SARS-CoV-2 infection in patients with autoimmune hepatitis

Background & Aims Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and coronavirus disease 2019 (COVID-19) continues to have a devastating impact across the globe. However, little is known about the disease course in patients with autoimmune hepatitis (AIH). Methods Data for patients with AIH and SARS-CoV-2 infection were combined from 3 international reporting registries and outcomes were compared to those in patients with chronic liver disease of other aetiology (non-AIH CLD) and to patients without liver disease (non-CLD). Results Between 25th March and 24th October 2020, data were collected for 932 patients with CLD and SARS-CoV-2 infection including 70 with autoimmune hepatitis (AIH). Fifty-eight (83%) patients with AIH were taking ≥1 immunosuppressive drug. There were no differences in rates of major outcomes between patients with AIH and non-AIH CLD, including hospitalization (76% vs. 85%; p = 0.06), intensive care unit admission (29% vs. 23%; p = 0.240), and death (23% vs. 20%; p = 0.643). Factors associated with death within the AIH cohort included age (odds ratio [OR] 2.16/10 years; 1.07–3.81), and Child-Pugh class B (OR 42.48; 4.40–409.53), and C (OR 69.30; 2.83–1694.50) cirrhosis, but not use of immunosuppression. Propensity score matched (PSM) analysis comparing patients with AIH with non-AIH CLD demonstrated no increased risk of adverse outcomes including death (+3.2%; -9.2%–15.7%). PSM analysis of patients with AIH vs. non-CLD (n = 769) demonstrated increased risk of hospitalization with AIH (+18.4%; 5.6–31.2%), but equivalent risk of all other outcomes including death (+3.2%; -9.1%–15.6%). Conclusion Patients with AIH were not at increased risk of adverse outcomes despite immunosuppressive treatment compared to other causes of CLD and to matched cases without liver disease. Lay summary Little is known about the outcomes of COVID-19 in patients with autoimmune hepatitis (AIH), a rare chronic inflammatory liver disease. This study combines data from 3 large registries to describe the course of COVID-19 in this patient group. We show that AIH patients do not appear to have an increased risk of death from COVID-19 compared to patients with other forms of liver disease and compared to patients without liver disease, despite the use of medications which suppress the immune system.


Background
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and coronavirus disease 2019  continues to have a devastating impact across the globe. However, little is known about the disease course in patients with autoimmune hepatitis (AIH).

Methods
Data for patients with AIH and SARS-CoV-2 infection were combined from three international reporting registries and outcomes were compared to those with chronic liver disease of other aetiology (non-AIH CLD) and to patients without liver disease (non-CLD).

Results:
Between 25 th March and 24 th October 2020, data were collected for 932 patients with CLD and SARS-CoV-2 infection including 70 with autoimmune hepatitis (AIH). Fifty-eight (83%) of AIH patients were taking one or more immunosuppressive drug. There were no differences in rates of major outcomes between AIH and non-AIH CLD including hospitalization (76% vs 85%; p= 0.06), ICU admission (29% vs. 23%; p=0.240), and death (23% vs. 20%; p=0.643). Factors associated with death within the AIH cohort included age (OR 2. 16 Little is known about the outcomes of COVID-19 in patients with autoimmune hepatitis (AIH), a rare chronic inflammatory liver disease. This study combines data from three large registries to describe the course of COVID-19 in this patient group. We show that AIH patients do not appear to have an increased risk of death from COVID-19 compared to patients with other forms of liver disease and compared to patients without liver disease, despite the use of medications which suppress the immune system.

Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and resultant coronavirus disease 2019 (COVID- 19) continues to have a devastating impact across the globe [1]. Since the onset of the pandemic, the scientific and clinical community have strived to understand the contributions of specific disease phenotypes to SARS-COV-2 susceptibility and subsequent adverse outcomes [2]. Recently, large international and multicentre cohorts have shown baseline liver disease severity and alcohol related liver disease (ALD) to be independently associated with COVID-19 mortality, with decompensated cirrhosis representing a particularly high-risk group [3][4][5][6]. In contrast, several studies have demonstrated no significant increased risk of critical COVID-19 in patients with previous liver transplantation despite high rates of immunosuppression [7][8][9]. However, no studies have yet evaluated the disease course and outcomes specifically for patients with autoimmune hepatitis (AIH), a large proportion of which will be on concurrent immunosuppressive agents.
The clinical impact of pre-existing immunosuppression in COVID-19 remains complex and incompletely defined. Observations in inflammatory bowel disease and rheumatological conditions have suggested a more severe disease course in those under maintenance treatment with thiopurines or corticosteroids, respectively [10,11]. Furthermore, a multicentre study in Spain demonstrated a higher incidence of SARS-CoV-2 in immunosuppressed liver transplant recipients compared with the general population [8]. In contrast, dexamethasone now has an established role in the management of hospitalised patients with COVID-19 potentially through modification of the hyperactive immune response, and has been shown to reduce mortality by a third in intubated patients and a fifth in those requiring supplemental oxygen therapy [12].
J o u r n a l P r e -p r o o f Current expert recommendations advocate against the routine modification of immunosuppressive therapy in patients with AIH both before and after SARS-CoV-2 infection [13,14]. However, there is little evidence beyond expert consensus and very small cohorts to support these recommendations [15,16]. Furthermore, given the resurgence of the virus in many areas of the world, clinicians and policy makers are being forced to carefully risk stratify patients to establish who may benefit most from enhanced physical and social distancing. A detailed understanding of the clinical course of COVID-19 in patients with AIH is therefore urgently required.
The current study represents an international collaborative effort, bringing together data from three large-scale reporting registries: The European Association for Study of the Liver (EASL) supported COVID-Hep registry, the American Association for the Study of Liver Diseases (AASLD) supported SECURE-cirrhosis registry and the European Reference Network on Hepatological Diseases (ERN RARE-LIVER). To our knowledge, we describe the epidemiology, presentation, disease course and outcomes of the largest cohort of patients with SARS-CoV-2 infection and AIH and offer statistical comparisons with liver disease of other aetiologies and to a contemporaneous cohort of patients without liver disease testing positive for SARS-COV-2.
J o u r n a l P r e -p r o o f

Setting and study design
We combined the data from three multinational registries for patients with laboratory confirmed SARS-CoV-2 and AIH collected between March 25th 2020 and 24th October 2020. These registries included the R-LIVER COVID-19 registry (co-ordinated by the European Reference Network on Hepatological Diseases [ERN RARE-LIVER], with R-LIVER being the general registry of ERN RARE-LIVER), the SECURE-Cirrhosis registry (co-ordinated by University of North Carolina, USA, and supported AASLD), and the COVID-Hep.net registry (co-ordinated by University of Oxford, UK, and supported by EASL). All three registries were widely advertised through the communication channels of multiple endorsing gastroenterology and hepatology societies, direct emails to hepatology providers, and through social media. Submitting clinicians were asked to complete a case report form of clinical data at the end of their patient's disease course, defined as resolution of clinical signs of COVID-19, discharge from hospital, or death.
All three registries used an online reporting form which was identical for COVID-Hep and SECURE-cirrhosis, but different for the R-LIVER COVID-19 registry; copies of both data collection tools are available as a Supplementary annex. In order to centralise and amalgamate overlapping report form information, case data from the R-LIVER report form was re-entered onto the online COVID-Hep report form via www.COVID-Hep.net.
Whereas the R-LIVER COVID-19 registry collected data exclusively for autoimmune liver disease and other rare liver diseases, COVID-Hep.net and SECURE-Cirrhosis registries also simultaneously collected identical data for patients with laboratory confirmed SARS-CoV-2 and all other aetiologies of chronic liver disease (CLD). The current study includes 745 cases of CLD and SARS-CoV-2 infection, including 42 patients with AIH, which are included in a recently J o u r n a l P r e -p r o o f published analysis from the COVID-Hep/SECURE-cirrhosis registry. However, this prior publication did not include any analysis on AIH patients. In addition, two recent publications from Italy partially contained descriptive data from 4 patients being included in the registries of this study [16,17]. Contributing centres of ERN RARE-LIVER were asked to report the total number of AIH patients of their institution and to report monthly even if no cases of SARS-CoV-2 were identified in their AIH patients.
To provide a comparison cohort of patients without CLD, data were also extracted using the SECURE-cirrhosis/COVID-Hep data collection tool from electronic patient records of consecutive patients testing positive for SARS-CoV-2 over the same time period at Oxford University Hospitals NHS Foundation Trust (OUHFT), an organization of four hospitals in and around Oxford in the UK, and from the University of North Carolina Hospitals (UNCH). Positive cases from OUHFT and UNCH were defined as detection of SARS-CoV-2 by reverse transcriptase polymerase chain reaction (RT-PCR) on nasopharyngeal swabs. Any cases of SARS-CoV-2 infection in patients with pre-existing chronic liver disease who were identified from the electronic health records whilst compiling the non-CLD cohort were subsequently incorporated in the CLD cohort. To minimize potential reporting bias, data extraction for the non-CLD cohort was performed by investigators blinded to the clinical characteristics and outcomes reported in CLD patients. All data for both CLD and non-CLD cohorts were uploaded real-time to the same secure, online, data capture tool.
All submitted report forms for all cohorts were manually reviewed to assess for data quality, completeness and inconsistencies and in some instances, submitting clinicians were contacted and asked to provide additional data where appropriate. When combining the R-LIVER COVID- patients with AIH and coexisting liver disease (e.g. AIH with alcohol-related liver disease) were excluded from the analysis. This was due to internationally varying diagnostic criteria and difficulty defining the predominant liver disease phenotype of variant syndromes from the data capture tools. Characteristics and outcomes of patients with variant syndromes and AIH with coexisting liver disease are presented in Supplementary table 1.

Variables and definitions
Liver disease stage and aetiology was classified by the reporting clinician. Those with cirrhosis were then further sub-categorized by the reporting clinician according to Child-Turcotte-Pugh class (CTP-A, CTP-B, CTP-C). Throughout this paper the following terminology will be used to define the groups; total CLD cohort (CLD), autoimmune hepatitis cohort excluding variant syndromes and IgG4-related disease (AIH), CLD without AIH (non-AIH CLD), and patients without liver disease (non-CLD).
Obesity was defined as a body mass index (BMI) of >30 kg/m 2 ; where data on BMI was unavailable obesity was assumed to be absent. For analysis of ethnicity, only white ethnicity (as the majority classification) as compared to other ethnicities was considered in analysis. For the non-CLD cohort, where ethnicity was not recorded, white ethnicity was assumed [18]. To evaluate the effect of the AIH on the COVID-19 disease course we compared major outcomes in patients with AIH to both non-AIH CLD and non-CLD cohorts using propensity score 1:2 matched samples via a nearest neighbour approach (individuals matched according to similar propensity score) [19]. Covariables included in the propensity score model were selected based on their known associations with severe COVID-19 [2,20], whilst aiming to provide matched variance ratios of between 0.5-2.0 . In AIH versus non-AIH CLD the variables included were age in years, interactions with age, sex, and baseline liver disease severity (CLD without cirrhosis, CTP-A, CTP-B, CTP-C). Covariables included in the propensity score matched model of AIH versus non-CLD were age, interactions with age, sex, hypertension, COPD, heart disease and diabetes. Propensity score matched analysis was performed using the teffects function in Stata. The average treatment effect on the treated (ATET) was calculated with robust Abadie-Imbens standard errors [19]. All statistical analyses were conducted using Stata v15.1 (College Station, TX). Similar statistical methods have recently been used to evaluate outcomes in the CLD and liver transplantation cohort from the COVID-Hep and SECURE-Cirrhosis registries [3,7].  Table 1). Of note, two paediatric AIH patients were submitted to the registry but were excluded from the analysis; one female aged 5-years and one male aged 10-years, both of which had a mild disease course and were not hospitalized. A total of 862 non-AIH CLD were included ( Table 2 with PBC (2%). Rates of major co-morbidities are presented in Table 2; the number of patients in the CLD cohort with unknown BMI who were assumed to be non-obese for the analysis was 116/932 (12%).
Period prevalence rates of SARS-CoV-2 in AIH patients

Non-CLD cohort
Within the same time period as the CLD and AIH cohort, data were collected using an identical case report form for 793 consecutive non-CLD patients of which a total of 769 cases remained after exclusions (OUHFT 614; UNCH 155) (Supplementary figure 1). The non-CLD cohort differed significantly from the AIH cohort with regards to age, sex, ethnicity, smoking status, baseline serum creatinine and rates of comorbidities including diabetes mellitus, hypertension, obesity and renal function (Supplementary table 4). The number of patients in the non-CLD cohort with unknown BMI who were assumed to be non-obese for the analysis was 158/769 (20%). The non-CLD cohort presented here includes 614 patients from OUHFT included in a previously published analysis comparing COVID-19 outcomes in liver transplant recipients [7].

Outcomes
Hospitalization, intensive care unit admission and death in AIH cohort When comparing AIH with non-AIH CLD there were no significant differences in the rates of all major outcomes including hospitalization (76% vs 85%; p= 0.060), ICU requirement (33% vs. 31%; p=0.788), ICU admission (29% vs. 23%; p=0.240), new requirement for renal replacement therapy (6% vs. 4%; p=0.522), invasive ventilation (13% vs. 17%; p=0.504) and death (23% vs. 20%; p=0.643) (Fig. 2A). The discrepancy between the rates of ICU requirement and ICU admission are accounted for by a proportion of severe cases being deemed inappropriate for ICU admission or due to lack of ICU availability. Furthermore, rates of mortality did not differ between AIH and non-AIH CLD when stratified by baseline CTP class (Fig. 2B). Within the 16 AIH patients who died, the major causes of death were COVID-19 lung disease 9 (56%), liverrelated 5 (31%) and cardiac-related 2 (13%) and these did not differ significantly from those dying in the non-AIH CLD cohort. A summary of baseline characteristics and rates of major  (Table 1).

Data included in multivariable analysis was available for all patients in all categories.
Associations between AIH and mortality within total CLD cohort  (Table 2).
Specifically, a diagnosis of AIH was not associated with mortality. In addition, when a separate analysis was repeated to include only AIH patients who were immunosuppressed as a variable, there remained no significant association with death (OR 1.27; 95%CI 0.49-3.34; p=0.623). The total CLD cohort presented here includes 745 patients included in a previously published analysis, however this prior analysis did not consider AIH as a variable in any logistic regression models [3].
Propensity score matched analysis comparing AIH with non-AIH CLD To further assess for an association between an AIH and major outcomes, a propensity score matched model was constructed including the variables age, interactions with age, sex, and baseline liver disease severity (CLD without cirrhosis, CTP-A, CTP-B, CTP-C) in order to compare AIH with non-AIH CLD. Using this model, AIH conferred no additional risk compared with non-AIH CLD patients across all major outcomes including hospitalization, ICU admission, and death (Fig. 3A). Patient characteristics for the non-AIH CLD cohort after propensity score matching are presented in Supplementary figure 6.
Propensity score matched analysis comparing AIH with non-CLD cohort.
A propensity score matched analysis was also performed comparing rates of major outcomes for AIH versus the non-CLD cohort derived from OUHFT (UK) and UNCH (USA) during the same time period. Variables selected for propensity score matching included age, interactions with age, sex, hypertension, COPD, heart disease and diabetes. This demonstrated a significant increased risk of hospitalization for AIH patients (+18.4% 95%CI 5.6-31.2%; p=0.005), but no J o u r n a l P r e -p r o o f increased risk of all other major outcomes including ICU admission, and death (Fig. 3B). Patient characteristics for the non-CLD cohort after propensity score matching are presented in Within our combined registry dataset, multivariable analysis of 70 AIH patients showed that only advancing age and baseline decompensated cirrhosis were independent risk factors for death.
Importantly there was no association between mortality and the use of immunosuppression.
Furthermore, we detected no association between AIH and death in a multivariable analysis of CLD patients, and demonstrated comparable outcomes between AIH and propensity score matched cohorts without liver disease, and with CLD of other aetiology. The rates of major outcomes and lack of an association between immunosuppression and death in those with AIH is consistent with similar findings in the liver transplant population [7,8]. The reasons for this are currently unknown, but presumably local or systemic immunity against SARS-CoV2 infection is Up until this work, only small case series of patients with AIH and COVID-19 have been published. In a report from across several regions in Italy, 10 AIH patients on immunosuppressive treatment showed a clinical course of COVID-19 comparable to that of nonimmunosuppressed patient [16]. Furthermore, telephone-based surveys in Northern Italy did not detect an increased COVID-19 mortality in patients with AIH [21]. These publications were important early signals for patients and their clinicians, but interpretations have been limited by very small sample sizes and localisation to one geographical area. This has led to difficulties in formulating clear recommendations on the approach to social distancing and use of immunosuppressive medication during the pandemic, which has by extension caused significant anxiety and uncertainty for patients. Although our data shows that AIH patients are not more susceptible to death, many of these patients are currently completely avoiding social contact or "shielding", which may reduce exposure to high SARS-CoV-2 viral load infections associated with worse clinical outcomes. It is now critical to vaccinate these patients against SARS-CoV-2 infection, and to monitor their response to immunisation which may be attenuated by immunosuppressive therapy.
Of note, we did observe greater rates of hospitalisation in AIH patients compared to a matched cohort without liver disease. This may relate to AIH patients being generally more likely to be hospitalized compared to patients without liver disease and therefore subject to higher rates of routine inpatient SARS-CoV-2 testing. In addition, it may reflect a background level of patient and clinician anxiety regarding the uncertain impact of immunosuppression on the COVID-19 disease course leading to a lower threshold for hospital admission. These anxieties can also be inferred from the fact that azathioprine maintenance therapy was discontinued in 25% of AIH J o u r n a l P r e -p r o o f patients following SARS-CoV-2 infection. However, we do not have follow-up clinical data to assess the secondary impact of these medication changes on liver disease activity, including the rates of AIH flares.
The strengths of the current study include the international nature of case submissions which gives a truly global perspective on the impact of SARS-CoV-2 infection in patients with AIH.
Clinician reporting also minimises the risk of misclassification although we accept that centralised diagnostic criteria for cirrhosis are lacking and that assessment of Childs-Turcotte-Pugh class includes subjective components. An additional strength is the comparison of AIH cases with a matched group of contemporaneous patients without liver disease from the UK and USA (the two largest contributors of AIH cases), which strengthens the argument that these patients are not at higher risk of adverse COVID-19 outcomes. However, our findings must be interpreted in the context of the study's potential limitations. Despite being the largest collection of AIH patients with SARS-CoV-2 to date, the total numbers remain relatively small thus constraining the number of variables included in propensity score matching and limiting interpretations with respect to variables associated with death within the AIH cohort. Secondly, registry data is vulnerable to reporting bias, leading to over-representation of patients with more severe liver disease and more severe COVID-19, and it is noteworthy that a majority of case submissions were derived from tertiary care centres (Supplementary tables 7 and 8). The high rates of cirrhosis in AIH patients (54%) found in the current study may be an indication of this reporting bias given that previous studies have described a cirrhosis rate of 20-30% in those with AIH [22,23]. However, despite the inclusion of patients with more severe baseline liver disease, we were still unable to identify an increased risk of severe COVID-19 and death in the AIH cohort suggesting that the true risk is likely to be even lower than reported here. It is also notable that despite the study having international reach, the AIH cohort is predominantly derived from western populations which may limit external generalisability of the results to other J o u r n a l P r e -p r o o f geographical areas. Similarly, the classification of what constitutes an 'intensive care unit' and the parameters used to determine the clinical requirement for ICU are likely to vary between international institutions. Lastly, although a time-dependent analysis (e.g. Cox regression and Kaplan-Meier curves) may have been preferable over logistic regression in evaluating mortality, this approach was not possible due to the lack of accurate data on the duration from positive laboratory COVID-19 diagnosis to death in the COVID-Hep/SECURE cirrhosis registries.
In summary, this study involving more than 1700 patients, helps characterize the COVID-19 disease course and risk of adverse outcome in 70 patients with AIH. Through multiple comparisons with non-AIH CLD and non-CLD cohorts, we demonstrate that AIH does not confer major additional susceptibility to adverse outcomes following SARS-COV-2 infection despite the potential reporting of cases with more severe liver disease. In this patient group, age and baseline liver disease severity remain the most important determinants of outcome in contrast to the use of immunosuppression were no negative impact was detected. This should provide some reassurance to patients and clinicians and lends weight to recommendations that immunosuppressive medication should not routinely be modified or discontinued during the course of COVID-19.
J o u r n a l P r e -p r o o f      Plot shows propensity-score matched analyses for major outcomes following SARS-CoV-2 infection for AIH compared to non-AIH CLD.
Variables selected for propensity score matching were age, interactions with age, sex, hypertension, COPD, heart disease and diabetes.  • There were no differences in rates of major adverse COVID-19 outcomes including hospitalization, intensive care unit (ICU) admission, and death between AIH patients and those with other aetiologies of liver disease.
• When compared to patients without liver disease in propensity score matched analysis, patients with AIH had higher rates of hospitalization but no increased risk of ICU admission or death despite potential reporting of AIH cases with more severe baseline liver disease.
• Independent risk factors for death in AIH patients were age and baseline liver disease severity, but not the use of immunosuppression.