Abstract
Background:
Nonalcoholic fatty liver disease (NAFLD) and HIV are independently associated with cardiovascular disease (CVD). However, the factors associated with NAFLD in persons living with HIV (PWH) and whether CVD is more frequent in PWH with NAFLD are currently unknown.
Methods:
From the Partners HealthCare Research Patient Data Registry, we identified PWH with and without NAFLD between 2010 and 2017. NAFLD was defined using validated histological or radiographic criteria. CVD was defined by an ICD-9 diagnosis of coronary artery disease, myocardial infarction, coronary revascularization, peripheral vascular disease, heart failure, transient ischemic attack, or stroke and was confirmed by clinician review. Multivariable logistic regression was performed to examine the relationship between NAFLD and CVD.
Results:
Compared with PWH without NAFLD (n = 135), PWH with NAFLD (n = 97) had higher body mass index and more frequently had hypertension, obstructive sleep apnea, diabetes mellitus, dyslipidemia, coronary artery disease, and CVD (P < 0.01 for all). PWH with NAFLD were also more likely to have CD4+ T-cell counts (CD4) <200 cells/mm3. In multivariable models, the presence of NAFLD was significantly associated with CVD (adjusted odds ratio 3.08, 95% confidence interval: 1.37 to 6.94) and CD4 <200 cells/mm3 (adjusted odds ratio 4.49, 95% confidence interval: 1.74 to 11.55).
Conclusion:
In PWH, CVD was independently associated with prevalent NAFLD after controlling for traditional CVD risk factors. NAFLD was also associated with CD4 <200 cells/mm3, suggesting that immune dysfunction may be related to NAFLD. Both CVD and low CD4+ count as risk factors for NAFLD require prospective evaluation.
Keywords: NAFLD, HIV, PWH, CAD, CVD
INTRODUCTION
Nonalcoholic fatty liver disease (NAFLD), is a frequent cause of chronic liver disease worldwide, affecting up to 25% of the population. Among persons with HIV (PWH), the prevalence of NAFLD may be as high as 40%.1–3 However, the risk factors for NAFLD in PWH and specifically the association between NAFLD and cardiovascular disease (CVD) among PWH are unknown. We therefore aimed to compare the cardiometabolic profiles and CVD prevalence in PWH with NAFLD to PWH without NAFLD. We hypothesized that compared to PWH alone, PWH with NAFLD would have less favorable cardiometabolic profiles and more prevalent CVD, even after controlling for traditional CVD risk factors and that CVD would be associated with NAFLD.
METHODS
Study Population
We performed a cross-sectional study comparing PWH with NAFLD to PWH without NAFLD. Data were extracted using the Partners Research Patient Data Registry and electronic medical records. In the current study, individuals were identified by an ICD-9 diagnosis of HIV from 2010 to 2017. Within this cohort of PWH, a separate query identified those with an ICD-9 diagnosis of NAFLD. All diagnoses were confirmed by clinician review. NAFLD diagnoses were restricted to biopsy and imaging to reduce misclassification. Biopsy diagnoses were made by liver histology demonstrating >5% macrovesicular hepatic steatosis. Imaging diagnoses were made using previously validated radiographic criteria for noncontrast computerized tomography (CT) (liver:spleen hounsfield unit ratio <1) or for ultrasonography (diffuse increase in hepatic echogenicity).4,5 These determinations were made by clinician review of radiology reports. A second reviewer validated findings in 20% of cases. Individuals identified in the original query without an ICD-9 code for NAFLD and with clinician-confirmed normal liver imaging and/or histology were selected as non-NAFLD controls.
Individuals who had not been seen in the Partners Healthcare system for >10 years, with >20% missing data or with less than 2 outpatient visits per year were excluded. Adults with other forms of chronic liver disease, including hepatitis B or C infection and those with significant alcohol use (>20 g daily for men or >10 g daily for women) were also excluded.
Data Collection
All data were extracted at the time of NAFLD diagnosis for the PWH with NAFLD group and at time of normal imaging or biopsy results for the control non-NAFLD PWH group. Demographic, clinical, and laboratory data were collected. Lifetime antiretroviral therapy (ART) exposure, CD4+ count, and time since diagnosis of HIV were all collected. Individual ART medication exposures were separated into their respective classes.
Diabetes mellitus (DM) was identified by ICD-9 code and confirmed by hemoglobin A1C values ≥ 6.5% and/or use of diabetic medications. Dyslipidemia was identified by ICD-9 code and confirmed by either low-density lipoprotein (LDL) >160 mg/dL, high-density lipoprotein <40 mg/dL for men and <50 mg/dL for women, triglycerides level ≥150 mg/dL, or use of lipid lowering medication. Hypertension (HTN) and obstructive sleep apnea (OSA) were also identified by ICD-9 codes. CVD was defined by an ICD-9 diagnosis of coronary artery disease (CAD), congestive heart failure (CHF), peripheral vascular disease (PVD), stroke, transient ischemic attack (TIA), myocardial infarction, or coronary revascularization. These diagnoses were confirmed by clinician review.
Statistical Analysis
Descriptive statistics were calculated for the demographic and clinical characteristics. Categorical variables were described as numbers with percentages and continuous variables as mean with SDs. Univariate comparisons were performed using t-tests or Wilcoxon rank-sum tests for continuous variables, depending on the normality of the distribution, and categorical variables were analyzed using χ2 tests or Fisher exact tests.
To examine the association between NAFLD and CVD we constructed a multivariable logistic regression model that included CVD. Additional covariates for the multivariable models were selected using a threshold of P < 0.01 from univariate analyses and were retained in a step-wise, backward selection process. These covariates included body mass index (BMI), HTN, OSA, smoking, dyslipidemia, DM, CD4+ - cell count <200 cells/mm3, and diagnosis of HIV in the last 10 years.
Data were missing in less than 10% of cases. Missing data were imputed from the mean value of the variable. All data were analyzed using SAS Studio software, Version 3.71 (SAS Institute, Cary, NC).
RESULTS
Baseline Characteristics
Two hundred thirty-two adults with HIV met study inclusion criteria, 97 with NAFLD and 135 without NAFLD. Within the NAFLD group, 13 (13%) were diagnosed by biopsy and 84 (87%) by imaging (50 by ultrasound, 30 by CT, and 4 by MRI). Baseline demographics are shown in Table 1. There were no significant differences in age, sex, race, or tobacco use between groups. The cohort was predominantly men and white, and the mean age was 54 years in both groups.
TABLE 1.
Baseline Characteristics and Cardiovascular Outcomes in Individuals With HIV With and Without NAFLD
| Characteristics | HIV and NAFLD (n = 97) | HIV without NAFLD (n = 135) | P |
|---|---|---|---|
| Demographics and clinical features | |||
| Male sex, n (%) | 74 (76%) | 103 (76%) | 0.90 |
| Race/ethnicity, n (%) | 0.18 | ||
| Black | 16 (16%) | 36 (27%) | |
| White | 58 (60%) | 69 (51%) | |
| Hispanic | 12 (12%) | 18 (13%) | |
| Others | 11 (12%) | 12 (9%) | |
| Age, years (mean ± SD) | 54 ± 9 | 54 ± 7 | 0.68 |
| Smoking, n (%) | 10 (10%) | 23 (17%) | 0.15 |
| BMI, kg/m2 (mean ± SD) | 31.0 ± 6 | 27.8 ± 6 | <0.001 |
| HTN, n (%) | 51 (53%) | 47 (35%) | 0.007 |
| OSA, n (%) | 14 (14%) | 6 (4%) | 0.008 |
| T2DM, n (%) | 26 (27%) | 18 (13%) | 0.01 |
| Diagnosis of HIV in last 10 years, n (%) | 28 (29%) | 23 (17%) | 0.02 |
| Laboratory values, mean ± SD | |||
| Total cholesterol, mg/dL | 184 ± 44 | 134 ± 43 | 0.32 |
| HDL, mg/dL | 40 ± 14 | 52 ± 19 | <0.001 |
| LDL, mg/dL | 103 ± 36 | 110 ± 35 | 0.14 |
| Triglycerides, mg/dL | 254 ± 274 | 140 ± 76 | <0.001 |
| AST, U/L | 48 ± 52 | 27 ± 13 | <0.001 |
| ALT, U/L | 61 ± 88 | 25 ± 13 | <0.001 |
| Alkaline phosphatase, U/L | 110 ± 132 | 84 ± 29 | 0.03 |
| Total bilirubin, mg/dL | 0.72 ± 1 | 0.56 ± 1 | 0.20 |
| Platelets, ×109/L | 218 ± 68 | 223 ± 63 | 0.62 |
| Albumin, g/dL | 4.0 ± 1 | 5.0 ± 7 | 0.30 |
| INR | 1.0 ± 0.5 | 1.0 ± 0.2 | 0.57 |
| CD4+ count <200 cells/mm3, n (%) | 17 (18%) | 9 (7%) | 0.01 |
| Exposure to HIV drug classes, n (%) | |||
| Protease inhibitors | 48 (49%) | 78 (58%) | 0.21 |
| ntegrase inhibitors | 34 (35%) | 58 (43%) | 0.22 |
| Non-nucleoside reverse transcriptase inhibitors | 64 (66%) | 89 (66%) | 0.99 |
| Nucleoside reverse transcriptase inhibitors | 92 (95%) | 131 (97%) | 0.50 |
| Exposure to HIV medications that predispose to metabolic syndrome, n (%) | |||
| Didanosine | 8 (8%) | 13 (10%) | 0.70 |
| Stavudine | 14 (14%) | 25 (19%) | 0.41 |
| CVD outcomes | |||
| CAD | 21 (22%) | 6 (4%) | <0.001 |
| CHF | 9 (9%) | 5 (4%) | 0.08 |
| PVD | 8 (8%) | 3 (2%) | 0.06 |
| Stroke | 5 (5%) | 3 (2%) | 0.28 |
| TIA | 4 (4%) | 3 (2%) | 0.46 |
| MI | 6 (6%) | 1 (1%) | 0.02 |
| Coronary revascularization | 11 (11%) | 3 (2%) | 0.004 |
| Cardiac-related death | 1 (1%) | 0 (0%) | 0.418 |
| Composite CVD* | 29 (30%) | 14 (10%) | <0.001 |
Composite CVD includes: CAD, CHF, PVD, stroke, TIA, MI, and coronary revascularization.
ALT, alanine aminotransferase; AST, aspartate aminotransferase; CAD, coronary artery disease; CHF, congestive heart failure; HDL, high density lipoprotein; HTN, hypertension; INR, international normalized ratio; MI, myocardial infarction; PVD, peripheral vascular disease; T2DM, type 2 diabetes mellitus; TIA, transient ischemic attack.
Metabolic Factors Associated With NAFLD
Compared with PWH without NAFLD, PWH with NAFLD had higher mean BMI (31 ± 6 vs. 28 ± 5 kg/m2, P < 0.001) and were more likely to have HTN (53% vs. 35%, P = 0.007), OSA (14% vs. 4%, P = 0.008) and DM (27% vs. 13%, P = 0.01). PWH with NAFLD also had higher mean triglycerides (254 ± 273 vs. 140 ± 76 mg/dL, P < 0.001) and lower mean high density lipoprotein cholesterol (40 ± 14 vs. 52 ± 19 mg/dL, P < 0.001) (Table 1).
HIV-specific Factors Associated With NAFLD
PWH with NAFLD were more likely to have a CD4 <200 cells/mm3 compared with PWH without NAFLD (18% vs. 7%, P = 0.01), and more likely to have been diagnosed with HIV in the last 10 years (28% vs. 17%, P = 0.03). 73% of those with CD4 <200 cells/mm3 had detectable HIV viral load. There were no differences in class of ART exposure between PWH with NAFLD and PWH without NAFLD (Table 1).
Cardiovascular Disease and NAFLD
PWH with NAFLD were significantly more likely to have CVD (30% vs. 10%, P < 0.001) compared with PWH without NAFLD (Table 2). Among those with CVD, myocardial infarction and coronary revascularization were significantly more common among PWH with NAFLD (6% vs. 1%, P = 0.02; 11% vs. 2%, P = 0.004, respectively). On multivariate analysis, to assess factors associated with NAFLD and including BMI, HTN, OSA, smoking, dyslipidemia, DM, CD < 200, and diagnosis of HIV in the last 10 years, CVD, BMI, and a CD4 <200 cells/mm3 were significantly associated with NAFLD (Table 2). A second analysis was conducted limiting NAFLD diagnoses to MRI, CT, or biopsy (excluding ultrasound) and the findings remained significant.
TABLE 2.
Factors Associated With NAFLD in Individuals With HIV
| Variable | OR (95% CI) |
|---|---|
| BMI | 1.10 (1.04 to 1.17) |
| HTN | 1.36 (0.71 to 2.60) |
| OSA | 1.89 (0.59 to 6.02) |
| Smoking | 0.74 (0.30 to 1.78) |
| Dyslipidemia | 1.67 (0.89 to 3.14) |
| T2DM | 1.13 (0.51 to 2.52) |
| CD4+ count <200 | 4.67 (1.82 to 12.02) |
| Diagnosis of HIV in last 10 yrs | 1.00 (0.96 to 1.03) |
| CVD* | 3.08 (1.37 to 6.94) |
Composite CVD includes: CAD, CHF, PVD, stroke, TIA, MI, and coronary revascularization.
CAD, coronary artery disease; CHF, congestive heart failure; CI, confidence interval; HTN, hypertension; MI, myocardial infarction; OR, odds ratio; PVD, peripheral vascular disease; T2DM, type 2 diabetes mellitus; TIA, transient ischemic attack.
DISCUSSION
NAFLD is an increasing concern in the treatment and management of individuals with HIV. The current study used a cross-sectional cohort of PWH to identify metabolic and HIV-specific factors associated with NAFLD. Among PWH, there was a significant association between NAFLD and CVD that persisted after accounting for traditional CVD risk factors. In addition, higher BMI and CD4 <200 cells/mm3 were associated with NAFLD in PWH. Our findings have important implications to identify those at highest risk of NAFLD among PWH and highlight the important relationship between CVD and NAFLD in PWH.
To our knowledge, there are only 2 other studies conducted in the HIV population examining the relationships between NAFLD, HIV, and CVD. One cross-sectional study demonstrated that NAFLD was associated with higher coronary artery calcium (CAC) scores in PWH (odds ratio 3.8, 95% confidence interval: 1.5 to 9.6).6 This study only evaluated the association between CAC and Framingham risk scores, however, and did not assess CVD outcomes. A second cross-sectional study determined cardiovascular risk using the Framingham equation and the PROCAM score (based on the Prospective Cardiovascular Munster study), as well as CAC scores. In contrast to our study, CVD, as determined by these scores, was not associated with NAFLD in PWH.7 Our study builds on these prior analyses using hard clinical outcomes as opposed to surrogate markers of CVD.
We also observed that CD4 <200 cells/mm3 was significantly associated with NAFLD in PWH after controlling for other metabolic factors. 73% of those with low CD4 counts had detectable viral load, despite most of them being on ART. To date, published data regarding the relationship between CD4 and NAFLD are inconsistent. In contrast to our study, some have not demonstrated a relationship between low CD4 and NAFLD. In the Multicenter AIDS Cohort Study cohort, nadir CD4 was not associated with NAFLD8; however, the odds of NAFLD in men with HIV was actually lower than in men without HIV, a finding that is inconsistent with most other epidemiologic studies and could be the reason our findings were different. Similarly, in a case control study conducted in China, neither nadir nor current CD4 was associated with NAFLD.9 The cohort in this study was conducted in a 94% Chinese population, however, and therefore may not be generalizable to the general population or the more racially diverse composition of our study. It is important to note that poorer HIV control often leads to hepatic inflammation that could confound fibrosis assessment, thereby making CD4 count that much more important to screen for in PWH with NAFLD.10
In our study, exposure to specific ART classes was not associated with the presence of NAFLD. Many of the older classes of ARTs have demonstrated a strong association with components of the metabolic syndrome, including insulin resistance and dyslipidemia, as well as with hepatocyte injury, likely mediated through oxidative stress.8,11–14 A longer duration of exposure to these older ART agents, therefore, likely increases the risk of the development of NAFLD. Many of the newer ART agents, however, are not associated with these same effects. We did not specifically examine duration of exposure to particular drugs, and perhaps varying exposures to older and newer generation agents clouded this relationship among our cohort.
There were several limitations to our study. First, our analysis was cross-sectional and therefore can report only associations and not establish causality. It is difficult to ascertain whether NAFLD’s association with CVD in PWH is secondary to NAFLD itself or whether there may be other factors no captured in our analysis that may explain this association. In addition, our study used ICD-9 code diagnoses to capture clinical variables, which could lead to misclassification. However, we minimized this potential risk by conducting a detailed, clinician-led review of all patient records. Also, the diagnosis of NAFLD in our study was made predominantly by imaging, with a smaller subset having biopsy-confirmed diagnoses. Although these imaging modalities have been previously validated for diagnosing NAFLD, they have well-established limitations, including decreased sensitivity and specificity with increased BMI.15–18 Future studies would include those only with histologic data or use newer elasto-graphic imaging for the noninvasive assessment of fibrosis.
Despite these limitations, our study had several strengths. It is one of the few cross-sectional studies conducted in PWH with NAFLD to examine the association between CVD and NAFLD. In addition, in contrast to previous studies that have used surrogates of CVD, our study assessed hard outcomes and used clinician review to confirm all documented clinical, laboratory, and diagnostic data.
In summary, in this cohort of PWH, CVD, BMI, and CD4 count <200 cells/mm3 were significantly associated with NAFLD, even after controlling for traditional risk factors. These results have important implications for screening for NAFLD in PWH and identify novel factors associated with NAFLD in PWH. Future prospective work will be important to further understand these findings and to better understand the longitudinal impact of HIV on the natural history of liver disease and risk of progression to cirrhosis.
Acknowledgments
Supported by K23 DK122104 (T.G.S.).
K.E.C. consults for BMS, Gilead and Novo Nordisk. She also has grant funding from Boeheringer-Ingelheim and Novartis. J.L. has served as a consultant to Gilead Sciences and Merck, and receives research support from Gilead Sciences. The remaining author have no conflicts of interest to disclose.
Footnotes
Presented at: (1) Conference on Retroviruses and Opportunistic Infections (CROI); March 7, 2018; Boston, MA. (2) American Association for the Study of Liver Diseases (AASLD); November 9, 2018; San Francisco, CA.
REFERENCES
- 1.Younossi Z, Koenig A, Abdelatif D, et al. Global epidemiology of non-alcoholic fatty liver disease-meta-analytic assessment of prevalence, incidence and outcomes. Hepatology. 2015;64:73–84. [DOI] [PubMed] [Google Scholar]
- 2.Younossi Z, Henry L. Contribution of alcoholic and nonalcoholic fatty liver disease to the burden of liver-related morbidity and mortality. Gastroenterology. 2016;150:1778–1785. [DOI] [PubMed] [Google Scholar]
- 3.Maurice JB, Patel A, Scott AJ, et al. Prevalence and risk factors of nonalcoholic fatty liver disease in HIV-monoinfection. AIDS. 2017;31: 1621–1632. [DOI] [PubMed] [Google Scholar]
- 4.de Lédinghen V, Vergniol J, Foucher J, et al. Non-invasive diagnosis of liver steatosis using controlled attenuation parameter (CAP) and transient elastography. Liver Int. 2012;31:848–855. [DOI] [PubMed] [Google Scholar]
- 5.Castera L, Vilgrain V, Angulo P. Noninvasive evaluation of NAFLD. Nat Rev Gastroenterol Hepatol. 2013;10:666–675. [DOI] [PubMed] [Google Scholar]
- 6.Crum-Cianflone NF, Krause D, Wessman D, et al. Fatty liver disease is associated with underlying cardiovascular disease in HIV-infected persons. HIV Med. 2011;12:463–471. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Guaraldi G, Squillace N, Stentarelli C, et al. Nonalcoholic fatty liver disease in HIV-infected patients referred to a metabolic clinic: prevalence, characteristics, and predictors. Clin Infect Dis. 2008;47:250–257. [DOI] [PubMed] [Google Scholar]
- 8.Price JC, Seaberg EC, Latanich R, et al. Risk factors for fatty liver in the multicenter AIDS cohort study. Am J Gastroenterol. 2014;109:695–704. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lui G, Wong VWS, Wong GLH, et al. Liver fibrosis and fatty liver in Asian HIV-infected patients. Aliment Pharmacol Ther. 2016;44: 411–421. [DOI] [PubMed] [Google Scholar]
- 10.Kim HN, Nance R, Van Rompaey S, et al. Poorly controlled HIV infection: an independent risk factor for liver fibrosis. J Acquir Immune Defic Syndr. 2016;72:437–443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Coronel-Castillo CE, Qi X, Contreras-Carmona J, et al. Nonalcoholic fatty liver disease and nonalcoholic steatohepatitis in HIV infection: a metabolic approach of an infectious disease. Expert Rev Gastroenterol Hepatol. 2019;13:531–540. [DOI] [PubMed] [Google Scholar]
- 12.Sutinen J, Häkkinen AM, Westerbacka J, et al. Increased fat accumulation in the liver in HIV-infected patients with antiretroviral therapy-associated lipodystrophy. AIDS. 2002;16:183–193. [DOI] [PubMed] [Google Scholar]
- 13.Ahmed D, Roy D, Cassol E. Examining relationships between metabolism and persistent inflammation in HIV patients on antiretroviral therapy. Mediators Inflamm. 2018;2018:6238978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lana LGC, Junqueira DRG, Perini E, et al. Lipodystrophy among patients with HIV infection on antiretroviral therapy: a systematic review protocol. BMJ Open. 2014;4:e004088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Shannon A, Alkhouri N, Carter-Kent C, et al. Ultrasonographic quantitative estimation of hepatic steatosis in children with NAFLD. J Pediatr Gastroenterol Nutr. 2011;53:190–195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cheah MC, McCullough AJ, Goh GBB. Current modalities of fibrosis assessment in non-alcoholic fatty liver disease. J Clin Transl Hepatol. 2017;5:261–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Park YS, Park SH, Lee SS, et al. Biopsy-proven nonsteatotic liver in adults: estimation of reference range for difference in attenuation between the liver and the spleen at nonenhanced CT. Radiology. 2011; 258:760–766. [DOI] [PubMed] [Google Scholar]
- 18.Li Q, Dhyani M, Grajo JR, et al. Current status of imaging in nonalcoholic fatty liver disease. World J Hepatol. 2018;10:530–542. [DOI] [PMC free article] [PubMed] [Google Scholar]