Carotid Artery Thickness Is Associated With Chronic Use of Highly Active Antiretroviral Therapy in Patients Infected With Human Immunodeficiency Virus

A 3.0 Tesla Magnetic Resonance Imaging Study

TM LaBounty; WD Hardy; Z Fan; R Yumul; D Li; R Dharmakumar; A Hernandez Conte


HIV Medicine. 2016;17(7):516-523. 

In This Article


The study recruited 26 HIV-infected subjects on HAART from a single academic medical centre and 20 HIV-negative controls. Inclusion criteria were: male gender, age of 35–55 years, and (for cases only) continuous HAART for ≥ 3 years. HAART was defined as the use of at least three different anti-HIV medications to achieve viral suppression. Exclusion criteria included history of cardiovascular disease (coronary artery disease, myocardial infarction, stroke or prior revascularization), hepatitis C virus infection, diabetes, prior injecting drug use, prolonged interruptions of HAART (≥ 3 months), prior AIDS-defining illness and contraindications to MRI. HIV-negative controls were recruited from the same community as the subjects. Control subjects were recruited via subjects' referral of friends and acquaintances in similar social circles, to approximate the characteristics of those with HIV infection based upon age, race, sexual orientation, lifestyle factors, drug use and medical history, as done previously.[19] This study was reviewed and approved by our Institutional Review Board and all subjects provided written informed consent.

Subjects with HIV infection had high medication compliance (mean 99%; range 90−100%) and 100% had undetectable HIV RNA by polymerase chain reaction (PCR) testing. The mean (±SD) CD4 T-cell count was 682 ± 468 cells/μL (range 242 to 2597 cells/μL). The mean (±SD) duration of HIV diagnosis was 16.8 ± 8.1 years (range 4–30 years), and the mean (±SD) duration of HAART was 13.4 ± 7.3 years (range 3–28 years). Protease inhibitors had been used in 24 of 26 patients during their treatment history; current protease inhibitors were used in 18 of 24 patients, with all 18 patients utilizing ritonavir as a "booster" PI.

An a priori power analysis was performed to determine the required sample size. To decrease the required sample size of the study, we a priori planned to evaluate the left and right carotid arteries in the same analysis to double our sample size, with a plan to adjust for the potential clustering effect. Based on prior estimates of mean (±SD) carotid intimal medial thickening of 0.62 ± 0.11 mm and 0.70 ± 0.10 between individuals without HIV infection and individuals with HIV infection on HAART,[15] a sample size of 25 subjects in each group with two carotid arteries each was determined to provide 85–89% power to detect a difference in mean carotid artery thickness between groups, using a two-tailed analysis and an alpha of 0.05, and after accounting for a clustering effect for the two arteries with a range of intra-cluster correlations from 0.50 to 0.70. During a pause in enrolment because of changes in study personnel, we performed an interim analysis when 46 of the 50 planned subjects were enrolled. At that time, we identified significant differences between groups, and we terminated subject enrolment early. As we observed significant differences for the separate left and right carotid arteries, a combined analysis that considered the left and right as independent arteries but attempted to correct for intra-subject clustering was no longer needed, and we instead performed separate analyses of the left and right carotid arteries. This eliminated the potential error and statistical assumptions required to estimate the clustering effect for a combined analysis.

All subjects and controls completed a detailed medical and social history questionnaire. For patients with HIV infection, medical records were reviewed to provide details regarding HIV medical and treatment history. A physical examination was performed by an experienced infectious disease specialist to assess anthropometric variables, including the presence of lipodystrophy, defined as the pathological presence (lipoaccumulation) or absence (lipoatrophy) of adipose tissue in various anatomical locations consistent with HAART-associated side effects, consistent with reported in previous literature.[20] If lipodystrophy was present, further physical evaluation was performed to determine the presence of lipoaccumulation and/or lipoatrophy. Height, weight, waist circumference and hip circumference were measured for each subject.

Subjects were instructed to abstain from caffeine, alcohol and vigorous exercise for at least 24 h before all MRI procedures. A fasting, venous blood sample was obtained for the measurement of glucose, lipids, a complete blood count and a basic metabolic panel. In control subjects, HIV testing was performed to confirm the absence of HIV infection. All subjects with HIV infection had prior serological testing that confirmed the diagnosis. Framingham cardiac risk factor scoring was calculated according to standard criteria, accounting for age, cholesterol, smoking history and blood pressure.[21]

All MRI was performed using a 3.0-Tesla whole-body scanner (Magnetom Verio; Siemens Healthcare, Erlangen, Germany) and a bilateral four-channel carotid surface coil (Machnet BV; Eelde, The Netherlands). Subjects were scanned in a head-first supine position. This study utilized images without the use of intravenous contrast. Multislice two-dimensional (2D) time-of-flight imaging was first performed to localize the left and right carotid artery bifurcation. The acquired images were also used to create minimal-intensity-projection images that were later used for positioning the vessel wall scan planes perpendicular to the axis of each common carotid artery. Typical scan parameters were as follows: repetition time/echo time = 22/5 ms; flip angle = 52°; slice thickness = 3 mm with an inter-slice gap of 3 mm; 30 slices; matrix size = 256 × 204; field of view 220 × 175 mm2; in-plane spatial resolution = 0.86 × 0.86 mm2.

Dedicated T1-weighted dark-blood imaging was performed separately for the left and right common carotid arteries, with imaging obtained over a 36 mm length in 2 mm increments proximal to the bifurcation of each common carotid artery. A multi-slice 2D T1-weighted turbo spin echo sequence with spatial pre-saturation band-based dark-blood preparation was used. Sequence parameters were: axial imaging orientation; 18 slices; slice thickness = 2 mm; matrix size = 256 × 256; field of view = 160 × 160 mm2; in-plane spatial resolution = 0.625 × 0.625 mm2; repetition time/echo time = 800/12 ms; echo train length = 7; and echo train duration = 63 ms. Chemically selective fat suppression was applied to improve the definition of the outer wall boundary and avoid chemical shift artifacts.

As done previously, a short-axis image was selected for each common carotid artery immediately inferior to the carotid bulb and within 2 cm of the bifurcation.[22] Existing literature has demonstrated good agreement between carotid intimal medial thickness and MRI utilizing semi-automated measurements of the carotid wall.[18] Although proprietary automated tools have been developed, these are not commercially available or validated, so we averaged the diameter of the arterial wall from eight evenly distributed sites manually measured on the short-axis image. In addition, the cross-sectional external area and luminal area of each common carotid artery area were manually traced on the same short-axis images, with the difference of these representing the cross-sectional wall area including the vessel wall and any plaque (Fig. 1). The wall area was indexed to the external area (wall area/external area), as described previously.[23] The presence of any visible carotid artery plaque was defined as visible wall thickening on any slice that was observed in at least two consecutive slices. Carotid artery distensibility was calculated in a manner consistent with the literature, using the mean diameter derived from tracing the cross-sectional area of each artery at end-systole and end-diastole [distensibility coefficient = (2 × change in diameter/end-diastolic diameter/pulse pressure][22] using bright-blood cine images. A single blinded experienced reader performed all carotid measurements, using OsiriX version 5.8.1 for Mac OS X (OsiriX Foundation, Geneva, Switzerland). To minimize any bias, cases and controls were read in a random sequence, and the reader was blinded to all clinical variables including HIV status.

Figure 1.

Measurement of common carotid artery wall thickness by magnetic resonance angiography. Both patients were 45-year-old men with no cardiovascular risk factors. The left common carotid mean wall thickness was 0.88 mm in the patient with HIV infection on chronic highly active antiretroviral therapy (HAART) (a), and 0.82 mm in the control patient (b). The external carotid area and luminal area are traced for each (labelled), and the wall thickness represents the mean of the eight diameters (two are labelled).

Our primary endpoint was a comparison in mean carotid artery thickness between groups. Secondary endpoints included comparisons of the cross-sectional wall area, indexed wall area, the presence of visible carotid artery plaque, and carotid artery distensibility. Comparisons of continuous variables were performed using Student's t-test for variables with normal distributions or the Mann–Whitney U-test for variables without a normal distribution. Fisher's exact test was used for comparisons of categorical variables.

We further assessed clinical and laboratory characteristics among those with HIV infection to determine which of these variables might be associated with increased carotid artery wall thickness. These comparisons were performed using linear regression analysis that utilized the mean value of the left and right common carotid arteries for each patient. Statistical analyses were performed using ibm spss version 20 (IBM Corporation, Armonk, NY, USA) for Mac OS X.