Metabolic and Adverse Events Report from the 13th CROI

By Graeme Moyle, MD, and Brian Boyle, MD

The 13^th Conference on Retroviruses and Opportunistic Infection (13^th CROI) provided new insights into differences between antiretroviral (ARV) agents with regard to their likelihood of causing adverse events and impact on metabolic parameters. The most important studies are summarized in this article.

PI Association with Myocardial Infarction (MI)

Updated results of the 23,000 patient D:A:D:S study were reported including data through February, 2005. (N Friis-Møller , Abstract 144) This study is one of the largest examining ARV toxicity, and it has over 94,000 person-years of follow-up with 345 myocardial infarction (MI) events having occurred during that time.  The average exposure to ARV therapy at the time of the analysis included approximately 6 years of exposure to nucleoside analogs (NAs), three years cumulative exposure to protease inhibitors (PIs) and one-year cumulative exposure to non-nucleoside reverse transcriptase inhibitors (NNRTIs).

The relative risk of MI per cumulative year on combination ARV therapy was 1.16, consistent with previous analyses of D:A:D:S. The relative risk of MI per year of PI exposure was also 1.16, and the relative risk remained significant in an evaluation of individuals who had only received PIs but never NNRTIs.  Adjustment by baseline lipids reduced the relative risk to 1.1, which, while remaining a significant association with cumulative years of PI exposure, suggests that at least some of the impact of the PIs on risk of MI was mediated through the impact of these agents on lipids.

There was no significant association of MI with cumulative years of exposure to NNRTIs, regardless of whether the individual had had prior PI exposure. (Figure 1). Since this report focused on the class effects on PIs and NNRTIs, details of individual agents within each class were not described.

The release of these data is likely to heighten interest in the relative impact of specific combinations of antiretrovirals or individual agents on the lipid profile.

Figure 1. D:A:D:S Study: Effect of PI and NNRTI on MI Risk

Lopinavir/ritonavir (LPV/r) in Healthy Volunteers

The effect of lopinavir/r (LPV/r) [Kaletra] on endothelial function and lipids in 6 healthy volunteers was reported. (J Grubb, Abstract 742). Four weeks exposure to LPV/r did not affect forearm blood flow and was associated with improvements in endothelial and non-endothelial dilatation and nitrogen oxide production.  These are consistent with some limited data in persons with HIV infection where similar improved flow mediated dilatation has been observed. These data would indicate that the impact of protease inhibitors on cardiovascular risk is not mediated to endothelial dysfunctions.

LPV/r led to increases in HOMA and fasting insulin levels, indicating relative insulin resistance, and, as summarized in Figure 2, had a significant impact on lipids.

Consistent with other studies in healthy volunteers, LPV/r leads to elevations in triglycerides and to a lesser extent cholesterol.  No impact on HDL cholesterol is observed.

Figure 2. Changes in Lipids in Healthy Volunteers on LPV/r

The absence of an impact on the endothelium was further supported from longitudinal data in which infected and uninfected individuals screened for carotid intima media thickness over a three-year follow up. One third of the individuals in the study were receiving a PI.  The design of the study was of a case control nature. No impact on HIV infection or protease inhibitor use on the progression rates of carotid intima media thickness was observed. (J Currier, Abstract 145).

BMS 089

Of particular interest from a toxicity standpoint was the outcome of the BMS 089 study which compared the use of ritonavir-boosted atazanavir (ATV)/r (300/100 mg) with unboosted ATV [Reyataz] (400 mg) with each combined with lamivudine (3TC) [Epivir] and extended release stavudine (d4T XR) [Zerit XR] as initial therapy. (N Malan N, Abstract 107LB).

The rate of drug-related adverse events of moderate or higher intensity was similar between the two groups.  However, adverse event-related discontinuations occurred more commonly in the ATV/r group (8%) as compared with the ATV alone arm (1%).  This was primarily related to persistent clinical signs of hyperbilirubinemia.  Scleral icterus of any grade was reported in 23 percent of ATV/r recipients but only 13 percent of ATV alone recipients.  These rates of icterus are higher than previously reported.

When considering metabolic changes, these again favored the unboosted approach. (Figure 3. The mean increase in total cholesterol was 15% for ATV/r as compared with 6% for ATV alone (p<.01), with 16% of ATV/r and 11% of ATV recipients experiencing a least a  1 grade increase in National Cholesterol Education Program (NCEP) total cholesterol category. 

Changes in triglycerides were a mean increase of 26% with ATV/r versus a 3% decline with ATV alone (p< 0.01).

These data suggest that given the similar efficacy of these two ATV-based approaches, the use of unboosted ATV is particularly attractive for individuals at high risk of metabolic disturbances.  As d4T XR has never been commercialized, the broad applicability of these data to current practice is challenging.  Clearly ATV must be administered with RTV boosting with tenofovir (TDF) [Viread]-based regimens.  However, when ATV is combined with zidovudine (ZDV)[Retrovir]/3TC or abacavir (ABC)[Ziagen]/3TC it maybe co-administered unboosted. Further investigation of the relative metabolic effects of these more current combinations is now required.

Figure 3. BMS 089: Lipid Changes to Week 48 

ACTG 5095

Investigation of the relative contributions of efavirenz (EFV) [Sustiva] and ABC to metabolic changes observed during initial therapy with ZDV/3TC was reported from the ACTG A5095 study. This study compared regimens of ZDV/3TC with either ABC, EFV or ABC plus EFV.  Comparisons of all three groups were possible through week 24 at which time the ZDV/3TC/ABC group was discontinued due to relatively poorer efficacy. 

The median changes in lipid parameters by intent to treat analysis for the ZDV/3TC/ABC, ZDV/3TC/EFV and quadruple therapy arms were, respectively, for triglycerides -1, +7 and +18 mg/dl, for total cholesterol +5, 23, 29milligram/dl, for HDL-c 5, 10, 10 mg/dl and LDL-c 1, 9, 14 mg/dl.

Changes in triglycerides and LDL cholesterol significantly favored the ZDV/3TC/ABC arm (p < 0.05).  Changes in lactate, glucose and insulin resistance by the homeostasis model did not differ between groups.

Follow-up through week 96 did not reveal any significant contribution from ABC to changes in these metabolic parameters relative to the ZDV/3TC/EFV group. Individuals in the ZDV/3TC/ABC group who intensified treatment with either EFV or TDF after week 24 were observed to have significantly greater changes through week 96 in the EFV group in total cholesterol (+34 verses -2 mg/dl), HDL-c (+4 vs. -2 mg/dl), and LDL-c (+23 vs. +5mg/dl) but not triglycerides (+32 vs. +17mg/dl), respectively.

Overall, these data suggest that ABC plays a limited role in metabolic changes during ARV therapy relative to the effects observed with EFV or ADV/3TC.  The NNRTI-based regimen tends to be associated with greater increases in both HDL and LDL cholesterol.  The limited impact of TDF on lipids, reported in previous studies, was also observed in the intensification phase of this study. (C Shikuma C, Absract 746).

Treatments for Fat accumulation and Lipoatrophy

Several studies evaluated the potential to treat fat accumulations using insulin sensitizing agents such as metformin and rosiglitazone. In a randomized, placebo-controlled study comparing the effects of metformin (1000 mg BID), rosiglitazone (4 mg QD), both or placebo in 105 individuals with elevated waist-hip ratio or waist circumference plus evidence of insulin resistance indicated that over 16 weeks of therapy there were modest improvements in insulin sensitivity with each of the three interventions, the largest impact be observed in the combined arm. (K Mulligan K, Abstract 147). No significant impact on visceral fat mass was observed.  However, some trends were noted towards modest increases in subcutaneous and limb fat with rosiglitazone and decreases in these parameters with metformin.

Figure 4. ACTG 5082: 16 week comparison of metformin and rosiglitazone  

This unfavorable effect of metformin on peripheral limb fat was confirmed by a second study using 1500 milligrams of metformin BID in individuals with elevated waist circumferences but without hyperinsulinemia. (R Kohli, Abstract 148).

The potential use of the Ppar-gamma agonist pioglitazone for the management of lipoatrophy was reported in a 130 patient, randomized, double-blind, placebo-controlled trial.  Participants in the study had good biological control of HIV (CD4 > 200 cells/mm³ and HIV RNA <400 copies/ml) on stable ARV therapy for a minimum of six months.  Randomization was to 30mg of pioglitazone once-daily or placebo. The groups were well matched at baseline and all patients were followed by DEXA and CT scanning for 48 weeks.

Limb fat and limb skinfold thickness increased to a greater degree in individuals receiving pioglitazone.  No differences in adverse events were reported. The magnitude of change in limb fat mass reported in the pioglitazone group is similar to the effects reported in studies involving switching away from thymidine analog based regimens.

Renal Toxicity

Five posters using a range of methodologies examined the issue of whether renal dysfunction occurs during TDF therapy. All the studies evaluated renal function using calculated creatinine clearances using either weight dependent Cockcroft-Gault or weight independent MDRD equations or both.

In patients receiving a combination of amprenavir (APV) [Agenerase]/r, ABC and additional NRTIs plus either EFV    (n = 38) or TDF (n=76) a modest, but statistically significant, difference in calculated GFR emerged over time.

Individuals entering the study had median calculated GFRs of 107.4 and 108.2, in the EFV and TDF groups, respectively. By week 24, GFR had risen by 12.7 in the EFV group and declined by 9.9 in the TDF group.  At week 48, the changes in calculated GFR from baseline were -0.4 and -11.1.  Differences between the groups were statically significant at both these time points. 

A multivariable model indicated that, after adjusting for baseline GFR, the only predictor of a decline in GFR are was TDF use. (M Thompson, Abstract 777). Of note, however, is that the median GFR remained comfortably within the normal range in both groups throughout the study.

Other cohort studies found a range of associations with renal function abnormalities (renal insufficiency/reduced GFR, hypophosphatemia) in individuals receiving TDF. In particular, associations were reported with concomitant or prior use of nephrotoxic agents, such as amphotericin B, injection drug use, hypertension, advanced HIV disease and concomitant administration of didanosine (ddI) [Videx]. (J Guest, Abstract. 778; J Heffelfinger, Abstract. 779; H Crane, Abstract 780).

The problem with many of these cohort studies, however, is that they included relatively small numbers of individuals and there are likely to be significant and confounding channeling or ascertainment bias as to why individuals in the study were receiving TDF in specific clinical circumstances.

Large randomized, controlled trials in individuals with normal renal function at study entry have not found that there is a higher risk of renal toxicity with TDF than the comparator NA (e.g., ZDV or d4T), and these studies have followed individuals carefully for the development of hypophosphatemia or creatinine elevations or changes in GFR.

In the international expanded access program for TDF, involving 10,343 individuals and post marketing surveillance with 455,392 patient years of exposure, serious renal events were reported in 0.57 percent of individuals. 

Risk factors identified in this surveillance program included concomitant administration of nephrotoxic medication and low CD4 cell count. (M Nelson, Abstract 781). Taken together, the results of these studies suggest that clinically relevant renal toxic events are uncommon in individuals receiving TDF-containing regimens and that the frequency of these events may be further diminished by the avoidance of co-administration of TDF with nephrotoxic agents and didanosine.

02/21/06