The aim of this study was to quantify the biomechanical basis for vertebral fracture risk in elderly men and women. A bone is likely to fracture when the loads imposed are similar to or greater than its strength. To quantify this risk, we developed a fracture risk index (FRI) based on the ratio of the vertebral body compressive load and strength. Loads were determined by upper body weight, height, and the muscle moment arm, and strength was estimated from cross-sectional area (CSA) and volumetric bone mineral density (vBMD). With loads less than the strength of the bone, the FRI remains <1. For any given load, once bone strength diminishes due to a falling vBMD, the FRI will increase. Should FRI approach or exceed unity, structural failure of the vertebra is likely. We measured vertebral body CSA vBMD of the middle zone of third lumbar vertebra by lateral and posteroanterior (PA) scanning using dual-energy X-ray absorptiometry (DXA) and calculated vertebral compressive stress (load per unit area) in 327 healthy men and 686 healthy women and 26 men and 55 postmenopausal women with vertebral fractures. Activities that require forward bending of the upper body caused ∼10-fold more compressive stress on the vertebra compared with standing upright. Men and women had similar peak vBMD in young adulthood. Because men have greater stature than women, the loads imposed on the vertebral body are higher (3754 ± 65 N vs. 3051 ± 31 N; p < 0.001). However, because CSA also was higher in men than women, peak load per unit CSA (stress) did not differ by gender (317.4 ± 4.7 N/cm2 vs. 321.9 ± 3.3 N/cm2, NS). The FRI was similar in young men and women and well below unity (0.42 ± 0.02 vs. 0.43 ± 0.01; NS). Gender differences emerged during aging; CSA increased in both men and women but more so in men, so load per unit area (stress) diminished but more so in men than in women. vBMD decreased in both genders but less so in men. These changes were captured in the FRI, which increased by only 21% in men and by 102% in women so that only 9% of elderly men but 26% of elderly women had an FRI ≥ 1. Men and women with vertebral fractures had an FRI that was greater than or equal to unity (1.03 ± 0.13 vs. 1.35 ± 0.13; p < 0.05) and was 2.04 SD and 2.26 SD higher than age-matched men and women, respectively. In summary and conclusion, young men and women have a similar vBMD, vertebral stress, and FRI. During aging, CSA increases more, and vBMD decreases less in men than in women. Thus, fewer men than women are at risk for fracture because fewer men than women have these structural determinants of bone strength below a level at which the loads exceed the bone's ability to tolerate them. Men and women with vertebral fractures have FRIs that are equal to or exceed unity. The results show that a fracture threshold for vertebrae can be defined using established biomechanical principles; whether this approach has greater sensitivity and specificity than the current BMD T score of −2.5 SD is unknown.
AS AGE advances, the proportion of men and women sustaining fractures increases.1-3) The lifetime risk for any fracture is about 50% in women and 30% in men.(2) Three factors are believed to account for the lower incidence of fractures in men than in women. First, men have larger bones than women, which confer greater strength in bending.4-6) Second, men lose less bone during aging so that bone mineral density (BMD) at any age is higher in men than women.(7) Third, the loss of bone occurs by less architectural disruption in men than in women.(8) Although this third factor remains controversial because men and women with same trabecular volumetric BMD (vBMD) measured by peripheral quantitative computed tomography (pQCT) have similar bone strength, despite purportedly less architectural disruption.(9)
One way to examine the relative importance of vertebral size and BMD on fracture risk is the application of biomechanics. In young adulthood, men do have bigger bones than women. Thus, the absolute load tolerated by men is greater than the load tolerated by women but this is not the loading scenario that occurs in life. Men have greater body weight (BW), body height, and larger muscles attached to the bone so that, at least for the vertebral body, the load imposed per unit cross-sectional area (CSA) of bone is the same in men and women.(6, 10) Fractures are likely to occur when the load imposed on the bone exceeds the bone strength. The loads encountered in daily life are a function of BW, height, and bending moments. The ability to resist them is a function of the CSA and vBMD of the bone.(11)
In this study, we hypothesized that structural failure emerges during aging in men and women because of the changing relationship between the load on the bone and its strength; loads will be less well tolerated because bone loss will reduce bone strength in both men and women. We also hypothesized that fractures occur less commonly in men than in women because the net loss of bone is less in men so that bone strength declines less in men than in women.
To address these hypotheses, we measured the changes in the size and vBMD of the vertebral body that account for the age-related increase in bone fragility and compared these changes between men and women. Based on engineering principles and the concept of a “factor of risk” developed by several investigators,(12, 13) we applied a fracture risk index (FRI) or quantitative measure of bone fragility applicable to both genders using structural determinants of bone strength. We asked (i) How do different loading scenarios affect the lumbar vertebral body compressive stress (load per unit area)? (ii) Is vertebral body compressive stress or FRI different in young men and women or in elderly men and women? (iii) Do men and women have a similar fracture threshold?
MATERIALS AND METHODS
We studied 327 healthy men and 686 healthy women aged 18-92 years: 85 men and 282 premenopausal women aged between 18 and 43 years (young subjects), 85 men and 209 women aged between 44 and 59 years, and 157 men and 195 women aged between 60 and 92 years (elderly subjects). All subjects were healthy volunteers without history of spine and/or hip fractures recruited from the local community (Melbourne, Australia) by advertisement as part of the ongoing research. We also excluded persons with illnesses known to affect bone metabolism (e.g., thyroid disease, hyperparathyroidism, Paget's disease, or chronic liver or renal disease) and those receiving medications known to affect bone density such as corticosteroids.
We also studied 26 men aged 41-84 years and 55 postmenopausal women aged 46-85 years with one or more vertebral fractures based on a reduction of vertebral body height by >20% on a lateral radiograph. The patients were recruited from the Metabolic Bone Clinic of the Austin and Repatriation Medical Center in Melbourne. We excluded patients with a fracture of the third lumbar vertebra (the measurement site) and those taking medication known to affect bone. All participants gave informed consent. The study was approved by the ethics committee of the Austin and Repatriation Medical Center.
Measurements of vertebral body dimensions, bone mineral content, and vBMD
Bone mineral content (BMC), areal BMD (aBMD), and dimensions of the third lumbar vertebra (L3) were measured by posteroanterior (PA) and lateral scanning using dual-energy X-ray absorptiometry (DXA; DPX-L, Version 1.3z; Lunar Corp., Madison, WI, USA). Vertebral body width was obtained from the PA scan. Vertebral body depth and height were derived from the lateral scan. Vertebral body height (the average of anterior, middle, and posterior heights) was measured manually during lateral scanning using the DXA ruler function. The average vertebral body depth was derived from the vertebral body area during lateral scanning divided by vertebral body height. Vertebral body CSA was calculated as π × width/2 × depth/2.(14)
Measurement of the vertebral body midportion omits end plate sclerosis and was used for biomechanical calculations. The middle one-third of the vertebral body was measured to determine BMC, CSA, and the middle vertebral body depth.(10) vBMD of the middle one-third vertebral body was calculated as middle vertebral body BMC/volume, where volume = CSA × one-third height.(10) This volume calculation assumes that the vertebral body is an ellipsoid cylinder(14) and is the most accurate method of predicting vertebral body volume measured by submersion in vitro.(15) The CV for these measurements ranged between 1.5% and 5.7% based on scanning 15 subjects twice within 3 months.
Measurements of spinal extensor muscle moment arm
Loads on the lumbar vertebrae are imposed through the moment arm of the spinal extensor muscles. The moment arm is derived from the lateral image of the DXA scan. We identified two posterior landmarks: (i) the posterior edge of the higher density region (the transverse process) and (ii) the anterior edge of the high density region. Two lengths from each scan were measured: length 1, the distance from the anterior edge of the vertebral body to the anterior edge of the high density region; and length 2, the distance from the anterior edge of the vertebral body to the posterior edge of the high density region. The extensor muscle moment arm is the distance from the center of the vertebral body to the middle lamina. Thus, we estimated the moment arm (d) as the average of length 1 and length 2 minus one-half of the middle vertebral body depth, that is, (length 1 + length 2 − midvertebral body depth)/2. The CV ranged from 1.6% to 1.9%.
Calculation of lumbar spine stress
Mechanical stress (load per unit area) on the spine is determined by the upper BW and the extensor muscle force divided by the vertebral body CSA. The extensor muscle force is a function of spinal extensor muscle moment arm and the angle of bending forward (θ). The stress on the spine increases the further the trunk is flexed, that is, with increasing θ. The term θ = 0° represents perfect posture, and θ = 90° is the worst scenario, where the spine is stressed maximally because the weight of the upper body creates a forward bending moment that is supported by the extensor muscles of the lower back. A third scenario simulated is rising from a chair, a common daily activity during which the lumbar spine flexed to about 50°.(16) We calculated stress on the third lumbar vertebral body for the three scenarios described previously: (i) stress 1, standing erect (θ = 0°), (ii) stress 2, rising from a chair (θ = 50°); and (iii) stress 3, bending forward (θ = 90°).
This calculation is based on the assumption that the center of upper BW is directly above the third lumbar vertebral body centrum, that is, there is no forward bending moment and that the spinal muscle forces are zero (Fig. 1). Axial compressive stress on the third lumbar vertebral body was where 0.455 BW is the proportion of BW in newtons (N = kg ∗ 9.81) above the third lumbar vertebrae,(17) and CSA is the average vertebral body CSA.
In the following sections, we consider scenarios in which the extensor muscles posterior to the vertebral column are contracted.
Rising from a chair:
During the sit-to-stand movement, the torso flexes forward and extends upward until an upright posture is achieved. The maximum stress on the spine occurs when forward flexion reaches 50° from vertical at which point the weight of the upper body creates a forward bending moment that is supported by the extensors of the lower back. Stress on the L3 vertebral body when rising from a chair was calculated as where H is body height in centimeters and 0.186 H is the distance from the L3 vertebral body to the center of mass of the upper body, 50° is the angle of forward flexion, and d is the extensor muscle moment arm as defined previously. This equation is based on the assumption of static equilibrium, that is, the moment caused by the extensor muscle force (F2 × d; Fig. 1) is exactly equal to the forward bending moment caused by the upper BW (F1 × D), thus F2 = F1 × D/d.
Leaning forward at a 90° angle causes considerable stress on the lumbar spine and is most likely of the three scenarios considered here, to cause a fracture.
This equation is based on the same assumptions as stress 2. In this case, the axial force on the L3 vertebral body is equal to the extensor muscle force (F2 in Fig. 1).
Calculation of FRI
Fracture is likely to occur when the load exceeds the vertebral body strength. The loads on the vertebral body encountered in daily life are a function of upper BW, height, and extensor muscle moment arm. The ability to resist load is a function of vertebral body CSA and its vBMD. The FRI is defined as the peak vertebral body stress (load per unit area) divided by the vertebral strength. Consequently, an FRI ≥ 1 indicates high risk of fracture and lower values reflect increasing margins of safety. For the vertebral body, the FRI should take into account vBMD. Vertebral body strength S (in N/cm2) is given by where ρash is the vertebral body volumetric ash density excluding the end plates.(18) The exponent 1.6 describes the nonlinear relationship between failure stress and ash density.(19) DXA vBMD is a measure of bone mineral per volume, as is ash density and ρash, which was measured excluding vertebral body end plates and is virtually identical to the DXA measurement of the middle one-third vertebral body vBMD. Therefore, vBMD may be substituted for ρash and FRI is calculated as
Stress 3 (θ = 90°) was used in this equation because that loading scenario created the greatest vertebral stress and, thus, the highest risk of fracture.
The data were expressed in absolute terms and as SD scores. The T score is the observed value minus young mean value divided by the SD in the young normal subjects and expresses the group mean as the number of SDs above or below the young normal mean (of zero). The Z score is the observed value minus age-predicted mean divided by the SD in the age-matched controls and expresses the group mean as the number of SDs above or below the age-matched normal mean (of zero). Polynomial regression analysis was used to determine the relationship between FRI and age. One sample t-test was used to determine whether the T or Z scores for a given trait differed from zero. Two sample t-tests were used to determine the significance of any difference within and between sites in healthy subjects and patients with fractures. A Mann-Whitney U test was used to analyze measurements of FRI, because these measurements included several data points that appeared to deviate from a normal distribution. Results were regarded as statistically significant at the 5% level (two-tailed).
The results are presented in full in the Table 1 (mean ± SEM; p values). Vertebral body CSA was about 24.8% greater in young men than in young women and increased with age by 9.7% in men and by 3.3% in women. Thus, elderly men had 32.6% greater CSA than elderly women. There was no gender difference in vertebral body vBMD between young men and women. Midvertebral body vBMD decreased by 15.7% in men and by 34.2% in women across age, being 22.6% lower in elderly women than in elderly men (p < 0.001).
Table Table 1.. Age, Height, Weight, Third Lumbar Vertebral Body CSA, PA-aBMD and Lateral Scanning (Lat)-aBMD, Middle One-Third Vertebral Body vBMD, Stress (Load per Unit Area) for the Three Loading Scenarios, and FRI in Healthy Young Men and Women (Range, 18-43 Years), Elderly Men and Women (Range, 60-92 Years), and Patients With Vertebral Fractures
Activities that involve contraction of the spinal extensor muscles generate considerable vertebral loading; load per unit CSA (stress) on the vertebral body was 8-fold higher when rising from a chair (θ = 50°) and 10-fold higher for bending forward (θ = 90°) compared with perfect posture (θ = 0°; Table 1).
Despite the gender difference in peak vertebral body CSA in young adulthood there were no gender differences in peak load per unit CSA (stress) at any loading scenarios because the larger bone was subjected to correspondingly larger absolute loads in men than in women (3754 ± 65 N vs. 3051 ± 31 N during bending forward; p < 0.001). Load per unit CSA (stress) decreased with age both in men and in women but more so in men because CSA increased more in men than in women reducing the load per unit area more in men than in women. The absolute fall in load per unit CSA (stress 3) was 13.2% in men and 5.8% in women (Table 1).
There was no gender difference in FRI in young adulthood for any loading scenarios. Given the greater increase in CSA and lesser fall in vBMD in men than in women during aging, the FRI increased by only 21% in men and by 102% in women so that only 9% of elderly men but 26% of elderly women had the FRI ≥ 1 (Fig. 2).
In men and women with vertebral fractures, load per unit area (stress) was reduced relative to age-matched controls rather than increased (despite the smaller vertebral body CSA) because BW and height were lower. The FRI was equal to or exceeded unity in men and women with fractures and increased relative to age-matched controls (Fig. 3). There was overlap when comparing FRI in patients with fractures and controls (Fig. 4).
We asked three questions. First, we found that activities that require forward bending, such as rising from a chair or bending forward to pick up an object from the floor, cause up to 10-fold more compressive stress (load per unit CSA) on the spine compared with standing upright. Second, vertebral body compressive stress was no different in young men and young women, regardless of the loading scenario. However, during aging, the FRI increased less in men than in women because the greater increase in CSA with aging reduced stress more greatly and the lesser fall in vBMD maintained bone strength more greatly in men than in women. Consequently, fewer elderly men than elderly women are at risk of fracture by having determinants of bone strength such as bone size and vBMD below a level at which loads on the bone exceed its strength. Third, men and women with fracture had FRI values that were equal to or exceeded unity, the threshold at which fracture is likely. Moreover, the FRI was higher than age-matched controls, suggesting that this threshold value of unity, derived based on established biomechanical principles, may be a rational and useful cut-off value to identify persons at risk for fracture.
The results indicated that vertebral stress varies greatly with different activities. In fact, for the three loading scenarios examined, L3 vertebral stress varied by an order of magnitude. The largest vertebral stresses occur when contraction of spinal extensor muscles is required to support the upper BW during forward bending at the waist. Stress on the spine caused by standing upright was small and conferred little fracture risk. In loading scenarios that included forward bending, the majority (92-100%) of the spinal stress was caused by extensor muscle force (Fig. 1, F2). Nontraumatic vertebral fracture is most likely to be caused by strong contraction of extensor muscles; hence, the muscle force needs to be measured to assess vertebral fracture risk. Muscle forces during different activities can be estimated from direct electromyographic measurements(20) or from calculations based on the assumption of static equilibrium, as was done in this study. Estimated spinal compressive force was about 3800 N in young men bending forward. Another study has shown that the spinal compressive force can exceed 6000 N in young men squat-lifting weights.(20) Based on this information, we can establish a gradation of vertebral fracture risk associated with loading scenarios—upright standing is low risk, bending forward increases risk about 10-fold over upright standing, and weight lifting increases risk about 1.5-fold over bending forward. Others have categorized vertebral fracture risks for several other activities such as lifting and opening a window.(13)
Absolute loads on the vertebra were higher in men than in women because of their greater stature. However, regardless of the loading scenario, the load per unit area imposed on the vertebral body was no different in young men and women. This is consistent with the notion of scaling—a higher vertebral CSA in men is associated with their larger stature. vBMD and FRI did not differ in young men and women. Hence, at peak young adulthood, vertebrae are appropriately sized for the loads placed on them, the loads on the bone are well below their strength, and fractures are uncommon in young men and women.
It is during aging that the gender difference develops in the ratio of load/strength (FRI). The load on the spine changes little during aging in each of the loading scenarios studied but vertebral load per unit area (stress) decreased because of increasing CSA. CSA increased in both genders, but more so in men, so that load per unit area (stress) diminished in both genders but more so in men. vBMD decreased in both genders but less so in men. These age-related changes were captured in the FRI, which increased with age in both genders but by only 21% in men and 102% in women, reflecting the combined effects of a greater increase in CSA and lesser net loss of vBMD in men than in women. These results suggest that vertebral bodies are appropriately adapted to imposed loading in young individuals but do not remain so during aging and so become susceptible to fracture.
For men and women with vertebral fractures, the FRI was higher than age-matched controls. Men with fractures had FRI values above the mean, whereas FRI values for women with fractures overlapped those of women with no fractures. This overlap reflects the uncertainty in the FRI estimation. Causes for uncertainty may reflect the heterogeneity of activity levels associated with loading and the many factors that contribute to the structural integrity of the spine such as qualitative properties of bone not captured in the vBMD measurement such as trabecular connectivity or the presence of intervertebral disc disease.(21)
It was assumed that fractures are caused by those activities causing the highest compressive loads on the lumbar spine. Although this is a reasonable assumption, it is difficult to estimate the peak spinal loads for different individuals because of the variance in daily activities. Furthermore, it is not possible to account for the severity of falls or other traumatic events that might cause large loads on the spine. Our current analysis of spinal stress includes activities that are common among individuals, that is, rising from a chair or bending forward to pick up something such as an object. This analysis could be improved by a better knowledge of daily spine-loading activities in young and elderly people.
The FRI was equal to or greater than unity in men and women with vertebral fractures, suggesting that men and women with vertebral fractures have a similar fracture threshold based on biomechanical calculations (FRI ≥ 1). Women with fractures had a higher FRI than men with fractures. In addition, more women without fractures had an FRI ≥ 1. Thus, women can achieve higher FRI before sustaining vertebral fractures. The reason for this finding is unclear. It is possible that women do not engage in activities that cause large loads on the spine as frequently as men do and, thus, impose less biomechanical risk.
To summarize, fracture risk is the same in young women and men but, during aging, subperiosteal bone formation increases vertebral CSA more in men than in women and vBMD decreased less in men than in women. Because of the fall in vBMD and the rise in CSA, the FRI increased less in men than in women so fewer elderly men than elderly women had FRI equal to or greater than unity. The lower incidence of vertebral fractures in men than in women is the result of fewer men than women having an FRI above unity.
We thank research nurse Jan Edmonds for her assistance with subject recruitment during this study and senior technologist Patricia D'souza for her technical assistance.