Osteoporosis International. Original Article. Bone Mineral Density and Vertebral Fractures in Men

Osteoporos Int (1999) 10:265 270 ß 1999 International Osteoporosis Foundation and National Osteoporosis Foundation Osteoporosis International Original Article Bone Mineral Density and Vertebral Fractures in Men E. Legrand 1, D. Chappard 2, C. Pascaretti 1, M. Duquenne 3, C. Rondeau 1, Y. Simon 1, V. Rohmer 3, M.-F. Basle 2 and M. Audran 1 1 Service de Rhumatologie, CHU d Angers; 2 LHEA Laboratoire d Histologie-Embryologie, CHU d Angers & Faculté de Médecine; and 3 Service de Médecine Interne et Endocrinologie, CHU d Angers, Angers, France Abstract. In women, many studies indicate that the risk of vertebral fragility fractures increases as bone mineral density (BMD) declines. In contrast, few studies are available for BMD and vertebral fractures in men. It is uncertain that the strength of the relationship between BMD and fractures is similar in magnitude in middleaged men and in postmenopausal women. In the present study, 200 men (mean age 54.7 years) with lumbar osteopenia ( <71.5) were recruited to examine the relationships between spine BMD and hip BMD and the associations of BMD with vertebral fractures. Lumbar BMD was assessed from L2 to L4, in the anteroposterior view, using dual-energy X-ray densitometry. At the upper left femur, hip BMD was measured at five regions of interest: femoral neck, trochanter, intertrochanter, Ward s triangle and total hip. Spinal radiographs were analyzed independently by two trained investigators and vertebral fracture was defined as a reduction of at least 20% in the anterior, middle or posterior vertebral height. Spinal radiographs evidenced at least one vertebral crush fracture in 119 patients (59.5%). The results of logistic regression showed that age, femoral and spine BMDs were significant predictors of the presence of a vertebral fracture. Odds ratios for a decrease of 1 standard deviation ranged from 1.8 (1.3 2.8) for spine BMD to 2.3 (1.5 3.6) for total hip BMD. For multiple fractures odds ratios ranged from 1.7 (1.1 2.5) for spine BMD to 2.6 (1.7 4.3) for total hip BMD. In all models, odds ratios were higher for hip BMD than for spine BMD, particularly in younger men, under 50 years. A <72.5 in the femur (total femoral site) was associated with a 2.7-fold increase in the risk of Correspondence and offprint requests to: Erick Legrand, MD, Service de Rhumatologie, CHU d Angers, F-49033 Angers Cédex, France. Fax: +33 2 41 35 35 70. vertebral fracture while a <72.5 in the spine was associated with only a 2-fold increase in risk. This study confirms the strong association of age and BMD with vertebral fractures in middle-aged men, shows that the femoral area is the best site of BMD measurement and suggests that a low femoral BMD could be considered as an index of severity in young men with lumbar osteopenia. Keywords: Bone mineral density; Hip; Men; Osteopenia; Osteoporosis; Vertebral fracture Introduction Osteoporosis has been defined as a disease characterized by low bone mass and microarchitectural deterioration of bone tissue, leading to enhanced bone fragility and a consequent increase in fracture risk [1]. In men the impact of osteoporosis has been often underestimated. Several recent epidemiologic studies have shown that about 30% of hip fractures occur in men. Perhaps as a result of higher prevalence of concomitant disease, the mortality associated with hip fractures in elderly men is at least twice as great as in women [2,3]. It has been estimated in the USA that the lifetime risk of hip fracture is about 6% and the risk of vertebral fracture about 5% in 50-year-old white men [2]. Furthermore, a recent multicenter European study has shown that the prevalence of vertebral deformity is similar in both sexes and that young men, aged 50 64 years, have a higher prevalence of deformity compared with similarly aged women [4]. The same findings were reported by Davies et al. [5] in a cohort of 529 men and 899 women: men in their fifties have a 29% prevalence of vertebral

266 E. Legrand et al. deformity compared with 10% in women [5]. Finally, sequelae of spine fractures have been reported in 63 men: a loss of height was documented in 49%, kyphosis in 54%, pain on standing in 64%, and 52% of the patients used analgesics daily [6]. Longitudinal studies are clearly the best way to examine the relationships between bone mineral density (BMD) and fracture risk in a general population. In women, many studies using single- or dual-photon absorptiometry indicate that the risk of vertebral fragility fractures increases as BMD declines [7 9]. The strength of this relationship is so clear that a value for BMD more than 2.5 SD below the young adult average value has been suggested to define osteoporosis in Caucasian postmenopausal women [10]. In contrast, few longitudinal studies are available in men for BMD and vertebral fractures [11 13]. The incidence of new vertebral fractures is very low in young and middleaged men, under 75 years: Nguyen et al. [13] observed only 10 vertebral fractures in 617 men (aged 60 74 years) during a follow-up period of 5 years. Gärdsell et al. [11] observed 17 vertebral fractures in 654 men during a follow-up period of 11 years and only 13 fractures in men under 70 years. Furthermore, men are exposed to greater trauma during their working life and it has been suggested that a number of vertebral fractures in middle-aged men are traumatic [4]. As a result of these methodologic problems, it is uncertain that the strength of the relationship between BMD and fractures is similar in magnitude in middle-aged men and in postmenopausal women. It has recently been shown by Huang et al. [14] that cross-sectional analyses of BMD and fractures (in women) can yield results similar to those based on long-term longitudinal studies. A crosssectional study of men with osteopenia can offer the opportunity to observe a higher number of fracture cases, to compare the accuracy of different BMD measurement sites and to study the relationships between BMD and fracture risk. In clinical practice, these results could be useful to validate information about the optimal measurement site and fracture threshold. In the present study, we have examined in a cohort of 200 men with lumbar osteopenia, the distribution of BMD at five regions of interest in the hip, the relationships between spine BMD and hip BMD and finally the associations of BMD with vertebral fracture. evaluate the presence of vertebral fractures. Their ages ranged from 26 to 80 years (mean 54.7 years). Bone Densitometry We measured BMD (area density in g/cm 2 ) using dualenergy X-ray absorptiometry (DXA; Hologic QDR 2000 densitometer, Hologic, Waltham, MA) operating in fan beam mode. The BMD was measured for all the patients on the same densitometer, by the same operator. Quality control scans were performed daily, using the manufacturer-supplied anthropomorphic spine phantom; the long-term coefficient of variation <2 years) is 0.40%. Lumbar BMD was assessed from L2 to L4, in the anteroposterior view, but fractured vertebrae were excluded from the analysis. At the upper left femur, hip BMD was measured at five regions of interest: femoral neck, trochanter, intertrochanter, Ward s triangle and total hip. The mean precision error of the DXA measurement is less than 1.5% for the lumbar spine and less than 2% for the hip BMD. The results were expressed in grams per square centimeter, as a Z-score and as a, using the manufacturer s reference values. Radiographic Assessment Anteroposterior and lateral spinal radiographs were taken at the time of the densitometry. They were analyzed independently by two trained investigators who were unaware of the patient s status. A patient was classified as having a vertebral fracture if both readers independently found a definite fracture. He was classified as normal if both readers independently found that the films were normal. When the readers disagreed, the films were reviewed in conference by both investigators. A vertebral fracture was defined as a reduction of at least 20% in the anterior, middle or posterior vertebral height. The criteria were thus: (1) in anterior wedge deformity, ratio of anterior/posterior height <80%; (2) in concavity deformity, ratio of middle/ posterior height <80%; (3) in compression deformity, ratio of posterior height/posterior height of the adjacent vertebra <80%. Materials and Methods Patients From January 1993 to December 1997, 335 men were referred in our unit (by general practitioners or private rheumatologists) for measurement of BMD because they had risk factors for osteoporosis, obvious osteopenia or vertebral fractures on radiographs. Two hundred male patients with lumbar osteopenia (BMD more than 1.5 SD below the young adult value) were included in a prospective study to examine BMD at the hip and to Statistical Analysis Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS release 6.1.3, SPSS, Chicago, IL). All results are expressed as the mean ± 1 SD. The nominal significance level was set at 0.05. Subjects were categorized according to the presence or absence of a vertebral fracture. Comparisons between group A (patients without a vertebral fracture), group B (patients with one vertebral fracture) and group C (patients with at least 2 vertebral fractures) were tested by multivariate analysis of variance and adjusted for age

Bone Mineral Density in Men 267 and body mass index (BMI). Pearson correlation coefficients were generated to examine the relationships among BMD measurements at the different sites. Stepwise linear regression was used to analyze the relationships between the number of fractures and age, BMI and BMD. The associations of BMD with vertebral fracture(s) were explored using logistic regression analysis: the outcome variable was the presence of vertebral fracture(s) (yes/no) or multiple fracture (yes/ no) and the possible predictor variables were age, BMI and BMD at the different regions. Backwards stepwise algorithms were used to arrive at the best model. Variables were included in the final model if the p value was <0.05. Results Radiographic Vertebral Evaluation Anteroposterior and lateral spinal radiographs evidenced at least one vertebral crush fracture in 119 patients (59.5%): 38 patients had only one vertebral fracture, 36 had 2 fractures, 13 patients had 3 fractures and 32 patients had 4 to 10 fractures. There was no fracture in 81 men (40.5%). Bone Mineral Density Femoral BMD and lumbar spine BMD are shown in Table 1. BMD, expressed as a Z-score or a, was lower in the lumbar spine than in the proximal femur. All Table 1. Densitometric data (mean values ± 1 SD), expressed in grams per square centimeter, Z-score and, in 200 men Site BMD g/cm 2 Z-score Neck 0.67±0.10 71.40±0.92 72.72±0.90 Trochanter 0.57±0.10 71.45±0.87 72.01±0.89 Intertrochanter 0.89±0.17 71.58±1.07 72.28±1.13 Ward s triangle 0.44±0.12 71.33±0.89 73.14±0.96 Total femur 0.76±0.13 71.54±0.99 72.29±1.03 Lumbar spine 0.74±0.10 72.87±0.87 73.35±0.88 The Z-score and values were calculated using the manufacturer s reference values. BMD measurements at the different femoral regions and at the lumbar spine were significantly correlated (p<0.0001 for all correlations). As seen in Table 2, the correlation was relatively low between spine BMD and femoral sites. After adjusting for age and BMI, stepwise linear regression showed that the trochanter site was the region with the higher correlation with lumbar spine measurement (r = 0.43, r 2 = 0.18, p<0.001). Bone Mineral Density and Prevalence of Vertebral Fractures As seen in Table 3, after adjusting for age and BMI there was no significant difference between groups A and B in femoral or spine BMD. In contrast, the multivariate analysis of variance showed that the BMD mean value was significantly lower at all sites in group C (patients with at least 2 fractures). The results of logistic regression showed that age, femoral BMD (at any site) and spine BMD were significant predictors of the presence of vertebral fracture(s) (Table 4). The two best models included age, BMI and trochanter or total femoral BMD. These models were able to predict the presence of fracture in 69.4% of the cases (trochanter BMD) or 68.4% of the cases (total femoral BMD). Patients with at least 3 vertebral fractures (n = 45) were categorized as multiple fractures. The results of logistic regression showed that age, femoral BMD (at any site) and spine BMD were significant predictors of the presence of multiple vertebral fractures. The odds ratios were higher for hip BMD at all sites than for spine BMD. Analysis was repeated in men under 50 years (n = 75) and men over 50 years (n = 125). The results showed that odds ratios were higher for hip BMD (particularly at the trochanteric site and at the total femoral site) in younger men (Table 5). Finally, we observed that a <72.5 in the femur (total femoral site) was associated with a 2.7-fold increase in the risk of vertebral fracture and a 2.0-fold increase in the risk of multiple vertebral fracture (fracture number >2). In contrast, a <72.5 in the spine was associated with only a 2-fold increase in risk of vertebral fracture and with a 1.3-fold increase in risk of multiple vertebral fractures. Table 2. Correlations between bone mineral density measurements at the different regions of interest Neck Trochanter Intertrochanter Total femur Ward s triangle Spine Neck 0.75 0.77 0.82 0.78 0.41 Trochanter 0.86 0.91 0.68 0.46 Intertrochanter 0.98 0.68 0.44 Total femur 0.71 0.45 Ward s triangle 0.35 Values represent Pearson correlation coefficients (p<0.0001 for all values).

268 E. Legrand et al. Table 3. Age, body mass index and bone mineral density (mean values ± 1 SD) in patients without vertebral fracture (group A), patients with only one fracture (group B) and patients with at least two fractures (group C) Group A: no fracture Group B: one fracture Group C: at least two fractures (n = 81) (n = 38) (n = 81) Age (years) 47.6±14.6 54.6±10.9 { 56.0±16.6 { BMI (kg/m 2 ) 24.2±3.9 23.5±4.1 { 24.3±4.5* Neck (g/cm 2 ) 0.70±0.09 72.4±0.8 0.68±0.08 72.7±0.7 0.63±0.09* { 73.2±0.9 Trochanter (g/cm 2 ) 0.61±0.08 71.7±0.8 0.59±0.08 71.9±0.8 0.53±0.10* { 72.4±0.9 Intertrochanter (g/cm 2 ) 0.96±0.15 71.8±1.0 0.90±0.16 72.2±1.1 0.83±0.17* { 72.7±1.1 Ward s triangle (g/cm 2 ) 0.49±0.10 72.7±0.8 0.44±0.11 73.1±0.9 0.40±0.11 { 73.5±0.9 Total femur (g/cm 2 ) 0.81±0.10 71.9±0.9 0.77±0.10 72.2±0.9 0.71±0.13* { 72.7±1.0 Lumbar spine (g/cm 2 ) 0.77±0.08 73.0±0.8 0.75±0.09 73.3±0.8 0.70±0.10 { 73.7±0.9 *.Statistically significant at the p<0.01 level, versus group B. {.Statistically significant at the p<0.001 level, versus group A. Table 4. Associations of age, bone mass index and bone mineral density with vertebral fractures in logistic regression models in 200 men Predictor At least one vertebral fracture Multiple fractures (at least three) Odds ratio (95% CI) p Odds ratio (95% CI) p Age (10 year increase) 1.5 (1.1 1.9) 0.005 1.3 (0.9 1.7) 0.005 BMI (1 kg/m 2 increase) 1.0 (0.9 1.9) NS 1.1 (1.0 1.2) NS Neck BMD (1 SD decrease) 1.9 (1.3 2.8) 0.002 2.3 (1.5 3.7) 0.004 Trochanter BMD (1 SD decrease) 2.2 (1.5 3.3) 0.001 2.5 (1.6 3.9) 0.001 Intertrochanter BMD (1 SD decrease) 2.1 (1.4 3.2) 0.003 2.6 (1.7 4.4) 0.005 Ward s triangle BMD (1 SD decrease) 2.1 (1.4 3.2) 0.002 2.5 (1.6 4.0) 0.005 Total femoral BMD (1 SD decrease) 2.3 (1.5 3.6) 0.001 2.6 (1.7 4.3) 0.001 Spine BMD (1 SD decrease) 1.8 (1.3 2.8) 0.001 1.7 (1.1 2.5) 0.01 BMI, body mass index The odds ratios shown represent the increase in odds of vertebral fracture corresponding to a 10 year increase in age, a 1 kg/m 2 increase in BMI or a 1 SD decrease in BMD. Table 5. Associations of age, body mass index and bone mineral density with vertebral fractures in logistic regression models in 200 men under 50 years (n = 75) and over 50 years (n = 125) Predictor Age <50 years Age >50 years Odds ratio (95% CI) p Odds ratio (95% CI) p Age (10 year increase) 1.6 (0.6 3.9) 0.05 1.5 (0.8 2.7) 0.03 BMI (1 kg/m 2 increase) 1.0 (0.9 1.2) NS 1.1 (0.8 1.1) NS Neck BMD (1 SD decrease) 1.8 (1.0 3.5) 0.05 2.0 (1.2 3.3) 0.01 Trochanter BMD (1 SD decrease) 3.1 (1.5 6.6) 0.003 2.0 (1.2 3.2) 0.005 Intertrochanter BMD (1 SD decrease) 2.4 (1.3 4.7) 0.01 2.0 (1.2 3.4) 0.01 Ward s triangle BMD (1 SD decrease) 2.9 (1.4 5.7) 0.005 1.8 (1.1 3.0) 0.02 Total femoral BMD (1 SD decrease) 3.1 (1.7 8.4) 0.005 1.9 (1.2 3.2) 0.01 Spine BMD (1 SD decrease) 2.0 (1.1 3.5) 0.02 1.9 (1.2 3.1) 0.01 BMI, body mass index. The odds ratios shown represent the increase in odds of vertebral fracture corresponding to a 10 year increase in age, a 1 kg/m 2 increase in BMI or a 1 SD decrease in BMD.

Bone Mineral Density in Men 269 Bone Mineral Density and Number of Vertebral Fractures We examined the relationships between BMD and the number of vertebral fractures in patients with at least one vertebral fracture (groups B and C). Statistical analysis showed that the fracture number was correlated with age, femoral BMD (at any site) and spine BMD. However, the stepwise linear regression analysis included trochanter BMD and age in the best model for predicting the number of fractures. In this model (r = 0.46, adjusted r 2 = 0.20, p<0.0001), trochanter BMD explained most of the variance in fracture number (partial r 2 = 0.15, p<0.0001), with a significant but lower age interaction (partial r 2 = 0.05, p=0.003). Discussion In the present study we examined the value of BMD at the spine and at five sites in the hip in a large cohort of males with lumbar osteopenia. We observed that measurements from each region of the proximal femur were significantly correlated with the spine BMD measurement. However, correlation coefficients were relatively low, as previously described in normal and osteoporotic women by Griffin et al. [15] and Mazess et al. [16]. In other words, spine BMD measurement does not adequately predict femoral BMD in men. Second, we observed that BMD was lower in the lumbar spine. When expressed as a Z-score, there was, for example, a difference of nearly 1.4 SD between spine BMD and total hip BMD. These results, which are in agreement with a previous report [17], strongly suggest that bone loss is not uniform in osteopenic middle-aged men, there being a preferential impact on trabecular bone and relative preservation of cortical bone. The third striking observation was that the decrease in BMD, at any site, was clearly associated with an increased vertebral fracture risk: after adjusting for age and BMI, odds ratios for a decrease of 1 SD ranged from 1.8 to 2.3, depending on the measurement site. The magnitude of this effect agrees with previous studies in nonselected men [12,13,18] or in postmenopausal women [7 9]. To test the hypothesis that spine osteoarthritis could affect the ability of spine BMD to predict fractures, we compared odds ratios for BMD in men under or over 50 years. On one hand, we observed that the odds ratio for spine BMD were nearly the same, indicating that osteoarthritis does not really affect the accuracy of spine BMD measurement in men aged from 50 to 80 years. On the other hand, the odds ratio for femoral BMD at the trochanteric and total femoral areas were higher in young men under 50 years. This result strongly suggests that a low BMD in the total femur or trochanter, which probably reflects trabecular and cortical bone loss, is an important criterion for estimating the severity of osteopenia in young men. Several authors have shown that the risk of vertebral deformity is as great, and even greater, in young men as in women [4,5] and suggested that these deformities are in part traumatic. This is a crucial methodologic problem in assessing the influence of low BMD in vertebral fracture risk in men. However, the higher the number of vertebral deformities in a patient, the higher is the probability that these deformities are osteoporotic and not traumatic. We observed that decreased BMD was clearly associated with the risk of multiple vertebral fractures and that femoral BMD (at any site) was more strongly associated with multiple fractures than spine BMD. Jones et al. [19] have shown that in elderly men (after 60 years) femoral BMD was more consistently associated with vertebral deformity than spine BMD; they supposed that this was due to the high prevalence of spinal degenerative diseases in elderly men. Our results strongly suggest that total femoral area is also the preferred site of BMD measurement to predict vertebral fracture risk in young and middle-aged men. Finally, we observed that the number of fractures, which reflects the severity of osteoporosis, was correlated with age and femoral or spine BMD. However, the best model, which included age and trochanter BMD, explained only 20% of the variance in fracture number. This result suggests that in addition to bone mass, bone quality and bone architecture could play a crucial role in men with osteopenia. Using a model of vertebral trabecular bone architecture, Jensen et al. [20] have shown that the apparent bone stiffness varies by a factor of 5 10 from a perfect cubic lattice to a network of maximal irregularity, even if trabecular bone volume remains constant. Using histomorphometric analysis of iliac bone biopsies, we have recently shown that altered trabecular bone microarchitecture was strongly associated with vertebral fractures in male osteoporosis [21]. To predict the severity of osteoporosis, BMD measurements have to be combined with information about bone quality and particularly trabecular microarchitecture. In summary, the present study confirms the strong association of age and BMD with vertebral fractures in young and middle-aged men, shows that the femoral area is the best site of BMD measurement for estimation of fracture risk and suggests that a low femoral BMD could be considered as an index of severity in young men with lumbar osteopenia. A <72.5 in femur, which is associated with a 2.7-fold increase in the risk of vertebral fracture, could be a rational diagnosis and treatment threshold in men. References 1. Consensus development conference. Diagnosis, prophylaxis and treatment of osteoporosis. Am J Med 1993;94:646 50. 2. Melton LJ III, Chrischilles EA, Cooper C, Lane AW, Riggs BL. How many women have osteoporosis? J Bone Miner Res 1992; 7: 1005-10. 3. Kellie SE, Brody JA. Sex-specific and race-specific hip fracture rates. Am J Public Health 1990;80:326 8. 4. O Neill TW, Felsenberg D, Varlow J, Cooper C, Kanis JA, Silman AJ and the European Vertebral Osteoporosis Study

270 E. Legrand et al. Group. The prevalence of vertebral deformity in European men and women: the European vertebral osteoporosis study. J Bone Miner Res 1996;11:1010 8. 5. Davies KM, Stegman MR, Heaney RP, Recker RR. Prevalence and severity of vertebral fracture: the Saunders county bone quality study. Osteoporos Int 1996;6:160 5. 6. Scane AC, Sutcliffe AM, Francis RM The sequelae of vertebral crush fractures in men. Osteoporos Int 1994;4:89 92. 7. Cummings SR, Black DM, Nevitt MC, Browner W, Cauley J, Ensrud K, et al. Bone density at various sites for prediction of hip fractures. Lancet 1993;341:72 5. 8. Gärdsell P, Johnell O, Nilsson BE. Predicting various fragility fractures in women by forearm bone densitometry: a follow-up study. Calcif Tissue Int 1993;52:348 53. 9. Melton LJ, Atkinson EJ, O Fallon WM, Wahner HW, Riggs BL. Long-term fracture prediction by bone mineral assessed at different skeletal sites. J Bone Miner Res 1993;8:1227 33. 10. Kanis JA, Melton LJ III, Christiansen C, Johnston CC, Khaltaev N. The diagnosis of osteoporosis. J Bone Miner Res 1994;9:1137 41. 11. Gärdsell P, Johnell O, Nilsson BE. The predictive value of forearm bone mineral content measurements in men. Bone 1990;11:229 32. 12. Ross PD, Kim S, Wasnich RD. Bone density predicts fracture risk in both men and women [abstract]. J Bone Miner Res 1996;11(Suppl 1):S127. 13. Nguyen TV, Eisman JA, Kelly PJ, Sambrook PN. Risk factors for osteoporotic fractures in elderly men. Am J Epidemiol 1996;144:255 63. 14. Huang C, Ross PD, Wasnich RD. Short-term and long-term fracture prediction by bone mass measurements: a prospective study. J Bone Miner Res 1998;13:107 13. 15. Griffin MG, Rupich RC, Avioli LV, Pacifici R. A comparison of dual energy radiography measurements at the lumbar spine and proximal femur for the diagnosis of osteoporosis. J Clin Endocrinol Metab 1991;73:1164 9. 16. Mazess RB, Barden H, Ettinger M, Schultz E. Bone density of the radius, spine, and proximal femur in osteoporosis. J Bone Miner Res 1988;13:107 13. 17. Vega E, Ghiringhelli G, Mautalen C. Bone mineral density in osteoporotic men with vertebral fractures [abstract]. J Bone Miner Res 1994;9:S383. 18. Lunt M, Felsenberg D, Reeve J, Benevolenskaya L, Cannata J, Dequeker J, et al. Bone density variation and its effects on risk of vertebral deformity in men and women studied in thirteen European centers: the EVOS study. J Bone Miner Res 1997;12:1883 94. 19. Jones G, White C, Nguyen T, Sambrook PN, Kelly PJ, Eisman JA. Prevalent vertebral deformities: relationship to bone mineral density and spinal osteophytosis in elderly men and women. Osteoporos Int 1996;6:233 9. 20. Jensen KS, Mosekilde Li, Mosekilde L. A model of vertebral trabecular bone architecture and its mechanical properties. Bone 1990;11:417 23. 21. Legrand E, Chappard D, Pascaretti C, Krebbs S, Basle MF, Audran M. Trabecular bone architecture is a determinant of vertebral fractures in men [abstract]. Osteoporos Int 1998;8:14. Received for publication 27 October 1998 Accepted in revised form 22 February 1999