Abacavir and heart disease: SMART study supports an abacavir-associated increased risk of cardiovascular disease

Simon Collins, HIV i-Base

Jens Lundgren from the INSIGHT research group presented an analysis of nucleoside toxicity and cardiovascular disease from the SMART treatment interruption study. [1]

This issue was one of the most discussed clinical topics of the meeting as GSK also presented an analysis from their trial database. [3]

In February 2008, the D:A:D study showed an increased risk of cardiovascular risk from current or recent use of abacavir – a finding that was unexpected and challenging for people skeptical of a cohort study identifying a new effect with an as yet unexplained mechanism. [4]

Cautious reactions to the D:A:D results looked for validation from other studies which were not allayed by GSK’s more limited dataset, originally published as a letter to the Lancet. [5]

The SMART researchers analysed patients in the continuous treatment arm of SMART, by use of NRTIs relating to the previous D:A:D findings: using abacavir (but not ddI) n=1019, using ddI (but not abacavir) n=643, and other NRTI combinations with netheir abacavir nor ddI (n=2882). Baseline characteristics of these three groups were similar, including common cardiovascular risks (~4% prior CVD, 40% current smokers, 35% ischemic abnormalities and 7% diabetic). Lipid lowering drugs and blood pressure medications were each used by just under 20% of patients. 15% patients had >5 cardiovascular risk factors.

In multivariate analyses adjusting for CVD risk factors, all four categories of cardiovascular disease defined by the group showed increased hazard ratios (HR) for abacavir compared to other NRTIs (see Table 1).

Table 1: Adjusted HR of cardiovascular event with abacavir use vs. other NRTIs in SMART

CVD category No. of events Adj. HR (95% CI)
Clinical and silent MI, stroke, surgery for coronary artery disease (CAD), and CVD death 70 1.8 (~1.1-3.2)
Clinical MI as defined in D:A:D 19 4.3 (1.4-13.0)
CVD, major, expanded version (major CVD plus peripheral vascular disease, congestive heart failure (CHF), drug treatment for CAD, and unwitnessed deaths). 112 1.9 (1.0-3.1)
CVD, minor (CHF, peripheral vascular disease or CAD requiring drug treatment). 58 2.7 (1.3-2.9)

Importantly, the results were similar when patients receiving tenofovir were used as reference group and when the approximate 10% of patients with events in both D:A:D and SMART databases were excluded from the analysis.

The SMART study also showed a strong association between elevated levels of some inflammation biomarkers with levels of viral load rebound and risk of serious event. In this analysis, patients using abacavir had D-dimer and IL-6 that were 27% and 16% higher at study baseline than patients in the reference group using other NRTIs (both p=0.07).

These levels could have been higher for reasons unrelated to abacavir use and will need to be examined in a study looking prospectively at changes in these biomarkers in patients starting abacavir. Interestingly, the HEAT study from GSK reported reductions in hs-CRP and IL-6 in both the abacavir/3TC and tenofovir/FTC over 48 and 96 weeks with no differences seen between the two arms. [2]

Similar to D:A:D, the clinical significance from abacavir use was greatest in patients with the highest underlying cardiovascular risk factors. Those patients with five or greater cardiovascular risks or ischemic abnormalities on ECG showed three-fold increased risk from using abacavir compared to other NRTIs (both HR 3.1).

Earlier in the conference, GSK, reported that their retrospective meta analysis from 54 phase 2 and 3 abacavir registrational studies did not find an association between cardiovascular events and either abacavir or non-abacavir use. [3]

While this was important from a regulatory perspective – any safety signal requires a company to look at their own dataset – the limitations of both this database and the presented analysis were unlikely to resolve the concerns highlighted by D:A:D and SMART.

Of the 54 trials, only 13 were randomised for abacavir use, 33 included abacavir in background regimens and 8 did not include abacavir. Just over 14,000 adults and 500 children were included. Events were identified by a search for cardiovascular related events and rates in naive and experienced patients were calculated per 1000 person years.

No differences were seen in the relative rates by abacavir use for any cardiovascular event (RR=0.59; 0.35-1.01; p-value=0.055) or any MI (RR=0.863; 0.40-1.86; p=0.706).

Myocardial infarctions (MI) were only identified in 16 patients using abacavir (10 using non-PI and 6 on PI-containing regimens. Of the 11 MIs in the non-abacavir group, all used PI-based regimens except for one patient using an NNRTI-based combination.

Several limitations were raised concerning this data. Firstly, that there were too few events to have statistical power to detect an association either way. Many cardiovascular risks were not recorded at baseline, including smoking status, hypertension, HDL and LDL. Patient numbers were much lower (~7000 and 4500 PYFU in the abacavir and non-abacavir groups), and importantly median follow-up time was less than one year.

By comparison, D:A:D included 33,000 patients who needed to be followed for seven years (160,000 PYFU) until there was sufficient power to make associations to a single-drug effect.

Secondly, patients in clinical trials are and were generally younger, healthier, with lower cardiovascular risks. Interestingly, GSK did not present an analysis relating to the comparator regimens used in these studies, which were PI-based, and therefore already carried a higher risk of CVD.


The significance of these results from the SMART study, which are already published as a fast track paper in the 12 September edition of AIDS [6], is that they support the earlier D:A:D findings in two ways. They report a similar association between current or recent abacavir use and cardiovascular disease; and they found that the clinical impact was most significant in patients with highest underlying CVD risk.

BHIVA guidelines have, for several years, included the recommendation to routinely assess CVD risk using Framingham calculators on first diagnosis, prior to starting treatment and annually thereafter. Taken together, the D:A:D and SMART results suggest that for patients at the highest CVD risk (>20% 10-year Framingham), abacavir only be used when alternative options are not available.


  1. Lundgren J, Neuhaus J, Babiker et al. Use of nucleoside reverse transcriptase inhibitors and risk of myocardial infarction in HIV-infected patients enrolled in the SMART study. Late breaker abstract THAB0305.
  2. Smith K, et al. Similarity in efficacy and safety of abacavir/lamivudine (ABC/3TC) compared to tenofovir/emtricitabine (TDF/FTC) in combination with QD lopinavir/ritonavir (LPV/r) over 96 weeks in the HEAT Study. 17th IAC Mexico City, 2008. Poster LBPE1138.
  3. Cutrell A et al. Is abacavir (ABC)-containing combination antiretroviral therapy (CART) associated with myocardial infarction (MI)? No association identified in pooled summary of 54 clinical trials. Oral abstract WEAB0106.
  4. D:A:D Study Group, Sabin CA et al. Use of nucleoside reverse transcriptase inhibitors and risk of myocardial infarction in HIV-infected patients enrolled in the D:A:D study: a multi-cohort collaboration. Lancet. 2008 Apr 26;371:1417-26.
  5. Cutrell A et al. Abacavir and the potential risk of myocardial infarction. The Lancet. 2008 Apr 26;371:1413.
  6. Lundgren et al. Use of nucleoside reverse transcriptase inhibitors and risk of myocardial infarction in HIV-infected patients. AIDS 2008, 22: F17-24.

Links to other websites are current at date of posting but not maintained.