icon-    folder.gif   Conference Reports for NATAP  
  19th Conference on Retroviruses and
Opportunistic Infections
Seattle, WA March 5 - 8, 2012
Back grey_arrow_rt.gif
CROI Report: Bones, Frailty, and Vitamin D
  Todd Brown, MD, PhD
Associate Professor of Medicine
Division of Endocrinology and Metabolism
Johns Hopkins University

Investigation into bone disease, frailty, and vitamin D continues to gather steam in the HIV world. While there were no major breakthroughs at the 19th CROI, there were multiple studies presented which improve our understanding of the epidemiology, pathogenesis and optimal management of these common problems. Mike Yin from Columbia University gave a superb talk summarizing our current understanding of osteoporosis in HIV-infected persons. I encourage interested readers to click on the following link to view the webcast (http://retroconference.org/static/webcasts/2012/, scroll to bottom). Here, I'll try to summarize some of the relevant studies presented in Seattle, try to put them in the context of our current knowledge, and provide implications for the care of the HIV-infected patients.

a couple of highlights from this interesting report:

"SUN study....found that 38% were pre-frail and 5% were frail with the most common criteria being exhaustion and physical inactivity. .....associated with non-white race and higher depression scores.....age in this cohort was 47 years injection drug-users with a mean age of 48 years, the overall prevalence of frailty was 9% (12% in the 351 HIV-infected persons and 7.6% in the 879 HIV-uninfected persons.....associated with all-cause mortality with HIV- infected, frail persons having a an 8 fold increased risk in mortality .......in this study with average age of 52, the Low Functioning HIV+ on ART had lower bone mineral density and lower lean mass than the High Functioning group HIV+ on ART....As highlighted in the previous study, low muscle mass is central to the concept of frailty. Guaraldi quantified muscle mass in 1877 HIV-infected patients....vs.....normative data for the general Italian population......low muscle mass......associated with mortality, independent of age and sex.

The optimal dosing schedule for vitamin D replacement is not clear and many strategies have been tried.....In a single arm study of vitamin D replacement (50,000 IU of D3 twice weekly for 5 weeks followed by 2000 IU D3 daily for 7 weeks) from monotonously sunny Los Angeles in 106 HIV infected patients (median 25D level of 28 ng/mL), 25OH D increased by a median of 23.5 ng/mL (IQR 16, 31), with a non-significant trend towards decreasing fasting glucose, but no change in other parameters of glucose metabolism"


Effect of Antiretroviral Therapy on Bone:

Tenofovir: Tenofovir disoproxil fumarate (TDF) decreases bone mineral density (BMD) by about 1-2% compared to other ART agents in randomized trials and a recent study by Bedimo (AIDS, 2012) in the Veterans Administration System showed that TDF was associated with an increased risk of fracture compared to other agents. Most of the data to date regarding TDF's effect on BMD have been derived from studies of ART initiation, where specific ART effects need to be disentangled from the effects of immune reconstitution on bone (discussed below). Relatively few studies have examined the effect of TDF on patients who are virologically suppressed.

Aoife Cotter (125LB) presented a clinical trial in which 54 virologically suppressed subjects were randomized to continue their AZT/3TC-containing regimen or switch to TDF/FTC (called the PREPARE study). In those who remained on AZT/3TC, BMD at the femoral neck did not change over 48 weeks, whereas in those who switched to TDF, BMD decreased by a median of 2% in the lumbar spine (p=0.01 vs AZT/3TC) and 1.5% in the femoral neck (p=0.16 vs AZT/3TC). This decrease in BMD was associated with an acceleration in bone turnover as measured by serum markers of bone resorption and bone formation. These findings are consistent with the STEAL study (Martin, Clinical Infectious Diseases, 2009) showing a negative effect of TDF on BMD compared to ABC/3TC in patients with virologic suppression. What was remarkable in the PREPARE study (author's note: this is a study name with UAM (Unknown/Unstated Acronym Meaning)) is that the comparator strategy (AZT) has also been linked to bone loss both in vitro (Pan, AIDS Res Hum Retroviruses, 2004) and in a study investigating ART initiation (van Vonderen, AIDS, 2009). Presumably, the patients enrolled in PREPARE already lost bone with the initiation of an AZT-containing regimen and those who switched to TDF/FTC lost additional bone.

The effect of TDF on bone is heterogeneous and in the future it is going to be important to be able to identify persons who will go on to develop more substantial bone loss with TDF, either through clinical and demographic factors, genetic markers, or biomarkers. Given the fact that bone loss with TDF is associated with rapid changes in markers of turnover, in PREPARE and other studies, it is conceivable that early changes in these markers could eventually be used for this purpose. However, the variability of these markers (both analytic and biological) has been a barrier to their use in the clinical assessment and management of osteoporosis in an individual patient, but this is an important area of research with real potential clinical application.

Raltegravir: If TDF is bad for the bone, are there alternatives for those at higher baseline risk of fracture? The TROP study (author's note: another UAM title) (#878) was a single arm trial which investigated the effect of substituting raltegravir (RAL) for TDF in 37 HIV-infected persons (97% male) with viral suppression and osteopenia/osteoporosis. Over 48 weeks, BMD increased by 3% in the spine and 2.5-2.7% in the total hip accompanied by reductions in markers of bone turnover. These interesting data suggest that RAL may have either neutral or salutatory effects on BMD and support the findings of the SPIRAL study (author's note: yet another UAM study) in which total BMD, hip BMD, and femoral neck BMD increased in those who switched from a boosted PI (PI/r) to RAL, with no change in those who continued PI/r (Curran, AIDS, 2012), as well as the PROGRESS study (PROtease/InteGRasE Simplification Study) in which ART-naïve patients initiating LPV/r+RAL had no significant loss of total body BMD and significantly smaller reduction in derived spine BMD, compared to those who received LPV/r+TDF/FTC ( Qaqish, IAS, 2011, TULBPE021).

In the SPIRAL study, switching to RAL was associated with decreases in markers of inflammation (#834). Whether the apparent salutary effects of RAL on BMD are mediated by changes in systemic inflammation is unclear. Further mechanistic studies are needed to understand the effect of RAL on bone and further randomized clinical trials, following up on the TROP study is needed to confirm the efficacy of a switch to RAL in those with low BMD. Also, as new integrase inhibitors make their way into clinical practice, it will be important to evaluate their effects on bone.

Immune Activation and Bone:

One of the major hypotheses as to why BMD in HIV-infected persons may be compromised is that chronic inflammation and immune activation increase osteoclast activity and decrease osteoblast activity, leading to net bone resorption and lower BMD. Indeed, other inflammatory diseases, such as rheumatoid arthritis, systemic inflammation is associated with lower BMD, which can be improved with anti-TNF therapy. Some studies regarding immune activation and bone were presented. Gazzola (#879) examined cellular markers of immune activation and senescence in 62 HIV-infected persons (25% ART-naïve, 75% on stable HAART) and found that those with osteopenia/osteoporosis at the spine or femoral neck had higher HLA-DR expression on CD4 and CD8 cells compared to those with normal BMD, but CD38 expression or CD38/DR co-expression was similar between groups (from Jules: HLR-DR & CD38 are both markers of activation). However, somewhat inexplicably, those with osteopenia/osteoporosis had a lower frequency of CD28- CD4 and CD8 cells (from jules: senescence markers), suggestive of less immune senescence. These associations held up after multivariable adjustment. Given the small sample size, the cross-sectional nature, and the dichotomous characterization of BMD around a cutpoint with unclear clinical significance (i.e T-score< -1), clearly further studies are needed understand these findings and put them into clinical context.

In the baseline analysis for the A5260s (a substudy of ACTG A5257 which compares the effects of RAL vs ATV/r vs DRV/r in persons initiating ART with TDF/FTC), we found no association between CD38 or HLA-DR expression on CD4 or CD8 cells and spine or hip BMD in 330 ART-naïve persons about to start ART (#875). As the field moves forward, it will be important to attempt to standardize the protocols for measuring these cellular markers, as there appears to be substantial between-lab variability.

Untreated HIV: Less Bad to the Bone than Expected?

With high levels of immune activation and markers of systemic inflammation, it would be reasonable to hypothesize that untreated HIV would be associated with a major decline of bone mineral density. However, the SMART study (Grund, AIDS, 2010) suggested that this might not be the case. In this study, those who were randomized to treatment interruption had an attenuated decrease in BMD compared to those who continued ART. Although there is currently limited longitudinal BMD data in untreated HIV-infected persons, two cross-sectional studies presented at CROI support the idea that untreated ART may not have major adverse effects on bone. Hileman (#880) found no difference in BMD between 43 untreated HIV-infected persons (median known HIV duration 3.5 years) and 35 age-, sex-, and race-matched controls. Similarly, the 10% prevalence of low BMD (Z-score < -2) in the baseline data for A5260s (#875) was identical for those reported in the iPrEX study (Liu, PLOSone, 2011), a population that is younger, but demographically similar. Clearly, longitudinal studies are needed in untreated HIV-infected persons.

If untreated HIV is not bad to the bone, why not? One possibility is that adequate T-cell numbers and function are required for bone resorption. Indeed, T-cell nude mice who received ovariectomy do not experience estrogen dependent bone loss (see Pacifici, Bone, 2010 for a good review) and as Igho Ofotukun presented as last year's CROI, in an animal model of immune reconstitution, T cells transferred into T-cell knockout mice led to marked increases in bone turnover and profound decreases in BMD within 12 weeks of T-cell transfer. Taken together, these findings suggest that fewer or dysfunctional T-cells in untreated HIV may serve to protect BMD, until the immune system is reconstituted. Another potential mechanism for BMD protection during untreated HIV is the osteoprotegerin (OPG)/RANKL ligand system. These osteoblast-secreted factors are best known for playing a major role in the coupling of bone formation and resorption. RANKL that is secreted by osteoblasts binds to RANK on the cell-surface of osteoclast progenitors leading to osteoclast differentiation and activation, and thus bone resorption. OPG is also secreted by osteoblasts and binds to RANKL preventing RANK-binding and osteoclast activation. This mechanism is exploited for the newest osteoporosis drug, denosumab, which is a monoclonal antibody to RANKL that causes profound decrease in bone resorption by essentially acting like OPG.

OPG is also secreted by activated immune cells and previous studies have shown that OPG concentrations decrease after ART-initiation (Brown, Antiviral Therapy, 2011), suggesting that along with its effect on immune activation, ART also decreases OPG concentrations. Indeed, in a case control study of HIV-infected persons who were ART-naïve, ART-treated with a low CD4 cell count, and ART-treated with a high CD4 cell count (Bandiera, #874), OPG concentrations, as well as markers of immune activation (CD38+ on CD4 and CD8 cells) were higher in the untreated group compared to the treated groups. In the A5260s substudy, higher OPG concentrations in untreated HIV-infected persons were significantly associated with higher BMD, suggesting a protective effect of OPG on BMD. Further work is clearly needed to understand the OPG/RANKL system in HIV-infected persons with and without treatment and the effect of treatment on these parameters and its relationship to BMD and fractures.


Frailty is defined as a "clinical state of vulnerability to stressors resulting from aging-associated declines in resiliency and physiologic reserves and a progressive decline in the ability to maintain a stable homeostasis" (Fried, 2005). It is a critical concept in the geriatrics world, in that early identification of frailty and knowledge of its underlying pathogenesis could in theory prevent disability. It has been hypothesized that for a given age the prevalence of frailty would be higher among HIV-infected persons compared to HIV-uninfected control populations. Several abstracts at the meeting addressed this issue.

Although there are different definitions of frailty, one common definition uses 5 different criteria: 1) unintentional weight loss, 2) physical inactivity, 3) exhaustion, 4) weakness (as measured by grip strength), and 5) slowness (as measured by walk speed). Impairments in 3 or more of these criteria is considered "frail", whereas 1 or 2 of these criteria is considered "pre-frail". In the geriatric literature, the presence of frailty has been associated with multiple negative health outcomes, including falls, fractures, hospitalization, functional disability, and death.

Studies in HIV-infected populations investigating the prevalence and correlates of frailty are beginning to emerge. The SUN study (Study to Understand the Natural History of HIV/AIDS in the Era of Effective Therapy) (#858) determined the prevalence of frailty in 308 HIV-infected persons and found that 38% were pre-frail and 5% were frail with the most common criteria being exhaustion and physical inactivity. In a multivariate analysis, the presence of these abnormalities was associated with non-white race and higher depression scores. Given that the frailty criteria were validated in much older populations, it is unclear if pre-frailty and frailty have the same clinical and prognostic meaning in younger population. The median age in this cohort was 47 years with 50% of the values between 41 and 53 years.

The prevalence of frailty can vary with the population studied. In the ALIVE (AIDS Linked to the Intravenous Experience) study , a Baltimore-based cohort of injection drug-users with a mean age of 48 years, the overall prevalence of frailty was 9% (12% in the 351 HIV-infected persons and 7.6% in the 879 HIV-uninfected persons (adjusted odds ratio HIV+ vs HIV-, 1.73, 95% confidence interval, 1.28, 2.33). Over 3.5 years of follow-up both HIV-status and the presence of frailty were associated with all-cause mortality with HIV-infected, frail persons having a an 8 fold increased risk in mortality (adjusted hazard ratio (aHR) 8.14, 95% confidence interval (4.09,16.2) compared to the HIV-uninfected, non-frail persons, confirming the relevance of the frailty classification in this relatively young cohort and its interaction with HIV-status.

Erlandson (#860) used a combination of the frailty criteria defined above and performance of the widely used Short Physical Performance Battery (SPPB) to define a group as low functioning (actual definition unclear)(n=31) and compared them to matched high functioning HIV-infected controls (n=49). Both cases and controls were receiving antiretroviral therapy. While the measures of overall adiposity and fat distribution were similar in the Low Functioning and High Functioning groups, the Low Functioning group had lower bone mineral density and lower lean mass than the High Functioning group, associated with a lower IGF-1, suggesting specific dysfunction in the GH/IGF-1 axis. What will be particularly challenging in future studies will be to try tease apart the causes and consequences of low physical performance, as these may be interdependent (eg low physical performance leads to lower muscle mass which leads to worsened physical performance and so on). Longitudinal studies are clearly needed to address this issue.

As highlighted in the previous study, low muscle mass is central to the concept of frailty. Guaraldi (#861) quantified muscle mass in 1877 HIV-infected patients attending a single clinic in Italy using the DXA-derived, fat free mass index (FFM/height2). Similar to body mass index, this gives an approximation of muscle mass normalized for height. Using normative data for the general Italian population, they were also able to calculated FFM index Z-scores (i.e compared to an age- matched population) and T-scores (compared to a young normal population, ~30 years). Although the numbers of events was small, a lower FFMi and having a FFMi Z-score < -2 were each associated with mortality, independent of age and sex. The effects of other variables (egs CD4 cell count, presence of other co-morbidities, etc) were not discussed.

Vitamin D:

Vitamin D continues to be a hot topic and several posters addressed issues relating to prevalence and risk factors for deficiency, associations with outcomes, and effect of vitamin D replacement.

Two studies assessed the prevalence and risk factors of vitamin D deficiency in very different parts of the world. In cross-sectional study of 483 patients from a clinic in New York City (#881), 59% had 25OHD levels < 20 ng/mL (low vitamin D levels), whereas 24% had levels between 20-30 ng/mL (vitamin D insufficiency). In addition, to hyperparathyroidism, vitamin D deficiency was associated with efavirenz use (aOR 3.1, 95% CI: 1.3, 7.1). On the other side of the world, remarkably similar results were found. In a Thai cohort (#882) of 673 HIV-infected patients, 30% had 25OHD levels < 20 ng/mL, whereas 40.6% had levels between 20-30 ng/mL. Again, EFV use was associated with vitamin D deficiency.

Vitamin D and Non-Skeletal Outcomes:

In the Women's Interagency HIV Study (WIHS), vitamin D deficiency was associated with CD4 recovery at 24 weeks in women initiating ART, independent of race, viral load response, BMI, and other factors (#885). In the PEARLS study (Prospective Evaluation of Antiretrovirals in Resource Limited Settings)(#886), a study of ART initiation in 9 countries and four continents, the prevalence of low vitamin D levels ranges from 23% in Brazil to 78% in Thailand. Low vitamin D levels were associated with clinical HIV progression and virologic failure, independent of other factors. While both of these studies are provocative, it still remains possible that vitamin D deficiency is a marker of poor health, rather than a causal factor. Vitamin D replacement studies are clearly needed to establish vitamin D as the cause of these adverse HIV outcomes.

Is the vitamin D effect in the genes?: In a pediatric ACTG treatment study assessing the efficacy of combination NRTI in children (median age 2.3 years), certain polymorphisms related to viamin D (Bsm-1 and DHCR7) were independently associated with HIV-disease progression (#996) .

It is possible that some of the effect of vitamin D deficiency may be mediated through enhanced systemic inflammation. In a cross-sectional study of 204 white and 151 black ART-naïve HIV-infected patients (#883), lower 25D levels (lowest tertile) were associated with higher hsCRP and soluble TNF receptor 2 (but not TNFR2, TNF-alpha, or IL-6). There appeared to be a differential association by race between 25D and post-glucose load insulin, LDL cholesterol, triglycerides, with white patients with lower vitamin D levels showing higher post-glucose load insulin, lower LDL and higher triglycerides, and black patients showing no association.

Examining associations between vitamin D and various outcomes stratified by race is an important. While 25D levels are lower in blacks than whites in both HIV-infected and uninfected populations, associations with vitamin D levels and more clinical endpoints may differ by race. For example, higher risk of fracture was recently associated with higher risk of fracture in African American women in the Women's Health Initiative (Cauley, JBMR, 2011), while the opposite was seen in white women. Similarly, in a racially-diverse, community based cohort of men in Boston, lower vitamin D levels were associated with lower BMD in white men, but not African-American men (Hannan, JCEM, 2008). Vitamin D screening and treatment recommendations, however, such as the Endocrine Society, do not separate recommendations by race. Clearly, further work both in HIV-infected and uninfected populations is needed to understand differential effects of vitamin D deficiency by race.

Vitamin D Replacement:

The optimal dosing schedule for vitamin D replacement is not clear and many strategies have been tried. Some practitioners do a load and maintain approach (high dose for a limited period of time, followed by lower regular doses to maintain levels in target range), others try daily dosing (used in many replacement studies in older population), and still others tailor the replacement depending on the baseline vitamin D concentration. Several studies focused on vitamin D replacement.

In a single arm study of vitamin D replacement (50,000 IU of D3 twice weekly for 5 weeks followed by 2000 IU D3 daily for 7 weeks) from monotonously sunny Los Angeles in 106 HIV infected patients (median 25D level of 28 ng/mL), 25OH D increased by a median of 23.5 ng/mL (IQR 16, 31), with a non-significant trend towards decreasing fasting glucose, but no change in other parameters of glucose metabolism, although, interestingly, adiponectin decreased. While this dosing regimen was clearly efficacious in improving 25 D concentrations in almost all participants, there was significant variability in the 25D response. From a previous presentation of this study at the 13th Adverse Drug Reaction and Co-morbidities in HIV in Rome (PO51), factors associated with successful repletion included white race, non-smoking status, older age, and baseline 25OHD. ART regimen, including efavirenz exposure was not associated with the treatment response.

Another supplementation study from France (#887) dosed vitamin D3 according to the baseline 25D level (10-30 ng/mL, 100,000 units monthly x 4 months (4 doses total);< 10 ng/mL, 100,000 units every 2 weeks x 4, then 100,000 units monthly for two months (6 doses total)). Lower vitamin D levels at follow-up were associated with non-Caucasian race, zidovudine use (but not efavirenz use).

In a separate French study, population pharmacokinetic studies were carried out on 135 HIV-infected patients receiving vitamin D supplementation (average dose 63,300 units per month -the actual dosing schedule was not specified) over 12 months. It was concluded that endogenous production of 25D was 20% lower in non-white patients and lower in the non-summer months. Exposures to various ART regimens were not associated with Vitamin D response. From the modeling, they concluded that 100,000 units per month is optimal for "safety and efficacy". The caveat of course is that it's difficult to conclude about safety or efficacy without specifically examining clinically related outcomes. While it's reassuring that high intermittent dosing increases 25D and does not lead to hypercalcemia, this is not the same as saying that this replacement decreases the risk of falls, fractures, BMD, HIV progression, etc. The reader should keep in mind that 500,000 units of D3 given to older women (non-HIV-infected) also increased 25D into the "target range", but also was associated with an increased risk falls and fractures (Sanders, JAMA, 2010). There is much that we need to learn about optimal vitamin D replacement.