Subclinical Cardiovascular Disease in HIV Controller and Long-Term Nonprogressor Populations

RM Brusca; DB Hanna; NI Wada; JN Blankson; MD Witt; LP Jacobson; L Kingsley; FJ Palella Jr; M Budoff; TT Brown; K Anastos; JM Lazar; WJ Mack; P Bacchetti; PC Tien; Y Golzar; M Plankey; E Golub; RC Kaplan; WS Post


HIV Medicine. 2020;21(4):217-227. 

In This Article

Abstract and Introduction


Objectives: Elite controllers (ECs), viraemic controllers (VCs), and long-term nonprogressors (LTNPs) control HIV viral replication or maintain CD4 T-cell counts without antiretroviral therapy, but may have increased cardiovascular disease (CVD) risk compared to HIV-uninfected persons. We evaluated subclinical carotid and coronary atherosclerosis and inflammatory biomarker levels among HIV controllers, LTNPs and noncontrollers and HIV-uninfected individuals in the Multicenter AIDS Cohort Study (MACS) and the Women's Interagency HIV Study (WIHS).

Methods: We measured carotid plaque presence and common carotid artery intima-media thickness (IMT) in 1729 women and 1308 men, and the presence of coronary artery calcium and plaque in a subgroup of men. Associations between HIV control category and carotid and coronary plaque prevalences were assessed by multivariable regression analyses adjusting for demographics and CVD risk factors. Serum inflammatory biomarker concentrations [soluble CD163 (sCD163), soluble CD14 (sCD14), galectin-3 (Gal-3), galectin-3 binding protein (Gal-3BP) and interleukin (IL)-6] were measured and associations with HIV control category assessed.

Results: We included 135 HIV controllers (30 ECs) and 135 LTNPs in the study. Carotid plaque prevalence and carotid IMT were similar in HIV controllers, LTNPs and HIV-uninfected individuals. HIV controllers and LTNPs had lower prevalences of carotid plaque compared to viraemic HIV-infected individuals. The prevalence of coronary atherosclerosis was similar in HIV controllers/LTNPs compared to HIV-uninfected and viraemic HIV-infected men. Controllers and LTNPs had higher concentrations of sCD163 and sCD14 compared to HIV-uninfected persons.

Conclusions: Subclinical CVD was similar in HIV controllers, LTNPs and HIV-uninfected individuals despite elevated levels of some inflammatory biomarkers. Future studies of HIV controllers and LTNPs are needed to characterize the risk of CVD among HIV-infected persons.


The widespread use of antiretroviral therapy (ART) to treat HIV infection is associated with extended survival,[1] which has resulted in greater age-related morbidity and mortality from noninfectious diseases, especially cardiovascular disease (CVD).[2–4] Traditional CVD risk factors, such as smoking,[5] contribute to increased CVD risk; however, adverse effects of ART[6,7] and elevated levels of inflammation and immune activation[8–11] are also implicated in higher CVD-associated morbidity and mortality among HIV-infected individuals.

HIV controllers are a rare subset of the HIV-infected population who control viral replication in the absence of ART.[12,13] Long-term nonprogressors (LTNPs) are a related subset, classified by longitudinal stability of CD4 T-cell counts rather than by viral replication. HIV controller and LTNP are overlapping phenotypes; however, important distinctions exist between the groups. While LTNPs by definition do not show clinical progression of HIV infection in the absence of ART, they can have nonsuppressed or even high-level viraemia.[14] HIV controllers have chronic low-level viraemia; however, they may demonstrate progression via decreasing CD4 T-cell counts over time.[14,15] The study of HIV controllers and LTNPs can help to elucidate the proportional contributions of ART, HIV replication, immunosuppression, and excess inflammation to subclinical CVD in HIV-infected persons.

As with the broader HIV-infected population, HIV controllers and LTNPs exhibit increased levels of inflammation and immune activation/dysregulation compared to HIV-uninfected individuals.[16,17] Moreover, both HIV controllers and LTNPs have increased T-cell activation compared to HIV-infected noncontrollers[18,19] even when compared to HIV-infected noncontrollers with ART-induced viral suppression.[20] CVD risk among HIV controllers is less clear. Some smaller studies have shown a greater burden of subclinical carotid[21] and coronary[22] atherosclerosis among elite controllers compared to HIV-uninfected individuals. Although HIV infection may play a role in the development of atherosclerotic CVD, even when naturally controlled and independent of traditional CVD risk factors or ART, conclusive data among a larger sample of HIV controllers, including viraemic controllers (VCs) and LTNPs, are lacking.

Carotid artery plaque and carotid intima-media thickness (IMT, measured by carotid ultrasonography) and coronary artery plaque [measured by noncontrast computed tomography (CT) and coronary CT angiography (CTA)] are validated, noninvasive measures of subclinical atherosclerosis and vascular disease associated with increased risk for CVD events.[23–25] We performed carotid ultrasound in the Multicenter AIDS Cohort Study (MACS) and the Women's Interagency HIV Study (WIHS), and cardiac CT scans in the MACS to assess subclinical CVD, and we measured inflammatory biomarker concentrations among a subset of MACS and WIHS participants. The purpose of this analysis was to compare the prevalence and extent of subclinical carotid and coronary atherosclerosis, as well as inflammatory biomarker levels, among HIV controllers, LTNPs, HIV-infected noncontrollers and HIV-uninfected individuals.

Methods. Study Design and Inclusion Criteria: This study included participants from the MACS and WIHS, two ongoing prospective observational cohort studies of HIV-infected and at-risk, HIV-uninfected men and women, respectively.[26,27] The MACS includes men who have sex with men and began enrolment in 1984, and the WIHS, which includes women, began in 1994. Each study includes semi-annual visits with structured interviews, physical examinations, and blood and urine collection. Participants with no known history of coronary artery disease were recruited in 2004 for participation in a vascular substudy, which included carotid B mode ultrasound in MACS and WIHS[28] and noncontrast cardiac CT scans in MACS.[7] Additional MACS participants were recruited for a vascular substudy in 2010 if they were between the ages of 40 to 70 years without a history of cardiac surgery or coronary revascularization. Some of these men also completed coronary CTA. The studies were approved by the institutional review boards of all participating sites. All participants provided informed consent.

Carotid Examination: Data were collected at a baseline visit in 2004–2006 or in 2010–2012. High-resolution B-mode carotid artery ultrasound was performed to measure carotid plaque presence and mean far wall common carotid IMT, as previously described.[28]

Cardiac CT Scans: Coronary artery calcium (CAC) was measured from noncontrast cardiac CT scans among MACS participants from 2010 to 2013,[29,30] and coronary CTA was performed, as previously described.[31] Trained readers, blinded to participant characteristics and HIV serostatus, analysed the CT images.[32] Each coronary segment was classified as normal or containing noncalcified, mixed (<50% of plaque area occupied by calcium) or calcified plaque. The coronary atherosclerosis measure used in analyses was coronary plaque presence, which was defined as the presence of one or more of any type of plaque in any coronary segment. Stenosis in each segment was defined as none, 1–29%, 30–49%, 50–69% or ≥ 70%.

HIV Control and Other Variables: HIV infection was determined by serological testing [enzyme-linked immunosorbent assay (ELISA)] and confirmed using western blot. The HIV control categories were HIV-uninfected, HIV controller, LTNP, HIV-infected noncontroller with undetectable viraemia [viral load < 50 HIV-1 RNA copies/mL (MACS) or < 80 copies/mL (WIHS)], and HIV-infected noncontroller with detectable viraemia. For the two HIV-infected noncontroller categories, the viral load at the baseline carotid ultrasound visit was used to define undetectable versus detectable viraemia. HIV controllers were defined as elite controllers (ECs) or viraemic controllers (VCs). Table 1 outlines these definitions. ECs had undetectable viral load (as defined above), and VCs had viral load < 2000 copies/mL, each in the absence of ART. ECs were included within the category of VC for the majority of analyses, unless otherwise specified, because of the small numbers of ECs. LTNPs were defined as HIV-infected individuals with CD4 T-cell counts ≥ 500 cells/μL for ≥ 5 years while not on ART. HIV controllers and LTNPs who met criteria for multiple control definitions were included in each control category for the relevant analysis. Many HIV controllers and LTNPs no longer met the criteria at the time of the vascular study visit. For the primary analyses, participants who ever satisfied the criteria for an HIV controller and/or LTNP category were included, even if they did not meet the criteria at the time of the vascular study (see Discussion). HIV controllers and LTNPs were combined for CT analyses because of the small sample size.

Self-reported variables included age, race/ethnicity, income, education, history of injecting drug use, current alcohol use, current cigarette smoking, history of diabetes mellitus (DM), and current use of antihypertensive and lipid-lowering medications. Measured CVD risk factors included body mass index (BMI), systolic blood pressure, total and high-density lipoprotein (HDL) cholesterol, and fasting glucose levels. In the MACS, DM was defined as fasting serum glucose ≥ 126 mg/dL or self-reported use of DM medications. The WIHS expanded on this definition to also include DM self-report or haemoglobin A1c ≥ 6.5%, and confirmation by subsequent report of DM medication or laboratory parameter.[33] CVD risk factors were based on values recorded at the study visit closest to the vascular study visit. HIV clinical characteristics included baseline and nadir CD4 T-cell count, history of clinical AIDS, and cumulative duration of ART use.

Inflammatory Biomarker Measurements: Inflammatory biomarker concentrations were collected for an ancillary study among 1281 individuals from MACS and WIHS who had serial carotid ultrasound scans.[34] Characteristics of this cohort have been previously described.[34] Biomarkers were chosen based on their established association with CVD risk.[9,10,34–37] Concentrations of soluble CD163 (sCD163), soluble CD14 (sCD14), interleukin-6 (IL-6), galectin-3 (Gal-3) and galectin-3 binding protein (Gal-3BP) were measured using ELISA in stored frozen sera collected at the core study visit closest to the vascular study visit. All assays were performed at the University of Vermont in duplicate then averaged on single assay product lots. These analyses were restricted to HIV controllers and LTNPs with biomarker measurements available who met the exposure criteria at the time of the carotid ultrasound because of the variability of biomarker levels relative to ART use and viraemia status.

Statistical Methods: Associations of HIV control category with the presence of carotid and coronary plaque were assessed by estimation of prevalence ratios using modified Poisson regression.[38] Associations between HIV control category and carotid IMT were assessed using linear regression. Models were adjusted for age, race/ethnicity, current cigarette smoking, MACS/WIHS centre, education, current alcohol use and CVD risk factors (BMI, systolic blood pressure, total and HDL cholesterol, current use of antihypertensive or lipid-lowering medications, and DM). We used IVEWARE software to implement multiple imputation (five imputation data sets) based on multivariate sequential regression separately for WIHS and MACS data, to account for the 1% of values that were missing.[39] Regression results for each imputation data set were pooled using standard methods to account for increased variability as a consequence of the imputation process. Generalized gamma regression was performed in STATA(StataCorp LLC, College Station, TX, USA) to assess associations between HIV control category and biomarker concentrations, adjusting for the same covariates as above. Shape and scale parameters were held constant, while the beta parameter was allowed to vary by exposure. Therefore, the percentage difference in biomarker concentrations associated with exposure category applies to all percentiles of the biomarker distribution.