| |
Associations between integrase strand-transfer inhibitors and cardiovascular disease in people living with HIV: a multicentre prospective study from the RESPOND cohort consortium
|
| |
| |
Download the PDF here
Download the PDF here
July 1 2022
new study just published & was presented at CROI 2021 - "Although the potential for unmeasured confounding and channelling bias cannot fully be excluded, INSTIs initiation was associated with an early onset, excess incidence of cardiovascular disease in the first 2 years of exposure, after accounting for known cardiovascular disease risk factors.
If the association is indeed causal, it could imply that cardiovascular disease develops quickly after initiation of INSTIs in individuals with a specific underlying vulnerability; however, unmeasured confounding might have affected our findings.
These early findings call for analyses in other large studies, and the potential underlying mechanisms explored further.....In summary, although we cannot exclude possible channelling bias and residual confounding, after accounting for cardiovascular disease risk factors, we found that INSTI exposure was associated with an aIRR of 1⋅85 for cardiovascular disease events in the first 6 months after INSTI initiation compared with no exposure. The association persisted up to 24 months of use, although with lower incidence rates than during the first 6 months of exposure. The association appeared similar in individuals with high and low estimated cardiovascular disease risk and across a wide range of sensitivity analyses. These early findings call for analyses in other large studies to verify our findings, and for the potential underlying mechanisms to be explored.
The increased likelihood of starting an INSTI in individuals with an increased estimated 5-year cardiovascular disease risk score indicates at least some degree of confounding by indication, with individuals at risk of cardiovascular disease preferentially starting an INSTI-based regimen. However, notably, the association between INSTI use and incident cardiovascular disease remained after adjusting for cardiovascular disease risk profiles, including use of abacavir and other antiretroviral drugs previously associated with cardiovascular disease. Furthermore, the association was similar for both individuals with high and low estimated 5-year cardiovascular disease risk, reflected by a non-significant interaction test, suggesting that the findings cannot alone be explained by confounding by indication. Nevertheless, the absence of such an interaction warrants a cautious interpretation because of the test's limited statistical power.
RESPOND's observational nature does not allow us to establish causality of the found association. However, we examined possible mediator effects in exploratory analyses, adjusting for any effects of time-updated BMI, hypertension, diabetes, dyslipidaemia, and chronic kidney disease. These adjustments showed no attenuation in the aIRR of cardiovascular disease events; therefore, none of these factors are likely to mediate this increased incidence rate, consistent with these factors leading to cardiovascular disease via slow development of atherosclerosis, and so would not account for the rapid increase in the incidence rate of cardiovascular disease seen here. Additionally, previous findings from RESPOND analyses examining incident dyslipidaemia27 and hypertension with INSTI exposure28 have not shown an increase in these events within a period that precedes or matches the increased incidence rate of cardiovascular disease seen here; however, the time of the event was not the main focus of those analyses. Nevertheless, here we focused on the potential association between cardiovascular disease and INSTI exposure more broadly, not restricting the population to those potentially experiencing weight gain. However, understanding the potential effects of INSTI-related weight gain is of increasing clinical interest. Therefore, future investigations from the RESPOND study will examine potential associations between weight gain, cardiovascular disease risk factors, and cardiovascular disease incidence in greater detail for the population who have weight gain related to INSTIs.
Overall, the absence of an attenuated effect after adjusting for BMI and other known risk factors for cardiovascular disease suggests one of two possible explanations: either that we have not captured the risk factors for cardiovascular disease through which INSTIs act to increase cardiovascular disease adequately, or that the association is in fact not causal, and is due to unmeasured risk factors in the INSTI-exposed population. A third possible explanation for our findings is that INSTIs can increase rates of cardiovascular disease events via a different mechanism unrelated to known risk factors. Such an effect could be similar to drug-induced platelet hyper-reactivity, which has been suggested as the mechanism linking abacavir to cardiovascular disease,12 or the antibody-mediated clot formation and thrombocytopenia seen in vaccine-induced immune thrombotic thrombocytopenia.29 However, introducing time-updated platelet counts into our model did not affect the aIRR, although we cannot adequately address thrombocyte function and other potential pathways in this study. We encourage further examinations of the possible underlying mechanism for the associations observed here in mechanistic studies.
Contrary to our findings here, no andomized clinical trials assessing INSTIs have reported a short-term increase in the incidence of cardiovascular disease.14, 15, 16, 17 Nevertheless, although andomized clinical trials are essential to determine ART efficacy and safety, they do generally not have the large sample size or duration of follow-up needed to uncover rarely occurring events such as cardiovascular disease. Although investigations of the occurrence of cardiovascular disease events with INSTI exposure are still scarce, a recently published US-based analysis showed no association between INSTI use and cardiovascular disease.22 Nevertheless, the analysis had a retrospective design, did not assess incidence of cardiovascular disease stratified by exposure time, and excluded cardiovascular disease events occurring in the first 3 months of INSTI initiation. Therefore, an immediate effect might have been overlooked and further diluted by not accounting for events shortly after INSTI initiation. Additionally, a 2017 analysis from the US Veterans Health Administration cohort, assessing potential cardioprotective effects of atazanavir, reported hazard ratios of myocardial infarction and stroke that were lower for atazanavir than for INSTIs, in line with our findings.21However, the study period of the analysis spanned 2003-15 and the INSTI group was relatively small, including only a small number of individuals treated with second-generation INSTIs.21
Our analysis has several limitations. First, because this is an observational study we cannot exclude the potential for residual confounders or channelling bias, as already discussed. We have applied the same methods developed and used in D:A:D pharmacovigilance analyses, adjusting for several potential confounders, and did numerous sensitivity analyses that gave consistent results, and interpreted the results of these cautiously and conservatively.5, 6, 7, 8
Nevertheless, propensity-score matching could have been used as an alternative to traditional regression analyses, although such methods also have their limitations. Second, we did not have adequate analytical power to restrict the analyses to only include ART-naïve individuals or provide reliable estimates for individual INSTIs used at present, as also seen from post-hoc power calculations. Therefore, we assessed INSTIs collectively as a class for a combined population of ART-naïve and ART-experienced individuals.
Because RESPOND only includes individuals naïve to INSTIs before Jan 1, 2012, we could not directly examine the association between cardiovascular disease and use of non-nucleotide reverse transcriptase inhibitors or boosted protease inhibitors with the same analytical approach because very few individuals within the cohort were naïve to these two drug classes by Jan 1, 2012 (data not shown), and thus the statistical power was insufficient."
EACS2021: Incidence of hypertension in people living with HIV receiving InSTI versus other third-drug ART regimens in the RESPOND cohort
CROI2021: ASSOCIATION BETWEEN INTEGRASE INHIBITORS (InSTIs) AND CARDIOVASCULAR DISEASE (CVD)
CONCLUSION: The INSTIs examined were associated with a 2.5 times greater incidence of CVD in the first 6 month of exposure when compared to no INSTI exposure, after accounting for known CVD risk factors, and across a wide range of sensitivity analyses. These findings call for further investigations in mechanistic studies and other large populations of PLWH seen in routine clinical care.
Summary
Background
Although associations between older antiretroviral drug classes and cardiovascular disease in people living with HIV are well described, there is a paucity of data regarding a possible association with integrase strand-transfer inhibitors (INSTIs). We investigated whether exposure to INSTIs was associated with an increased incidence of cardiovascular disease.
Methods
RESPOND is a prospective, multicentre, collaboration study between 17 pre-existing European and Australian cohorts and includes more than 32 000 adults living with HIV in clinical care after Jan 1, 2012. Individuals were eligible for inclusion in these analyses if they were older than 18 years, had CD4 cell counts and HIV viral load measurements in the 12 months before or within 3 months after baseline (latest of cohort enrolment or Jan 1, 2012), and had no exposure to INSTIs before baseline. These individuals were subsequently followed up to the earliest of the first cardiovascular disease event (ie, myocardial infarction, stroke, or invasive cardiovascular procedure), last follow-up, or Dec 31, 2019. We used multivariable negative binomial regression to assess associations between cardiovascular disease and INSTI exposure (0 months [no exposure] vs >0 to 6 months, >6 to 12 months, >12 to 24 months, >24 to 36 months, and >36 months), adjusted for cardiovascular risk factors. RESPOND is registered with ClinicalTrials.gov, NCT04090151, and is ongoing.
Findings
29 340 people living with HIV were included in these analyses, of whom 7478 (25⋅5%) were female, 21 818 (74⋅4%) were male, and 44 (<1%) were transgender, with a median age of 44⋅3 years (IQR 36⋅2-51⋅3) at baseline.
As of Dec 31, 2019, 14 000 (47⋅7%) of 29 340 participants had been exposed to an INSTI.
During a median follow-up of 6⋅16 years (IQR 3⋅87-7⋅52; 160 252 person-years), 748 (2⋅5%) individuals had a cardiovascular disease event (incidence rate of 4⋅67 events [95% CI 4⋅34-5⋅01] per 1000 person-years of follow-up).
The crude cardiovascular disease incidence rate was 4⋅19 events (3⋅83-4⋅57) per 1000 person-years in those with no INSTI exposure, which increased to 8⋅46 events (6⋅58-10⋅71) per 1000 person-years in those with more than 0 months to 6 months of exposure, and gradually decreased with increasing length of exposure, until it decreased to similar levels of no exposure at more than 24 months of exposure (4⋅25 events [2⋅89-6⋅04] per 1000 person-years among those with >24 to 36 months of exposure).
Compared with those with no INSTI exposure, the risk of cardiovascular disease was increased in the first 24 months of INSTI exposure and thereafter decreased to levels similar to those never exposed (>0 to 6 months of exposure: adjusted incidence rate ratio of 1⋅85 [1⋅44-2⋅39]; >6 to 12 months of exposure: 1⋅19 [0⋅84-1⋅68]; >12 to 24 months of exposure: 1⋅46 [1⋅13-1⋅88]; >24 to 36 months of exposure: 0⋅89 [0⋅62-1⋅29]; and >36 months of exposure: 0⋅96 [0⋅69-1⋅33]; p<0⋅0001).
Interpretation
Although the potential for unmeasured confounding and channelling bias cannot fully be excluded, INSTIs initiation was associated with an early onset, excess incidence of cardiovascular disease in the first 2 years of exposure, after accounting for known cardiovascular disease risk factors. These early findings call for analyses in other large studies, and the potential underlying mechanisms explored further.
|
|
| |
| |
|
|
|
