To study the structural basis of bone fragility in men, we compared bone size and volumetric bone mineral density (vBMD) of the third lumbar vertebra and femoral neck in 95 men with spine fractures, 127 men with hip fractures, and 395 healthy controls using dual-energy X-ray absorptiometry (DXA). The results were expressed in absolute terms and age-specific SD scores (mean ± SEM). In controls, vertebral body and femoral neck width increased across age, being 0.46 ± 0.11 SD and 0.91 ± 0.08 SD higher in elderly men than in young men, respectively (both, p < 0.001). Men with spine fractures had reduced vertebral body width (−0.45 ± 0.10 SD; p < 0.01) but not femoral neck width (−0.15 ± 0.10 SD, NS). Men with hip fractures had reduced femoral neck width (−0.45 ± 0.11 SD; p < 0.01) and vertebral body width (−0.25 ± 0.10 SD; p < 0.05). The deficits in bone volume (BV) exaggerated the deficits in bone mineral content (BMC) by 40% at the vertebrae in men with spine fractures and by 9% at the femoral neck in men with hip fractures. vBMD deficits were greater at the vertebrae in men with spine fractures than in men with hip fractures (−1.37 ± 0.08 SD vs.−0.70 ± 0.10 SD, respectively; p < 0.01) but were similar at the femoral neck (−0.93 ± 0.10 SD and −0.76 ± 0.11 SD, respectively, NS), despite the men with spine fracture being 10 years younger. Bone fragility leading to spine or hip fractures in men may be the result of fracture site-specific deficits in bone size and vBMD that have their origins in growth, aging, or both.
IN THE study of the pathogenesis of bone fragility of women or men, patients are grouped together because they have “one or more nontraumatic” spine fractures or a hip fracture “sustained in a fall from no greater than the standing position.” The patients are consistently reported to have deficits in bone mineral content (BMC) or areal bone mineral density (aBMD) usually attributed to excessive bone loss.(1)
Because neither BMC nor aBMD fully account for bone size,(2,3) any deficit in BMC or aBMD in patients with reduced bone size and fractures will be exaggerated.(4–8) In patients with larger bones, any deficit in BMC or aBMD will be either obscured or underestimated leading to the erroneous conclusion that neither reduced accrual nor excessive bone loss occurred.
Thus, the failure to recognize influence of bone size on the BMC and aBMD measurement leads to (1) failure to consider the independent contribution of bone size to bone fragility; (2) failure to consider the potential contribution of reduced periosteal expansion during growth, aging, or both to any deficit in bone size in old age; and (3) flawed conclusions regarding the existence or magnitude of reduced accrual and bone loss to the deficit in volumetric BMD (vBMD).(3,4,9)
Women with spine fractures may have reduced vertebral body size.(4–7) This deficit in bone size overestimates the deficit in BMC by 48% and aBMD by 16%.(4) Men with spine fractures also may have reduced vertebral size, but the contribution of the reduced bone size to the deficit in BMC or aBMD has not been quantified.(8) There have been no studies examining the deficit in vBMD in men with spine fractures and none examining bone size and vBMD in men with hip fractures.
To address these issues and gain insight into the pathogenesis of bone fragility in men with fractures, we compared bone size, BMC, aBMD, and vBMD in men with spine or hip fractures with age-matched controls. We asked the following questions. (1) Does bone size increase with advancing age in healthy men? (2) Do men with spine or hip fractures have reduced bone size at the fracture site? (3) Are there differences in bone size, BMC, aBMD, and vBMD of the vertebrae and the femoral neck that distinguish men with spine or hip fractures?
MATERIALS AND METHODS
We studied 95 men aged 45–90 years with one or more nontraumatic spine fractures. The patients were recruited from the Metabolic Bone Clinic of Austin & Repatriation Medical Centre. All the patients had symptomatic spine fractures confirmed radiologically by a reduction in vertebral body height by greater than 20%. We excluded patients with secondary osteoporosis because of thyroid diseases, primary hyperparathyroidism, multiple myeloma, corticosteroid therapy, and chronic liver or renal disease. We excluded patients taking medication known to affect bone and patients with fractures of the third lumbar vertebra (the site of measurement of bone size and vBMD).
We also studied 127 men aged 45–95 years with hip fractures (58 cervical, 59 trochanteric, and 10 unspecified fractures) sustained after a fall from no greater than the standing position. Hip fractures caused by severe trauma (motor vehicle accident) or local pathology were excluded. About 80% of patients were recruited within 12 months of hip fracture; 20% were recruited within 1-3 years of fracture. We excluded patients taking medication or having illness known to affect bone.
Three hundred and ninety-five healthy male controls aged 17–91 years with no history of spine and hip fractures were recruited from the local community by advertisement as part of the ongoing research in this department. Subjects taking medication or having illness known to affect bone were excluded. All cases and controls were whites. Informed consent was obtained from all the participants. The study was approved by the Human Research Ethics Committee of the Austin and Repatriation Medical Center.
Measurement of BMC, size, aBMD, and vBMD
BMC, aBMD, and bone size of the third lumbar vertebra and femoral neck were measured by dual-energy X-ray absorptiometry (DXA; DPX-L, Version 1.3z; Lunar Corp., Madison, WI, USA). Vertebral body volume (V) was estimated as V = A3/2, where A is the projected area of the third lumbar vertebra obtained by posteroanterior (PA) scanning.(2) This measurement overestimates(10) but correlates with volume determined in vitro by submersion (r = 0.80; p < 0.001; standard error of estimate (SEE) = 6.2 cm3; Tabensky and Seeman, unpublished data, 1996). Vertebral vBMD = total BMC (of the vertebral body plus posterior processes)/V. Vertebral body width and height were obtained from the PA scan. The CV ranged from 1.1% to 1.7% for these measurements.
Femoral neck volume was determined assuming the region was cylindrical; V = π × (W/2) × h, where W is femoral neck width at the middle point of the femoral neck axis length (FNAL) using the DXA ruler option (Fig. 1), and h is the height of the scanned region of the femoral neck. As h is held constant (1.5 cm), femoral neck volume is determined by width only. FNAL was the distance between the cortical rim of the femoral head to the periosteal surface of the greater trochanter. All manual analyses of FNAL and width were made by one investigator (Y.D.). Measurements were performed on the unfractured femoral neck. The CV was 1.0% for FNAL and 1.6% for femoral neck width based on scanning 10 subjects twice within 3 months. The CV ranged from 2.6% to 3.1% for BMC, volume, aBMD, and vBMD at the femoral neck.
The data were expressed in absolute terms and as the number of SD scores above or below the young normal or age-matched normal mean (T or Z score, respectively). Multiple linear regression analysis was used to adjust for the confounding factors of body height and weight.(11) One-sample t-tests were used to determine whether the T or Z scores differed from zero. Two-sample t-tests were used to determine the significance of any difference comparing fracture cases with controls or spine with hip fracture cases. Bonferroni correction of multiple comparisons was carried out.(12) Results were regarded as statistically significant at the 5% level (two-tailed).
BMC depends on bone size. To derive the volume-adjusted BMC or aBMD in patients and age-matched controls, we regressed BMC or aBMD on volume in men aged 18–43 years assuming that BMC and bone size are stable during these years.(4) Volume-adjusted BMC = observed BMC + (mean volume − observed volume) × slope. Mean vertebral body volume (cm3) and slope (g/cm3) used in the formula to derive the volume-adjusted BMC were 69 cm3 and 0.34 g/cm3, respectively. The corresponding values for mean femoral neck volume and slope were 14 cm3 and 0.15 g/cm3, respectively. Similarly, volume-adjusted aBMD = observed aBMD + (mean volume − observed volume) × slope. Mean vertebral body volume (cm3) and slope (g/cm2 per cm3) used in the formula to derive the volume-adjusted aBMD were 69 cm3 and 0.0077 g/cm2 per cm3 and mean femoral neck volume and slope were 14 cm3 and 0.00015 g/cm2 per cm3, respectively. The deficit in BMC or aBMD in patients with fractures before size adjustment is caused by smaller bone size and/or less bone in the bone; the deficit after size adjustment reflects reduced accrual, excessive bone loss, or both (ignoring bone loss after the fracture); the difference between the two is the BMC or aBMD difference attributable to the difference in bone size.
Results (mean ± SEM; p values) are presented in absolute terms, T scores, and Z scores in the tables. For brevity and clarity, the following results are presented as group means only; SEMs and p values are provided for data not shown in the tables.
Vertebral and femoral neck size in healthy controls and patients with fractures
As shown in Fig. 2 and Table 1, vertebral body and femoral neck width increased across age in healthy men (r = 0.17 and r = 0.40, respectively; both, p < 0.001), being 0.46 SD and 0.91 SD higher in the elderly men than in the young men, respectively. Vertebral body volume was independent of age (r = −0.01, NS) because vertebral body width increased (r = 0.17) while vertebral body height decreased across age (r = −0.17; p < 0.001). Men with spine fractures had reduced vertebral body volume (−0.73 SD) because of reduced vertebral body width (−0.45 SD) and height (−0.35 SD) but no deficit in femoral neck width (−0.15 SD). The deficit in vertebral body width was greater than the deficit in femoral neck width in men with spine fractures (p < 0.02). Vertebral body volume remained lower after adjusting for body height and weight (−0.21 ± 0.13 SD; p = 0.10), while femoral neck volume remained no different from controls.
Table Table 1.. Age, Height, Weight, BMI, L3 Vertebral Body Width, Height, Volume, BMC, aBMD, and vBMD; FNAL, Width, Volume, BMC, aBMD, and vBMD in Young Men, Men with Spine or Hip Fractures, and Their Respective Age-Matched Controls
Men with hip fractures had reduced femoral neck width (−0.45 SD) and vertebral body width (−0.25 SD) relative to age-matched controls. Femoral neck and vertebral body volume also were reduced (−0.42 SD and −0.37 SD, respectively). After adjusting for body height and weight, femoral neck volume remained significantly reduced (−0.37 ± 0.12 SD; p < 0.004) while vertebral body volume was no different from age-matched controls (−0.05 ± 0.12 SD, NS). There was no significant difference of age, height, and weight or in the size of the deficits in femoral neck and vertebral size in patients with cervical and trochanteric fractures.
Comparing the deficits in men with spine and hip fractures, the deficit in vertebral body volume in men with spine fractures was greater than the deficit in vertebral body volume in men with hip fractures (p < 0.01); there was no difference in femoral neck volume in men with spine or hip fractures. However, the deficit in vertebral body volume in men with spine fractures was greater than the deficit in femoral neck volume in men with hip fractures (−0.73 SD vs. −0.42 SD; p < 0.01).
As shown in Table 2, reduced vertebral size was responsible for about 40% of the deficit in vertebral BMC in men with spine fractures but only 9% of the deficit in femoral neck BMC in men with hip fractures. Similarly, about 19% of the deficit in spinal aBMD in men with spine fractures was accounted for by the smaller vertebra. However, there was no contribution of the reduced femoral neck size to the deficit in femoral neck aBMD in patients with hip fractures.
Table Table 2.. The Observed and Volume-Adjusted Values for BMC (g) and aBMD (g/cm2) at the L3 Vertebra in Men with Spine Fractures and at the FN in Men with Hip Fractures
FNAL was unchanged with age in healthy men (r = 0.05, NS), and was not reduced in men with spine (−0.04 SD) or hip fractures (0.18 SD; Table 1).
vBMD in patients with spine fractures versus hip fractures
As shown in Table 1 and Fig. 3, vertebral and femoral neck vBMD was reduced in men with either fracture. However, the deficit in vertebral vBMD in men with spine fractures was greater than in men with hip fractures 10 years more advanced in age. Femoral neck vBMD was similar in men with spine and hip fractures, despite this age difference. Matching men with spine and hip fractures by age (between 65 to 75 years) did not affect the results (vertebral vBMD, 0.24 ± 0.01 g/cm3 vs. 0.29 ± 0.01 g/cm3 and p < 0.01; femoral neck vBMD, 0.28 ± 0.01 g/cm3 vs. 0.28 ± 0.01 g/cm3, NS). There were no significant differences in vBMD or bone size in younger men versus older men with spine fractures, or younger men versus older men with hip fractures (data not shown). There were no differences in vertebral or femoral neck vBMD in men with cervical and trochanteric fractures.
We report that (1) advancing age was associated with increased vertebral body and femoral neck width in healthy men. Relative to age-matched controls, men with spine fractures had reduced vertebral body size and men with hip fractures had reduced femoral neck size. Deficits in vertebral body size were greater in men with spine fractures than in men with hip fractures while deficits in femoral neck size occurred in men with hip fractures only. (2) The smaller bone size overestimated the deficit in BMC. There was more contribution of bone size to the deficit in BMC in men with spine fractures than in men with hip fractures. (3) The smaller bones also had reduced vBMD. (4) Despite a 10-year difference in age, men with spine fractures had greater vertebral vBMD deficits and similar femoral neck vBMD deficits compared with men with hip fractures. This site specificity of the deficits in vBMD persisted after matching spine and hip fracture cases by age.
Bone size is an independent determinant of bone strength.(5,10,13–16) We confirmed several cross-sectional and longitudinal studies of increasing bone size in vertebral body and femur in men with advancing age.(17–23) This increase in bone size offsets the fall of bone strength conferred by bone loss.(20,23) Vertebral body width increased by 0.5 SD while femoral neck width increased by 0.9 SD, increments that are likely to be conservative as secular changes in bone size underestimate the age-related increase.
The importance of bone size has been underestimated. For example, gender differences in vertebral body compressive strength disappear after accounting for gender differences in bone size.(24) BMD plus the cross-sectional area of bone is a better predictor of bone strength than BMD alone at the spine and hip.(13,14,16) Reduced vertebral size is found in men and women with spine fractures(5,8); about half of the deficit in vertebral BMC in women with spine fractures was explained by the difference in bone size.(4) We found that vertebral size was reduced in men with spine fractures, and femoral neck size was reduced in men with hip fractures relative to age-matched controls, but the deficit in femoral neck size in men with hip fractures was less than the deficit in vertebral size in men with spine fractures. Moreover, femoral neck size did not differ in men with spine or hip fractures.
About 40% of the deficit in vertebral BMC in spine fractures was because of the smaller vertebra, while the contribution of the smaller femoral neck size to the deficit in femoral neck BMC is less (9%). Increased femoral neck width has been reported in one study of men with hip fractures.(25) We suggest that the greater deficits in vertebral size may be more important in the pathogenesis of spine fractures than femoral neck size in hip fractures because vertebral size disperses loads so stress (load per unit area) diminishes with age in men. Bone size still may be important in hip fracture pathogenesis, but relatively less so than bone size in spine fracture pathogenesis because the trauma is critical in hip fractures.
Whether the reduced bone size in men with spine or hip fractures is caused by reduced attainment of peak bone size or by failed periosteal expansion during advancing age is unclear. We suggest that these size deficits are likely to be growth-related rather than age-related because (1) failed periosteal expansion should produce greater deficits in older fracture cases than younger fracture cases but this was not observed, and (2) the offspring and relatives of patients with spine fractures have reduced aBMD, which may be, in part, caused by reduced bone size and reduced mineral accrual.(9,26,27)
The site specificity of the deficits in bone size in patients with fracture may be related to the normally disparate patterns of growth of the axial and appendicular skeleton. Leg growth proceeds more rapidly than trunk growth before puberty and then decelerates during puberty as trunk growth accelerates.(28) Thus, exposure to disease or risk factors before puberty may affect the legs; exposure to illness or risk factors delaying pubertal growth may affect the spine.(28,29)
vBMD was reduced relative to the age-predicted mean at all sites in both types of fracture. However, the deficits in vBMD also were site-specific; men with spine fracture had greater deficits at the spine than men with hip fractures while deficits at the femoral neck were similar (despite the age difference). Matching cases with spine and hip fractures by age did not affect the site specificity of the deficits in vertebral vBMD. A greater deficit in aBMD of the femoral neck than of the spine in women with hip fractures has been reported by several investigators.(30–33) Whether these deficits are growth or age related is uncertain but, again, finding reduced aBMD in offspring and relatives of men and women with spine fractures is also consistent with the possibility that the deficits are, at least in part, growth related.
Although BMC, aBMD, and vBMD are equally predictive of fracture risk and bone strength,(10,34) the problems produced by the expressions of bone mass such as BMC and aBMD should not be ignored. The failure to account for bone size obscures the structural diversity in patients with fractures, overestimates the deficit attributable to reduced accrual or excess bone loss in patients with smaller bones, and obscures any deficit in patients with larger bones. The failure to account for bone size also obstructs progress in the study of the growth-related or age-related mechanisms responsible for producing reduced bone size, an independent determinant of bone strength.
Longer hip axis length has been suggested to be an independent determinant of risk of hip fracture in women in some,(35–37) but not all,(25,38) studies. There also is evidence to suggest that women with hip fractures have greater hip axis length than women with spine fractures.(37) We found no difference in FNAL in men with spine and hip fractures, suggesting that this trait may not be a risk factor for hip fracture in men.
Patients with cervical or trochanteric fractures may differ in several ways.(39,40) Women with trochanteric fractures may be older and may have lower body weight and lower BMD than women with cervical fractures(40) and have a greater deficit in trabecular bone.(32) Duboeuf et al. reported that hip axis length was longer in cervical fractures than in trochanteric fractures,(39) while femoral neck width was lower in women with trochanteric fractures relative to controls. We found no significant differences in age, height, weight, FNAL, bone size, and vBMD in men with cervical fractures compared with trochanteric fractures.
As stated in Materials and Methods and as discussed recently,(4) the Carter et al.(2) method overestimates vertebral body volume and therefore underestimates vBMD, but this is proportional in patients with fractures and controls. In a subset of 26 men with spine fractures and 157 controls, measured by lateral scanning, which estimates vertebral body volume more accurately relative to submersion volume gold standard in vivo,(10) the deficit in vertebral body volume in men with spine fractures relative to age-matched controls was about −0.8 SD and the deficit in vBMD was −1.5 SD (Y. Duan and E. Seeman, unpublished data, 2000), almost identical to the results reported in this study (−0.7 SD and −1.4 SD, respectively). Whether the estimate of femoral neck volume is accurate is uncertain. Assumption of circularity is crude and further studies are needed to examine femoral neck size in patients with hip fractures.
Whether one etiologic factor is responsible for site- and gender-specific abnormalities is unclear. Riggs et al. proposed that estrogen deficiency is likely to play a central role in age-related bone loss.(41) Testosterone deficiency results in reduced peak bone size and, perhaps with estrogen deficiency, may contribute to reduced peak vBMD and bone loss in men. In the setting of reduced peak bone size and reduced peak vBMD, bone loss is likely to be tolerated poorly, resulting in thinning and complete loss of trabeculae, loss of connectivity, vertebral fragility, and spine fractures after minimal trauma. In the setting of reduced peak vBMD, normal or larger bone size may protect against the compressive loads at the spine but not the later occurrence of hip fractures as cortical bone loss continues producing bone fragility at the hip during aging when the incidence of falls increases. A better understanding of the genetic and environmental factors contributing to the diverse pathogenesis of primary and secondary osteoporosis may be obtained attention to growth-related and age-related factors that regulate bone size and volumetric density and recognition of the limitations of bone densitometry.(3)
We thank senior technologists Ms. Vanessa De Luca and Patricia D'Sousa for technical assistance during this study.