At CROI 2013, statins use in HIV management was the focus of a poster discussion and two late breaker oral presentations.
Most usefully, this poster session included an overview by Steven Grinspoon from Massachusetts General Hospital prior to the main discussion and a summary afterwards by Priscilla Hsue from University of California, San Francisco.
The introduction covered both general population and HIV positive issues. Data on statins comes largely from non-HIV populations where they are widely used to reduce cardiovascular risk by inducing plaque regression and reducing LDL cholesterol. However, the Jupiter study looked at primary prevention in the general population (with normal LDL: men were over 50 and women over 60) and reported clinical benefits based on their anti-inflammatory properties, that were independent of the lipid effect.
In a recent meta analysis of 76 studies in the general population, including >170,000 patients with 14,878 deaths, statin use significantly reduced cardiovascular-related mortality by about 20% and all cause mortality by 10%. Side effects included increased liver enzymes and CPK and were generally mild, but 17 randomised clinical trials reported a 9% increased risk of development of incident diabetes (OR 1.09; 95% CI 1.02-1.17, p=0.001).2
In the Jupiter study, patients had no prior cardiovascular disease and normal LDL (<130 mg/dL) but increased CRP (>2 mg/mL). CVD deaths were reduced by 50% using rosasuvastatin, with a slight reduction in deaths from cancer (0.4% vs 0.7%), but statistically significantly higher rates of new diabetes (3% vs 2.4%) were also reported, with the mechanism for this unknown.3
However, how and when HIV may have an impact on statin use and optimal use of statins in HIV positive people is unclear. While statins are widely recommended for lipid-lowering in HIV management, data in HIV positive people, especially long term safety data, is limited.
A review of the efficacy of lowering LDL in HIV positive people reported broadly similar results to HIV negative studies, even though this was slightly lower in some HIV studies (by approximately 2% less, but statistically significant). Also, although tolerability was good, rate ratios for some of the complications were also slightly (and significantly) higher compared to HIV negative studies.4
Some statins are commonly contraindicated (simvastatin and lovastatin) due to drug-drug interactions with some antiretrovirals and dose adjustment is often required with others.5 While rosuvastatin (10 mg) is more effective than pravastatin (40 mg) at lowering LDL and triglyrcerides in patients on protease inhibitors, drug levels were increased in HIV negative studies using atazanavir/ritoanvir.
Similarly, interesting preliminary data from patients on ART reported survival benefits from comorbidities (relative Hazard Ratio 0.33 (0.14, 0.76), p=0.009) in the 15% of patients on statins.6 Statins may have potential antiretroviral activity and may disrupt CCR5 and RANTES and decrease CCR2 expression in monocytes.
Four posters were also selected for discussion but although these are summarised below, all had methodological issues that highlighted the need for further research.
Line Rasmussen from Odense University Hospital presented results from a Danish observational study of HIV positive people starting ART from 1998 with undetectable viral load within 6 months, and statin use before and after a cormorbidity diagnosis (CVD, renal disease or diabetes mellitus), compared to the general population.7 A reduction in all cause mortality was only reported in patients with diagnosed comorbidities, and only in a censored model when VL >400 copies/mL (adj mortality RR 0.32, 95%CI 0.10-0.99; without censoring aMMR was 0.57, 95%CI 0.28-1.15).
No difference was seen without a comorbidity diagnosis. However, this was a retrospective study and statin prescription is likely to have been for a clinical indication (suggesting confounding factors), suggesting that the beneficial impact in this analysis may have been underestimated.
Henning Drechsler from the University of Texas Southwestern Medical Centre reported on the impact of statins on mortality and non-AIDS complications in >25,000 patients on ART in the Veteran's Association (VA) cohort, stratifying by virological suppression and choice of statin.8
Approximately 35% of patients used statins. There were 6435 deaths during a median follow up of 6.3 years (>25,000 PY). There were also 2199 cardiovascular events (acute myocardial infarction and cerebrovascular accident), 5011 malignancies, 3196 cases of chronic kidney disease (CKD), and 610 fragility fractures.
Adjusted multivariate analysis showed significant reductions in deaths with statin use, irrespective of viral suppression but with a greater benefit in patients using atorvastatin or rosuvastatin but no effect was seen on CVD (which was inexplicably increased), malignancy or fracture outcomes. As with the Danish study, confounding by clinical indication was suggested for the lack of effect seen in these results. The lack of mortality data in the VA cohort limited further understanding of the cause of death.
Vincenzo Spagnuolo from the San Raffaele Hospital, Milan, presented a retrospective longitudinal data on the incidence of Type-2 diabetes mellitus (DM) in 5380 HIV positive patients on ART (n=726 using statins) followed for 9.8 (IQR 4.4-14.9) years.9
Significant differences at baseline included median age of the statin users was slightly older (50.7 vs 45.7 years), with longer use of ART (14.1 vs 9.6 years) and significantly worse lipid profiles.
The overall crude incidence rate (IR) for new DM was 2.82 per 1000 PYFU (95%CI 2.74-2.91). Approximately 14% patients used statins (81% rosuvastain, 18% pravastatin) for a median of 24.4 months (IQR 10.2-42.4). However DM occurred significantly less in 12/726 (1.7%) statin users (IR 1.32; 95%CI 0.68-2.18) compared to 150/4654 (3.2%) not statin users (IR 3.40; 95%CI 2.86-3.95); p = 0.020.
In multivariate analysis, older age, higher BMI, basline triglycerides and glucose and prior use of d4T or ddI were all significantly associated with increased risk of DM, with higher CD4 nadir and statin use being protective. Data on choice of statin and dose were not available.
This study reported very low rates of DM, below general population incidence, but only 5% of participants had BMI >30 kg/m2. This was raised in the question session, including how DM was diagnosed.
In contrast to this study, the final poster presented by Kenneth Lichtenstein reported an increased risk of DM with statin use in 4962 patients at 9 U.S. clinic sites in the HIV Out Patients Study (the HOPS cohort) who were prospectively followed from 2002-2011.10
Of these patients, 590 received statins during follow-up of a mean 49.4 months (IQR 21.7-78.1). New DM was diagnosed in 355 patients (crude incidence of 3.06 per 100 person-year while prescribed statins vs. 1.16 otherwise). In multivariate analysis, per year of statin use was associated with a 10% increased risk of incident DM (HR 1.10; 95%CI 1.01, 1.30), p=0.038. Other significant factors included older age, Hispanic race, BMI > 30 kg/m2 but not with individual ARVs or protease inhibitor use.
The conclusion from this study was that this showed the importance of routinely monitoring glucose in HIV positive patients and that these findings would be unlikely to reduce appropriate statin use for hyperlipidaemia and statin use should not be avoided when clinically warranted.
Questions relating to the methodology of this study included whether patients on statins would routinely be monitored more frequently for glucose tests, that higher diagnosis might relate to use of single elevated glucose results and that the U.S. background population risks, including BMI might be higher compared to Italians.
The full panel discussion emphasised the importance of separating use when clinically indicated from use when not clinically indicated (in the belief there are other benefits). While there are no data to support not using statins when indicated, the data are currently unknown for non-indicated use and this was not answered by any of the retrospective studies discussed in the session. It also included a comment that the assumption that inflammation is on the causal pathway for cardiovascular disease may be incorrect as there is little data supporting this in the general population.
In summary, Priscilla Hsue emphasised the strong associated between even modest LDL reductions in the general populations reducing major coronary events and mortality, with the impact on risk correlating to the degree of LDL reductions.11,12
HIV studies are largely small and observational compared to large randomised studies in the general population. In general population studies, statins are largely beneficial even in patients with diabetes, with similar benefits irrespective of DM status.13 However, the 9% increase risk of incident diabetes reported has a low absolute risk: requiring 255 patients (95%CI 150-852) to be treated for four years to produce on new case of DM.14
Finally, there were two late breaker oral presentations to the main conference.
In the first, Grace McComsey from Case Western Reserve University and colleagues reported early 24-week results on impact of rosuvastatin on cellular activation markers in treatment experienced patient on ART (~ median 6 years; IQR 3-10), but with normal LDL cholesterol (<130 mg/dL).15
This is a 96-week study looking at the impact of rosuvastatin on cardiovascular and bone health in 147 adults randomised to rosuvastatin 10 mg or matching placebo. Entry criteria included evidence of immune activation and/or inflammation (defined as CD8+CD38+HLA-DR+ >19% and/or hsCRP >2 mg/L, respectively).
At 24-weeks, rosuvastatin only significantly reduced one out of five inflammation markers (Lp-PLA2, with no impact on IL-6 or hsCRP) and two out of six markers of monocyte activation (sCD14 and CD14dimCD16+TF+ monocytes). Rosuvastatin had no impact on a panel of markers of lymphocyte activation or coagulation.
Further results from this study are needed in this mixed population. Although LDL was normal, median BMI was 26 and >30 kg/m2 in 25% of participants, >60% are current smokers, ART experience is extensive with 50% of current PI-based treatment.
The second late-breaker presented results from a randomised phase 4 study, designed to show superiority of the more recently approved pitavastatin (approved 2010 in the UK) compared to pravastatin (available as generic for at least five years) based on the primary endpoint of greater reductions in LDL cholesterol at 12 weeks.16
Although pitavastatin has reduced potential for CYP-mediated drug-drug interactions, metabolism is glucuronidation via UGT1A3 and UGT2B7. Pitavastatin is only marginally metabolized by CYP2C9 and to a lesser extent by CYP2C8
The study was sponsored by Kowa Pharmaceuticals, manufacturers of pitavastatin, and enrollment criteria included dyslipidaemia defined as LDL-C between 130-220 mg/dL and triglycerides ≤400 mg/dL.
Mean (SD) percentage changes in LDL at week 12 were -31.5 mg/dL (15.3) vs -20.9 mg/dL (15.2), mean difference -9%, p<0.001. Similarly significant differences were seen in reductions in Apo B and total cholesterol, but no difference were seen for HDL or triglyceride levels. No results were given in terms of the percentages of patients achieving target lipid levels. Tolerability results were similar for both drugs.
Taken together, the posters in this session did not addto what is already a confusing field, due to problems in the study designs.
The first study (Rasmussen) found no effect for primary prevention, but did find an effect in those with a prior co-morbidity. This would be slightly at odds with the general population where there is a (albeit relatively small) benefit of statin use in thegeneral population.
The second study (Drechsler) didn't find any association with CVD -- but did adjust for LDL as a time-updated covariate. This is on the causal pathway between statin use and MI development, so adjusting for this as a time-updated covariate is likely to remove the very effect you're looking for (like adjusting for time-updated CD4 count when assessing the effect of ART on mortality). This makes interpretation difficult. For example, they report that statins were protective of mortality in the univariate analysis which would not normally be expected as people who receive statins generally have multiple risk factors for CVD, many of which are also risk factors for mortality (high BMI, smoking, family history of CVD,older age, male gender). A study that doesn't find a negative association of mortality with statins in unadjusted analysis is unusual.
Most studies also have too limited data to be able to remove confounding in multivariate analyses. The same issue is relevant for the Spagnuolo study (as well as the issues about under-ascertainment).
The final study (Lichtenstein) is most convincing, although whether statins had causal relationship to the increased DM risk, or were just a consequence of unmeasured confounding), remains unclear.
Unless stated otherwise, references are to the Programme and Abstracts for the 20th conference on Retroviruses and Opportunistic Infections (CROI), 3-6 March 2013, Atlanta.
Links to other websites are current at date of posting but not maintained.