Associations between Anatomic Fat Depots with Total, Calcified, Non-Calcified and Mixed Coronary Plaque in the Multicenter AIDS Cohort Study (MACS)
Reported by Jules Levin
CROI 2012 Seattle March 5-8
Frank J. Palella Jr*1, Xiuhong Li2, Lisa Jacobson2, Todd Brown2, Lawrence Kingsley3, Mallory Witt4, Matthew Budoff5, Adrian Dobs2, Susheel
from Jules: this study reports less body fat (lipoatrophy) is associated with coronary plaque. There are 2 types of plaque, calcified & non-calcified. Different types of fat depots (less fat in SAT, more fat in belly) are differentially associated with different types of plaque.
Background: Body fat depots are associated with coronary artery disease (CAD). HIV+ persons may be at greater CAD risk.
Methods: The Multicenter AIDS Cohort Study (MACS) performed coronary CT angiography on 452 HIV+ and HIVŠ men. Plaque was graded in coronary segments to generate scores for total, non-calcified, mixed, and calcified plaque. We measured abdominal (visceral and subcutaneous) adipose tissue, thigh fat and liver fat with non-contrast CT scans. Fatty liver was defined as mean liver HU <40. Logistic regression examined associations between fat depots and the top quartile of plaque scores compared with the combined lower quartiles, adjusted for age and HIV status, then CAD risk factors; then tested for interactions by HIV status. Correlations with serum interleukin-6 (IL-6) levels were made.
Results: Greater visceral adipose tissue was associated with higher total plaque scores (OR 1.03/10 units, p = 0.009) in the entire cohort without interaction by HIV status. When CAD risk factors were added, this attenuated (OR 1.02, p = 0.24). Associations between subcutaneous fat and thigh fat with total plaque scored differed by HIV serostatus (p = 0.03 for both interactions) with trends toward greater total plaque scores with more subcutaneous fat in HIVŠ (p = 0.12) and an inverse association in HIV+ for total plaque scores and thigh fat (OR 0.86, p = 0.04) and for total plaque scores and subcutaneous adipose tissue (p = 0.13). Fatty liver was associated with greater total plaque scores (OR 3.64, p = 0.013); after adjusting for CAD risk factors this attenuated (OR 2.78, p = 0.076). A positive trend in HIV+ existed between fatty liver and greater total plaque scores in adjusted models (OR 3.06, p = 0.117). Greater visceral adipose tissue was associated with more non-calcified plaque (p <0.05), but significance attenuated in adjusted models (p = 0.54). In HIV+ less subcutaneous adipose tissue was associated with greater mixed plaque and persisted after adjustment (OR 0.97, p = 0.03); more subcutaneous fat tended to be associated with greater mixed plaque in HIVŠ (p = 0.20). There was a positive association between fatty liver and mixed plaque (OR 2.99, p = 0.034) that attenuated after adjustment for CAD risks (OR 2.46, p = 0.1). HIV+ were more likely than HIVŠ (OR 3.56, p = 0.048) to have an association between fatty liver and mixed plaque. Among HIV+, in fully adjusted models there was a borderline significant positive association between fatty liver and greater mixed plaque (OR 3.57, p = 0.058). Adjustment for IL-6 levels did not modify associations.
Conclusions: Differential associations exist between adiposity and subclinical coronary plaque amount and type by HIV status. In HIV+ men lesser subcutaneous fat or fatty liver correlate with greater total plaque scores and mixed plaque; in HIVŠ men more subcutaneous fat tended to be associated with more plaque. Visceral adipose tissue is associated with non-calcified plaque.
TPS- total plaque score
NCP= non-calcified plaque score
CP= calcified plaque score
MP=mixed plaque score
TF= thigh fat
VAT- visceral (belly) fat
SAT= subcutaneous fat
· In the general population, increased visceral abdominal adiposity (VAT) has been associated with increased risks for pro-atherogenic serum lipids, Insulin resistance (a major cardiovascular disease [CVD] risk factor), and coronary artery disease (CAD).
· HIV-infected (HIV+) persons experience body fat changes, both lipoatrophy and lipoaccumulation, that may impact risk for CAD; in this population, associations have been demonstrated between increased VAT (as well as decreased subcutaneous and limb fat) with both pro-atherogenic serum lipids and Insulin resistance, but not clearly with coronary plaque or clinical CVD endpoints
· Subclinical atherosclerosis can be detected non-invasively with coronary CT angiography (CTA) which allows assessment of the presence, extent and composition of coronary artery plaque.
·We hypothesized that associations exist between anatomic fat depot amounts (VAT , subcutaneous, and liver) and coronary plaque, but that these associations differ by HIV serostatus.
Multicenter AIDS Cohort Study (MACS)
·Ongoing multi-center prospective observational study of U.S. men who have sex with men (MSM) established by NIH in 1984.
·MACS sites include Baltimore MD, Chicago IL, Los Angeles CA, and Pittsburgh PA, and Washington DC.
·Participants seen semi-annually with interviews, physical examinations and blood draws.
·The MACS Cardiovascular Disease Substudy assesses CAD risk using non-invasive coronary CT angiography (CTA) in HIV infected (HIV+) and uninfected HIV-) men.
·Inclusion criteria: MACS participants, age 45-70, no prior cardiac surgery or coronary artery PTCA or stent, estimate GFR> 60 ml/min/m2, no contrast allergy.
· Coronary CTA performed in 343 HIV + and 176 HIV- men using prospective ECG triggering.
· Plaque graded in each of the 15 AHA predefined coronary segments.
· Plaque composition: non-calcified, mixed or calcified.
· Total plaque score (TPS) = sum of each plaque size score in each coronary segment.
· Non-calcified plaque score (NCP) = sum of plaque size score of each non-calcified plaque.
· Mixed plaque score (MP)= sum of plaque size score of each mixed plaque.
· Calcified plaque score (CP)= sum of plaque size score of each calcified plaque
· Abdominal visceral [VAT], subcutaneous [SAT]), thigh fat (TF) and liver fat were measured with non-contrast CT scans.
· Fatty liver (FL) was defined as mean liver Houndsfield Units (HU) <40.
·Associations between plaque presence (total, calcified, non-calcified, and mixed) and individual anatomic fat depots [visceral abdominal fat (VAT), subcutaneous fat (SAT), thigh fat (TF) and fatty liver (FL)] assessed with logistic regression adjusting for age, race/ethnicity, HIV serostatus and, in additional models, CVD risks
·Associations between log transformed plaque scores and fat depots assessed with linear regression among those men with plaque present, with the same adjustments as listed above
·Among HIV+ men, associations were tested with HIV disease clinical parameters
·Testing for interaction by HIV serostatus was performed
NOTE for Table 3 and FIgures: Associations between body fat depots and coronary plaque were obtained after controlling for HIV serostatus, age, race, and CVD risks (hyperlipidemia, smoking status, hypertension, and diabetes mellitus).
Fat variables were centered and scaled per 10 unit increase.
Liver fat variable set equal to 1 if liver mean fat was greater than or equal to 40 HU, 0 if less than 40 HU.
In Table 3, mean change refers to the mean change in log plaque score per 10 unit increase in SAT or TF. Values shaded gray indicate analytic models in which full adjustment for CVD risks were not undertaken because non-significant associations were seen in models adjusted only for age, race and HIV serostatus.
Values highlighted green in Table 3 indicate associations that were significant or borderline significant.
In analytic models adjusted for age, race, HIV serostatus, and CVD risks we found the following:
(p< 0.05 for all comparisons mentioned unless otherwise indicated)
- Among HIV-infected men:
· a negative association between SAT and log TPS in linear models (lesser SAT among HIV-infected men associated with greater TPS ) with an interaction by HIV serostatus (p=0.066), in linear models adjusted for CVD risks.
· a positive association between VAT and log TPS that did not persist after adjustment for CVD risks in linear models
· a negative association between TF amount and log TPS in linear models
· positive association between the presence of FL and log TPS in linear models
· a negative association between SAT and log NCP in linear models. The
interaction between HIV serostatus and SAT was significant when adjusted for age and race, but was borderline significant when additionally adjusted for CVD risk factors (p=0.0718)
· a positive association between VAT and NCP in logistic models adjusted for age and race that is less apparent in logistic and linear models adjusted for CVD risks.
· there was a positive association between VAT and CP>0 in logistic models with a borderline significant interaction by HIV serostatus (p=0.0723)
· a positive association between FL and MP>0 in logistic models.
· a negative association between SAT and MP>0 in logistic models and between SAT and log MP in linear models. There was a significant interaction by HIV serostatus.
- In contrast, among HIV- men,
· there was a trend toward a positive relationship between SAT and log MP in linear models adjusted for CVD risks (p=0.06).
Acknowledgment: The MACS CVD study is funded by NHLBI: RO1 HL095129-01 (Post). The MACS is funded by NIAID, with additional supplemental funding from NCI. UO1-AI-35042, UL1-RR025005 (GCRC), UO1-AI-35043, UO1-AI-35039, UO1-AI-35040, UO1-AI-35041. Disclosure: Richard.T. George: Consultant/Advisory Board; Modest; ICON Medical Imaging.