We sought to quantify the association of transfusion-associated HCV infection and history of alcohol abuse with development of cirrhosis among patients followed from the time of acute HCV infection.

Methods

Top Methods Results Discussion Author & Article Info References

The source of this investigation, described in detail elsewhere (13), is a long-term follow-up study of acute non-A, non-B hepatitis in patients included in earlier prospective studies of transfusion-related HCV infection. All patients provided informed consent, and the study was approved by the relevant institutional review boards. Briefly, non-A, non-B hepatitis was identified by otherwise unexplained elevations in serum alanine aminotransferase (ALT) levels developing 2 to 24 weeks after transfusion, in the absence of serologic evidence of hepatitis A or B (13). The ALT level had to be elevated on two consecutive measurements, and at least one of the values had to exceed twice the upper limit of normal.

Stored sera from case-patients and controls from three of the prospective studies—the second Veterans Administration Cooperative study (14), the Transfusion-Transmitted Viruses study (15), and the National Institutes of Health Blood Bank Study (16)—were tested for anti-HCV by enzyme immunoassay (HCV EIA 2.0, Abbott Laboratories, North Chicago, Illinois). Transfusion-associated HCV infection was diagnosed if anti-HCV appeared and persisted, was temporally related to elevated ALT levels, and was confirmed by using the supplementary recombinant immunoblot assay (RIBA version 3.0, Ortho Diagnostics, Raritan, New Jersey). Because repository samples were not optimally stored for minimizing nucleic acid loss, HCV RNA testing was not performed.

Controls consisted of transfused patients from the original three prospective studies who had not developed hepatitis (13). Controls were matched to case-patients by initial treatment center, sex, ethnicity, use of hepatitis immune globulin, presence or absence of a history of alcoholism, age, number of units of blood transfused, and date of transfusion.

We excluded patients with anti-HCV–positive samples that predated the index transfusion and those with anti-HCV reactivity but no confirmatory test. Anti-HCV–positive/RIBA-negative and anti-HCV–negative patients were classified as having transfusion-associated non-A, non-B, non-C hepatitis. Controls with anti-HCV in repository samples were excluded.

We used a multifaceted approach to gather information on each patient's history of liver disease from participation in the prospective studies through initiation of the follow-up study (13). Vital status was ascertained through searches of death registries, and copies of death certificates were obtained. Patients or their designated proxies (if the patient was deceased or incompetent) were invited to participate in a personal interview that included questions about previous hospitalizations. Authorization for release of medical records was obtained, and information on all hospitalizations at each facility identified was requested. Medical records, death certificates, and biopsy and autopsy reports were reviewed by trained abstractors for the diagnosis of cirrhosis, which was based on the occurrence of features of portal hypertension and overt clinical manifestations of cirrhosis. Evaluation of a sample of records indicated a high level of agreement between abstractors and one of the investigators.

Potential risk factors were derived primarily from personal interview. A composite variable aimed at identifying a history of heavy alcohol abuse was designed to minimize the possibility of misclassifying "unexposed" patients. The composite variable was based on one or more of the following criteria: loss of friends, family, or a job because of drinking; admitting to ever having a problem with alcoholism; evidence of heavy drinking abstracted from medical records; or quantification of usual intake of more than 80 g of alcohol per day during the years when the patient drank.

Univariate methods (the Fisher exact, exact Fisher–Freeman–Halton [17], and Wilcoxon rank-sum tests) were used to compare the distribution of characteristics between patients included in and those excluded from the analysis and to examine associations between potential risk factors and development of cirrhosis. The strength of the association between risk factors and cirrhosis was assessed by using the odds ratio obtained from multiple logistic regression. The likelihood of developing cirrhosis associated with a combination of risk factors was estimated from the logistic model (18).

Results

Top Methods Results Discussion Author & Article Info References

Of 1030 patients who were followed up from the original prospective studies that had collected repository samples, 836 (81.2%) were included in this analysis. We excluded 108 patients whose repository samples were exhausted, 79 with ALT levels or HCV assays that did not permit reliable classification of hepatitis status, 1 with cirrhosis that predated the index transfusion, and 6 with data inconsistencies that could not be resolved.

As shown in Table 1, included and excluded patients differed only in median number of blood units originally transfused. The number of units received could affect the probability of developing acute transfusion-associated hepatitis or its severity but would probably not influence development of cirrhosis decades later.

Preliminary analyses indicated that development of cirrhosis was unrelated to tattooing, ear piercing, occupational exposures, extended travel to areas in which hepatitis is endemic, and use of injection or other illegal drugs. Information on prescription drug use was unavailable for analysis. Development of cirrhosis differed (P < 0.05) according to transfusion-associated hepatitis status, study cohort, race/ethnicity, units of blood originally transfused, history of heavy alcohol abuse, and vital status at follow-up (Table 2). The absolute risk for developing cirrhosis was higher among patients with transfusion-associated HCV infection (17.0% [35 of 206 patients]) (P < 0.001) than among those with transfusion-associated non-A, non-B, non-C hepatitis infection (3.2% [3 of 95 patients]) or controls (2.8% [15 of 535 patients]) but did not differ between the latter two groups (P > 0.2). The median time from index transfusion to initiation of follow-up (15.6 years overall) did not differ by cirrhosis status, transfusion-associated hepatitis status, or race/ethnicity (P 0.10).

Logistic regression modeling indicated that patients with transfusion-associated HCV infection were 7.8 times (95% CI, 4.0 to 15.1 times) more likely to develop cirrhosis than controls, after adjustment for history of heavy alcohol abuse and study cohort. Patients with transfusion-associated non-A, non-B, non-C hepatitis were at increased risk for developing cirrhosis, but this finding was not significant (odds ratio, 1.4 [95% CI, 0.4 to 5.0]). Patients with transfusion-associated HCV were 5.6 times (CI, 1.6 to 19.3 times) more likely to develop cirrhosis than those with non-A, non-B, non-C hepatitis. A history of heavy alcohol abuse was associated with an increased risk for developing cirrhosis (odds ratio, 4.0 [CI, 2.1 to 7.7]).

The Hosmer–Lemeshow goodness-of-fit test indicated that the model fit the data well (P > 0.2). From this model, it was estimated that patients with both transfusion-associated HCV infection and a history of heavy alcohol abuse were 31.1 times (CI, 11.4 to 84.5 times) more likely to develop cirrhosis than controls without a history of alcohol abuse. Patients with transfusion-associated non-A, non-B, non-C hepatitis and a history of heavy alcohol abuse were also more likely than controls without such a history to develop cirrhosis (odds ratio, 5.5 [CI, 1.3 to 24.3]).

Although cirrhosis was less common among African-Americans (2.2%) than among persons of other ethnicities (7.2%), addition of race/ethnicity, follow-up duration, units of blood originally transfused, vital status, and receipt of additional transfusions to the model changed the risks associated with hepatitis status and history of heavy alcohol abuse only slightly. None of these factors made a significant independent contribution to the model (data not shown).

Discussion

Top Methods Results Discussion Author & Article Info References

Patients who contracted transfusion-associated HCV infection were at increased risk for developing cirrhosis (odds ratio, 7.8), and this risk increased substantially if the patient also had a history of heavy alcohol abuse (odds ratio, 31.1). Although numerous reports have identified a strong role of alcohol in promoting progression of liver disease among persons with chronic HCV infection (4-12), our findings provide a quantitative measure to assess the strength of this association.

Some patients who were originally classified as having acute non-A, non-B hepatitis tested negative for HCV infection. These patients developed cirrhosis more frequently than controls, but not significantly so. This finding suggests that the elevated ALT levels observed during the original prospective studies were caused by intrinsic liver disease. It remains to be determined, however, whether these patients had been infected with a virus other than hepatitis A, B, or C virus or whether the elevated enzyme levels resulted from other causes (for example, nonviral infection or exposures to toxins).

An advantage of our study is that follow-up could be done among patients from earlier prospective studies that used a similar protocol to diagnose acute hepatitis virus infection and that periodically monitored ALT values for an extended period. In addition, serum repositories were maintained by three of the original studies, which permitted subsequent serologic and molecular testing for HCV infection.

Death certificates, medical records, and autopsy and biopsy reports are subject to possible bias due to omission of information on liver disease; however, omission of such data was unlikely to have differed between study groups. Although patients with non-A, non-B hepatitis may have had close medical follow-up, the prevalence of cirrhosis did not differ between controls and patients with transfusion-associated non-A, non-B, non-C hepatitis, suggesting that this possibility did not greatly influence the results.

Patients are unlikely to have overestimated or fabricated a history of alcohol abuse, but they may have underestimated such a history because of social stigma. Our inability to establish the exact timing and duration of alcohol abuse created additional uncertainty about the accuracy of this variable. We asked patients about the usual quantities of beer, wine, and liquor consumed per week during the time when they drank, but we could not determine whether the alcohol was consumed during a brief period or over an extended period.

In general, the potential limitations of our study may have led us to underestimate the risk for developing cirrhosis associated with transfusion-related HCV infection and a history of heavy alcohol abuse. These results stress the need to counsel patients with HCV about their drinking habits. Finally, although it was not significant in multivariate analysis, the lower frequency of cirrhosis observed in African-Americans compared with persons of other ethnicity warrants further examination, since this phenomenon has been observed previously (19).