Uncertainties in Managing Special Circumstances
Long-term management of autoimmune hepatitis can be complicated by treatment dependence, pregnancy and increasing risk of hepatocellular carcinoma. These contingencies are either unaddressed by the guidelines of the AASLD and BSG or the recommendations are discrepant ( Table 3 ).
Treatment Dependence
The average duration of conventional corticosteroid therapy required to induce clinical and laboratory resolution and histological improvement to normal or near-normal tissue is 22 ± 2 months (range, 1–180 months; median, 15 months). Eleven per cent of patients meet these criteria within 6 months; 63% do so within 18 months; and 77% do so within 24 months. The frequency of this response decreases to 3% for each 6 months of treatment beyond 2 years, and 14% of patients require continuous therapy for >36 months. Continuous therapy for more than 3 years without achieving clinical, laboratory and histological resolution is associated with progression to cirrhosis (54%) and need for liver transplantation (15%). The guidelines of the AASLD and BSG agree that long-term maintenance therapy is justified in this small subgroup ( Table 3 ). The discrepancies between the guidelines relate to the optimal time for introducing this alternative strategy. Both guidelines are uncertain about the nature of the maintenance regimen.
The guidelines of the AASLD suggest that conventional corticosteroid therapy be maintained for 36 months and that long-term maintenance therapy adjusted to laboratory evidence of disease activity be instituted thereafter if an end point of treatment had not yet been achieved. The BSG guidelines propose that continuous long-term maintenance therapy, preferably with azathioprine (2 mg/kg daily), be considered for young adults and patients with anti-LKM1 or antibodies to soluble liver antigen (anti-SLA) after maximum clinical and laboratory improvement on conventional treatment. This recommendation assumes that young patients, especially those with certain serological markers, are treatment-dependent and cannot be withdrawn from therapy even after meeting end-point criteria.
Young adults with autoimmune hepatitis (aged <40 years) do respond more slowly to conventional corticosteroid therapy than elderly patients (aged ≥60 years). Only 64% of adults ≤40 years achieve clinical, laboratory and histological remission within 24 months compared to 94% of patients aged ≥60 years. Furthermore, anti-SLA do identify individuals who commonly relapse after corticosteroid withdrawal, and they may indicate a genetic propensity related to HLA DRB1*03 for severe disease and treatment dependence. In contrast, the association of anti-LKM1 with treatment dependence remains controversial.
The decision point must counterbalance the uncertain risks of disease progression associated with the use of nonstandard maintenance regimens against the consequences of indefinite standard therapy and the limited prospect that it will ultimately inactivate the disease ( Table 5 ). Clinical judgment favours the continuation of standard treatment for no longer than 24 months in patients aged ≥60 years and no longer than 36 months for adults aged ≤40 years. Age and serological status should not preclude withdrawal from therapy if end-point criteria are met. Young age and serological status, including the presence of anti-SLA, have been inconsistent predictors of treatment dependence, and they should not preclude the effort.
The long-term maintenance strategy will depend on the degree of residual disease activity. Standard therapy with prednisone alone or in combination with azathioprine can be modified by reducing the daily dose of prednisone by 2.5 mg per month as azathioprine is either added to the regimen or its dose increased to 2 mg/kg daily. The prednisone dose can then be reduced until instability in the laboratory tests precludes further reduction. Prednisone may ultimately be fully withdrawn if azathioprine treatment proves sufficient. The use of calcineurin inhibitors or mycophenolate mofetil for long-term maintenance extends the management strategy beyond clinical precedence and cannot be encouraged unless there is disease progression.
Azathioprine in Pregnancy
Azathioprine has been classified in the US as a category D drug in pregnancy because high doses in animals have produced congenital malformations ( Table 5 ). Active azathioprine metabolites have been demonstrated in the neonates of treated mothers with autoimmune hepatitis, and azathioprine may have been a factor in the occurrence of foetal complications in some pregnant women with the disease. These observations caution against the use of azathioprine during pregnancy, but they have been countered by small retrospective observational studies in autoimmune hepatitis and inflammatory bowel disease that have minimised the foetal risk.
The occurrence of birth defects in 2 of 31 pregnancies (6%) in British women with autoimmune hepatitis could not be attributed to azathioprine therapy. Stillbirths and foetal malformations, ascertained by questionnaire involving 63 pregnancies in 35 treated women with autoimmune hepatitis, occurred at frequencies similar to those in healthy Swedish mothers, and 30 of 35 live German births developed normally despite the in utero exposure of 11 children to azathioprine. Similar results have been reported in 155 azathioprine-treated pregnant women with inflammatory bowel disease.
These retrospective experiences, which in one instance included pregnant women who had actually stopped or reduced their immunosuppressive therapy without physician consultation, have supported, but not established, the safety of azathioprine in pregnancy. This uncertainty has been reflected in the recommendations of the AASLD, which has recommended avoidance of the drug in pregnancy, and the BSG, which has recommended individualised treatment and 'minimal adjustment' of the immunosuppressive regimen.
The clinical judgment must counterbalance the category D status of azathioprine and the uncertainty of its safety to the foetus against the importance of the drug in the management of the mother's disease ( Table 5 ). Importantly, azathioprine is used mainly as a corticosteroid-sparing agent in autoimmune hepatitis, and it is not an essential medication. Corticosteroids alone induce clinical, laboratory and histological improvements as frequently as the combination regimen with azathioprine, and low-dose prednisone therapy (median dose, 7.5 mg daily) can suppress disease activity over the long-term. Furthermore, autoimmune hepatitis during pregnancy typically subsides as high blood oestrogen levels strengthen the anti-inflammatory cytokine pathway, and the disease may not require the conventional drug regimen.
The management decision in pregnancy must be directed at minimising the actual and theoretical risks to the mother and foetus, reducing parental anxieties and maintaining disease control. These goals can be achieved in most instances by avoiding azathioprine therapy during pregnancy ( Table 5 ). Post-partum flares of autoimmune hepatitis are common (12–86%) and worrisome, and they require resumption of standard treatment immediately after delivery.
Surveillance for Liver Cancer
Hepatocellular carcinoma occurs in 1–9% of patients with autoimmune hepatitis and cirrhosis, and the annual incidence is estimated at 1.1–1.9%. Surveillance is considered effective if it extends life by more than 3 months, and it is deemed cost-effective if it increases quality-adjusted life by 1 year at an expense less than $50 000 (USD). Guidelines for managing hepatocellular carcinoma have recommended surveillance in those patients with an annual incidence of hepatocellular carcinoma that exceeds 1.5%. The estimated annual incidence of hepatocellular carcinoma in autoimmune hepatitis is at the edge of this criterion, and surveillance for hepatocellular carcinoma has not been formally endorsed in autoimmune hepatitis. The AASLD and BSG guidelines for autoimmune hepatitis encourage surveillance in otherwise healthy individuals, and they differ from each other mainly in the recommended components of the surveillance strategy ( Table 3 ).
Both the AASLD and BSG guidelines recommend hepatic ultrasonography every 6 months as the cornerstone of surveillance in autoimmune hepatitis. Hepatic ultrasonography has a sensitivity of 65–80% and specificity >90% for hepatocellular carcinoma. The sensitivity for hepatocellular carcinoma is reduced to 38% in nodular cirrhosis and 63% in small early-stage tumours, and it is also compromised by obesity. Operator expertise is another variable that affects the performance of this surveillance tool, especially in detecting small potentially curable tumours of less than 2 cm. Experiences that can detect hepatocellular carcinomas with an average size of 1.6 ± 0.6 cm and diameter less than 3 cm in >98% of cases set a standard that must be emulated.
The recommendations of the AASLD for cancer surveillance in autoimmune hepatitis do not include serum alpha fetoprotein testing, whereas those of the BSG do ( Table 3 ). The argument against performing serum alpha fetoprotein determinations as a surveillance tool is based mainly on the observations that only 10–20% of small tumours present with an abnormal serum alpha fetoprotein level and the number of false positive results generate evaluations that can be misleading, potentially hazardous and expensive. The addition of a serum alpha fetoprotein determination to hepatic ultrasonography increases the frequency of tumour detection by 9%, but it also increases the frequency of false positive findings by 2.4-fold and decreases the positive predictive value of surveillance by 2.2-fold.
The argument in favour of serum alpha fetoprotein testing is that as many as 20% of hepatocellular carcinomas undetected by hepatic ultrasonography have been associated with an increase in serum alpha fetoprotein level and that the addition of the serum alpha fetoprotein determination to hepatic ultrasonography has increased the tumour detection rate in a large, albeit flawed, prospective surveillance study. Furthermore, the combination of serum alpha fetoprotein testing and hepatic ultrasonography has had better performance parameters for detecting hepatocellular carcinoma than either study alone, and a meta-analysis of cohort studies has also supported the combination of tests. The pooled sensitivity for hepatocellular carcinoma for hepatic ultrasonography and serum alpha fetoprotein determination was 70% compared to 63% for hepatic ultrasonography.
Each argument suffers from the absence of strong prospective studies that have been designed and executed to answer the salient question. Extrapolations from retrospective studies using the serum alpha fetoprotein level as a diagnostic tool rather than a surveillance instrument are inappropriate, and theoretical projections that estimate performance parameters from similarly flawed experiences are highly speculative. The key clinical question is how complementary are hepatic ultrasonography and serum alpha fetoprotein determinations in a surveillance strategy. The performance parameters of both tests as a surveillance unit must be determined, and the complementary value of results in each modality clarified.
The decision point in patients with autoimmune hepatitis and cirrhosis must balance the uncertain value of surveillance and the variable performance parameters of the surveillance tools against the prospect of detecting a small (<2 cm), early-stage, curable hepatocellular carcinoma ( Table 5 ). Surveillance for hepatocellular carcinoma in selected patients has been recommended by the AASLD, EASLand the Asian Pacific Association for the Study of Liver Diseases, and it can be justified in autoimmune hepatitis despite the absence of a formal recommendation for this disease.
The issue of surveillance is unsettled in autoimmune hepatitis, and the potential value of surveillance cannot be lost while the issue is debated. The inclusion of a serum alpha fetoprotein determination every 6 months with hepatic ultrasonography combines two deficient modalities, which may be complementary and increase the diagnostic yield. The availability, simplicity, cost and standardisation of the assay for serum alpha fetoprotein support its performance as the debate regarding the usefulness of the test continues and other biomarkers emerge ( Table 5 ).
Commitment to performance of the surveillance strategy is another factor that must be considered in the decision process. In the US, only 20–30% of patients with cirrhosis who later develop hepatocellular carcinoma have undergone surveillance; only 54% of patients with hepatocellular carcinoma who have undergone surveillance have been evaluated by hepatic ultrasonography; and only 12% of patients at risk for hepatocellular carcinoma maintain surveillance on a regular basis after the first year. Compliance with the surveillance recommendation rests heavily on the socio-economic status of the patient, subspecialty practice of the physician and academic affiliation of the health provider. Surveillance without commitment by the patient and physician defeats the objective of the strategy, and the level of commitment must influence the decision to proceed, especially for a recommendation that is weakly endorsed in autoimmune hepatitis.
The cost-effectiveness of cancer surveillance in autoimmune hepatitis and the commitment to a surveillance strategy might be increased if high-risk patients for hepatocellular carcinoma could be better targeted. Patients at risk for hepatocellular carcinoma in autoimmune hepatitis have cirrhosis for at least 5 years, features of portal hypertension, and immunosuppressive treatment for more than 3 years. These selection criteria have not been validated prospectively, and the decision to institute a surveillance programme in patients with autoimmune hepatitis and cirrhosis remains an independent clinical judgment.