INSTIs and cardiovascular events in the RESPOND cohort: early increase at six months normalises after two years

Kirk Taylor, HIV i-Base

New data from the international multi-cohort RESPOND study published in The Lancet HIV reports a link between cardiovascular disease (CVD) and integrase inhibitors (INSTIs). [1]

Over a  median follow-up of six years, the incidence of CVD for people on INSTI-containing regimens was approximately twice as high as for people not using INSTIs: at 8 vs 4 events per 1,000 person years. This association was greatest at six months but normalised after two years.

The RESPOND consortium collected data from on CVD events from 17 Australian and European cohorts. Baseline demographics of the INSTI-naive participants (n=29,340) included male (74%). female (25%). transgender (<1%); White (70%), Black (10%). The median age was 44 (IQR: 36 to 51) and recent CD4 counts were 524 (IQR: 357 to 715) cells/mm3. [1]

Retrospective data was included for 5 years prior to enrolment and prospectively updated annually thereafter. CVD events were centrally validated for those that occurred in the year prior to enrolment and thereafter. This included fatal and non-fatal myocardial infarction, strokes, and invasive coronary procedures (coronary angioplasty or stenting, bypass surgery and carotid endarterectomy).

Participants on INSTI-based regimens were taking dolutegravir (62%), elvitegravir (24%), raltegravir (24%) or bictegravir (6%). The study was not powered to detect associations for specific INSTIs.

Over a median follow-up was 6.16 years (IQR: 3.87 to 7.52) there were 748 events giving an incidence of CVD of 2.5%. Myocardial infarction was most common (40%), followed by stroke (30%) and invasive procedures (30%). The incidence rate across the cohort was 4.67 events (95% CI: 4.34 to 5.01) per 1,000 person years.

CVD incidence was reported at baseline 6, 12, 24, 36 and ≥36 months. At 6 months, CVD incidence increased from 4.19 (95%CI: 3.83 to 4.57) to 8.46 (95% CI: 6.58 to 10.71) events per 1000 person years for people on INSTI-based regimens.

However, there was no cumulative link to use of INSTI-based ART. Instead, CVD risk decreased after six months and was no longer significant at two years.

Baseline risk factors (smoking, hypertension, diabetes, kidney disease, smoking status, dyslipidaemia and age) were good predictors of CVD events during follow-up. However, the higher early association with INSTI-based ART remained after adjustment for traditional CVD risks above and HIV history including CD4 counts and HIV treatment.

The INSTI signal remained in several sensitivity analysis except when the model only included centrally adjudicated CVD events (n=145). In this case the adjusted incidence rate ratio dropped to 1.37 (95%CI: 0.89 to 2.12) and was no longer significant (p=0.22).

Unexplained mechanism for CVD risk

As an observational study, RESPOND is unable to identify causal links for CVD risk.

The early increase of risk means that slower progressing conditions (e.g. atherosclerosis) are unlikely to be at play. The authors also ruled out dyslipidaemia, hypertension, and immune reconstitution syndrome.

The accompanying editorial further reflects on the mechanism. It notably rules out INSTI-associated weight gain, as adjustment for BMI did not explain the increased CVD risk. [2]

Insomnia is linked to both INSTI use and CVD risk and warrants further investigation. Biomarker and switch studies have also noted complex changes of inflammatory profiles for INSTIs that may explain transient rises of CVD risk.

If these changes are short-lived, it is possible that modification of traditional CVD risk factors could reduce the signal at 6 months.


This INSTI-naive cohort was largely treatment experienced. Traditional risk factors for CVD were also significantly associated with risk of CVD events.

The results therefore describe an increased initial risk of CVD of the first six months that then steadily normalises over the next 18 months. The early risk though remained after adjusting for baseline CVD risk, including use of PIs and abacavir.

Established links between CVD and other HIV drugs include a cumulative association with indinavir, lopinavir, and darunavir and an association with recent use of abacavir in people at higher baseline CVD risk.

The discussion is not able to explained the observed short-term increase in CVD including a platelet mechanism (similar to abacavir). It also notes the lask of association between INSTIs and early CVD in regulatory phase 3 studies and calls for further research in larger studies.

INSTIs are recommended as first-line therapy in BHIVA guidelines as they have a high barrier to genetic resistance, rapidly lower viremia and restore CD4 counts. [3]

The RESPOND consortium was formed in 2017 with the earliest data from 2012 before INSTIs were recommended as first-line ARVs. Dolutegravir received FDA approval in 2013 and was the most frequently used INSTI in the current study. Bictegravir was approved 5 years later, and less data is available for this structurally similar INSTI. More data points are required to evaluate signals for specific INSTIs.

A meta-analysis of eight randomised trials investigated links between DTG and CVD. [4]

Serious adverse cardiovascular events were 0.7% for DTG and 0.4% for other ARVs (raltegravir or efavirenz). The increased relative risk of 1.6 was not significant and only one event was considered DTG-related. The majority of serious adverse events (19/23) were in people with underlying CVD risk factors.   

The authors suggest that studies of platelet function for people on INSTIs should be investigated to evaluate this as a potential mechanism. A poster from CROI 2019 reported a modest reduction of platelet function for HIV negative volunteers that received DTG for one week. [5] Further studies of platelet function in HIV positive people on INSTI-based regimens are required to explore this mechanism.

There are similarities to the abacavir and CVD risk reported by the D:A:D study. The risk for people on abacavir regimens was only in people that already had elevated CVD risk, largely driven by age. This is a major reason why the RCTs by GSK did not find a link. [6]

The earliest reports from D:A:D initially reported a signal for ART and then later by ARV class. Both these initial signals were misleading though because longer follow-up was needed to show associations to individual drugs rather than drug classes. RESPOND is not currently able to report on individual drugs.

Until then, the most optimistic outcome, as suggested in the accompanying editorial comment, is that a class effect is not confirmed in future studies. [2]


  1. Neesgaard B et al., Associations between integrase strand-transfer inhibitors and cardiovascular disease in people living with HIV: a multicentre prospective study from the RESPOND cohort consortium. The Lancet HIV. DOI:10.1016/S2352-3018(22)00094-7. (07 June 2022).
  2. Venter WDF et al. Integrase inhibitors hand us a new HIV clinical puzzle. The Lancet HIV. DOI:10.1016/S2352-3018(22)00122-9. (07 June 2022).
  3. BHIVA guidelines on antiretroviral treatment for adults living with HIV-1 2022 – consultation version. (09 June 2022).
  4. Hill AM et al. Risks of cardiovascular or central nervous system adverse events and immune reconstitution inflammatory syndrome, for dolutegravir versus other antiretrovirals: meta-analysis of randomized trials. Current opinion in HIV and AIDS. DOI:10.1097/COH.0000000000000445. (March 2018).
  5. Taylor KA et al., Platelet function after dolutegravir and/or darunavir/cobicistat in healthy subjects. CROI 2019. 4-7 March 2019, Seattle. Poster 640.
  6. Collins S. Increased risk of myocardial infarction associated with abacavir and ddI. HTB (1 February 2008).

Links to other websites are current at date of posting but not maintained.