Seropositivity for Anti-HCV Core Antigen is Independently Associated With Increased All-Cause, Cardiovascular, and Liver Disease-Related Mortality in Hemodialysis Patients

Background It is not known whether chronic or past hepatitis C virus (HCV) infection contributes to the high mortality rate in hemodialysis patients. Methods This prospective study of 1077 adult hemodialysis patients without hepatitis B virus infection used Poisson regression analysis to estimate crude and sex- and age-adjusted rates (per 1000 patient-years) of all-cause, cardiovascular, infectious disease-related and liver disease-related mortality in patients negative for HCV antibody (group A), patients positive for HCV antibody and negative for anti-HCV core antigen (group B), and patients positive for anti-HCV core antigen (group C). The relative risks (RRs) for each cause of death in group B vs group C as compared with those in group A were also estimated by Poisson regression analysis after multivariate adjustment. Results A total of 407 patients died during the 5-year observation period. The sex- and age-adjusted mortality rate was 71.9 in group A, 80.4 in group B, and 156 in group C. The RRs (95% CI) for death in group B vs group C were 1.23 (0.72 to 2.12) vs 1.60 (1.13 to 2.28) for all-cause death, 0.75 (0.28 to 2.02) vs 1.64 (0.98 to 2.73) for cardiovascular death, 1.64 (0.65 to 4.15) vs 1.58 (0.81 to 3.07) for infectious disease-related death, and 15.3 (1.26 to 186) vs 28.8 (3.75 to 221) for liver disease-related death, respectively. Conclusions Anti-HCV core antigen seropositivity independently contributes to elevated risks of all-cause and cause-specific death. Chronic HCV infection, but not past HCV infection, is a risk for death among hemodialysis patients.

To assess the contribution of past and chronic HCV infection among hemodialysis patients, we estimated the relative risks of all-cause and cause-specific death attributable to HCV antibody seropositivity and anti-HCV core antigen seropositivity.

METHODS Participants
The eligible participants were adult hemodialysis patients who participated in the KAREN study, a populationbased prospective study that has been conducted since 2003 in northern Japan (Figure 1). 28 A total of 1214 adult hemodialysis patients (80.6% of all hemodialysis patients in the study area; age 22 to 95 years; 779 men and 435 women) are included in the KAREN study. The participants in the KAREN study are patients who were undergoing adult hemodialysis in April 2003. A total of 137 patients who were positive for hepatitis B surface antigen were excluded. Ultimately, data from 1077 patients were analyzed. We ascertained the vital status of all subjects in a 5-year followup survey ( Figure 2). All participants gave written informed consent to participate. This study was approved by the Medical Ethics Committee of Iwate Medical University and conducted in accordance with the guidelines of the Declaration of Helsinki.

Data collection
The initial investigations in the KAREN Study consisted of a questionnaire, review of medical records, measurements of blood pressure and anthropometric data, and blood testing. The data gathering methodology was previously described. 21,28 Information on HCV antibody serology testing was collected by reviewing medical charts. 21 Results of anti-HCV antibody tests could not be obtained from chart review for 50 patients. Frozen serum samples from those patients were thawed and anti-HCV antibody tests were performed using a second-generation assay (Architect HCV, Abbott Laboratories, Japan). Frozen samples from patients who were positive for anti-HCV antibody were thawed, and HCV core antigen tests were performed using a chemiluminescent enzyme immunoassay (Lumispot Eiken HCV antigen, Eiken Chemical Co., LTD, Japan). 21 Outcomes Follow-up studies were performed annually at each center. Committee, based on the summaries. Disagreements regarding cause of death were discussed, and the final determination was reached by consensus. We identified the 3 major causes of death (cardiovascular, infectious disease-related, and liver disease-related) using codes form the Tenth Revision of the International Classification of Diseases (ICD-10; Table 1).

Classification and definition
The participants were divided into 3 groups based on the results of HCV antibody testing and anti-HCV core antigen testing at the baseline survey. Group A consisted of patients who were negative for anti-HCV antibodies (n = 968). Group B consisted of patients who were positive for anti-HCV antibodies and negative for anti-HCV core antigen antibodies (n = 55). Group C consisted of patients who were positive for both anti-HCV antibodies and anti-HCV core antigen antibodies (n = 79). These 3 groups were selected because they roughly corresponded to patients without HCV infection (group A), patients with past HCV infection (group B), and patients with chronic HCV infection (group C). 21 High blood pressure was defined as systolic blood pressure (SBP) in the highest quartile of this study population (SBP ≥169 mm Hg). Low blood pressure was defined as SBP in the lowest quartile (<140 mm Hg). Diabetes was defined as a nonfasting plasma glucose level of 200 mg/dL or higher, a plasma HbA1c of 6.5% or higher, use of antidiabetic medication, or a combination thereof. Dyslipidemia was defined as serum total cholesterol (TC) of 220 mg/dL or higher, serum high-density lipoprotein cholesterol (HDL-C) level less than 40 mg/dL, use of antidyslipidemic medication, or a combination thereof. High body mass index (BMI) was defined as a BMI of 27.5 kg/m 2 or higher. Low BMI was defined as a BMI less than 18.5 kg/m 2 . High-sensitivity C reactive protein (hs-CRP) level was considered high if it was in the highest quartile (≥3.6 mg/L  defined as a serum albumin level less than 3.5 mg/dL. A smoking habit was defined as current smoking. Regular drinking was defined as alcohol consumption on 5 or more days per week.

Statistical analysis
Risk factor-related variables were expressed as sex-and age-adjusted means plus 95% CI and compared across HCV infection status groups using analysis of covariance (ANCOVA). The hs-CRP level was expressed as a sex-and age-adjusted geometric mean plus 95% CI. The χ 2 test was used to compare frequencies.
We defined the follow-up period as the period from the initial survey to the first outcome or the end of observation. Individuals who were free of outcomes in the 5-year followup study were administratively censored. The cumulative probability of each cause of death was estimated using the Kaplan-Meier method, and differences in the cumulative probability of death were assessed by the log-rank test. Crude mortality rates and sex-and age-adjusted mortality rates were estimated in the 3 groups (groups A, B, and C) by Poisson regression analysis in which multivariate-adjusted mortality rate ratios and their 95% CIs were calculated in groups B and C, with those of group A serving as reference. The variables used in the multivariate adjustment were traditional risk factors, including age, male sex, high BMI, dyslipidemia, diabetes, high blood pressure, history of myocardial infarction, stroke, or malignant disease, smoking habit, and regular drinking habit (model A). Hemodialysis-related risk factors, including low BMI, low blood pressure, high CRP level, and hypoalbuminemia, were also additionally used as explanatory variables in model B. All P values were 2-tailed, and values less than 0.05 were considered to indicate statistical significance. The statistical package PASW (version 18.0, IBM Japan Inc., Tokyo, Japan) was used for the statistical analysis. Table 2 shows the baseline characteristics of the patients, stratified by HCV infection status. The proportions of patients in groups A, B, and C were 90.0%, 3.6%, and 6.5%, respectively. As compared with patients in group A, those in group C had significantly lower serum TC, serum low-density lipoprotein cholesterol (LDL-C), serum albumin, and serum creatinine levels, and lower platelet and white blood cell (WBC) counts (P < 0.05 for all tests). Patients in group B had significantly lower systolic blood pressure, TC, and LDL-C levels, and lower platelet and WBC counts than did patients in group A (P < 0.05 for all tests).

RESULTS
The proportion of current smokers in group C was the highest of the 3 groups (χ 2 = 6.47, P = 0.03). There was no significant difference among groups in the proportions of patients with chronic glomerulonephritis (χ 2 = 5.66, P = 0.06), diabetic nephropathy (χ 2 = 3.06, P = 0.22), diabetes mellitus (χ 2 = 5.41, P = 0.07), or past histories of myocardial infarction (χ 2 = 1.65, P = 0.44), stroke (χ 2 = 0.93, P = 0.63), or malignancy (χ 2 = 4.12, P = 0.13). There were 4233 observed patient-years, after 5 years of follow-up. The mean and median follow-up periods were 3.9 and 4.9 years, respectively. A total of 406 patients died during the 5-year observation period. Figure 3 shows the Kaplan-Meier estimated cumulative probabilities of death for patients in the 3 groups. The cumulative probability of all-cause death (upper left) in group C was significantly higher as compared with group A Fibrosis and cirrhosis of liver C22.0 Liver cell carcinoma (P = 0.007, log rank test), but not as compared with group B (P = 0.174). Group C also had higher probabilities of cardiovascular death (upper right, P = 0.033) and liver disease-related death (lower right, P < 0.001) as compared with group A. Group B did not have significantly higher probabilities of cardiovascular death (P = 0.118), infectious disease-related death (P = 0.775), or liver disease-related death (P = 0.457). Table 3 shows the number of deaths, crude mortality rates, and sex-and age-adjusted mortality rates per 1000 patientyears (95% CIs), and relative risks for death expressed as sex-and age-adjusted relative mortality rate ratios (95% CIs) in group B and group C as compared with the reference (group A). The crude mortality rates in groups A, B, and C were 92.7, 94.0, and 147, respectively. Sex-and age-adjusted mortality rates (95% CI) in groups A, B, and C were 71.9 (62.6 to 81.3), 80.4 (37.9 to 123), and 156 (104 to 207), respectively. The relative risks (95% CI) for all-cause, cardiovascular, infectious disease-related, and liver diseaserelated death in group C were 2.16 (1.53 to 3.07), 1.98 (1.19 to 3.28), 2.46 (1.27 to 4.76), and 30.8 (5.34 to 178), respectively. In contrast, group B did not have significantly higher risks for death, with the exception of liver diseaserelated death (RR, 13.7; 95% CI, 1.24 to 152). Table 4 shows the relative risks for each cause of death in groups B and C, as compared with the reference (group A), expressed as multivariate-adjusted mortality rate ratios. The RRs for all-cause and cardiovascular death in group C were approximately 2.0, which indicated significantly higher risks for such deaths in group C. The RR for infectious diseaserelated death in group C was also approximately 2.0, although the result was of marginal significance (P = 0.051 after

DISCUSSION
In this study, we estimated crude and sex-and age-adjusted rates for all-cause death and cause-specific death in hemodialysis patients who were negative for HCV antibodies, those who were positive for HCV antibodies, and those who were positive for both HCV antibodies and anti-HCV core antigen antibodies. We also calculated the relative risks of all-cause death and cause-specific death in patients positive for HCV antibodies only and patients positive for both HCV antibodies and anti-HCV core antigen antibodies as compared with patients who were negative for anti-HCV antibodies. These 3 groups roughly correspond to patients without HCV infection (group A), patients with past HCV infection (group B), and patients with chronic HCV infection (group C). Therefore, the results showed higher risks of all-cause, cardiovascular, infectious disease-related, and liver disease-related death among the chronic HCV subgroup, whereas past HCV infection was not associated with increased risk of any cause of death, except liver disease-related death.
Most prior studies investigated only the relative risks of all-cause and/or cause-specific death attributable to HCV infection among hemodialysis patients, without further differentiating between past and chronic HCV infection. 19,22,24,25 In a meta-analysis, Fabrizi et al found that the adjusted RR for all-cause mortality due to HCV infection was 1.34. 27  Patients in group C had a significantly higher probability of death than did patients in group A (P = 0.007, log rank test), but there was no significant difference in probability of death between groups A and B (P = 0.174). The upper right graph shows the cumulative probability of cardiovascular death. Patients in group C had a significantly higher probability of cardiovascular death than did patients in group A (P = 0.033); the probability of cardiovascular death did not significantly differ between groups A and B (P = 0.118). The lower left graph shows the cumulative probability of infectious disease-related death, which did not significantly differ among the 3 groups. The lower right graph shows the cumulative probability of liver disease-related death. Patients in group C had a significantly higher probability of liver disease-related death as compared with patients without HCV infection (P < 0.001). The probability of liver disease-related death in group B did not significantly differ from that of group A (P = 0.457).
HCV infection increased mortality among hemodialysis patients: the multivariate-adjusted RR for all-cause death attributable to chronic HCV infection was 2.0. 23 Other studies reported a multivariate-adjusted RR of death attributable to HCV infection (including past HCV infection) between 1.2 and 1.6. 27 The lower relative risks in studies that assessed HCV status using antibody-based techniques may be due to underestimation related to the inclusion of patients with past HCV infection only. In our study, the RR of all-cause death due to chronic HCV infection, as determined by quantitative estimation of HCV core antigen, was 1.83 after traditional risk factor adjustment, which is similar to the RR of 2.0 reported by Stehman-Breen et al. Taken together, both previous studies and the present study suggest that it is mainly chronic HCV infection that increases the risks of all-cause and causespecific death.
The causes of death that contribute to increased mortality among hemodialysis patients with chronic HCV infection were not fully identified in previous studies. In a metaanalysis, Fabrizi et al showed that HCV-seropositive hemodialysis patients had higher rates of liver diseaserelated death than their seronegative counterparts, but that  Crude and sex-and age-adjusted mortality rates (95% confidence intervals) are expressed as /1000 person-years. Adjusted mortalities and relative risks (expressed as relative mortality rate ratios) were estimated by Poisson regression analysis.
cardiovascular and infectious disease-related morality rates were similar. 27 The studies included in their meta-analysis all cited cardiovascular death as the most common cause of death in dialysis patients. Excess deaths attributable to HCV infection cannot be explained by an increase in the number of HCV-attributable liver disease-related deaths. Whether cardiovascular death (the most common cause of death) and infectious disease-related death (the second most common cause of death) increase mortality among hemodialysis patients is also important, as is the contribution of liver disease-related death.
We are unable to explain the increased risks of cardiovascular death and infectious disease-related death among hemodialysis patients who were positive for anti-HCV core antigen antibodies in the present study. Crosssectional analysis of baseline data provides some clues regarding possible mechanisms that might explain the association between anti-HCV core antigen positivity and increased cardiovascular and infectious disease-related mortality risk. Despite being younger, patients who were positive on the anti-HCV core antigen test had lower levels of serum lipids and albumin as compared with patients who were negative on the HCV antibody test. These findings suggest that hemodialysis patients who were positive for anti-HCV core antigen antibodies had hypocholesterolemia and hypoalbuminemia. Thus, insufficient levels of serum cholesterol and albumin might be associated with a malnutrition-inflammation syndrome activated by chronic HCV infection. Such a syndrome might lead to immune dysfunction, resulting in an increased risk of cardiovascular and infectious disease-related death. [19][20][21][22][23][24][25][26][27][28][29][30][31] Several limitations in our study should be noted. Because we enrolled only 70 patients who were positive for anti-HCV core antigen, we were not able to perform an accurate sex-stratified risk assessment of cause-specific death. Sex differences in the risk of each cause of death might exist, and the relationships should therefore be re-examined in larger cohort studies or in meta-analyses using data from patients whose chronic HCV infection status is precisely defined. Lack of high-sensitivity, quantitative HCV-RNA data from patients who were positive for HCV antibody and negative for HCV core antigen is a major limitation of this study. It is possible that hemodialysis patients who are negative for HCV core antigen nevertheless have very low levels of HCV-RNA; however, the possibility of missing such cases is very low because none of the population-based controls in our previous study were simultaneously positive for HCV-RNA and HCV antibody and negative for HCV core antigen. 21 Therefore, we believe that the results of the current study are not distorted by the lack of HCV-RNA data.
Because second-generation ELISA became available as a clinical diagnostic tool in 1992, patients who began hemodialysis treatment before 1992 might have had more exposures to infection and a higher incidence of HCV infection. It remains to be clarified whether HCV infection, and a long history of hemodialysis treatment, independently increase the risk of death. In a separate analysis, we estimated the risk of each cause of death attributable to HCV infection only in patients who started hemodialysis treatment after 1992. The results were similar to those from analyses of all subjects (data not shown).
We determined HCV infection status based on baseline information only. Changes in HCV infection status (eg, incident HCV infection during the observation period) were not considered. Previous studies showed that the incidence of HCV infection in hemodialysis patients was lower than 0.5% per year. 11,32 The risk of death attributable to HCV infection may have been underestimated, and putative underestimation of relative risks of death is not negligible.
Despite its limitations, our simple and economic method of determining HCV infection status provided sufficient results to discern a difference in mortality between patients with past versus chronic HCV infection. Furthermore, limiting the analysis to patients with chronic HCV infection enabled us to show that an increased risk of death among hemodialysis patients with HCV infection was due to an increased risk of cardiovascular and infectious disease-related deaths, as well as the increased risk of liver disease-related death. We conclude that more attention should be paid to chronic HCV infection in hemodialysis patients.