Metabolics and Aging at CROI 2014
David Alain Wohl, MD - The University of North Carolina
At the feast of information regarding things metabolic, cardiovascular, and inflammatory served at this year's CROI, separating the entrees from the appetizers was not always easy given the heaping helpings of data presented at the many oral and poster sessions. What matters most to patients and their health care providers is less the morsels of what may be than the meat of what we know to be. Therefore, I review below what we learned at CROI about HIV metabolics and aging that we did not know the day before the meeting. My top picks are:
· HIV and other factors besides HIV cause cardiovascular disease, conditions associated with aging, and frailty
· Risk of heart attack and stroke now no different for HIV+ and HIV- patients (at least in a large HMO in California)
· Abacavir association with myocardial infarction persists in D:A:D cohort
· Vitamin D and calcium supplementation attenuates bone loss associated with starting Atripla
· Atazanavir/ritonavir causes more bone loss than darunavir/ritonavir or raltegravir and had higher discontinuation rate
· Statin helps reduce immune activation as well as inflammation and boost bone density but at a cost of insulin resistance
· How not to fight inflammation
· Tesamorelin reduces liver fat
· Testosterone levels decline over time equally in men with and without HIV
HIV and other factors besides HIV cause cardiovascular disease, conditions associated with aging, and frailty
For a while now we have been trying to quantify what HIV and its therapies contribute to cardiovascular disease (CVD) and other conditions associated with aging. At this year's CROI there was plenty of that, as described below, and the results collectively suggest a remarkably similar impact of HIV on the end-organ disease no matter the population or outcome, with a 1.5 to 2-fold increased risk typically reported.
At the same time, we are seeing in the data an increasing appreciation for the looming influence of other forces at play than those related to the virus. These include the obvious (smoking, substance use, hypertension, dyslipidemia), but also underappreciated risks not found on the list of traditional factors adjusted for in models of determinants of cardiovascular disease.
Two reports from the Veterans Aging Cohort Study (VACS) presented at CROI looked at the contribution of depression to heart disease. In the first, data from over 81,000 participants (a third were HIV+) without a history of known CVD were examined (1). Approximately 19% of the HIV+ and 23% of the HIV- veterans in the cohort were diagnosed with major depression during the analysis period. Analyses adjusted for demographics, hypertension, diabetes, lipids, smoking, cocaine abuse, alcohol abuse, body mass index, HCV infection, renal disease, anemia, statin use, and atrial fibrillation, and atrial flutter.
Major depression was found to be independently associated with heart failure in both HIV+ and HIV- participants. The hazard ratios for heart failure ranged from 1.19 to 1.68 with the 95% confidence intervals mostly overlapping. A number of other risk factors associated with heart failure were also identified, some, like renal disease, profoundly so (see table). Typical is the finding of an association between heart disease and cocaine use - a significant finding given the prevalence of abuse of this drug and the number of studies linking HIV to CVD that do not measure cocaine use or smoking of crack and marijuana.
A second related study from the same cohort looked a subset of over 27,000 VACS participants free of CVD to determine if clinical depression was linked to myocardial infarction (MI) (2). Over 6 years of follow-up, 23% had a diagnosis of depression and 1.3% suffered an acute MI. In models adjusting for the same covariates as in the heart failure study plus HIV-specific factors depression was significantly and independently associated with MI - accounting for a 31% increased risk above other known factors.
That there is more to the aging puzzle than a pesky virus and an exuberant immune response was unwittingly revealed in a report from the ALIVE Cohort (3). This observational study follows HIV- and HIV+ (29%) people in Baltimore who are active or former substance users. In the analysis presented at CROI, markers of inflammation (IL-6 and sTNFR1) were examined for their relationship to a standard definition of frailty (Freid LP, et al. 2001). In HIV+ participants a derived inflammatory index based on the markers was strongly associated with frailty after adjustment for confounders (only age, sex, race, education) with a hazard ratio of 2.55 (95% CI: 1.65, 2.95). However, among the HIV- the same inflammatory index appeared to be just as predictive with a hazard ratio of 2.74 (95% CI: 1.88, 3.99). When asked, what factors were associated with inflammation in the uninfected participants, the presenter stated that it was the obvious risks: substance abuse, comorbid disease, HCV. This punctuates an important point: HIV is not the only cause of inflammation and in this study where the controls have many of the same stressors and hard knocks as HIV+ people, the groups were indistinguishable as far as the effects of inflammation on health.
Each of these studies have their limitations but their results are hardly surprising. Depression, stress, and previous psychological trauma have been linked to adverse health outcomes in people without HIV infection. For example, children subjected to high levels of stress have been documented to suffer adverse health consequences into adulthood (see: http://opinionator.blogs.nytimes.com/2013/10/30/protecting-children-from-toxic-stress/ and http://www.natap.org/2012/HIV/032912_03.htm and a fascinating article in the New Yorker: http://www.newyorker.com/reporting/2011/03/21/110321fa_fact_tough). In our quest to find a viral smoking gun to solve the case of heightened CVD risk in people living with HIV, we have largely ignored these harder to measure 'soft' risks. This despite the fact that people who are HIV+ are more likely to have experienced toxic levels of stress and trauma, and their mental health sequelae.
Other studies presented at the conference punctuated the influence of lifestyle choices in contributing to abnormally heightened inflammation. In a report from a team at Harvard, smoking and obesity were each significantly associated with higher levels of inflammatory markers including IL-6, sCD14, sTNFRI & II (4). Similar findings linking smoking and inflammation were seen in the SUN cohort of HIV+ patients (5). In that study, heavy alcohol intake was associated with elevations of the coagulation marker D-dimer.
Overall, these findings do not mean that there are no effects of HIV on end-organ well being, either directly or mediated through immune activation and/or pro-inflammatory/pro-coagulation actions. Several studies make a strong case for a role of the virus. But, of the excess CVD and other diseases of aging in people with HIV infection, how much is due to these less than obvious factors? This is something we need to get a better handle on.
So, lets look at what we learned about potential direct effects of HIV on organ health were. A series of cohort studies nicely plumbed this issue at CROI. These included:
· A follow-up from the Boston hospitals study published in 2007 by Dr. Virginia Triant (6) and colleagues continues to show an association between HIV infection and increased risk of CVD (7). Records of HIV+ patients and controls matched on age, race and gender were searched for diagnosis codes for MI, stroke, angina, and coronary revascularization. Rates of a composite CVD endpoint were greater for HIV+ patients [16.7 events per 1000 PYs (95% CI 15.1-18.5]) versus the HIV- controls (10.7 [95% CI 10.2-11.2]). As in the original study, there were limited data available on lifestyle risk factors for CVD; rates of smoking were "approximated" rather than actually recorded.
· A third VACS study focused exclusively on women and found higher rates of CVD in those with HIV infection (8). As in the other VACS presentations, those without known CVD were included and ICD-9 codes, here for MI, heart failure, and stroke, were sought. Of the 2,190 women, 32% were HIV+. During a median 6 years of follow-up there were 83 events (mostly heart failure) and 53% were experienced by those with HIV infection. Modeling was conducted to determine if HIV was associated with CVD independent of a number of putative risks and in the most rigorous of these there was a significant association between being HIV+ and CVD in women (hazard ratio was 2.8 (95% CI: 1.7, 4.6). Other independent risks were age, hypertension, and cocaine use. Interestingly, when comparing HIV- women and HIV+ women with and without controlled HIV RNA no significant difference in rates of CVD between the HIV- patients and the HIV+ women with HIV RNA levels <500 copies/mL was found. In contrast, a higher risk was found for those with a higher viral load and those not on ART.
In addition to these cohort studies there were several presentations that centered on the pathogenesis of CVD in the setting of HIV infection:
· A report from the Multicenter AIDS Cohort Study (MACS) found markers of inflammation and immune activation correlated with evidence of atherosclerosis using CT angiography in men with HIV (n=139) but not those who were HIV negative (n=92) (9). The study was conducted at the Los Angeles study site and all the HIV+ participants had undetectable HIV RNA levels on ART. There were no significant differences in the prevalence of coronary plaque (calcified and non-calcified) or coronary calcium score in the HIV+ and HIV- men. However, the HIV+ men were a bit younger (mean age 52 versus 55 for the HIV-). The HIV+ men did have higher levels of the monocyte activation markers sCD163 and sCD14 and in the HIV+ but not the HIV- patients levels of markers of immune activation and sCD163 were associated with calcified plaque and calcium score.
This study was a teaser for a larger MACS study of coronary plaque published after the conference in the Annals of Internal Medicine by Dr. Wendy Post and colleagues that deserves some discussion (10). In this cross-sectional study, 618 HIV+ and 383 HIV- men without a history of cardiac surgery/intervention and between the ages of 40 to 70 years underwent CT for calcium scoring; CT angiography to assess for plaque was performed in 450 of the HIV+ and 309 of the HIV- participants who had no evidence of chronic renal insufficiency. There were some significant differences between the HIV+ and HIV- men even in this cohort of MSM including in demographic characteristics, lipids, BMI, and current smoking status. After adjusting for age, race, CT scanning center, cohort and CAD risk factors, the association between HIV and presence of coronary calcium became "borderline significant" (i.e., not so much) (PR, 1.12 [CI, 0.99 to 1.26]). CT angiography found plaque in an amazing 77.6% of HIV+ men and 74.4% of HIV- men. After full adjustment for major confounders, there remained a higher prevalence of plaque in those who are HIV+ (PR, 1.13 [CI, 1.04 to 1.23]). Importantly, HIV+ men were also more likely to have noncalcified plaques (the most vulnerable to rupture) (PR, 1.25 [CI, 1.10 to 1.43]). Older age was associated with noncalcified plaque in the HIV+ but not HIV- men, and this seemed to drive the overall differences between these groups (see figure). Actual coronary stenosis was seen in 17% of HIV+ and 15% of HIV- and no significant differences were found after adjustment. Longer duration of ART (PR, 1.09 [CI, 1.02 to 1.17]; P < 0.007) and lower nadir CD4+ cell count (PR, 0.80 [CI, 0.69 to 0.94]; P< 0.005) were associated with coronary stenosis greater than 50%. Overall, this study adds to others showing a possible increased burden of noncalcified coronary plaque among those living with HIV infection, especially as they age.
· Other studies demonstrated not only increased aortic inflammation as indicated by FDG PET scans in HIV+ people (11), as was reported by Dr. Steve Grinspoon and his team at the 2013 CROI, but also FDG uptake in the spleen and bone marrow in patients with controlled HIV (12). The presence of inflammation in these areas and the correlation with markers of inflammation and monocyte activity are suggestive of a possible pathogenic mechanism underlying CVD in HIV+ people.
Risk of heart attack and stroke now no different for HIV+ and HIV- patients (at least in a large HMO in California)
These data on risk of CVD are pretty scary and many living with HIV can wonder what they can do to reduce their risk of heart attacks, strokes, coronary plaques, and inflamed spleens. Some good news came from, where else? California. Researchers at the Kaiser Permenante group found rates of MI and stroke among HIV+ and HIV- members had started to converge and in recent years were no different (13, 14).
The authors posit that the reduction in CVD among HIV+ patients is likely a result of a combination of factors including CVD risk factor reduction through medications and lifestyle modification, use of more lipid-friendly ART, and reduced immunodeficiency with earlier treatment of HIV infection. This seems reasonable and may indicate that we are already on the right track to reducing any excess CVD disease burden accompanying HIV.
Abacavir and MI Risk After the Fall
In 2008 researchers from the D:A:D cohort, a large observational study of HIV+ patients mostly in Europe, first reported an association between abacavir use and risk of MI. In an update at CROI the research group looked at this association five years later with an underlying assumption that reduced use of abacavir since their initial announcement would channel those at greatest risk of MI away from this drug (15). Therefore, any continued association would, the authors imply, be more convincing. As expected, abacavir use in the cohort declined among those with high and moderate risk for CVD based on Framingham 10 year risk score. However, use of abacavir in those at low CVD risk has remained largely unchanged. As was reported in 2008, the risk of MI continued to be close to double for those treated with abacavir, even with adjustment for additional confounders. However, what is often missed is that the relative risk of MI then and now was greatest in those not at high CVD risk. As there has been little change in the use of abacavir in those with greatest relative risk of MI, it may not be too surprising that the association persists despite less use by those for whom the link to MI was weaker. It is also reassuring that the overall rate of MI in this cohort remains rather low with 94 events during over 31,000 patient-years. It is unclear if this analysis adds much to the does-abacavir-cause-MI debate and those who feel comfortable using the drug will likely continue to do so.
Vitamin D and calcium supplementation attenuates bone loss associated with starting Atripla
In most studies of ART in which bone mineral density (BMD) has been tracked, a sharp decline in BMD is observed. The cause of this rapid decline in bone density is unclear and some have hypothesized that ART triggers an immune reconstitution like syndrome that effects cells responsible for bone remodeling. In addition, there may contributions to bone loss by ART itself. Tenofovir produces deeper drops in BMD than other drugs in its class and efavirenz has been shown to reduce levels of 25,OH vitamin D. However, a BMD drop is also seen during therapy with other agents. To determine if supplementation with vitamin D and calcium can blunt the loss of bone after initiation of ART, the US AIDS Clinical Trials Group (ACTG) conducted a randomized placebo controlled trial of this intervention in patients starting Atripla (16). Participants (n=165) were assigned 4000 IU of vitamin D plus 1000 mg of calcium or matching placebo for 48 weeks. Bone loss of the hip and spine were seen in both arms but the drop was 50% lower at the hip in the intervention group compared to the placebo (p <0.001). At the spine, where there is more variability in BMD, there was a trend favoring the vitamin D/calcium group but this did not reach statistical significance. For many providers this is a fairly low cost and simple intervention, although the potential effect of added pill burden on medication adherence should not be ignored. Some wonder whether supplementation will work or is necessary when starting other regimens and undoubtedly others will try to find out. Sharing these data with patients and offering supplementation regardless of regimen seems prudent.
Atazanavir/ritonavir causes more bone loss than darunavir/ritonavir or raltegravir and had higher discontinuation rate
In our clinic at UNC some providers reach for atazanavir/ritonavir and some for darunavir/ritonavir when a protease inhibitor is called for. As these boosted PIs have not been compared head to head, who could argue with either preference? Results of the ACTG trial A5257, presented in Boston may change that (17). ART-naïve (n=1810) patients were randomized 1:1:1 to atazanavir/ritonavir, darunavir/ritonavir, or raltegravir - all with Truvada. At 96 weeks, the two PI regimens were equivalent in terms of virologic failure by intent to treat (changes in therapy were permitted and did not count as failure); however, the proportion with a viral load < 50 copies/mL at this time point by snapshot analysis (off therapy = failure) showed a markedly lower rate of viral suppression for atazanavir (63%) versus darunavir (73%), and raltegravir (80%). Toxicity was the main contributor to the atazanavir outcome - there was a 9.2% difference in tolerability failure rate between this PI and darunavir. Hyperbilirubinemia was a very common reason (47%) for the discontinuation of atazanavir but so was gastrointestinal toxicity (25% compared to 14% for darunavir). In addition, decreases in BMD were greater for atazanavir than darunavir (18). These results, coupled with reports of kidney and gall stones due to crystalized atazanavir, and possible links between the PI and renal issues tips the balance in favor of darunavir.
Statin helps reduce immune activation as well as inflammation and boost bone density but at a cost of insulin resistance
Once it was clear that people living with HIV had, on average, higher levels of inflammatory markers, interventions to reduce inflammation began to be studied. Statins, the darlings of the anti-inflammatory world, were an obvious first choice: they lower bad cholesterol, are associated with reductions in western civilization mortality, and they are known to have anti-inflammatory properties. In addition, there has been some evidence that they can increase BMD.
In the SATURN trial, ART-receiving patients (n=147) with evidence of heightened immune activation (CD8+CD38+HLA-DR+ >19%) or inflammation (hsCRP > 2mg/L) and no need to receive statin therapy were randomized to rosuvastatin or placebo (19). By 48 weeks, hip BMD increased in the intervention arm by 0.6% but had declined by the same amount in the placebo control group - leading to a significant difference (P = 0.017). A similar effect was seen at the greater trochanter but not the spine, where BMD did not budge much in either group.
Rousuvastatin was associated with a significant decline in a number of markers of inflammation and immune activation including the monocyte activation marker sCD14 and in the proportion of CD14dimCD16+TF+ monocytes, the vascular inflammation marker Lp-PLA2, and levels of interferon gamma-induced protein 10 (IP-10) as well as the proportions of activated CD4+ and CD8+ T cells (20).
But, the statin was associated with significant increases in fasting glucose 8% (P = 0.0017 versus baseline, P = 0.217 versus 3.3% rise with placebo), fasting insulin 52% (P = 0.0006 versus baseline, P = 0.0055 versus 5.5% rise with placebo), and insulin resistance as measured by HOMA-IR 72% (P = 0.0015 versus baseline, P = 0.0068 versus 14.5% rise with placebo). One person was diagnosed with diabetes mellitus but was in the placebo arm.
The effect of statins on glucose metabolism is a topic drawing intense interest given the ubiquity of these medications. This study makes clear that we may have to weigh the balance of the benefits of these drugs with their toxicities and tolerability issues.
How not to fight inflammation
As mentioned above, inflammation has become public enemy # 1 and so as good scientists and clinicians we look for a pill that we can use to take that sucka down. At CROI a row of posters reporting results of interventions aimed at snuffing out immune activation and inflammation could have been called Disappointing Results Alley. It was a lonely place.
Sevelemer, a non-absorbable polymer that when orally administered may decrease lipopolysaccharide (LPS) levels was studied in 40 HIV+ patients with CD4+ cell counts of 400/mm3 or greater and detectable HIV RNA levels (>50 copies/mL) and not taking ART (21). At the end of 8 weeks there was no significant change in LPS, sCD14 (another marker of microbial translocation), or markers of inflammation/coagulation. LDL cholesterol levels did drop, which is nice.
Rifimaxin is a non-absorbable antibiotic that may also decrease LPS and microbial translocation. Four weeks of this agent was studied in 49 people with HIV, CD4+ cell counts <350/mm3, on ART with viral loads below detection limit (22). Changes in levels of markers of interest were compared to 29 control subjects who were not dosed but were followed and 20 HIV-negative controls from the MACS cohort. At the end of the day there was little change in any of the outcomes with rifimaxin treatment. Gut microbial translocation seemed to be modestly and transiently reduced and IL-6 and CRP shifts were small and not even impressive to the researchers.
Mesalamine an anti-inflammatory salicylate used to treat some forms of inflammatory bowel disease, was studied in a small cross over trial of 33 patients on stable ART but with CD4+ counts <350/mm3 (23). Bottom line, did not do much of anything.
One of the few things, other than ART and rosuvastatin, that was found to impact markers of inflammation was a probiotic (24). Patients (n=30) on ART were randomized to a lactobacillus and bifidobacterium probiotic cocktail or a control arm. After 8 weeks, there were significant declines in CRP and D-dimer levels in the probiotic arm (IL-6 trended down). Interestingly, LPS and scCD14 did not change appreciably, suggesting the probiotic may not have been acting on microbial translocation. That changes in the gut flora can lead to reductions in inflammation, perhaps independent of any transit of bacteria across the mucosal has been suggested in other disease states and it is likely we will see much more about manipulations to the gut microbiome in people living with HIV.
Tesamorelin reduces liver fat
Fatty liver disease is prevalent in HIV+ patients. To date, there has been little to offer those diagnosed with non-alcoholic steatohepatitis to reverse this condition. Long ago, a study examining pioglitazone for lipodystrophy found decreases in liver fat in those treated with this anti-diabetic agent when scans of the belly were examined. However, this approach has not been pursued. At CROI 2104, Steven Grinspoon presented data on behalf of his colleagues examining the effects of tesamorelin on liver fat (25). This growth hormone releasing factor has been approved for the treatment of excess visceral abdominal fat in HIV+ persons. In this study of HIV+ patients on ART with a waist circumference of at least 95 cm (37 inches) and a waist-to-hip ratio of at least 0.94 for men and a waist circumference of at least 94 cm and a waist-to-hip ratio of at least 0.88 for women, 28 were assigned to tesamorelin and 22 to placebo. At 24 weeks, liver fat dropped by 40% in the tesamorelin group and rose 27% in the control group (p = 0.004). As expected, visceral fat declined in the active therapy arm. There was a trend toward a reduction in carotid intimal thickness with the drug but in this small study this finding is hard to interpret. Overall, the tesamorelin was tolerated although three subjects assigned the drug discontinued treatment due to an adverse event. While no one is suggesting that this is an attractive or even viable therapeutic approach to fatty liver disease, the study suggests that interventions targeting visceral fat may have other metabolic benefits.
Testosterone levels decline over time equally in men with and without HIV
Many men with HIV infection believe their levels of testosterone are insufficient and the cause of their problems with libido, erectile dynamics, and the ability to obtain 6-pack abs. Not a few are prescribed topical or injectable testosterone. Our friends in the MACS cohort again turned to the power of that comes from studying both HIV+ and HIV- individuals to examine whether there are differences between the two in terms of testosterone changes that occur with aging (26). Stored serum of HIV+ men (n=182), 45 years an older before starting ART, not taking supplemental testosterone, with available specimens pre-ART and at least two after initiating ART were tested for free testosterone. Sera from HIV- men (n=267) matched to the HIV+ subjects by age, race, site, and calendar time for the pre-ART and on-ART samples served as the comparator. Overall, testosterone levels were not significantly different between HIV+ and HIV- men but when looked at by the time of day that the specimen was drawn, morning levels were lower for the HIV+ men, but by the afternoon the levels were indistinguishable by HIV status. Over a median of 6 years of observation, testosterone dropped in both HIV+ and HIV- men and there was no difference in the rate of this decline. The lower levels of this hormone in the morning among the HIV+ men mirrors an effect seen with aging, and therefore may not be inconsequential. However, the results are mostly reassuring in that men do not suffer accelerated declines in testosterone as they get older when HIV+.
At this CROI there was an impressive constellation of research findings that help to quantify the various contributors to non-AIDS related morbidity in HIV+ people. The introduction of novel mechanisms of disease including mental health issues and gut flora add important pieces to the puzzle. Interventions were tested that may have clinical implications. At the risk of being Rumsfeld-esque, we learned more about some of the things we do not know (the known unknowns), but remain challenged with the unknowns we do not (yet) know.
1. White J, et al. Abstract 726
2. Khambaty T, et al. Abstract 735
3. Piggott D, et al. Abstract 762
4. Krishnan S, et al. Abstract 757.
5. Cioe PA, et al. Abstract 732
6. Triant VA, Lee H, Hadigan C, Grinspoon SK. Increased acute myocardial infarction rates and cardiovascular risk factors among patients with human immunodeficiency virus disease. J Clin Endocrinol Metab. 2007 Jul;92(7):2506-2512.
7. Triant V, et al. Abstract
8. Womack J, et al. Abstract 734
9. Daar E, et al. Abstract 730
10. Post WS, et al. Associations between HIV infection and subclinical coronary atherosclerosis. Ann Intern Med. 2014; 160:458.-67
11. Tawakol A, et al. Abstract 130.
12. Hsue PY, et al. Abstract 131
13. Klein D, et al. Abstract 737
14. Klein D, et al. Abstract 741
15. Sabin C, et al. Abstract 747LB
16. Overton T, et al. Abstract 133
17. Landovitz L, et al. Abstract 85
18. Brown T, et al. Abstract 779
19. McComsey GA, et al. Abstract 134
20. Funderburg N, et al. Abstract 335
21. Sandler NG, et al. Abstract 337
22. Tenorio AR, et al. Abstract 339
23. Somsouk M, et al. Abstract 341
24. Ortiz AM, et al. Abstract 83
25. Stanley T, et al. Abstract 135
26. Slama L, et al. Abstract 754