|Home | About | Journals | Submit | Contact Us | Français|
Whether hepatitis C (HCV) confers additional coronary heart disease (CHD) risk among Human Immunodeficiency Virus (HIV) infected individuals is unclear. Without appropriate adjustment for antiretroviral therapy, CD4 count, and HIV-1 RNA, and substantially different mortality rates among those with and without HIV and HCV infection, the association between HIV, HCV, and CHD may be obscured.
We analyzed data on 8579 participants (28% HIV+, 9% HIV+HCV+) from the Veterans Aging Cohort Study Virtual Cohort who participated in the 1999 Large Health Study of Veteran Enrollees. We analyzed data collected on HIV and HCV status, risk factors for and the incidence of CHD, and mortality from 1/2000–7/2007. We compared models to assess CHD risk when death was treated as a censoring event and as a competing risk. During the median 7.3 years of follow-up, there were 194 CHD events and 1186 deaths. Compared with HIV−HCV− Veterans, HIV+ HCV+ Veterans had a significantly higher risk of CHD regardless of whether death was adjusted for as a censoring event (adjusted hazard ratio (HR)=2.03, 95% CI=1.28–3.21) or a competing risk (adjusted HR=2.45, 95% CI=1.83–3.27 respectively). Compared with HIV+HCV− Veterans, HIV+ HCV+ Veterans also had a significantly higher adjusted risk of CHD regardless of whether death was treated as a censored event (adjusted HR=1.93, 95% CI=1.02–3.62) or a competing risk (adjusted HR =1.46, 95% CI=1.03–2.07).
HIV+HCV+ Veterans have an increased risk of CHD compared to HIV+HCV−, and HIV−HCV− Veterans.
HIV infection is associated with increased risk of CHD.1, 2 Whether HCV infection confers additional risk over and above that of HIV infection remains unclear and the reports are inconsistent.3–5 Using data from participants in the Veterans Aging Cohort Study (VACS) “Virtual Cohort”6 and the 1999 Large Health Survey of Veteran
Enrollees7 we first examined the association between HIV+HCV+ Veterans and the risk of future CHD events compared to HIV−HCV− Veterans who were demographically and behaviorally similar. Next, we compared HIV+HCV+ with HIV+HCV− Veterans in order to also account for baseline, duration of, and recent use of class of antiretroviral therapy, and baseline and recent HIV-1 RNA and CD4 count levels. Finally, because HIV+HCV+ patients are at higher risk of death than HIV+HCV− and HIV−HCV− patients,1, 8 we constructed competing risk models to ensure that the association between HIV, HCV, and incident CHD was not obscured by an excess (competing) risk of death.
The Virtual Cohort is a cohort of HIV infected and age, gender, race/ethnicity, and clinical site matched HIV uninfected participants identified from United States Department of Veterans Affairs (VA) administrative data in the fiscal years 1998–2003 using a modified existing algorithm.6 This cohort consists of data from the immunology case registry; the VA HIV registry; the pharmacy benefits management database, a VA centralized database of outpatient prescriptions; the decision support system, a national database of VA clinical and financial data including laboratory data; and the National Patient Care Database. The 1999 Large Health Study of Veteran Enrollees, a survey administered between June 1999 and January 2000, was designed to assess the health status of Veterans in the Veterans Health Administration. It contained the Veterans RAND 36 Item Health Survey (VR-36) instrument and other measures of sociodemographic and economic status.7, 9 The institutional review boards at the University of Pittsburgh, Yale University, and the West Haven VAMC approved this study.
All participants in the Virtual Cohort / Large Health Study data set (n=13,250) were eligible for the present study. After excluding participants with baseline cardiovascular disease (CVD) and cancer, failure to answer baseline questions about cancer, and women due to small numbers, our final sample was 8579 participants.
We categorized participants into one of four groups: HIV+HCV+, HIV+HCV−, HIV−HCV+, and HIV−HCV− uninfected (referent). HIV infection was defined as a participant with ≥ 1 inpatient and/or ≥ 2 outpatient International Classification of Diseases (ICD)-9 codes for HIV infection and confirmed by the participant’s presence in the immunology case registry.6 HCV infection was defined as a positive HCV antibody test or ≥ 1 inpatient and/or ≥2 outpatient ICD-9 codes for this diagnosis.10
We used ICD-9 codes (410–411) to identify incident acute myocardial infarction and unstable angina events (CHD) events from January 2000 (entry into the Large Health Study) to July 2007. We chose these ICD-9 codes based on the positive predictive value of the diagnosis of acute myocardial infarction in the Veterans Health Administration11 and the high agreement between these ICD-9 codes (both inpatient and outpatient ICD-9 codes) and formal reviewer adjudication in the Cardiovascular Health Study.12 Follow-up time was to a CHD or death event or the last known visit within the VA system. In contrast to a prior study that used only inpatient VA hospitalization data to assess the risk of cardiovascular diseases among HIV infected Veterans,13 the present study incorporates both inpatient and outpatient ICD-9 codes. Outpatient ICD-9 codes were used to capture participants who had non-VA hospitalized CHD events but subsequently received their follow-up CHD care as part of the VA outpatient clinical care system. In the Veterans Aging Cohort Study, all coronary heart disease events are being formally adjudicated using established protocols as part of an NHBLI funded collaborative research initiative. This process is not yet complete. We have observed after reviewing charts of nearly 4,000 participants that approximately 90% of the acute myocardial infarctions occurring within and outside the VA hospital system were captured by using inpatient and outpatient ICD-9 codes 410 and 411. We confirmed deaths using the VA vitals status file; the Social Security administration death master file, the Beneficiary Identification and Records Locator Subsystem, and the VHA medical SAS inpatient datasets.
Sociodemographic data included age, race/ethnicity, and level of education. Hypertension was defined using ICD-9 codes. Diabetes was diagnosed using a combination of glucose measurements, diabetic medication use, and or ICD-9 codes. We calculated body mass index (kg/m2) using self-reported height and weight. Smoking was categorized as history of current smoking, past smoking or never smoking. Hypercholesterolemia was defined as use of a HMGCoA reductase inhibitor medication; a total cholesterol value >200 mg/dL; or ICD-9 codes. History of cocaine dependence and abuse was defined using ICD-9 codes (304.20–304.23 and 305.6–305.63). History of alcohol dependence and abuse were defined using ICD-9 codes based on prior work in the VACS. 14
Among the HIV infected Veterans we collected data on baseline and recent CD4 cell counts and HIV -1 RNA. Baseline CD4 count and HIV -1 RNA measurements were from 180 days prior to and up to 180 days after the time of enrollment in the Large Health Study and recent measurements were the CD4 count and HIV -1 RNA collected closest to the date of the incident CHD event, mortality event, or the date of last follow up observation. Duration of antiretroviral therapy (ART) was calculated using the number of days a participant was taking ART. Duration of ART, calculated based on prescription refill data in the pharmacy benefits management database, was available from the time of enrollment into the large health study through June 2005. We calculated the duration of ART by drug classes: protease inhibitors (PI), non-nucleoside reverse transcriptase inhibitors (NNRTI), and nucleoside reverse transcriptase inhibitors (NRTI). We also analyzed recent ART because recent studies suggest that ART taken within 6 months of a CHD event is significantly associated with an increased risk of CHD15, 16. Recent ART was defined as the use of ART by drug classes within 180 days of an incident CHD or mortality event or the date of last ART prescription recorded among those who did not have a CHD or mortality event. All ART medications that were available on the VA formulary during the time period of this study were considered in the analyses.
Descriptive statistics for all variables by HIV and HCV status were assessed using t-tests or its nonparametric counterpart for continuous variables and chi-square test or Fisher’s exact test for categorical variables. When treating death as a censoring event, Cox proportional hazard models were used to estimate the hazard ratio (HR) and 95% confidence intervals (CI) for incident CHD associated with HIV and HCV status after adjusting for confounders. The proportional hazards assumption was assessed using the Grambsch-Therneau method.17 We calculated age and race/ethnicity adjusted incident CHD per 1000 person-years and mortality rates per 100 person-years. Multiple imputation was used to generate 5 data sets with complete covariate values to increase the robustness and efficiency of the estimated hazard ratio.
This sample represents HIV participants at the earliest period of combination ART. The mortality rates among those infected with HIV were substantially higher compared to Veterans not infected with HIV. Similarly, mortality rates among HIV+HCV+ were higher than those HIV−HCV+. Cox proportional hazard models assume “non informative censoring”; this means that patients censored (due to death or loss to follow up) do not depend on the prognosis of developing CHD. This assumption is violated if HIV+HCV+ Veterans have both substantially higher mortality rates and are at greater risk of CHD than HIV+HCV− and HIV−HCV− Veterans. Therefore, we conducted secondary analyses to ensure that the high mortality rates were not obscuring the association between the HIV and HCV status and incident CHD. To do this we constructed regression models that incorporated death as a competing risk. Fine and Gray methodology requires that participants who experience the competing event (i.e., death) would never have died of CHD or developed the event of interest (i.e., CHD) if they had not died of a non CHD death and completed the follow up period. 18 This seems an implausible assumption for our analyses. Thus, to adapt this technique to account for differing underlying risks of CHD, we needed to identify among those participants who died, probability of either dying of CHD or developing CHD in the follow up period. We calculated propensity scores using a multivariable logistic regression model to assess the likelihood of developing CHD given an observed set of CHD risk factors (age, race, body mass index, hypertension, diabetes, current smoking, hypercholesterolemia, and cocaine use).19 The model’s calculated c statistic (0.75—which represents good prediction of the likelihood of developing CHD), was used to assess the predictive capacity of the model.20 We then used the propensity score to identify, among those that died, who were at a high risk of dying of CHD or developing CHD in the follow up period. Each person was then reclassified into one of three categories: died of CHD; did not die of CHD but would have developed CHD in the follow up period; or did not die of CHD and did not develop CHD in the follow up period. Once participants were reclassified; we could then apply the Fine and Gray technique to our sample. It is important to note that during the median follow up period of 7.3 years, 48 of the 254 HIV+HCV+ Veterans who died had propensity score estimates consistent with death from CHD or developing CHD during the follow up period if they have not died from another cause.
For analyses restricted to HIV infected participants, we added HIV-1 RNA, CD4 count, and ART status, and traditional risk factors into the model. Our analyses considered baseline, duration of and recent ART as well as baseline and recent CD4 count and HIV -1 RNA. As ART data were only available through June 2005, these analyses were truncated to 2005.
We excluded participants who reported at baseline CHD, congestive heart failure, and stroke (n=3,116), cancer (except non-melanomatous skin cancer, n=856), or failed to answer questions regarding prevalent cancer (n=860). Women were excluded due to limited numbers (n=276). After these exclusions, our final sample size was 8579 participants (28% HIV infected). Among the study participants, the prevalence of all established cardiovascular risk factors differed by HIV and HCV status (p≤0.001 for all Table 1). Compared to HIV+HCV+ Veterans, the number of days on NNRTI and NRTI use and the prevalence of recent use of NRTI use was higher among HIV+HCV− Veterans (p<0.05 for all, Table 1).
During the 7.5 years of follow-up (median 7.3 years), there were 194 CHD events (2.3%) and 1186 deaths (13.8%). The age and race/ethnicity adjusted mortality rate per 100 person years was 4.59 for HIV infected Veterans (HIV+HCV+ and HIV+HCV− combined, 95% CI=4.47–.71 person years) and 1.38 for HIV uninfected (HIV−HCV+ and HIV−HCV− combined, 95% CI=1.35–1.40). The age and race/ethnicity adjusted CHD rate was 4.67 per 1000 person years for HIV infected (HIV+HCV+ and HIV+HCV− combined,95% CI=4.56–4.79 per 1000 person years) and 3.23 per 1000 person years for uninfected (HIV−HCV+ and HIV−HCV− combined, 95% CI=3.18–3.28 per 1000 person years). Compared with HIV uninfected Veterans (HIV−HCV+ and HIV−HCV− combined), HIV infected Veterans (HIV+HCV+ and HIV+HCV− combined) had an increased risk of incident CHD (Figure 1).
Compared with HIV−HCV− Veterans, HIV+HCV+ Veterans had significantly higher risks of incident CHD (Figure 2), whether death was censored or treated as a competing risk (Table 2). We examined the association between HIV+HCV+ and incident CHD using models that adjust for death as a competing risk (Table 2) because CHD and mortality rates were highest among co-infected Veterans.
Compared with HIV+HCV− Veterans, HIV+HCV+ Veterans had a higher risk of incident CHD, regardless of whether death was censored or treated as a competing risk (Table 3). This association persisted even after adjusting for traditional risk factors; baseline, duration of, and recent antiretroviral therapy (Table 3, models 1, 2, and 3, respectively); and baseline and recent HIV viral load and CD4 count (Table 3 model 3). Recent PI use (HR=2.15, 95% CI=0.99–4.65), recent NNRTI use (HR=1.40, 95% CI=0.70–2.77), and recent HIV-1 RNA>500 (HR=1.29, 95% CI=0.68–2.45) were not significantly associated with an increase in CHD. At baseline, classes of ART and duration of PI and NNRTI were also not significantly associated with CHD in multivariable models. Duration of NRTI per year was associated with a reduced risk of CHD (HR=0.69, 95% CI=0.53–0.90). The association between ART and mortality by class of ART was(HR=0.97, 95% CI=0.88–1.07) for PI; (HR=0.97, 95% CI=0.88–1.07) for NNRTI; and (HR=0.69, 95% CI=0.63–0.76) for NRTI, respectively.
Our findings suggest that HIV+HCV+ Veterans have an increased risk of CHD after adjustment for traditional CHD risk factors, use of ART; CD4 count and HIV -1 RNA compared with HIV+HCV−, and HIV−HCV− Veterans. This association persisted whether death was treated as a censored event or a competing risk.
Currently there are no prospective data describing the risk of incident CHD among HIV+HCV+ individuals compared with HIV− HCV− individuals who are similar demographically and behaviorally. The data describing the association between HCV and CHD among HIV infected individuals are sparse, the results are inconsistent and none of these studies used competing risk models.3, 4 However our results (Table 3, Model 2, with death as a censored event) are very consistent with another study of 19, 424 HIV infected Veterans which reported a significant association between HCV infection and cardiovascular diseases (HR=1.20, 95% CI=1.04–1.38) after adjusting for hypertension, age, type 2 diabetes, and smoking.5
We used competing risk models in the present study because HIV+ HCV+ Veterans had higher adjusted incident mortality rates. We conducted these analyses to ensure that mortality was not obscuring the association between HIV, HCV and incident CHD. Whether the referent group was HIV−HCV− or HIV+HCV− Veterans, HIV+HCV+ Veterans had a significantly increased risk of CHD. This association persisted regardless of whether a Cox or competing risk model was employed. Of note, whether the competing risk model increased (Table 2) or modestly decreased (Table 3) the estimation of CHD risk among HIV+HCV+ Veterans was largely determined by the mortality rates of and the propensity estimates for CHD death or developing CHD in the follow up period among those that died of a non CHD death among the referent groups. In our analyses among HIV infected Veterans, ART was associated with an increased risk of CHD and a reduced risk of mortality. Thus, ART almost certainly influenced the association between HIV, HCV, and the risk of CHD.
This analysis was not designed to determine whether HCV monoinfection is associated with increased risk of CHD. Among HIV uninfected individuals, several studies report that HCV infection is associated with a higher prevalence of CHD risk factors,21, 22 carotid atherosclerosis23–25 and CHD. 26, 27 Other studies, report no association.28–30 Our results did not demonstrate a significant association between HIV−HCV+ infection and incident CHD. However, the confidence intervals around these estimates were wide and did not preclude a clinically important difference in risk (up to 70–80% increased HR compared to HIV−HCV−). The width of these confidence intervals may be attributable to, the small number of CHD events among the HIV−HCV+ infected Veterans in this study.
The mechanism by which the HIV virus influences CHD risk remains unknown. One hypothesis suggests that chronic HIV infection causes microbial translocation of intestinal bacterial products leading to increased immune activation.31 Interestingly, markers of microbial translocation are also associated with the progression of HCV and cirrhosis, each of which may also stimulate an increased immune response.32 Whether HIV and HCV viruses increase the risk of CHD risk via microbial translocation and increased immune response is not known.
The present study has limitations. First, as our population consists entirely of men, our results may not be generalizable to women. Using ICD-9 codes for CHD diagnoses may have resulted in some misclassification, however, prior work suggests these codes have high positive predictive values and demonstrate good agreement with formal chart review adjudication processes. Moreover, the ICD-9 codes used in this analysis represent both inpatient and outpatient records. This is important because CHD events occurring outside the VA system could not be captured if only VA inpatient hospitalization data were used to identify CHD events. Third, cause of death data were not available, however, our competing risk models were specifically designed to identify among those that died who were a greatest risk of CHD death and to incorporate that risk in our analyses. Fourth, without HCV viral RNA values, we could not account for those individuals who might have spontaneously cleared their HCV infection or had false positive tests. Fifth, we did not incorporate treatment for HCV infection, however, our prior studies in clinical settings shows that treatment rates are low for HCV mono-infected people and even lower for HIV HCV co-infected patients.33
In conclusion, HIV+HCV+ Veterans have an increased risk of incident CHD after adjustment for traditional CHD risk factors, and HIV factors including antiretroviral therapy, CD4 count, and HIV-1 RNA-1 RNA compared to HIV+HCV−, and HIV−HCV− Veterans. Further investigations should focus on the mechanism of HCV infection and the risk of CHD among HIV infected individuals.
Funding Sources: National Institute on Alcohol Abuse and Alcoholism, 2U10 AA 13566 and K23 AA015914, and National Heart Lung & Blood Institute 1RO1HL095136-03
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Journal Subject Codes:  Chronic ischemic heart disease;  Epidemiology;  Risk Factors
Conflict of Interest Disclosures: None