Abnormal Myocardial Function Is Related to Myocardial Steatosis and Diffuse Myocardial Fibrosis in HIV-Infected Adults

Abstract

Background. Impaired cardiac function persists in the era of effective human immunodeficiency virus (HIV) therapy, although the etiology is unclear. We used magnetic resonance imaging (MRI) to measure intramyocardial lipid levels and fibrosis as possible contributors to HIV-associated myocardial dysfunction.

Methods. A cross-sectional study of 95 HIV-infected and 30 matched-healthy adults, without known cardiovascular disease (CVD) was completed. Intramyocardial lipid levels, myocardial fibrosis, and cardiac function (measured on the basis of strain) were quantified by MRI.

Results. Systolic function was significantly decreased in HIV-infected subjects as compared to controls (mean radial strain [±SD], 21.7 ± 8.6% vs 30.5 ± 14.2%; P = .004). Intramyocardial lipid level and fibrosis index were both increased in HIV-infected subjects as compared to controls (P ≤ .04 for both) and correlated with the degree of myocardial dysfunction measured by strain parameters. Intramyocardial lipid levels correlated positively with antiretroviral therapy duration and visceral adiposity. Further, impaired myocardial function was strongly correlated with increased monocyte chemoattractant protein 1 levels (r = 0.396, P = .0002) and lipopolysaccharide binding protein levels (r = 0.25, P = .02).

Conclusions. HIV-infected adults have reduced myocardial function as compared to controls in the absence of known CVD. Decreased cardiac function was associated with abnormal myocardial tissue composition characterized by increased lipid levels and diffuse myocardial fibrosis. Metabolic alterations related to antiretroviral therapy and chronic inflammation may be important targets for optimizing long-term cardiovascular health in HIV-infected individuals.

In the era of widespread use of antiretroviral (ARV) therapy for human immunodeficiency virus (HIV) infection, reports of higher than expected rates of systolic and diastolic dysfunction persist [1–5]. A recent meta-analysis of studies evaluating cardiac dysfunction during HIV infection in the context of ARV therapy identified left ventricular (LV) systolic dysfunction in 8% and diastolic dysfunction in as many as 43% of HIV-infected adults [1]. While the long-term clinical implications of subclinical cardiac abnormalities is not clear, Moyers et al [6] showed that significantly diminished LV function (eg, ejection fraction <40%) was a strong predictor of sudden cardiac death in a large cohort of HIV-infected patients. The etiology of impaired myocardial function in HIV–infected individuals is not yet fully understood, but studies have implicated traditional risk factors, such as age, hypertension, and smoking [1, 4], as well as ARV therapy [7] and direct effects of HIV [5, 8].

Prior investigation has demonstrated significant abnormalities in lipid deposition in various tissue compartments in HIV [9–14], and myocardial lipid deposition may also be increased. Advances in magnetic resonance spectroscopy (MRS) now permit reliable noninvasive quantification of intramyocardial fat content [15]. Myocardial steatosis is abnormal and likely represents subclinical myocardial injury that may ultimately lead to myocardial dysfunction [16–20]. Magnetic resonance tagging techniques also represent a valid and sensitive method to assess myocardial contraction through myocardial strain measurements [21]. For example, tagged MRI measurements can identify subclinical regional LV dysfunction (impaired strain) that corresponds to coronary atherosclerosis in adults without known cardiovascular disease [22].

We hypothesized that abnormal myocardial lipid accumulation and myocardial fibrosis in HIV-infected individuals may be related to impairment in cardiac function. Therefore, we examined intramyocardial lipid content, using MRS, in a cohort of HIV-infected adults and compared these findings with those for healthy controls, and we evaluated the relationships between measures of myocardial steatosis, fibrosis, and myocardial strain, as well as biomarkers of immune activation.

METHODS

Subjects

We prospectively evaluated 95 HIV-infected adults and 30 age-, sex-, and race-matched controls from April 2010 to May 2013 at the National Institutes of Health (NIH) Clinical Research Center in Bethesda, Maryland. Subjects were recruited through self-referral and in response to local advertisements. Participants were excluded if they had a known history of cardiovascular disease, a contraindication to MRI, or an estimated glomerular filtration rate of <60 cm³/min/1.73 m². There were no restrictions regarding ARV medication use or CD4⁺ T-cell count. Controls were documented as negative for HIV and were required to be healthy with no known significant medical conditions, including cardiovascular disease. Targeted recruitment of control subjects was performed to match the relative age (±5 years), sex, and racial distribution of the HIV-infected group at a ratio of approximately 3 to 1. Written informed consent was obtained from each participant, and the protocol was approved by the institutional review board of the National Institute of the Allergy and Infectious Diseases of the NIH.

Medical history, physical examination findings, and laboratory test results were obtained from each participant, including detailed review of ARV exposures and cardiovascular disease risk factors. Diagnosis of chronic hepatitis C virus (HCV) infection and diabetes were based on patient report and verified by medical records when available. Fasting lipid panel, glucose level, insulin level, homeostatic model of insulin resistance (HOMA-IR), CD4⁺ T-cell count, HIV load, and serologic biomarkers of inflammation, coagulation, and immune activation (eg, C-reactive protein [CRP], D-dimer, and pro-brain natriuretic peptide [pro-BNP], monocyte chemoattractant protein 1 [MCP-1], and lipopolysaccharide binding protein [LBP] levels) were determined. Subjects also underwent standard echocardiography.

Cardiac MRI/MRS

All studies were performed on a 3.0-T MR scanner (Verio; Siemens, Erlangen, Germany) with a 32-channel phased-array torso coil (In Vivo, Orlando, Florida) and combined with posterior coil elements. To quantify the intramyocardial triglyceride content, each participant underwent myocardial ¹H-MR spectroscopy (MRS; Figure 1). A 6–8-mL voxel was positioned in the interventricular septum. MRS was performed via electrocardiography–gated point-resolved spectroscopy (repetition time/echo time = [heart rate interval]/30 ms), with the navigator across the liver-lung interface to reduce the effect of breathing. One spectrum was recorded with water suppression (32 averages), and another spectrum (8 averages) was recorded without water suppression. The same sequence was used for liver fat measurement, with the MRS voxel placed over the right hepatic lobe. Eight averages of water-suppressed and 8 averages of no-water-suppressed spectra were acquired with breath holding. Fat content was quantified with Amares/MRUI and was related to the water signal in unsuppressed spectra, with findings expressed as the percentage of the fat signal relative to the water signal.

Figure 1.

Representative magnetic resonance spectroscopy findings in the interventricular septum with good spectra results showing triacylglycerol (TG) peak (A), peak fat signal with the water signal unsuppressed (B), and a suppressed water signal (C).

Myocardial strain was derived from grid-tagged cine MR images in the axial plane, with a temporal resolution of 35 msec. Strain images were analyzed using HARP (Diagnosoft, v.4.3.1; Morrisville, North Carolina) [23]. By convention, greater circumferential strain is more negative (indicative of greater shortening with contraction), while greater radial strain is more positive (indicative of more thickening with contraction). Therefore, impairment in circumferential strain is less negative (and indicates less shortening), and impairment in radial strain is less positive. The distribution of visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) were quantified by a single-shot fast spin echo sequence (slice thickness, 10 mm). Three transverse images were acquired at the level of the fifth lumbar vertebrae during 1 breath hold. VAT and SAT were identified by selecting the region of interest and thresholding the pixel signal intensity, using QMASS (QMass 7.2; Medis, the Netherlands).

Gadopentetate dimeglumine (0.15 mmol/kg; Berlex; Bayer Healthcare, New Jersey) was administered to assess myocardial scarring 15 minutes after contrast administration, using the late gadolinium enhancement method [24]. Myocardial T1 mapping was performed by means of the modified Look-Locker inversion recovery (MOLLI) sequence technique [25]. Single-slice T1 mapping was performed before and at 2 time points between 10 and 25 minutes after contrast injection. The MOLLI sequence acquired a set of 8 source images in the mid ventricle with 1 breath hold (11 heartbeats), allowing the reconstruction of 1 parametric T1 map [26]. A region of interest was manually drawn on the core of myocardium. Fibrosis index scores were evaluated by the extracellular volume fraction (ECV), calculated as [partition coefficient] × [1 − hematocrit], where the partition coefficient was determined by the slope of the linear relationship of (1/T_1myo vs 1/T_1blood) at precontrast and postcontrast time points.

Statistical Analysis

Group comparisons were performed using Student t tests or χ² statistics as appropriate. Nonnormally distributed variables (CRP, D-dimer, intramyocardial lipid content, pro-BNP, LBP, MCP-1, TIMP-1, and triglyceride levels) were log transformed to approximate a normal distribution and are presented as median and interquartile ranges, and group differences were tested using a Wilcoxon test. Two-sided P values of <.05 were used to determine statistical significance. Univariate linear regression analyses were calculated to identify variables associated with intramyocardial lipid content both in the entire study population and within HIV-infected subjects only. We evaluated the relationship between years of ARV exposure and years of subclasses of ARVs (nonnucleoside reverse-transcriptase inhibitors [NNRTIs] and protease inhibitors [PIs]). The study was not sufficiently large to evaluate individual ARV agents. Multivariate regression models were constructed to include all variables identified as statistically significant in univariate regression, as well as important potential confounders (eg, smoking pack-years, age, and use of lipid lowering therapy). Similar analyses were performed for myocardial fibrosis and cardiac strain parameters. Major findings were retested in separate subanalyses that excluded subjects with HCV infection, those with diabetes, and those not receiving ARV therapy. Statistical analyses were completed using SAS JMP, version 11.0.

RESULTS

The study groups were similar with respect to age, sex, and race (Table 1). The majority of HIV-infected subjects were receiving ARV therapy, and 84.2% had HIV loads below the limit of detection (<50 copies/mL). One subject not receiving ARV therapy was a known elite controller, and all others not receiving ARV therapy (n = 6) had detectable HIV. HIV-infected subjects were more likely to have ever smoked and had greater mean number of smoking pack-years. More HIV-infected subjects than controls were receiving therapy to decrease lipid levels (P = .06), and of the subjects receiving such therapy, 84% were receiving a statin. Metabolic parameters and biomarkers of inflammation and immune activation are presented in Table 2. The HIV-infected group had an increased fasting glucose level, compared with controls, but the groups were similar with respect to insulin level, HOMA-IR, and free fatty acid level. The mean low-density lipoprotein cholesterol was lower in the HIV-infected subjects, but there were no statistical differences in total or high-density lipoprotein cholesterol or triglyceride levels between the groups. Median D-dimer values were higher in the control group, but in both groups the upper quartile was within the normal range for the assay (<0.50 µg/mL fibrinogen equivalent units). MCP-1 and TIMP-1 levels were higher in the HIV-infected group, but there was no difference in LBP level between the groups.

Table 1.

Demographic and Clinical Characteristics Among Subjects With and Controls Without Human Immunodeficiency Virus (HIV) Infection

Characteristic	HIV-Infected Group (n = 95)	Control Group (n = 30)	P Value
Age, y	49 ± 10	46 ± 8	…
Sex
Male	71 (75)	22 (73)	…
Female	24 (25)	8 (27)
Race
White	27 (28)	8 (27)	…
Black	55 (58)	18 (60)
Hispanic	9 (10)	4 (13)
Mixed	1 (1)	0 (0)
Asian	3 (3)	0 (0)
Body mass indexa	28.0 ± 5.4	29.8 ± 4.3	.07
Smoking history
Current	18 (19)	2 (7)	.08
Ever	62 (65)	10 (33)	.002
Pack-years	11.3 ± 22.3	1.6 ± 3.7	.0001
Diabetes	9 (10)	0 (0)	.09
Hepatitis C	18 (19)	0 (0)	.01
Framingham risk score, %	4.5 ± 4.9	3.4 ± 3.2	.16
Blood pressure, mm Hg
Systolic	127 ± 13	123 ± 13	.13
Diastolic	79 ± 9	76 ± 10	.12
Current receipt of treatment to lower lipid level	33 (34.7)	5 (16.7)	.06
Duration HIV diagnosis, years	14 ± 8	…	…
ARV use
Current	88 (93)	…	…
Duration, y	9 ± 6	…	…
CD4+ T-cell count, cells/μL
Current	615 ± 276	904 ± 349	.0002
Nadir	235 ± 182	…	…
Undetectable HIV load (<50 copies/mL)	80 (84.2)	…	…

Data are mean ± SD or no. (%) of subjects.

Abbreviation: ARV, antiretroviral.

^a Defined as the weight in kilograms divided by the height in meters squared.

Table 2.

Metabolic and Inflammation Biomarkers Among Subjects With and Controls Without Human Immunodeficiency Virus (HIV) Infection

Biomarker	HIV-Infected Group (n = 95)	Control Group (n = 30)	P Value
Fasting glucose level, mg/dL	97 ± 18	92 ± 8	.03
Fasting insulin level, µIU/mL	9.9 ± 8.3	9.2 ± 5.5	.6
HOMA-IR finding, %	2.6 ± 2.8	2.1 ± 1.4	.3
Free fatty acid level, μEq/L	0.47 ± 0.21	0.47 ± 0.20	.99
Cholesterol level, mg/dL
Total	168 ± 30	181 ± 40	.11
HDL	47 ± 15	45 ± 13	.6
LDL	94 ± 29	112 ± 34	.01
Triglyceride level, mg/dL	117 (86–173)	91 (62–188)	.13
D-dimer level, µg/mL FEU	0.26 (0.22–0.37)	0.38 (0.23–0.48)	.03
CRP level, mg/L	1.76 (0.78–4.06)	2.06 (0.79–4.46)	.7
Pro-BNP level, pg/mL	29 (11–54)	17 (13–44)	.19
MCP-1 level, pg/mL	394 (294–779)	260 (195–395)	.0001
LBP level, ng/mL	413 (282–534)	358 (227–548)	.5
TIMP-1 level, ng/mL	11.6 (8.2–15.0)	8.8 (6.9–12.1)	.046

Data are mean ± SD or median (interquartile range).

Abbreviations: CRP, C-reactive protein; FEU, fibrinogen equivalent units; HDL, high-density lipoprotein; HOMA-IR, homeostatic model of insulin resistance; LBP, lipopolysaccharide binding protein; LDL, low-density lipoprotein; MCP-1, monocyte chemoattractant protein 1; pro-BNP, pro-brain natriuretic peptide; TIMP-1, tissue inhibitor of metalloproteinase 1.

HIV-infected individuals had a normal ejection fraction. However, there was evidence of subclinical systolic dysfunction, compared with control subjects. Systolic radial strain and strain rate, as well as epicardial-endocardial circumferential strain, were impaired, compared with controls (Table 3). In a multivariate regression analysis including age, sex, body mass index (BMI), use of therapy to reduce lipid levels, and smoking pack-years, HIV status (P = .004) was the only variable independently associated with decreased radial strain. In sensitivity analyses excluding HCV-infected subjects, those with diabetes, or those not receiving ARV therapy, HIV status remained independently associated with impaired radial strain. Impaired radial strain rate and epicardial-endocardial circumferential strain were associated with greater intramyocardial lipid levels (r = 0.19, P = .04; r = 0.20, P = .03) and fibrosis index (r = 0.26, P = .007; r = 0.22, P = .03), respectively. Duration of ARV exposure did not correlate with measures of cardiac strain.

Table 3.

Myocardial Measurements and Regional Adipose Measurements Among Subjects With and Controls Without Human Immunodeficiency Virus (HIV) Infection

Measurement	HIV-Infected Group (n = 95)	Control Group (n = 30)	P Value
Lipid content
Intramyocardial, %	1.14 (0.55–1.85)	0.58 (0.36–1.58)	.04
Intrahepatic, %	7.0 ± 10.4	10.5 ± 14.9	.23
Abdominal fat level, mL
Subcutaneous	622 ± 375	650 ± 281	.68
Visceral	499 ± 228	467 ± 151	.41
Myocardial extracellular volume index	0.28 ± 0.04	0.26 ± 0.03	.02
Ejection fraction, %	62 ± 6	63 ± 4	.4
Systolic circumferential strain
Percentage	−15.8 ± 2.7	−16.4 ± 2.5	.31
Rate	−93.9 ± 18.9	−100.4 ± 23.7	.23
Systolic radial strain
Percentage	21.7 ± 8.6	30.5 ± 14.2	.004
Rate	99.7 ± 41.0	121.3 ± 49.6	.048
Systolic epicardial-endocardial circumferential strain, %	7.0 ± 2.5	8.5 ± 2.8	.02

Data are mean ± SD or median (interquartile range).

Intramyocardial lipid content was significantly higher in HIV-infected subjects, compared with controls (Table 3). Intramyocardial lipid content was positively associated with age (r = 0.22, P = .01), fasting glucose level (r = 0.21, P = .02), triglyceride levels (r = 0.19, P = .04), and VAT volume (r = 0.37, P < .001). Intramyocardial lipid content did not correlate with CD4⁺ T-cell count, nadir CD4⁺ T-cell count, HIV load, or CRP, D-dimer, pro-BNP, MCP-1, or LBP levels. In a multivariate analysis, including age, sex, smoking pack-years, diabetes, glucose level, triglyceride level, use of therapy to reduce lipid levels, and VAT volume, HIV status remained an independent predictor of intramyocardial lipid content (P = .03), as did VAT volume (P = .0006) and female sex (P = .02). These observations did not change in subanalyses that excluded subjects with HCV infection or those not receiving ARV therapy.

Duration of ARV exposure was positively correlated with intramyocardial lipid content (r = 0.27, P = .007). No association between exposure to specific subclasses of ARVs (ie, NNRTIs or PIs) and intramyocardial lipid content was found. VAT volume was also correlated with years of ARV use (r = 0.31, P = .004). In a multivariate regression analysis, VAT volume (P = .02) and female sex (P = .02) were associated with intramyocardial lipid content, but years of ARV exposure was not significant (P = .2). We failed to identify a difference in mean intramyocardial lipid content (±SD) between HIV-infected patients with diabetes (n = 9) and those without diabetes (n = 86; 1.69% ± 1.13% and 1.42% ± 1.23%, respectively; P = .5). Similarly, within the HIV-infected cohort, there was no difference detected in intramyocardial lipid content, based on HCV status, history of illicit drug use, or current use of therapy to reduce lipid levels.

Focal myocardial scarring (identified by late gadolinium enhancement) was infrequent, and similar rates of focal scarring were present in the HIV-infected group, compared with the control group (8.6% and 7.7%, respectively; P = .8). In all cases, myocardial scar volume was <5% of the total myocardial mass. However, HIV-infected subjects had significantly greater indices of diffuse myocardial fibrosis (ie, greater extracellular volume index), compared with controls, as measured by MRI T1 mapping (Table 3). A multivariate regression model including HIV status and adjusting for age, sex, BMI, heart rate, systolic blood pressure, high-density lipoprotein and low-density lipoprotein cholesterol levels, use of therapy to reduce lipid levels, and hematocrit showed independent associations between myocardial fibrosis index scores and HIV status (P = .004), female sex (P = .002), and systolic blood pressure (P = .03). Again, these finding persisted when subjects with HCV infection, those with diabetes, or those not receiving ARV therapy were excluded. There was a positive correlation between fibrosis index and intramyocardial lipid content (r = 0.29, P = .005) among subjects with HIV infection. However, there was no association between fibrosis index and nadir CD4⁺ T-cell count, current CD4⁺ T-cell count, years of ARV exposure, HIV load, or the measured biomarkers of inflammation and immune activation.

Within the HIV-infected cohort, higher levels of pro-BNP and MCP-1 were correlated with decreased radial strain (pro-BNP, r = −0.28, P < .01; MCP-1, r = −0.43, P < .0001), as was increased LBP levels (r = −0.25, P = .02). Each correlation remained statistically significant in subsequent sensitivity analyses in which subjects with diabetes, those with HCV infection, or those not receiving ARV therapy were excluded. However, the correlations between pro-BNP, MCP-1, and LBP levels and radial strain were not observed within the control group, but this may have been limited by the smaller sample size of this group.

DISCUSSION

To investigate the potential relationship between subclinical cardiac dysfunction and myocardial lipid and fibrosis, this study evaluated subjects with a broad range of exposure to HIV and its therapies but without clinical cardiovascular disease. Despite normal ejection fraction, we identified significantly reduced systolic function (27% relative reduction in radial strain), greater intramyocardial lipid content (38%), and evidence of diffuse myocardial fibrosis (8%) in HIV-infected individuals, compared with age, sex, and race-matched controls. Impaired myocardial function as measured by strain parameters was associated with increased myocardial lipid level and fibrosis index in this cohort.

One brief report and 1 cohort study have used MRI/MRS to examine subjects with HIV infection receiving ARV therapy [27, 28]. In the United Kingdom cohort, subjects had more-overt cardiac disease than those in the current study (eg, 76% of the United Kingdom HIV-infected cohort had overt, focal myocardial scarring on MRI vs only 8% in the current study). Focal myocardial scarring represents macroscopic replacement of myocardium by collagen; the most common cause of focal scarring is myocardial infarction, but other conditions (eg, myocarditis, hypertension, and diabetes) are also associated with focal scarring. Although our US HIV-infected cohort had very little focal myocardial scarring, our cohort did have evidence of an expanded ECV index, which is associated with diffuse myocardial fibrosis. Diffuse myocardial fibrosis may represent a sequelae of subclinical myocarditis and, in the general population, is a histologic finding in the failing heart that is associated with adverse cardiac outcomes [29, 30]. The ECV index remained increased in HIV-infected subjects as compared to controls after adjustment for variables known to increase fibrosis index, such as age and lower hematocrit. Further, myocardial fibrosis index was positively correlated with intramyocardial lipid content but not with duration of ARV exposure, HIV viremia, or inflammatory markers in the HIV-infected subjects. Our findings suggest that cardiac fibrosis may be secondary to the downstream metabolic effects of HIV infection. Similar to the United Kingdom cohort, we found that the relationship between HIV infection and increased cardiac steatosis persisted after adjustment for potential confounders, such as smoking.

Myocardial strain was lower in HIV-infected subjects, compared with control subjects. Holloway et al [27] proposed their observed cardiac dysfunction in HIV-infected subjects was due to cardiac steatosis and fibrosis, but they did not report a correlation between myocardial strain and these parameters. We, however, observed overall correlations between cardiac steatosis and fibrosis with depressed cardiac strain in subjects with a minor degree of focal scarring that did not differ from age- and sex-matched controls (8%–9%). Our findings strongly suggest that diffuse rather than focal myocardial injury may be the sequela of HIV infection and myocardial steatosis.

The relationship between visceral abdominal fat volume and myocardial lipid content has been well established in obese populations [16, 31]. HIV infection is associated with increased central adiposity, which conveys a 2-fold increased risk in 5-year mortality [9], but the present study is among the first to characterize the relationship between VAT and intramyocardial steatosis in HIV-infected individuals. Increased VAT, in part because of the side effects of ARV exposure, was the strongest independent predictor of intramyocardial lipid level. This suggests that the altered metabolic activity of the visceral fat compartment may be a driving force of observed myocardial effects. Studies of non–HIV-infected obese populations have shown that increased intramyocardial triglyceride content, which correlates with the visceral fat, is inversely associated with stroke volume [16]. Therefore, visceral adiposity in HIV-infected individuals may increase the risk for clinical cardiac dysfunction through cardiac steatosis.

MCP-1 is a chemokine that is important in monocyte and macrophage migration and is a marker of chronic inflammation and immune activation. In patients with chronic heart failure, elevated MCP-1 levels are associated with a decreased left ventricular ejection fraction, and, in one study, higher MCP-1 levels predicted subsequent cardiac events [32]. We found that increased MCP-1 levels were strongly associated with the degree of impairment in cardiac strain. LBP, a marker of chronic inflammation and an acute phase protein made in response to lipopolysaccharide (LPS), was also associated with impaired cardiac strain in the HIV-infected cohort. In a case-control study, Sandler et al [33] found that soluble CD14, a marker of LPS-mediated monocyte activation, although not associated with cardiovascular events, was a predictor of mortality in HIV-infected individuals. Therapeutic interventions targeted at attenuating immune activation in chronic HIV infection are under active investigation. Our data support this as an approach that may modify or reduce end organ injury that accumulates through this process.

Previously uncharacterized differences related to sex in both cardiac steatosis and cardiac fibrosis indices were observed. Female sex was an independent predictor of intramyocardial lipid content in both HIV-infected subjects and controls. Earlier data in HIV-infected populations suggest that women may be more susceptible than their male counterparts to the metabolic effects of ARVs [34]. In the general population, female sex independently predicts higher myocardial fatty acid esterification and a lower percentage of fatty acid oxidation in the LV of the heart [35], suggesting that cardiomyocytes may have an increased fatty acid deposition-to-use ratio in females. Therefore, HIV-infected women receiving ARVs may be particularly susceptible to abnormal cardiac lipid deposition. As seen in large population studies [30], female sex was independently associated with greater myocardial fibrosis indices, compared with male sex.

The cross-sectional design of the present study limits the interpretation of observed associations and cannot establish causality. HIV-infected and control subjects were allowed to self-refer, which may have introduced bias to the sample selection. Subjects with known cardiovascular disease, though, were excluded, and therefore the abnormalities identified in cardiac steatosis and function are subclinical in nature. Prospective longitudinal studies evaluating these characteristics in patients initiating ARV therapy, as well as studies with clinical cardiovascular disease end points, are needed to fully appreciate the etiology and significance of cardiac steatosis, fibrosis, and impaired cardiac strain in HIV-infected individuals. Females were relatively underrepresented in the study population. Further studies focusing on cardiovascular disease in women are needed to better examine the relationship between HIV infection and cardiac steatosis.

Our study identified increased subclinical cardiac dysfunction in association with cardiac steatosis and fibrosis in HIV-infected adults. Given the known increased risk of cardiovascular disease in persons living with HIV [36], it is important to identify risk factors and create targeted strategies to prevent progression of global cardiac dysfunction. We demonstrate that increased VAT is a strong independent predictor of myocardial steatosis, and as such, reducing visceral adiposity should be a target for strategies of lifestyle modification and cardiovascular risk reduction. Further, impaired cardiac strain tracked with markers of chronic inflammation and immune activation, which may serve as targets for the development of therapeutic strategies to optimize long-term cardiovascular health in persons living with HIV.