THIS ISSUE of the Journal includes an article by Beck et al. entitled “Structural Trends in the Aging Femoral Neck and Proximal Shaft: Analysis of the NHANES III DCS Data.”(1) The authors show a change in the geometry of femurs with age, which appears to maintain the mechanical integrity of the femoral neck. Furthermore, the authors suggest that a loss in bone mass in the proximal femur may not be indicative of an increase in fracture risk, because of this geometric compensatory factor. These results, as well as others in the literature, encourage us to “step back” and review what we understand about fractures of the aging skeleton and what factors seem to be associated most with an increase in fracture risk.
Popular conventional wisdom suggests that the most dominant factor related to skeletal fragility due to aging or osteoporosis is reduced bone mass. However, on further questioning, many experts would caution that measures of bone mass or density alone cannot reliably predict fracture risk in patients. Given this controversy and the fact that hundreds of studies have been conducted to examine the factors associated with osteoporotic fractures, it seems appropriate to revisit current perspectives. The following statements were excerpted directly from the literature and illustrate the continuing disagreements about fracture risk factors.
The recent interim report from the World Health Organization (WHO) Task Force for Osteoporosis, although briefly mentioning microstructure, recommends using only bone mineral density (BMD) for determining fracture risk, and their research recommendations focus on methods to assess and alter BMD and bone turnover.(2)
“Mechanically, osteoporosis is rather simple: there is not enough bone tissue, and therefore the bone structure is weak and fractures occur. The complexity is in the bone remodeling biology that leads to diminished bone structure.”(3)
“ … the predictive ability of bone mass was … better than that of serum cholesterol for cardiovascular disease.”(4)
“ …the osteoporosis model proposed by the World Health Organization (WHO), which assumes that fragility depends only on a single mean value of bone mineral density (BMD) for a patient, is over-simplistic and requires upgrading to include indices representing the distribution of bone mineral.”(5)
“Risk factors not related to bone mass contribute significantly to hip fracture risk.”(6)
“Recent studies have shown that factors related to fall biomechanics may play as important a role in the etiology of hip fracture as age-related bone loss.”(7)
“ … there is no doubt that deterioration of trabecular architecture is a contributor factor to the bone fragility in osteoporosis … ”(8)
“Measurements of bone mineral density can predict fracture risk but cannot identify individuals who will have a fracture. We do not recommend a programme of screening menopausal women for osteoporosis by measuring bone density.”(4)
How has the community arrived at such opposing viewpoints about a single condition? Is it a single condition? Osteoporosis and skeletal fragility in general are extremely complex, cross many hierarchical levels, and may be impacted by many physiological variables. From a biomechanical perspective, there are a variety of factors to consider:
- (1)The loss of bone mass or BMD is an extremely important feature related to osteoporotic fractures. BMD explains a significant portion of the risk of osteoporotic fractures (9–11) and correlates significantly with bone strength. (12–16) The risk of fracture increases by 50-150% with each SD decrease in bone mass as measured by dual-energy X-ray absorptiometry (DXA).(4) In fact, the performance of BMD in predicting fracture is at least as good as that of blood pressure in predicting stroke and better than the use of serum cholesterol in predicting cardiovascular disease.(4) However, BMD by itself does not determine if an individual will sustain a fracture.(4,17) There is substantial overlap between the normal and osteoporotic populations.(18) The relationship between BMD and fracture risk is less consistent for vertebral fractures than for hip fractures.(14)
- (2)The geometry of the bones and the distribution of bone mass play an important role in determining bone strength and appear to differ in ways that parallel the risk of osteoporotic fracture.(19,20) The hip axis length has been found to be correlated with the risk of hip fracture,(9,21) and may be an explanation for racial differences in hip fracture(22) and the secular trends for increasing hip fracture since the 1950s.(23) The accompanying article and others(24) have shown aging trends in the geometry of the aging femur. Sandor,(5) Cody,(25) and McCubbrey(26) have shown that the distribution patterns of bone mass, not just its quantity, are important in predicting whole bone strength. Likewise, the cross-sectional area of vertebrae is important in determining their strength. (27–31) Although cortical thickness has been shown to decrease with age,(17) geometric changes, particularly periosteal expansion, (32–35) may more than compensate for this loss by increasing the structural resistance to load.
- (3)The force applied to each bone must be greater than its strength in order for a fracture to occur. Hayes et al.(36) have clearly described that there are two factors that determine whether a bone fractures, the applied load and the fracture load. They incorporate these into a fraction, applied load/fracture load, which describes the risk of fracture. The force applied to the bone is influenced by padding (artificial or body fat), the distribution of mass (including any carried weight), and the direction of the fall.(7,37) These are particularly important relative to hip fractures. Loading rate also has been shown to influence fracture strength,(38) by modifying the strength of the bone.
- (4)Trabecular bone microarchitecture, separate from density or bone volume fraction, is another important determinant of bone strength(8) and appears to differ between osteoporotic and normal individuals and with age.(19, 32, 33) Several studies measuring microarchitecture in two and three dimensions have correlated measures of density and architectural organization with trabecular bone stiffness and strength. (39–41) In addition, changes in architectural features with age and differences between males and females have been documented.(42,43) In a recent, more direct study, trabecular bone from hip fracture patients was found to be more anisotropic than a matched cohort of female subjects who died without fracture.(44)
- (5)Microdamage and/or increased remodeling density may play roles in weakening bone tissue, predisposing to fracture. (45–47)
- (6)Changes in the bone tissue properties may be associated with osteoporotic fracture. Shifts in the mineral content of the tissue will influence its stiffness(48) and strength, and have been related to bone fragility.(17, 49, 50) Recent data suggest that there may be greater variability in local mineral content (even without differences in average mineral content) in hip fracture patients compared with controls.(51)
- (7)Genetics, body size, and environmental factors may influence fracture risk through expression of many of the factors described previously. For example, it has been reported that up to 80-85% of BMD is determined by genetics.(52,53)
- (8)Currently approved therapies are able, at best, to increase bone density by up to 10%,(54) although the risk of fracture decreases by a much larger extent.(14)
- (9)Recent experiments investigating biomechanics at the level of the cell have found no evidence of factors at this scale that appear to contribute to increased fracture risk. It has been suggested that degradation in the ability of osteocytes to sense strain could be responsible for the loss of bone associated with osteoporosis. This change in the ability of the cell to sense strain could be caused by changes in the geometry of the osteocyte and its lacuna, changes within the cell, changes in the fluid flow surrounding the cell, etc. It has been shown that there are no differences in the shape of the osteocyte lacunae in osteoporotic individuals compared with those without fracture.(51) Other experimental measurements of potential biomechanical changes at this level are technically difficult and to our knowledge have not been attempted to date.
- (10)Achieving effective and efficient experimental designs in the study of osteoporosis is extremely difficult. First, it is difficult to define and then find a true control. Should the control be a young adult, age-matched without a fracture, or age-matched and BMD-matched without a fracture? Is tissue from a cadaver without fracture “normal”? Would a fracture have occurred within a few days had the individual lived longer? Second, it is often difficult to obtain experimental samples for in vitro research. Fractured vertebrae are not removed after fracture, and hip fractures are most often fixed without the need for arthroplasty. Tissue, which is obtained after fracture, therefore is not representative of all osteoporotic fractures. Prospective clinical trials are difficult and expensive because of the long follow-up times and the large number of individuals required to provide statistically significant and meaningful data. A further complication arises with the variations across different fracture sites. These variations have been hypothesized to be caused by two types of osteoporosis(55) and also may be a function of further site-specific factors.
- (11)The etiologies of osteoporosis, whether biomechanical, molecular, cellular, environmental, or otherwise, are still relatively unknown.
What is the “take-home” message? What does this say about the best methods of screening for fracture risk and the future of biomechanics research in osteoporosis?
For robust fracture risk screening, it appears that we can do better than BMD alone with current knowledge and technology. BMD by itself can predict fractures with a detection rate of 30-50% and a false positive rate of about 15%, using a cut-off of 1 SD below the mean BMD and actual hip fracture and BMD data.(4,56) Using a 0.5 SD difference in BMD between individuals who will and will not develop fracture, Law et al.(56) show a 6% detection rate and 2% false positive rate with a cut-off of 2 SD below the mean. Looking at the numbers from a different perspective, 85-95% of white women who fracture have BMD measurements less than 2.5 SD below normal.(57) By combining BMD and some “geometric” factors, either by looking at the size and shape of a bone or further describing the BMD by location within the bone, we can likely identify more accurately those who have or will sustain a fracture. For example, Sandor et al.(5) claim an accuracy of 90% by including the distribution of BMD. It is inappropriate to expect to achieve perfect identification of individuals who will fracture, because there is a varying amount of risk that is attributable to the chance of sufficient loading to fracture the bone (i.e., a significant fall). Although technology appears to be available to improve on BMD alone, clinical trials have not been completed that will enable its acceptance and general use. There is a need for prospective clinical trials to verify previous results in large populations and determine what geometry/location information provides the best fracture prediction. However, once it is indicated, it should require little work to incorporate into current screening protocols.
There are several characteristics of bone that have a potential to further improve our ability to identify those at risk of fracture. Tissue material properties may provide additional information about fracture risk, but it is unlikely that it will become routine in screening for osteoporosis because of its invasive nature. Genetic analysis may, in the future, provide additional methods of screening, and with currently identified candidate genes(53,58) and the recent announcement that the human genome has been mapped, progress may begin to accelerate. In addition, noninvasive means of measuring trabecular bone architecture may soon provide an additional level of insight for screening, although the resolution is currently insufficient. There are likely to be additional attributes that are discovered from ongoing basic science research that correlate with fracture risk. However, it is unclear whether these methods provide improved risk estimates beyond that from BMD and geometry/location.
Biomechanics teaches us that the risk of fracture is dependent on the geometry of the bone, its architecture (at all scales of hierarchy), its material properties and the distribution of material properties, and the character of the imposed load (magnitude, rate, and direction). Genetics can influence the geometry, material properties, architecture, or the propensity for an individual to alter these features in response to the environment or underlying physiological conditions. The environment provides the cues that may trigger an adaptation response over time or the acute conditions that are sufficient to cause a fracture. Noninvasive imaging techniques can provide measures of geometry and a correlate to macroscopic material properties (BMD). Until we have effective methods for measuring microarchitecture and genetic or other biomarkers for individual response dynamics, we should strive to use both geometry and BMD to predict susceptibility to fracture in patients.