Diabetes and Bone Loss at the Hip in Older Black and White Adults†
This paper was presented in part as an abstract at the 25th Annual Meeting of the American Society for Bone and Mineral Research in Minneapolis, Minnesota, USA, September 19-23, 2003.
Type 2 diabetes may be associated with elevated fracture risk, but the impact on bone loss is unknown. Analysis of 4-year change in hip BMD data from a cohort of white and black well-functioning men and women 70-79 years of age found that white women with diabetes had more rapid bone loss at the femoral neck than those with normal glucose metabolism.
Introduction: Type 2 diabetes may be associated with elevated fracture risk in older adults. Although type 2 diabetes is not associated with lower BMD, older diabetic adults have a higher prevalence of other risk factors for fracture, including more frequent falls, functional limitations, and diabetic complications. With this burden of risk factors, loss of BMD could place older adults with diabetes at higher risk of sustaining a fracture.
Materials and Methods: To determine if bone loss is increased with type 2 diabetes, we analyzed data from the Health, Aging, and Body Composition Study of white and black well-functioning men and women 70-79 years of age. Hip BMD was measured at baseline and 4 years later in 480 (23%) participants with diabetes, 439 with impaired glucose metabolism, and 1172 with normal glucose homeostasis (NG).
Results: Those with diabetes had higher baseline hip BMD and weight, but among white women, had more weight loss over 4 years. White women with diabetes lost more femoral neck and total hip BMD than those with NG in age-adjusted models. After multivariable adjustment, diabetes was associated with greater loss of femoral neck BMD (−0.32%/year; 95% CI: −0.61, −0.02) but not total hip BMD. In men and black women, change in hip BMD was similar for participants with diabetes and NG.
Conclusions: Despite having higher baseline BMD, diabetic white women, but not men or black women, had more rapid bone loss at the femoral neck than those with NG. This increased bone loss may contribute to the higher fracture risk observed in older diabetic women.
MOST, BUT NOT all,(1–3) cross-sectional studies have found average,(4–6) or somewhat elevated,(7–12) BMD in older adults with type 2 diabetes, even when accounting for larger body size. However, recent longitudinal studies have reported an increased risk of hip fracture associated with type 2 diabetes.(13–17) Diabetes also seems to increase fracture risk at other sites, including the proximal humerus, foot, and possibly ankle.(13,14,18) Although type 2 diabetes is not associated with lower BMD, older diabetic adults have a higher prevalence of other risk factors for fracture, including a tendency to fall, poor balance and gait, functional limitations, poorer vision, history of stroke, and peripheral neuropathy.(19–23) With this burden of risk factors, decrements in BMD could place older adults with diabetes at particularly high risk of sustaining a fracture.
Little information is available on longitudinal changes in bone mass in those with type 2 diabetes. In a study of 19 patients with type 2 diabetes, Krakauer et al.(2) found no loss of BMD at the radius over 12 years of follow-up. Among 4203 older white women in a prospective study of fractures, those with diabetes lost bone at the hip more rapidly than those without diabetes.(24) We analyzed results from 4 years of follow-up in the Health, Aging, and Body Composition Study (Health ABC Study) to determine if older black and white adults with type 2 diabetes or impaired glucose metabolism are losing bone at the hip more rapidly than normal controls and whether diabetes is an independent risk factor for bone loss.
MATERIALS AND METHODS
Health ABC participants
The Health ABC Study is a prospective cohort study investigating whether changes in body composition act as a common pathway by which multiple diseases affect morbidity, disability, and risk of mortality. The study has been previously described in a report on baseline BMD and diabetes.(12) Briefly, the cohort consists of 3075 men and women 70-79 years of age recruited at two centers: University of Pittsburgh and University of Tennessee, Memphis. Participants were recruited from a random sample of white Medicare beneficiaries and all age-eligible black community residents in Pittsburgh, PA, and Memphis, TN. Participants were excluded if they reported any difficulty with activities of daily living, walking up 10 steps without resting, or walking a quarter of a mile. The study procedures were approved by the institutional review boards, and written informed consent was provided by all participants.
The baseline examination took place during 1997-1998. Participants returned for annual visits. These analyses include BMD measurements at baseline and at the fourth follow-up visit (exam 5), ∼4 years later. At exam 5, 263 participants were deceased. We excluded participants with missing information on diabetes status at baseline (N = 88) or those diagnosed with diabetes after baseline (N = 63). Of the surviving participants, 21% (N = 570) did not have a hip BMD scan at both the baseline and fourth follow-up visit. Of those who did not have scans, 414 (73%) had a home or phone visit instead of a clinic visit, 79 (14%) missed the visit, 46 (8%) had a clinic visit but did not have a BMD scan, and 31 (5%) had a technically inadequate baseline or follow-up scan. Compared with the 2091 participants included in these analyses, the 984 deceased or excluded participants were older at baseline (73.9 versus 73.5 years, p < 0.001), more likely to be black (51% versus 37%, p < 0.001), less likely to report very good or excellent health (36% versus 48%, p < 0.001), and less likely to have normal glucose metabolism (46% versus 56%, p < 0.001). However, baseline BMD at the total hip and femoral neck did not differ.
At the baseline interview, participants were asked, “Has a doctor ever told you that you have diabetes or sugar diabetes?” Women were asked not to include diabetes that only occurred during pregnancy. An inventory of current medications was taken during the interview, including any hypoglycemic medications. Within several weeks of the interview, participants provided a blood sample after an overnight fast (≥8 h). Previously diagnosed diabetes was found in 306 participants who reported a physician diagnosis of diabetes or use of hypoglycemic medications. An additional 174 participants were found to have diabetes at baseline, based on a fasting glucose ≥126 mg/dl or a 75-g, 2-h oral glucose tolerance test (OGTT) ≥200 mg/dl.
Impaired glucose metabolism
Participants with a fasting glucose <126 and ≥110 mg/dl and those with an OGTT <200 and ≥140 mg/dl were considered to have impaired glucose metabolism, based on the American Diabetes Association definition.(25) At baseline, 439 participants had impaired glucose metabolism (IG).
BMD was measured at the proximal femur using DXA (QDR 4500A, software version 9.03; Hologic). DXA quality assurance measurements, including use of daily and cross-calibration phantoms, were performed at both study sites to ensure scanner reliability. The precision of DXA scans of the hip is 1-2%.(26) Measurements were obtained at the baseline exam and at the fourth follow-up exam (exam 5). The average time between the two scans was 4.0 ± 0.1 years.
Weight was measured with a calibrated balance beam scale at baseline and was repeated at each annual clinic visit. Weight change was calculated as the difference between weight measured at the fourth follow-up visit and at baseline. Height was measured at baseline with a Harpenden stadiometer. Body mass index was calculated as weight divided by height squared (kg/m2). Body composition was obtained from whole body scans using DXA (software version 8.21a, QDR 4500A; Hologic).
A questionnaire administered by a trained interviewer during the baseline visit recorded smoking history and physical activity in the past 7 days. Participants were asked to bring medications used in the previous week to the clinic visit. Thiazide diuretics, estrogen, statins, oral steroids, bisphosphonates, calcitonin, raloxifene, and calcium and vitamin D supplement use were coded according to the Iowa Drug Information System (IDIS).(27) In these analyses, bisphosphonates, calcitonin, and raloxifene were grouped together as “osteoporosis medications.” Oral estrogen and other bone-active medications listed above were considered as separate variables.
Grip strength was assessed with a handheld dynamometer (Jamar; TEC, Clifton, NJ, USA). Two trials were performed on each hand; the average of the four trials was used as a measure of grip strength. Walking speed (m/s) was determined from a 20-m walk at a normal pace.
Laboratory measurements were performed on baseline specimens at the Laboratory of Clinical Biochemistry at the University of Vermont. For measurements of plasma glucose and fasting insulin, blood was drawn after an overnight fast (≥8 h). Immediately after, participants ingested 75 g glucose in solution (glucola), and a second blood sample was drawn 2 h later. The fasting insulin and 2-h glucose tolerance tests were not conducted on participants who self-reported a history of diabetes. Creatinine and hemoglobin A1c (A1C) were measured in serum specimens using standard laboratory procedures. Interleukin (IL)-6 was measured by ultrasensitive ELISA (R&D Systems, Minneapolis, MN, USA). TNF-α was measured by an ultrasensitive, solid-phase sandwich ELISA using a monoclonal antibody specific for TNF-α (R&D Systems). Serum levels of C-reactive protein (CRP) were measured by ELISA based on purified protein and polyclonal anti-CRP antibodies (Calbiochem, San Diego, CA, USA). Blind duplicate analyses for IL-6, TNF-α, and CRP showed an average interassay CV of 10.3%, 15.8%, and 8.0%, respectively. Glomerular filtration rate (GFR) was estimated using the abbreviated Modification of Diet in Renal Disease (MDRD) study equation that includes serum creatinine, age, sex, and race.(28) An estimated GFR <60 ml/min/1.73 m2 was defined as renal insufficiency.(28)
Baseline characteristics of the cohort are presented separately for those with diabetes, impaired glucose metabolism, and normal glucose metabolism. χ2 tests were calculated for categorical variables, and ANOVA was used for continuous variables to test for any statistical differences among the three groups.
Annualized absolute change in BMD was calculated as the difference between the results of the baseline and follow-up hip scans divided by the time between scans in years. Annualized percent change in BMD was calculated by dividing the annualized absolute change by the baseline BMD. Annualized absolute change and percent change in BMD were considered as outcomes in the statistical models. Because results were similar, only the results for annualized percent change in BMD are presented.
The associations between the predictor variables (diabetes and impaired glucose metabolism) and the outcome of change in BMD were assessed using generalized linear regression models. Multivariable models for change in BMD, using baseline characteristics, were constructed with backward elimination, retaining variables associated with change in BMD at p < 0.05. Variables were selected for initial entry into the multivariable models if they were known to be correlated with change in BMD from previous studies or if they were associated with change in BMD at the total hip or femoral neck (p < 0.1) and with diabetes or impaired glucose metabolism (p < 0.1) in these data. The initial variables were age, baseline BMD, baseline weight, weight loss from baseline to exam 5, lean mass, fasting insulin, CRP, IL-6, TNF-α, renal insufficiency, current cigarette smoker, grip strength, walking speed, and baseline use of vitamin D supplements, calcium supplements, oral estrogen (women only), osteoporosis treatments (bisphosphonates, calcitonin, or raloxifene), thiazide diuretic, statin, or oral steroid. All statistical analyses were performed in SAS Version 8 (SAS Institute, Cary, NC, USA).
Those with diabetes (DM; N = 480) had higher baseline total hip and femoral neck BMD and weight than those with normal glucose homeostasis (NG). Among white women, those with DM experienced more weight loss during follow-up than NG participants. DM participants had higher levels of inflammatory cytokines (CRP and IL-6) than NG participants (Table 1). Among those with DM, 14% were treated with insulin and 37% were using oral hypoglycemics but not insulin.
Table Table 1.. Characteristics of Older Women and Men by Glucose Metabolism Status, Health, Aging, and Body Composition Study
At the femoral neck, white women with DM lost 0.47% (95% CI: −0.75, −0.19) more BMD per year than NG women in models adjusted for age (Table 2). Additional adjustment for weight change during follow-up resulted in the largest attenuation of the association between DM and change in BMD (−0.36% per year) of any individual covariate. In multivariable models with further adjustment (for age, baseline femoral neck BMD, weight change from baseline to exam 5, IL-6, current smoker, walking speed, statin use, oral estrogen use, renal insufficiency, and vitamin D supplement use), white women with DM lost 0.32% (95% CI: −0.61, −0.02) more BMD per year on average than those with NG. In multivariable models, white women with impaired glucose metabolism (IG) lost 0.18% (95% CI: −0.43, 0.06) per year more BMD than those with NG, but the difference was not statistically significant. In black women and in white and black men, change in femoral neck BMD was similar for those with DM and NG and for those with IG and NG in models adjusted for age and in multivariable adjusted models.
Table Table 2.. Adjusted Differences in Percent Change per Year in BMD, Comparing Participants With DM or IG and Participants With NG
At the total hip, white women with DM lost more bone than those with NG in age-adjusted models (−0.23% per year; 95% CI: −0.47, 0.00), but the difference was reduced after multivariable adjustment (−0.19% per year; 95% CI −0.44, 0.05) and was no longer statistically significant. Adjustment for weight loss resulted in the largest attenuation in the association between DM and bone loss at the total hip among white women (−0.11% per year; 95% CI: −0.33, 0.11). There was no difference in bone loss at the total hip for black women with DM and NG. However, black women with IG had preservation of bone compared with NG in age-adjusted models (0.29% per year; 95% CI: 0.01, 0.57), but the association was somewhat reduced and no longer statistically significant after multivariable adjustment (0.22% per year; 95% CI: −0.04, 0.47). Men with DM or IG did not have a different rate of bone loss compared with NG in age-adjusted or multivariable adjusted models (Table 2).
In this cohort, older white women with DM had more rapid bone loss over 4 years at the femoral neck and the total hip than women with NG. We did not observe an effect of DM on the rate of bone loss for black women or for black or white men at the femoral neck or total hip. In each of the race and sex groups, those with DM had a higher baseline BMD at the total hip, as previously reported by Strotmeyer et al.,(12) and at the femoral neck.
Factors that might contribute to higher cross-sectional BMD in type 2 DM include obesity, hyperinsulinemia, and in women, increased androgen levels associated with lower levels of sex hormone-binding globulin (SHBG).(29,30) Our finding of more rapid bone loss in diabetic white women suggests that the factors contributing to increased BMD become less influential at some point in the progression of DM. Possible mechanisms that could lead to bone loss in older diabetic adults include weight loss,(31) hypercalciuria,(32) impaired renal function,(33) lower endogenous insulin levels as DM progresses,(30) microvascular complications,(34) elevated cytokines,(35) and reduced physical activity.(36)
The increased rate of bone loss associated with type 2 DM in white women was accounted for in part by weight loss. Controlling for weight loss resulted in the largest attenuation of the associations between DM and bone loss at the femoral neck and total hip among white women. Among white women, those with DM experienced more weight loss than those with IG or NG. Although older adults with DM have higher average weight, other studies have also reported greater weight loss.(37,38) Weight loss is strongly associated with bone loss and with increased bone resorption in studies of older women and men.(31,39,40) This association may be caused by decreased loading of the skeleton. In addition, because weight loss is also accompanied by bone loss at non-weight-bearing sites, the bone loss may be caused by reduced levels of sex hormones or other hormones associated with fat and bone metabolism. Measurement artifacts in fan beam DXA introduced by weight loss and the resulting changes in body composition may also account for some of this association. Because DXA determines BMD by comparing the bone with surrounding soft tissue, changes in fat composition with weight loss may alter the BMD results. Changes in beam hardening with reduced tissue thickness after weight loss could also introduce artifacts. However, because the weight loss was modest, the effect on beam hardening and changes in fat and lean tissue at the hip should be minimal.(41) Whether the association between weight loss and bone loss is caused by an actual increase in bone loss, a measurement artifact, or some combination of these, femoral neck bone loss in white women remained independently associated with DM in models adjusted for weight loss.
In the Study of Osteoporotic Fractures, a cohort of older white women, DM was associated with more rapid bone loss at the total hip and femoral neck.(24) In the same cohort, DM was also associated with greater weight loss (unpublished data) as well as higher baseline BMD at the femoral neck(9) and distal radius.(8) Keegan et al.(42) reported more rapid bone loss among those with DM in older women with low baseline BMD. In contrast, Krakauer et al.(2) found no loss of BMD at the radius, based on z scores, over 12 years of follow-up of 19 older patients with type 2 DM. However, these patients also gained weight, on average, during follow-up.
We considered several other factors that might account for increased bone loss in older white women with DM. Although the differences were not statistically significant, those with DM were less likely to report use of medications for osteoporosis and of vitamin D supplements and were more likely to report use of oral steroids. Diabetic participants also had slower walking speed, a factor associated with increased bone loss in univariate models. Similarly, inflammatory cytokines may contribute to bone loss(35) and were elevated in participants with DM. However, inclusion of these factors in the multivariable models did not substantially attenuate the association between DM and loss of BMD at the femoral neck or total hip in white women. Our measurement of physical activity was not associated with bone loss in univariate analyses and was not studied further.
The association between DM and bone loss was seen in white women but not men or black women. This may be explained in part by the lack of association between weight loss and DM in these groups. Whereas white women with DM experienced greater weight loss during follow-up than those with NG, there was little difference in weight loss between men or black women with DM and NG. White women in this study did not seem to have more severe DM as reflected in A1C levels and duration of DM (Table 1).
There is limited evidence that more rapid bone loss as well as lower baseline BMD is associated with increased fracture risk.(43–45) In a study of bone loss at the forearm in postmenopausal women, Riis et al.(43) reported that those with a faster rate of bone loss had double the risk of an osteoporotic fracture, independent of baseline BMD levels. They reported an average difference in forearm bone loss between the fast and normal rates of bone loss of 0.6% per year, similar to the difference we found at the femoral neck between white women with DM and NG. In our age-adjusted models for white women, the difference in the rate of bone loss for those with DM and NG was −0.47% per year at the femoral neck. Over 10 years, this difference in bone loss at the femoral neck for white women in Health ABC with DM and NG would be ∼25% of an SD, which corresponds with a 30% increased risk of hip fracture, based on cross-sectional results in a study of older white women.(46)
Previous studies reporting that increased bone loss is associated with fracture risk were not conducted in a diabetic population. Those with DM tend to have higher BMD and weight, and it is not known if the moderate degree of additional bone loss identified in this study will increase fracture risk in individuals with higher BMD. However, weight loss, a strong predictor of bone loss, has been shown to predict fracture risk even in those with higher baseline weight.(47) Given the increased fracture risk associated with type 2 DM, assessment of change in BMD in older diabetic women, particularly those who are losing weight, may be prudent.
Antiresorptive treatments known to preserve bone and reduce fracture rates in older women seem to be effective in those with DM.(42,48,49) Little is known about the association between glycemic control and bone loss in older adults with DM.(50) Further research is needed to determine if improved glycemic control will preserve bone in older adults with DM.
The study has several strengths, including measurement of fasting glucose and OGTT to determine DM and impaired glucose metabolism, longitudinal measurements of BMD and weight, measurement of medications that might contribute to bone loss in older diabetic adults, and inclusion of men and black participants. Health ABC was limited to well-functioning participants, able to walk 0.25 mi and climb stairs. In addition, our results are limited to those participants who had a BMD scan at the fourth follow-up visit, a group that is likely to be in better health than the full cohort. Indeed, those who did not have a BMD scan at the fourth follow-up visit and were therefore excluded from these analyses were more likely to report poorer health at baseline and were less likely to have NG. It is likely that those who did not have a follow-up scan lost more bone because they were in poorer health. but we do not know if the difference in rate of bone loss between those with DM and NG was similar in the excluded participants. Thus, our results may not apply to older adults with poorer health, particularly those with poor mobility or other functional disabilities.
Despite having higher baseline BMD, white women with DM, but not men or black women, had more rapid bone loss at the femoral neck and total hip than those with NG. The association was partly accounted for by greater weight loss in the diabetic participants. However, at the femoral neck, DM and bone loss remained independently associated after multivariable adjustment. This increased bone loss may contribute to the higher fracture risk observed in older diabetic white women.
This study was supported by N01-AG-6-2101, N01-AG-6-2103, and N01-AG-6-2106 and an American Diabetes Association Junior Faculty Award (AVS). The authors thank Lisa Palermo for statistical analyses.