No increased risk of diabetes with integrase inhibitors in people starting HIV treatment

Among people with HIV taking antiretroviral therapy for the first time, integrase inhibitors are not associated with an increased risk of incident diabetes, according to a French study recently published in the Journal of Antimicrobial Chemotherapy. This result was observed after comparing HIV-treatment regimens in which the third agent was either an integrase inhibitor, a protease inhibitor or a non-nucleoside reverse transcriptase inhibitor (NNRTI).

As many previous studies have highlighted an association between the development of diabetes and the duration of antiretroviral therapy with certain drugs, the investigators wanted to verify if this was specifically seen with integrase inhibitors. This question is of particular interest as it has also been reported that integrase inhibitor-based regimens are associated with greater weight gain than NNRTI or protease inhibitor-based regimens, and as weight gain is often associated with diabetes. A US study last year did find a link between integrase inhibitors and diabetes.

The study

Participants’ medical records were collected retrospectively from the Dat’AIDS cohort study, a collaboration of 23 HIV clinics in France and its overseas territories. The study included adults living with HIV who started their very first antiretroviral therapy between 2009 and 2017.

Antiretroviral therapy was defined as two nucleoside reverse transcriptase inhibitors and a third drug from another antiretroviral family: an integrase inhibitor, a boosted protease inhibitor, or an NNRTI. Patients who had been diagnosed with diabetes before starting antiretroviral therapy were excluded from the study.

Incident diabetes was defined as a notification of new diabetes in a participant’s medical history, a glycated haemoglobin (or HbA1c, a haemoglobin linked to sugar) level over 7.5%, or starting diabetes treatment. Routine blood sugar measurements were not taken into account as it was unknown if patients had been fasting or not at the time they were performed.

Baseline data

Among the 19642 participants, there were 5606 women, 13,654 men and 202 transgender people.

Some risk factors known to be linked to diabetes were reported:

• In people with the highest body mass index (BMI, >30), more started with a protease inhibitor (52%) than an NNRTI (29%) or integrase inhibitor (19%). At first sight, this would seem surprising, as protease inhibitors are associated with metabolic issues, but the results may just reflect a limitation in therapeutic choices when patients began treatment.  
• In people with dyslipidaemia, more started with a protease inhibitor (53%) than an NNRTI (35%) or integrase inhibitor (12%).
• The same trend was found with hypertension: 53% of patients with hypertension started a protease inhibitor, 27% an NNRTI, and 20% an integrase inhibitor.

Antiretroviral choices were comparable in people without these risk factors. All of these variables were collected before the start of antiretroviral therapy.

The 3403 participants on an integrase inhibitor had a median follow-up time of 572 days. The integrase inhibitors taken were elvitegravir (35%), raltegravir (30%), dolutegravir (35%) and bictegravir (approved in 2018, less than 1%).

Diabetes and risk factors

A total of 265 cases of diabetes occurred in the study. Per third HIV medicine, these cases were distributed as below:

• 31/3403 (0.91%) of patients receiving an integrase inhibitor.
• 77/5601 (1.37%) of patients receiving a non-nucleoside reverse transcriptase inhibitor.
• 157/10458 (1.50%) of patients receiving a protease inhibitor.

In the researchers’ first analysis, incidence of diabetes increased with age, BMI, hypertension, having developed AIDS, starting treatment in an earlier period, female sex, hepatitis co-infection, treatment at an overseas HIV clinic rather than in mainland France, African or South American place of birth, and being heterosexual.

However, diabetes was less frequent in the context of active smoking, alcohol consumption and drug abuse. Also, surprising as this may sound, no difference was found with regards to dyslipidaemia.

In multivariate analysis, which took into account a range of other factors that may influence the results, no difference was found between the third agents in terms of diabetes occurrence. However, diabetes was more common in people starting in 2009-2011 than in later years. The geographical origin of participants (not being from Africa, South America or France) was also a protective factor. However, a BMI of >30 kg/m² and being older than 46 years were associated with an increased incidence of diabetes.

"Incident diabetes was found in this cohort at a lower rate than sometimes previously reported.

The date of diabetes notification was clearly established for 197 participants, out of the 19,300 for whom a follow-up time was available. This allowed the investigators to proceed with a survival analysis for this subset of participants in which the incidence of diabetes was 4.1/1000 person-years follow-up – which translates into finding 4.1 cases of diabetes a year in a clinic cohort of 1000 patients.

Again, as with the previous two models, no difference was found in incident diabetes according to the type of antiretroviral therapy taken by patients, whereas BMI >30 kg/m² and age over 37 years read out as risk factors for diabetes. Also, hypertension and having developed AIDS were significantly associated with the occurrence of diabetes.

Importantly, contrary to the multivariate analysis, this survival analysis did not find an association between the year of starting antiretrovirals and incident diabetes.

In their discussion, the authors highlight that incident diabetes was found in their cohort at a lower rate than sometimes previously reported, which might be explained by the reduction of antiretroviral medications’ toxicities over time.

The investigators emphasise that they did not observe an increased risk of incident diabetes with the use of integrase inhibitors. However, since integrase inhibitors have been associated with weight gain, these results need to be confirmed with longer follow-up.