The introduction of highly active antiretroviral therapy (HAART) has considerably reduced the mortality of human immunodeficiency virus (HIV)-infected patients in developed countries. Consequently, the HIV prevalence in the general population is growing, and the age of the HIV-infected individuals is increasing, to the point that by 2015, the median age of HIV-infected individuals in the US is expected to exceed 50 years . The improved survival of HIV-infected patients has already been associated with a shift in the underlying cause of death among these patients, with lesser representation of ‘AIDS-related causes’ and greater representation of ‘non-AIDS-related’ deaths . Low bone mass and osteoporotic fractures have also emerged among the chronic problems affecting this aging HIV-positive population.
A number of studies have described accelerated bone loss and higher rates of osteopenia and osteoporosis among HIV-infected individuals than in the general population [3,4]. Rates of osteoporotic fractures have also been shown to be higher among HIV-infected patients than age-matched controls [5,6]. Whereas HIV infection itself has adverse skeletal effects, HAART may also contribute to accelerated bone loss [4,7–9]. Previous studies have suggested that antiretroviral drugs differ in their impact on bone health: tenofovir (TDF) has been found to be associated with a greater decline in bone mineral density (BMD) than stavudine  or abacavir . Prophylactic use of TDF has also been shown to cause a small but significant decline in BMD in HIV-uninfected individuals . Also, earlier studies had suggested that exposure to protease inhibitors decreased BMD [4,13], and it has been recently suggested that atazanavir is associated with increased risk of osteoporosis, compared to efavirenz . Finally, antiretroviral initiation has been shown to be associated with a rapid and significant increase in levels of serologic markers of increased bone turnover (which might signify increased bone fragility) [14–16].
Although these findings have raised concern for increased risk for osteoporotic fractures, there has never been an evaluation of the osteoporotic fracture risk associated with cumulative exposure to TDF and other antiretroviral drugs.
Our source of data was the Veterans Health Administration (VHA)'s Clinical Case Registry (CCR), spanning a 21-year period, from 1988 to 2009. The CCR database aggregates detailed demographic, diagnostic, therapeutic and healthcare utilization data on all HIV-infected patients from all VHA facilities to the unique patient level. It comprises Veterans Administration specific codes for healthcare utilization (hospitalizations and clinic visits), National Pharmacy Benefits Management (PBM) codes for drug utilization, as well as current procedural terminology and ICD-9 codes for procedures and diagnoses, respectively.
Patient-days of antiretroviral use prior to osteoporotic fracture event were calculated for each antiretroviral drug, and survival analyses done to predict new osteoporotic fracture. PBM data were used to identify supply of each antiretroviral. Exposure time was determined for each antiretroviral drug or class of antiretrovirals by calculating the number of days covered by each prescription. It was defined as time of cumulative exposure to the antiretroviral or class of antiretrovirals from their initial prescription to the first occurrence of one of the following: development of the first osteoporotic fracture episode or death; discontinuation of the antiretroviral; last recorded patient encounter; 31 December 2009 (date of censure of the dataset).
The following demographic variables were extracted: age, race (Black, White or other), and sex. Body mass index (BMI) was calculated annually based on height and weight. For patients whose BMI was missing during certain years, we imputed the missing data by assuming a linear trend in BMI with respect to time. Patients with a value of less than 20 kg/m2 were classified as having low BMI. Patients with diabetes mellitus or tobacco use were identified by the presence of at least one of the ICD-9 codes indicated in Table 1 listed as their discharge or outpatient diagnoses.
Patients were classified as having chronic kidney disease (CKD) if their estimated glomerular filtration rate (eGFR) was below 60 ml/min per 1.73 m2 by the modificaton of diet in renal disease method: GFR (ml/min per 1.73 m2) = 175 × (Scr)−1.154 × (Age)−0.203 × (0.742 if female) × (1.212 if African American).
Our primary outcome was incident vertebral, hip and wrist fractures (selected on the basis of their likelihood of being related to osteoporosis), and referred to herein after as ‘osteoporotic fractures’ . These fractures were identified using ICD-9 codes outlined in Table 1. For patients with multiple osteoporotic fracture events, only the first one was counted.
A validation of ICD-9 codes for ascertainment of fractures in the Veterans Administration databases (compared to a review of the radiology reports) was done in a recently presented study . It showed a positive predictive value and a negative predictive value of 0.64 and 1.00, respectively, with a level of agreement of 0.97.
We summarized the CCR data by mean, median, standard deviation (SD) and proportion. Annual and age group-specific incidence rates were computed for osteoporotic fracture. Univariate Cox survival models were used to assess the marginal association between osteoporotic fracture and various risk factors, such as age at cohort entry, race, smoking, BMI, CKD, hepatitis C virus (HCV) and cumulative antiretroviral exposure. The annual BMI and CKD measurements were included into the model as time-dependent covariates. Categories explored were cumulative exposure to tenofovir (TDF), abacavir (ABC), zidovudine (ZDV) or stavudine (D4T) (ZDV/D4T) any boosted protease inhibitors (rPI), and non-nucleoside reverse transcriptase inhibitors (NNRTIs). Two multivariable models were constructed to examine the association of these antiretroviral exposures with osteoporotic fracture: model 1 (MV1), controlling for age, race, tobacco use, diabetes, CKD, HCV and BMI; and model 2 (MV2), controlling model 1 variables and concomitant exposure to other antiretroviral drugs. Sex was not included in the model since over 98% of the study population is male. Statistical significance was declared at P less than 0.05.
Two separate analyses were conducted using these survival models for the entire study population, and only for patients entering the cohort in the HAART era (since 1 January 1996).
In the HAART era, we also examined the osteoporotic fracture risk associated with specific protease inhibitors with over 10 000 person-years of exposure in the database: nelfinavir (NFV), indinavir (IDV), atazanavir (ATV) and lopinavir/ritonavir (LPV/RTV) in univariate and multivariate models 1 and 2 as above.
All statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, North Carolina, USA).
Study population and treatment exposure
We identified 56 660 patients who used VHA services for HIV disease during the study period and were included in CCR, including 22 005 who had clinic/outpatient or inpatient discharge data within the past 12 months of observation (1 January 2009–31 December 2009). They contributed a total of 305 237 person-years of follow-up. Among them, 39 277 (69.4%) had at least at least 1 month of ART exposure, and total ART exposure in the cohort was 164 414 person-years. The proportion of male patients was 98%.
Patients with osteoporotic fracture had a higher median age than those without (46 vs. 44 years), were more likely to be white (57 vs. 45% of those without osteoporotic fracture), smokers (56 vs. 32%), to have diabetes (25 vs. 15%), a BMI below 20 (49 vs. 33%) and have HCV co-infection (51 vs. 31%) (P < 0.0001 for all comparisons).
Rates of osteoporotic fractures
A total of 951 individual patients sustained at least one osteoporotic fracture during the period of observation (124 vertebral, 486 wrist and 341 hip). Rates for both hip and vertebral fractures (per 1000 person-years) increased progressively from the 18–29-year age group (0.02 and 0.00, respectively) to the 70+-year age group (4.49 and 1.65, respectively). The rates of wrist fractures increased progressively from the 18–29-year age group (0.11 per 1000 patient-years) to the 50–59-year age group (2.41), then declined in the later age groups (1.64 for the 60–69 years and 1.20 for the 70+ years).
Cumulative antiretroviral use and risk of osteoporotic fractures: 1988–2009
Unadjusted and adjusted hazard ratios for osteoporotic fracture associated with cumulative exposure to any antiretroviral therapy and traditional osteoporotic risk factors are presented in Table 2.
Figure 1 presents hazard ratios for osteoporotic fracture associated with different individual antiretroviral drugs and antiretroviral classes, in the univariate and multivariate models described above.
Tenofovir exposure (46 062 person-years) was associated with a yearly hazard ratio for osteoporotic fracture of 1.08 (95% CI 1.02–1.15, P < 0.001) in univariate model, 1.06 (0.99–1.12, P = 0.079) in MV1 and 1.06 (0.99–1.14, P = 0.106) in MV2. Boosted protease inhibitor exposure (41 336 person-years) was associated with hazard ratio of 1.06 (1.01–1.12, P = 0.015) in univariate model, 1.04 (0.99–1.10, P = 0.142) in MV1 and 1.03 (0.97–1.09, P = 0.349) in MV2. Exposure to ABC (24 251 person-years), ZDV/D4T (94 595 person-years) or NNRTI (59 857 person-years) was not significantly associated with increased risk of osteoporotic fracture in univariate or multivariate models.
Cumulative antiretroviral use and risk of osteoporotic fractures in the HAART era: 1996–2009
There were 32 439 patients who entered the cohort in the HAART era (191 258 person-years). As expected, the proportion of patients exposed to antiretroviral therapy was significantly higher among patients who entered the cohort in the HAART era (83.6%, compared to 69.4% in the entire cohort). The rate of osteoporotic fractures was significantly higher in the HAART era (4.09 events/1000 patient-years) compared to the pre-HAART era (1.61 events/1000 patient-years).
For patients who entered the cohort in the HAART era, TDF exposure (38 009 person-years) was associated with a yearly hazard ratio for osteoporotic fracture of 1.16 (95% CI 1.08–1.24, P < 0.001) in univariate model, 1.13 (1.05–1.21, P = 0.001) in MV1 and 1.12 (1.03–1.21, P = 0.011) in MV2. Boosted protease inhibitor exposure (32 109 person-years) was associated with hazard ratio of 1.11 (1.05–1.18, P = 0.001) in univariate model, 1.08 (1.01–1.15, P = 0.026) in MV1 and 1.05 (0.97–1.13, P = 0.237) in MV2. Exposure to ABC (18 885), ZDV/D4T (68 376 person-years) or NNRTI (48 943 person-years) was again not significantly associated with increased risk of osteoporotic fracture in univariate or multivariate models (Fig. 2).
Since exposure to TDF and rPI was the only one significantly associated with increased osteoporotic fracture risk in univariate analysis, we then evaluated the effect of cumulative exposure to both TDF and rPI, and that of exposure to individual rPIs. For this analysis, we determined four different exposure categories: exposure to neither TDF nor rPI (referent category); exposure to TDF, but not rPI; exposure to rPI, but not TDF; and concomitant exposure to TDF and rPI. Concomitant exposure to both TDF and rPI associated with a greater osteoporotic fracture risk (hazard ratio 1.16, CI 1.04–1.30) than exposure to either TDF without rPI (hazard ratio 1.11, CI 1.01–1.21) or rPI without TDF (hazard ratio 1.10, CI 1.01–1.22).
Regarding exposure to individual protease inhibitors, we selected those with the highest person-years of exposure in the database: indinavir (IDV; 12 124 person-years), atazanavir (ATV; 12 685 person-years), nelfinavir (NFV; 14 356 person-years) and lopinavir/ritonavir (LPV/RTV; 15 319 person-years). Only LPV/RTV was associated with significantly increased osteoporotic fracture risk in univariate model (hazard ratio 1.17, CI 1.08–1.26, P < 0.0001). The association remained significant in MV1 (hazard ratio 1.13, CI 1.04–1.22, P = 0.005) and barely in MV2 (hazard ratio 1.09, CI 1.00–1.20, P = 0.051). Exposure to other protease inhibitors (boosted or unboosted) was not associated with significantly increased osteoporotic fracture risk (Fig. 3). Also, cumulative exposure to RTV (regardless of dosing used for boosting) was not predictive of fracture risk (hazard ratio 1.06, CI 0.97–1.15, P = 0.200) (Fig. 3). Also, as the dose of RTV used for boosting of other protease inhibitors varies, we examined whether RTV dose was associated with osteoporotic fracture risk, and it was not (hazard ratio 1.009, P = 0.279).
The goal of this study was to evaluate the association of antiretroviral exposure and other risk factors with incident osteoporotic fracture among HIV-infected patients. Using a cohort of 56 660 HIV-infected patients followed for a mean of 5.4 patient-years, we found for the first time that exposure to TDF and rPI was associated with a modestly increased osteoporotic fracture risk. After controlling for traditional osteoporotic fracture risk factors, these associations were no longer statistically significant. However, stronger associations between either TDF or rPI with osteoporotic fracture were found when analysis was limited to patients who entered the cohort in the HAART era. TDF remained independently predictive of osteoporotic fracture risk (12% higher risk per year of exposure) after controlling for traditional osteoporotic fracture risk factors and concomitant antiretroviral drug used.
Furthermore, we found evidence of an interaction between TDF and rPI in their association with osteoporotic fracture risk. Concomitant exposure to both TDF and rPI associated with a greater osteoporotic fracture risk than exposure to either TDF without rPI, or rPI without TDF.
Among individual protease inhibitors, only LPV/RTV was found to be associated with a significantly increased risk of osteoporotic fracture. Whereas these could be explained by concomitant use of RTV with LPV, neither RTV alone nor boosted ATV or IDV was associated with increased risk.
The pathogenesis of HIV-associated increased risk of osteoporosis is likely multifactorial and is incompletely understood. Individuals with HIV infection have a high prevalence of traditional risk factors for osteoporosis – including low BMI, poor nutrition, frequent glucocorticoid use, smoking, and vitamin D deficiency – as well as co-infection with HCV [6,19]. HIV infection itself may independently be associated with an increased risk of osteoporosis, as has been shown in studies evaluating BMD in patients naive to antiretroviral therapy [8,20,21]. This might be mediated through the presence of chronic inflammation and its effects on bone remodeling, and imbalances in the gonadal and calciotropic hormonal systems that regulate bone formation and resorption. ART, particularly immediately after the initiation of antiretroviral therapy, may also directly contribute to low BMD [3,4,22,23].
Previous studies have suggested that antiretroviral drugs differ in their impact on bone health: TDF has been found to be associated with a greater decline in BMD than stavudine  or abacavir . Antiretroviral regimens have also been shown to increase expression of markers of bone turnover. Our findings suggest for the first time that these antiretroviral associations with either decreased BMD or increased bone turnover markers might be reflected in increased risk of osteoporotic fracture. These findings, if confirmed, might warrant consideration of osteoporotic fracture risk in decisions in ART initiation among HIV-infected patients. Hansen et al.  showed that in the HAART era, only HAART-exposed HIV patients had a higher risk of low-energy fractures than uninfected patients. Exposure to TDF was not independently predictive of higher fracture risk. However, unlike their analysis, we have been able to model cumulative exposure to TDF and other antiretroviral drugs prior to fracture events. We have also been able to adjust for concomitant exposure to other antiretroviral drugs.
Consistent with data from the general population osteoporotic fracture was independently associated with advancing age, race other than Black, low BMI, and smoking in HIV-infected individuals. These traditional osteoporotic fracture risk factors are much more important in predicting fracture risk among HIV-infected patients than antiretroviral exposure. We did not find CKD and diabetes to be independently associated with an increased risk of osteoporotic fractures after controlling for other risk factors, although both were significantly associated with increased fracture risk in the univariate analysis.
Whereas we found significant increase in fracture rates in the HAART era, cumulative antiretroviral may not entirely account for the HAART-to-pre-HAART eras increased risk. Greater fracture rates, higher (significant) hazard ratio for TDF and rPI in the HAART era could be due to longer survival, and exclusion of most patients with no antiretroviral, mono, or dual antiretrovirals. Also, lower BMD was associated with controlled HIV replication in a recent study , and controlled viremia is much more likely in the HAART era.
With increased survival on HAART and the aging of the HIV population, the morbidity and mortality related to non-AIDS-related diseases are rapidly increasing, with cardiovascular disease (CVD) and liver disease (mostly HCV-related) being one of the most common causes of morbidity and mortality. Bone disease should be added to the list. Interventions used to mitigate that risk in the general population (calcium and vitamin D) have been shown to improve BMD of HIV-infected patients  and are advocated in HIV guidelines [26,27]. Future studies will need to determine whether these and/or changes in antiretroviral therapy will result in decreased risk of osteoporotic fractures. As a cautionary tale, recent data suggest that changes in BMD by such interventions are very poorly predictive of osteoporotic fracture risk in the general population .
The strengths of our study include a large sample size (more than 56 000 patients) and high number of osteoporotic fracture events (over 900), a uniform data collection on exposures and outcomes across Veterans Administration system, and a long follow-up time including pre-HAART and HAART eras.
As limitations, ours is a retrospective cohort study. Osteoporotic fracture events were not ascertained (only ICD-9 code used – validated in other Veterans Administration studies ). BMD was not evaluated and fractures cannot be proven to be osteoporotic in nature; however, these, or similar definitions, have been used in other epidemiological studies evaluating fracture risk [5,6,24]. Finally, as our study population is overwhelmingly male, the results might not be generalizable to women with HIV disease. Also, owing to the retrospective nature of the study, the associations observed are not necessarily causative. However, these findings suggest osteoporotic fractures are an important cause of morbidity and mortality in HIV, and that whereas traditional osteoporotic fracture risk factors are the most important contributors to the risk, further attention should be placed on antiretroviral drugs.
The study was funded by Veterans Administration MERIT grant I01 CX000418–01A1.
Conflicts of interest
R.B. received research grants from Merck & Co, Tibotec Therapeutics, Bristol Myers Squibb and Abbott Pharmaceuticals. He also served as scientific advisor or consultant for EMD Serono, Merck & Co, Gilead, Tibotec Therapeutics and AIDS Arms. The other authors report no relevant conflict of interest.
1. Luther VP, Wilkin AM. HIV infection in older adults
. Clin Geriatr Med
2. Sackoff JE, Hanna DB, Pfeiffer MR, Torian LV. Causes of death among persons with AIDS in the era of highly active antiretroviral therapy: New York City
. Ann Intern Med
3. Looker AC, Wahner HW, Dunn WL, Calvo MS, Harris TB, Heyse SP, et al. Updated data on proximal femur bone mineral levels of US adults
. Osteoporos Int
4. Tebas P, Powderly WG, Claxton S, Marin D, Tantisiriwat W, Teitelbaum SL, et al. Accelerated bone mineral loss in HIV-infected patients receiving potent antiretroviral therapy
5. Triant VA, Brown TT, Lee H, Grinspoon SK. Fracture prevalence among human immunodeficiency virus (HIV)-infected versus non-HIV-infected patients in a large U.S. healthcare system
. J Clin Endocrinol Metab
6. Womack JA, Goulet JL, Gibert C, Brandt C, Chang CC, Gulanski B, et al. Increased risk of fragility fractures among HIV infected compared to uninfected male veterans
. PLoS One
7. Cazanave C, Dupon M, Lavignolle-Aurillac V, Barthe N, Lawson-Ayayi S, Mehsen N, et al. Reduced bone mineral density in HIV-infected patients: prevalence and associated factors
8. Brown TT, Qaqish RB. Antiretroviral therapy and the prevalence of osteopenia and osteoporosis: a meta-analytic review
9. McDermott AY, Terrin N, Wanke C, Skinner S, Tchetgen E, Shevitz AH. CD4+ cell count, viral load, and highly active antiretroviral therapy use are independent predictors of body composition alterations in HIV-infected adults: a longitudinal study
. Clin Infect Dis
10. Gallant JE, Staszewski S, Pozniak AL, DeJesus E, Suleiman JM, Miller MD, et al. Efficacy and safety of tenofovir DF vs. stavudine in combination therapy in antiretroviral-naive patients: a 3-year randomized trial
11. McComsey GA, Kitch D, Daar ES, Tierney C, Jahed NC, Tebas P, et al. Bone mineral density and fractures in antiretroviral-naive persons randomized to receive abacavir-lamivudine or tenofovir disoproxil fumarate-emtricitabine along with efavirenz or atazanavir-ritonavir: Aids Clinical Trials Group A5224s, a substudy of ACTG A5202
. J Infect Dis
12. Mulligan K, Glidden D, Gonzales P, Ramirez-Cardich M-E, Liu A, Namwongprom S, et al. Effects of FTC/TDF on bone mineral density in seronegative men from 4 continents: DEXA results of the Gobal iPrEx study
. In 18th Conference on Retroviruses and Opportunistic Infections
. Boston, MA; 2011.
13. Mondy K, Yarasheski K, Powderly WG, Whyte M, Claxton S, DeMarco D, et al. Longitudinal evolution of bone mineral density and bone markers in human immunodeficiency virus-infected individuals
. Clin Infect Dis
14. van Vonderen MG, Lips P, van Agtmael MA, Hassink EA, Brinkman K, Geerlings SE, et al. First line zidovudine/lamivudine/lopinavir/ritonavir leads to greater bone loss compared to nevirapine/lopinavir/ritonavir
15. Ofotokun I, Weitzmann N, Vunnava A, Sheth A, Villinger F, Zhou J, et al. HAART-induced immune reconstitution: a driving force behind bone resorption in HIV/AIDS
. In 18th Conference on Retroviruses and Opportunistic Infections
. Boston, MA; 2011.
16. van Vonderen MG, Mallon P, Murray B, Doran P, van Agtmael MA, Danner SA, et al. Changes in bone biomarkers in ARV-naïve HIV+ men randomized to NVP/LPV/r or AZT/3TC/LPV/r help explain limited loss of bone mineral density over the first 12 months after ART initiation
. In 18th Conference on Retroviruses and Opportunistic Infections
. Boston, MA; 2011.
17. Gage BF, Birman-Deych E, Radford MJ, Nilasena DS, Binder EF. Risk of osteoporotic fracture in elderly patients taking warfarin: results from the National Registry of Atrial Fibrillation 2
. Arch Intern Med
18. Womack J, Goulet J, Gibert CL, Brandt C, Mattocks K, Rimland D, et al. HIV-infection and fragility fracture risk among male veterans
. In 17th Conference on Retroviruses and Opportunistic Infections
. San Francisco, California; 2010. pp. Abstract #129.
19. El-Maouche D, Mehta SH, Sutcliffe C, Higgins Y, Torbenson MS, Moore RD, et al. Controlled HIV viral replication, not liver disease severity associated with low bone mineral density in HIV/HCV co-infection
. J Hepatol
20. Bruera D, Luna N, David DO, Bergoglio LM, Zamudio J. Decreased bone mineral density in HIV-infected patients is independent of antiretroviral therapy
21. Dolan SE, Huang JS, Killilea KM, Sullivan MP, Aliabadi N, Grinspoon S. Reduced bone density in HIV-infected women
22. Leslie WD, Bernstein CN, Leboff MS. AGA technical review on osteoporosis in hepatic disorders
23. Gallego-Rojo FJ, Gonzalez-Calvin JL, Munoz-Torres M, Mundi JL, Fernandez-Perez R, Rodrigo-Moreno D. Bone mineral density, serum insulin-like growth factor I, and bone turnover markers in viral cirrhosis
24. Hansen AB, Gerstoft J, Kronborg G, Larsen CS, Pedersen C, Pedersen G, et al. Incidence of low- and high-energy fractures in persons with and without HIV-infection: a Danish population-based cohort study
25. McComsey GA, Kendall MA, Tebas P, Swindells S, Hogg E, Alston-Smith B, et al. Alendronate with calcium and vitamin D supplementation is safe and effective for the treatment of decreased bone mineral density in HIV
26. McComsey GA, Tebas P, Shane E, Yin MT, Overton ET, Huang JS, et al. Bone disease in HIV infection: a practical review and recommendations for HIV care providers
. Clin Infect Dis
27. European AIDS Clinical Society Guidelines 2011, Version 6.0.
28. Rabenda V, Bruyere O, Reginster JY. Relationship between bone mineral density changes and risk of fractures among patients receiving calcium with or without vitamin D supplementation: a meta-regression
. Osteoporos Int