Clinical Outcome of Hepatitis C as a Function of Mode of Transmission

Transcription

1 Clinical Outcome of Hepatitis C as a Function of Mode of Transmission STUART C. GORDON, 1,2 NASSER BAYATI, 1 AND ANN L. SILVERMAN 1,2 Abbreviations: HCV, hepatitis C virus, ALT, alanine aminotransferase. From the 1 Division of Gastroenterology-Hepatology, Department of Medicine, William Beaumont Hospital, Royal Oak, Michigan; and 2 Wayne State University School of Medicine, Detroit, MI. Received November 18, 1997; accepted March 17, Address reprint requests to: Stuart C. Gordon, M.D., Division of Gastroenterology- Hepatology, William Beaumont Hospital, 3601 West 13 Mile Road, Royal Oak, MI Fax: (248) Copyright 1998 by the American Association for the Study of Liver Diseases /98/ $3.00/0 Several reports suggest that posttransfusion hepatitis C causes more aggressive histological activity than disease that is acquired via other routes. We sought to determine whether mode of transmission affects disease outcome. We studied the demographics, presenting laboratory data, and clinical course of 627 consecutively evaluated nonalcoholic patients with chronic hepatitis C. Two hundred eighty-two patients (45%) were transfusion recipients, 262 (42%) acquired the disease via other routes of percutaneous exposure, and 83 (13%) were without risks. Liver histology was available in 463 patients (215 transfusion recipients, 195 non-transfusion recipients, and 53 who were were without risks) and showed noncirrhosis in 274 (59%), cirrhosis in 173 (37%), and hepatocellular carcinoma in 16 patients (4%) who also had underlying cirrhosis. Duration of follow-up was 1 to 25 years (mean, 48 months; median, 21 years). One hundred eighteen of 173 (68%) cirrhotic patients were transfusion recipients; 40 of 173 (23%) cirrhotic patients acquired infection via other percutaneous exposure, and the remainder were without known risk factors (P F.001). Among the 215 patients with blood transfusions for whom histology was available, 118 of 215 (55%) had cirrhosis and 89 of 215 (41%) were noncirrhotic (P F.001); 8 transfused patients (4%) had hepatocellular carcinoma. In the percutaneous group, 40 of 195 (21%) of the patients were cirrhotic versus 153 of 195 (78%) who were noncirrhotic (P F.001); 2 patients (1%) had hepatocellular carcinoma. During the follow-up period, 59 of 189 (31%) of the cirrhotic patients (including those 16 individuals with hepatocellular carcinoma) developed hepatic decompensation. By univariate analysis, the risk of liver failure was related to age at viral acquisition, but by logistic regression analysis, only mode of transmission, and not age or estimated disease duration, predicted risk of liver failure. Patients with posttransfusion hepatitis C were more likely to develop decompensation than individuals who were not transfusion recipients (relative risk, 3.921; CI to 7.015). Serum albumin, prothrombin time, and platelet count at presentation were independent laboratory predictors of subsequent hepatic decompensation. The rate of hepatocellular carcinoma development among all cirrhotic patients during the follow-up period was 1.2% per year. Patients with posttransfusion hepatitis C are at greater risk of cirrhotic decompensation than those individuals with non-transfusion-acquired disease. The risk of liver failure is more closely related to the mode of transmission than to age at viral acquisition or to the duration of infection. (HEPATOL- OGY 1998;28: ) Knowledge of the clinical outcome of hepatitis C will provide important information regarding treatment and public health guidelines. The heterogeneity of the disease process and the lack of prospective longitudinal studies beyond 20 years compromise data regarding the natural history of this viral infection. Disease progression is thought to be a function of both host-related factors (age at exposure, coexistent alcoholism, and immune status) and perhaps virusrelated factors (genotype, and possibly viral load). The sequential progression from infection to cirrhosis and hepatocellular carcinoma has been described, 1 but the frequency of this sequence remains controversial. The recent discovery of this virus, coupled with the long duration of the disease, mandates that most outcome studies be retrospective in nature. Even in this setting, studies are further compromised by referral bias, assessment of coexistent or prior alcoholism, histological versus clinical outcomes, and analyses based on modes of transmission. The observation that posttransfusion hepatitis is morphologically worse than disease that is acquired sporadically was made in Jove et al. 2 reported that transfusion-related non-a, non-b hepatitis was histologically most severe in transfusion-acquired infection. Shortly after the discovery of the hepatitis C virus (HCV), we confirmed 3 that the histopathology of chronic hepatitis C was a function of mode of transmission. We showed that individuals with transfusionacquired hepatitis C had more aggressive histological inflammatory activity than the hepatitis that resulted from prior injection drug use. In a review of the natural history of community-acquired hepatitis C, Alter et al. 4 also found that histologically advanced disease was more likely in blood transfusion recipients than injection drug users. More recently, a review of 6,664 patients with chronic hepatitis C 5 showed that mode of viral acquisition was the most important predictor of outcome. This study observed that transfusion recipients had the highest rate of development of cirrhosis, and this finding was independent of disease duration. To explain these observations, we postulated 3 that the severity of the initial pathological lesion was a predictor of subsequent histological disease activity. It is logical to as- 562

2 HEPATOLOGY Vol. 28, No. 2, 1998 GORDON, BAYATI, AND SILVERMAN 563 sume, therefore, that clinical outcomes should parallel these pathological observations. To validate and extend our original findings regarding histology, we attempted to identify clinical and laboratory features that might predict disease outcome in a large cohort of patients with chronic hepatitis C. We also sought to establish the frequency and incidence of hepatic decompensation, and the rate at which hepatocellular carcinoma develops in these patients. Specifically, the aim of this study was to determine whether our previous observations concerning histology would translate into clinical outcomes: does posttransfusion hepatitis C pursue a more aggressive clinical course than disease that is non-transfusionrelated? PATIENTS AND METHODS Study Population. We reviewed the medical records of 667 consecutive patients referred to William Beaumont Hospital for further evaluation of chronic hepatitis C infection between 1989 and January This medical center is located in suburban Oakland County, Michigan, and represents a primarily middle- and uppermiddle-class socioeconomic population. We previously reported 6 that the prevalence of alcoholic liver disease in our population is low. The diagnosis of chronic hepatitis C was based on standard serological tests or documentation of viremia, compatible clinical history, liver histopathology, or a combination of these factors. Sixty-six of 667 (10%) anti-hcv-positive patients presented with persistently normal aminotransferase values 7 ; 38 of 66 were also HCV-RNA-positive by polymerase chain reaction testing and were included in the analysis, whereas 28 of 66 had no evidence of viremia and were subsequently excluded from this study. Twelve additional patients who were also excluded from analysis included 2 who were anti-human immunodeficiency virus-positive, 4 who were hepatitis B surface antigen-positive, and 6 whose daily alcohol consumption was greater than 60 g/d for more than 5 years. We assessed demographic and historical information, including transfusion history and lifetime alcohol consumption, from the initial patient interview. Thus, 627 patients with chronic HCV infection were studied (284 females, 343 males; age range, years; mean age, 47.7 years). The Human Investigation Committee of our institution approved this study. Internal Validation Cohort. Because age at viral acquisition, number of transfusions, and habitual injection drug use may affect outcome analysis, we also examined the role of mode of transmission in a well-defined, age-matched cohort of our patient population. In this separate analysis, we studied the above-outlined clinical and histological parameters of 134 patients: 81 transfusion recipients (one time, less than or equal to two units of blood), and 53 former injection drug users (less than 1 year). The mean age at acquisition was 26.8 and 25.5 years among the transfusion recipients and injection drug users, respectively (P.6). Disease Duration and Follow-up. When patients reported a known mode of transmission, we determined disease duration based on time since blood transfusion or time of icteric hepatitis. For individuals who reported prior injection drug use, we made an estimate of disease duration based on approximate first year of illicit drug use. For individuals who reported more than one date of blood transfusions, a mid-point time was selected as the time of viral acquisition unless a bout of clinical posttransfusion hepatitis was described. Transfusion recipients who also reported prior injection drug use (four individuals) were designated as posttransfusion hepatitis patients. Injection drug users who received blood transfusions after 1990 were considered to have non-transfusion-associated disease. We calculated the duration of follow-up from the patient s first clinical presentation to a physician for evaluation of chronic hepatitis, as determined by a review of medical records, until the end of the observation period (September 1, 1997). The mean duration of follow-up was 48 months (range, 1-25 years); the median duration of follow-up was 21 years (SD 9.53 years). One hundred forty-eight patients were treated with interferon alfa during the course of the follow-up period, including 52 who were cirrhotic. Treatment response was defined as the normalization of serum alanine transaminase (ALT) at the end of antiviral therapy. Clinical Variables and Laboratory Tests. Demographic features studied included age (at both presentation and presumed disease acquisition), gender, mode of transmission, and estimated duration of infection. Laboratory parameters measured included serum ALT, albumin, prothrombin time, and platelet count at presentation. Hepatic decompensation was defined as variceal bleeding, ascites, encephalopathy, hepatocellular carcinoma, or need for transplantation. Histopathology. Liver histology was available in 463 of 627 patients (74%), including 58 patients with coagulopathy secondary to cirrhosis that precluded biopsy. Histological severity was classified as noncirrhosis in 274 (59%), and cirrhosis in 173 (37%). In addition, hepatocellular carcinoma developed in 16 patients (4%), all of whom also had underlying cirrhosis. Four of these 16 patients presented with liver cancer as an initial finding, and 12 developed this tumor during the course of the follow-up period. Statistical Analysis. All data provided from the medical records were entered into the Excel Spreadsheet program, and data analysis was performed using the Systat 5.0 (Systat, Inc., Evanston, IL) and SAS (version 6.12) programs. We performed univariate analyses with Fisher s Exact test, and the exact Pearson 2 test and Student s t test to compare frequencies and means. Group means are reported as mean SD. We performed multivariate analysis of prognostic factors involved in hepatic decompensation by logistic regression analysis, computing survival curves according to the Kaplan-Meier estimation method. 8 Logistic regression was used to compute the odds ratios, and 2 tables were used to compute the relative risks, wherever appropriate. To perform survival analysis, we compared curves statistically with the log rank method. 9 Because the precise date of development of cirrhosis was not obvious, the starting date for Kaplan-Meier survival probability was the date of presumed transmission of the virus. Complete sets of variables were necessary for survival analysis, and therefore to prevent bias, we omitted from analysis those cases with missing data, e.g., unknown risk factor group. The estimation of survival analysis between risk groups was therefore based on retrospective data 10 and was calculated for the influence of significance variables by the Cox regression model. 11 The Mantel-Haenszel 2 test was used to calculate the difference in clinical outcome between multiple transfusion recipients and those individuals who were transfused on just one occasion. Data are presented as mean SD. RESULTS Patient Characteristics and Modes of Transmission. Two hundred eighty-two patients (45%) acquired hepatitis C via previous transfusion (including 30 who were transfused on more than one occasion); 262 (42%) from other percutaneous exposure (including 249 who admitted to prior injection drug use and 13 from health care exposure); 83 (13%) reported no known risk factors for viral hepatitis. Figure 1 shows the distribution of age of hepatitis C acquisition and age at presentation according to gender. The overall mean age at presentation was years for women and years for men (P.001). Prior blood transfusion was more common in women (178 of 282 [63%]) than in men (104 of 282 [37%]; P.001), while percutaneous transmission (primarily injection drug use) was more common in men (187 of 262 [71%]) than women (75 of 262 [29%]; P.001). Figure 2 shows the age at acquisition as a function of mode of transmission. Although the transfused group acquired disease at a later age, the duration of disease was similar in both groups. The mean age of viral acquisition among transfusion

3 564 GORDON, BAYATI, AND SILVERMAN HEPATOLOGY August 1998 FIG. 1. Distribution of age of hepatitis C acquisition and age at presentation according to gender. (J), female; (6), male. FIG. 3. Distribution of liver histopathology as a function of age at presentation and estimated disease duration. (J), age at presentation (P.001); (6), disease duration (P.5). recipients was years, whereas the mean age of viral acquisition among the nontransfused group was years (P.001). The presumed duration of disease ( years and years for the transfused group and the nontransfused group, respectively) was not significantly different (P.2) We found no significant difference in disease duration based on mode of transmission or histological pattern. Histopathology. Figure 3 shows the distribution of liver histopathology as a function of age at presentation and estimated duration of disease. The age at presentation was years for noncirrhotic patients, years for cirrhotic patients, and years for hepatocellular carcinoma (P.001; odds ratio, 1.05; CI 1.03 to 1.73). Histological severity was not related to disease duration. Figure 4 shows histological severity of chronic hepatitis C as a function of mode of transmission. One hundred eighteen of 173 (68%) cirrhotic patients reported prior blood transfusion, but only 40 of 173 (23%) cirrhotic patients acquired infection via other percutaneous exposure (P.001). In the transfused group, 118 of 215 (55%) of those who underwent biopsy were cirrhotic, and 89 of 215 (41%) were noncirrhotic; the remaining 8 patients (4%) had hepatocellular carcinoma. In the percutaneous group, 40 of 195 (21%) patients were cirrhotic versus 153 of 195 (78%) who were noncirrhotic; the remaining 1% had hepatocellular carcinoma (P.001; odds ratio, 0.187; CI 0.09 to 0.39). The 164 patients who did not undergo liver biopsy included 66 transfusion recipients, 65 without prior transfusion, and 33 with no risk factors. Clinical Outcome and Morbidity. During the follow-up period, 59 of 189 (31%) of the cirrhotic patients (including 16 patients with hepatocellular carcinoma) developed hepatic decompensation. Eight of these patients presented initially with signs of decompensation; 4 of these 8 had hepatocellular carcinoma at presentation. The 59 patients included 42 who died from liver-related causes and 17 patients who underwent transplantation. Twelve cirrhotic patients developed hepatocellular carcinoma during the follow-up period, yielding an incidence of hepatoma development of 1.2% per year. Ten additional patients died from non-liver disease. Table 1 shows the group of 59 patients who decompensated compared with 568 patients who have not yet developed liver failure. By univariate analysis, both the mean age at presentation and the mean age of presumed viral acquisition were significantly older in the decompensated group, whereas the presumed duration of disease was similar in both groups. Patient gender was not related to risk of decompensation FIG. 2. Distribution of age at acquisition and duration of disease as a function of mode of transmission. (J), transfused group; (6), percutaneous group. FIG. 4. Histological severity of chronic hepatitis C as a function of mode of transmission. (J), transfusion; (6), percutaneous; ( ), unknown.

4 HEPATOLOGY Vol. 28, No. 2, 1998 GORDON, BAYATI, AND SILVERMAN 565 TABLE 1. Characteristics of Compensated Versus Decompensated Patients With Chronic Hepatitis C Decompensated (n 59) Compensated (n 568) P Relative Risk Confidence Interval (95%) Gender (male/female) 27/32 316/ (M) 0.429, (F) 0.986, Mean age of acquisition (yr) * 1.014, Mean diagnosis interval (yr) * 0.975, Mean age of presentation (yr) * 1.025, Mode of transmission Transfusion 45 (76%) 237 (42%) , Percutaneous 9 (15%) 253 (44%) , Unknown 5 (9%) 78 (14%) , ALT (U/L) * 0.999, Albumin (g/dl) * 0.030, Platelet count ( 1,000) * 0.964, Prothrombin time (INR) * 1.963, *Odds ratio. (P.147). Overall, however, females (relative risk, 1.038) were more likely than males (relative risk, 0.699) to develop liver failure. The transfused group was more likely to develop hepatic decompensation than those who acquired the disease by other means. There was no correlation between either histological or clinical outcome and the primary disease (e.g., cardiovascular surgery, obstetric, accident, bleeding ulcer, etc.) that necessitated the original blood transfusion (data not shown). There was no relationship between single- or multipleepisode blood transfusions and either histological or clinical outcome. Among the 30 individuals who were transfused on more than one occasion, 46.3% had cirrhosis, whereas 43.7% of the 252 one-time transfusion recipients were cirrhotic (P.785); nor was there a significant relationship between number of transfusion episodes and hepatic decompensation: 14.98% and 21.45% of single-time and multiple-occasion transfusion recipients, respectively, developed liver failure (P.4). Although, by univariate analysis, the age at viral acquisition predicted risk of developing hepatic decompensation (P.0004; odds ratio 1.031), this risk was more strongly related to the mode of acquisition via transfusions (P.001; relative risk 3.932). By logistic regression (multivariate) analysis, only mode of transmission predicted the risk of subsequent liver failure. The serum albumin, platelet count, and prothrombin times at initial presentation were independent predictors of the risk of subsequent hepatic decompensation, whereas the initial serum ALT value was not related to clinical outcome. Therapy. There was no difference in clinical outcome or frequency of hepatocellular carcinoma because of the use of interferon alfa in this patient population. The 148 patients treated with antiviral therapy during the follow-up period included 52 cirrhotics; 13 of these 52 experienced a biochemical response to treatment with only three sustained responders. Hepatic decompensation occurred only among cirrhotic patients, and we found no difference in clinical outcome between patients who did or did not receive treatment (P.285). Development of Cirrhosis and Event-Free Survival. Twenty-five years after the acquisition of hepatitis C, the Kaplan-Meier 8,10 cumulative risk of cirrhosis was 52.5% among transfusion recipients and was 35% in the percutaneous, primarily injection drug use group (log rank, P.001). Such determination could not be established for those patients who lacked risk factors, because the time of viral acquisition could not be ascertained. Kaplan-Meier probability for liver decompensation in all hepatitis C patients was 11.6% and 31% after 25 and 40 years of disease, respectively. This probability in the cirrhotic population was 21.9% and 46.6% after 25 and 40 years of disease, respectively. Figure 5 shows the Kaplan- Meier event-free survival curve for the 544 patients for whom mode of transmission was known. The cumulative risk of liver decompensation after 25 years was 17% in the transfused group versus 6% in the percutaneous, nontransfused group (log rank, P.0001). This risk was not related to gender (log rank, P.2), prior antiviral therapy (log rank, P.4), or disease duration (P.7). Among cirrhotic patients, the risk of developing hepatoma after 5 years of follow-up was 8.4% (1.2% annually). Liver-specific mortality among all cirrhotic patients was 15.5% and 35% after 25 and 40 years, with a mean survival of 42.6 years after acquisition of infection. Internal Validation Cohort. In the subset of patients (brief drug use, less than or equal to two-unit transfusions) who were age-matched for disease acquisition, the disease duration was 19.6 years in the transfused group and 17.3 years in injection drug users (P.1). Transfusion recipients tended to be female (69%), whereas former injection drug users were FIG. 5. Cumulative probability of liver decompensation as it relates to mode of transmission. Twenty-five years after acquisition of hepatitis C, the risk of hepatic decompensation in the transfused and parenteral groups is 17% and 6%, respectively. Dashed line, percutaneous; solid line, blood transfusion.

5 566 GORDON, BAYATI, AND SILVERMAN HEPATOLOGY August 1998 more often male (64%) (P.001). Liver histology at presentation showed cirrhosis in 46 of 134 (34%), and noncirrhosis in 88 of 134 (66%). Despite the similar age at viral acquisition and disease duration, cirrhosis was more frequent among transfusion recipients (33 of 91 [41%]) than in the injection drug use group (12 of 53 [23%]; P.03). The relative risk for the development of cirrhosis in the transfused group was 1.85 (CI ). DISCUSSION The attempts to define the natural history of hepatitis C have resulted in two contradictory viewpoints. Tong et al. 1 described the experience of a referral center in the United States and studied a group of nonalcoholic adults (mean age, 57 years) who had posttransfusion hepatitis C. This report observed that 51% of the patients had cirrhosis at presentation; evolution to hepatic decompensation was frequent (15.3% mortality during the follow-up period). This relatively high rate of cirrhosis among transfusion recipients referred for evaluation of hepatitis C closely parallels the results of the present study. Furthermore, our liver-related mortality rate (42 of 189 [22.2%]) among cirrhotic patients during the follow-up period is similar to the results observed by Tong et al. 1 The fundamental bias created by such nonconcurrent historical analyses, which lack control groups, is the notion that index cases may not represent the cohort. Less-severe cases may have not come to medical attention, and referral to tertiary medical centers skews outcome data. Prospective studies primarily involving patients infected via contaminated blood or blood products describe a generally benign clinical course of patients followed for up to 15 to 20 years, with a modest increase in liver-related morbidity. 12,13 The inclusion criteria of such studies likewise confound interpretation of the natural history data, because several variables influence the outcome of chronic hepatitis C. Prospective posttransfusion studies that determine liverrelated mortality rates via death-certificate information 12 may underestimate the prevalence of liver-related morbidity. The exclusion of decompensated cirrhotic patients, 14 or of nontransfusion-related disease, 1 may skew interpretation of retrospective data. Other studies 15,16 examined the histological progression of liver fibrosis as it relates to various host factors, but did not report actual clinical outcomes. The definition and assessment of cumulative alcohol consumption is exceedingly difficult and may relate to cultural factors in different countries. Although long-term prospective studies are preferred, such data will require decades to evolve. Furthermore, this information when it does emerge will be limited to transfusion-related hepatitis C. No prospective analyses of injection drug use-acquired hepatitis C will be forthcoming because of the unstable nature of this patient population and because precise times of disease onset cannot be determined with certainty. Any attempt to ascertain differences in clinical outcomes based on mode of transmission, therefore, must come from retrospective analyses and estimates of disease duration. The present report describes 627 consecutively evaluated patients seen at one medical center, all of whom were referred for evaluation of hepatitis C. Because our institution is not a tertiary or a liver transplant center, and is located in a nonurban setting, we believe that our particular cohort reflects a representative cross-section of patients with chronic hepatitis C as seen in an American community. Ascertainment bias is inherent in any retrospective cohort analysis, and thus the clinical outcome of patients not yet diagnosed with chronic viral hepatitis cannot be assessed. The present study group represents virtually all patients referred for evaluation of a positive anti-hcv test over a 7-year period. Nevertheless, the referral bias of older individuals, who are more apt to seek medical attention, probably created a spuriously high percentage of posttransfusion hepatitis patients in this study. That 45% of the patients in this series were transfusion recipients does not imply a similar prevalence of posttransfusion hepatitis C in this geographic region. In addition, the 249 former injection drug users represent the largest description of clinical outcome of this subset of chronic hepatitis C patients. As we previously reported, 3,6 our group of mostly educated, suburban, former drug users differed from other reported injection drug using populations in that their disease was not influenced by the immunological and socioeconomic factors often seen among urban illicit drug users. The present study essentially confirms and validates previous histological observations regarding the severe nature of transfusion-acquired hepatitis C 2-5 and extends these morphological findings to their clinical correlates. An additional bias created by examining clinical outcome of hepatitis C as a function of mode of transmission is the confounding variable of age. Injection drug use is essentially limited to young adult men, whereas blood transfusion is generally the risk factor among older individuals. Recent reports concur that age at acquisition appears to predict progression of this disease, 14,15,17 and these findings all validate much earlier studies that noted poor outcomes among the elderly who acquired viral hepatitis in general. 18 Because of these two disparate populations afflicted with hepatitis C, it is therefore crucial to analyze both factors age at infection and mode of transmission as individual predictors of outcome. By multivariate analysis, we found that mode of acquisition was the most important demographic predictor of subsequent hepatic decompensation. In the report by Fattovich et al., 14 both transfusion recipients as well as intravenous drug users fared better than those who acquired disease via other routes, but only 9 of 384 patients in that study were designated as injection drug users. The study concluded that mode of acquisition of infection in fact does influence prognosis and suggested further investigation. Similar to recent reports, 5,17,19 we also found no consistent relationship between estimated disease duration and features of morbidity. This remarkable observation suggests that clinical outcomes may not be significantly dependent on the factor of time. The finding also lends credence to the hypothesis that the severity of the initial pathological lesion may be an important predictor of clinical outcome. 3-5 Despite the lack of relationship between disease duration and risk of hepatic decompensation, the present study found a clear relationship between distribution of liver histopathology and age at presentation (Fig. 3), with hepatocellular carcinoma patients presenting at a mean age of 66.3 years. These findings contrast with those of Tong et al., 1 who observed that the development of hepatocellular carcinoma was more strongly related to the actual number of years following transfusion than to age per se. The rate at which hepatocellular carcinoma develops among cirrhotic patients remains controversial; studies from the West do not replicate

6 HEPATOLOGY Vol. 28, No. 2, 1998 GORDON, BAYATI, AND SILVERMAN 567 the strikingly higher frequency of hepatocellular carcinoma among Japanese patients with hepatitis C. 20,21 We found that 8.4% of our cirrhotic patients developed hepatocellular carcinoma, with a yearly incidence of 1.2%, and this figure closely parallels the 1.4% yearly incidence found by Fattovich et al. 14 Logistic regression analysis (Table 1) found that the initial serum albumin, platelet count, and prothrombin time at presentation were independent predictors of the risk of subsequent hepatic decompensation. These are expected findings, observed by others, 14,16 and reflect the notion that these laboratory variables are surrogate markers of underlying cirrhosis. Initial serum ALT levels were of no prognostic importance in this study. Although we did not specifically study the role of interferon alfa therapy in this study, we found that treated patients did not have any difference in risk of hepatic decompensation or hepatocellular carcinoma, either by univariate or multivariate analysis. A finding observed by others, 14 this observation does not establish whether interferon alfa therapy affects survival in cirrhotic patients with chronic hepatitis C; such determination will require randomized and prospective long-term studies. Among transfusion recipients, 54% presented to medical attention with established cirrhosis, whereas only 21% of former drug users had cirrhosis at presentation (P.001). The concept that mode of acquisition affects outcome implicitly suggests that the infecting viral load, or inoculum size, is an influential factor. We previously showed 22 that there was no relationship between mode of transmission and HCV-RNA levels, as subsequently measured at the time of clinical presentation. Others, however, have demonstrated higher viral levels among transfusion recipients. 23 Nevertheless, it is doubtful whether the quantity of infecting viral particles bears any relationship to subsequent serum viral levels when measured years later. Route of transmission might represent a surrogate marker of viral genotype; we did not determine genotypes in this study. Serfaty et al. 24 reported that genotype 1b was predominant in patients infected by transfusion, whereas genotype 3a was common among injection drug users. This study concluded, however, that genotype 1b does not influence the development of cirrhosis among chronic hepatitis C patients. The report by Tong et al. 1 concluded that chronic posttransfusion hepatitis C was a progressive disease that often led to serious morbidity, and the conclusion of the Fattovich et al. 14 study was that life expectancy is relatively long among patients with cirrhotic HCV. We found, in summary, that this viral infection may pursue two distinct clinical courses depending on its mode of transmission. To our knowledge, this concept is novel, i.e., the source of viral infection has not been implicated as a predictor of outcome in other diseases, such as acquired immune deficiency syndrome, that can be transmitted via transfusions. Also inferred by our findings is the possibility that chronic hepatitis C will in general become a less-severe illness in the future because of the elimination of the risk of posttransfusion disease. Knowledge of which groups are at highest risk for liver disease will help to identify those individuals most likely to benefit from antiviral therapy. Acknowledgment: The authors thank Matthew S. Schwartz for excellent assistance in data collection. REFERENCES 1. Tong MJ, El-Farra NS, Reikes AR, Co RL. Clinical outcomes after transfusion-associated hepatitis C. N Engl J Med 1995;332: Jove J, Sanchez-Taplas M, Bruguera M, Mas A, Costa J, Barrera JM, Rodes J. Posttransfusional versus sporadic non-a, non-b chronic hepatitis: a clinicopathological and evolutive study. Liver 1988;8: Gordon SC, Elloway RS, Long JC, Dmuchowski CF. The pathology of hepatitis C as a function of mode of transmission: blood transfusion vs. intravenous drug use. HEPATOLOGY 1993;18: Alter MJ, Margolis HS, Krawczynski K, Judson FN, Mares A, Alexander WJ, Hu PYH, et al. The natural history of community-acquired hepatitis C in the United States. N Engl J Med 1992;327: Roudot-Thoraval F, Bastie A, Pawlotski JM, Dhumeaux D, et al. Epidemiological factors affecting the severity of hepatitis C virus-related liver disease: a French survey of 6,664 patients. HEPATOLOGY 1997;26: Kodali V, Gordon SC, McCray DG, Silverman AL. Cryptogenic liver disease in the United States: further evidence for non-a, non-b, non-c hepatitis. Am J Gastroenterol 1994;89: Silverman AL, Piquette DL, Filipiak CL, Neill JS, Bayati N, Gordon SC. Alfa-interferon treatment of HCV RNA positive patients with normal or near-normal ALT levels. Am J Gastroenterol 1997;92: Kaplan EL, Meier P. Nonparametric estimation from incomplete observation. J Am Stat Assoc 1958;53: Peto R, Pike MC. Conservatism of the approximation (O-E)/E in the logrank test for survival data on tumor incidences data. Biometrics 1973;29: Allison PD. Survival analysis using the SAS system: a practical guide. Cary, NC: SAS Institute Inc., Cox DR. Regression models and life tables (with discussion). J R Stat Soc B 1972;34: Seeff LB, Buskell-Bales Z, Wright EC, Durako SJ, Alter HJ, Iber FL, et al. Long-term mortality after transfusion-associated non-a, non-b hepatitis. N Engl J Med 1992;327: Mattson L, Sonnerborg A, Weiland O. Outcome of acute symptomatic non-a, non-b hepatitis: a 13-year follow-up study of hepatitis C virus markers. Liver 1993;13: Fattovich GF, Giustina G, Degos F, Tremolada F, Diodati G, Almasio P. Morbidity and mortality in compensated cirrhosis type C: a retrospective follow-up study of 384 patients. Gastroenterology 1997;112: Poynard T, Bedossa P, Opolon P. Natural history of liver fibrosis progression in patients with chronic hepatitis C. Lancet 1997;349: Yano M, Kumada H, Kage M, Ikeda K, Shimamatsu K, Inoue O, Hashimoto E, et al. The long-term pathological evolution of hepatitis C. HEPATOLOGY 1996;23: Merrican I, Sherlock S, McIntyre N, Dusheiko GM. Clinical, biochemical and histological features in 102 patients with chronic hepatitis C virus infection. Q J Med 1991;86: Fenster LF. Viral hepatitis in the elderly: an analysis of 23 patients over 65 years of age. Gastroenterology 1965;49: Lee DH, Jamal H, Regenstein FG, Perillo RP. Morbidity of chronic hepatitis C as seen in a tertiary care medical center. Dig Dis Sci 1997;42: Nishiguchi S, Kuroki T, Nakatani S, Morimoto H, Takeda T, Nakajima S, Shiomi S, et al. Randomized trial of effects of interferon alpha on incidence of hepatocellular carcinoma in chronic active hepatitis C with cirrhosis. Lancet 1995;346: Okuda K, Fujimoto I, Hanai A, Urano Y. Changing incidence of hepatocellular carcinoma in Japan. Cancer Res 1987;47: Gordon SC, Kodali VP, Silverman AL, Dmuchowski C, Urdea MS, Chan C, Wilber JC. Levels of hepatitis C virus RNA and liver histology in chronic type C hepatitis. Am J Gastroenterol 1994;89: Lau JYN, Davis GL, Kniffen J, Qian K-P, Urdea MS, Chan CS, Mizokami M, et al. Significance of serum hepatitis C virus RNA levels in chronic hepatitis C. Lancet 1993;341: Serfaty L, Chazouilleres O, Poujol-Robert A, Mourand-Joubert L, Dubois C, Chreitien Y, Poupon RE, et al. Risk factors for cirrhosis in patients with chronic hepatitis C virus infection: results of a case-control study. HEPATOLOGY 1997;26: