icon-    folder.gif   Conference Reports for NATAP  
  Conference on Retroviruses
and Opportunistic Infections (CROI)
February 13-16, 2017, Seattle WA
Back grey_arrow_rt.gif
Frailty, Bones, and Vitamin D: CROI 2017
  Todd T. Brown, MD, PhD
Professor of Medicine and Epidemiology
Johns Hopkins University

CROI 2017 saw a building of momentum for studies related frailty and osteoporosis. Not only has the quantity of studies increased, but the quality has jumped too. Vitamin D-related studies also made a good showing this year at CROI, moving beyond small observational studies to examining the effects of vitamin D supplementation in randomized clinical trials. As in previous years, my focus will be on studies which have the most clinical relevance and potential for translation into clinical practice.
What is TBS and why should I care about it?
Conventional dual-energy x-ray absorptiometry (DXA) is commonly used in clinical practice to assess fracture risk and to determine whether intervention is needed. It's a relatively cheap test with little radiation exposure, and lower bone mineral density (BMD) has been consistently associated with fracture risk in multiple populations, including those with HIV. However, BMD by DXA only explains about 50% of fracture risk. Some of risk that DXA does not explain may be related to non-skeletal factors such as falls, but some is related to aspects of bone that are not captured by conventional DXA, such as bone quality.
One important aspect of bone quality is the bone microarchitecture. This can be assessed with specialized techniques, such as high resolution CT, but because of expense and radiation exposure, clinical utility has been limited. Another technique to assess bone microarchitecture in the trabecular spine is with the trabecular bone score (TBS). TBS uses the image obtained from a conventional spine DXA to estimate characteristics of trabecular bone microarchitecture, including trabecular number, trabecular separation, and trabecular connectivity. Basically, in the digitized image, the grey-level texture of a given pixel is compared to adjacent pixels. If the patient's vertebral body has lots of intact trabeculae that are well connected with small spaces between trabeculae, the TBS (a unitless measure) will be high. Bone with poor trabecular microarchitecture will have a low TBS. Normal is considered ≥ 1.35; Degraded ≤ 1.20 and in between is Intermediate. The important thing is that the TBS gives information about fracture risk that is independent of conventional BMD and thus can make risk estimation more precise. For this reason, the FRAX score (an algorithm to estimate a persons 10 y risk for fracture, similar to the Framingham Risk Score for CVD) can be done with TBS to refine the 10 year risk of fracture. If TBS software is installed on the DXA machine, the TBS score will be generated without any additional burden to the patient. The TBS software can also be used to generate the TBS score retrospectively on DXA images that have already been obtained. (For more on TBS, the reader is referred to two excellent reviews: http://www.sciencedirect.com/science/article/pii/S8756328215001957; http://link.springer.com/article/10.1007/s12018-016-9203-7.
There were several studies at CROI which used TBS to help answer different questions.
• TBS can improve fracture risk prediction in HIV-infected women (#132): This study examined 900 HIV-infected and 248 HIV-uninfected women enrolled in the WIHS cohort and followed over a 10 year period. Based on their risk factors for fracture at baseline, the investigators calculated the expected risk of fracture using the FRAX equation (https://www.shef.ac.uk/FRAX/). They then compared this to the number of fractures that were observed during the same period and calculated the observed/estimated (O/E) ratio. If FRAX predicted fractures perfectly, the O/E would be 1. If the O/E >>>1, the actual risk of fracture is being underestimated by the risk prediction tool. In the WIHS, the O/E was 17 when using FRAX with the clinical risk factors alone. This is big league (or is it bigly?) underestimation. When the "secondary osteoporosis" box was checked in FRAX, the O/E came down to 14 ( a little bit of an improvement). This means that checking "secondary causes" for an HIV-infected person will give a better estimate of true fracture risk. (It should be noted that the "secondary cause" button only changes the estimate when FRAX is used without BMD).
Now here's the big take home message from the study: When the subset with DXA data were analyzed (∼25%, a limitation of the study), the O/E ratio came down to 2.74 when BMD was added to the FRAX score. This suggests that adding BMD really helped risk prediction in this population (and that FRAX without using BMD grossly underestimates fracture risk). The authors then recalculated FRAX with both the BMD and the TBS and the O/E came down to 1.32. This suggests that poor bone microarchitecture is significantly contributing to fracture risk in this population, independent of BMD. If available, TBS along with BMD may be very useful to help guide treatment decisions.
CROI: Bone density and trabecular bone score improve fracture prediction in HIV-infected women - (03/06/17)
• Bone Microarchitecture is Compromised in HCV infected Persons (#677):
Both HIV infection and HCV infection have been associated with a higher risk of osteoporosis and fracture. This study examined conventional DXA and TBS in four groups of patients from the Veteran's Administration Hospital: HIV/HCV co-infected (n=57), HCV-mono-infected (n=123), HIV mono-infected (n=174), and uninfected (n=178). In the multivariable analysis, positive HIV status and HCV status were both independently associated with lower BMD at the total hip and the femoral neck. Interestingly, treated HCV was associated with higher BMD suggesting that HCV treatment may have salutary effect on bone. The associations at the spine were quite different. Neither HCV nor HIV status were associated with lower lumbar spine BMD. However, HCV infection (but not HIV) was associated with lower TBS. Taken together, this may suggest that HCV has both detrimental effects on bone density and trabecular microarchitecture. The big caveat here, however, is that this is a cross-sectional study and it's impossible to know whether the effect of HCV is due to HCV per se or an unmeasured confounder that is associated with HCV (eg, prior IV drug use, poor nutrition, etc). Examining TBS after HCV cure will help to determine whether HCV is playing a causal role in microarchitectural deterioration seen with HCV.
CROI: Bone Micro-Architectural Changes and Fracture Risk Prediction in HIV and HCV - (03/06/17)
• TDF is not only Bad for BMD, but it is Bad for Trabecular Microarchitecture too (#681): In multiple studies in many different populations, ART initiation with TDF leads to decreases in BMD. In this single center, single arm study, 37 ART-naïve patients received a regimen of TDF/FTC/DRV/r. As expected, BMD at both the spine and the hip dropped over 48 weeks. In addition, TBS decreased 2.5% from 1.4 to 1.32. This is first investigation that I'm aware of examining TBS with ART initiation. Not surprisingly, ART initiation with TDF/FTC/DRV/r led to trabecular microarchitecture deterioration and may be a mechanisms that leads to fracture in the first few years after ART initiation (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3652631/). Another novel feature of this study was the use of microindentation, which measures an important aspect of bone quality: bone material strength (i.e, how resistant the bone tissue is to fracture). Microindentation is a minimally invasive technique to assess bone material strength in vivo. After numbing up the skin and periosteum over lying the anterior tibia with local anesthesia, the investigator inserts the probe into the skin, through the periosteum, and rests it on the bone. The device is the triggered to make a very small microindentation into the bone. The deeper the probe goes into the bone, the poorer the bone quality and the lower the bone material strength index (BMSi). It may sound like a tough sell to IRBs and potential participants, but the indentation is extremely small (∼300 microns) and the participant doesn't feel much. When I had it done on myself, the most painful thing was pulling the Band-Aid off my hirsute leg. This study is the only longitudinal study using microindentation in HIV that I'm aware of. The results were a little confusing. BMSi stayed relatively flat until W24 when it INCREASED, suggesting improved bone material strength from W24 to W48. Clearly, this needs to be replicated, but it may suggest that there may be some compensatory improvement in bone material properties, as other aspects of bone strength are taking a hit. If ever a time the "further studies are needed" stock phrase were appropriate, this would be it. Here's a nice review on microindentation (https://www.ncbi.nlm.nih.gov/pubmed/27975078).
- (03/08/17)
Interventions for Bone Health:
• Getting off of TDF: TDF can be bad to the bone. One strategy to improve bone health in HIV-infected person is to switch off of TDF. We have seen multiple studies at the last few CROIs re: TAF. A switch to TAF from TDF generally improves BMD by about 2%. There were two studies that examined the switch in specialized populations.
• Swtich to TAF in Women (#443). In this analysis, 212 women who had initiated ART with TDF/FTC/ATV/r and were successfully treated for at least 48 weeks were randomized to either continue this regimen or switch to TAF/FTC/Elvitegravir/cobicistat. Over 48 weeks, lumbar spine BMD increased by 2.8% in the TAF group and stayed flat in the TDF group (p<0.001). At the hip, BMD increased by 2.1% in the TAF group, but also increased in the TDF group 1.3%. The reasons for the increase at the hip in the TDF group was unexpected and unexplained, but overall these data suggest that there is substantial BMD gain in women who switch from TDF to TAF. The data are unique in that most studies in HIV-infected persons examining bone are predominately men. Since osteoporosis is such an important condition for women's health, these data are an important contribution.
• Switch to TAF in Those with Low BMD: From a clinical perspective, the patients in whom a switch off of TDF may have the largest impact are those who have low BMD to begin with. We looked at two Gilead studies in which TDF-treated persons were switched to TAF and focus on those 214 persons who had a T-score < -2 at the spine, hip, or femoral neck. Over 96 weeks, the median increase in lumbar spine BMD was 2.54% and 2.34% at the total hip, corresponding to increases in T score of 0.19 and 0.14, respectively. Of the 86 participants who had osteoporosis at baseline and follow-up DXA scan, 23% no longer had BMD in the osteoporotic range at the 96 week follow up visit. This study was instructive when thinking about bisphosphonates in patients with low BMD who have low absolute risk of fracture (no previous fracture, younger age, etc). In these patients, making a switch off of TDF is a good option. In this way, the use of bisphosphonates can possibly be deferred or avoided.
• Switch to Rilpivirine/dolutegravir: In the SWORD study, 1024 ART-treated persons were randomized to either to switch to RPV/DTG or continue on their ART regimen, 70% of whom were taking TDF. While the regimens were equivalent in their virologic effects, those who switched to RPV/DTG had significant decreases in bone turnover markers, consistent with a salutary effect. The authors stated that the BMD data will be presented at future meetings, but the preliminary suggestion is that a switch to RPV/DTG may be a reasonable alternative to TDF to improve bone health.
• Zoledronic Acid to Eliminate Bone Loss at ART initiation (#684): At last year's CROI, Igho Ofotokun showed that a single dose of pre-ART zoledronic acid could eliminate the bone loss over 48 weeks. This year, the 144 week data were presented showing that the effect persisted. Spine BMD remained about 5% lower than baseline in the placebo group but remained flat in the ZOL group. It was reported that similar effects were seen at the hip. These data provide rationale for using single dose zoledronic acid prior to ART initiation to preserve BMD. Personally, I use this strategy for people starting ART who have other risks for fracture (age, previous fracture, low BMD). Since the effects of this strategy last at least 3 years, I know I can wait until then to do a repeat DXA and think about additional treatment.
CROI: Switching from TDF To TAF in HIV-Infected Adults With Low BMD: A Pooled Analysis - (04/05/17)
CROI: Efficacy and Safety of Switching to E/C/F/TAF in Virologically Suppressed Women - (02/26/17)
CROI: SWORD 1 & 2: Switch to DTG + RPV Maintains Virologic Suppression Through 48 Weeks, a Phase III Study - (02/16/17)
CROI: Antiretroviral-induced Bone Loss is Durably Suppressed by A Single Dose of Zoledronic Acid - (03/29/17)
• Falls more common in HIV-infected persons (#930):
Falls are a very important geriatric event which has huge implications for morbidity and mortality. HIV-infected persons have a high prevalence for many risk factors for falls including cognitive dysfunction, substance abuse, polypharmacy, peripheral neuropathy, and low muscle mass, among others. Julie Womack and colleagues in the VACS reported on incident falls in the VA population by HIV serostatus. Since we all fall now and again, falls can be a difficult event to capture. In this study, they focused on medically relevant falls (falls for which medical attention was sought), and captured these by using a machine learning algorithm that scoured the VA medical record (radiology and progress notes) to identify falls. In this way, they were able to have data on > 130,000 people between 1996 and 2009. What they found was that overall, the rate of falls was higher in the the HIV-infected persons compared to the HIV-uninfected persons. In their multivariable analysis, an interesting pattern appeared in those < 50, the HIV-infected persons actually had a lower risk of falls compared to the HIV-uninfected persons. However, for those ≥50 years, the exact opposite was seen: HIV-infected persons had a higher risk of falls vs the HIV-uninfected. These two groups were not compared directly in the analysis presented in the poster, but were compared relative to the HIV-uninfected group < 50 years. The odds ratio was 1.18 in the HIV-infected ≥ 50 vs HIV-uninfected< 50 and 1.13 in the HIV-uninfected ≥ 50 vs the HIV-uninfected <50. It was reported that there was an HIV x Age interaction. This means that in older people, but not younger people, HIV-status was associated with a higher risk of falls. This is the first large study of falls in HIV-infected persons and has important implications. Fall risk reduction with attention to the home environment, minimization of external factors leading to falls (polypharmacy, overtreatment for hypertension, substance abuse, etc), and strength and balance training are going to be critical to improve health in the aging HIV population.
• Frailty associated with falls in HAILO (#666): HAILO (The HIV infection, Aging, and Immune Function Long-term Observational Study) is a longitudinal study of older HIV-infected person who have participated in the previous ACTG trials. In this analysis, the HAILO investigators examined whether frailty and pre-frailty as assessed by the Fried criteria (https://www.ncbi.nlm.nih.gov/pubmed/11253156) was related to falls. At baseline, 6% of the population (59 of 967) were categorized as frail, whereas 39% were pre-frail. Overall, 18% reported at least 1 fall in a 48 week period and 7% reported multiple falls. Compared to the non-frail people, frail people were marked more likely to have recurrent falls (OR 17). Pre-frail people were also more likely to have recurrent falls (OR 3.8). Interestingly, the effect of frailty status on fall risk was observed only in the 40% of the population who also had peripheral neuropathy. These findings suggest that those with peripheral neuropathy who may be experiencing decreases in strength, energy, and vitality are an important population on whom to intervene to prevent falls.
• Frailty predicts poor outcomes better than neurocognitive impairment (#665): In another HAILO analysis, investigators sought to determine whether neurocognitive impairment (NCI), frailty, or the combination was a better predictor of falls, disability, and mortality. In this population, 2% were both frail and had NCI, 4% were frail without NCI, and 14% had NCI without frailty. While both frailty and NCI were associated with falls/disability/mortality, the associations with frailty (with or without NCI) and the composite outcome were stronger than with NCI, suggesting that frailty may be the driving factor behind these outcomes.
• Cognitive complaints associated with falls in WIHS (#361): Investigators in the WIHS examined whether subjective cognitive complaints (problems with memory/concentration or confusion/inability to perform mental tasks) were associated with falls in HIV-infected and HIV-uninfected women. The average age of the women was ∼48 years and ∼11% had cognitive complaints with no difference by HIV serostatus. HIV-infected and HIV-infected women reported a similar frequency of falls. In multivariable analysis, cognitive complaints were strongly associated with falls in both HIV-infected and HIV-uninfected women. However, in the HIV-infected women, the association between cognitive complaints and falls was lost after adjustment for comorbidities, suggesting a confounding or mediating role. It could be that controlling or preventing comorbidities may help to minimize cognitive problems and this could in turn help to reduce falls. Clearly, these links need to be explored further.
• Frailty Progression and Frailty Recovery in HIV-infected and HIV-uninfected Injection Drug Users (#133): Frailty is not necessarily a one-way street, but should rather be considered a dynamic state. This is important since frailty can be viewed as an target for intervention and it is indeed possible to reverse frailty and perhaps reduce the risk of downstream adverse events. In the ALIVE cohort, a Baltimore-based cohort of injection drug users, Damani Piggott investigated factors associated with frailty progression and frailty recovery from one semi-annual visit to the next. He found that increasing number of co-morbidities and HIV-status were associated with increased odds of frailty progression and a decreased odds of frailty recovery. Drilling down more to explore the effect of HIV disease severity, he found that HIV-infected persons who had undetectable viremia, no prior AIDS, or CD4 cell count > 500, were similar to HIV-uninfected persons in their lower odds of frailty progression and higher odds of frailty recovery. Also, those with lower systemic inflammation also had lower risk of frailty progression and a higher risk of recovery. Taken together, these findings suggest that uncontrolled HIV infection and poor immunologic status are the most important drivers of frailty progression and inability to revert once frailty is present and underlines the importance of continued engagement with HIV treatment.
CROI: Frailty Status and Risk of Falls in HIV-Infected Older Adults in the ACTG A5322 Study - (03/29/17)
CROI: Frailty Progression and Recovery among Persons Aging with HIV and Substance Use - (02/18/17)
CROI: Frailty Stronger Than Neurocognitive Impairment in Predicting Falls-Disability-Death - (02/26/17)
Vitamin D:
There were several important randomized clinical trials of vitamin D supplementation at this year's CROI and one observational study of note.
• High dose vitamin D increases spine BMD in HIV-infected young adults (#685LB): In an Adolescent Trials Network Study, 214 young HIV-infected adults treated with a TDF-containing regimen were randomized to receive either either monthly D3 (50000 units) + daily multivitamin with 400 IU of D3 or a monthly placebo + daily multivitamin. Over 48 weeks, the 25OHD went from a median of 16 ng/mL at baseline to 37 ng/mL in the high dose vitamin D group and 21 ng/mL in the low dose group. Spine BMD increased by 1.15% in the high dose group but did not change significantly from baseline in the low dose group, although the between group difference was not statistically significant. There was no change in hip BMD in either arm, although these data were not presented. Although these data may suggest a benefit of high dose vitamin D, it's important to note that the between group comparison at the lumbar spine was not statistically significant. It also is a little tough to put these changes into context as BMD may be increasing normally during this time in young adulthood. Showing the data as z-scores would help answer this question. Nevertheless, monthly vitamin D could indeed help preserve bone health in adolescents/young adults with HIV.
• Higher dose vitamin D does not prevent bone loss to a greater extent compared to lower dose in HIV-infected post-menopausal women (#682): In this randomized controlled trial, 82 HIV-infected post-menopausal women received 3000 IU D3 daily or 1000 IU daily over 12 months. In both groups, bone mineral density decreased significantly with similar patterns of changes in bone macro and microarchitecture by CT. 25 OHD levels were ∼20 ng/mL at baseline and increased to ∼25 ng/mL in the low dose group and ∼30 in the higher dose group. These findings suggest that vitamin D at these two doses did not differ in their effects on bone. However, without an untreated control group, it's difficult to determine whether vitamin D (at either dose) actually attenuated post-menopausal bone loss or did not have any demonstrable effect. While a placebo-treated third arm may have been helpful in the interpretation of this study, it may be considered unethical to randomize some women to a placebo group. From this we can conclude that 1000 IU daily is probably just as effective as 3000 IU daily at least over the short term using BMD as a surrogate. However, we should not conclude that vitamin D had no effect and therefore is not worth recommending at all. I'm sticking with my 1000 IU daily recommendation.
• Effect of Vitamin D Deficiency on the Effect of Statins (#617): In this secondary analysis of the SATURN trial, the investigators sought to determine whether the effects of rosuvastatin on LDL, monocyte activation and inflammatory markers, and carotid IMT differed based on baseline 25OH D. Compared to those with a baseline 25OHD < 20 ng/mL, those with a 25OHD ≥ 20 ng/mL had larger decreases in LDL, monocyte activation and carotid IMT over 96 weeks, suggesting that universal supplementation with 800-1000 IU daily, which will get most people to a 25OHD level ≥ 20 ng/mL, may also be important to maximize the effects of statins. It should be noted that for some markers examined (HOMA-IR, T-cell activation markers, oxidized LDL, sCD14), the effect of rosuvastatin did not differ by baseline 25OHD status. Nevertheless, overall, this is a thought-provoking result which may influence our practice.