeletal size, mass, and internal architecture that occur during growth and aging in men and women of different racial groups and (ii) to describe the differences in skeletal size, mass, and internal architecture found in men with fractures relative to men without fractures. Because the modeling and remodeling of the periosteal and endosteal (endocortical, intracortical, and trabecular) surfaces during growth and aging determine the external and internal structure of bone, insight into the mechanisms responsible for the development of bone fragility in men and women can be gained by comparing and contrasting the age-, gender-, and race-specific patterns of modeling and remodeling that occur on these bone surfaces.
The growth of the periosteal surface defines the external bone size—an independent determinant of bone strength. The growth of the endocortical surface relative to the periosteal surface determines cortical thickness. The subsequent expansion of the periosteal surface and endocortical remodeling during aging determines the extent of cortical thinning and the distance of the mass of cortical bone relative to the neutral axis of the long bone in old age. The development of trabecular numbers during growth, and the thickening of the trabeculae during pre- and peripubertal growth establishes peak trabecular bone density and the size of surface available for remodeling during aging. The subsequent intensity of remodeling and the remodeling imbalance between bone formation and resorption within each BMU on the trabecular surfaces during aging establish the extent of trabecular bone loss and the degree of trabecular thinning, loss of trabecular connectivity, and so trabecular bone fragility.
II. Comparing Men and Women of Different Races
Fractures may occur less commonly in men than in women because bone fragility is less or trauma is less common or less severe. The mechanisms that are more likely to contribute to the lower bone fragility in men than women include the following: (i) attainment of a higher peak bone mass and size at the completion of growth, (ii) less bone loss as a percentage of the (higher) peak bone mass in men, (iii) trabecular bone loss by thinning caused by reduced bone formation in men (Trabecular plate perforation and loss of connectivity is primarily the result of a menopause-related increase in the extent of remodeling and perhaps resorption depth in women.), (iv) less endocortical resorption, (v) greater periosteal expansion during aging, thus increasing bone size and strength, and counteracting cortical bone thinning due to endocortical resorption, and (vi) perhaps less intracortical porosity (Seeman, 1994, 1995).
A. Growth in Size, Mass, and Volumetric Density of the Axial Skeleton
Males have bigger bones than females, and bone size is an independent determinant of bone strength (Ruff and Hayes, 1988). Whether gender and racial differences in bone size are present before birth or shortly thereafter is uncertain. Rupich et al. (1996) suggest gender and ethnic differences in total body bone mineral content (BMC) and areal bone mineral density (BMD) are present in infants aged 1 to 18 months. Gilsanz et al. (1988), using quantitative computed tomography (QCT) in a study of 196 healthy children aged 4 to 20 years, reported that the cross-sectional area of vertebral bodies was 17% greater in boys at Tanner stage I and higher throughout childhood and adolescence. There were no gender differences in vertebral height. By contrast, blacks matched by height have longer legs and shorter sitting height than whites. Vertebral height was less in black men and women when com-pared to their white counterparts. Vertebral width was similar in black and white men and in black and white women (Gilsanz et al., 1998). This suggests that there may be race-specific factors regulating vertebral height and gender-specific factors regulating vertebral width.
Areal BMD is greater in the spine in men than women in large part because vertebral width (not height) is greater. The greater amount of bone gained during growth in men than women builds a bigger skeleton, but not necessarily a denser skeleton. For volumetric BMD (the amount of bone contained within the bone) to increase during growth, the increase in bone mass (relative to the population mean for bone mass) must be greater than the increase in bone size (relative to the population mean for bone size). This may occur in predominantly trabecular structures such as the vertebral body by increasing trabecular numbers or thickness or by increasing the true (material) density of the trabeculae themselves (Seeman, 1998).
Volumetric BMD is no different in men and women of the same race (i.e., at peak, trabecular number and thickness are the same in white men and women and in black men and women). Gilsanz et al. (1994), reported no differences in cancellous or cortical BMD in 25 women and 18 men aged between 25 and 46 years. Vertebral bodies in women had a lower crosssectional area (7.9 ± 1.1 versus 10.9 ± 1.3 cm2, P < 0.001) and volume (22.4 ± 2.4 versus 30.9 ± 2.6 cm P < 0.001) than men. As a consequence, mechanical stresses within vertebral bodies were predicted to be 30-40% greater in women than in men. Similarly, neither trabecular number nor thickness differ in South African black men and women nor in Japanese men and women (Schnitzler et al., 1990). Fugii et al. (1989), using QCT, showed that Japanese men and women have similar trabecular volumetric BMD. By contrast, volumetric BMD is greater in blacks than whites of the corresponding gender because blacks have thicker trabeculae than whites (Han et al., 1996; Parfitt, 1998). Thus, males may have greater peak vertebral bone strength than females of the same race as a result of greater vertebral width, not vertebral BMD. Blacks may have greater peak vertebral bone strength than whites of the corresponding gender because of greater trabecular BMD, despite the smaller vertebral size. (Blacks have a shorter vertebral body containing more bone as a result of thicker trabeculae.)
Before puberty, trabecular volumetric BMD at the spine is similar in boys and girls, blacks and whites, and is independent of age during the prepubertal years. During puberty, trabecular volumetric BMD increases comparably in males and females of a given race but increases more greatly in blacks than whites (Gilsanz et al., 1998). The increase in trabecular BMD in males and females, as well as the greater increase in trabecular BMD in blacks than whites, is likely to be the result of increased trabecular thickness, not numbers (Han et al., 1996; Parfitt, 1998). Fugii et al. (1989), using QCT, showed that Japanese men and women had lower trabecular BMD than their white gender counterparts. The authors suggest that the differences were greater than can be explained by differences in the methods of measurement used in the two countries. Whether the Japanese have thinner or fewer trabeculae than whites is unknown because no histomorphometric data are available.
Thus, trabecular numbers are independent of race and gender (trabecular numbers are the same in males and females, blacks and whites). Trabecular thickness is independent of gender, being similar before puberty and increasing by a similar amount in males and females at puberty. Trabecular thickness is race-specific, being greater in blacks than whites and probably greater in whites than Japanese. The similarity in thickness in both genders suggests that estrogen may be the common factor regulating trabecular endosteal formation in males and females. Why does trabecular thickness increase more greatly in blacks than whites at puberty (and perhaps more greatly in whites than Japanese at puberty)? Why is vertebral height less and leg length greater in blacks? Why is leg length less in Japanese than whites, whereas trunk length is similar? An increased sensitivity or early exposure to estrogen in blacks may result in earlier fusion of epiphyses, producing shorter vertebra with thicker trabeculae. However, this would produce shorter legs in blacks than whites. Early exposure to estrogen produces smaller bones with higher volumetric BMD in animals (Migliaccio et al., 1996).
B. Growth in Size, Mass, and Volumetric Density of the Appendicular Skeleton
The longer (2 years) prepubertal growth in boys, the more rapid pubertal growth spurt (reaching 10-12 cm/yr in boys and 8-10 cm/yr in girls), and the longer duration of puberty in boys all contribute to size differences. Only 3 cm of the 13-cm difference in height between men and women is attributable to pubertal growth, 10 cm is attributable to prepubertal growth, and most of this difference is in leg length (Cameron et al., 1982; Preece et al., 1992). Thus, on average, men are taller than women because they have longer legs rather than longer trunks. Similarly, black men matched by height with white men have longer legs (Gilsanz et al., 1998). Whether the greater length is present at birth or emerges prior or during puberty is unclear.
Periosteal growth accelerates at puberty in males, enlarging bone diameter. In females, periosteal diameter ceases to expand at puberty, whereas endocortical contraction narrows the medullary cavity. Any existing gender difference in bone width before puberty increases further at puberty by this mechanism. Thus, bone length is greater in males because of the longer prepubertal and intrapubertal growth, whereas bone width is probably greater because of the androgen-mediated increase in periosteal expansion. Similarly, long bone width is greater in blacks than whites because of greater width at birth and/or greater periosteal expansion before or during puberty. Why blacks have wider and longer femurs than whites of the corresponding gender is uncertain.
There is little evidence to support the notion that cortical thickness is greater in men than women or greater in blacks than whites. On the contrary, femoral cortical thickness is similar both in young men and women and in blacks and whites. Gilsanz et al. (1998) reported that white boys and girls have the same cortical thickness at the midfemur. Similarly, cortical thickness of the midfemur is similar in black males and females and is no different from their white counterparts.
Cortical thickness is the net result of the relative growth of the periosteal and endocortical surfaces. Before puberty, periosteal expansion proceeds rapidly, whereas endocortical (medullary) diameter expands modestly and then contracts, perhaps more in females than in males. Cortical thickness does not differ in males and females because the greater periosteal expansion in males is accompanied by greater endocortical expansion before puberty and less endocortical contraction during puberty. Females achieve the same cortical thickness as males because 25% of final thickness is the result of endocortical contraction during puberty, whereas 75% is the result of periosteal expansion. Cortical thickness in males is largely the result of periosteal expansion.
Similarly, for blacks to have the same cortical thickness as whites (despite the larger bone diameter), either endocortical expansion before and during puberty must be greater or endocortical contraction at puberty must be less in blacks. Garn et al. (1972) studied 4379 whites and 1589 blacks. Black men had 7% higher subperiosteal, 30% higher medullary, and 3% higher resultant cortical areas than whites. Black women had 14% higher subperiosteal, 49% higher medullary areas, and 7% higher cortical areas than white women. Whether differences of this magnitude account for the ethnic and gender differences in fracture rates is unknown. By contrast, in a study of 950 South African blacks and 782 whites, Solomon (1979) found that blacks had lower cortical area than whites (despite a lower fracture incidence).
Femoral midshaft BMC and areal BMD increase during growth because size increases. Proximal femur BMC and areal BMD are higher in men than women and higher in blacks than whites because the femur is longer and wider in men than women and in blacks than whites. For predominantly cortical structures like the femur or radius, the volumetric BMD will increase if the amount of bone in the growing bone increases more (relative to its population mean) than the increase in external volume (relative to its population mean). This may occur by increasing cortical thickness or increasing the true density of the cortical bone. Midshaft femoral volumetric BMD is constant during growth, even during puberty. Similarly, radial volumetric BMD is constant during growth (Zamberlan et al., 1996). (Note that vertebral trabecular BMD increases at puberty.)
This constancy implies that the increase in size is matched by a commensurate increase in mass within the periosteal envelope of the growing long bone. Although the bones in males and females differ in size, they do not differ by volumetric BMD (Lu et al., 1996). Although vertebral volumetric BMD is greater in blacks than whites, midshaft volumetric BMD was not higher in blacks than whites in the study by Gilsanz et al. (1998). This information suggests that any difference in bone fragility in childhood and early adulthood may be a function of the gender and racial differences in bone size rather than “density.”
Thus, to summarize, femur length and width is greater in men than women and greater in blacks than whites. Cortical thickness is independent of gender and race. The greater diameter of the midfemur in the male than female and in blacks than whites results in the larger bone having a greater perimeter, and so the greater mass of cortical bone is placed farther from the neutral axis of the long bone conferring greater bone strength in men than women and in blacks than whites. Vertebral size is greater in men than women because vertebral width, not height, is greater. Men and women have the same trabecular number and thickness; at peak, bone strength is greater because bone size is greater. Vertebral size is less in blacks because vertebral height is less; width is similar. Blacks have thicker trabeculae; the vertebrae are smaller in blacks, but trabecular BMD is greater because the trabeculae are thicker.
The surfaces that form these dimensions and structures behave differently because they are regulated differently. Comparative studies within and between genders and races is likely to give insight into the genetic and environmental factors regulating these surfaces. An understanding of the hormonal regulators of periosteal and endocortical growth and remodeling in men and women and blacks and whites may contribute to the development of new drugs that increase periosteal growth (increasing the bending strength of cortical bone), increase endocortical apposition (increasing cortical thickness), or reduce endocortical resorption (preventing cortical thinning).
C. Delayed Puberty
Delayed puberty in males may result in increased femur length because of delayed epiphyseal fusion. Bone width may be reduced because periosteal growth is androgen-dependent. Whether delayed puberty results in reduced volumetric BMD is uncertain. Moore et al. (1997) report normal volumetric BMD in adult males with a history of delayed puberty. If volumetric BMD is reduced, this must be the result of reduced cortical width in long bones which may be the result of continued endocortical expansion despite reduced periosteal expansion (due to androgen deficiency) or of failed endocortical contraction (a process that may be dependent on estrogen synthesis from testosterone in males).
Finkelstein et al. (1992, 1996) suggest that men with constitutionally delayed puberty may have a low peak areal BMD in adulthood. In the first study, there were two control groups, 21 men 2 years younger and 39 men 2 years older than the cases. Lumbar spine areal BMD in the cases was 1.03 g/cm2—0.10 g/cm2 less than the younger controls and 0.05 g/cm2 less than the 60 controls combined. Results for the 39 older controls were not provided but must have been lower than the younger controls to bring the mean from 1.13 g/cm2 in the younger controls to 1.08 g/cm2 for all 60 controls; lumbar spine areal BMD should be about 1.03 g/cm2 in the older controls, which is no different than the cases. In the follow-up study conducted 2 years later in 18 men, radial and spinal areal BMD were reduced. Femoral neck areal BMD was lower than in the controls: 0.88 ± 0.11 versus 0.98 ± 0.14 g/cm2 (P < 0.02).
These subjects may have suffered from hy- pogonadotrophic hypogonadism rather than delayed puberty; they exercised 48 ± 104 miles/week, 35% ran more than 15 miles/week, and 57% were regular weight lifters. Whether the deficits are the result of reduced bone size is also unclear (Seeman, 1997,1998). If delayed puberty reduces bone width, areal BMD may be reduced because of the smaller size. If bone length is increased (because of delayed epiphyseal closure), BMC or areal BMD may be higher or no different than the controls. For example, Luisetto et al. (1995) reported that 42 patients with Klinefelters’ syndrome had normal areal BMD (z scores: lumbar spine, 0.5 SD; femoral neck, 0.002 SD; total femur, 0.2 SD). Failure to account for size may have resulted in finding no deficit at the proximal femur (probably a larger bone than in the controls).
III. Changes in Bone Size, Mass, and Volumetric Density during Aging
Bone remodeling occurs on the trabecular surfaces, on the endocortical surfaces, within the cortices of bone, and on the periosteal surfaces. The purpose of remodeling is to maintain bone strength. Remodeling imbalance, the failure to replace the old bone with the same amount of new bone, is the morphological basis of bone loss. Remodeling imbalance (i) on trabecular surfaces results in trabecular thinning, perforation, and loss of connectivity; (ii) on the endocortical surface results in cortical thinning; and (iii) within cortical bone results in increased cortical porosity. Periosteal appositional growth partly offsets the bone loss occurring on the endosteal surfaces.
A. Trabecular and Cortical Bone Loss
The amount of trabecular bone lost in women and men is similar whether assessed by histomorphometry of the iliac crest or quantitative computed tomography of the spine (Kalender et al., 1989; Meunier et al., 1990). Trabecular bone loss occurs mainly by thinning in men and mainly by loss of connectivity in women (Aaron etal., 1987). Loss of connectivity may be less in men than in women because there is no comparable menopause-related increase in remodeling intensity. In women, loss of trabeculae occurs because of the increased surface extent of remodeling and may produce perforation and complete loss of trabeculae. As trabeculae are lost, there is less trabecular surface available for remodeling, and trabecular bone loss slows. In men, progressive thinning of trabeculae may increase the trabecular surfaces available for remodeling. This may result in continued trabecular bone loss in men. Studies of vertebral trabecular bone loss in elderly men and women using QCT are not available.
Bone loss from the proximal femur is detected soon after attainment of peak areal BMD in cross-sectional studies using densitometry in men and in women. In part, the changes reported in cross-sectional studies may be an artifact of a fall in cellularity and an increase in adipose cells of the marrow in the proximal femur (Kuiper et al, 1996). Studies are needed to clarify when bone loss commences at this site. Trabecular bone may reach its peak earlier than cortical bone and may start to decline while cortical mineral accrual and consolidation is still occurring. The age of attainment of peak areal BMD and commencement of bone loss may vary by race (Looker et al., 1995).
Cortical bone loss is the result of endocortical remodeling imbalance, intracortical remodeling imbalance and periosteal apposition. Thus, the same amount of bone may be lost by fundamentally different mechanisms in males and females, in different races, and in the same individual at different times of life. For example, the 150 g less total bone “loss” in men than women is the net result of less endocortical bone resorption in men than women, greater periosteal apposition in men than in women, and less intracortical porosity in men than women.
The greater periosteal apposition in men than in women results in a bone with a larger cross-sectional area. Ruff and Hayes (1988) studied 99 tibia from 73 individuals and 103 femora from 75 individuals. Changes per decade included the following, (i) Medullary area (reflecting endosteal resorption) increased by 7% in men and 8% in women, (ii) Subperiosteal area (reflecting periosteal deposition) increased by 2.5% in men and 1.1% in women, (iii) Cortical area decreased by 1.6% in men and 7% in women, (iv) The polar second moment of area (bending rigidity), increased by 2.1% in men but declined by 3.3% in women.
Secular trends may obscure a true increase in periosteal diameter. Height and bone width have increased in the last 70 years. In a cross-sectional study, this age-related increase in bone width in (earlier born) current 80 year olds may bring bone width to equal the (later born) current 20 year olds. Femoral width increases more greatly in men than in women across age (Y. Duan and E. Seeman, unpublished data, 1998). However, secular increases in bone width have been reported in women in other studies (Looker et al., 1995). The differing observations in cross-sectional studies may reflect either measurement error or the heterogeneous nature of secular changes in growth. Depending on the community studied, increases are found in one or both genders and in one or both upper and lower body segment lengths (Bakwin, 1964; Meredith, 1978; Cameron et al., 1982; Tanner et al., 1982; Malina and Brown, 1987).
Prospective studies suggest that bone loss accelerates, rather than decel-erates, in old age. Ensrud et al. (1995) showed that rates of decline in areal BMD at the proximal femur, measured in 5698 community-living white women aged over 61 years, increased fivefold in women 70 to 85 years old. Jones et al. (1994) suggest that rates of bone loss at the proximal femur increase with advancing age, based on 241 men and 385 women followed an average of 2.5 years (range 1 to 4 years). More rapid rates of bone loss were not found at the lumbar spine, perhaps because of coexistent osteoarthritis and, in part, because of the loss of trabecular surface. Hannan et al. (1994) found similar rates of diminution (percent per year) in 437 men and 698 women 68 to 98 years old: —0.69 ± 0.15 versus —0.68 ± 12 (femoral neck); —0.45 ± 0.17 versus —0.53 + 0.15 (trochanter); and —0.88 ± 0.23 versus —0.94 ±0.12 (Wards triangle).
Secondary hyperparathyroidism is likely to contribute to accelerated cortical bone loss in men and women. Increased endocortical and intracortical remodeling increase the surface available for resorption in cortical bone and may explain the increasing rate cortical bone loss in the elderly. The increased numbers of sites undergoing remodeling on the progressively increasing surface contributes to accelerating bone loss because of the negative bone balance within each remodeling unit. Reduced cortical areal BMD is also the result of increased porosity. Laval-Jeantete^a/. (1983) report cortical porosity of the humerus increased from —4% in white men and women aged 40 years to ~ 10% in over 80 year olds. The fall in apparent density with age correlated with porosity. True mineral density (ash weight per volume unit of bone free of vascular channels) was unchanged.
B. Relative Contributions of Peak Bone Mass and Bone Loss to Bone Mass in Old Age
Men have a net gain of 1200 g calcium to build their skeleton and lose 100 g net ( — 8%). Women have a net gain of 900 g calcium and lose 250 g ( — 30%). Thus, bone mass in old age, when fractures occur, is determined more by the amount of bone gained during growth than lost during aging. Compared to women, men gain 300 g more calcium during growth to build a bigger skeleton than women (1200 — 900 g). Because the difference in net loss is 150 g (250 — 100 g), of the 450 g (1100 — 650 g) greater total bone calcium in elderly men than women, 300 g is attributable to the greater net gain, whereas only 150 g is attributable to the lesser net amount of bone lost during aging in men. Of this lesser amount lost, a proportion will be attributable to greater loss in women than men, and the remainder will be attributable to greater bone formation in men.
Because puberty occurs about 2 years later in boys than girls, the pre-pubertal contribution to total BMC in young adulthood in males is 80%, the pubertal contribution is 20%. In girls, prepubertal and pubertal growth each contribute about 50% of the total BMC in adulthood (Gordan et al., 1991). Thus, a smaller proportion of total BMC at maturity may be sex-hormone- dependent in men than women. If so, then delayed puberty or hypogonadism during growth may be less deleterious in males than in females. If the amount of bone loss caused by hypogonadism is a function of the amount of bone gained during puberty, then hypogonadism during aging in males should be less deleterious than in females.
Comparisons of different racial/ethnic groups suggest that differences in total BMC in old age—when fractures occur—are constituted by different combinations of bone gain and loss (Looker et al., 1995). For example, comparing the total BMC of the proximal femur in old age in whites, blacks, and Mexican-American men, blacks have higher BMC than whites in old age because they gain more and lose similar amounts. Blacks have higher BMC than Mexican-Americans because they gain more but lose more, attenuating the difference in peak BMC. White men gain more than Mexican-Americans but lose more, and so Mexican American men have higher BMC than whites in old age. The gains and losses are net changes; the relative contributions of the endosteal and periosteal surface modeling and remodeling to both peak BMC and net bone loss are unknown.
The changes are region-specific (Perry et al., 1996). Data are available in women, not men. Upper body calcium net losses were 48 g in blacks and 70 g in whites, whereas lower body net losses were 146 g in blacks (14%) and 160 g in whites (16%). Thus, net bone loss is greater in whites than blacks in absolute terms and as a percentage of their (lower) peaks. Net loss relative to peak at the upper body in whites (70 g, 24%) was twice that in blacks (48 g, 14%), but similar net losses occurred in the lower body. The higher BMC in blacks than whites in the postmenopausal years is largely accounted for by the greater net gain in BMC by blacks during growth than by the greater net loss in whites during aging. How much of the greater net loss in whites than blacks is accounted for by more endocortical resorption in whites than blacks and/or more periosteal gain in blacks than whites is unknown.
C. Hip Axis Length
Hip axis length (HAL) is purported to be an independent predictor of hip fracture. These data are based mainly on studies in women. There is little evidence that men with hip fractures have shorter HAL than age-matched controls. Gomez (1994) studied 188 men and 300 women with hip fractures aged 75 ± 7 years. There was no difference in HAL between men or women with hip fractures relative to gender-matched controls. The neck-shaft angle was lower in the fracture cases. Femoral neck cross-sectional area was higher in the men with hip fractures (12.1 ± 2.3 cm) than in the controls (9.6 ± 2.1 cm, P < 0.001). This is an unusual observation because greater cross-sectional area is associated with greater bone strength. HAL was greater in men than women.
The finding of a shorter HAL in women compared to men is difficult to reconcile with the higher hip fracture rate in women. Racial differences in hip fracture incidence are also attributed to differences in HAL. Blacks have longer legs and shorter trunks than whites. The finding of a shorter HAL in blacks may be a conservative error; HAL may be even shorter if adjustment is made by leg length rather than total height. Asians have similar trunk length but shorter leg length than whites. A shorter HAL reported in Japanese may not be observed after adjustment for total height compared to whites after adjustment for leg length. There are secular trends in upper and/ or lower body segment lengths in whites, blacks, and Asians. Secular increases have been reported in upper and lower body segments in females and males, and secular increases in HAL may parallel the changes in segment lengths (referenced earlier). Whether they are responsible for the increase in hip fracture incidence is uncertain. Whether HAL is an independent predictor of hip fracture will not be resolved until a prospective randomized trial is performed by stratifying HAL and matching the groups for BMD, age, height, weight, and menopausal status.
IV. Comparing Men with and without Fractures
Areal BMD is reduced at most sites in men with fractures. Men with spine fractures have reduced areal BMD at the spine and proximal femur, whereas men with hip fractures have reduced areal BMD at the proximal femur with more modest deficits at the spine (in part because arthrosis arti- factually increases BMD at that site). Because the areal BMD measurement does not entirely correct for differences in bone size, the deficits in areal BMD may be, in part, the result of reduced bone size in men with fractures. The remaining deficit between men with fractures and those without fractures (after accounting for differences in bone size) may be caused by reduced accrual, excessive bone loss, or both.
A. Reduced Bone Size
Reduced vertebral body width, but not femoral neck width, can be found in men with spine fractures (Vega etal., 1998), and reduced femoral neck width, but not spine width, may be present in men with hip fractures (Y. Duan and E. Seeman, unpublished data, 1998). Approximately 16-20% of the deficit in areal BMD at the spine in men with spine fractures and at the proximal femur in men with hip fractures is explained by the smaller bone size. Vertebral body and femoral neck width increase as age advances (Y. Duan and E. Seeman, unpublished data, 1998). Thus, the smaller bone size may be the result of the attainment of a reduced peak bone size, failure of periosteal apposition during advancing age, or both.
The site selectivity of the deficit in size may be related to reduced regional growth. Before puberty, growth in leg length is more rapid than growth in spine length. During puberty, growth in spine length increases, while leg epiphyses fuse and growth velocity slows. Illness, deficiency in growth hormone or IGF-1, and sex steroid deficiency may have differing effects depending on the age of exposure to the illness or hormone deficiency state. Regions growing more rapidly may be more adversely affected than those that either have not started their growth spurt or have completed it. Illness before puberty may have a greater effect on the growth in size, mass, and volumetric BMD of the legs than on that of the spine; illness during puberty may have greater effects on growth in size, mass, and volumetric BMD of the axial skeleton (Bass etal., 1998). Growth hormone, IGF-1 deficiency, or testosterone deficiency during early growth may result in reduced femoral neck width. Sex steroid deficiency in puberty may result in reduced expansion of the vertebral width. These hypotheses remain untested.
B. Less Bone in the Bone—Reduced Accrual and Excessive Bone Loss
The deficit in areal BMD remaining after accounting for the contribution of reduced bone size reflects the reduced amount of bone in the bone— reduced volumetric BMD. Reduced volumetric BMD may be the result of reduced accrual, excessive bone loss, or both. A reduction in the peak volumetric BMD (reduced accrual) in long bones may occur if endocortical expansion is excessive relative to periosteal expansion. Alternatively, thinner cortices may result if endocortical contraction during puberty is reduced, perhaps as a result of sex hormone deficiency. A reduced volumetric BMD in trabecular sites may result if growth of primary and secondary spongiosa is disturbed. For instance, testosterone or estrogen deficiency may prevent trabecular thickening during pubertal development and so contribute to reduced trabecular BMD.
Excessive bone loss may also be responsible for the reduced volumetric BMD in patients with fractures. An imbalance between bone resorption and formation at the basic modeling unit (BMU) is the morphological basis for bone loss. Thus, for there to be “excessive bone loss” in patients with fractures, bone balance at the BMU must be more negative than the negative bone balance in controls on one or more of its three endosteal surfaces— endocortical, intracortical, and trabecular. The imbalance will be greater if resorption depth is greater, if formation is lower, or if both are present. Alternatively, if the remodeling rate is higher, bone loss may be greater, even though the negative bone balance is no different. The amount of bone removed from the skeleton as a result of the remodeling imbalance on one or more of these surfaces and the rate of remodeling will be modified according to the extent of subperiosteal bone formation in men with fractures relative to controls.
This potential heterogeneity in the pathogenesis of the deficit in volumetric BMD is poorly characterized. The relative contributions of reduced accrual versus excessive bone loss is unknown and may vary from patient to patient. Whatever the deficit attributable to excessive bone loss, little data are available defining the possible heterogeneous morphological basis of the greater bone loss. Is trabecular bone loss more rapid in men coming to sustain spine fractures? If so, is this caused by an increased depth of resorption within each remodeling site, reduced bone formation within the remodeling site, or increased numbers of remodeling sites?
Do men who sustain hip fractures lose cortical bone more rapidly than the controls? If so, is this the result of increased endocortical resorption, reduced endocortical formation, or both? The greater porosity may be caused by greater numbers and/or larger canals—the former reflecting increased intracortical remodeling and the latter reflecting increased BMU imbalance caused by larger remodeling units (increased resorption or reduced formation).
C. Histomorphometry and Reduced Bone Formation
Histomorphometric studies in men are difficult to interpret because most sample sizes are small, the studies often lack controls, the men with fractures often range in ages by several decades, and men with primary and secondary osteoporosis are often combined. Several studies have been instructive. Mosekilde (1988) reports a reduction in the thickness and loss of horizontal trabeculae in vertebral specimens. There was loss of vertical trabeculae with an increase in intertrabecular space in women, not men. No compensatory thickening of vertical trabeculae was observed. Mellish et al. (1989) report that trabecular thinning occurred with advancing age in 49 men and 47 women. Perforation occurred in both sexes but more so in women. Parfitt etal. (1983) report the age-related decline in iliac crest trabecular bone volume in men and women occurred by a reduction in trabecular density, but not trabecular thinning. In patients with vertebral fractures, trabecular bone volume deficits (of 38% in women and 48% in men relative to age-predicted mean values) occurred by reduction in trabecular density (of about 30% in both sexes). Trabecular thinning contributed to both, although perhaps more so in men (deficits were 18% in men and 7% in women). In patients with hip fractures, the deficits in trabecular bone volume (of 27% in women and 23% in men below age-predicted mean values) occurred by reduction in trabecular thickness (28% in men and 17% in women). The deficit in trabecular density was 11% in women. No deficit in trabecular density occurred in men (relative to age predicted value), but a deficit of 37% relative to the young normals was observed.
Bordier et al. (1973) studied 11 cases of osteoporosis in young adult men. Bone formation was reduced when assessed by quantitative histology, by calcium-45 accretion rate, and by serum alkaline phosphatase. Active bone resorption surfaces (the proportion of surface exhibiting osteoclasts within their lacunae) were normal in 8 of 11 subjects and increased in 3. Zerwekh et al. (1992) showed that, relative to controls, 16 nonalcoholic eugonadal men with osteoporosis had reduced bone volume (11.4 ± 4% versus 23.2 ± 4.4%), reduced osteoid surfaces (5.6 ± 3.9% versus 12.1 ± 4.6%), osteoblastic surfaces (2.0 ± 2.3% versus 3.9 ± 1.9%), and reduced bone formation rate (0.004 ± 0.001 mm3/mm2/yr versus 0.01 ± 0.006 mm3/mm2/yr). Clarke et al. (1996) reported histomorphometric data in 43 healthy men 20 to 80 years old. Cancellous volume of the iliac crest decreased by 40%, as did osteoblastosteoid interface (19.2%), and double- and single-labeled osteoid (18.6% and 18.0%, respectively). On multiple regression analysis, the log-free androgen index and body weight best predicted the age-related decline in cancellous bone volume (r2 = 0.19, P =0. 015). Johansson et al. (1997) reported that 11 men with idiopathic osteo-porosis aged 43 ± 9 years had reduced wall thickness (48.3 ± 7.2 jjim versus 61.7 ± 5.4 (xm, P 0.001), reduced resorption depth (54.4 it 3.8 |xm versus 60.7 ± 5.3 |xm, P < 0.01), and a negative balance ( — 6.04 ± 9.8 p,m versus 0.96 ± 3.2 fim, P < 0.05), relative to 11 controls aged 31 ± 10 years. Preosteoblastic resorption depth correlated positively with wall thickness in controls (r = 0.82, P < 0.01) and negatively with wall thickness in patients (r = —0.56, P = 0.07).
Thus, loss of trabeculae contribute to age-related bone loss and the pathogenesis of fractures in both men and women. Loss of trabeculae is likely to be the main mechanism in vertebral fractures in men and women and in trabecular thinning in hip fractures in men and women. Thus, it is likely that both thinning and loss of trabeculae contribute to bone loss. Bone loss in men is likely to be caused by reduced bone formation rather than increased bone resorption.
D. Cellular Evidence of Reduced Bone Formation
The reduced bone formation may be caused by reduced osteoblastic progenitor cell availability. Bergman etal. (1996) reported that cultured marrow stromal mesenchymal stem cells from male mice aged 24 months yielded 41% fewer osteogenic progenitor cell colonies than cells from 4 month old mice. Cultures from older animals had a threefold higher basal proliferative rate, measured by 3H-thymidine uptake, relative to cultures from young mice, but the increase in proliferation in response to serum stimulation was tenfold in cultures from young animals and nonsignificant in cultures from older mice. The age-related decrease in osteoblast number and function may be caused by a reduction in the number and proliferative potential of stem cells.
Jilka et al. (1996) used the murine model of accelerated senescence and osteopenia, SAMP6, to determine whether the age-related decrease in bone mass is associated with reduced osteoblastogenesis. Osteoblast progenitor numbers were normal in SAMP6 marrow at one month of age but decreased threefold at 3 to 4 months. This reduction was temporally associated with decreased bone formation (determined using histomorphometry) and decreased BMD. In ex vivo bone marrow cultures, osteoclastogenesis was decreased in tissue from SAMP6 mice but was restored by addition of osteoblastic cells from normal mice, suggesting that the osteoclastogenesis defect was secondary to impaired osteoblast formation. Kajkenova et al. (1997) suggest that a change in the differentiation program of multipotential mesenchymal progenitors may explain the reduced osteoblastogenesis in SAMP6. Ex vivo marrow cultures from SAMP6 mice aged 3 to 5 months had increased numbers of colony-forming unit—adipocytes (14.7-fold in unstimulated cultures) and of fully differentiated marrow adipocytes relative to control SAMR1 mice. The number of colony-forming unit—fibroblasts did not differ. In long-term cultures, the adherent stromal layer capable of supporting hematopoiesis was generated more rapidly, more nonadherent myeloid progenitors were generated, and more IL-6 and colony stimulating activity was produced.
Weinstein et al. (1997) report that orchiectomy of SAMR1 mice induced fourfold increases in activation frequency, bone formation rate per bone perimeter and per bone area, increases in osteoclast number, percent osteoclast perimeter, and cancellous osteoclast number, whereas orchiectomized SAMP6 mice showed no such changes. In SAMR1, cancellous bone area decreased by 61 %, and trabecular spacing increased by 160%; these parameters changed similarly in SAMP6 but were blunted in magnitude.
Mineralized perimeter in lumbar vertebrae increased in SAMR1, with augmentation in formation rate and appositional rate. No such changes occurred in SAMP6. Global BMD decreased 6.6% in SAMR1 and was unchanged in SAMP6. Orchiectomy increased formation of colonies of fibro- blastoid cells (CFU-F) and of colonies producing mineralized bone nodules (CFU-OB) in ex vivo bone marrow cultures from SAMR1 but not SAMP6 mice. Thus, cells of osteoblastic lineage are essential mediators of skeletal changes following orchidectomy.
Marie et al. (1991) showed that reduced osteoblastic proliferative activity may be responsible for reduced bone formation in men with osteoporosis. Thymidine incorporation into DNA was normal in cells of normal subjects and patients with normal bone formation but was reduced in cells isolated from patients with osteoporosis with reduced bone formation (doubly labeled surfaces, mean wall thickness, osteoblast surface, and mineral apposition rate). Synthetic activity, assessed by osteocalcin responsiveness to vitamin D, was normal. These studies substantiate the importance of reduced bone formation in the pathogenesis of bone loss in men.
Bone fragility in men in old age may be the result of reduced bone size or architectural changes accompanying bone loss such as cortical thinning, trabecular thinning, and loss of connectivity. Men may have fewer spine fractures than women because their peak bone size is greater; greater vertebral width (not height) confers greater breaking strength. Although men have bigger bones, the amount of bone in the (bigger) bone—the peak vertebral volumetric trabecular BMD (trabecular number and thickness)—is the same in men and women. Vertebral body fragility increases less in men than in women during aging because the loss of trabecular bone proceeds primarily by thinning caused by reduced bone formation, whereas increased remodeling in women caused by menopause contributes to trabecular thinning and loss of connectivity.
Men may have fewer hip fractures than women, in part, because proximal femur bone mass and size is greater in men. Femur length and width is greater in men than women because prepubertal growth is two years longer, and pubertal growth velocity is more rapid and ceases later in males. Although femoral neck and shaft cortical thickness is the same in men and women, the cortex is placed farther from the neutral axis in men conferring greater bending strength. The higher proximal femoral BMC and areal BMD in men is the result of greater bone size; volumetric cortical BMD does not differ by gender. Femoral neck axis length is longer in men than in women in absolute terms, an observation difficult to reconcile with lower hip fracture rates in men. Net cortical bone loss during aging is less in men than in women because endocortical resorption is less, and periosteal apposition may be greater; the latter increases bone size, offsetting the bone fragility conferred by cortical thinning.
Bone remodeling in old age increases in men (and remains elevated in women) perhaps because of secondary hyperparathyroidism, calcium mal-absorption, and vitamin D deficiency, particularly in house-bound subjects. The increasing cortical porosity and endocortical remodeling “trabecularize” cortical bone, increasing the surface available for remodeling. Cortical bone loss accelerates as a result of the increased remodeling activity and negative bone balance in each remodeling unit, predisposing it to hip fractures.
Men with spine fractures have reduced vertebral width (not height) relative to controls; femoral neck width is normal. Men with hip fractures have reduced femoral neck width; vertebral size is normal. Reduced size accounts for —16-20% of the deficit in areal BMD. The site specificity of the deficits in bone size may be the result of a regional deficits in bone growth, failed periosteal apposition during aging, or both. The residual deficit in volumetric BMD—the amount of bone in the (smaller) bone in both types of fracture— may be caused by reduced accrual, excessive bone loss, or both. Reduced accrual may be caused by (i) reduced cortical thickness, itself the net result of excessive expansion of endocortical surface relative to the periosteal expansion before puberty, failed endocortical contraction during puberty, (ii) reduced development of trabecular numbers in early growth, or (iii) reduced trabecular thickening in puberty.
Studies of racial/ethnic patterns of skeletal growth, aging, and fracture rates in males are scarce. Blacks have fewer fractures than whites. Regional differences in bone mass and HAL are in part, caused by differences in bone size. Blacks have shorter trunks and shorter (not wider) vertebral bodies than whites. (Men have wider, not shorter, vertebrae than women.) Blacks have wider and longer femurs with the same cortical width. (Men also have wider and longer femurs with similar cortical width compared to women.) Because the femur is wider, the greater cortical mass is farther from the neutral axis of the long bone in blacks. Asians have similar trunk length as whites but shorter legs. Vertebral trabecular BMD is higher in blacks than whites because of greater trabecular thickness (not numbers) so that blacks have less surface and a lower bone turnover than whites. Because blacks do not appear to lose less bone than whites, bone balance at the BMU may be more negative in blacks. The thicker trabeculae may preserve connectivity during aging.
Periosteal expansion during growth and aging determine external bone size, an independent determinant of bone strength; endosteal (intracortical, endocortical, trabecular) modeling and remodeling establish cortical thickness and porosity, trabecular number, thickness, and connectivity. Understanding the structural basis of bone fragility in men requires the study of the modeling and remodeling of the surfaces of the axial and appendicular skeleton in men and women of different races.
There are many unresolved questions. One of the most fundamental questions is whether the differences in fracture rates between men and women, between men with fractures and men without fractures, and between races can be explained by the structural differences observed between these groups. If bone size is a critical determinant of its breaking strength, then what genetic and environmental factors contribute to the variance in bone size between men and women of the same race, and between individuals of the same gender but different race? What factors account for the variance in trabecular numbers and their thicknesses? Why do blacks have a greater increase in trabecular thickness at puberty than whites of the corresponding gender? What is the role of estrogens in trabecular and endocortical bone remodeling? What factors regulate periosteal expansion during aging? Do black men lose less bone than white men in absolute terms, or as a percentage of their higher bone mass? Of this putative lesser bone loss, what proportion is caused by less resorption, what proportion is caused by greater bone formation?