An analysis of the global START trial has found that beginning treatment with a regimen including Sustiva (efavirenz) is associated with an increased risk of suicidal behaviours. This finding is particularly reliable compared with those of other studies that have examined the drug’s link with suicide because it comes from a randomised controlled trial, which allowed the study’s authors to better control for variables that might have otherwise clouded their findings.
Sustiva (efavirenz) is included in the single-tablet regimen Atripla (efavirenz/tenofovir disoproxil fumarate/emtricitabine). The antiretroviral (ARV) has fallen out of favour in the US because of its troublesome side effects, which include nightmares. Among all the single-tablet regimens manufactured by Gilead Sciences that include tenofovir disoproxil fumarate, Atripla is the only one that the company did not revise by swapping in tenofovir alafenamide, the updated, safer version of tenofovir.
The randomised controlled START trial, published in 2015, included 4,684 HIV-positive individuals from around the world who had not yet started ARV treatment and had a CD4 count above 500. A total of 271 (5.8%) of them had a psychiatric diagnosis. The participants were randomly assigned to begin ARVs immediately or to begin on a deferred basis, waiting until their CD4 count dropped to 350 or below, until they developed Aids or other serious illnesses or until they met qualifications for starting treatment according to local guidelines.
The trial prespecified that 3,515 of the participants (75%) would receive Sustiva as part of their ARV regimen when they began treatment – 40% of those with a psychiatric diagnosis were prespecified to receive Sustiva, compared with 77% of those without such a diagnosis. Among those prespecified to receive Sustiva, 3.1% had a pre-existing psychiatric diagnosis, compared with 13.9% of those prespecified for other ARVs. A respective 5.2% and 16.8% of each group used psychotropic medications.
Twenty-eight members of the immediate treatment group and 25 members of the deferred treatment group reported suicidal behaviours during an average follow-up of 3.2 years. This translated to rates of 0.39 and 0.34 per 100 cumulative years of follow-up, respectively. This meant that, overall, there was no statistically significant difference in the rate of suicidal behaviours between the two groups, meaning any apparent difference might have been driven by chance.
The researchers, led by Alejandro Arenas-Pinto at the Medical Research Council Clinical Trials Unit, University College London, looked at the rate of suicidal behaviours among those randomised to start treatment immediately, looking at the entire study period, and compared that rate among those in the deferred treatment arm only among the period before those individuals started ARVs.
Among those prespecified to receive Sustiva, 18 participants in the immediate treatment group reported suicidal behaviour, for a rate of 0.36 per 100 cumulative years. (The researchers excluded from their analysis one individual in this group who reported suicidal behaviour but who did not actually end up taking Sustiva.) By comparison, four individuals in the deferred treatment arm who were prespecified to take Sustiva reported suicidal behaviours during the period before they started treatment, for a rate of 0.10 per 100 cumulative years.
The study authors calculated that taking Sustiva was associated with a 3.31-fold greater risk of suicidal behaviours.
By comparison, among those prespecified to receive regimens that did not include Sustiva, the rate of suicidal behaviours was higher, at a respective 0.56 and 0.66 per 100 cumulative years in the immediate and deferred treatment arms. (Which makes sense, since this group had a higher rate of psychiatric diagnoses.) But the difference between these rates was not statistically significant.
Among the 109 participants who had a prior psychiatric diagnosis and were prespecified to receive Sustiva, none in the deferred group reported suicidal behaviour during the period before they started treatment, compared with six of those in the immediate treatment arm after they’d started Sustiva (for a rate of 2.7 per 100 cumulative years).
The strongest predictor of suicidal behaviour among those in the immediate treatment group was a prior psychiatric diagnosis, which was associated with a 12.5-fold increased risk of such behaviour among those prespecified to receive Sustiva and a 9.3-fold increased risk among those prespecified to receive other ARVs. However, whether they were prespecified to receive Sustiva or other ARVs did not affect this increased risk in a way that was statistically significant.
Among those prespecified to receive Sustiva, heavy alcohol use was associated with a 4.6-fold increased risk of suicidal behaviour and recreational drug use was associated with a 2.6-fold increased risk, while each relative addition of 10 years of age decreased the risk by 49%.
Background: Randomized trials have shown increased risk of suicidality associated with efavirenz (EFV). The START (Strategic Timing of Antiretroviral Treatment) trial randomized treatment-naive human immunodeficiency virus (HIV)–positive adults with high CD4 cell counts to immediate vs deferred antiretroviral therapy (ART).
Methods: The initial ART regimen was selected prior to randomization (prespecified). We compared the incidence of suicidal and self-injurious behaviours (suicidal behavior) between the immediate vs deferred ART groups using proportional hazards models, separately for those with EFV and other prespecified regimens, by intention to treat, and after censoring participants in the deferred arm at ART initiation.
Results: Of 4684 participants, 271 (5.8%) had a prior psychiatric diagnosis. EFV was prespecified for 3515 participants (75%), less often in those with psychiatric diagnoses (40%) than without (77%). While the overall intention-to-treat comparison showed no difference in suicidal behavior between arms (hazard ratio [HR], 1.07, P = .81), subgroup analyses suggest that initiation of EFV, but not other ART, is associated with increased risk of suicidal behavior. When censoring follow-up at ART initiation in the deferred group, the immediate vs deferred HR among those who were prespecified EFV was 3.31 (P = .03) and 1.04 (P = .93) among those with other prespecified ART; (P = .07 for interaction). In the immediate group, the risk was higher among those with prior psychiatric diagnoses, regardless of prespecified treatment group.
Conclusions: Participants who used EFV in the immediate ART group had increased risk of suicidal behavior compared with ART-naive controls. Those with prior psychiatric diagnoses were at higher risk.
Alejandro Arenas-Pinto, Birgit Grund, Shweta Sharma, Esteban Martinez, Nathan Cummins, Julie Fox, Karin L Klingman, Dalibor Sedlacek, Simon Collins, Patricia M Flynn, William M Chasanov, Eynat Kedem, Christine Katlama Juan Sierra-Madero, Claudia Afonso, Pim Brouwers, David A Cooper