Autoimmune Hepatitis: Effect of Symptoms and Cirrhosis on Natural History and Outcome Jordan J. Feld,1 Huong Dinh,1 Tamara Arenovich,1 Victoria A. Marcus,2 Ian R. Wanless,1 and E. Jenny Heathcote1 Although the natural history of autoimmune hepatitis (AIH) has been characterized, little is known about patients who present asymptomatically. Consequently, whether they require immunosuppressive therapy with its associated complications is unclear. To compare the natural history of asymptomatic AIH with symptomatic AIH, a large cohort of patients from a single center was examined. All patients with a clinical diagnosis of AIH were reassessed using the revised criteria of the International Autoimmune Hepatitis Group. Liver histology, response to therapy, and survival were assessed. Patients asymptomatic at presentation (n ⴝ 31) had lower serum aminotransferase, bilirubin, and immunoglobulin G (IgG) values at baseline. Half of the asymptomatic patients received no therapy, and their survival was no different from that of the total cohort. Ten-year survival was 80.0% (62.5%-97.5%) in the asymptomatic group and 83.8% (75.1%-92.6%) in the symptomatic patients (P ⴝ NS). Survival to liver-related endpoints at 10 years was similar in both groups: 89.5% (75.7%100%) asymptomatic and 83.8% (75.1%-92.6%) symptomatic patients (P ⴝ NS). Patients with cirrhosis at baseline had poorer 10-year survival (61.9% [CI 44.9%-78.9%]) than those without cirrhosis at presentation (94.0% [CI 87.4%-100%]) (P ⴝ .003) regardless of whether they presented with symptoms or whether they received immunosuppressive therapy. In conclusion, patients with AIH who are asymptomatic at presentation have a good prognosis and may not require immunosuppressive therapy. Cirrhosis on initial liver biopsy portends a poor prognosis in all patients with AIH. (HEPATOLOGY 2005;42:53-62.)

A

utoimmune hepatitis (AIH) is a relapsing inflammatory disease of the liver associated with hypergammaglobulinemia, serum autoantibodies, and hepatitis with lymphoplasmacytic infiltration on liver biopsy.1 Although the natural history, prognosis, and treatment of symptomatic AIH have been well characterized for several decades, little is known about the asymptomatic variant of this disease, which is diagnosed with increasing frequency because of the use of routine liver biochemical testing.

Abbreviations: AIH, autoimmune hepatitis; ANA, anti-nuclear antibodies; SMA, smooth muscle antibodies; LKM, liver-kidney-microsomal antibodies; HCV, hepatitis C virus; Ig, immunoglobulin; AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase; AMA, antimitochondrial antibody; ULN, upper limit of normal. From the 1Departments of Medicine and Pathology, University Health Network, University of Toronto, Toronto, Canada; and the 2Department of Pathology, Montreal General Hospital, Montreal, Canada. Received December 2, 2004; accepted April 5, 2005. Address reprint requests to: Jordan J. Feld, 10 Center Drive, Room 9B16, MSC 1800 Liver Disease Branch, NIDDK, NIH, Bethesda, MD 20892. E-mail: [email protected]; fax: 301-402-0491. Copyright © 2005 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/hep.20732 Potential conflict of interest: Nothing to report.

A broad spectrum of clinical manifestations ranging from mild nonspecific symptoms to fulminant hepatic failure has been described in AIH. Although many patients present for medical attention at the onset of symptoms, others are found to have the disease through investigation of incidental liver test abnormalities. Whether all asymptomatic AIH progresses to symptomatic disease is unclear, and the optimal management of asymptomatic patients with AIH is currently unknown. The value of immunosuppressive therapy with glucocorticoids, alone or in combination with azathioprine, was well documented in numerous studies of AIH in the early 1970s.2-7 Although steroids have greatly improved the survival of patients with AIH, whether they prevent fibrosis and ultimate progression to cirrhosis is unclear. In addition, corticosteroid therapy is associated with numerous significant toxicities.1 Consequently, whether asymptomatic patients with AIH, patients unlikely to have been recruited to the initial treatment trials, benefit from treatment with longterm immunosuppressive therapy is unknown. To better understand the clinical features, natural history, and prognosis of asymptomatic AIH, a large cohort of patients from a single site given a clinical-pathological diagnosis of AIH was reviewed. 53

54

FELD ET AL.

Patients and Methods Study Population. The charts of all patients given a clinical diagnosis of AIH at the Toronto Western Hospital Liver Clinic between the years 1970 and 2002 were retrospectively reviewed. The study was approved by the University Health Network ethics review board. Data were collected and a score was calculated for each patient based on the International Autoimmune Hepatitis Group revised scoring system.8 Patients with a pretreatment score of less than 10 or a post-treatment score of less than 12 were excluded from further analysis. All patients without a liver biopsy within 6 months of diagnosis were excluded. Clinical notes were carefully reviewed to document the presence or absence of symptoms at the time of diagnosis as well as at subsequent clinic visits. To be considered asymptomatic, patients needed to be free of even nonspecific symptoms such as fatigue, arthralgias, or abdominal pain. Any prescribed treatment was recorded, and all toxicities as well as responses were noted. Treatment. Immunosuppressive therapy with glucocorticoids or azathioprine was recommended for all symptomatic patients. Untreated asymptomatic patients were not offered therapy; however, if treatment had been initiated by the referring physician, such treatment was continued. Asymptomatic patients who developed symptoms during follow-up were started on immunosuppressive therapy. If patients remained in remission for 2 years with no relapse on tapered doses, immunosuppressive therapy was discontinued. If relapse occurred after treatment withdrawal, therapy was restarted. Laboratory Methods. All patients were tested for the non– organ-specific antibodies known to be associated with AIH. Anti-nuclear antibody (ANA) and smooth muscle antibody (SMA) titers were determined by immunofluorescence, and a titer of 1:40 or greater was considered positive. In the appropriate clinical setting, patients were tested for liver-kidney-microsomal antibody (LKM) using immunofluorescence. Antimitochondrial antibody testing was done by immunofluorescence until 1998, and thereafter by M2 ELISA (Varelisa, Pharmacia & Upjohn Diagnostics, Uppsala, Sweden). Baseline and follow-up blood work included serum aminotransferases, alkaline phosphatase, parameters of liver function, serum immunoglobulins, as well as standard hematological indices. Patients were screened for evidence of viral hepatitis. Antibodies to hepatitis C virus (HCV) were determined using second- or third-generation ELISA (Axsym System, Abbott Diagnostics, Chicago, IL). Patients with antiHCV antibodies were subsequently screened for HCV RNA using nested polymerase chain reaction (PCR; Am-

HEPATOLOGY, July 2005

plicor HCV Test, Roche Diagnostics, Mannheim, Germany). Hepatitis B surface antigen was identified using radioimmunoassay or ELISA (Axsym System, Abbott Diagnostics). When deemed appropriate, hepatitis A virus (HAV) immunoglobulin M (IgM) was performed. All liver histology specimens were reviewed by 2 pathologists, and a score was given based on the AIH revised scoring system. The presence or absence of cirrhosis was documented using standard histological criteria. No specific index has been designed to assess severity of inflammation in AIH. The pathologists at our institution routinely use the lobular inflammation component of the hepatic activity index for this purpose, and consequently, these values are reported. Statistical Analyses. Endpoints were defined as liver transplantation, liver-related death, and all death. The time to endpoint was calculated using the date of first clinic visit as the onset of disease. Cox regression survival curves were developed for each cohort examined, and from these, 10-year survival data were compiled. Patients were considered to have achieved a “Complete Response” to treatment if either there was symptomatic improvement associated with a normalization of serum aspartate and alanine aminotransferases (AST/ALT), bilirubin, and serum IgG values within 1 year of initiating treatment, which was sustained for a minimum of 6 months on maintenance therapy or if they had an improvement in symptoms together with at least 50% improvement of all liver test results during the first month of treatment, with AST/ALT levels continuing to fall to less than twice the upper normal limit within 6 months of starting therapy. Patients with improvement in liver biochemistry or symptoms, not meeting the criteria for a “Complete Response,” were deemed to have a “Partial Response” to therapy. Relapse was defined using the definitions outlined in the International Autoimmune Hepatitis revised scoring system.8 Groups were compared using 2-factor ANOVA for numerical data and chi-square analyses for categorical data. To examine which factors affect survival in AIH, Cox-Regression with time-dependent covariates was used. Variables identified by univariate modeling as affecting survival were used to create a multivariate CoxRegression model. The factors evaluated were baseline laboratory data (AST, ALT, alkaline phosphatase [ALP], bilirubin, platelet count, IgG), sex, age at presentation, pre-/post-AIH scoring system and AIH biopsy scores, presence or absence of symptoms at presentation, cirrhosis at presentation, and treatment. Data are presented as the mean ⫾ standard deviation along with range in both the text and tables.

HEPATOLOGY, Vol. 42, No. 1, 2005

Fig. 1. Age at presentation. The number of patients with AIH divided by decade of age at first presentation and separated into symptomatic and asymptomatic patients is shown. Unlike in previous studies, a bimodal age at presentation was not seen in either group of patients.

Results Of the 135 patients who had been given a clinicalpathological diagnosis of autoimmune hepatitis, 9 were excluded. Three of these patients gave a history of recent treatment with minocycline, no information was available for 2 patients after their initial visit, and the other 4 patients did not satisfy the requirements for a diagnosis of AIH based on their calculated scores using the revised AIH scoring system. The age at diagnosis ranged from 9.5 to 75.6 years (mean, 43.5 ⫾ 16.6 years) and was distributed fairly uniformly across decades (Fig. 1), with 15 (11.9%) patients presenting in the pediatric age range (⬍18 years), 5 of whom had type II AIH with positive anti-LKM antibodies. There was a female predominance (75.4%), and although most patients were white (73.8%), a number of other racial groups were represented: 7 (5.5%) black, 8 (6.3%) Asian, 8 (6.3%) East Indian, and 10 (7.9%) of other backgrounds. Patients were followed up for a mean of 8.00 ⫾ 6.86 years (range, 0.6-29.3 years). Seventy-one patients (56.3%) had a “definite” diagnosis of AIH according to the AIH revised scoring system, whereas 55 patients (43.7%) had a “probable” diagnosis of AIH. The mean score for the patients was 15.7 ⫾ 3.2 before treatment and 18.1 ⫾ 4.2 after treatment. Five patients (4.0%) had type 2 AIH with positive LKM antibodies. Two patients had concomitant hepatitis C infection with documented HCV antibodies and HCV RNA positivity. Hepatitis C testing was unavailable in 7 patients (7.4%), because they were seen before the introduction of the assay. Six of these 7 patients had high-titer IgG, ANA/ SMA of at least 1:80, and a “definite” diagnosis of AIH by the AIH scoring system; none had risk factors for HCV acquisition. Consequently, these patients likely had AIH and were therefore included in the analysis. The other patient had negative autoantibodies and a minimally elevated IgG with an AIH score of 12. Given the diagnostic

FELD ET AL.

55

uncertainty, this patient was excluded. A total of 5 (4.0%) patients were anti-mitochondrial antibody (AMA) positive. None of the AMA⫹ patients had an ALP ⬎ 1.5⫻ the upper limit of normal (ULN), all had normal IgM, and none had duct injury or ductopenia on liver biopsy, suggesting that they did not have primary biliary cirrhosis or clear evidence of an overlap syndrome. None developed symptoms, signs, or laboratory markers typical of primary biliary cirrhosis during follow-up.9 Most patients (98, 78.4%) were treated with immunosuppressive therapy: glucocorticoid monotherapy in 33 (26.4%) and glucocorticoids plus azathioprine (approximately 1 mg/kg/day) in 65 (52.0%). The mean duration of steroid therapy was 4.1 ⫾ 4.7 years (range, 0.2-24.0 years) compared with 3.62 ⫾ 4.2 years (range, 0.1-23.4 years) for azathioprine. A complete response to treatment was achieved in 83 patients (84.7%), whereas 13 (13.3%) achieved a partial response, and only 2 patients (2.0%) did not respond to therapy. Half of the patients relapsed after either withdrawal of therapy or a dose reduction; however, all relapsers responded to retreatment. Relapsers were indistinguishable from non-relapsers at baseline. Eleven (8.7%) patients were able to remain off all immunosuppressive therapy after 2 years of quiescent disease with no relapse during follow-up (mean, 3.3 years; 2-9 years) (Table 1). Nine patients (7.1%) taking prednisone and 10 patients (7.9%) taking azathioprine reported a significant adverse side effect necessitating dose reduction or discontinuation of the offending agent. Endpoints. A total of 20 patients (16.1%) reached an endpoint during the course of study. Ten patients (8.1%) underwent orthotopic liver transplantation, 8 (6.5%) died as a result of a complication of their liver disease, and 2 (1.6%) died of a myocardial infarction believed in both instances to be unrelated to their underlying liver condition. The calculated survival to death or liver transplantation for the entire cohort was 89.6% (83.7%-95.4%) at 5 years and 83.0% (75.1%-90.8%) at 10 years. Five- and 10-year survival to liver-related death or transplantation was 90.5% (85.9%-96.1%) and 85.1% (77.6%-92.5%), respectively (Fig. 2). Asymptomatic Patients. Thirty-one of the 124 patients (25.0%) were asymptomatic at initial presentation, and all were identified through investigation, including liver biopsy, of incidental abnormal liver tests. Asymptomatic patients constituted 2 distinct groups: 8 (25.8%) with inactive (“burned out”) cirrhosis with no signs of hepatic decompensation/portal hypertension and nearnormal liver enzymes, and 23 (74.2%) with elevated liver enzymes associated with inflammatory activity but no evidence of cirrhosis on biopsy. Eight (25.8%) initially asymptomatic patients went on to develop symptoms

56

FELD ET AL.

HEPATOLOGY, July 2005

Table 1. Baseline Characteristics of Asymptomatic and Symptomatic Patients With AIH

Mean age at presentation (years) Mean follow-up (years) Baseline Laboratory Values AST (⬍40 IU/L) ALT (⬍40 IU/L) ALP (⬍112 IU/L) Bilirubin (⬍23 mM) Albumin (35–50 g/L) IgG (5–13 g/L) Platelets (150–400 ⫻ 103) ANA ⫹ (ⱖ1:40) SMA ⫹ (ⱖ1:40) LKM ⫹ (ⱖ1:40) AMA ⫹ (ⱖ1:40) Cirrhosis at presentation Mean AIH score—pre/post treatment Mean AIH Biopsy Score Mean lobular inflammation score† Received treatment Pred alone Pred ⫹ Aza Complete response Partial response Relapse

Asymptomatic at Presentation (n ⴝ 31)

Symptomatic at Presentation (n ⴝ 94)

P

48.5 ⫾ 12.30* (15.3–75.6) 6.03 ⫾ 5.81 (0.3–19.6)

41.63 ⫾ 17.42* (9.5–72.9) 8.63 ⫾ 7.08 (0.3–29.3)

⬍.05 NS

254 ⫾ 353* (34–1259) 252 ⫾ 270* (11–1146) 123 ⫾ 49 (52–254) 16 ⫾ 12* (4–24) 43 ⫾ 5 (29–55) 21.3 ⫾ 10.6* (3.4–63) 226 ⫾ 86 (43–412) 23 (74.4%) 22 (71.0%) 0 2 (6.45%) 8 (25.8%) 14.6/16.0* 2.61 ⫾ 1.96 2.53 ⫾ 0.88* 15 (48.4%)* 3 (9.7%) 12 (38.7%) 10 (66.7%)* 5 (33.3%)* 7 (46.7%)

515 ⫾ 644* (20–3600) 490 ⫾ 641* (18–4091) 160 ⫾ 108 (41–590) 80 ⫾ 116* (2–526) 38 ⫾ 6 (21–50) 28.1 ⫾ 14.2* (9–100) 203 ⫾ 91 (60–420) 67 (71.2%) 64 (68.1%) 5 (5.3%) 3 (3.19%) 34 (36.2%) 15.9/18.8* 2.69 ⫾ 2.18 2.00 ⫾ 0.68* 84 (89.4%)* 24 (25.5%) 60 (63.8%) 74 (88.1%)* 8 (9.5%)* 42 (50.0%)

.05 .05 NS .05 NS .05 NS NS NS NS NS NS ⬍.05 NS ⬍.05 ⬍.001 NS NS .03 .01 NS

NOTE. Baseline characteristics for asymptomatic and symptomatic patients with autoimmune hepatitis (AIH) are shown. AIH score and AIH biopsy score refer to values calculated from the International Autoimmune Hepatitis Group Revised Scoring System.8 HAI biopsy score refers to the Hepatic Activity Index score for lobular inflammation and is scored out of a maximum of 4. Abbreviations: LKM, liver-kidney-microsomal antibody; Pred, prednisone; Aza, azathioprine; NS, nonsignificant (P ⬎ .05). *P ⬍ .05 between the 2 values. †Values refer to the lobular inflammation score of the Hepatic Activity Index.

during follow-up: jaundice in 3 patients, arthralgias in 3 patients, ascites in 1 patient, and onset of fatigue severe enough to warrant treatment in 1 individual. In this group, symptoms developed a mean of 2.00 ⫾ 2.46 years

Fig. 2. 10-year survival for total cohort— death or transplantation. A Cox regression survival curve measuring survival for the entire cohort. The 10-year survival was 83.0% (75.1%-90.8%).

(range, 0.8 to 8 years) after diagnosis. Treatment was instituted in all 8 patients, with good response. Asymptomatic patients were older at presentation (asymptomatic 48.5 ⫾ 12.3 years vs. symptomatic 41.63 ⫾ 17.42 years, P ⬍ .05) (Table 1). Asymptomatic patients had lower pre- and post-treatment AIH scores than symptomatic subjects, and this difference was reflected in a greater proportion of asymptomatic patients with a diagnosis of “probable” AIH (64.5% asymptomatic vs. 37.2% symptomatic) than “definite” AIH (35.5% asymptomatic vs. 63.8% symptomatic) according to the revised AIH scoring system (P ⬍ .05). Cirrhosis was present at diagnosis in a similar proportion of asymptomatic (25.8%), and symptomatic (36.2%) patients and mean follow-up was similar for both groups (asymptomatic mean of 6.03 ⫾ 5.81 years; range, 0.30 to 19.6 years vs. symptomatic mean 8.63 ⫾ 7.08 years; range, 0.3029.3 years). In keeping with their milder clinical presentation, the asymptomatic cohort had significantly lower serum aminotransferase, bilirubin, and immunoglobulin levels than their symptomatic counterparts at baseline. However, despite the absence of symptoms, some patients had very high ALT and IgG levels at baseline (ALT range, 11-

HEPATOLOGY, Vol. 42, No. 1, 2005

1,146 IU/L; IgG range, 3.4-63 g/L). Similar proportions in both groups were ANA, SMA, LKM, and AMA positive (Table 1). No baseline laboratory values were associated with the development of symptoms or ultimate outcome (Table 5). No permanent spontaneous remissions were noted in either symptomatic or asymptomatic patients. Although the AIH biopsy scores were similar (asymptomatic 2.61 ⫾ 1.96 and symptomatic 2.69 ⫾ 2.18), the symptomatic patients had more active histology as evaluated by the lobular inflammation score of the hepatic activity index (symptomatic 2.54 ⫾ 0.88 and asymptomatic 2.00 ⫾ 0.68, P ⫽ .04). A majority (56.3%) of asymptomatic patients had interface hepatitis, with 50% (9) of these patients showing rosettes on biopsy. All asymptomatic patients had a lymphoplasmacytic infiltrate, and none had confluent necrosis on liver biopsy. Similarly, in the symptomatic group, all had a lymphoplasmacytic infiltrate, 66 (70.2%) had interface hepatitis (22 [33.3%] with resetting), and 6 (6.4%) patients had confluent necrosis. Fifteen of the asymptomatic patients (48.4%) received immunosuppressive therapy, consisting of prednisone and azathioprine in 80% and prednisone monotherapy in 20%. Therapy was initiated at the onset of symptoms in 8 patients and by the referring physician in 7 individuals. The other 16 asymptomatic patients (51.6%) were not treated, as this was the practice pattern at our institution. Whereas a higher percentage (89.4%) of symptomatic patients did receive therapy, a proportion were not treated because of refusal of recommended therapy (n ⫽ 4), an acute presentation leading to liver transplantation (n ⫽ 1), a contraindication to therapy (n ⫽ 1), or symptoms related to portal hypertension with burned-out cirrhosis (n ⫽ 5). When immunosuppressive therapy was given, the duration of therapy was similar between patients with symptomatic and asymptomatic disease; however, biochemical complete response rates were higher in the symptomatic patients (88.1% symptomatic vs. 66.7% in the asymptomatic group) (P ⫽ .03). A similar percentage of patients in each group went on to reach an endpoint during follow-up (asymptomatic 5/31 [16.1%], symptomatic 15/94 [16.0%]). Among patients asymptomatic at presentation, 3 developed progressive liver disease. Two of these patients underwent successful transplantations and the third died awaiting transplantation. One of these patients had inactive, burned-out cirrhosis on liver biopsy and received immunosuppressive therapy only after developing ascites, jaundice, and liver transaminase elevation. The other 2 initially asymptomatic patients that went on to reach liver-related endpoints were both treated with prednisone and azathioprine from the time of diagnosis (i.e., before

FELD ET AL.

57

Fig. 3. Ten-year survival for symptomatic versus asymptomatic patients with AIH. A Cox regression survival curve comparing actuarial survival among symptomatic and asymptomatic patients with AIH. Death of any cause and liver transplantation were used as endpoints. Two of three asymptomatic patients who reached an endpoint during follow-up died of unrelated myocardial infarction. The 10-year survival was 80.0% (62.5%-97.5%) for asymptomatic patients and 83.8% (75.1%-92.6%) for symptomatic patients (P ⫽ NS).

the onset of symptoms). Two other asymptomatic patients died of myocardial infarctions seemingly unrelated to their underlying liver disease (neither of these individuals received treatment). Of the 15 patients symptomatic at presentation who went on to reach an endpoint, 8 underwent liver transplantation and 7 died of progressive liver disease. Endpoints occurred a mean of 9.33 ⫾ 8.01 years (range, 0.5-25 years) after diagnosis. Eleven of the 15 patients received treatment while 4 had inactive, burned-out cirrhosis with only complications of portal hypertension and therefore were not treated with immunosuppressive therapy. The overall 5- and 10-year survival was 87.2% (73.7%-100%) and 80.0% (62.5%-97.5%), respectively, for the asymptomatic group compared with 90.2% (83.8%-96.7%) and 83.8% (75.1%-92.6%), respectively, in the symptomatic group (P ⫽ NS) (Fig. 3). The overall 5- and 10-year survival for liver-related outcomes (liver death or transplant) was 91.3% (79.8%-100%) and 89.5% (75.7%-100%), respectively, for asymptomatic patients compared with 90.2% (83.8%-96.7%) and 83.8% (75.1%-92.6%), respectively, for symptomatic patients (P ⫽ NS) (Table 4, Fig. 4). There was no difference in the baseline characteristics or outcome between asymptomatic patients who received treatment and those who did not (Table 2). Cirrhosis. Liver biopsy demonstrated the presence of

58

FELD ET AL.

HEPATOLOGY, July 2005

Fig. 4. Ten-year survival to liver-related endpoints for symptomatic and asymptomatic patients with AIH. A Cox regression survival curve comparing survival to a liver-related endpoint among symptomatic and asymptomatic patients with AIH. Endpoints included liver-related death and liver transplantation. The 10-year survival to a liver-related outcome was 89.5% (75.7%-100%) for asymptomatic patients compared with 83.8% (75.1%-92.6%) for symptomatic patients (P ⫽ NS).

cirrhosis in 42 patients (33.9%) at the time of diagnosis. Initial laboratory data was similar between patients with and without cirrhosis, with the exception of lower platelet counts in the group with cirrhosis (158 ⫾ 94 with cirrho-

sis and 221 ⫾ 79 without cirrhosis, P ⬍ .05). Patients with cirrhosis had similar demographics, AIH total scores, and AIH scoring system biopsy scores to those without cirrhosis at baseline. Patients without cirrhosis had more active histology than those with cirrhosis at presentation as assessed by the hepatic activity index (with cirrhosis, 2.15 ⫾ 0.81; without cirrhosis, 2.68 ⫾ 0.63, P ⫽ .028). This finding reflects that many of the patients with cirrhosis at presentation had burned-out inactive disease. Similar proportions of both groups were symptomatic at presentation (81.5% with cirrhosis and 73.1% without cirrhosis), and they were treated similarly with comparable complete response and relapse rates. Duration of follow-up was similar in both groups (Table 3). The major difference noted between those with and without cirrhosis at baseline was the proportion reaching an endpoint during follow-up. Sixteen patients with cirrhosis (38.1%) reached an endpoint, with 9 undergoing orthotopic liver transplantation, 5 dying of end-stage liver disease, and 2 dying of heart disease. In contrast, only 4 patients (4.8%) without cirrhosis at presentation reached an endpoint during follow-up: 1 patient required a liver transplant, and 3 died of progressive liver disease (P ⬍ .05). The 5- and 10-year survival was 76.3% (62.8%89.8%) and 61.9% (44.9%-78.9%), respectively, for all patients with cirrhosis compared to 96.7% (92.3%100%) and 94.0% (87.4%-100%), respectively, for patients without cirrhosis (P ⫽ .003) (Fig. 5). The 5- and

Table 2. Characteristics of Treated and Untreated Asymptomatic Patients Mean age at presentation (years) Mean follow-up (years) Baseline Laboratory Values AST (⬍40 IU/L) ALT (⬍40 IU/L) ALP (⬍112 IU/L) Bilirubin (⬍23 mmol/L) Albumin (35–50 g/L) IgG (5–13 g/L) Platelets (150–400 ⫻ 103) ANA ⫹ (ⱖ1:40) SMA ⫹ (ⱖ1:40) Cirrhosis at presentation Mean AIH score—pre/post treatment Mean AIH biopsy score Mean lobular inflammation score† Endpoints Transplant Liver death All death

Treated (n ⴝ 15)

Untreated (n ⴝ 16)

P

46.5 ⫾ 12.57 (23–68) 7.73 ⫾ 5.82 (0.92–20.8)

51.2 ⫾ 12.77 (31–75) 4.48 ⫾ 5.48 (0.5–22.61)

NS NS

275 ⫾ 375 267 ⫾ 282 135 ⫾ 59 18 ⫾ 13 41 ⫾ 5 21.3 ⫾ 10.6 222 ⫾ 67 12 (80.0%) 11 (73.3%) 1 (6.7%)* 15.7/18.4* 2.61 ⫾ 1.88 2.00 ⫾ 0.63 2 (13.3%) 2 (13.7%) 0 0

234 ⫾ 343 239 ⫾ 266 111 ⫾ 35 15 ⫾ 9.9 43 ⫾ 5 18.0 ⫾ 8.1 229 ⫾ 104 13 (81.3%) 11 (68.8%) 7 (43.8%)* 13.7/13.7* 1.56 ⫾ 1.67 2.00 ⫾ 0.71 3 (18.8%) 0 1 (6.3%) 3 (18.8%)

NS NS NS NS NS NS NS NS NS .018 ⬍ .05 NS NS NS NS NS NS

NOTE. Among asymptomatic patients with AIH, a proportion received immunosuppressive therapy, and others did not. Baseline characteristics as well as outcomes are shown for both groups. AIH score and AIH biopsy score refer to values calculated from the International Autoimmune Hepatitis Group Revised Scoring System.8 Abbreviations: LKM, liver-kidney-microsomal antibody; NS, nonsignificant (P ⬎ .05). *P ⬍ .05 between the 2 values. †Values refer to the lobular inflammation score of the Hepatic Activity Index.

HEPATOLOGY, Vol. 42, No. 1, 2005

FELD ET AL.

59

Table 3. Baseline Characteristics of Patients With AIH With and Without Cirrhosis

Mean age at presentation (years) Mean follow-up (years) Baseline Laboratory Values AST (⬍40 IU/L) ALT (⬍40 IU/L) ALP (⬍112 IU/L) Bilirubin (⬍23 mmol/L) Albumin (35–50 g/L) IgG (5–13 g/L) Platelets (150–400 ⫻ 103) ANA ⫹ (ⱖ1:40) SMA ⫹ (ⱖ1:40) LKM ⫹ (ⱖ1:40) Mean AIH score—pre/post treatment Mean AIH biopsy score Mean lobular inflammation score† Treatment Pred alone Pred ⫹ Aza Complete response Partial response Relapse

Cirrhosis at Presentation (n ⴝ 42)

No Cirrhosis at Presentation (n ⴝ 83)

P

44.8 ⫾ 20.02 (9.5–75.2) 8.24 ⫾ 7.24 (0.5–28.68)

42.9 ⫾ 14.7 (10.4–72.7) 7.74 ⫾ 6.36 (0.5–29.35)

NS NS

406 ⫾ 526 377 ⫾ 502 160 ⫾ 128 49 ⫾ 70 39.2 ⫾ 5 26.1 ⫾ 12.3 158 ⫾ 94* 34 (80.9%) 23 (54.8%) 2 (4.8%) 15.1/17.2 2.64 ⫾ 2.12 2.15 ⫾ 0.81* 31 (73.8%) 10 (23.8%) 21 (50.0%) 25 (80.6%) 4 (12.9%) 14 (45.2%)

467 ⫾ 629 453 ⫾ 617 145 ⫾ 80 71 ⫾ 117 39.4 ⫾ 7 26.2 ⫾ 14.4 221 ⫾ 79* 65 (77.4%) 63 (75.0%) 3 (3.6%) 15.9/18.5 2.69 ⫾ 2.13 2.68 ⫾ 0.63* 68 (81.9%) 16 (19.3%) 52 (62.3%) 59 (86.8%) 9 (13.2%) 20 (29.4%)

NS NS NS NS NS NS ⬍ .05 NS NS NS NS NS ⬍ .05 NS NS NS NS NS NS

NOTE. The baseline characteristics and response to treatment of patients with and without cirrhosis at initial liver biopsy are shown. AIH score and AIH biopsy score refer to values calculated from the International Autoimmune Hepatitis Group Revised Scoring System.8 Abbreviations: LKM, Liver-kidney-microsomal antibody; Pred, prednisone; Aza, azathioprine; NS, nonsignificant (P ⬎ .05). *P ⬍ .05 between the 2 values. †Values refer to the lobular inflammation score of the Hepatic Activity Index.

10-year survival for liver-related outcomes was 78.7% (65.6%-91.8%) and 67.2% (50.6%-83.9%), respectively, for patients with cirrhosis compared with 96.7% (92.3%-100%) and 94.0% (87.4%-100%), respectively,

Fig. 5. Ten-year survival for patients with AIH with and without cirrhosis at baseline. A Cox regression survival curve comparing survival among patients with AIH with and without cirrhosis at baseline. Death of any cause and liver transplantation were used as endpoints. The 10-year survival was 61.9% (CI, 44.9%-78.9%) for patients with cirrhosis compared with 94.0% (CI, 87.4%100%) for patients without cirrhosis at baseline (P ⫽ .003).

for patients without cirrhosis at presentation (P ⫽ .003) (Table 4). For the Cox-regression analysis, 112 of the 125 patients were assessed. Nine were excluded because of missing data and 5 were removed because of censoring before the first occurrence of a death or transplantation. By univariate analysis, baseline laboratory values, AIH score, AIH biopsy score, and gender had no effect on survival and were therefore removed from the model (Table 5). The final multivariate model was created using the presence of symptoms and cirrhosis at presentation, age at presentation, and treatment (Table 6). Symptomatic and asymptomatic patients were found to be at equal risk of death or liver transplantation (RR, 0.85; CI, 0.23-3.13; P ⫽ .804). In contrast, cirrhosis at presentation significantly increased the risk of death or transplantation (RR, 6.79; CI, 1.91-24.16; P ⫽ .003), as did greater age at presentation (RR, 1.03; CI, 1.00-1.07; P ⫽ .037). Although the RR of 1.03 for age at presentation is small, the risk is 1.03 times greater with each additional year of age and consequently is very relevant with large age differences. Treatment with immunosuppressive agents was not found to affect outcome (RR, 0.38; CI, 0.12-.1.25; P ⫽ .111); however, this is likely a result of significant bias in treatment decisions (Table 6). Most patients were treated, and those who did not receive immunosuppres-

60

FELD ET AL.

HEPATOLOGY, July 2005

Table 4. Outcome of Patients With AIH by Subgroup Asymptomatic at Presentation (n ⴝ 31)

Endpoints Transplant Liver death All death 5-Year survival 10-Year survival 5-Year liver-related survival† 10-Year liver-related survival†

Symptomatic at Presentation (n ⴝ 94)

5 (16.1%) 2 (6.5%) 1 (3.2%) 3 (9.7%) 87.2% (CI 73.7–100%) 80.0% (CI 62.5–97.5%) 91.3% (CI 79.8–100%) 89.5% (CI 75.7–100%)

15 (16.0%) 8 (8.5%) 7 (7.4%) 7 (7.4%) 90.2% (CI 83.8–96.7%) 83.8% (CI 75.1–92.6%) 90.2% (CI 83.8–96.7%) 83.8% (CI 75.1–92.6%)

P

NS NS NS NS NS .80 NS .22

Cirrhosis at Presentation (n ⴝ 42)

16* (38.1%) 9* (14.2%) 5 (11.9%) 7* (16.7%) 76.3%* (CI 62.8–89.8%) 61.9%* (CI 44.9–78.9%) 78.7%* (CI 65.6–91.8%) 67.2%* (CI 50.6–83.9%)

No Cirrhosis at Presentation (n ⴝ 83)

4* (4.8%) 1* (1.2%) 3 (3.6%) 3* (3.6%) 96.7%* (CI 92.3–100%) 94.0%* (CI 87.4–100%) 96.7%* (CI 92.3–100%) 94.0%* (CI 87.4–100%)

P

⬍.0001 ⬍.0001 .02 ⬍.05 .003 ⬍ .05 .003

All Patients (n ⴝ 125)

20 (16.0%) 10 (8.3%) 8 (6.4%) 10 (8.3%) 89.6% (CI 83.7–95.4%) 83.0% (CI 75.1–90.8%) 90.5% (CI 84.9–96.1%) 85.1% (CI 77.6–92.5%)

NOTE. The outcomes of symptomatic and asymptomatic patients with AIH are compared, as are those of patients with and without cirrhosis at initial liver biopsy. The final column shows the outcomes for all patients with AIH included in the study. Abbreviation: NS, nonsignificant (P ⬎ .05). *P ⬍ .05 between the 2 values. †Survival to either liver transplant or liver-related mortality.

sive therapy were for the most part patients with burnedout, inactive cirrhosis. Consequently, this result should not be interpreted as suggesting that immunosuppressive therapy is not of value in patients with AIH.

Discussion Little is known about the natural history of asymptomatic patients with AIH, and therefore, the role of immunosuppressive therapy with its potential toxicities has remained unclear. This review of a large cohort of patients with AIH, 25% of whom were entirely symptom free at

Table 5. Univariate Cox Regression Survival Analysis—Death or Liver Transplantation Factor

Wald Statistic DF

Cirrhosis at presentation 12.62 Treatment 9.49 Age at presentation 9.49 Presence of symptoms 0.627 Sex 0.021 ALT 0.017 AST 0.544 ALP 0.566 Plt 0.059 IgG 0.469 AIH biopsy score 0.065 Pretreatment AIH score 2.27 Post-treatment AIH score 0.566

P

Risk Ratio (95% CI)

1 1 1 1 1 1 1 1 1 1 1 1

⬍ .0001 .002 .002 .429 .884 .895 .461 .452 .808 .494 .065 .132

7.35 (2.450–22.22)* 4.476 (1.725–11.62)* 1.052 (1.019–1.187)* 0.661 (0.237–1.842) 0.932 (0.361–2.407) 1.001 (0.999–1.001) 1.001 (0.999–1.001) 1.002 (0.997–1.006) 0.999 (0.994–1.005) 0.987 (0.952–1.024) 0.976 (0.812–1.170) 1.022 (0.993–1.052)

1

.452

1.002 (0.997–1.006)

NOTE. A univariate Cox regression analysis of factors affecting patient outcome. Only the presence of cirrhosis, age at presentation, and treatment affected outcome. Presence of symptoms was used in the multivariate model as well to ensure no small effect was overlooked. *P ⬍ .05 between the 2 values.

initial presentation, describes the natural history of this clinically silent variant of AIH. Early descriptions of AIH documented a very aggressive disease with a high mortality rate.1 The introduction of immunosuppressive therapy, either using glucocorticoids alone or in combination with azathioprine, had a very significant effect on long-term survival2-7; however, a high incidence of adverse effects was also reported. Although not specifically stated, it is likely that patients recruited to the original treatment trials of AIH, initiated back in the 1960s, were all symptomatic with severe disease. The mean bilirubin at presentation ranged from 46.8 to 88.4 mmol/L3,4 in early studies compared with 16 ⫾ 12 mmol/L in our asymptomatic cohort. Therefore, whether asymptomatic patients with AIH would accrue the same risks:benefits from therapy is unclear. Our results demonstrate that asymptomatic AIH is not uncommon and consists of 2 groups of patients: those with burned-out inactive cirrhosis and those with mildly active ongoing hepatitis. Asymptomatic patients had lower liver Table 6. Multivariate Cox Regression Survival Analysis—Death or Liver Transplantation Factor

Wald Statistic

DF

P

Risk Ratio (95% CI)

Cirrhosis at presentation Age at presentation Treatment Presence of Symptoms

8.74 4.35 2.53 0.06

1 1 1 1

.003 .037 .111 .804

6.79 (1.91–24.16)* 1.03 (1.00–1.07)* 0.38 (0.12–1.25) 0.85 (0.24–3.13)

NOTE. A multivariate Cox regression analysis of factors affecting patient outcome. The presence of cirrhosis and older age at baseline increase the risk of reaching an endpoint while the presence of symptoms and the use of immunosuppressive therapy do not affect outcome. *P ⬍ .05 between the 2 values.

HEPATOLOGY, Vol. 42, No. 1, 2005

enzyme and IgG elevations and lower AIH and hepatic activity index scores, but were otherwise indistinguishable from symptomatic patients. During follow-up, most had persistently abnormal liver tests, with no permanent spontaneous remissions observed. Eight (25.8%) of the initially asymptomatic patients went on to develop symptoms during the study period, highlighting the importance of close follow-up in this group of patients. Unfortunately, no markers identified patients who became symptomatic. We found that asymptomatic patients had a good prognosis, with an overall 80.0% 10-year survival. Although the survival curves for asymptomatic and symptomatic patients appear to separate early (Fig. 3), it is important to recognize that this apparent difference relates to the 2 early cardiac deaths in the asymptomatic group and does not reflect a worse prognosis from AIH. The 10-year survival to a liver-related endpoint for asymptomatic patients was 89.5%, which compares very favorably to that of the symptomatic patients (83.8%) (Fig. 4). Previous studies have found that patients with AIH who present with an abrupt onset of severe symptoms fare poorly,10-12 and those with milder disease at presentation may have a better prognosis.13-15 A recent report from Israel found that asymptomatic patients with AIH fared well but that most developed symptoms over time despite the fact that all were treated. They found that they responded well to treatment and required lower doses of corticosteroids to induce disease remission.16 In contrast, we found asymptomatic patients less likely to respond to treatment (66.1% asymptomatic vs. 88.1% symptomatic, P ⬍ .05), and only a minority of our patients developed symptoms during a similar duration of follow-up. The practice of the liver unit in this study was not to treat asymptomatic patients. In total, approximately half (15/31) of the asymptomatic cohort received immunosuppressive therapy at some point in their course: 8 for the development of symptomatic disease and 7 had therapy initiated at the time of diagnosis by the referring physician. Although this variation in practice pattern is a potential source of bias, those given and those not given treatment had similar baseline liver biochemical profiles and biopsy scores (Table 2). The 10year survival of the treated and untreated asymptomatic patients appears similar, but the numbers are too small for statistical analysis. Among those who received treatment from the time of diagnosis, 2 patients progressed to end-stage liver disease and required liver transplantation, whereas in the untreated group, 1 patient died of complications of portal hypertension and 2 patients died of heart disease. Of the 8 patients who were asymptomatic at first diagnosis but subsequently developed symptoms during follow-up, all received treatment, and 1 went on to require a liver transplant years after starting therapy. The patient that required liver

FELD ET AL.

61

transplantation had inactive cirrhosis with near-normal liver enzymes at diagnosis, and therefore it is unlikely that immunosuppressive therapy would have altered his course. Our data suggest that it may be safe to follow asymptomatic patients with a strategy to institute immunosuppressive treatment if symptoms develop over time. No evidence-based data are available to guide treatment decisions for asymptomatic patients with AIH. Czaja1 has suggested that asymptomatic patients with minimal enzyme elevations can safely be observed, whereas those with aminotransferase levels of 10-fold normal or 5-fold normal with twice-normal IgG concentrations, merit treatment. In addition, he argues that patients with confluent necrosis on initial liver biopsy should be treated, because this histological finding has been associated with the subsequent development of cirrhosis.1 We did not find that the degree of liver enzyme or IgG elevation was predictive of outcome, and none of the asymptomatic patients had confluent necrosis on liver biopsy. Unfortunately no controlled data exist regarding the treatment of “mild” or asymptomatic AIH, and therefore all recommendations are necessarily based on the original, old studies of chronic active hepatitis.3,17 The rationale for using treatment in asymptomatic disease would be to prevent the development of fibrosis and ultimately cirrhosis, with its complications. Little evidence has been found that this can be achieved. In a large study examining this issue, Roberts et al.18 found that 40% of patients progressed to cirrhosis despite immunosuppressive therapy. Dufour et al.19 found that in a small number of patients, fibrosis and even cirrhosis was reversible with treatment of AIH, and although this is an important finding, in this small, uncontrolled study, no comment was made on progression of fibrosis in the face of therapy. A recent report from Czaja et al.20 found that fibrosis scores were stable or improved during steroid therapy in most patients with AIH. Because these patients were all likely symptomatic and therefore required therapy, there was no control group of untreated patients for comparison. Whether patients with mild or asymptomatic disease would accrue the same benefit from immunosuppressive therapy remains unclear. In keeping with their milder clinical course and lower aminotransferase elevations, the asymptomatic patients also had less inflammation on liver biopsy, as reflected by lower lobular inflammation activity scores. Although the AIH biopsy scores were similar between the 2 groups, it is important to note that the AIH biopsy score was designed specifically to confirm the diagnosis of AIH rather than to assess the severity or activity of the disease. For the group as a whole, one third of patients were found to have established cirrhosis at presentation, which is similar to that reported in previous studies of AIH.16 Before the recognition of the benefit of corticosteroid

62

FELD ET AL.

therapy, cirrhosis at presentation was cited as a predictor of poor outcome in AIH, with 5- and 10-year survival rates of 40% and 18%, respectively.21-23 As diagnostic criteria and treatment for AIH became better defined, survival rates among patients with AIH and cirrhosis improved.24 In the most recent report on the natural history of AIH from the Mayo clinic, Roberts et al.18 found that the 10-year survival of patients with cirrhosis was similar to that of patients without cirrhosis at baseline (93%). Although we used the same methods as the Mayo group to diagnose AIH, when our entire cohort was analyzed, the survival results for patients with cirrhosis at diagnosis were less favorable. Patients with cirrhosis at presentation had a worse outcome than those without cirrhosis (61.9% vs. 94.0% 10-year survival, P ⫽ .003) and were 6.79 (CI, 1.91-24.16) times more likely to reach an endpoint during follow-up (Table 6). Roberts et al. attributed their successful outcome (93% 10-year survival) to the immunosuppressive therapy given to all of their patients with cirrhosis and AIH. Although in our study, a proportion (26%) of patients without cirrhosis did not receive treatment, this group did not account for the reduced survival. Of the 16 patients with cirrhosis who reached endpoints, all but 3 had received treatment, 2 of whom did not die of liver disease. Our study and that from the Mayo Clinic were of similar size and design, and no demographic or clinical differences were identified between the 2 cohorts to account for the worse prognosis among our patients with cirrhosis. However, we used Cox regression rather than Kaplan-Meier analysis to assess survival. By controlling for multiple variables in a time-dependent manner, our survival data are more robust, albeit apparently less optimistic. A recent study by Verma et al.25 also found that patients with cirrhosis at presentation were more likely to die or develop complications of their liver disease during follow-up. Our results suggest that asymptomatic type 1 AIH is not uncommon and is generally associated with a favorable outcome. Most patients will not develop symptoms during follow-up, and they appear to do well without immunosuppressive therapy at least for as long as they remain asymptomatic. The slow progression and rarity of asymptomatic AIH make a prospective randomized controlled trial very difficult to perform; however, our findings should be confirmed in other cohorts before the “wait and see” approach to asymptomatic AIH is widely adopted. Our data confirm previous findings that cirrhosis at presentation portends a poor prognosis in AIH.

References 1. Czaja AJ. Natural history, clinical features, and treatment of autoimmune hepatitis. Semin Liver Dis 1980;4:1-11.

HEPATOLOGY, July 2005

2. Cook GC, Mulligan R, Sherlock S. Controlled prospective trial of corticosteroid therapy in active chronic hepatitis. Q J Med. 1971;40:159-85. 3. Soloway R, Summerskill DM, Baggenstoss AH, Geall M, Gitnick G, Elveback L, et al. Clinical, biochemical, and histological remission of severe chronic active liver disease: a controlled study of treatments and early prognosis. Gastroenterology 1972;63:820-833. 4. Murray-Lyon IM, Stern RB, Williams R. Controlled trial of prednisone and azathioprine in chronic active hepatitis. Lancet 1973;1:1735-1737. 5. Stellon J, Keating J, Johnson PJ, McFarlane IG, Williams R. Maintenance of remission in autoimmune chronic active hepatitis with azathioprine after corticosteroid withdrawal. HEPATOLOGY 1988;8:781-784. 6. Wang KH, Czaja AJ. Prognosis of steroid treatment in HBsAg negative chronic active hepatitis in postmenopausal women. Gastroenterology 1989;97:1288-1293. 7. Johnson PJ, McFarlane IG, Williams R. Azathioprine for long-term maintenance of remission in autoimmune hepatitis. N Engl J Med 1995;333: 958-963. 8. Alvarez A, Berg A, Bianchi AB, Bianchi L, Burroughs K, International Autoimmune Hepatitis Group. International Autoimmune Hepatitis Group Report: review of criteria for diagnosis of autoimmune hepatitis. J Hepatol 1999;31:929-938. 9. Czaja AJ, Muratori P, Muratori L, Carpenter HA, Bianchi FB. Diagnostic and therapeutic implications of bile duct injury in autoimmune hepatitis. Liver Int 2004;24:322-329. 10. Mistilis SP, Skyring AP, Blackburn CRB. Active chronic hepatitis. Am J Med 1970;48:484-495. 11. Czaja AJ, Rakela J, Ludwig J. Features reflective of early prognosis in corticosteroid-treated severe autoimmune chronic active hepatitis. Gastroenterology 1988;95:448-453. 12. Nikias GA, Batts KP, Czaja AJ. The nature and prognostic implications of autoimmune hepatitis with an acute presentation. J Hepatol 1994;21:866871. 13. Thaler H. The natural history of chronic hepatitis. In: Schaffner F, Sherlock S, Leevy CM (eds). The Liver and Its Diseases. New York: Stratton Intercontinental Medical Books, 1974:207-215. 14. Schalm SW, Korman MG, Summerskill WH, Czaja AJ, Baggenstoss AH. Severe chronic active liver disease: prognostic significance of initial morphologic patterns. Am J Dig Dis 1977;22:973-980. 15. Degroote J, Fevery J, Lepoutre L. Long-term follow-up of chronic active hepatitis of moderate severity. Gut 1978;19:510-513. 16. Kogan J, Safadi R, Ashur Y, Shouval D, Ilan Y. Prognosis of symptomatic versus asymptomatic autoimmune hepatitis: a study of 68 patients. J Clin Gastroenterol 2002;35:75-81. 17. Geal MG, Schoenfield LD, Summerskill WHJ. Classification and treatment of chronic active liver disease. Gastroenterology 1968;55:724-729. 18. Roberts SK, Therneau TM, Czaja AJ. Prognosis of histological cirrhosis in type 1 autoimmune hepatitis. Gastroenterology 1996;110:848-857. 19. Dufour JF, DeLellis R, Kaplan MM. Reversibility of hepatic fibrosis in autoimmune hepatitis. Ann Intern Med 1997;127:981-985. 20. Czaja AJ, Carpenter HA. Progressive fibrosis during corticosteroid therapy in autoimmune hepatitis. HEPATOLOGY 2004;39:1631-1638. 21. Willcox R, Isselbacher K. Chronic liver disease in young people: clinical features and course in thirty-three patients. Am J Med 1961;30:185-195. 22. Read AE, Sherlock S, Harrison CV. Active “juvenile” cirrhosis considered as part of a systemic disease and the effect of corticosteroid therapy. Gut 1963;4:378-393. 23. Page AR, Good RA, Pollara B. Long-term results of therapy in patients with chronic liver disease associated with hypergammaglobulinemia. Am J Med 1969;47:765-774. 24. Davis GL, Czaja AJ, Ludwig J. Development and prognosis of histologic cirrhosis in corticosteroid-treated HBsAg-negative chronic active hepatitis. Gastroenterology 1984;87:1222-1227. 25. Verma S, Gunuwan B, Mendler M, Govindrajan S, Redeker A. Factors predicting relapse and poor outcome in type I autoimmune hepatitis: role of cirrhosis development, patterns of transaminases during remission and plasma cell activity in the liver biopsy. Am J Gastroenterol 2004;99:1510-1516.