We prospectively studied a group of 189 consecutive IFN-treated chronic hepatitis C patients (95 males, 94 females, mean age 58 ± 18 years) who entered an IFN therapeutic program at our institution from September 1996 to July 1998 (Table 1). Criteria for the diagnosis of chronic HCV infection were: 1) presence of anti-HCV antibodies and polymerase chain reaction (PCR) positivity for HCV-RNA, 2) histological confirmation of chronic hepatitis C, 3) exclusion of other causes of chronic liver diseases (alcoholism, Wilson's disease, drugs, hemochromatosis, a1 antitrypsin deficiency, autoimmune hepatitis, PBC). Criteria for inclusion in the IFN treatment program were the generally recognized ones (transaminase levels over two times the upper limit, age between 18 and 70 years, absence of pregnancy and psychiatric history or other chronic severely invalidating conditions). None of the patients had received immunosuppressive or immunostimulatory therapy before entry into the study. All the patients were negative for human immunodeficiency virus (HIV) antibodies and hepatitis B surface antigen (HBsAg). Disease activity and stage were evaluated according to Scheuer's histological score. None of the patients were receiving proton pump inhibitors or anti-H₂ antagonist drugs. Informed consent was obtained from each patient and approval for the study protocol was granted by the Ethical Committee of our institution.

All the patients received 6 MU of recombinant IFN daily for the first month, followed by 6 MU each alternate day for 5 mo and then 3 MU each alternate day for a further 6 mo. In the event of side-effects, the dosage was decreased. If side effects were severe or sustained, IFN treatment was suspended.

Serum samples were analyzed in all the patients for presence of gastric parietal cell autoantibodies and gastrin at the following time points: baseline, after 3 and 6 mo of treatment, the end of IFN treatment (i.e. at 12 mo, or at the time of suspension, if earlier), 12 mo after suspension of IFN. Multiple endoscopic gastric biopsies were always performed when positive anti-gastric parietal cell autoantibodies and/or serum levels of gastrin were found to be over 3 times the normal upper limit, presence of gastrinoma was excluded by the secretin stimulation test. In all patients, Helicobacter pylori status was serologically tested before the start of treatment. Criteria for diagnosis of autoimmune gastritis were according to Sidney classification system[18,19].

Serum samples were analyzed for routine biochemical liver function tests with a multichannel autoanalyzer. HBsAg and anti-HBs and anti-hepatitis B core (HBc) antibodies were tested using a commercial solid-phase radioimmunoassay (Abbott Laboratories, North Chicago, IL). HIV determination was done according to a standard enzyme-linked immunosorbent assay (ELISA) procedure (HIV ELISA, Abbott Laboratories, North Chicago, IL). Anti-HCV antibodies were tested using a second-generation ELISA procedure (ELISA-2, Ortho Diagnostic Test Systems, Raritan, NJ). HCV RNA was detected by nested PCR analysis using primers from the highly conserved 5' non-coding region of the HCV genome (Shindo et al, 1991). HCV genotype was determined by InnoLipa.

Anti-nuclear, anti-mitochondrial, smooth-muscle and liver-kidney microsome-1 auto-antibodies (ANA, AMA, SMA and LKM-1) were determined using indirect immunofluorescence assays on unfixed cryostat frozen sections of mouse liver, kidney and stomach. Sera were screened for anti-parietal cell autoantibodies (APCA) by immunofluorescence reactivity with paraffin-embedded sections of mouse stomach and for H⁺/K⁺-ATPase autoantibodies by ELISA. A positive result required a titer of at least 1/40. Basal serum gastrin and thyroid serum hormone (TSH) concentration were measured by radioimmunological assay (RIA), which detects sulfated and non-sulfated human heptadecapeptide (hG-17), as well as human big gastrin (hG-34). Gastrin results were expressed as pg/ml. Thyroid microsomal and thyroglobulin autoantibodies (TMA and TGA) were analyzed using hemagglutination tests (Ames, Elkhart, IN, USA). The cut-off points for both TMA and TGA were 1:100.

Serum immunoglobulin G (IgG) response to H. pylori purified antigens was measured by ELISA, the cut-off value used was an optical density ratio > 1.0. The presence or absence of H. pylori was also defined by histological examination of multiple gastric biopsy specimens from the antrum, fundus, or cardia in all the patients with positive APCA and/or elevated levels of gastrin. All biopsy specimens were fixed in Hollande's fixative and stained with H&E and Giemsa.

Intent-to-treat analysis was adopted. To analyze associations among groups the Fisher's exact test and the χ² test with Yates' correction were used. A two-tailed P value less than 0.05 was considered significant.

The entire 12-month treatment schedule was completed by 168/189 (89%) of patients. In 21 patients, the IFN dose was reduced due to the severity of side effects (severe thrombocytopenia and/or severe leukopenia with neutrophil count < 800/mm³, continuous fever unresponsive to paracetamol, or severe myalgia). At least three months of treatment were completed by all but two of the patients (one suspended due to severe depression and suicidal tendency, the other due to side effects coupled with lack of motivation). None of the patients who discontinued therapy was positive for APCA.

In 164/189 (87%) patients, no abdominal symptoms were reported. The most frequently reported symptoms were epigastric pain and/or abdominal discomfort. Presence of abdominal symptoms was not affected by positivity/negativity for H. pylori.

At the start of treatment, APCA positivity was detected in 3 (1.6%) patients, and hypergastrinemia (i.e. serum gastrin levels over three times the normal upper limit) was found in a further 9 (5%). At the end of treatment, these incidences rose to 25 (13%) and 31 (16%) patients, respectively (both P < 0.001 vs baseline values). Thus, 22 patients developed both APCA and hypergastrinemia, mostly by the third month of therapy. Twelve months after suspension of IFN, APCA and hypergastrinemia were still present in 7 (4%) and 14 (7%) patients, respectively (P = 0.002 and P = 0.01, respectively, vs end of IFN treatment; both P = not significant vs baseline). During IFN treatment, females were more prone to have increased APCA (P = 0.017) or increased serum gastrin levels (P = 0.011) than males.

Serum gastrin levels (defined as median ± Standard Error) increased during administration of IFN (from 52 ± 10.4 pg/mL to 57 ± 17.2 pg/ml, P = 0.001). Twelve months after withdrawal of therapy, serum gastrin levels (56 ± 12.6 pg/ml) were still higher than those at baseline (P = 0.001), although they were significantly lower than those at the end of treatment (P = 0.001). Serum TSH levels increased during administration of IFN (from 2.6 ± 0.13 MCU/mL to 3.2 ± 0.17 MCU/ml, P = 0.02), and subsequently remained higher than the baseline values at 12 mo from withdrawal of therapy (3.2 ± 0.3 MCU/ml, P = 0.02 vs baseline, P = not significant vs withdrawal). As can be seen from Table 2 and Table 3, the behavior of the antithyroid autoantibodies TPO was very similar to that of APCA. By contrast, IFN did not affect the non-organ-specific antibodies ANA, SMA and AMA.

At baseline, 61/189 (32.3%) patients had a positive serum IgG response to H. pylori whole-cell antigen. These included 3 of the 9 (33.3%) patients with autoimmune gastritis and 12 of the 31 (38.7%) who developed biochemical signs of autoimmune gastritis during treatment.

Multiple endoscopic gastric biopsies were performed when positive APCA and/or serum levels of gastrin were found to be over 3 times the normal upper limit. At biopsy, all the patients with either baseline APCA positivity (n = 3) or hypergastrinemia (n = 9) showed a histological picture consistent with autoimmune gastritis. Among the 22 patients who developed both APCA positivity and hypergastrinemia during IFN therapy, the histology of the fundus was consistent with autoimmune gastric atrophy in 13 (59%). Four other patients who were persistently consistent with presence of gastric atrophy maintained histological lesions at 12 mo from IFN withdrawal.

At baseline, 115/189 (61.0%) patients were found to be infected with HCV genotype 1 or 4, while the remaining 74 (39.0%) were infected with more IFN-responsive strains. Overall, 37/189 (19.5%) patients showed a long-term virological response to IFN (at 12 mo from withdrawal). In particular, long-term response was observed in 8/109 (7.3%) patients with genotype 1, 21/39 (35%) with genotype 2, 7/14 (50%) with genotype 3, and 1/6 (16%) with genotype 4, respectively. No difference in responsive rate was observed among the patients who developed hypothyroidism and/or hypergastrinemia and those who did not. No relationships were observed between HCV genotype and the development of autoimmune gastritis.