Bone Density and the Risk of Fractures: Title and subTitle BreakShould Treatment Thresholds Vary by Race?

Age, female sex, slender body habitus, and white race are well-known risk factors for osteoporotic fractures.¹ Middle-aged and older black men and women have higher bone mass and substantially lower fracture rates than whites.² Partly because of their reduced risk, blacks have only recently been included in prospective studies of osteoporosis with measurements of bone mineral density (BMD) and fracture incidence.^{2 - 3}

Bone mineral density is highly predictive of fracture risk in white women.⁴ A criterion for diagnosing osteoporosis is a BMD of 2.5 or more SDs below the mean for healthy young adults (T score ≤−2.5).⁵ These norms are sex-specific to account for the substantially higher BMD and lower fracture risk in men compared with women.⁶ In 2001 a panel of experts, convened by the International Society for Clinical Densitometry, concluded that too few data were available to recommend using separate, race-specific norms of BMD to define osteoporosis for nonwhites.⁶ They found a paucity of data and conflicting associations of BMD with fracture risk in nonwhites. Furthermore, they pointed out that racial and ethnic groups are difficult to define.

Since then, more data have been collected from nonwhite research participants. In this issue of JAMA, Cauley et al³ report on 636 black and 7334 white women followed up for a mean (SD) of 6.1 (1.5) years in the Study of Osteoporotic Fractures (SOF). They found that low femur BMD measured by dual-energy x-ray absorptiometry was associated with an increased risk of nonspinal fractures for both black and white women. The age-adjusted relative risk of fracture increased proportionally for blacks and whites with each SD decrease in BMD at the hip. Adjusting also for body weight attenuated but did not eliminate the association of decreasing BMD with fracture rate in blacks. However, at each level of BMD, fracture rates for black women were 30% to 40% lower than for white women. When adjusted for BMD and other risk factors that differed between the 2 groups, black women in the SOF had less than half the risk of fracture compared with white women.

Another observational study, the National Osteoporosis Risk Assessment study, recently reported 1-year fracture incidence for a multiethnic sample of 197 848 postmenopausal female primary care patients (7784 black) after baseline heel, forearm, or finger BMD measurements.² Within each ethnic group after adjustment for other risk factors, peripheral BMD had a strong, inverse relationship to fracture incidence, but the absolute fracture risk differed among ethnic groups. As in the SOF, compared with white women, black women had approximately half the prevalence of low BMD in the osteoporotic range and half the incidence of fracture. Surprisingly, Asian American women had BMD similar to white women, but only 32% the adjusted relative risk of fracture.²

It is important that data on BMD and fracture risk are now more inclusive, but not surprising that low BMD predicts fractures in all ethnic groups studied. The findings^{2 - 3} that older black and Asian study participants, compared with whites, had lower BMD-specific fracture risk, even after adjustment for many known risk factors, provides a stimulus for further research that may explain these disparities.

On the basis of the new data, Cauley et al³ advocate race-specific norms of BMD for defining osteoporosis. What would be the consequences of such a change? For any population group with higher peak BMD in young adults, using a population-specific reference group will result in a larger proportion diagnosed as having osteoporosis than if norms from a group with lower peak BMD are used.⁷ This is the case for blacks. For example, calculating a T score comparing measured BMD for black participants in the SOF with BMD norms for healthy young black women changed the proportion with T scores in the osteoporotic range from 9.8% (based on norms for white women) to 13.5%³ (vs 24% of white participants who had osteoporosis, based on norms for US white women). If low BMD were the only criterion for instituting osteoporosis treatment, then applying race-specific norms to SOF participants would result in almost 40% more black women being treated for osteoporosis, although these would still represent less than one seventh of the black women in the study compared with almost one fourth of the white women.

Because their fracture risk was lower, black women’s low BMD was not as sensitive or specific for predicting fractures. As a result of using race-specific BMD norms, the sensitivity of a T score below −2.5 as a predictor of fracture for black SOF participants increased from 19% to 24% while the specificity remained low at 16%. In contrast, a T score below −2.5 was 36% sensitive and 32% specific for predicting fractures in white SOF participants (calculated from Table 2).³ Current guidelines recommend screening BMD with dual-energy x-ray absorptiometry in women of all races who are older than 65 years to identify patients for treatment with bone-stabilizing drugs to prevent fractures.⁸

An important issue is whether evidence about ethnic differences, such as that from the SOF and the National Osteoporosis Risk Assessment study, indicates a different threshold for treating patients with the same BMD based on differences in skin color, facial features, or self-identified racial groups. This approach is fallible to the extent that race is a nonbiological category, an extremely crude surrogate for biological, environmental, cultural, and behavioral differences among individuals and human populations.^{9 - 10}

Genetic traits vary in gradients across populations, or may cluster in populations that have been isolated.^{11 - 12} Diet and many other behaviors and exposures that affect health also vary among ethnic and cultural groups, changing with migration and contact with other cultures. Socially constructed racial categories partition environments and resources, resulting in health disparities.⁹ The words “white” and “black” mislead by lumping groups of individuals with heterogeneous ancestry, histories, and environmental exposures.⁹ For example, Somali women living in the United States have lower adjusted BMD than African American women.¹³ White women in France have lower peak bone mass than US white women.⁷ Scientists and clinicians will do best by avoiding race proxies for other biological, social, and cultural constructs.¹¹

Given the imprecision of defining race, why should biomedical and clinical science explore racial and ethnic disparities? First, to eliminate injustice.¹⁴ Second, understanding complex mechanisms of observed differences in health among racial or ethnic groups can be applied to individualize medical care.¹⁰ A polygenic, multifactorial problem such as osteoporosis results from genetic, environmental, behavioral, and societal processes.⁹ Ascertaining differences in the natural history of fractures among population groups is a first step toward being able to define clinically important measures of individual differences in fracture susceptibility.¹⁵ To fully personalize treatment, it will also be important to understand variations in treatment responsiveness,¹⁶ adverse effects of treatment, and patient values or utilities for various outcomes.^{15 ,17}

Ethnic differences in fracture risk might also provide insights into determinants of bone health. These include peak BMD and other determinants of bone size and strength, factors that cause or prevent bone loss, and variables related to the risk of falls—all operate over a lifetime. Several clinically important mechanisms could explain observed differences among ethnic groups in the risk of fracture.

Body Mass Index. Part of the lower risk of fracture among the black women in the SOF was accounted for by their 10-kg greater body weight.^{1 ,3} Several known mechanisms may help to explain this finding. Increased skeletal loading and muscle mass increase bone size and trabecular density.¹⁸ Fat protects postmenopausal women against bone loss through higher estrogen exposure.¹⁹ Obesity and other comorbidities can be associated with limited physical activity that might diminish the probability of fractures, independent of BMD. Furthermore, wearing padding over the greater trochanter has been shown to reduce the risk of femur fracture with falls²⁰ ; adiposity might have a similar effect.

Bone Architecture. Bone size, microarchitecture, and geometry affect fracture risk and may differ among population groups.³ Measuring bone mineral apparent density to approximate 3-dimensional density corrects more completely for bone size than the 2-dimensional BMD.²¹ The SOF found bone mineral apparent density to be a slightly better predictor of age-adjusted fracture risk than BMD for blacks but not for whites.³ Some investigators have measured differences in femur geometry that would indicate decreased hip fracture risk in black compared with white women.^{18 ,21} Future refinements of bone imaging may assess bone strength more precisely.

Genetic Variation. Peak BMD is highly heritable as a polygenic trait.^{22 - 23} Genetic variation in a variety of metabolic pathways contributes to bone loss and the origin of osteoporosis.^{16 ,22 - 23} Ethnic differences in candidate gene polymorphisms related to bone metabolism are beginning to be explored.²⁴ Before predictive genotyping becomes clinically feasible,¹⁶ biochemical indices of bone turnover might become useful for identifying when osteoporosis treatment is needed.⁵

Environment and Behavior. Lifelong environmental and behavioral factors, subject to cultural and familial variation, interact with physiology to influence fragility and fracture risk. For example, perhaps due to a higher prevalence of lactose intolerance, black women in the SOF had lower dietary calcium intake than white women, yet they were less likely than whites to use calcium supplements.³ In northern climates, dark-skinned individuals require more vitamin D and are more prone to osteomalacia. Hormonal exposures influencing bone turnover are likely to vary by race, including earlier menarche in blacks compared with whites, differences in reproductive history, lactation, hormonal contraception, menopause, and perhaps body burden of pollutants with hormonelike effects.^{19 ,25} Physical activity in adolescents and adults and physical capacity in older age vary by race and social class and may influence body mass index, bone and muscle mass, and fracture risk.^{26 - 27}

Prevention and Treatment. Prevention of osteoporotic fractures involves more than risk assessment by BMD screening or otherwise. Studies have shown that the use of medications to treat osteoporosis is infrequent relative to the prevalence of the condition,²⁸ even after fractures have occurred,^{29 - 30} and that blacks are less likely to receive treatment than whites.³¹ Some women become vigilant and seek treatment after a fracture at midlife, while others minimize its significance.³² Comorbidities, prevalent with aging³³ and more common among black compared with white participants in the SOF,³ often interfere with adherence to treatment.³⁴ Clinical attention and financial resources for screening and treatment must be prioritized among multiple diseases.^{34 - 35} These issues are compounded by disparities in access to medical care, safe environments, and ability to afford prescription drugs.^{14 ,35}

More accurate clinical assessment to determine the best options for patients is the ultimate goal.¹⁵ Data showing higher fracture risk with low BMD in all ethnic groups is an important step. If, besides BMD, bone geometry, body composition, bone metabolism, physical capacity, fall risk, and eventually genotype¹⁶ are race-related variables determining fracture risk,³ measurements related to these factors could be evaluated clinically. Research will be needed to test their value. This step will be more appropriate than using race as a variable to determine treatment threshold.