Recent studies on HIV, ART and osteoporotic fracture risk
Muirgen Stack, HIV i-Base
Two studies published earlier this year in the journal AIDS added to the accumulating data on the complex relationship between bone health, HIV and antiretroviral treatment (ART).
Tenofovir associated with increased fracture risk
In the first, Roger Bedimo and colleagues reported on the relationship between osteoporotic fracture risk and cumulative exposure to ARVs. They reported that cumulative exposure to tenofovir was independently predictive of increased risk of osteoporotic fracture (12% higher risk per year of exposure) after controlling for traditional osteoporotic risk factors and concomitant ART. 
This was a retrospective analysis from patients treated from 1988 to 2009 in the US Veterans Health Administration clinical case registry. ICD-9 diagnostic codes were used to identify osteoporotic fractures (defined as wrist, vertebral or hip fracture) after patients had been diagnosed with HIV. Cumulative ART exposure (drug or class) was defined from initial prescription to the first recorded fracture.
Multivariate analyses used two models: model 1 (MV1), controlled for age, race, tobacco use, diabetes, chronic kidney disease (CKD), hepatitis C virus (HCV) and BMI; model 2 (MV2), controlled MV1 variables and concomitant exposure to other antiretroviral drugs.
From over 56,600 patients indentified in this predominently male (98%) cohort, 39,277 (69.4%) had at least 1 month of antiretroviral therapy (ART) exposure with the total ART exposure in the cohort being 164,414 person-years (PY). A total of 951 individual patients sustained osteoporotic fractures. Multiple fractures were censored after the first event.
Patients with osteoporotic fracture had a slightly higher median age than those without (46 vs. 44 years), were more likely to be white (57% vs. 45% of those without fracture), a BMI below 20 (49% vs. 33%) and have HCV co-infection (51% vs. 31% (p<0.0001 for all comparisons).
Tenofovir exposure (46,062 PY) was associated with a yearly hazard ratio for osteoporotic fracture of 1.08 (95% CI 1.02-1.15, p<0.001). Exposure to abacavir, AZT or d4T or NNRTIs were not significantly associated with increased risk of osteoporotic fracture in univariate or multivariate models.
For the 32,439 patients who entered the cohort in the HAART era, tenofovir exposure (38,009 PY) was associated with a yearly hazard ratio (HR: 95%CI) for osteoporotic fracture of HR 1.13 (1.05-1.21, p=0.001) in MV1 and HR 1.12 (1.03-1.21, p=0.011) in MV2. Boosted protease inhibitor exposure PI/r (32,109 PY) was associated with HR 1.08 (1.01-1.15, p=0.026) in MV1 but was not significant at HR 1.05 (0.97-1.13, p= 0.237) in MV2. Exposure to abacavir, AZT, d4T or NNRTI was again not significantly associated with increased risk of osteoporotic fracture in either model.
Concomitant exposure to both tenofovir and PI/r was associated with a greater osteoporotic fracture risk (HR 1.16; 95%CI 1.04-1.30) than exposure to either tenofovir without PI/r (HR 1.11, 95%CI 1.01-1.21) or PI/r without tenofovir (HR 1.10; 95%CI 1.01-1.22).
Of the protease inhibitors, only lopinavir/ritonavir (15,319 PY) was associated with significantly increased osteoporotic fracture risk in MV1 (HR 1.13; 95%CI 1.04-1.22, p=0.005) and barely in MV2 (HR 1.09; 95%CI 1.00-1.20, p=0.051).
ART (including tenofovir) protective of fracture risk
The second study was a nested case-control study by Linda Mundy and colleagues and reported a reduced risk of fracture in HIV positive people on ART (including tenofovir). 
This was a nested case-control design in a cohort of almost 60,000 HIV positive people (approximately 25% women) enrolled from 1997 to 2008 in a US medical insurance database. ART was prescribed to 51% of patients at some point and was more common from 2003-2008 (72%) than 1998-2003 (29%). Cumulative ART exposure was again derived from prescription history. During this period, 2,411 individuals were identified with closed non-traumatic fractures according to ICD-9 codes and were matched by age and sex to 9144 HIV positive controls without fractures.
Variables included in the analysis included excess alcohol use, low physical activity, low body weight, hepatitis C virus (HCV) infection, excess steroid use and treatment for osteoporosis with bisphosphonates.
In this study, fracture risk was significantly reduced in people exposed to ART (OR 0.64, 95% CI 0.58-0.71; p<0.0001). Furthermore, reduced risk for fracture was associated with exposures to both NRTI and NNRTI drug classes, with a pattern of incremental reduction of risk with increased duration of exposure. A null effect was associated with those exposed to protease inhibitors (PI), but this effect was reduced after extended exposure of 18 months or more in a subset of patients.
Fracture risks were also reported for individual drug exposure. Reduced risk was reported for efavirenz, FTC, 3TC, tenofovir and AZT. Increased risk was reported for darunavir, delavirdine and saquinavir. After an initial increase in risk, nevirapine, ddI, nelfinavir, ritonavir and d4T were associated with a reduced risk after increasing the duration of exposure. Null or uncertain risk for fracture was associated with amprenavir, atazanavir, enfuvirtide, fosamprenavir, indinavir, lopinavir, tipranavir, and T-20 (although limited data was available for some of these drugs).
A sub-analysis of 8,879 cases enrolled in care after 2001, assessed the exposure-response relationship to abacavir and tenofovir, two drugs that have previously been associated with altering BMD levels. No statistically significant association was reported even after 12 months of cumulative exposure to either drug. Fracture risk after 12 months of exposure to tenofovir was not significant (aOR 1.08 95% CI 0.83-1.40).
In the discussion the authors note the complexity of estimating time-dependent drug-specific risks over different time periods, especially given the dynamics of bone metabolism with age. However, ART exposure, including by class and drug was generally protective of fracture risk suggesting an overall benefit of treatment. Although the number of fractures with darunavir was low, the associated was notable (aOR = 1.93, 95% CI = 1.05–3.56; global P-value = 0.043) and may warrant further study.
Comparing the studies
Both studies evaluated a similar number of patients and used the same diagnostic codes to identify reported fractures through retrospective methods. Antiretroviral use was ascertained through prescription history and both studies included fracture risk variables in their analysis.
There were differences however, as Bedimo et al, failed to include prior fracture events in their results, which is an important absence (Mundy et al found this to be significantly higher, statistically as a prediction of subsequent fracture risk). However, Bedimo et al only used reported vertebral, hip and wrist fractures (selected on the basis of their likelihood of being related to osteoporosis) whereas Mundy et al identified any non-traumatic fractures and grouped them as cases, limiting the inference that all were osteoporotic in origin. Neither study evaluated BMD so none of the reported fractures can be proven to be definitively osteoporotic.
The Bedimo et al study period was 21 years and included the distinction of ARV use during the HAART and pre HAART eras, whereas the Mundy et al study period was only 11.25 years.
Bedimo et al included age, race, tobacco use, diabetes, BMI, HCV co-infection and cumulative ARV exposure as variables. Sex was not included as 98% of the cohort was male.
Mundy et al included a much larger number of variables into their analysis; age, sex, geographic census region, year of enrollment, excess alcohol use, low physical activity, HIV-related conditions (CDC category A/B/C), prior fractures, low body weigh, lipodystrophy, hepatitis B/C virus, and prescription drug exposures (proton pump inhibitor, glucocorticosteroid excess, vitamin D/calcium, bisphosphates) against ARV drug exposure. However, race and tobacco use were included.
Mundy et al evaluated a wider variety of ARV drugs on fracture risk, including all of the main classes whereas, Bedimo et al analysed a smaller selection but crucially were able to study their effects on fracture risk over a longer period of time.
The primary result and conclusion from Bedimo et al was that tenofovir remained independently predictive of osteoporotic fracture risk (12% higher risk per year of exposure) after controlling for traditional osteoporotic risk factors and concomitant antiretroviral drug used during the HAART era.
Despite finding a statistically significant result for tenofovir use increasing fracture risk, Bedimo et al failed to find a statistically significant result on cumulative ARV use (per year of exposure) increasing fracture risk in their multivariable analysis 0.99 (0.95-1.04; p=0.77).
These findings are not necessarily contradictive, nor supportive. As both HIV infection itself and ARV use has been associated with increasing fracture risk and lowering BMD, it may be that two physiological effects are happening but not being recorded in the data. Retrospective analyses from cohort database, especially over such a long period with likely underreported of both events and lifestyle factors has well described limitations.
ART lowers the impact of the HIV infection on the body, but is used over time, as BMD reduces due to ageing. Until clear causality is established, it is difficult to weight specific risk on certain ARVs, particularly when other more established risk factors are added into the analysis and the statistical significance of each ARV on fracture risk starts to decline.
The caution on using tenofovir in patients with highest risk of fracture (previous history of fracture, osteoporosis, FRAX score) is still probably warranted. 
The START study that randomises patients in early infection (CD4 >500 cells/mm3) to either immediate or deferred (until CD4 <350) ART is already two-thirds enrolled. The bone sub-study in START, with both biomarker and DEXA results will produce a prospective dataset to help answer many of these increasingly important questions.
- Bedimo R et al. Osteoporotic fracture risk associated with cumulative exposure to tenofovir and other antiretroviral agents. AIDS: 26(7):825-31 (24 April 2012).
- Mundy L et al. Overall benefit of antiretroviral treatment on the risk of fracture in HIV: nested case control analysis in a health-insured population. AIDS 2012, Volume 26 – Issue 9 – p 1073–1082. (01 June 2012).
- Andrew Carr and Jennifer Hoy. Low bone mineral density with tenofovir: does statistically significant mean clinically significant? Clinical Infectious Diseases 2010; 51(8): 973-975.