do Thymidine Analogues, Abacavir, Didanosine and Lamivudine Contribute to the Risk of Myocardial Infarction? The D:A:D Study
C Sabin1, S Worm2, R Weber3, W El-Sadr4, P Reiss5, R Thiebaut6, S DeWit7, M Law8, A Phillips1, Jens Lundgren*2,9, and the D:A:D Study Group
1Univ Coll London and Royal Free Hosp, UK; 2Univ of Copenhagen, Denmark; 3Univ Hosp Zurich, Switzerland; 4Colombia Univ, New York, NY, US; 5Academic Med Ctr, Amsterdam, The Netherlands; 6Inst for Publ Hlth, Epi and Devt, Univ Victor Segalen Bordeaux 2, France; 7St Pierre Univ Hosp, Brussels, Belgium; 8Univ of New South Wales, Sydney, Australia; and 9Rigshospitalet, Copenhagen, Denmark
Background: Attention has focused mainly on the role
of protease inhibitors and risk of myocardial infarction (MI). Despite the
known association of thymidine analogues (TA) (zidovudine and stavudine) with
dyslipidaemia and insulin resistance, the question of whether they may also be
associated with an increased risk of MI remains unanswered.
Methods: D:A:D is a collaboration of 11 prospective
cohorts; includes data on 33,347 patients, of whom 517 developed a MI during
157,912 person-years. Poisson regression assessed the effect of cumulative,
recent (current or within last 6 months) and past (>6 months ago) use of TA,
abacavir (ABC), didanosine (ddI), and lamivudine (3TC) after adjustment for
demographic, known cardiovascular (CV) risk factors, cohort, calendar year,
body mass index (BMI) and use of other antiretrovirals (ARV).
Results: We observed 89% of follow-up and 98% of MI
among patients exposed to ≥1 NRTI. Neither cumulative nor recent use of
the 2 TA or 3TC was associated with risk of MI. However, cumulative use of ABC
and ddI were each associated with an excess risk of MI (relative risk/year of
use = 1.14 [95%CI, 1.08 to 1.21, p <0.01; and 1.06, 1.01 to 1.12, p
= 0.03; respectively]). In a model including recent use, recent use of ABC and
ddI predicted risk of MI (1.90, 1.47 to 2.45, p <0.01; and 1.49, 1.14
to 1.95, p <0.01; respectively), but not cumulative use. In other
models, recent, but not past, use of ABC and ddI predicted risk of MI. Rates of
MI (/1000 person-years) for patients with recent versus no recent use of ABC
and(/) ddI were 3.3/2.1 vs 1.2/1.5, 9.8/9.1 vs 7.1/7.5 and 31.3/20.8 vs
11.2/14.9, in those with low, moderate and high predicted Framingham CV risk,
respectively (p >0.2 for interaction). At MI diagnosis, the CV risk
profile was comparable irrespective of type of ARV regimen used. The risks of
MI associated with recent ABC and ddI use were seen regardless of duration of
use and remained after adjustment for HIV-RNA levels, CD4 count, dyslipidaemia
and other metabolic factors. Preferential use of ABC and ddI in patients with
an a priori elevated CV risk appears not to explain the findings.
Conclusions: TA were not associated with risk of MI.
Unexpectedly, recent use of ABC and ddI were associated with increased risk of
MI, by 90% and 49%, respectively. The excess risks of MI associated with ABC
and ddI use were most pronounced—in absolute terms—in patients with high
underlying CV risk. Although it is impossible to rule out bias as an
explanation, if these associations are causal, the unknown biological
mechanism(s) appears reversible upon cessation of these drugs.