Non-Infectious Complications of HIV and Antiretroviral Therapy - IAS 2013 (Kuala Lumpur, Malaysia)
Roger Bedimo, MD
VA North Texas Health Care System
Rumors had it that the 7th IAS Conference on HIV Pathogenesis, Treatment and Prevention (IAS 2013) was on pace for breaking all-time low attendance records. Officially, there were 5000 attendees, but it certainly felt like fewer than that. If so, one has to wonder about the impact of the choice of location of future conferences. The setting was very pleasant though.
Chronic Complications of HIV and HAART: Making Old Bones With HIV
Non-infectious co-morbidities are contributing for a growing share of the mortality of HIV infected patients in the HAART era. It is assumed that higher levels of immune activation, systemic inflammation, and immunosenescence promote earlier onset and accelerated course in this population. However, pathogenic mechanisms remain to be clearly elucidated. Further reductions in HIV-associated morbidity and mortality will require better understanding of disease mechanisms and co-factors associated with these non-infections complications to guide preventive and therapeutic interventions. It is nice to see that in the past couple of years, a growing focus is being placed on exploring the mechanisms of bone, kidney and cardiovascular disease. It's even more important that markers of inflammation, kidney dysfunction, bone turnover and bone mineral density are also now frequently planned safety endpoints in the evaluation of new drugs and regimens.
1. Bone Disease
Osteoporosis and osteoporotic fractures (OF) have emerged as common complications encountered in the aging HIV-infected population(1-2). HIV infection itself and its treatment with Highly Active Antiretroviral Therapy (HAART) are associated with accelerated bone loss and higher fracture risk. However, determinants and mechanisms of this increased risk remain to be elucidated.
a. Increased risk of osteoporosis and fractures in HIV patients, even in adolescents:
It is well recognized that "traditional" risk factors for decreased bone density are over-represented among HIV-infected patients, including smoking, low BMI and hypogonadism. Knobel et al (3) add to the growing body of literature confirming that even after controlling for these risk factors, HIV infection remains an independent predictor of decreased bone density. In a population-based cohort study, they identified 2,489 HIV-infected patients among 1,118,156 subjects aged 40 or older in Catalonia. Following a median follow-up time of 3 years, they calculated hip and all fracture incidence in the HIV and non-HIV patients. HIV status was associated with a HR 4.7 (2.4-9.5; p< 0.001) for hip fractures and 1.8 (1.2-2.5; p=0.002) for all fractures, adjusting for several factors including age, gender, body mass index (BMI), smoking status, alcohol consumption, oral glucocorticoid use, and co-morbid conditions (as the Charlson Index).
Also, in a small study of 47 perinatally infected Thai adolescents (4), low bone mineral density (BMD) and low vitamin D levels were significantly prevalent, and increased even further on repeat measurements 12 to 24 months later. The proportion of adolescents with BMD z-scores ≤ -1.5, ≤ -2.0, and ≤ -2.5 of 17.0%, 12.8% and 10.6%, respectively; 13(27.7%). Those proportions increased to 27.7%, 10.6% and 8.5%, respectively, at the second measurement. Given that these subjects are yet to achieve their peak bone mass, the findings suggest they may never do, magnifying their fracture risk as they age.
b. Increased bone turnover predicts low BMD in HIV infected patients:
Two prospective cohort studies presented on Wednesday showed increased bone turnover markers (BTM) among HIV-infected patients. Cotter et al. (5) measured BMD and markers of bone resorption (C-terminal cross-linking telopeptide of type 1 collagen (CTX-1)) and bone formation (osteocalcin (OC), procollagen type 1 amino-terminal propeptide (P1NP)) in 210 HIV-infected patients compared to 254 uninfected controls, with a median age of 46 years. HIV was associated with lower BMD and higher bone turnover. Furthermore, Cotter et al showed that a weak association between HIV and low BMD remained after controlling for increased turnover. It's therefore likely that other mechanisms than increased turnover drive declining BMD in HIV, calling for caution about targeting these markers for therapeutic interventions.
Maalouf et al. (6) measured BMD and BTMs in 168 subjects including HIV/HCV co-infected, HIV mono-infected, HCV mono-infected and uninfected subjects. All patients were older than 40, all HIV patients were virologically suppressed on HAART and all HCV patients were HCV treatment naïve. They also showed a significant association of HIV infection with decreased BMD and increased markers of bone resorption (CTX) and bone formation (OC and bone specific alkaline phosphatase (BSAP)). Higher levels of bone turnover markers were significantly correlated with lower BMD.
IAS: HIV[ART] and HCV infections independently contribute to lower bone mineral density but have different effects on bone turnover markers - (07/05/13)
c. Hepatitis C also decreases bone mineral density, but appears to not do so through increased bone turnover:
HCV co-infection is a significant risk factor for osteoporotic fractures in several cohorts of HIV-infected patients (1,2). Whether HCV does so by further decreasing BMD and/or bone turnover among HIV patients remains unclear. To address this, Maalouf et al (6) had four groups of patients: HIV/HCV co-infected, HIV mono-infected, HCV mono-infected patients, and uninfected controls. This allowed the group to show that both HIV and HCV independently lower BMD and T-scores (smaller contribution for HCV). The HIV impact on BMD could be explained by increased turnover (increased resorption and formation markers). The authors also measured regulatory cytokines: receptor activator of nuclear factor kappa-B ligand (RANKL) and osteoprotegerin (OPG). RANKL involved in formation and activation of osteoclasts while OPG is decoy receptor competing with RANKL. Neither was increased in the HIV patients compared to controls. So, the increased turnover doesn't appear to be driven by RANK/RANKL/OPG system, which couples bone formation and resorption. In contrast, HCV infection was not associated with increased bone resorption, leaving its effect on decreased BMD largely unexplained.
d. Impact of different antiretroviral regimens on BMD and bone turnover:
It's important to note the since all HIV patients in the above studies were on HAART, the differences in BMD between them and HIV negative patients could be in part or entirely attributed to the antiretroviral regimen.
Jennifer Hoy (7) presented data from SECONDLINE where 210 patients were randomized to receive lopinavir/ritonavir (LPV/r) + raltegravir (RAL) or LPV/r+2-3 nucleoside reverse transcriptase inhibitors (N(t)RTIs) as second-line therapy. The incidence of osteopenia (7.6%) and osteoporosis (2.0%) assessed at the proximal femur over the 48 weeks were similar between groups. However, the nucleoside-free group had a significantly lower % drop in BMD at week 48 at the femoral neck:-5.2 (-6.7, -3.8) vs. -2.9 (-4.3, -1.5); p= 0.0001 and at the lumbar spine: -4.2 (-5.7, -2.7) vs. -2.0 (-3.5, -0.6); p= 0.0006.
The findings are consistent with the safety analysis of the RADAR study (8), which randomized 83 HIV treatment naïve patients to receive either tenofovir/emtricitabine (TDF/FTC) or raltegravir (RAL), either in combination with ritonavir-boosted darunavir (DRV/r). Changes in subtotal BMD differed significantly (+1.02% with RAL vs. -0.76% with TDF/FTC, p=0.002). CTx and P1NP increased significantly in the TDF/FTC group (+112%, resp. +85%, p=0.01 for both), but not in the RAL group (+20%, p=0.22; and +11%, p=0.33). Week 16 changes in BTM predicted week 48 changes in BMD (R= -0.394, p=0.003 for CTx; and R= -0.477, p< 0.001 for P1NP). The findings suggest that NRTI-free regimens will be associated with less marked declines in bone mineral density and a smaller impact on bone turnover than TDF-containing regimens. Furthermore, it might be relevant to measure changes in BTMs as predictors of future bone loss.
2. Kidney Disease
Like other non-infectious complications, the incidence of chronic kidney disease is difficult to assess. Also, the relative contributions of aging, antiretroviral therapy and intrinsic effect of HIV are difficult to assess in the causation of CKD.
a. Incidence of CKD in HIV: Impact of aging and HAART:
A large analysis (9) of 11 NA-ACCORD cohorts (including 37,073 patients) showed significantly high rates of end-stage renal disease (ESRD) among HIV-infected patients (almost 4-fold higher than the US general population). As has been shown before, the greatest contribution was from African Americans, while there was no statistically significant difference in the risk of ESRD by HIV status for Hispanics. Age- and sex-standardized incidence ratios (SIRs) for ESRD were 4.4 (3.88, 5.00) among Blacks, 1.89 (1.28, 2.51) among Whites, and 1.61 (0.84, 2.38) among Hispanics.
Mandalia et al (10) used a linear mixed model analysis to assess the relationship between renal function, HIV-1 infection, ART and aging. They grouped patients ages into decades calculated mean eGFR values (by MDRD) by decade for ART naïve PLHIV and those on first-line ART regimens and compared with creatinine clearances of HIV uninfected people. This analysis showed a progressive decline in mean eGFR from 80 ml/min in 40-49 years old to 59 ml/min in those older than 50. Also, the effect of antiretroviral regimen on eGFR was only apparent in patients older than 50 where regimens with boosted PI were associated with a greater eGFR decline than those on NNRTI: mean difference in slope was 5.8 ml/min (95% CI: 3.9 to 7.7, p< 0.001).
IAS: End-Stage Renal Disease Incidence 2 to 4 Times Higher With HIV in North America - written by Mark Mascolini - (07/02/13)
b. Tenofovir nephrotoxicity across geographic settings:
A lot of attention continues to be placed on the impact of tenofovir (TDF) - by far the most widely used NRTI in resource-rich settings - on kidney disease. Guilleme et al (11) from British Columbia presented a retrospective evaluation of changes in renal and lipids parameters among 225 adults that were switched by their treating physician from TDF+3TC/FTC to ABC+3TC-based HAART. For the majority of patients (56%), the third drug was atazanavir (boosted or unboosted), and all had VL<200 copies/mL at the time of switch. For creatinine, eGFR, phosphorus, and urine albumin to creatinine ratio (UACR), baseline results were the individual's worst values within 12 months before starting ABC. Repeat measurements were evaluated at 3, 6, and 12 months following switch.
By month 3 following the switch, there was a significant improvement in serum creatinine (-12 μmol/L; (-22, -4, p<0.001)), and eGFR (+9 mL/min (+4, +19; p<0.001)). The improvement in kidney function was maintained up to month 12. Improvements in renal function were similar whether the patients were on ATZ or not. There were also statistically significant improvements in serum phosphorus and UACR from month 3, there were sustained up to month 12.
Almost all patients remained virologically suppressed after the switch (98.4% of patients maintained a plasma VL <200 at month 12). These results will have to be contrasted with a meta-analysis of over 40,000 patients presented at the meeting by Lee et al (12) showing that TDF-containing regimens were associated with a statistically significantly greater likelihood of virologic success than ABC-containing regimens.
Most data on TDF-associated renal dysfunction has been derived from Europe and North America. Dravid et al (13), presented data from India at the conference. They enrolled 743 patients on TDF-containing and 340 on TDF-sparing regimens, followed for a median of 21 months and 33 months respectively. They defined Acute Kidney Injury as serum creatinine > 2 mg/dl, Cr Cl decrease to < 50 ml/min or GFR decrease > 50% of baseline (RIFLE criteria 2002). Patients with GFR value < 60 ml/min(MDRD equation), 6 months after TDF discontinuation were classified as having Chronic kidney disease (CKD).
During follow-up, there was a mean decline in GFR on TDF-containing regimens was 5.29 ml/min/year (vs. 1 ml/min/year for TDF-sparing regimens). TDF use with Protease inhibitors had a greater decline in GFR compared to TDF with Non-nucleoside reverse transcriptase inhibitors (9.19 ml/min/year vs. 4.18 ml/min/year). AKI developed in 36 patients (incidence 4.8 %) leading to TDF discontinuation. Median time to developing AKI After TDF discontinuation for > 6 months (18/36 patients), 9 showed complete recovery of renal function while 9(50%) progressed to CKD.
c. Is renal tubular dysfunction associated with bone mineral loss?
Finally, it remains unclear whether the possible renal tubular dysfunction (RTD) mediated by TDF drives at least in part the increased risk of osteoporosis also associated to this drug. An interesting study from the King College of London (14) evaluated rates of renal tubular dysfunction (RTD) in a cohort of 438 HIV patients by the use of urinary retinol binding protein (RBP) excretion presented as a ratio with urinary creatinine (RBPCR). By this assay, 65 (15.4%) of patients had mild to moderate RTD and another 22 (5.2%) had severe RTD. A total of 79% of patients in this cohort had tenofovir (TDF) exposure. Unsurprisingly, RTD was correlated with TDF exposure (adjusted coefficient: 0.41; p=0.01) and inversely correlated with eGFR (Adj. Coeff. -0.02; p=0.01). With measurement of DXA scans and markers of bone turnover (CTX and P1NP), they then showed that RTD predicted neither BMD nor increased bone turnover.
3. Cardiovascular Disease
Steve Deeks highlighted the the increasing role of impact of cardiovascular disease in the morbidity and mortality of HIV-infected patients at the opening lecture with a mention of the recently published VA study by Freiberg et al (15). In this large VA cohort, after adjusting for traditional cardiovascular risk factors, HIV disease was associated with a 50% increased risk in incident myocardial infarctions. Drivers of this increased risk - according to this and several other presentations at the conference - include "persistent inflammation/immune dysfunction, subtle but cumulative treatment toxicity, excess co-morbidity (non-AIDS events), and clinical aging". Not mentioned was the still controversial impact of type and duration of antiretroviral therapy. Other fundamental questions remain to be answered including which risk model to utilize, and whether HIV itself should be included as a cardiovascular risk factor in assessing cardiovascular risk scores.
IAS: Aging & Cure - a talk by Steve Deeks, MD at the IAS Conference in Kuala Lumpur - (07/03/13)
a. Predictors of carotid intimal thickness and fatty heart in HIV+ vs. HIV- patients
Ricky Hsue (16) assessed the predictive value of traditional cardiovascular risk factors, lipoprotein particle size distributions and inflammatory markers on carotid intimal thickness (CIMT) in 307 patients (179 HIV+ and 128 HIV-). They determined that the calculated heart age of HIV patients was higher than predicted (by 6.7 years, compared to their HIV negative counterparts), despite lower lipid levels and smoking rates. Perhaps a testament to how aggressively providers are addressing cardiovascular risk in HIV patients, they were also more likely to be on lipid lower and antihypertensive medications.
CIMT was measured at the common carotid artery (CCA) and the internal carotid artery (ICA). The group has previously shown that CIMT progressed faster in HIV-infected patients than in control (although the rate of progression was faster at the carotid bulb - not reported in the current analysis) (17). Here, they show that factors predicting higher CIMT were different in HIV+ and HIV- patients:
In HIV+ patients, LDL particle size (p=0.04)[from jules: small dense LDL is culprit, highly correlated with CIMT] - but not LDL level - was associated with CIMT measurement at the ICA. The reverse was the case for HIV- patients. Both HDL (p=0.04) and large HDL particle number (p=0.02) were significantly correlated with CIMT values at the CCA for the HIV-, but neither was for HIV+. As for inflammatory markers, IL-6 was predictive of CIMT at the ICA for HIV+ while hsCRP was predictive for HIV-. Use of abacavir or indinavir, or duration of lopinavir were not predictive of CIMT.
Heart age was the only significant predictor of CIMT at the CCA for HIV+ patients, however for HIV- patients, age, HTN, HTN medications, and history of MI/stroke were all significant predictors.
Framingham risk factors were strongly correlated with CIMT at both the CCA (p=0.008) and ICA (p=<0.0001) in both HIV+ and HIV- individuals (CCA p=0.001) and (ICA p=0.03). D:A:D risk calculator was less correlated with CCA CIMT (p=0.05) than the Framingham risk calculator (p=0.008).
The authors then made a case for re-stratification of cardiovascular risk by CIMT, lipid sub-particles, and/or inflammatory markers found significant in Multivariable Regression analysis to determine if predictive of atherosclerotic regression as measured by CIMT 1 year later.
IAS: Independent Predictors (slides) of Carotid Intimal Thickness Differ Between HIV+ and HIV- Patients with Respect to Traditional Cardiac Risk Factors, Risk Calculators, Lipid Subfractions, and Inflammatory Markers (slide presentation) - (07/09/13)
IAS: Independent Predictors of Carotid Intimal Thickness Differ Between HIV+ and HIV- Patients with Respect to Traditional Cardiac Risk Factors, Risk Calculators, Lipid Subfractions, & Inflammatory Markers (poster) - (07/09/13)
Cardiac steatosis (fatty heart) has been found in patients with diabetes and obesity. Lai (18) assessed its prevalence by proton magnetic resonance spectroscopy among 167 HIV-infected patients aged 21 and above with no clinical evidence of cardiomyopathy or CAD, and evaluated its determinants.
The median myocardial triglyceride content was 0.76% (IQR:0.43-1.76%). Two of 167 had diabetes and approximately 50% were overweight or obese. Of HIV specific factors, baseline viral load and duration of protease inhibitor use were associated with the accumulation of myocardial triglyceride. Interestingly, this was not statistically associated with the presence of coronary plaque. So, it's unclear what the prognostic significance of the steatosis is in HIV patients.
b. HAART exposure and incident cardiovascular disease
Further igniting the debate over the role of abacavir on cardiovascular disease in HIV, Young et al (19) presented an analysis of the Swiss cohort. They assessed the cardiovascular risk of 11,625 patients enrolled in the cohort from April 2000 to October 2012, 4474 of whom were exposed to abacavir. A total of 350 patients reached the composite endpoint of myocardial infarction, cardiovascular related death or invasive cardiovascular procedure. The authors used two different statistical approaches to estimate the risk associated with abacavir exposure.
First, in a conventional Cox model, while cumulative exposure to abacavir was not associated with increased risk of events (HR: 1.04, CI: 0.96 - 1.10), recent use - within 6 months - was (HR: 1.52; CI: 1.13 - 2.04). The authors then used a new Cox model that estimates a flexible curse based on a one knot spline constrained to equal zero at both ends of a period of influence. This, as the authors explained to me in plain English, means that the model accounts for a lag time before exposure and an event. The effects then increases with exposure time. They argue that this model provided a better fit for the data. In the revised analysis yielded the following HR for effect of ABC: 1.04 (0.99-1.09) for cumulative exposure (per year), 3.36 (2.04-5.53) for any use within the last 1 to 6 months, and 0.42 (0.25-0.69) for current use. In other words, it's not the recent use but immediate past use that was deleterious which they suggest argues against increased inflammation as a mediator of this risk.
The effect of statin (HMG-CoA reductase inhibitors) therapy has been evaluated due to their pleiotropic effects beyond lipid-lowering, including decreases in chronic inflammation and immune activation. Recent evidence suggests that statins could reduce mortality and non-AIDS complications, including malignancies (20, 21). It remains difficult to tease out the real impact of statins as owing to the difficulty in controlling for several biases in these cohort studies. An obvious one being that only patients in regular care get prescribed statins. Nonetheless, Spagnuolo et al (22) presented an analysis of an Italian cohort that adds to this field. They included 5357 HIV patients followed for a median of 10.3 (4.8-15.1) years and who were on HAART for a median of 10 years. They observed 375 cancers (194 AIDS-defining malignancies - ADM; and 181 non-AIDS defining malignancies - NADM). Only 14% of the patients were statin users, mostly rosuvastatin (the proportions have been found to be much higher in most US cohorts). They found a remarkably lower incidence of malignancies among the statin users. In a multivariate analysis including age, years of HAART use, total cholesterol, CD4 count and viral load, any use of statin was associated with a cancer HR of 0.54 (0.46 - 0.69; p<0.001). Cancer occurred in 12/740 (1.6%) statin users and in 363/4617 (7.9%) no-statin users (p< 0.0001).
1. Yin et al., AIDS 2010; 24:2679-2686;
2. Young et al., Clin Infect Dis 2011; 52:1061-1068
3. Knobel et al. IAS 2013; Abstract MOAB0205
4. Tanchaweng et al. IAS 2013. Abstract MOPDB0103
5. Cotter et al. IAS 2013. Abstract MOPE077
6. Maalouf et al. IAS 2013. Abstract WEAB0204
7. Hoy et al. IAS 2013. Abstract WELBB05
8. Bedimo et al. IAS 2013. Abstract WEPE512
9. Athoff et al. IAS 2013. Abstract MOPE082
10. Mandalia et al. IAS 2013. Abstract MOPE081
11. Gilleme et al. IAS 2013. Abstract WEPE0202
12. Lee et al. IAS 2013. Abstract WEAB0104
13. Dravid et al. IAS 2013. Abstract WEAB0201
14. Hamzah et al. IAS 2013. Abstract MOPE076
15. Freiberg. JAMA Intern Med. 2013 Apr 22;173(8):614-22
16. Hsue et al. IAS 2013. Abstract WEAB0206
17. Hsue et al. J Am Heart Assoc. 2012 Apr;1(2)
18. Lai. IAS 2013. Abstract MOPE068
19. Young et al. IAS 2013. Abstract MOPE070
20. Moore et al. PLoS One. 2011;6(7):e21843
21. Overton et al. Clin Infect Dis. 2013 May;56(10):1471-9
22. Spagnuolo et al. IAS 2013. Abstract WEPE504.