Combination antiretroviral therapy is associated with visceral fat accumulation, subcutaneous fat loss [1,2] and reduced bone density in HIV-infected patients . In a recent study using dual energy X-ray absorptiometry (DXA), truncal adiposity was not related to bone density among ambulatory HIV-infected patients treated with combined antiretroviral therapy . However, bone density was not compared to clinical indices of lipodystrophy status, more specific indices of visceral and subcutaneous fat, or other antiretroviral therapies, including nucleoside reverse transcriptase inhibitors (NRTI) and non-nucleoside reverse transcriptase inhibitors (NNRTI).
Among non HIV-infected patients, weight and body composition are strong predictors of bone density . However, the effects of lipodystrophy, antiretroviral therapy and fat deposition on bone density in the HIV-infected population remain unknown. We therefore sought to determine bone density in HIV-infected patients with and without lipodystrophy and to examine the relationship between bone density and changes in regional and whole body composition in these populations.
Bone density, indices of bone turnover and calcium metabolism, and body composition were measured in both ambulatory HIV-infected men who had been characterized as to lipodystrophy status (HIV-infected with lipodystrophy, n = 21; HIV-infected without lipodystrophy, n = 20), and matched healthy HIV-negative control subjects (n = 18). All groups were comparable in terms of body mass index (BMI) and age. Data on growth hormone dynamics have been published previously in this group of patients .
HIV-infected men were recruited from the multi-disciplinary HIV clinic at Massachusetts General Hospital and via community advertisement from October 1999 to June 2000. HIV status was verified via ELISA and Western blot testing in all subjects. Exclusion criteria included: BMI < 20 kg/m2, diabetes mellitus, hemoglobin < 9 g/dl, age > 60 years or age < 18 years, and prior administration of testosterone, growth hormone, anabolic hormones, glucocorticoids, antidiabetic agents, or megestrol acetate within 6 months of the study period. Subjects with HIV infection receiving antiretroviral therapy were required to have been on a stable regimen for at least 6 weeks. AIDS diagnosis was determined according to 1993 case criteria established by the Centers for Disease Control and Prevention (CDC) .
Degree of fat deposition in the trunk and neck or fat atrophy in the extremities and face was rated by a single investigator based on a 0–2 point scale with 0.5 point increments. A score of 0 indicated that no discernible change in fat was present, and a score of 2 signified the presence of severe fat deposition or atrophy. A composite score was determined by summing the scores at each of the four sites (trunk, neck, face and extremities). Patients with a waist-to-hip ratio ≥ 0.95 and physical evidence of significant fat deposition (score > 1.5 for the trunk or neck) or fat atrophy (score > 1.5 for either the extremities or face) were determined to have lipodystrophy. Non-lipodystrophic subjects were defined as persons with a waist-to-hip ratio < 0.95 and without significant changes in fat (individual score for each site < 1.5). Patients who met only one of the two criteria of lipodystrophy were excluded. Lipodystrophy status was assigned at a screening visit prior to determination of bone density, markers of bone turnover and calcium homeostasis. The lipodystrophy scale has been used by our laboratory in previous studies [5,7] and the lipodystrophy score determined from this scale correlates strongly with anthropometric and radiologic assessments of fat such as waist-to-hip ratio (r, 0.81;P < 0.0001) and quantitative computed tomography (QCT)-measured ratio of intra-abdominal (visceral) fat area to total abdominal cross-sectional area (r, 0.72;P < 0.0001).
Body composition and bone density assessment
Whole body lean mass, fat mass, and bone density were assessed using DXA (Hologic-4500 densitometer, Hologic, Inc., Waltham, Massachusetts, USA). Lumbar spine bone density was measured by single-slice cross-sectional abdominal QCT scanning at the lumbar spine as described by Katznelson . A lateral scout image standardized the location of the single slice image at the mid-L4 pedicle. Scanning specifics were uniformly set (80 kV, 70 mA, 2 sec, 1 cm slice thickness, 48 fields of view). Attenuation was calibrated to an internal phantom containing K2HPO4.
Subcutaneous and intra-abdominal fat was quantified with image analysis software (ALICE, 4.3.9, 1999, Parexel Corp, Waltham, Massachusetts, USA). The body perimeter was identified on the scans using a crude manual trace to encompass the entire torso, followed by a ‘shrink wrap’ algorithm that identified the surface. A manual trace was then performed through the internal surface of the abdominal wall musculature. Fat was identified as all pixels with attenuation values between −250 and −50 Hounsfield Units. All fatty tissues between the body surface and the deep musculature was considered subcutaneous and all fatty tissue deep to the muscles was considered intra-abdominal.
The ratio of intra-abdominal (visceral) fat area to total cross-sectional area, the ratio of visceral to subcutaneous abdominal fat areas (VAT : SAT) and the patient visceral to total abdominal fat areas (VAT : TAT) were also determined from the single slice cross-sectional QCT scan.
Blood sampling was performed after an overnight 12 h fast. CD4 cell counts were measured by flow cytometry (FACS scan analyzer, Becton-Dickson Immunocytochemistry Systems, San Jose, California, USA). Serum osteocalcin, a marker of bone formation and osteoblast function, was measured using a two-site immunoradiometric assay (Nichols Institute Diagnostics, San Juan Capistrano, California, USA). The lower detection limit of the assay was 0.5 ng/ml, and the intra-assay coefficient of variation (CV) was 3.2–5.2%. Serum parathyroid hormone (PTH), which regulates calcium homeostasis and bone resorption, was assessed using a two-site immunoradiometric assay (Nichols Institute Diagnostics) with a lower detection limit of 1 pg/ml and an intra-assay CV of 1.8–3.4%. Serum calcium and phosphorus levels and urine creatinine were determined by standard techniques. Serum 25-hydroxyvitamin D was measured following an acetonitrile extraction via a radioimmunosorbent assay (RIA) equilibrium procedure (INCSTAR, Stillwater, Minnesota, USA) based on an antibody with specificity to 25-hydroxyvitamin D. This assay can detect vitamin D levels ≤ 3 ng/ml; the intra-assay CV was 9.6–13.4%. Testosterone levels were measured by RIA with an intra-assay CV of 5–12% (Diagnostics Products Corp., Los Angeles, California, USA).
Urine deoxypyridinoline and urine N-telopeptide, sensitive markers of bone resorption, were measured. Urine deoxypyridinoline was determined from a 24 h collection in a preservative-free container using a competitive enzyme immunoassay with a lower detection limit of 1.1 nmol/l and an intra-assay CV of 4.3–8.8% (Metra Biosystems, Mountain View, California, USA). Urine N-telopeptide concentrations were also determined from a 24 h urine collection by a competitive inhibition ELISA test with a lower detection limit of 20 nM bone collagen equivalents and an intra-assay CV of < 20% (Ostex International, Seattle, Washington, USA). Urine deoxypyridinoline and N-telopeptide concentrations were corrected for urinary creatinine.
Lipodystrophy status, HIV status, and protease inhibitor (PI) use were coded as dichotomous variables. Bone density measurements, anthropometric dimensions, CD4 cell counts, and bone markers were treated as continuous variables. The student's t test was used to compare continuous variables amongst the different groups. The chi-squared test was used for categorical variables. In a secondary analysis, we compared clinical parameters by current PI status and duration of total PI use. Bone density, body composition and bone turnover data were compared by univariate logistic regression analyses. Intra-abdominal fat was evaluated as a predictor of bone density in the entire study population of HIV infected patients using a multivariate regression analysis, with age, BMI, historical lowest body weight, PI use and extremity fat as co-variates. Bone density data from two additional HIV-negative control subjects were rejected as extreme outliers using the Dixon criterion  and were not used in the analysis.
Demographic and clinical data based on lipodystrophy categorization are presented in Table 1. Bone density, as measured by QCT, was reduced significantly in the HIV-positive lipodystrophy versus HIV-negative control groups and HIV-positive lipodystrophy versus HIV-positive non-lipodystrophy groups, but was not different between HIV-positive non-lipodystrophy versus HIV-negative control groups. The difference in bone density between HIV-positive lipodystrophy patients and HIV-negative control subjects was approximately one standard deviation. The differences in bone density between the groups remained significant (HIV-positive lipodystrophy versus HIV-positive non-lipodystrophy, P = 0.04; HIV-positive lipodystrophy versus HIV-negative controls, P = 0.04), controlling for minority status in the individual groups. Using a criterion of < 110 mg/cm3 as an indicator of increased fracture risk , we observed that 24% of HIV-positive lipodystrophy patients versus 5% of HIV-positive non-lipodystrophy patients versus 0% of the HIV-negative controls were below the fracture risk threshold (P = 0.03). Total body bone density measured by DXA was not significantly different between the groups. No significant differences were seen between the groups for age, race, BMI, historical lowest body weight, NNRTI use, CD4 cell counts, viral load, history of opportunistic infection or diagnosis with AIDS. PI and NRTI use were more prevalent among those diagnosed with lipodystrophy.
There was no difference by lipodystrophy categorization in any bone turnover marker. 25-Hydroxyvitamin D concentrations were slightly but significantly increased (P = 0.03) in HIV-infected subjects with lipodystrophy compared to HIV-negative control subjects, but there were no differences in calcium or phosphorus levels between the groups (data not shown). In addition, 25-hydroxyvitamin D concentrations did not correlate with bone density (Table 2). Testosterone levels were not different in the HIV-lipodystrophy group as compared to HIV-negative control subjects [mean ± SD, 21.8 ± 8.1 versus 20.4 ± 5.0 nmol/l;P = 0.52] and did not correlate with bone density (Table 2).
Univariate regression analyses for QCT bone density and clinical variables among all HIV-infected individuals (n = 41) are shown in Table 2. Bone density was not associated with either duration or current use of PI or NRTI therapy. In contrast, bone density was increased in current NNRTI users versus non-users [mean ± SD, 158 ± 28 versus 136 ± 31 mg/cm3;P = 0.04], but no association was seen between bone density and duration of NNRTI use (Table 2). Bone density was inversely associated with age and CD4 cell count.
Comparison of bone density with body composition indices demonstrated that volumetric bone density was inversely associated with the visceral fat (Fig. 1) but was not associated with indices of subcutaneous fat, extremity fat, and total truncal fat by DXA. In a multivariate regression analysis among all HIV-infected subjects, abdominal visceral fat remained a significant predictor of bone density controlling for age, BMI, historical lowest body weight, PI use and extremity fat (Table 3).
In a subanalysis by PI status among all HIV-infected patients, no differences in bone density or bone turnover markers were observed. QCT bone density [mean ± SD), 144 ± 33 versus 139 ± 29 mg/cm3] was not different between current PI users and non-PI users. Serum testosterone (mean ± SD, 22.7 ± 8.1 versus 26.3 ± 10.7 nmol), osteocalcin, urinary N-telopeptide, urinary deoxypyridinoline, 25-hydroxyvitamin D and PTH were likewise not different in current PI users versus non-PI users (P > 0.1 for all comparisons). Similarly, no association was observed between markers of bone turnover and the duration of antiviral medication exposure (Table 4). Serum calcium was associated positively with duration of PI use (r, 0.38;P = 0.02) but calcium levels were within normal limits in all patients.
Excess abdominal visceral fat deposition and subcutaneous fat atrophy have been well described in HIV-infected patients receiving combined antiretroviral therapy , but the relationship between changes in body composition and bone density remains unknown in this population. Paton et al. did not observe significant osteopenia among HIV-infected men investigated prior to the introduction of combined antiretroviral therapy . In contrast, Tebas et al. recently demonstrated reduced bone mass in HIV-positive patients receiving PI. However, the mechanism of reduced bone density in HIV lipodystrophy remains unknown.
In this study, we demonstrated reduced lumbar spine volumetric bone density among patients with HIV lipodystrophy. Lumbar spine bone density was strongly predicted by increased visceral adiposity among HIV-infected patients. In contrast to the study by Tebas et al. and in agreement with the recent report of Billaud et al. , we demonstrate no independent effect of PI status on bone density controlling for change in visceral fat. Although osteopenia has been noted in patients with HIV infection, to our knowledge, this is the first report of an association between excess visceral fat deposition and reduced bone density in HIV lipodystrophy.
We used single-slice QCT at the L4 pedicle to measure bone density in our study. Single slice QCT bone mineral measurement of the lumbar spine is the most sensitive method for determining bone density and fracture risk . Our HIV lipodystrophy population demonstrated reduced QCT bone density as compared with HIV-infected non-lipodystrophic men and normal age-matched HIV-negative controls. The difference in bone density between groups was on the order of magnitude of a single standard deviation, suggesting moderate bone loss that may predispose HIV-infected patients with lipodystrophy to an increased fracture risk.
Increased visceral fat and loss of abdominal and peripheral subcutaneous fat are common features of HIV-related lipodystrophy. Although increased visceral fat and reduced subcutaneous fat often co-exist, patients may differ in phenotypic expression and location of fat distribution. To account for such individual differences, we examined the relationship between the visceral fat and bone density among all HIV-infected patients independent of lipodystrophy categorization. We found that increased visceral fat was strongly associated with reduced bone density (Fig. 1). The relationship between visceral fat and bone density has not been reported previously and was robust, remaining significant when controlling for age, BMI, lowest body weight, PI use, and extremity fat.
In contrast, we found no relationship between bone density and loss of peripheral subcutaneous fat in univariate or multivariate modeling. Our data differ from a recent report showing a relationship between lumbar spine bone density and change in subcutaneous fat by DXA in HIV-infected patients on highly active antiretroviral therapy (HAART) . Use of regional lumbar bone density instead of the more specific measurement of volumetric bone density may account for such differences.
Lumbar spine bone density determined by QCT was reduced whereas total body bone density determined by DXA was not reduced in the lipodystrophy patients. This difference probably reflects known differences in bone composition measured by the two techniques used in this study. Quantitative computed tomography scanning of the lumbar spine measures trabecular bone whereas assessment of total body bone density by DXA measures primarily cortical bone. Trabecular bone is distinct from cortical bone and may be lost at a greater rate than cortical bone in patients with HIV lipodystrophy.
Prior reports demonstrate low osteocalcin levels with increased disease severity among HIV-infected patients [16,17] suggesting that a defect in bone formation may be the mechanism by which reduced bone density occurs. In contrast, the HIV-infected lipodystrophic men in our study were healthy and did not have a prior history of increased opportunistic infection or lower CD4 cell counts as compared to HIV-infected non-lipodystrophic men. A similar proportion of men in each group (50%) met the CDC definition of AIDS. The osteopenia in our HIV-infected patients with lipodystrophy may represent a distinct entity from the osteopenia illustrated in the wasted and ill AIDS patients. More recently, Tebas et al. reported increased serum osteocalcin in subjects receiving PI-based HAART . In contrast, our data do not suggest a relationship between osteocalcin levels and PI use. Furthermore, our data do not show a relationship between markers of bone turnover and bone density in patients with HIV lipodystrophy.
One potential mechanism to explain the observation of reduced bone density in patients with HIV lipodystrophy is an increase in marrow fat. Abnormal marrow or whole body fat metabolism may affect bone cell differentiation or the marrow cytokine environment in patients with HIV lipodystrophy. A strong inverse association was observed between increased abdominal visceral fat and reduced lumbar spine bone density. Prior studies have shown that fat cells not only play a role in lipid metabolism but also in osteogenesis  and hematopoiesis . Thus, disruption of normal fat metabolism may result in abnormal osteogenesis. In addition, leptin, a centrally acting peptide produced in fat cells, has been found to inhibit bone formation via a central inhibitory pathway . Further studies on fat metabolism and leptin are needed to clarify potential effects on bone turnover. Although fat metabolism may play a role in the HIV osteopenia, we did not find abnormalities in bone turnover and calcium homeostasis in our patients to suggest abnormal bone metabolism in HIV-associated lipodystrophy.
Alternatively, increased marrow fat might affect the measurement of spinal bone density in the absence of a true reduction in bone density. Single energy QCT is affected by increased adiposity in the bone marrow . For each 10% increase in marrow fat volume, bone density is underestimated by 7 mg/cm3 . Therefore, a 30% increase in bone marrow fat could account for the reduction in bone density reading by QCT seen in our HIV-infected lipodystrophy patients. Indeed, increased marrow fat may be one factor contributing to the recently reported increased prevalence of avascular necrosis among asymptomatic HIV-infected individuals on antiretroviral therapy .
The clinical significance of reduced bone density in HIV-infected patients with lipodystrophy relates to the potential for increased fractures in this population. The potential impact of reduced bone density and/or increased marrow fat on the development of fractures in this population remains unknown. Although anecdotal reports of spinal fracture in HIV-infected patients have been published , fractures were not reported in a recent Australian cohort of 221 otherwise healthy HIV-infected men evaluated for lipodystrophy . However, further longitudinal studies with larger numbers of patients are necessary to assess the true fracture risk in this population.
Other possible etiologies of reduced bone density in patients with HIV infection were not explored in this study. Carr et al. recently demonstrated an association between acidemia and bone loss in HIV-infected patients . Acidemia is well known to increase mineral dissolution  and might affect bone density in this population. In addition, the effect of antiretroviral therapy on osteogenesis remains unclear. Although our data do not suggest a difference in bone density related to PI or NRTI status, Wang et al.  recently reported that PI may directly inhibit bone formation and prevent osteoclast differentiation. Lastly, HIV has been found to infect osteoblasts and fibroblasts directly, and may itself contribute to bone pathology.
In conclusion, the etiology of reduced bone density among patients with HIV lipodystrophy remains unknown. We demonstrate for the first time a clear association between excess visceral fat deposition and bone density in this population. Fat accumulation in the marrow and/or the related effects of abnormal adipogenesis on the marrow microenvironment may contribute to reduced bone density in HIV-infected patients on combination antiretroviral therapy. Additional studies are needed to determine whether marrow fat is increased in HIV-infected patients with lipodystrophy and whether abnormal fat metabolism contributes to low bone density in this population.
Supported in part through NIH grants R01 DK 59535, DK 49302, DK07477, M01-RR-01066 and Serono Labs.
1. Carr A, Samaras K, Burton S. et al
. A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors. AIDS 1998, 12: F51 –F58.
2. Saint-Marc T, Partisani M, Poizot-Martin I. et al
. Fat distribution evaluated by computed tomography and metabolic abnormalities in patients undergoing antiretroviral therapy: preliminary results of the LIPOCO study. AIDS 2000, 14: 37 –49.
3. Tebas P, Powderly WG, Claxton S. et al
. Accelerated bone mineral loss in HIV-infected patients receiving potent antiretroviral therapy. AIDS 2000, 14: F63 –F67.
4. Kanis J, Johnell O, Gullberg B. et al
. Risk factors for hip fracture in men from southern Europe: the MEDOS study. Mediterranean Osteoporosis Study.
Osteoporos Int 1999, 9: 45 –54.
5. Rietschel P, Hadigan C, Corcoran C, et al. Assessment of growth hormone dynamics in human immunodeficiency virus-related lipodystrophy. J Clin Endocrinol Metab 86
6. Centers for Disease Control and Prevention. Revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR 1992, 41: RR -17.
7. Hadigan C, Corcoran C, Basgoz N. et al
. Metformin in the treatment of HIV lipodystrophy syndrome. JAMA 2000, 284: 472 –477.
8. Katznelson L, Rosenthal DI, Rosol MS. et al
. Using quantitative CT to assess adipose distribution in adult men with acquired hypogonadism. Am J Roentgenol 1998, 170: 423 –427.
9. Dixon WJ, Dixon WJ, Massey FJ Jr (eds). Introduction to Statistical Analysis
edn. New York: McGraw-Hill Book Company, Inc., 1957.
10. Cann CE, Genant HK. Cross-sectional studies of vertebral mineral using quantitative computed tomography. J Comput Assist Tomogr 1982, 6: 216. 216.
11. Silva M, Skolnik PR, Gorbach SL. et al
. The effect of protease inhibitors on weight and body composition in HIV-infected patients. AIDS 1998, 12: 1645 –1651.
12. Paton NIJ, Macallan DC, Griffin GE. et al
. Bone mineral density in patients with human immunodeficiency virus infection. Calcif Tissue Int 1997, 61: 30 –32.
13. Billaud E. Osteopenia and osteoporosis in HIV patients: role of antiviral therapy?
Presented at 40 th Meeting of ICAAC
. Toronto, September 2000.
14. Lang TF, Li J, Harris ST. et al
. Assessment of vertebral bone mineral density using volumetric quantitative CT. J Comput Assist Tomogr 1999, 23: 130 –137.
15. Nolan D, Upton R, James I, et al. Longitudinal analysis of bone mineral density (BMD) in HIV-infected patients treated with HAART: changes in BMD correlate with change in subcutaneous fat; with an additional independent effect of indinavir therapy to increase BMD. Second International Workshop on Adverse Drug Reactions and Lipodystrophy in HIV.
Toronto, September 2000 [abstract 031].
16. Aukrust P, Haug CJ, Ueland T. et al
. Decreased bone formative and enhanced resorptive markers in human immunodeficiency virus infection: indication of normalization of the bone-remodeling process during highly active antiretroviral therapy
. J Clin Endocrinol Metab 1999, 84: 145 –150.
17. Serrano S, Marinoso ML, Soriano JC. et al
. Bone remodeling in human immunodeficiency virus-1 infected patients. A histomorphometric study.
Bone 1995, 16: 185 –191.
18. Tebas P, Yarasheski KE, Whyte M, et al. Serum and urine markers of bone mineral metabolism in HIV infected patients taking protease inhibitor containing potent antiretroviral therapy. Second International Workshop on Adverse Drug Reactions and Lipodystrophy in HIV.
Toronto, September 2000 [abstract 029].
19. Dexter TM, Allen TD, Lajtha LG. Conditions controlling the proliferation of haemopoietic stem cells in vitro. J Cell Physiol 1977, 91: 335 –344.
20. Howlett CR, Cave J, Williamson M. et al
. Mineralization in in vitro cultures of rabbit marrow stromal cells. Clin Orthop 1986, 213: 251 –263.
21. Ducy P, Amling M, Takeda S. et al
. Leptin inhibits bone formation through a hypothalamic relay: a central control of bone mass. Cell 2000, 100: 197 –208.
22. Richardson ML, Genant HK, Cann CE. et al
. Assessment of metabolic bone diseases by quantitative computed tomography. Clin Orthop 1985, 195: 224 –238.
23. Faul DD, Couch JL, Cann CD. et al
. Composition-selective reconstruction for mineral content in the axial and appendicular skeleton
. J Comput Assist Tomogr 1982, 6: 202 –203.
24. Masur H, Miller KD, Jones EC, et al. High prevalence of avascular necrosis (AVN) of the hip in HIV infection: magnetic resonance imaging of 339 asymptomatic patients. 38 th IDSA Meeting
. New Orleans, September 2000 [abstract 15].
25. Stephens EA, Das R, Madge S. et al
. Symptomatic osteoporosis in two young HIV-positive African women. AIDS 1999, 13: 2605 –2606.
26. Carr A, Miller J, Cooper DA. Osteopenia in HIV-infected men: association with lactic academia and lower weight pre-antiretroviral therapy. Second International Workshop on Adverse Drug Reactions and Lipodystrophy in HIV.
Toronto, September 2000 [abstract 032].
27. Grinspoon SK, Baum HBA, Kim V, Coggins C, Klibanski A. Decreased bone formation and increased mineral dissolution during acute fasting in young women. J Endocrinol Metab 1995, 80: 3628 –3633.
28. Wang MWH, Teitelbaum SL, Tebas P, Powderly WG, Ross FP. Indinavir inhibits bone formation while ritonavir inhibits osteoclast differentiation and function. Retrovirus
2001. Chicago. February 2001 [abstract 541].
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
HIV; antiretroviral therapy; lipodystrophy; bone density