Influence of Hepatitis G Virus Infection on the Severity of Liver Disease and Response to Interferon-α in Patients with Chronic Hepatitis C

Methods

Patients

We enrolled 228 patients (143 men and 85 women [mean age ±SD, 40 ± 12 years]) who had previously participated in three controlled trials of interferon-α at our center [15-17]. Eighty-seven patients had a history of blood transfusion, 57 were intravenous drug users, and 84 had no known source of infection. All patients received interferon-α (alfa-2a [Roferon, Hoffman-LaRoche, Neuilly, France]; alfa-2b [IntronA, Schering-Plough, Levallois, France]; or alfa-n1 [Lymphoblastoid, Wellferon, Burroughs Wellcome, Issy-les-Moulineaux, France]). Three interferon-α regimens were defined according to the total dose received. Dose 1 corresponded to a dosage of 3 million U three times a week for 3 months, dose 2 corresponded to a dosage of 3 million U three times a week for 6 months, and dose 3 corresponded to a dosage of 3 million U three times a week for 12 months or 5 million U three times a week for 6 or 12 months. All patients were born in France and lived in France at the time of the study. All patients tested positive for antibody to HCV on third-generation enzyme-linked immunosorbent assay. These results were confirmed by recombinant immunoblot assay [Ortho Diagnostic Systems, Roissy, France]. Diagnosis of chronic hepatitis C was based on the following criteria: 1) persistently elevated serum alanine aminotransferase levels for more than 6 months before randomization; 2) no evidence of hepatitis B virus infection [absence of detectable hepatitis B surface antigen]; 3) exclusion of other causes of chronic liver disease [alcoholism, hepatotoxic drugs, autoimmune chronic hepatitis, hemochromatosis, Wilson disease, and α-1 antitrypsin deficiency]; and 4) liver histologic examination showing lesions characteristic of chronic hepatitis. No patient had a history of decompensated cirrhosis (ascites, bleeding esophageal varices, or hepatic encephalopathy), and all were negative for anti-human immunodeficiency virus (HIV) antibodies.

Response to interferon-α therapy was defined biochemically. Sustained response was defined as normalization of serum alanine aminotransferase levels during treatment and during the 6-month follow-up period. Response with relapse was defined as normalization of serum alanine aminotransferase levels during treatment, with elevation occurring after the end of treatment. No response was defined as elevated serum alanine aminotransferase levels at the end of the treatment.

We recorded the following patient characteristics: sex; age; source and duration of infection; pretreatment serum levels of alanine aminotransferase, γ-glutamyltransferase, and ferritin; liver histologic characteristics; pretreatment serum levels of HCV RNA and HGV RNA; and HCV genotype. In patients who became infected as a result of blood transfusion or intravenous drug use, the duration of infection was estimated to be the interval between the date of transfusion or the date of the onset of intravenous drug use and the date of initiation of treatment. Patients with no known parenteral exposure were not considered for this variable. Liver biopsy specimens were obtained from all patients within the 6 months before initiation of treatment. The histologic preparations were examined in a blinded manner by the same pathologist and were scored for fibrosis and necroinflammatory activity according to the criteria proposed by Knodell and colleagues [18]. Activity of chronic hepatitis was considered to be low when the necroinflammatory activity score was 6 or less; activity was considered to be mild or high when the necroinflammatory activity score exceeded 6.

Laboratory Studies

All virologic assays were done on aliquots of the same serum samples, which were kept frozen until they were used.

Measurement of Serum Hepatitis G Virus RNA and Hepatitis C Virus RNA

Serum HGV RNA and serum HCV RNA were quantitated to evaluate the levels of HGV and HCV viremia before treatment.

Serum HGV RNA quantitation was done with an experimental branched-DNA (bDNA) assay (Chiron Corp., Emeryville, California). In this assay, additional sensitivity was achieved by using amplimers made with iso-C and iso-G nucleotides and by using a preamplifier sequence to create an interface between target probe overhangs and the bDNA amplifier [19]. The capture and label extenders are located in the relatively conserved sequence of the 5′ untranslated region of the HGV genome according to the HGV isolates described by Linnen and colleagues [3] and cover approximately 300 bases. The molecular quantitation cutoff of the assay is 0.05 × 10⁶ genome equivalents/mL. All samples were run in duplicate.

Serum HCV RNA quantitation was done with the improved quantitative bDNA-HCV RNA assay (Quantiplex HCV RNA 2.0, Chiron Diagnostics, Eragny, France). In this assay, refined oligonucleotide probe sets that are based on sequence variation among disparate HCV isolates are incorporated [20]. This newly developed assay is more efficient for HCV RNA quantitation of HCV genotypes 2 and 3 and leads to an equivalent quantification for HCV genotypes 1 through 6. The quantitation cutoff of the assay is 0.2 × 10⁶ genome equivalents/mL. All samples were run in duplicate.

To obtain comparable quantitation on bDNA assays for HGV RNA and HCV RNA, the standards for both assays were value-assigned against RNA transcripts that were synthesized to include the entire region recognized by the respective probe sets. The transcripts were independently characterized and quantified as described elsewhere [21]. To assess the quality of the transcripts, the preparations were run on agarose gels that contained formaldehyde; the gels were dried and scanned using an Ambis 4000 Radiologic Imager (Ambis, Inc., San Diego, California). The preparations of HGV RNA and HCV RNA transcripts that were used to define the quantification in the bDNA assays were greater than 80% full length and contained less than 3% free nucleotides. The analytical methods used to quantify RNA transcripts, phosphatase analysis, hyperchromicity, and absorbance at 260 nm produce values that agree within 10%. The transcripts were subsequently tested in the respective bDNA assays to determine the signal generated per attomole of RNA transcripts.

In HCV RNA and HGV RNA assays, a genome equivalent is defined as the amount of RNA in a sample that generates the same signal as one molecule of the characterized transcript.

Detection of Serum Hepatitis G Virus RNA and Hepatitis C Virus RNA

Serum HGV RNA and serum HCV RNA were detected using a qualitative method at the end of treatment and 6 months after treatment in order to look for a correlation between the biochemical and virologic responses.

Serum HGV RNA was detected by using reverse-transcription polymerase chain reaction with primers in the 5′ end of the HGV genome according to the method of Linnen and colleagues [3]. The detection of serum HCV RNA was done by using reverse-transcription polymerase chain reaction with primers in the 5′ noncoding region of the HCV genome [22]. Detection was done at the initiation of treatment in all patients with detectable serum HGV RNA and in all patients with detectable serum HCV RNA.

Genotyping of Hepatitis C Virus

Genotyping of HCV was done in the 5′ untranslated region of the HCV genome by using reverse hybridization with the line probe assay [23] (InGeN, Rungis, France). In reverse hybridization, the biotinylated amplification products obtained are hybridized to oligonucleotides directed against the variable region of the 5′ untranslated region and are immobilized as parallel lines on membrane strips. Incubation with streptavidin labeled with alkaline phosphatase then allows detection of the hybrids. The HCV line probe assay contains 15 probe lines that allow identification of HCV types 1 to 5 and HCV subtypes.

All serum specimens were stored at 4 °C immediately after collection, were centrifuged through a paraffin plug after formation of the clot within 2 hours of sampling, and were frozen at −80°C until quantitation and genotyping were done.

Statistical Analysis

Dichotomous variables were compared by using a chi-square test or a Fisher exact test for small sample size. To study the relation between two qualitative variables when adjusting for a third variable (bivariate analysis), the Mantel-Haenszel test was used in the absence of interaction. Interaction was studied by using the Breslow-Day test. Quantitative variables were compared by using the Student t-test and the Kruksal-Wallis rank test for small groups. A Spearman test was used to determine the correlation between the different characteristics.

Results

Forty-eight of 228 patients (21%) in our study were infected with HGV, as assessed by the bDNA method. The prevalence differed with regard to the source of infection: It was higher in intravenous drug users (18 of 57 [32%]) than in patients with a history of blood transfusion (16 of 87 [18%]) or in patients with no known risk factor (14 of 84 [17%]) The difference, however, was not statistically significant (P = 0.08). The median serum HGV RNA level was 33 × 10⁶ genome equivalents/mL (range, 0.13 to 114 × 10⁶ genome equivalents/mL). The median serum HCV RNA level was 2.99 × 10⁶ genome equivalents/mL (range, 0.20 to 31 × 10⁶ genome equivalents/mL).

Patient Characteristics according to Hepatitis G Virus Status

The clinical HCV virologic and histologic features of patients with and patients without HGV infection are compared in Table 1. Infection with HGV was found more frequently in men than in women (P = 0.05). The patients with HGV infection did not statistically significantly differ from those without HGV infection with regard to age; source and duration of infection; serum levels of alanine aminotransferase, γ-glutamyltransferase, ferritin, and HCV RNA; and histologic lesions.

Distribution of HCV genotype differed significantly between patients with HGV infection and those without HGV infection (P = 0.02). In particular, HCV genotype 3a was found more often in patients with HGV infection than in those without HGV infection (44% compared with 22%). To determine whether the association between HGV infection and HCV genotype 3a was related to the source of infection, we did a bivariate analysis of the relation among HCV genotype, source of infection, and HGV infection. When adjusted for the source of infection, the association between HCV genotype 3a and HGV infection remained significant (P = 0.02); when adjusted for HCV genotype, the association between intravenous drug use and HGV infection was no longer significant (P = 0.2).

The occurrence of histologic lesions was similar in the two patient groups. Chronic active hepatitis with mild or high activity was found in 67% of patients with HGV infection and in 73% of those without HGV infection, chronic active hepatitis with low activity was found in 27% of patients with HGV infection and in 22% of those without HGV infection, and chronic active hepatitis with cirrhosis was found in 6% of patients with HGV infection and 5% of those without HGV infection. According to histologic examination, no morphologic variables were particularly associated with HGV infection.

Response of Hepatitis G Virus to Interferon-α Therapy

Of the 48 patients who had detectable serum HGV RNA at the start of treatment, 35 (73%) had no detectable serum HGV RNA at the end of the treatment and 7 of 46 (15%) had no detectable serum HGV RNA 6 months after treatment.

The biochemical response to interferon-α therapy did not differ between patients with and those without HGV infection. Among patients without HGV infection, 48% had no response, 37% had a response with relapse, and 15% had a sustained response; among patients with HGV infection, 54% had no response, 33% had response with relapse, and 13% had a sustained response (P > 0.2). The presence of HGV infection did not influence the association of HCV genotype and serum HCV RNA levels with sustained response to interferon-α therapy, as described elsewhere [24]. When adjusted for HGV infection, HCV genotypes 2a or 3a and low serum HCV RNA levels (<0.2 × 10⁶ genome equivalents/mL) remained significantly associated with a sustained response to interferon-α therapy (P < 0.001).

No correlation was found between biochemical response (defined as normalization of serum alanine aminotransferase levels) and serum HGV RNA response. At the end of treatment, serum HGV RNA was not detectable in 17 of 22 patients (77%) who had biochemical response compared with 18 of 26 patients (69%) who did not have biochemical response (P > 0.2). Six months after treatment, serum HGV RNA was not detectable in 3 of 7 (43%) patients who had biochemical response compared with 4 of 39 patients (10%) who did not have biochemical response (P = 0.06). On the contrary, a correlation was seen between biochemical response and serum HCV RNA response: At the end of treatment, serum HCV RNA was not detectable in 15 of 22 patients (68%) who had biochemical response compared with 7 of 26 patients (27%) who did not have biochemical response (P = 0.004). Six months after treatment, serum HCV RNA was not detectable in 7 of 7 patients (100%) who had biochemical response compared with 1 of 39 patients (3%) who did not have biochemical response (P < 0.001).

Serum Hepatitis G Virus RNA Levels according to Patient Characteristics

Table 2 shows serum HGV RNA levels according to the clinical, virologic, and histologic features of the study patients. Serum HGV RNA levels were significantly higher in patients 40 years of age or younger than in patients older than 40 years of age (P = 0.03). Serum HGV RNA levels did not differ according to sex, source and duration of infection, HCV genotype, or serum HCV RNA levels.

Serum HGV RNA levels were significantly lower in patients whose histologic score was high, particularly those whose histologic score was higher than 9 (P = 0.02). Serum HGV RNA levels were significantly lower in patients whose histologic necroinflammatory activity score was high, particularly in those whose histologic necroinflammatory activity score was between 7 and 9 (P = 0.03). Serum HGV RNA levels did not differ according to fibrosis score. No correlation was found between serum HGV RNA levels and serum alanine aminotransferase levels (r = −0.04).

The relation between pretreatment serum HGV RNA and HCV RNA levels and the response to interferon-α therapy are shown in the (Figure 1). Median serum HGV RNA levels did not differ among patients who had no response (23.4 × 10⁶ genome equivalents/mL [range, 0.180 to 70.0 × 10⁶ genome equivalents/mL]), those who had response with relapse (40.5 × 10⁶ genome equivalents/mL [range, 0.147 to 113 × 10⁶ genome equivalents/mL]), and those who had sustained response (20.5 × 10⁶ genome equivalents/mL [range, 0.173 to 112 × 10⁶ genome equivalents/mL]) (P > 0.2). Median serum HCV RNA levels differed significantly in patients who had no response (1.97 × 10⁶ genome equivalents/mL [range, 0.200 to 30.7 × 10⁶ genome equivalents/mL]), those who had response with relapse (2.99 × 10⁶ genome equivalents/mL [range, 0.200 to 27.4 × 10⁶ genome equivalents/mL]), and those who had sustained response (0.415 × 10⁶ genome equivalents/mL [range, 0.200 to 19.1 × 10⁶ genome equivalents/mL]) (P = 0.001).

Discussion

Our study indicates that in patients with chronic hepatitis C, the prevalence of HGV infection was 21% as assessed by the detection of serum HGV RNA by the bDNA assay. Infection with HGV was associated with male sex, a history of intravenous drug use, and HCV genotype 3a, independent of the source of infection. The median serum HGV RNA level was 33 × 10⁶ genome equivalents/mL (range, 0.13 to 114 × 10⁶ genome equivalents/mL); this value is 10-fold higher than the median value for serum HCV RNA levels (2.99 × 10⁶ genome equivalents/mL [range, 0.2 to 31 × 10⁶ genome equivalents/mL]). Infection with HGV did not influence the severity of liver disease or response to interferon-α therapy. Hepatitis G virus was sensitive to interferon-α therapy, but the biochemical response was correlated to HCV virologic response and not to HGV virologic response.

Prevalence of Hepatitis G Virus Infection

The high prevalence of HGV infection found in our study is consistent with the 20% to 24% prevalence reported in other studies [5, 9]. The prevalence of HGV infection was especially high in patients who had a history of intravenous drug use (32%); this population is therefore at a particularly high risk for HGV infection. We used serum HGV RNA to detect active HGV infection; this method probably underestimates the actual rate of HGV infection because many patients could have had HGV infection that had resolved. Indeed, the rate of HGV infection in intravenous drug users could have been expected to be higher than the rate of HCV infection (resulting from acquisition by the same route) found in this group. Studies on the detection of anti-HGV antibodies are needed to evaluate the actual rate of HGV infection in intravenous drug users, but no serologic assays are currently available.

It is noteworthy that HGV infection was associated with HCV genotype 3a independent of the source of infection. This observation suggests that HGV infection may persist more easily in patients infected with HCV genotype 3a than in patients infected with other HCV genotypes, particularly genotype 1b.

The prevalence of HGV infection was also high among patients with a history of blood transfusion (18%). Linnen and colleagues [3] reported a 1.7% prevalence of HGV infection in blood donors with normal alanine aminotransferase levels. These results emphasize that HGV is frequently transmitted by blood transfusion.

The prevalence of HGV infection was 17% among patients with no known parenteral exposure (sporadic cases). This finding is consistent with the view that sporadic cases of HGV infection may occur relatively frequently. As with HCV infection, these cases are probably related to some type of parenteral exposure. However, one cannot exclude the hypothesis that HGV might be transmitted by other routes, such as sexual contact or mother-to-infant transmission. Feucht and colleagues [9] reported a high rate of HGV transmission from mother to infant (33%), especially when the mothers were infected with HCV or HIV. The relatively high level of HGV viremia (10-fold higher than the serum HCV RNA level) found in our study suggests that infectivity might be higher for HGV than for HCV and might be responsible for a higher risk for transmission of HGV from mother to infant or through sexual contact. Indeed, the risk for mother-to-infant transmission of HCV was shown to be increased in women with high serum HCV RNA levels [25].

Severity of Liver Disease

Our results show that in patients with chronic hepatitis C, HGV infection does not influence the rate of HCV replication or the severity of liver disease. Serum alanine aminotransferase levels, serum HCV RNA levels, and the severity of liver histologic lesions in patients with HGV infection did not differ from those in patients without HGV infection. This result is consistent with the view that liver lesions are more strongly related to HCV infection than to HGV infection. Furthermore, these results support the argument that HGV infection has little or no pathogenic effect.

In our patients with HGV and HCV infection, histologic necroinflammatory activity was significantly higher in patients who had low serum HGV RNA levels than in those who had higher serum HGV RNA levels (Table 2). This finding reinforces the idea that HGV infection has little pathogenic effect on the liver. A direct mechanism of HGV infection on liver histologic findings seems unlikely; an indirect relation is more probable. In our study, lower serum HGV RNA levels were associated with older age. It is well established that the severity of histologic lesions is associated with old age [26]. Thus, the association between low serum HGV RNA levels and high histologic score might be related to patient age.

The low pathogenic effect of HGV infection was shown by Masuko and colleagues [6], who reported normal alanine aminotransferase levels in patients undergoing hemodialysis who had had persistent HGV infection for as long as 16 years. This hypothesis is reinforced by Linnen and coworkers [3], who reported a similar rate of HGV infection in 709 blood donors with normal alanine aminotransferase levels and 709 blood donors with elevated alanine aminotransferase levels (1.7% and 1.5%, respectively). Finally, our study confirms the hypothesis of a low pathogenic effect of HGV infection by the absence of correlation between serum HGV RNA response and biochemical response to interferon-α therapy. In contrast, the serum HCV RNA response was correlated with the biochemical response, confirming that the activity of liver disease was primarily related to HCV infection.

Response to Interferon-α Therapy

In agreement with the study by Tanaka and colleagues [14], our results suggest that HGV infection does not influence the response to interferon-α therapy. The rates of response to interferon-α therapy did not differ between patients with and those without HGV infection. In patients with chronic hepatitis C, HCV genotypes and serum HCV RNA levels are strongly associated with sustained response to interferon-α therapy. As we reported elsewhere [24], HCV genotype 2a or 3a and low serum HCV RNA levels are also independently associated with sustained response to interferon-α therapy. Indeed, in the present study, the presence of HGV infection did not affect the association among HCV genotype 2a or 3a, low serum HCV RNA level, and sustained response to interferon-α therapy.

To conclude, HGV infection is frequently found in patients with chronic hepatitis C. Hepatitis G virus infection does not, however, influence the course of hepatitis C or the response to interferon-α therapy. Thus, testing for serum HGV RNA has no clinical implications for prognosis or choice of therapy for patients with chronic hepatitis C.

Drs. Marcellin, Boyer, Pouteau, Castelnau, Benhamou, and Erlinger: INSERM U24 and Service d'Hepatologie, Hopital Beaujon, 100, Boulevard du Gal-Leclerc, 92118 Clichy, France.

Ms. Detmer and Drs. Collins, Kolberg, and Wilber: Chiron Corp., 4560 Horton Street, Emeryville, CA.

Dr. Degott: Service d'Anatomie et Cytologie Pathologiques, Hopital Beaujon, 100, Boulevard du Gal-Leclerc, 92118 Clichy, France.