Association between African American race/ethnicity and low bone mineral density in women with systemic lupus erythematosus

Authors


Abstract

Objective

To determine the association between race/ethnicity and bone mineral density (BMD) in women with systemic lupus erythematosus (SLE).

Methods

Women with SLE (n = 298), including 77 African Americans and 221 whites, completed this cross-sectional study conducted from 1996 to 2002. Hip and lumbar spine BMD were measured by dual-energy x-ray absorptiometry. Study participants completed a self-administered questionnaire and a physician completed the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (SDI). BMD results were expressed as Z scores. Analyses were performed to identify factors, including race/ethnicity, associated with low BMD defined as a Z score −1.0 or less at the hip or lumbar spine.

Results

African Americans compared with whites were younger at study visit (mean ± SD 39.7 ± 8.4 years versus 42.9 ± 11.6 years) and had higher SDI (mean ± SD 1.8 ± 2.0 versus 1.0 ± 1.6), but similar proportions of women were postmenopausal (31.2% versus 38.0%). African Americans had significantly lower mean BMD Z scores at the hip (−0.49 versus −0.07; group difference −0.41; 95% confidence interval [95% CI] −0.70, −0.13) and at the lumbar spine (−1.03 versus 0.10; group difference −1.13; 95% CI −1.48, −0.78) compared with whites. African American race/ethnicity was strongly associated with low BMD at the lumbar spine (adjusted odds ratio 4.42; 95% CI 2.19, 8.91) but not at the hip, adjusting for factors associated with low BMD.

Conclusion

African American women compared with white women with SLE had lower BMD at the hip and lumbar spine. African American race/ethnicity was associated with low BMD at the lumbar spine controlling for relevant clinical covariates.

INTRODUCTION

Women with systemic lupus erythematosus (SLE) are at increased risk for osteoporosis (1) and fractures (2). Moreover, the frequency of osteopenia is estimated to be as high as 25% among premenopausal women with SLE (3), with bone loss occurring at regions of cortical (4, 5) and trabecular bone (6). The basis for reduced bone mineral density (BMD) and higher fracture rates in some groups of women with SLE may be related to a combination of traditional osteoporosis risk factors (e.g., white race/ethnicity) and bone loss due to factors related to SLE (e.g., corticosteroid use). However, the association between race/ethnicity and BMD status in SLE has undergone minimal study.

In the general population, white race/ethnicity is an established risk factor for osteoporosis, whereas African American race/ethnicity is associated with comparatively higher peak bone mass (7, 8) and lower risk for fractures (9). Most published reports examining bone health in persons with SLE predominantly consist of white individuals, and there is evidence that white (10) and Asian race/ethnicity (11, 12) are associated with low BMD in SLE. In contrast, there is a paucity of information on the relationship between African American race/ethnicity and BMD status in patients with SLE. Unlike the general population, African American race/ethnicity may not confer a protective benefit against the development of osteopenia or osteoporosis in patients with SLE given the potential contribution of SLE-related factors for bone loss, particularly in the presence of any concurrent traditional osteoporosis risk factors.

For this study, we hypothesized that African American race/ethnicity is associated with similarly low BMD at the hip and lumbar spine as compared with white race/ethnicity in women with SLE. In addition, we examined the relationship of traditional osteoporosis risk factors and SLE-related factors, including race/ethnicity, as it relates to BMD in women with SLE.

PATIENTS AND METHODS

Study design, patient enrollment, and data collection.

A total of 298 women with SLE were recruited from the Chicago Lupus Database (n = 219) and the Pittsburgh Lupus Registry (n = 79) from 1996 to 2002 for this cross-sectional study. Designation of race/ethnicity was self declared, and participants consisted of non-Hispanic white and African American women. A total of 96% of the women met 4 components of the American College of Rheumatology (ACR) classification criteria for SLE (13, 14). Of the remaining 4% of women meeting <4 ACR classification criteria for SLE, all were white, had a clinical diagnosis of SLE, and met 3 components of the ACR classification criteria for SLE. Institutional review boards at the Chicago and Pittsburgh study centers approved the study protocol, and all participants provided informed written consent prior to participation. Patients attended a single clinic visit that included an interview, physical examination, and BMD measurements of the hip and lumbar spine.

Traditional osteoporosis risk factors and SLE-related factors for low BMD.

Study participants provided information on traditional factors related to osteoporosis, including age (at SLE diagnosis and at study visit); menopausal status; lifestyle factors (current smoking status, use of caffeine, use of alcohol); use of hormone preparations; calcium and vitamin D intake; and use of medications, including use of osteoporosis medication(s) (etidronate, alendronate, calcitonin), thyroid hormones, diuretics, and antiepileptics. Information on potential SLE-related factors for low BMD, including duration of SLE, cumulative disease damage as measured by the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Cumulative Disease Damage Index (SDI), and SLE renal disease (proteinuria >0.5 gm or 3+ protein, cellular casts, or positive renal biopsy result), was also collected. We scored disease damage using the SDI, a validated measure of damage occurring in SLE disease course regardless of cause, modified by excluding the osteoporosis/fracture item (1 point) from a possible maximum score of 46 (15). The history and duration of lupus renal involvement were defined using the 1982 ACR classification criteria for SLE (13). We also collected information on hydroxychloroquine and corticosteroid use, including past or current use of corticosteroids (including mean daily dose) and duration of use (including duration ≥1 year). For standardization, all corticosteroid preparations were converted to prednisone equivalents.

Bone mineral density measurements.

BMD measurements of the hip (n = 290) and lumbar spine (n = 295) were obtained by dual-energy x-ray absorptiometry (DXA) using a Hologic QDR-4500 (Chicago study center; Hologic, Bedford, MA) or Hologic QDR-2000 (Pittsburgh study center; Hologic). Total hip BMD measurement was performed, and the anteroposterior lumbar spine was measured from L1 to L4 to calculate the mean lumbar spine BMD. All BMD values (gm/cm2) for the hip and lumbar spine were expressed as BMD Z scores using the DXA manufacturer's age and race/ethnicity-specific female reference database. For this study, low BMD was defined as a BMD Z score of −1.0 or less.

Quality control.

To assess intra- and interinstrument variations of the 2 densitometers, we measured femur and lumbar spine phantoms on machines at both the Chicago and Pittsburgh study centers. Block phantoms of standardized densities were measured daily to test variation in system linearity. The intrainstrument coefficients of variation were <0.56% for the lumbar spine, 0.46% for the femur, and 0.44% for the block phantoms. The lumbar spine phantom measurement was 4% higher with the Hologic QDR-2000 as compared with the Hologic QDR-4500 model. Data analyses used uncorrected lumbar spine measurements, but sensitivity analyses using a 4% correction factor applied to the lumbar spine measurement obtained in women from the Pittsburgh Lupus Registry produced similar results.

Statistical analysis.

All statistical testing was performed using a 2-sided nominal alpha level of 0.05. Differences between African Americans versus whites and the associated 95% confidence intervals (95% CIs) were calculated for traditional osteoporosis risk factors and SLE-related factors for low BMD; 95% CIs excluding zero indicated significant differences. Univariate logistic regression was used to screen traditional osteoporosis risk factors and SLE-related factors of low BMD to identify the most relevant (P < 0.05) factors associated with low BMD at each anatomic site for inclusion in a multiple logistic regression model. Factors selected by univariate screening were evaluated for multicollinearity. We determined the strength of correlation between 11 factors that were significantly associated with low BMD in the univariate screening. To avoid multicollinearity, we parsimoniously selected a single factor most strongly associated with low BMD among those factors correlating (r ≥0.50) with ≥1 other factor(s) for inclusion in our final regression model, which also included study center (Chicago/Pittsburgh recruitment study centers), a design variable. Results of the multiple logistic regression model were presented as adjusted odds ratios (ORs) with corresponding 95% CIs; confidence interval excluding 1 indicated a statistically significant OR. Statistical analyses were completed using SAS software, version 9 (SAS Institute, Cary, NC).

RESULTS

Traditional risk factors for osteoporosis.

Of the 298 women with SLE included in this study, 77 were African American. Compared with white women, African Americans were significantly younger at both SLE diagnosis and at study visit, and African Americans had higher mean body mass index (BMI). However, both groups were similar in the proportion of women still menstruating (Table 1).

Table 1. Traditional osteoporosis risk factors stratified by race in women with systemic lupus erythematosus (SLE)*
FactorsAfrican American (n = 77)White (n = 221)Difference (95% CI)
  • *

    Values are the percentage unless otherwise indicated. 95% CI = 95% confidence interval.

  • Indicates the difference in proportions between African Americans and whites and the corresponding 95% CI unless otherwise stated. Negative results for differences indicate lower values for African Americans.

  • 95% CI excluding zero indicates a significant difference.

  • §

    Data available on 72 individuals.

  • Data available on 214 individuals.

  • #

    Data available on 219 individuals.

  • **

    Data available on 76 individuals.

  • ††

    Data available on 220 individuals.

Patient characteristics   
 Age at SLE diagnosis, mean ± SD (range) years29.4 ± 10.6 (6–52)34.1 ± 12.0 (6–68)−4.7 (−1.7, −7.7)
 Age at study visit, mean ± SD (range) years39.7 ± 8.4 (23–57)42.9 ± 11.6 (20–73)−3.2 (−0.4, −6.0)
 Body mass index, mean ± SD kg/m228.7 ± 7.0§26.2 ± 6.62.5 (4.3, 0.7)
 Still menstruating68.862.06.8 (−5.5, 19.0)
 Study center—Chicago site89.667.921.7 (12.6, 30.9)
Lifestyle factors   
 Currently smoking16.914.92.0 (−7.6, 11.6)
 Drink caffeine76.684.9#−8.3 (−18.9, 2.3)
 Drink alcohol33.858.2#−24.6 (−36.8, −12.0)
Medication and dietary supplement use   
 Taken female hormones10.5**23.1−12.6 (−21.4, −3.7)
 Taken calcium32.554.8−22.3 (−34.6, −9.9)
 Daily calcium intake (mean ± SD mg/day)957.3 ± 710.91,274.8 ± 697.9−317.5 (−500.1, −134.8)
 Taken vitamin D28.646.6−18.0 (−30.1, −6.0)
 Daily vitamin D intake (mean ± SD IU/day)135.1 ± 248.0208.1 ± 265.8−73.0 (−141.2, −5.0)
 Taken thyroid hormones5.222.7††−17.5 (−25.0, −10.1)
 Taken osteoporosis medications9.110.0−0.9 (−8.4, 6.7)
 Taken diuretics31.226.4††4.8 (−7.1, 16.7)
 Taken thiazide diuretics13.09.6#3.4 (−5.0, 11.9)
 Taken antiepileptics6.55.90.6 (−5.7, 6.9)

In terms of lifestyle factors, a similar proportion of women from both racial/ethnic groups currently smoked and consumed caffeine, but a lower proportion of African Americans used alcohol. However, among those who consumed caffeine or alcohol, African Americans consumed significantly less caffeine (69.9 mg/day versus 110.4 mg/day) and alcohol (0.8 gm/day versus 2.7 gm/day) compared with white women.

For medication and dietary supplement use, significantly lower proportions of African Americans had taken female hormones, calcium, or vitamin D as seen in Table 1. In addition, among participants reporting use of dietary supplements, the mean daily amounts of either calcium (957.3 mg/day versus 1,274.8 mg/day) or vitamin D (135.1 IU/day versus 208.1 IU/day) were less among African Americans than whites. Although a lower proportion of African Americans received thyroid hormones, we observed no significant group differences in their use of osteoporosis medications, diuretics (including thiazide diuretics), or antiepileptics.

SLE-related factors associated with low BMD.

Differences between African Americans and whites for SLE-related factors of low BMD and the corresponding 95% CIs are presented in Table 2. Although both groups of women were comparable in terms of their disease duration and proportion of women with SLE renal disease, African Americans had significantly greater cumulative disease damage as reflected by the mean SDI score. However, the overall SDI levels were relatively low for both groups, with African Americans having a median SDI of 1.0 (range 0–8), whereas whites had a median SDI of 0 (range 0–8).

Table 2. Systemic lupus erythematosus (SLE)-related factors associated with low bone mineral density stratified by race in women with lupus*
FactorsAfrican American (n = 77)White (n = 221)Difference (95% CI)
  • *

    Values are the mean ± SD unless otherwise indicated. 95% CI = 95% confidence interval; SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Disease Damage Index.

  • Indicates the difference in proportions between African Americans and whites and the corresponding 95% CI unless otherwise stated. Negative results for differences indicate lower values for African Americans.

  • SDI was modified by excluding the osteoporosis/fracture item. Data available on 75 African Americans and 216 whites.

  • §

    95% CI excluding zero indicates a significant difference.

  • Defined as proteinuria >0.5 gm or 3+ protein, cellular casts, or positive renal biopsy.

  • #

    Data available on 75 African Americans and 219 whites.

  • **

    Data available on 62 African Americans and 166 whites.

  • ††

    Data available on 71 African Americans and 172 whites.

SLE-related characteristics   
 Disease duration, years9.8 ± 8.48.3 ± 7.81.5 (−0.6, 3.5)
 SDI1.8 ± 2.01.0 ± 1.60.8 (0.3, 1.2)§
 SLE renal disease, %27.319.08.3 (−2.9, 19.5)
SLE-related medications   
 Taken hydroxychloroquine in the past, %#55.355.9−0.6 (−13.6, 12.3)
 Taken corticosteroids in the past, %81.877.44.4 (−5.8, 14.7)
 Duration of taking corticosteroids, years**8.7 ± 8.66.0 ± 7.02.7 (0.5, 4.8)§
 Current use of corticosteroids, %64.946.218.7 (6.3, 31.3)§
 Current dose of corticosteroids, mg/day††8.5 ± 10.16.2 ± 9.32.3 (−0.4, 4.9)

With regard to SLE-related medications, African Americans were similar to whites in the proportions having taken hydroxychloroquine and having any use of corticosteroids in the past as shown in Table 2. African Americans had longer use of corticosteroids and a greater proportion were currently taking corticosteroids, yet both groups were not significantly different in their mean current dose of corticosteroids. Moreover, African Americans were similar to whites in the frequency of those having taken corticosteroids for ≥1 year (67.5% versus 60.2%) as well as in their median current daily corticosteroid dose (6.0 mg/day versus 4.0 mg/day).

Bone mineral densitometry measurements.

Mean ± SD BMD values for both the hip (whites: 0.93 ± 0.15 gm/cm2, African Americans: 0.89 ± 0.14 gm/cm2) and lumbar spine (whites: 0.98 ± 0.13 gm/cm2, African Americans: 0.99 ± 0.16 gm/cm2) were expressed as Z scores because the majority of the women in both groups were premenopausal. We observed lower BMD Z scores for African Americans at the hip (−0.49 versus −0.07; group difference = −0.41; 95% CI −0.70, −0.13) and at the lumbar spine (−1.03 versus 0.10; group difference = −1.13; 95% CI −1.48, −0.78) compared with whites (Figure 1). The findings for BMD T scores at both anatomic sites, including African American versus white racial/ethnic group differences, were similar to those observed for BMD Z scores (data not shown).

Figure 1.

Mean bone mineral density (BMD) Z scores for the hip and lumbar spine in white and African American women with systemic lupus erythematosus. Error bars represent 95% confidence intervals for mean BMD Z scores. Open bars = whites; solid bars = African Americans. Available BMD Z score data for whites: hip (n = 219) and lumbar spine (n = 221). Available BMD Z score data for African Americans: hip (n = 74) and lumbar spine (n = 77). *P = 0.004. **P < 0.0001.

Univariate risk factors for low bone mineral density.

We performed univariate screening of traditional osteoporosis risk factors and SLE-related factors to determine the strength of association between individual factors and low BMD at the hip and lumbar spine (Table 3). Of the 27 factors screened, 11 factors were significantly associated with low BMD at ≥1 anatomic site(s). Among the significant factors, African American race/ethnicity was most strongly associated with low BMD at the lumbar spine (unadjusted OR 4.11; 95% CI 2.37, 7.15). Although African American race/ethnicity was not significantly associated with low BMD at the hip, the results approached statistical significance (OR 1.82; 95% CI 0.96, 3.43). Older age at study visit was marginally associated with low BMD at the hip. No significant relationship between still menstruating status and low BMD was observed at either anatomic site. As might be expected for weight-bearing sites, BMI was inversely associated with low BMD at both the hip and lumbar spine.

Table 3. Univariate factors associated with low bone mineral density (BMD) at the hip and lumbar spine in 298 women with systemic lupus erythematosus (SLE)*
FactorsLow hip BMDLow lumbar spine BMD
  • *

    Values are the odds ratio (95% confidence interval). Low BMD defined as a BMD Z score ≤−1.0. SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Disease Damage Index.

  • 95% confidence interval excluding 1 indicates a significant odds ratio.

  • Modified by excluding the osteoporosis/fracture item.

Patient characteristics  
 African American race1.82 (0.96, 3.43)4.11 (2.37, 7.15)
 Age at SLE diagnosis0.96 (0.93, 0.98)0.94 (0.92, 0.97)
 Age at study visit0.99 (0.96, 1.02)0.96 (0.94, 0.98)
 Still menstruating0.95 (0.52, 1.74)0.97 (0.58, 1.61)
 Body mass index (kg/m2)0.90 (0.85, 0.96)0.95 (0.91, 0.99)
 Study center0.96 (0.49, 1.88)0.61 (0.33, 1.12)
Lifestyle factors  
 Currently smoking0.49 (0.19, 1.31)0.47 (0.21, 1.05)
 Drink caffeine1.05 (0.47, 2.31)0.38 (0.21, 0.71)
 Drink alcohol1.55 (0.85, 2.82)0.74 (0.45, 1.23)
SLE-related characteristics  
 Disease duration in years1.06 (1.03, 1.10)1.03 (1.00, 1.06)
 SDI1.29 (1.11, 1.51)1.19 (1.04, 1.37)
 SLE renal disease1.93 (1.00, 3.72)2.24 (1.25, 4.00)
Medication and dietary supplement use  
 Taken female hormones1.16 (0.57, 2.38)0.75 (0.40, 1.43)
 Taken calcium1.16 (0.64, 2.10)1.06 (0.64, 1.75)
 Daily calcium intake1.00 (1.00, 1.01)1.00 (0.99, 1.00)
 Taken vitamin D1.25 (0.69, 2.27)1.13 (0.68, 1.87)
 Daily vitamin D intake0.13 (<0.001, >999)1.48 (0.03, 71.56)
 Taken osteoporosis medications2.08 (0.89, 4.83)1.75 (0.80, 3.81)
 Taken thyroid hormones0.91 (0.41, 2.01)0.50 (0.24, 1.05)
 Taken diuretics1.23 (0.65, 2.34)1.27 (0.73, 2.19)
 Taken thiazide diuretics0.60 (0.20, 1.79)0.95 (0.42, 2.15)
 Taken antiepileptics0.54 (0.12, 2.40)0.95 (0.33, 2.74)
SLE-related medications  
 Taken hydroxychloroquine in the past0.87 (0.48, 1.58)1.12 (0.67, 1.85)
 Any use of corticosteroids in the past2.10 (0.90, 4.91)3.00 (1.41, 6.38)
 Duration of taking corticosteroids1.05 (1.01, 1.09)1.03 (1.00, 1.07)
 Current use of corticosteroids2.43 (1.30, 4.55)2.65 (1.57, 4.49)
 Current dose of corticosteroids1.03 (1.00, 1.05)1.02 (1.00, 1.05)

Of the lifestyle factors, there was an inverse relationship between caffeine consumption and low BMD at the lumbar spine as seen in Table 3. In terms of SLE-related characteristics, higher disease duration, elevated SDI score, and presence of SLE renal disease were significantly associated with low BMD at 1 or both anatomic sites. We did not find any significant relationships between those factors examined in the medication and dietary supplement use category and low BMD at either the hip or lumbar spine. As for SLE-related medications, any use of corticosteroids in the past, duration of corticosteroid treatment, and current use of corticosteroids were associated with low BMD at either the hip and/or lumbar spine as shown in Table 3.

Adjusted risk factors for low bone mineral density.

The relative associations between factors related to low BMD are shown in Table 4 according to anatomic site based on the results of our multiple logistic regression model, which was constructed after screening for collinear relationships. Univariate risk factors excluded from the multiple logistic model due to their strong correlation with those factors included in the model were age at study visit (correlated with age at SLE diagnosis), disease duration and duration of taking corticosteroids (correlated with SDI), and any use of corticosteroids in the past (correlated with current use of corticosteroids).

Table 4. Adjusted results for factors associated with low bone mineral density (BMD) at the hip and lumbar spine in 298 women with systemic lupus erythematosus (SLE)*
FactorsLow hip BMD adjusted OR (95% CI)Low lumbar spine BMD adjusted OR (95% CI)
  • *

    Values are the adjusted odds ratio (95% confidence interval). Low BMD defined as a BMD Z score ≤−1.0. SDI = Systemic Lupus International Collaborating Clinics/American College of Rheumatology Disease Damage Index.

  • 95% confidence interval excluding 1 indicates a significant odds ratio.

  • Modified by excluding the osteoporosis/fracture item.

African American race1.54 (0.69, 3.46)4.42 (2.19, 8.91)
Age at SLE diagnosis0.98 (0.95, 1.01)0.96 (0.93, 0.99)
Body mass index (kg/m2)0.90 (0.84, 0.96)0.93 (0.89, 0.98)
Drink caffeine1.33 (0.53, 3.36)0.34 (0.17, 0.72)
SDI1.30 (1.08, 1.57)1.06 (0.89, 1.26)
SLE renal disease0.79 (0.33, 1.87)1.50 (0.71, 3.16)
Current use of corticosteroids1.48 (0.71, 3.09)1.54 (0.81, 2.92)
Study center0.95 (0.42, 2.15)0.71 (0.33, 1.53)

After adjusting for significant factors of low BMD, African American race remained strongly associated with low BMD at the lumbar spine (adjusted OR 4.42; 95% CI 2.19, 8.91), but not at the hip. In addition, age at SLE diagnosis and BMI were both inversely associated with presence of low BMD at the hip and lumbar spine, whereas caffeine use was inversely associated with low BMD only at the lumbar spine. Moreover, higher SDI score was associated with low BMD at the hip, but not at the lumbar spine. The remaining factors in the regression model did not demonstrate any notable association with low BMD at the hip or lumbar spine.

DISCUSSION

In the present study, mean BMD Z score at the lumbar spine was significantly lower in African American women with SLE, and the odds of having low BMD at the lumbar spine were more than 4 times higher for African Americans than whites after controlling for traditional osteoporosis factors and SLE-related factors of low BMD, including corticosteroid use. The finding of low BMD at the lumbar spine after adjusting for corticosteroids is noteworthy given that African Americans were younger at study visit and had a higher mean BMI, whereas the proportions of postmenopausal women were similar between the racial groups. At the hip, the mean BMD Z score was also significantly lower in African Americans than whites. The association between African American race/ethnicity and low BMD at the hip approached statistical significance in the univariate analysis, but we found no significant relationship between African American race/ethnicity and low BMD at the hip after adjusting for pertinent clinical factors. Prior to expressing BMD values as standardized Z scores, African Americans and whites showed comparable BMD values at the hip, a finding that suggests that bone mass of African American women may be at least similarly reduced as that of white women in the setting of SLE.

In African American women, the relatively lower BMD seen at the lumbar spine as compared with the hip may to some extent be related to variations in the type of bone comprising these 2 anatomic sites. The lumbar spine, which is composed predominantly of trabecular-rich bone, is more susceptible to the adverse effects of corticosteroids than those anatomic regions consisting of comparatively less trabecular bone, such as the hip (16). Reduced BMD at sites of trabecular bone has been linked to corticosteroid use in women with SLE, although not specifically in African Americans (12, 17, 18). In addition, among premenopausal women with SLE, reduced BMD at sites of both trabecular and cortical bone has been described, and the finding of significantly reduced lumbar spine BMD has been attributed to prior corticosteroid use (6). Conversely, there is evidence for trabecular bone loss in patients with SLE not explained by corticosteroid exposure (3). Moreover, trabecular bone loss involving the hip and lumbar spine has been observed in healthy premenopausal African American and white women (19). Therefore, the pattern of relatively lower BMD at the spine in African American women with SLE as seen in our study may be attributable in part to factors other than corticosteroid use alone.

Although the duration of corticosteroid use was longer for African Americans, similar proportions of women in both racial groups reported past use of corticosteroids and were currently taking corticosteroids with comparable current daily doses. Although we cannot completely discount the potential detrimental effects of longer duration of corticosteroid use on sites of trabecular bone, such as the lumbar spine, the results of our regression analysis indicate that corticosteroid exposure alone does not explain the association found between African American race/ethnicity and low BMD at the lumbar spine.

Our results are in relative contrast to those reported by Yee and colleagues who reported only non-African Caribbean race/ethnicity, which included whites, Asians (Indian subcontinent), and others, as opposed to African Caribbean race (black) to be associated with reduced BMD in patients with SLE undergoing BMD measurements of the femoral neck and lumbar spine (20). Although their study of 242 patients was not exclusively limited to women, ∼95% of study participants were women, and as a group, the mean age of patients, severity of disease as measured by the SDI, and corticosteroid burden appeared to be similar to women in our study. Their conclusion was based on 40 African Caribbeans with SLE, which comprised less than one-fifth of the entire study group. Also, the authors assessed low BMD, defined as osteopenia or osteoporosis, using BMD T scores without specifying according to anatomic site. By applying sex, race/ethnicity, and age-matched control reference data to determine Z scores in a group that includes premenopausal women, our results provide a more clinically relevant approach in comparing BMD values from individuals differing in age, race/ethnicity, and menopausal status (21). Moreover, reporting BMD findings specified by anatomic site offers further insight into bony regions that may be more susceptible to bone loss in the context of SLE.

In addition to African American race/ethnicity, several SLE-related factors were also associated with low BMD. For instance, higher SDI score and having SLE renal disease were associated with low BMD at the hip after adjusting for covariates in the multiple logistic regression model. One explanation might be that patients with more severe disease, as represented by the SDI, may experience more deleterious effects on bone, including bone loss. Interestingly, findings from prior investigations imply a potentially independent association between an elevated cumulative disease damage score and low BMD at the hip and lumbar spine in patients with SLE (22, 23), offering support for such an explanation. Alternatively, higher disease damage may reflect other risk factors for low BMD not assessed in the current study. For example, patients with more severe disease may be less physically active, and in turn may fail to reach their optimal peak bone mass or be at risk for bone loss. Further investigation will help clarify the importance of SLE per se as a unique risk factor for bone loss.

The absence of any significant associations between dietary supplementation, such as calcium and vitamin D intake, and low BMD was not entirely surprising. Given the potential for discordance between reported consumption of these dietary supplements and patients' serum levels, individuals may have insufficient levels of vitamin D and/or calcium necessary for maintaining optimal bone health, which may be reflected by reduced BMD status (24). With the exception of any diuretic use (e.g., taken diuretics), the proportions of women in either group receiving medications (e.g., thyroid hormones, osteoporosis medications, antiepileptics, thiazide diuretics) were relatively modest, which may have precluded identification of any underlying relationship between medication use and low BMD in our analysis.

The cross-sectional design of the current study did not allow us to ascertain if African American race/ethnicity is predictive of future bone loss in women with SLE. It is also conceivable that our findings of lower BMD, such as at the lumbar spine, may represent lower peak bone mass attained by women with SLE rather than bone loss per se. Despite the lower BMD Z scores for African Americans, the overall BMD levels for both racial/ethnic groups were not remarkably low, which likely reflects a relatively young group of women. Although BMD measurement remains an important method of assessing bone health, it provides only limited insight into potential underlying differences with respect to bone quality that may exist between African American and white women with SLE.

In summary, African American race/ethnicity is significantly associated with low BMD at the lumbar spine in women with SLE. Although African American race/ethnicity was not associated with low BMD at the hip, the finding of comparable mean hip BMD values and the significantly lower mean hip BMD Z score in African Americans than whites suggests that bone health of African American women may be at least similarly reduced as that of whites in the setting of SLE.

AUTHOR CONTRIBUTIONS

Dr. Lee had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Lee, Almagor, Dunlop, Ramsey-Goldman.

Acquisition of data. Chadha, Manzi, Spies, Ramsey-Goldman.

Analysis and interpretation of data. Lee, Almagor, Dunlop, Manzi, Ramsey-Goldman.

Manuscript preparation. Lee, Almagor, Dunlop, Manzi, Ramsey-Goldman.

Statistical analysis. Lee, Almagor, Dunlop, Ramsey-Goldman.

Acknowledgements

We thank Ms Li-Yung Lui and Dr. Jane Cauley for their assistance in the application of the bone mineral density reference database for this project.

Ancillary