Cardiovascular disease risk with integrase inhibitor-based antiretroviral therapy: A causal inference analysis in RESPOND
Bernard Surial, Bastian Neesgaard, Frédérique Chammartin, Lene Ryom, Kathy Petoumenos, Joanne Reekie, Roger Paredes, Robert Zangerle, Colette Smith, Charlotte Martin, Ferdinand Wit, Nadine Jaschinski, Cristina Mussini, Antonella d’Arminio Monforte, Antonella Castagna, Christina Carlander, Larysa Hetman, Elena Sohani, Clara Lehmann, Sean R Hosein, Jean Andre Van Wyk, Emiliano Bissio, Jens Lundgren, Andri Rauch, Huldrych F Günthard, Gilles Wandeler, Lauren GreenbergBackground
A previous RESPOND analysis suggested that integrase inhibitors (INSTIs) increase cardiovascular disease (CVD) risk compared with other antiretroviral therapy (ART). We re-analysed the data using target trial emulations examining the impact of methodological choices.
Methods
Treatment-naïve analyses included ART-naïve adults with detectable HIV viraemia and recent CD4 and HIV-RNA measurements. Treatment-experienced analyses included adults with suppressed HIV-RNA receiving INSTI-free ART. We estimated adjusted risk differences (RD) and risk ratios (RR) for CVD events comparing INSTI initiation or switching vs other ART using inverse-probability weighting and pooled logistic regression.
Results
Among 7111 treatment-naïve individuals, 95 CVD events occurred over a median of 62 months. RRs comparing INSTI initiation with starting other ART were 1.07 (95% confidence interval 0.29–4.93) at one year, 1.21 (0.41–3.91) at two years, and 1.87 (1.06–3.26) at six years. Corresponding RDs were 0.02% (-0.32–0.29), 0.09% (-0.43–0.53), and 1.03% (0.08–2.07). Among 22’921 treatment-experienced individuals, 800 CVD events occurred over a median of 37 months. RRs were 1.55 (1.16–2.00) at one year, 1.30 (1.07–1.62) at two years, and 0.93 (0.78–1.09) at six years. Corresponding RDs were 0.26% (0.08–0.47), 0.29% (0.03-0.58), and -0.22 (-0.66–0.25).
Conclusions
Switching to INSTIs was associated with a transient increase in CVD risk during the first two years among treatment-experienced individuals, whereas starting INSTIs in ART-naïve individuals was associated with a gradual increase over time, albeit based on few events. Absolute risk differences remained small, but the findings suggest a potential impact of INSTIs on the development of CVD.