Outcome of COVID-19 in Patients With Autoimmune Hepatitis: An International Multicenter Study

Study Design

We retrospectively evaluated data of patients with AIH who were diagnosed with COVID-19 between March 11 and November 12, 2020, from 34 centers in 10 countries. All participants independently identified patients and collected data from electronic health records and patients’ follow-up charts. Patients with AIH who were older than 16 years at the time of COVID-19 with a diagnosis confirmed by a PCR-based test were included in the study. Patients with typical radiological findings but with negative PCR tests were not included. All patients with AIH, or overlap with primary biliary cholangitis (PBC) or primary sclerosing cholangitis (PSC), were diagnosed and treated according to international guidelines.^{(8, 13)}

The presence of cirrhosis was diagnosed based on standard imaging studies (elastography, ultrasound, CT, or MRI), liver tissue examination, or clinical findings of portal hypertension or its complications. Complete biochemical response was defined as normal alanine aminotransferase (ALT), aspartate aminotransferase (AST) and immunoglobulin G (IgG) levels. Partial response was defined as a decrease of ALT or AST to below 2 times the upper limit of normal (ULN), and nonresponse was defined as persistently elevated transaminase levels more than 2 times the ULN despite appropriate immunosuppressive therapy.^{(8, 13)} High alcohol consumption was defined as more than 2 drinks/day for men and more than 1 drink/day for women.⁽¹⁴⁾

New-onset liver injury was defined as rising ALT values during COVID-19 and categorized as no or mild (< 2 × ULN), moderate (2-5 × ULN), or severe (>5 × ULN). New-onset ALT > 2 × ULN was used as the definition of liver injury for the study. Patients who were nonresponders to AIH therapy were excluded from new-onset liver injury analysis as they already had ALT > 2 × ULN levels before the COVID-19 diagnosis. The cutoffs for normal values of ALT were considered 25 U/L for women and 35 U/L for men.⁽¹⁵⁾ Severe COVID-19 infection was defined as a composite of intensive care unit (ICU) admission, ventilator use, and/or death, in line with previously reported COVID-19 data.⁽³⁾ The Harran University Hospital of Şanlıurfa was the coordinating center (HRE ID 2020-00682), and the local ethical review boards of the participating centers approved the study.

Data Collection

Collected patient data included general information on patients, autoimmune liver serology, types and doses of immunosuppression, the patient’s AIH response status at last follow-up before COVID-19, and presence of cirrhosis. At the time of COVID-19 diagnosis, body mass index (BMI), clinical features, and comorbid conditions were recorded. Laboratory values at COVID-19 diagnosis and at peak/worst values any time in the first month of COVID-19 were used to evaluate liver injury. Any modification in the dose or type of immunosuppressive drugs during COVID-19, highest care level, hospitalization time, specific COVID-19 therapies, and patient outcomes, providing a minimum of 4 weeks after COVID-19, were recorded.

In order to determine how AIH impacts clinical outcomes with COVID-19, we compared AIH patients to a control group of patients with non-AIH CLD and COVID-19. To identify a control group, we used data from a multicenter, observational study of adult patients with CLD and PCR-confirmed COVID-19. More details regarding the inclusion and exclusion criteria of this cohort have been published.⁽⁷⁾ Only patients without AIH from this cohort were included in the control group. None of the patients in the control group were on immunosuppression.

Statistical Analysis

Continuous variables are presented as mean with standard deviation or as median with range. The Student t test or the Mann-Whitney U test was used for comparisons, as appropriate depending on the distribution. Categorical variables are presented as numbers and percentages, and the chi-squared test was used for comparisons. Multivariable logistic regression analysis models were performed to predict the two major outcomes of the study (dependent variables): significant new-onset liver injury and severe COVID-19. According to the predefined statistical plan, we evaluated associations between each independent variable and outcome after having analyzed the number of outcome events. In addition to predefined sex and age in terms of their clinical relevance for the outcomes, independent variables with a statistically significant relationship (P < 0.1) with dependent (outcome) variables in the univariate analysis were then included in the multivariable models. Hosmer-Lemeshow goodness-of-fit statistics were used to assess model fit. P < 0.05 was considered to represent statistical significance.

We compared clinical outcomes of COVID-19 in patients with AIH and non-AIH CLD. We used propensity score analysis to identify a cohort of patients with non-AIH CLD and COVID-19 who were statistically matched on a 1:2 basis using the nearest neighbor approach. We matched AIH and non-AIH for crucial variables which are known to impact clinical outcomes with COVID-19 including age, gender, presence of cirrhosis, diabetes mellitus, hypertension, and heart diseases.

General Characteristics of the Study Population

Medical data of 115 patients with AIH who acquired COVID-19 were analyzed. Five patients were excluded: 3 patients had previously undergone liver transplantation, and 2 patients were concomitantly diagnosed with AIH and COVID-19 (Fig. 1). The final study group included 110 patients with AIH (80% female) with a median age of 49 (range, 18–85) years at COVID-19 infection. Four of these patients had been included in a previous study.⁽¹¹⁾ The general characteristics, clinical features, and outcomes of the patients are presented in Table 1. Twelve (10.9%) patients had overlap with PBC and 4 (3.6%), with PSC. Coexistence of other immune-mediated disorders was noted in 28 (25.5%) patients, including autoimmune thyroid diseases in 17 (15.4%) patients, inflammatory bowel diseases in 3 (2.7%), Sjögren syndrome in 2 (1.8%), systemic sclerosis in 2 (1.8%), rheumatoid arthritis in 2 (1.8%), systemic lupus erythematosus in 1 (0.9%), celiac disease in 1 (0.9%), ankylosing spondylitis in 1 (0.9%), and antiphospholipid syndrome in 1 (0.9%).

AIH Characteristics Before COVID-19

At last follow-up before COVID-19 diagnosis, 88 (80%) patients were complete responders, 19 (17.3%) were partial responders, and 3 (2.7%) were nonresponders. Thirty-two (29.1%) patients with AIH had features of cirrhosis. A total of 53 (48.2%) patients had comorbid conditions; diabetes mellitus (n = 27, 24.5%) and hypertension (n = 26, 23.6%) were the most common. Seven (6.4%) patients reported active smoking, and 2 (1.8%) had high alcohol consumption. Five (4.5%) patients had a history of malignancy, two cases of which were active during COVID-19.

Most patients (92.7%) were on immunosuppressive therapy prior to the COVID-19 diagnosis. Eight patients (7.3%) were not on immunosuppressive therapy; 6 patients had withdrawn therapy (3 by themselves and 3 by a physician), 1 patient was a nonresponder to available therapies, and 1 had inactive (“burned-out”) cirrhosis. At the time of COVID-19 diagnosis, 65 (59.1%) patients were on prednisone therapy (alone or in combination with other immunosuppressants) with a median dose of 5 (range, 2.5-60) mg/day. Among them, 25 (22.7%) patients were on prednisone ≥ 10 mg/day. The patients’ immunosuppressive therapies are presented in Table 1.

Management of AIH During COVID-19

The majority of patients (n = 96, 87.3%) were symptomatic at the time of COVID-19 diagnosis; cough (n = 68, 61.8%) and fever (n = 65, 59.1%) were the most commonly reported symptoms. Gastrointestinal symptoms (abdominal pain, diarrhea, nausea, and vomiting) were noted in 30 (27.2%) patients.

The dose or type of immunosuppression was modified in 33 (30%) patients during COVID-19 but remained unchanged in 69 (62.7%). General characteristics of the patients according to immunosuppressive therapy status are presented in Table 2. Among patients on azathioprine/6-MP monotherapy (n = 10), doses were reduced in 3 and discontinued in 7. Among patients on steroid monotherapy (n = 2), the dose was reduced in 1 and discontinued in 1. In patients treated with a combination of azathioprine/6-MP and steroids (n = 14), the dose of only azathioprine/6-MP was reduced in 1 and discontinued in 5, the dose of only steroids was reduced in 5, while the doses of both azathioprine/6-MP and steroids were reduced in 1 and discontinued in 2. Azathioprine was discontinued and tacrolimus reduced in a patient who was treated with both drugs. In 1 patient who was on triple immunosuppression, azathioprine was discontinued, and both the steroid and tacrolimus doses were reduced. Among patients treated with MMF and steroids (n = 4), only MMF was discontinued in 3, and both drugs were reduced in 1. The dose was reduced in a patient on MMF monotherapy.

Antibiotics were given to 40 (36.4%) patients during COVID-19 infection. Sixty-seven (60.9%) patients received specific therapy for COVID-19. Hydroxychloroquine (n = 29, 26.4%) and antivirals (n = 22, 20%) were the most commonly used therapies. High-dose steroids were given to 15 (13.6%) patients, and plasma exchange was performed in 3 (2.7%) patients. Tocilizumab and rituximab were used in 1 (0.9%) patient each. The details of COVID-19 therapies are presented in Table 1.

Liver Injury During COVID-19

New-onset liver injury was reported in 37.1% (33/89) of the patients: 16 (18%) had moderate and 17 (19.1%) had severe liver injury. New-onset liver injury was observed in 85.7% (6/7) of patients who were not on immunosuppression, in 51.7% (15/29) of patients whose immunosuppression was modified, and in 22.6% (12/53) of patients whose immunosuppression was unchanged during COVID-19. Predictors of liver injury are presented in Table 3. Multivariate regression analysis revealed that antivirals (defined in Table 1) were associated with new-onset liver injury (P = 0.041; OR, 3.36; 95% CI, 1.05-10.78) and that continued immunosuppression during COVID-19 reduced the probability of liver injury (P = 0.009; OR, 0.26; 05% CI, 0.09-0.71).

Outcomes of COVID-19 in Patients With AIH Compared to Non-AIH CLD

When the outcomes of patients with AIH (n = 110) were compared to a cohort of non-AIH patients with CLD (n = 220) and COVID-19 (Supporting Table S1), no statistically significant differences were noted regarding the rates of all-cause mortality (10% versus 11.5%; P = 0.852), severe COVID-19 (15.5% versus 20.2%; P = 0.231), need for supplemental oxygen (38.2% versus 42.2%; P = 0.553), and hospitalization (46.4% versus 50.0%; P = 0.560) (Fig. 2).

Seventeen AIH patients had severe COVID-19, and 11 patients died due to COVID-19. Predictors of severe outcomes in patients with AIH are presented in Table 4. After adjusting for age, gender, coexistence of PBC/PSC or extrahepatic autoimmune diseases, and AIH therapy response, the presence of cirrhosis was independently associated with severe COVID-19 (P < 0.001; OR, 17.46; 95% CI, 4.22-72.13). Patients with AIH and comorbid conditions (defined in Table 1) tended to have severe COVID-19 infection (P = 0.06; OR, 4.16; 95% CI, 0.94-18.36). Presence of liver injury and continued immunosuppression during COVID-19 were not associated with severe COVID-19.