Cardiovascular Disease & Risk Assessment in HIV - Editorial & New Study: "Presence, Characteristics, and Prognostic Associations of Carotid Plaque Among People Living With HIV"
Download the PDF here
Download the PDF here
see full study below following the Editorial
"Presence, Characteristics, and Prognostic Associations of Carotid Plaque Among People Living With HIV"
from Jules: I suggest reading this EDITORIAL before putting too much emphasis on the original published study below following the editorial
Imaging to End Points
Cardiovascular Disease Risk Assessment in HIV
Ahmed M. Gharib, Colleen Hadigan
From the Biomedical and Metabolic Imaging Branch, National Institute
of Diabetes and Digestive and Kidney Diseases (A.M.G.) and Laboratory
of Immunoregulation, National Institute of Allergy and Infectious
Diseases (C.H.), National Institutes of Health, Bethesda, MD.
"by 2030 as many as 73% of people living with HIV will be over the age of 50 years....cohort studies have consistently validated these observations, identifying ≈1.5-fold higher risk of acute myocardial infarction in people living with HIV compared with uninfected controls therefore, the prevalence of CVD, an anticipated consequence of aging, can be expected to rise. Strategies for accurate identification of at-risk patients and development of effective approaches to CVD risk reduction will become increasingly important for the care of people living with HIV in the future......25% of the HIV subjects had a history of cocaine exposure compared with only 4% of controls, and hepatitis C infection was present in 18% of the HIV subjects.....chronic cocaine use was associated with a 2-fold increased risk of coronary artery plaque in HIV.....it would be difficult to justify using this technology as a means of identifying or monitor asymptomatic HIV subjects at high risk for cardiovascular events."
The remarkable evolution of the HIV epidemic from a near uniformly fatal condition to a chronic viral infection now managed for many patients with 1 pill once a day marks one of the great accomplishments of medicine in recent decades. Life expectancy estimates for people living with HIV with access to antiretroviral therapy are beginning to approach life expectancy estimates of the general population.1 However, not long after the widespread use of combination antiretroviral therapy for the treatment of HIV, clinicians began to recognize undesirable effects of HIV therapy such as dyslipidemia and alterations in body fat distribution. These observations stimulated considerable attention from HIV providers and researchers to identify and understand cardiovascular disease (CVD) risk associated with HIV and its therapy.
See Article by Janjua et al
Early large cohort studies designed to evaluate CVD outcomes in the context of HIV demonstrated an increased risk of myocardial infarction in association with increasing years of antiretroviral therapy exposure, and observational patient registries showed that rates of myocardial infarction in people living with HIV are increased compared with uninfected contemporaries.2,3 Subsequent cohort studies have consistently validated these observations, identifying ≈1.5-fold higher risk of acute myocardial infarction in people living with HIV compared with uninfected controls. In these studies, the enhanced risk of CVD associated with a diagnosis of HIV was found to be independent of traditional CVD risk factors, which are often enriched in HIV-infected populations.4,5Considerable efforts are underway to optimize CVD risk assessment and to characterize the unique contribution that HIV and immune activation play in the pathophysiology of CVD in people living with HIV. Recent estimates for the aging characteristics of the HIV-infected population suggest that by 2030 as many as 73% of people living with HIV will be over the age of 50 years6; therefore, the prevalence of CVD, an anticipated consequence of aging, can be expected to rise. Strategies for accurate identification of at-risk patients and development of effective approaches to CVD risk reduction will become increasingly important for the care of people living with HIV in the future.
In the current issue of Circulation: Cardiovascular Imaging, Janjua et al7 present the results of a study that leverages a large multi-institutional patient data registry to expand our current knowledge of CVD risk in HIV and provides an important link between an observed biomarker of atherosclerosis, namely carotid plaque, and subsequent CVD events. This retrospective study used available data from diagnostic contrast-enhanced computed tomography (CT) scans from people living with HIV without history of atherosclerosis (n=209) and similar non-HIV-infected controls (n=168), to compare the presence and characteristics of carotid plaque and to determine the relationship between carotid plaque and subsequent CVD events.
The study shows that people living with HIV had significantly higher rates of any carotid plaque, noncalcified carotid plaque, and high-risk plaque compared with controls. However, there were important differences between the HIV group and controls. The majority (60%) of HIV subjects had ≥1 traditional CVD risk factors, whereas 64% of controls had no CVD risk factors identified (P<0.001). In addition, 25% of the HIV subjects had a history of cocaine exposure compared with only 4% of controls, and hepatitis C infection was present in 18% of the HIV subjects. Nontraditional risk factors for atherosclerotic disease such as cocaine and chronic hepatitis C infection are associated with carotid plaque in HIV.8 Further, chronic cocaine use was associated with a 2-fold increased risk of coronary artery plaque in HIV, independent of traditional risk factors and antiretroviral therapy exposure.9 Differences in routine practices for inquiring about and documenting traditional CVD risk factors and substance use by providers caring for people living with HIV and those without HIV infection may create reporting bias and underestimate the prevalence of these risk factors in the control group. Nonetheless, the mismatch in traditional CVD risk factors, as well as cocaine exposure, may account for the much of the observed difference between groups in the presence of carotid plaque.
The present report may not have had optimally matched controls or been sufficiently large to tease out the contribution of non-HIV risk factors toward the prevalence of carotid plaque; however, the novel and informative observation that carotid plaque predicted later CVD events advances the field of CVD investigation in HIV. In multivariable analysis adjusted for traditional CVD risk factors, the presence of any carotid plaque among HIV subjects was associated with an increased risk of a CVD events (hazard ratio, 2.84; confidence interval, 1.05-7.63; P=0.03). Similar associations were observed between noncalcified carotid plaque and high-risk plaque and the risk of subsequent CVD events in HIV subjects. These data are consistent with research in the general population. For example, Cao et al10 showed that the presence of intermediate- and high-risk carotid plaque on carotid ultrasound conferred approximately a 2-fold increased risk of CVD event after adjustment for traditional CVD risk factors in the Cardiovascular Health Study, a study of adults (≥65 years) without known CVD. In the younger MESA cohort (Multi-Ethnic Study of Atherosclerosis), carotid plaque score was also a significant predictor of CVD events after adjustment for traditional risk factors (hazard ratio, 1.27; 95% confidence interval, 1.16-1.40; P<0.001).11 The observed link between asymptomatic carotid plaque and subsequent CVD events in a cohort of people living with HIV supports the potential use of carotid plaque measurements as a biomarker of CVD risk stratification and CVD risk reduction.
The overwhelming majority of HIV research evaluating carotid atherosclerosis and carotid intima-media thickness has used ultrasound technology and not CT imaging, as used in the current report. Although both are noninvasive imaging approaches, the additional burden of radiation exposure and contrast administration limits the broader appeal for the use of CT as a general screening tool in large populations. Radiation risk is further accentuated by the presence of the thyroid gland in the field of view and may explain the conspicuously low number of carotid CT studies related to atherosclerosis in asymptomatic populations, particularly those that may require follow-up. Therefore, it would be difficult to justify using this technology as a means of identifying or monitor asymptomatic HIV subjects at high risk for cardiovascular events.
Admittedly, Janjua et al7 did not prospectively apply carotid CT as a general screening tool. The investigators evaluated carotid CT retrospectively among a population with clinically indicated neck CT data, but additional limitations to this approach exist. The use of <3 mm spotty calcifications and 1 mm diameter foci of low attenuation (<40 HU) as signs of high-risk plaque would require higher resolution acquisitions (more radiation) and perfectly timed contrast administration. The latter component is even more critical as some high-risk plaques may enhance or washout rapidly depending on the inflammatory status in addition to the wide range of variability in densities noted in atherosclerotic plaque.12-14 Many of the challenges associated with CT are overcome by high-field magnetic resonance imaging which has improved signal to noise and spatial and temporal resolution.15-17 The technical versatility of magnetic resonance imaging also allows for plaque characterization.12,14 The relative technical simplicity, mobility, and lower cost of ultrasound places this modality at an advantage for claustrophobic patients and locations that might not have the magnetic resonance imaging capabilities.12,14 Given these factors and available alternatives, translation of the findings of this study into clinical practice may be challenging.
The era of big data has arrived, and its application in health care to infectious diseases and HIV are underway.18 Although carotid CT may not be the optimal approach to prospective CVD risk screening, as access to large patient data registries with high-quality, reliable medical information and imaging data becomes more readily available, additional studies using retrospective data of this nature will be feasible. Triant et al3completed one of the earlier studies to use this approach to effectively demonstrate an increased risk of myocardial infarction in HIV relative to a general contemporary population. The application of this strategy to evaluate the role of incidental carotid plaque identified by CT to predict CVD events in HIV provides valuable new data that may be used to inform future research and management of CVD risk stratification and risk reduction in HIV.
As the HIV population ages, CVD associated morbidity and mortality is expected to rise. Multiple large observational cohorts have established a 1.5- to 2-fold increased risk of CVD in people living with HIV compared with uninfected controls. Excess CVD observed in HIV is likely attributable to multiple factors including traditional CVD risk factors, HIV-related inflammation and immune activation, and possible off-target effects of HIV therapy. Carotid plaque in the general population is predictive of CVD events, and Janjua et al7 demonstrate that this relationship also holds true in HIV subjects. Research efforts continue to focus on identification of biomarker and imaging correlates of CVD in HIV, and large prospective trials such as REPRIEVE (Randomized Trial to Prevent Vascular Events in HIV), a study of pitavastatin to reduce incident CVD, are underway to establish optimal strategies to address the excess CVD risk associated with HIV infection.
Presence, Characteristics, and Prognostic Associations of Carotid Plaque Among People Living With HIV
Circulation: Cardiovascular Imaging October 11, 2017
Sumbal A. Janjua, Pedro V. Staziaki, Balint Szilveszter, Richard A.P. Takx, Thomas Mayrhofer, Orla Hennessy, Hamed A. Emami, Jakob Park, Alexander Ivanov, Travis R. Hallett, Michael T. Lu, Javier M. Romero, Steven K. Grinspoon, Udo Hoffmann, Markella V. Zanni, Tomas G. Neilan
"There were no significant differences in age (45±10 versus 43±17 years, HIV+ versus HIV-; P=0.07), sex (72% versus 65% male; P=0.18), prevalence of diabetes mellitus (10% versus 11%; P=0.61), hyperlipidemia (15% versus17%; P=0.67), and cigarette smoking (33% versus 30%; P=0.57). Cardiovascular medication use was similar between the 2 groups........PLWHIV varied significantly with regards to race (more Black) and had increased illicit drug use, [from Jules: you can see below in Table 1 had much greater history of drug use: 8% to 15% of HIV+ vs zero in HIV-neg group] lower body mass index, and lower high-density lipoprotein as compared with uninfected controls. The mean duration of HIV was 14±7 years, 49% had an undetectable viral load, the mean CD4 count was 378±358 cells/mm3, and 93% (n=194) of the PLWHIV were on ART. Variables including duration of ART, CD4 nadir, and CD4 count were present in 93% of the cases (missing 7%), absolute CD8 count for 69% of the cases (missing 31%), and undetectable viral load for 88% of the cases (missing 12%). ......."
Background—Data from broad populations have established associations between incidental carotid plaque and vascular events. Among people living with HIV (PLWHIV), the risk of vascular events is increased; however, whether incidental carotid plaque is increased and there is an association between incidental carotid plaque, plaque characteristics, and vascular events among PLWHIV is unclear.
Methods and Results—Data from the multi-institutional Research Patient Data Registry were used. Presence and characteristics (high-risk plaque, including spotty calcification and low attenuation) of carotid plaque by computerized tomography among PLWHIV without known vascular disease were described. Data were compared with uninfected controls similar in age, sex, and cardiovascular risk factors, including diabetes mellitus, hyperlipidemia, and cigarette smoking to cases.
Primary outcome was an atherosclerotic cardiovascular disease event, and secondary outcome was ischemic stroke. Cohort consisted of 209 PLWHIV (45±10 years, 72% male) and 168 controls.
Using computerized tomography, PLWHIV without vascular disease had higher rates of any carotid plaque (34% versus 25%; P=0.04), noncalcified (18% versus 5%; P<0.001) and high-risk plaque (25% versus 16%; P=0.03).
Over a follow-up of 3 years, 19 atherosclerotic cardiovascular disease events (9 strokes) occurred. Carotid plaque was independently associated with a 3-fold increase in atherosclerotic cardiovascular disease events among PLWHIV (hazard ratio, 2.91; confidence interval, 1.10-7.7, P=0.03) and a 4-fold increased risk of stroke (hazard ratio, 4.43; confidence interval, 1.17-16.70; P=0.02); high-risk plaque was associated with a 3-fold increased risk of atherosclerotic cardiovascular disease events and a 4-fold increased risk of stroke.
Conclusions—There is an increase in incidental carotid plaque, noncalcified plaque, and high-risk plaque among PLWHIV, and the presence and characteristics of carotid plaque are associated with subsequent vascular events.
Among people living with HIV (PLWHIV), studies have shown a disproportionate vulnerability to select cardiovascular diseases, including myocardial infarction (MI) and stroke.1,2 Specifically, PLWHIV face a more than 2-fold greater risk of vascular events, including both MI and ischemic stroke as compared with uninfected individuals.3,4 The mechanisms involved in the increased risk of vascular events in HIV are incompletely understood,5 and more effective means of identifying PLWHIV at highest risk of incurring atherosclerotic cardiovascular disease (ASCVD) events may be of value.
See Editorial by Gharib and Hadigan
See Clinical Perspective
Computerized tomography (CT) allows for the detailed anatomic delineation of large- and medium-sized vessels and the reproducible early detection and characterization of atherosclerosis and atherosclerotic plaque.6 Among PLWHIV, studies using CT focused on coronary imaging have established that PLWHIV have an increased prevalence of coronary atherosclerosis, noncalcified, and high-risk coronary atherosclerotic plaque, and studies have shown an association between coronary atherosclerosis and plaque features with subsequent ASCVD events.7,8Among PLWHIV, there are no data testing whether HIV status is associated with an increase in carotid plaque using CT and whether the presence or characteristics of carotid plaque using CT in HIV is associated with an increase in subsequent ASCVD events. Therefore, we compared the presence and characteristics of incidental carotid plaque among a population of PLWHIV without known vascular disease to an uninfected control group similar with regards to age, sex, and cardiovascular risk factors, including diabetes mellitus, hyperlipidemia, and cigarette smoking to cases and tested the association between the presence of plaque and plaque features with adverse outcomes.
Leveraging data from a large US healthcare system, we tested, using CT, whether HIV status was associated with an increase in incidental carotid plaque, whether HIV status was associated with high risk and noncalcified plaque, and whether incidental carotid plaque and plaque characteristics were associated with subsequent vascular events among PLWHIV. In a cohort free of known vascular disease, we found that HIV status was associated with an increased rates of incidental carotid plaque, noncalcified plaque, and HRP as compared with uninfected controls. Additionally, in follow-up, among PLWHIV, the presence of carotid plaque, noncalcified plaque, and HRP was associated with an increased risk of both any atherosclerotic event and stroke after adjustment for cardiovascular risk factors.
Our hypothesis that there would be an increase in subclinical carotid plaque in HIV was based on prior data from several groups showing an increase in subclinical coronary plaque in HIV.19,20 For example, Lo et al19 showed in a study of 78 men living with HIV without a history of vascular disease that incidental coronary atherosclerosis by CT was almost 2-fold higher in men living with HIV as compared with demographic and risk factor uninfected men (59% versus 34%). Based on these findings, we hypothesized that there would be an increase in subclinical carotid plaque by CT among PLWHIV free of known vascular disease. The hypothesis was also supported by robust data using ultrasound for detection of carotid plaque in HIV and data from other inflammatory diseases associated with an increased risk of vascular events.21,22 For example, in a study of 1822 PLWHIV using serial ultrasound, Hanna et al21 showed that PLWHIV had a 61% greater risk of new focal carotid artery plaque formation over 7 years compared with uninfected controls, regardless of baseline vascular phenotype and after controlling for cardiometabolic risk factors.
There are also significant data in other diseases that share pathophysiological overlap with HIV. Specifically, the mechanisms of increased vascular risk in HIV are likely related to immune activation and an increase in inflammation,23,24 and in other diseases where an immune activation and inflammation play a key pathophysiological role, an increase in incidental carotid plaque has been reported.25,26 For example, in rheumatoid arthritis and diabetes mellitus, data have described an increase in incidental carotid plaque.27,28 We extend these findings to HIV and provide, to our knowledge, the first study characterizing carotid plaque by CT in HIV and demonstrate an increased number of incidental carotid artery plaque in PLWHIV as compared with an uninfected cohort.
Given its spatial resolution, CT is an excellent technique for characterizing atherosclerotic plaque composition, and data using CT in broad populations have shown that the association between atherosclerotic plaque in the coronaries and adverse outcomes is markedly influenced by plaque composition.29,30 Among PLWHIV, data using CT have shown that there is a marked increase in the prevalence of both noncalcified plaque31 and HRP in coronary arteries of PLWHIV.8,32 For example, Post et al20 showed in a study of 618 PLWHIV that the prevalence of any noncalcified coronary plaque was 63% as compared with 53% in uninfected controls. Similar, D'Ascenzo et al,31 in a meta-analysis of 9 studies (1229 PLWHIV and 1029 controls), showed that the rates of noncalcified plaque were >3-fold higher in PLWHIV (58% versus 17%). Prior to this current work, there were no data testing the prognostic significance of noncalcified plaque or HRP in the carotid arteries among PLWHIV. However, there are data testing the prevalence and prognostic significance of noncalcified plaque and HRP in the carotid in non-HIV populations.33 For example, Homburg et al evaluated the associations between carotid plaque ulceration and plaque characteristics in ischemic stroke patients and found that a lipid-rich core (noncalcified plaque) was associated with plaque ulceration, whereas calcification (plaque stability) was inversely associated with plaque ulceration. We extend these prior observations to PLWHIV and find that PLWHIV have an increase number of noncalcified carotid plaque and HRP and that these plaque characteristics are associated with an increase risk of subsequent clinical vascular events.
Data have established that the risk of vascular events is increased in HIV.5 For example, Chow et al showed that the incidence rate of ischemic stroke was 5.27 per 1000 person years in PLWHIV compared with 3.75 in uninfected patients (adjusted HR of 1.40) and that HIV remained an independent predictor of stroke after controlling for demographics and stroke risk factor.1 Additionally, studies in broad cohorts have demonstrated that the presence of carotid plaque is associated with an increased incidence of vascular events.
For example, Gepner et al34reported that in 6779 subjects without prior cardiovascular disease from the MESA (Multi-Ethnic Study of Atherosclerosis) cohort, carotid plaque presence by carotid ultrasound (c=0.782 to c=0.787; P=0.045) marginally improved prediction of stroke/transient ischemic attack, above the traditional cardiovascular risk factors. However, no prior study has connected carotid plaque to vascular events and stroke specifically in HIV. In this study, consistent with prior data, the baseline risk of a vascular event was higher in PLWHIV as compared with a uninfected cohort, and additively, we show that the adjusted risk for a vascular event in PLWHIV with carotid plaque or HRP was ≈3-fold higher than those without plaque.
Strengths of our study include the first ever CT-based assessment of the presence of incidental carotid plaque in HIV, as well as first reporting on quantitative and qualitative plaque carotid analysis in relation to events in this population. However, some limitations merit discussion. This was a retrospective study, and HIV-specific parameters were not recorded at the time of the CT but were gathered from the electronic medical record at a time close to the study. All efforts were made to collect the viral load assay data closest to the date of physician visit (ART data) if not on the same date. Nonadherence is a possible factor as well. We suspect that the majority of patients in whom detectable viral load was noted were experiencing blips-temporary low level, detectable viremia observed in the context of transient ART nonadherence. We consider it likely that transient nonadherence with ART is relatively common in this cohort, as it is in other contemporary cohorts. Indeed, a recent investigation published by the Swiss HIV Cohort team suggested that blips can be observed in ≤41% of HIV-infected patients to whom ART is prescribed. Because uninfected controls were generated from the original query, there may be an element of referral/selection bias. Additionally, cocaine use was only available for 65% of the subjects and its interpretation as significantly different between the cases and controls should be done with caution, and the observational nature precludes definitive determinations of causal drivers of outcomes. We also acknowledge that the type of CT was not prespecified and CT scans included arterial and venous phase CT studies, which differ in spatial resolution. However, CT angiography scans, with their higher resolution for plaque analysis, were more prevalent in the uninfected cohort.
PLWHIV without prior vascular disease have an increased number of carotid plaque, noncalcified plaque, and HRP as compared with uninfected controls, and in follow-up, the presence of carotid plaque, noncalcified plaque, and HRP is associated with an increased risk of adverse vascular outcomes. Further research is needed to identify the underlying mechanisms, both general or HIV specific, contributing to this increased risk and whether interventions, such as statins, can reduce risk in this vulnerable population.
Study Design and Patient Population
The study was retrospective in design, and our data set was developed from an established and validated research registry of PLWHIV and non-HIV controls.1,9-12 In brief, PLWHIV were identified via a broad initial search, which included the International Classification of Diseases Ninth Revision codes for HIV (International Classification of Diseases Ninth Revision-clinical modification codes 042, 043, or V08) used in the research registry plus HIV-related diagnosis-related codes (diagnosis-related groups: diagnosis-related group 488, 489, 702, 705, 707, 708, 710, 711, 714, 790) and specific search terms (eg, HIV, antiretroviral therapy [ART], AIDS) generating a broad initial sample size of (n) 9375 (Figure 1A). From this data set, we confirmed a diagnosis of HIV on chart review and identified all PWLHIV who had a neck CT with contrast. Subjects with a prior history of ASCVD, including MI, coronary artery disease, cerebrovascular disease (stroke and transient ischemic attack), prior carotid or peripheral vascular disease, and prior atrial fibrillation were excluded. Subjects with nondiagnostic/unavailable neck CT scans and subjects in whom the indication for the CT scan was stroke were also excluded. Control subjects were derived from the same database. In brief, 441 subjects were noted to be erroneously coded under the extended search term HIV and served as the control group. A through chart review demonstrated that these subjects had the word HIV in the medical record typically because of HIV testing and did not have HIV. Non-HIV control subjects with a prior history of ASCVD, coronary artery disease, and cerebrovascular disease were excluded (stroke and transient ischemic attack). Data collection, image analysis, and event adjudication were performed by 3 independent and blinded groups. One team performed electronic medical record review to ascertain data on baseline cardiovascular risk factors and HIV-specific parameters, another adjudicated clinical outcomes, and a third performed image analysis of the neck CTs. Our study was approved by the Institutional Review Board.
Traditional Cardiovascular Risk Factors
Baseline covariates were collected within 1 year of the neck CT scan by electronic medical record review. Data were collected on traditional cardiovascular risk factors, including hypertension, dyslipidemia, diabetes mellitus, and family history of coronary heart disease.13 The use of cardiovascular medications, body mass index, sleep apnea, creatinine levels, smoking, alcohol use, and illicit drug use were also collected.
Data were collected on duration since HIV diagnosis, duration of antiretroviral therapy, category of antiretroviral therapies, viral load, CD4 count, and nadir CD4 count as previously described.11 All efforts were made to collect the HIV-specific parameters from the electronic medical record at a time point closest to the CT scan.
CT Neck Image Analysis
The types of CT scans included CT neck with contrast, CT angiography, and positron emission tomography-CT with contrast, performed for a variety of indications. Indications for neck CT included assessment for trauma (27%), neck mass (17%), neck cancer (13%), lymphadenopathy (8%), abscess (7%), neck pain (7%), headache (2%), dysphagia (1%), and varied (12%). We assessed the presence of plaque, plaque composition (calcified, mixed, and noncalcified), degree of stenosis based on North American Symptomatic Carotid Endarterectomy Trial (NASCET) evaluation (minimal, 1%-30%; mild, 31%-50%; moderate, 51%-69%; severe, 70%-99%), presence of high-risk plaque (HRP) features, including spotty calcification (size of calcification <3 mm in all directions) and low attenuation (if circular region of interest with diameter of 1 mm shows area with mean attenuation <40 HU).14 A plaque score was computed by summing the maximum thickness of plaque in each segment (mm) on both sides according to Gupta et al.15 The carotid images were analyzed by 3 experienced readers (S.J., P.S., R.T.), blinded to the clinical data, on a cardiac workstation (Tera Recon, Foster City, CA). All readers completed standardized CT training to assess interobserver variability and had excellent interobserver agreement for presence of plaque (kappa, 0.92), grade of stenosis (kappa, 0.77), type of calcification (kappa, 0.74), low attenuation (kappa, 0.95), and spotty calcification (kappa, 0.69).
The primary outcome was an ASCVD event, defined as MI, coronary heart disease death, or ischemic stroke, in accordance with ASCVD end points defined in the 2013 American College of Cardiology/American Heart Association Risk Assessment Guidelines.16 The secondary outcome was ischemic stroke as defined by the American Heart Association/American Stroke Association consensus document.17 All events were adjudicated by blinded board-certified physicians using standardized definitions blinded to other variables.18
Continuous variables were presented as mean±SD or median and interquartile range, whereas categorical variables were presented as percentages. Parameters were compared between the 2 groups using Student's t test or Mann Whitney test for continuous variables and χ2 or Fisher exact test for categorical variables, as appropriate. Univariate and multivariate Cox proportional hazard regression analyses were performed to determine the association between the presence of any carotid plaque or HRP and outcomes among the entire cohort and then just limited to PLWHIV. The multivariate models were adjusted for the number of traditional cardiovascular risk factors, including hypertension, diabetes mellitus, hyperlipidemia, current smoking, and family history of coronary artery disease. Hazard ratios (HR) are presented with 95% confidence interval (CI). Kaplan-Meier curves were generated to compare event-free survival among PLWHIV and uninfected individuals with and without plaque. Between-group differences were assessed using the log-rank test. A P value <0.05 was considered significant. Statistical calculations were performed using SPSS software (SPSS version 22; IBM Corp., Armonk, NY) and STATA software (Version 13, StataCorp LP., College Station, TX).
Study flow diagrams for the derivation of both the PLWHIV and uninfected control individuals are presented in Figure 1A and 1B. After exclusion of subjects in each cohort for prior vascular disease, there were 209 PLWHIV and 168 controls. The distribution of type of CT scans included CT neck with contrast (65%), CT angiography (44%), and positron emission tomography-CT with contrast (4%). PLWHIV had a higher frequency of CT neck with contrast, while controls had a higher number of CT angiography. The indication categories in the 2 groups were similar for abscess, cancer, dysphagia, pain, mental status change (P<0.05 for all), whereas significant differences were found for the numbers in each group where the indication was lymphadenopathy, mass, and trauma (P>0.05 for all). Tables 1 and 2 describe the baseline characteristics of PLWHIV and uninfected controls. There were no significant differences in age (45±10 versus 43±17 years, HIV+ versus HIV-; P=0.07), sex (72% versus 65% male; P=0.18), prevalence of diabetes mellitus (10% versus 11%; P=0.61), hyperlipidemia (15% versus17%; P=0.67), and cigarette smoking (33% versus 30%; P=0.57). Cardiovascular medication use was similar between the 2 groups.
PLWHIV varied significantly with regards to race (more Black) and had increased illicit drug use, lower body mass index, and lower high-density lipoprotein as compared with uninfected controls. The mean duration of HIV was 14±7 years, 49% had an undetectable viral load, the mean CD4 count was 378±358 cells/mm3, and 93% (n=194) of the PLWHIV were on ART. Variables including duration of ART, CD4 nadir, and CD4 count were present in 93% of the cases (missing 7%), absolute CD8 count for 69% of the cases (missing 31%), and undetectable viral load for 88% of the cases (missing 12%).
Carotid Plaque Characteristics
PLWHIV free of known vascular disease had an increased rate of any carotid plaque (34% versus 25%; P=0.04), noncalcified carotid plaque (18% versus 5%; P<0.001), and higher plaque score (1.8±3.4 versus 1.2±2.7; P=0.03). PLWHIV also had a significantly higher number of any HRP (25% versus 16%; P=0.03), higher number of HRP features per patient (P=0.03), and a higher number of low attenuation plaque (16% versus 7%; P=0.007; Table 3 and Figure 2).
The primary outcome was an ASCVD event. Over a median follow-up of 3 years, there were 17 ASCVD (ischemic stroke, 9; MI, 6; coronary heart disease death, 2) events among 209 PLWHIV and 2 ASCVD events (coronary heart disease death, 2) among uninfected controls (P=0.002). The secondary outcome of interest was ischemic stroke. There were 9 ischemic strokes among PLWHIV as compared with 0 strokes among controls (P=0.005). Among PLWHIV, there were 10 ASCVD events among the 72 PLWHIV with carotid plaque (14%) as compared with 7 events among the 137 PLWHIV without plaque (5%; P=0.027). Similarly, among PLWHIV, there were 6 ischemic stroke events among the 72 PLWHIV (8%) with any carotid plaque as compared with 3 events among 137 PLWHIV (2%) without any plaque (P=0.046).
Among PLWHIV, there were 8 ASCVD events among the 53 PLWHIV with any HRP (15%) as compared with 9 events among the 156 PLWHIV without HRP (6%; P=0.032), and there were 5 ischemic stroke events among the 53 PLWHIV (9%) with any HRP as compared with 4 events among 156 PLWHIV (2%) without any HRP (P=0.048).
Univariate and Multivariate Modeling for Outcomes
Among the whole cohort, there was an unadjusted association between HIV status and ASCVD events (HR, 5.81; CI, 1.34-25.2; P=0.02) and ischemic stroke and presence of any carotid plaque with ASCVD events (HR, 3.32; CI, 1.34-8.29; P=0.01) and ischemic stroke (HR, 4.89; CI, 1.22-19.60; P=0.03). After adjustment for the number of cardiovascular risk factors, HIV status (HR, 5.22; CI, 1.20-22.67; P=0.027) and carotid plaque (HR, 2.77; CI, 1.08-7.12; P=0.03) were independent predictors of ASCVD events (Table 4).
In analysis confined to PLWHIV, the presence of any carotid plaque was associated with an almost 3-fold increased risk of ASCVD events (HR, 2.91; CI, 1.10-7.7; P=0.03) and a 4-fold increased risk of ischemic stroke (HR, 4.14; CI, 1.03-16.6; P=0.04). In multivariate analysis confined to PLWHIV, after adjustment for cardiovascular risk factors, the presence of any carotid plaque was associated with an increased risk for ASCVD events (HR, 2.84; CI, 1.05-7.63; P=0.03) and with a trend toward an increased risk for ischemic stroke (HR, 3.57; CI, 0.86-14.7; P=0.07; Table 5 and Figure 3A and 3B). Also among PLWHIV, the presence of any HRP was associated with an unadjusted 3-fold increased risk for ASCVD events (HR, 3.02; CI, 1.15-7.88; P=0.02) and a 4-fold increased risk of stroke (HR, 4.43; CI, 1.17-16.70; P=0.02). In multivariate analysis, after adjustment for number of cardiovascular risk factors, the presence of any HRP feature remained associated with an increased risk of ASCVD events (HR, 2.93; CI, 1.11-7.77; P=0.03) and with an increased risk of stroke (HR, 3.89; CI, 1.01-14.89; P=0.04; Table 6 and Figure 4A and 4B). We also evaluated the association of noncalcified plaque and events: the presence of noncalcified plaque was associated with an unadjusted 3-fold increased risk for ASCVD events (HR, 3.21; CI, 1.22-8.45; P=0.02) and stroke (HR, 3.71; CI, 0.992-13.85; P=0.051). In multivariate analysis, the presence of any noncalcified plaque remained associated with an increased risk of ASCVD events (HR, 3.36; CI, 1.26-8.91; P=0.02) and an independent association with stroke (HR, 4.65; CI, 1.17-18.53; P=0.03).