Lateral spine dual-energy x-ray absorptiometry (DXA) selectively measures the trabecular-rich vertebral bodies without the contributions of the cortical-rich posterior elements of the spine and is less affected by spinal degenerative disease than posterior-anterior DXA. We tested whether lateral DXA detects vertebral osteoporosis more often and is more sensitive to age-related bone loss than posterior-anterior DXA in 193 healthy, community-dwelling men aged 51–81 years (mean ± SD; 67 ± 8 years). All men had supine lateral, posterior-anterior, and proximal femur DXA scans on a Hologic QDR 2000 densitometer. A subset (n = 102) had repeat scans after 4 years to determine annualized rates of change in bone mineral density (BMD). Age was inversely and significantly associated with BMD in the midlateral (r = −0.27) and lateral (r = −0.24) but not posterior-anterior (r = 0.04) projections. Midlateral (−1.43 ± 3.48% per year; p = 0.0001), lateral (−0.27 ± 1.68% per year; p = 0.12), and hip (−0.19 ± 1.02% per year; p = 0.06) BMD decreased, whereas posterior-anterior BMD increased (0.73 ± 1.11% per year; p = 0.0001) during follow-up. When compared with normal values in 43 men aged 21–42 years, mean T scores were significantly lower with lateral (−1.47 ± 1.32) and midlateral (−1.57 ± 1.36) than posterior-anterior (−0.12 ± 1.30; p < 0.0001) DXA. Only 2.6% of the older men were considered osteoporotic (T score ≤ −2.5) at the posterior-anterior spine, whereas 11.0% were osteoporotic at the femoral neck, 22.5% at the lateral spine, and 24.6% were osteoporotic at the midlateral spine. We conclude that supine lateral DXA identifies considerably more men as osteoporotic and is more sensitive to age-related bone loss than posterior-anterior DXA. Spinal osteoporosis may represent a substantially greater health problem among older men than previously recognized.
OSTEOPOROSIS IS an increasingly important clinical and public health problem among older men. Recent data from the Third National Health and Nutrition Examination Survey indicate that 3–6% (1–2 million) of U.S. men aged 50 years and older have osteoporosis at the hip and are at increased risk of fracture.(1) Other recent longitudinal studies have shown that the rate of bone loss from the hip accelerates with advancing age in men,(2, 3) paralleling the exponential rise in hip fracture rates with age among elderly men.(4) Considerably less is known about the magnitude of osteoporosis and the pattern of age-related bone loss at other clinically relevant skeletal sites in older men such as the lumbar spine. Moreover, most estimates of vertebral bone loss in aging men are derived from cross-sectional observations using posterior-anterior dual-energy X-ray absorptiometry (DXA) measurements,(5–11) which are erroneously increased by spinal degenerative disease.(12–20) Such measurements may be especially problematic among elderly men, because severe spinal degenerative disease is more common among older men than women.(13, 16, 21)
Lateral DXA scanning measures bone density of the vertebral bodies without the contributions of the cortical-rich posterior elements of the spine and thus is less affected by spinal degenerative disease than posterior-anterior spine measurements.(12) We(22) and others(23–27) have shown that lateral DXA scanning detects spinal osteoporosis more often and is more sensitive to age-related bone loss than posterior-anterior spine DXA in women. Comparable studies have not been reported in men. Thus, we compared the ability of supine lateral and posterior-anterior spine DXA to detect osteoporosis and age-related bone loss in healthy, community-dwelling older men.
MATERIALS AND METHODS
All of the older men in this study were participants in a study of the determinants of bone mineral density (BMD) in 523 white men aged 50 years and older.(28) Details of the study population and recruitment procedures are published elsewhere.(28) Briefly, participants were recruited primarily from population-based lists of age-eligible voters in the Monongahela Valley, 30 miles southeast of Pittsburgh, PA. Men who were unable to walk without assistance or who had a history of bilateral hip replacement were ineligible to participate. All 523 participants completed a baseline clinic examination including a measurement of spine and hip BMD on either a QDR 1000 or QDR 2000 densitometer (Hologic, Inc., Waltham, MA, U.S.A.) between 1991 and 1992. Of the 523 men who participated in the baseline clinic examination, 209 were scanned on the QDR 2000 and had both posterior-anterior and lateral spine BMD measurements. Of these 209 men, 16 (8%) were eliminated from the present analysis because of poor positioning (6%), inability to delineate intervertebral disc spaces (<1%), severe deformities (<1%), or severe sclerosis (<1%). Thus, 193 men form the basis of the cross-sectional analyses in this report.
From July 1995 to September 1995, a subset of participants (n = 104) attended a second clinic examination after an average of 3.6 years (range, 3.5–3.8 years). These men were selected randomly from quartiles of baseline body weight (≤74.8 kg, >74.8 to < 83.0 kg, ≥83.0 to <91.2 kg, and ≥91.2 kg) to prevent oversampling lean or obese participants. Technically adequate BMD measurements were obtained in 102 participants. The 102 men who participated in the follow-up study were representative of the original cohort and form the basis of the longitudinal analyses in this report.
Forty-three healthy men aged 21–42 years were recruited through hospital advertisements to serve as a young normal control sample. All had histories of puberty before the age of 14 years, normal serum testosterone levels, and none were taking medications known to affect bone and mineral metabolism. Measurements of femoral neck (n = 32) and lumbar spine (n = 43) BMD in the posterior-anterior and supine lateral projections were made without repositioning on a Hologic QDR 2000 densitometer.
DXA of the lumbar spine (posterior-anterior and lateral) and proximal femur (total hip, femoral neck, and trochanter) was performed with a Hologic QDR 2000 densitometer. Participants remained in the supine position during posterior-anterior and lateral scans. Lateral BMD was determined for the entire vertebral body and in a slice approximately 1 cm thick within the midportion of the vertebral body to eliminate contributions from the superior, inferior, anterior, and posterior vertebral end plates, which contain predominately cortical bone. Baseline scans were reviewed by a physician (J.S.F.) to eliminate vertebra with clear deformities or areas of focal sclerosis from the analyses. Vertebrae with visible overlap from ribs or pelvis were eliminated from analysis of the lateral scans.
Follow-up scans were analyzed using the manufacturer's scan comparison software. Longitudinal machine stability was assessed during the study from plots of daily spine phantom scans. A weekly printout of quality control plots was used to detect both short-term inconsistencies and long-term drift. Quality control scans were evaluated to ensure that the BMD and bone mineral content of the scanner were within normal limits. The densitometer was stable over the course of the study.
The short-term in vivo reproducibility of posterior-anterior, lateral, and midlateral DXA measurements (i.e., the 95% confidence limits for the changes attributable to measurement error) is denoted 2σ, where σ is the SD of the differences between paired BMD measurements for 50 subjects who had repeat scans. The short-term in vivo reproducibilities of paired bone density measurements (i.e., 2σ) for the posterior-anterior, lateral, and midlateral spine measurements were 0.027, 0.038, and 0.057 g/cm2, respectively, corresponding to estimated SDS for individual measurements (i.e., σ/√2) of 0.010, 0.013, and 0.020 g/cm2 using all bones. Expressed as percentages, these values were 1.1, 2.0, and 3.4%, respectively. The in vivo coefficient of variation for femoral neck BMD is 1.2%.(29)
To compare the frequency with which lumbar spine and proximal femur DXA scans detect osteoporosis in older men, BMD values of the 193 men were expressed as T scores using the formula (X1 − Xnl)/SDnl and as a percent difference from young normal values using the formula [(X1 − Xnl)/Xnl] × 100, where X1 is the patient's value, Xnl is the mean value for our normal young men, and SDnl is the SD for normal young men. Because BMD increases progressively across L1-L4, T scores and the percent difference in BMD were calculated using normal values matched to the specific vertebra in each patient. Men were classified as osteoporotic (T score ≤ −2.5), osteopenic (T score, −1.0 to −2.5), and normal (T score > −1.0).(30) It should be noted that these criteria were developed specifically for women(30) and it is uncertain if they apply to men. The rate of change in spine and hip BMD was calculated from baseline and follow-up DXA measurements and expressed as an annualized percentage (% per year) rate of change from initial BMD values. The T scores and percent difference in BMD between the spine DXA measurements were compared using paired T tests. Differences in BMD across age groups (50–59 years,6 0–69 years, and ≥70 years) were tested for statistical significance using analysis of variance. Associations among BMD sites and between age and BMD were assessed with Spearman correlation coefficients. Values are presented as mean ± SD.
The mean (±SD) age, height, weight, and body mass index (BMI) of the older men were 67 ± 8 years, 174.3 ± 6.0 cm, 83.3 ± 12.1 kg, and 27.3 ± 3.3 kg/m2, respectively. The mean age, height, weight, and BMI of the young normal men were 28 ± 5 years, 178.7 ± 7.4 cm, 78.1 ± 10.4 kg, and 24.4 ± 2.5 kg/m2, respectively. Posterior-anterior, lateral, and midlateral spine BMD in the older men were 1.037 ± 0.149 g/cm2, 0.687 ± 0.105 g/cm2, and 0.592 ± 0.126 g/cm2, respectively. Posterior-anterior, lateral, and midlateral spine BMD in the young normal men were 1.052 ± 0.115 g/cm2, 0.814 ± 0.074 g/cm2, and 0.740 ± 0.083 g/cm2, respectively. Femoral neck BMD was 0.790 ± 0.127 g/cm2 in the older men and 0.972 ± 0.129 g/cm2 in the young normal controls. Mean lateral spine, midlateral spine, and femoral neck BMD were 15, 20, and 19% lower, respectively, in the older men compared with the young normal values (p < 0.001 for all comparisons), whereas mean posterior-anterior spine BMD was similar (Table 1). Mean lateral spine, midlateral spine, and femoral neck T scores were significantly different from zero whereas mean posterior-anterior spine T scores were not (Table 1).
Table Table 1. Percentage Difference in BMD and T Score in Men Aged 50 Years and Older Compared with Young Normal Men8
Correlations among skeletal sites ranged from 0.34 to 0.93 (Table 2). Correlations between the hip and spine generally were modest and weaker than among measurements within these skeletal sites. Correlations between hip and posterior-anterior spine BMD (r = 0.64–0.69) were stronger than between the hip and lateral (r = 0.40–0.48) or midlateral (r = 0.33–0.38) spine measurements. Correlations between the femoral neck and trochanter measurements were higher (r = 0.76) than spine measurements (r = 0.46–0.58), although the strongest association was between lateral and midlateral spine BMD (r = 0.93).
Table Table 2. Spearman Correlation Coefficients Among Bone Density Measurements at Different Skeletal Sites and with Age in Community-Dwelling Older Men (n = 193)8
The only skeletal sites not associated with age were the posterior-anterior spine and trochanter (Table 2). All other skeletal measurements tended to decline with age (Table 2). Mean lateral and midlateral spine BMD were 8% and 13% lower in the oldest (>70 years) compared with youngest (50–59 years) age groups, respectively (p < 0.005 for both) whereas posterior-anterior spine BMD did not differ between age groups in cross-sectional analyses (p = 0.51; Fig. 1). Annualized rates of bone loss were greatest when BMD was measured in the spine in the midlateral projection (−1.43 ± 3.48% per year; p = 0.0001). Annualized rates of loss were smaller, and of borderline significance when BMD was measured in the lateral spine projection (−0.27 ± 1.68% per year; p = 0.12) or the proximal femur (total hip= −0.19 ± 1.02% per year and p = 0.067; femoral neck=−0.15 ± 1.28% per year and p = 0.25; trochanter=−0.16 ± 1.06% per year and p = 0.13). In contrast, annualized changes in BMD increased significantly when spine BMD was measured in the posterior-anterior projection (0.73 ± 1.11% per year; p = 0.0001). The rate of decline in trochanteric BMD increased with advancing age (r = −0.21; p = 0.03). The rate of loss in lateral BMD was not associated with age (r = 0.01; p = 0.90), whereas the rate of decline in midlateral BMD decreased with advancing age (r = 0.24; p = 0.02).
Only 2.6% of the older men were classified as osteoporotic (T score ≤ −2.5) when BMD was measured in the lumbar spine in the posterior-anterior projection. In contrast, 22.5% of the men were classified as osteoporotic when BMD was measured in the lateral spine projection, and 11.0% were osteoporotic when BMD was measured at the femoral neck. Osteoporosis was found at either the femoral neck or posterior-anterior spine in 12.0% and femoral neck or lateral spine in 28.8% of the older men.
The diagnosis of osteoporosis was considered to be misclassified if men were osteoporotic (T score ≤ −2.5) at one skeletal site and normal (T > −1.0) at another site (Table 3). For example, of the 21 men considered osteoporotic at the femoral neck, 7 (33.3%) were considered normal in the posterior-anterior view and only 3 of these 21 men (14.3%) were considered osteoporotic based on their posterior-anterior spine measurement (Fig. 2). In contrast, only 2 of these 21 men (9.5%) were misclassified as normal with lateral spine DXA. Of the 108 men (56.5%) classified as osteopenic at the femoral neck, 74 (68.5%) were considered normal based on the posterior-anterior spine measurement and only 34 of these 108 men were considered normal based on their lateral spine measurements. Of the 43 men considered osteoporotic based on their lateral spine measurement, 19 (44.2%) were considered normal and only 4 (9.3%) were considered osteoporotic in the posterior-anterior spine projection (Fig. 3). Only five men were considered osteoporotic based on their posterior-anterior spine measurements, three of whom had femoral neck T scores < −2.5 and four of whom had lateral spine T scores < −2.5. Of the 76 men (39.8%) who were osteopenic on their lateral spine measurement, 21 (27.6%) were considered normal based on their the femoral neck measurement.
Table Table 3. Percentage of Community-Dwelling Older Men Misclassified as Osteoporotic According to World Health Organization Criteria and Skeletal Sitea8
Posterior-anterior spine DXA is used widely for the assessment of male osteoporosis in both clinical practice and research, but such measurements may underestimate the prevalence of vertebral osteoporosis and the magnitude of bone loss. Only 2.6% of our community-dwelling older men were classified as osteoporotic by posterior-anterior spine DXA, whereas 11% were considered osteoporotic when BMD was measured at the femoral neck and 23% were osteoporotic when BMD was measured in the lateral projection. Moreover, men diagnosed as osteoporotic at one skeletal site often were misclassified as normal at another site. This finding is not surprising given the modest correlation among skeletal measurements in the present study and previous reports of older women(31–34) and suggests that assessing BMD at only the posterior-anterior spine may underestimate the number of affected men. For example, prevalence estimates from the Third National Health and Nutrition Survey indicate that 3–6% of U.S. men aged 50 years and older have osteoporosis at the hip. However, in the present study, we found that 12% of the men over age 50 had osteoporosis at either the femoral neck or the posterior-anterior spine, and 29% had osteoporosis at either the femoral neck or the lateral spine. These data suggest that the overall number of men affected with osteoporosis in general and spinal osteoporosis in particular is substantial and that the diagnosis of osteoporosis may be missed in a significant number of older men if only posterior-anterior spine or femoral neck BMD is assessed.
National data for lumbar spine BMD or osteoporosis are not available for comparison, but our estimates of the prevalence of osteoporosis based on posterior-anterior spine measurements (2.6%) are only slightly greater than the prevalence (1.4%) among approximately 200 men aged 50 years and older in Rochester, MN.(35) The prevalence of osteoporosis based on posterior-anterior spine measurements was somewhat higher (4.2–8.5%) among men in southern California(36) and Indiana,(13) possibly because these men were older than those in our cohort. The prevalence of posterior-anterior spine osteoporosis in the present and these previous studies is much lower than the prevalence of vertebral deformities in recent radiographic surveys of older men.(37–39) For example, the prevalence of morphometrically defined vertebral deformity was 20% among European men aged 50 years and older(37) and 25% among Australian men aged 60 years and older.(38) The extent to which these deformities are related to osteoporosis or to trauma in early adult life is uncertain. Nevertheless, the prevalence of spinal osteoporosis detected by lateral DXA in the present study (23%) is similar to the prevalence of vertebral deformities in these radiographic surveys. Moreover, the prevalence of osteoporosis at either the lateral spine or the femoral neck in the present report (28.8%) is remarkably similar to estimates of the lifetime risk of osteoporotic fracture among men aged 60 years and older (29%).(40)
We found greater age-related declines in lumbar spine BMD when measured in the lateral and midlateral compared with posterior-anterior projections. Mean lateral and midlateral spine BMD were 15% and 20% lower, respectively, than in normal young men whereas posterior-anterior spine BMD was not different. These findings confirm and extend previous cross-sectional results in women(22–27) and a report in men,(41) which also found greater age-related decrements in lumbar spine BMD when measured in the lateral and midlateral compared with posterior-anterior projections. Moreover, midlateral spine BMD decreased whereas posterior-anterior spine BMD increased significantly during 4 years of follow-up. Similar increases in posterior-anterior spine BMD have been noted in other longitudinal studies of elderly men.(2, 42) These cross-sectional and longitudinal data show that supine lateral DXA is superior to posterior-anterior DXA for detecting age-related bone loss in older men.
Supine lateral and midlateral DXA may be more sensitive to age-related bone loss than posterior-anterior DXA because such measurements avoid the cortical-rich posterior elements of the spine and are less influenced by spinal degenerative disease.(12, 43, 44) Manifestations of spinal degenerative disease such as disc space narrowing, vertebral end plate sclerosis, and osteophytes at the spinous processes and facet joints are prevalent among the elderly(13, 16, 21) and may be more common among older men than women.(13, 16, 21) Such changes increase the apparent posterior-anterior(12–20) and to a lesser extent lateral(12, 43, 44) spine BMD. In fact, posterior-anterior BMD may be increased by 15–20% in men with spinal osteophytes,(15, 16) an effect sufficient to influence the ability of such measurements to discriminate osteoporotic men.(19) In contrast, midlateral spine DXA and quantitative computed tomography (QCT) measurements are not significantly influenced by spinal degenerative disease.(12) Thus, clinicians and investigators should consider measuring lumbar spine BMD with lateral DXA or QCT in older men, particularly among those with established spinal degenerative disease.
Cross-sectional studies suggest that supine lateral DXA also discriminates vertebral fractures better than posterior-anterior DXA.(23, 45–47) For instance, supine lateral and midlateral DXA yield greater odds ratios and area under the receiver operating characteristic curves than posterior-anterior DXA among postmenopausal women.(46) More recently, Jergas et al.(48) found a stronger association between incident vertebral fractures and BMD measured in the lateral decubitus compared with posterior-anterior positions.(48) To our knowledge, similar data have not been reported among men. However, lateral and midlateral DXA are correlated more closely with spinal QCT measurements than posterior-anterior DXA,(22, 27, 47, 49) and a small study of Japanese men found better discrimination of vertebral fractures with QCT than posterior-anterior DXA.(19) Nevertheless, additional cross-sectional and prospective studies are needed to define better the relationships between posterior-anterior and lateral spine DXA measurements and fracture risk in both men and women.
We found that the rate of bone loss at the hip tended to increase with advancing age, as has been observed in other recent longitudinal studies of older men(2, 3) and women.(50, 51) Much less is known about age-related vertebral bone loss in men, and most data are derived from cross-sectional observations using posterior-anterior DXA. In our cohort spinal bone loss was 14% per decade when measured longitudinally in the midlateral view, similar to cross-sectional estimates using QCT (12% per decade) in other studies of middle-aged and elderly men.(52) This rate of loss was substantially greater than the approximate 2–3% per decade rate of loss at the total hip and its subregions, consistent with the concept that bone loss occurs at an earlier age and greater rate at skeletal sites rich in trabecular bone among older men.(52) The rate of loss was less with lateral (∼3% per decade) than midlateral spine measurements, possibly because lateral DXA includes the cortical rim and vertebral body end plates, which are largely cortical bone. We were unable to show an acceleration in vertebral bone loss with age in either the lateral or midlateral projection. Moreover, the rate of bone loss at the midlateral spine was positively and significantly correlated with age, suggesting that vertebral bone loss may decelerate with advancing age in men. These observations suggest that the pattern of vertebral bone loss may be quantitatively and qualitatively different from hip bone loss in aging men, although this possibility requires confirmation in a larger sample of men and with other techniques that measure cortical and trabecular bone separately, such as volumetric QCT.(53, 54)
The relatively small sample size of the young normal reference population may be a potential limitation of the present study. All of the young men were confirmed to have normal testosterone levels, however, and all had histories of puberty before the age of 14 years. Moreover, the mean and SDs of their posterior-anterior and femoral neck BMD are comparable with the manufacturer's original reference data for young men. Normative data for lateral and midlateral spine BMD in men are not available from the manufacturer for comparison. All of the older men were healthy volunteers recruited from the community and were white. Thus, our findings may not apply to nonwhite or institutionalized men. Finally, it will be important to define the true magnitude and pattern of trabecular and cortical bone loss at the spine in aging men using serial QCT measurements.
In conclusion, lateral DXA identifies considerably more men as osteoporotic and is more sensitive to age-related bone loss than posterior-anterior spine DXA. The present results, together with recent vertebral fracture prevalence estimates, indicate that spinal osteoporosis may represent a substantially greater health problem among older men than previously recognized.
Supported in part by the United States Public Health Service grants AR 35582, DK 46204, 1P60 AR44811, and R29 DK43341.