144
Exposure to PI and NNRTI and Risk of Myocardial Infarction: Results from the D:A:D Study
Nina Friis-Møller*1, P Reiss2, W El-Sadr3, A D'Arminio Monforte4, R Thiébaut5, S De Wit6, R Weber7, E Fontas8, M Law9, A Phillips10, and and the D:A:D Study Group
1Copenhagen HIV Prgm, Hvidovre, Denmark; 2HIV Monitoring Fndn, Academic Med Ctr, Univ of Amsterdam, The Netherlands; 3CPCRA, Columbia Univ, Harlem Hosp, New York, NY, US; 4ICONA, L Sacco Hospital, University of Milan, Milan, Italy; 5Bordeaux Univ Hosp, INSERM U593, France; 6Ctr Hosp Univ St Pierre, Brussels, Belgium; 7Univ Hosp, Zürich, Switzerland; 8Ctr Hosp Univ Nice, Hosp de l'Archet, France; 9Australian HIV Observational Database, Natl Ctr in HIV Epidemiology and Clin Res, Sydney; and 10Royal Free and Univ Coll Med Sch, London, UK
Background:
Prior data from the D:A:D study showed a
progressive increase in the risk of myocardial infarction (MI) with longer
exposure to combination ART (cART). Here, we
investigate if this association differs by drug class and the possible
mechanisms for any association.
Methods:
Observational study of more than 23,400 HIV-infected patients from 11
cohorts in Europe, Australia,
and the United States.
Results are based on follow-up to February 2005. Incidence rates of first
prospective MI (/1000 person-years of follow-up), and relative rates (RR) for
factors associated with MI from Poisson regression models are reported. We used
2 different models to assess the association with years on cART,
and the association with years on protease inhibitors (PI) and non-nucleoside
reverse transcriptase inhibitors (NNRTI), separately. Both models are adjusted
for demographic factors and risk factors for MI.
Results:
By 2005, 345 patients had experienced an MI over 94,469 person-years
(3.65/1000 person-years). Increased time on cART was
associated with a risk of MI (adjusted RR 1.16/year of exposure [95%CI 1.09 to
1.23]). The risk of MI has decreased over calendar time (RR for 2003-2004 vs 1999 0.50 [0.32 to 0.77]); an effect that is removed by
adjusting for latest lipid levels (RR for 2003-2004 vs
1999 0.82 [0.49 to 1.37]). Total exposure to the PI drug class was 72,846
person-years (30,198 person-years in patients with no NNRTI exposure) and to
the NNRTI drug class 52,457 person-years (9808 person-years in patients with no
PI exposure). MI incidence increased from 1.53/1000 person-years in those not
exposed to PI to 6.01/1000 person-years in those exposed for >6 years
(RR/year of exposure: 1.17, 1.12 to 1.23). The incidence also increased
slightly with NNRTI exposure (RR/year:
1.07, 1.00 to 1.14). After adjustment for the other drug class and other
known risk factors for MI, the relative rate per year of PI exposure was 1.16
(1.10 to 1.23, p = 0.0001), while for
NNRTI it was 1.05 (0.98 to 1.13, p =
0.17). These associations persisted, although were reduced slightly, after
controlling for years of NRTI use. Total cholesterol (1.24/mmol/L [1.14 to
1.34]), HDL cholesterol (0.76/mmol/L [0.53 to 1.08]), and triglycerides
(1.42/doubling [0.85 to 2.38]) were associated with the risk of MI; adjustment
for these reduced the effect of PI and NNRTI exposure to 1.10 [1.03 to 1.17]
and 1.01 [0.93 to 1.10], respectively.
Conclusions:
Increased PI exposure is associated with an
increased risk of MI, which is partly explained by dyslipidemia. Conversely,
although there were fewer years of experience, we found no evidence that
increased NNRTI exposure is associated with risk of MI.
|
