Primary sclerosing cholangitis with moderately elevated serum‐IgG4 – characterization and outcome of a distinct variant phenotype

Immunoglobulin G4‐associated cholangitis (IAC) is characterized by distinctly elevated immunoglobulin G4 in serum (sIgG4) and responds well to corticosteroid therapy. Primary Sclerosing Cholangitis (PSC) is a progressive liver disease without causal treatment options usually not responding to immunosuppression. Increased serum levels of sIgG4 in patients with PSC, that do not meet criteria of IAC, have been reported in 10%‐25%. Therefore, we aimed to characterize this subgroup of patients in a retrospective, multicenter study.


| INTRODUC TI ON
Primary sclerosing cholangitis (PSC) is a rare progressive liver disease frequently leading to biliary cirrhosis and is associated with an increased risk of cholangiocarcinoma (CCA). PSC is usually categorized as autoimmune liver disease alongside autoimmune hepatitis, primary biliary cirrhosis 1 and Immunoglobulin (Ig)-G4-associated cholangitis (IAC). 2,3 Pathophysiology of PSC is widely unknown resulting in limited causal therapeutic options. Several pathomechanisms have been thoroughly investigated proposing a multifactorial etiology for PSC 4,5 reflecting the interface of microbiological, 6,7 genetic 8,9 and immunological dysregulations. [10][11][12] The clinical course of this cholestatic liver disease can be extremely variable. Modifiable risk factors accelerating disease progression are not sufficiently identified yet. Useful laboratory parameters for monitoring disease severity are cholestasis parameters like alkaline phosphatase. 13,14 In the initial diagnostic work-up of PSC, autoantibodies like antimitochondrial antibodies, anti-nuclear antibodies, smooth muscle antibodies, soluble-liver-antigen antibodies, liver kidney microsomal antigen antibodies, anti-neutrophil cytoplasmic antibodies and serum-IgG4 should be determined in addition to distinguish PSC from the other autoimmune liver diseases. Differentiating between those entities can pose a significant clinical challenge. Especially IAC can mimic PSC phenotypically and clinically rendering correct diagnosis difficult in clinical practice 15,16 ; but prognosis and therapeutical strategies of the two diseases differ profoundly. IAC as a hepatobiliary manifestation of IgG4-associated diseases usually presents with unambiguously increased IgG4-serum concentrations and responds well to corticosteroid therapy. 17 IgG4-associated cholangitis is a separate disease entity within the known cholangiopathies, characterized by typical findings according to HISORt or Japanese criteria. 15 Patients suffering from PSC typically show strong associations with inflammatory bowel disease (IBD), are of younger age and are highly susceptible for CCA and colorectal carcinomas and do not respond to corticosteroids. 18 Whereas Jaundice, abdominal pain and autoimmune pancreatitis or other IgG4-associated diseases occur more often in IAC. Patients with AIC seem to have a much lower risk to develop CCA, when compared with PSC. 17,19,20 Prognosis of IAC is therefore better as compared to PSC. Incidental findings of elevated serum-IgG4 (sIgG4) in PSC patients have been reported 16,21,22 and could influence transplant-free survival. 21 An IgG4-PSC subgroup has therefore been suggested. 21 Standard recommended therapeutic strategies for these patients do not exist and the significance of treatment with corticosteroids remains elusive. 23 We have therefore conducted a large multicenter retrospective clinical study with 289 PSC patients from three German university hospitals with a long-term follow-up to analyse the role of serum-IgG4 in the disease course of PSC patients and its correlation with specific clinical phenotypes.  The actuarial survival free of liver transplantation rate was estimated by Kaplan-Meier survival analysis. The differences between the actuarial estimates were analysed using the log-rank test. All tests were two-tailed and a P-value of .05 or less was considered statistically significant.

| Diagnostic criteria
All patients were diagnosed with PSC in accordance with the current European or German guidelines, 3

| IgG4-Measurement
Serum-IgG4-values were available for all included patients. Firstly, routinely documented IgG4-measurements were chosen for our analyses. Cases with first IgG4 value determined after liver transplantation (LT) or more than 6 months previous to or 15 years after the date of first diagnosis of PSC were excluded. In a subset of 27 patients subsequent IgG4-measurements were available.
In the Hannover cohort, IgG4 was determined via immunoturbidimetry using assays by Siemens Healthcare Diagnostics. The normal value range for these patients was set from 0.08 to 1.4 g/L according to the laboratory's specifications. The assays used in the University hospitals Essen and Bonn based on immunonephelometry by Siemens Healthcare Diagnostics to determine IgG4 in serum with normal values ranging from 0.03 to 2.01 g/L. After 2017 diagnostics in Bonn were changed to turbidimetric immunoassays from Binding Site with lower normal range values from 39 to 864 mg/L.
The patient-subgroup with elevated IgG4 (PSC-IgG4) was defined as that the first serum-IgG4 exceeded the upper limit of the normal range (ULN) according to the respective assay in these patients. Accordingly, the group of patients with normal IgG4-values is referred to as PSC-N. Since unit of measurement as well as normal range values differed between assay and centres the ratio of first measured Ig4 in relation to ULN was determined for non-parametric comparison.

| Subgroup of PSC patients with elevated IgG4serum-concentrations
The mean ratio of IgG4 in relation to the ULN in the entire cohort was 36%. Increased IgG4 serum-levels were observed in 14.5% of PSC patients (PSC-IgG4). Mean IgG4-level was 132% of the ULN in these patients. (see Figure 1). Sequential measurements of IgG4 at different time points during disease course were available in 27 patients (median: two measurements). IgG4 remained stable in 25 patients.
One patient with IgG4 within normal range in the first measure-  (Table S1).
The histogram of IgG4 distribution ( Figure 1) shows two separate peaks around the median of 0.3 for the PSC-N group and 1.32, respectively, reflecting the PSC-IgG4 group. The nadir in between peaks is set at a value of 0.9. When using 0.9 of the ULN instead of the ULN as cut-off value 15.9% of the patients in our cohort were identified as IgG4 (0.9) -PSC patients.
Distribution of age, gender, presence of an IBD (see Table 1) or IBD subgroup (CD, CU, Indeterminate colitis) (see Table 1) as well as small duct-phenotype were comparable between PSC-IgG4 and PSC-N groups (see Table 1). There were significantly less patients with IgG4-values above the ULN in the subgroup of patients with an AIH-phenotype (see Table 1).
Clinical data concerning concomitant medication were available for 100 patients of the Bonn and Hannover cohort. Corticosteroids were administered in 6.6% of all patients. Patients with PSC/AIHphenotype received systemic steroid therapy (75%) at time of first IgG4 measurement significantly more often than PSC patients without AIH-phenotype (11.4%) (P < .01). In cases of concomitant IBD 21.9% were treated with corticosteroids at time of first IgG4measurement. Patient treated with corticosteroids at time of IgG4 measurement, regardless of the indication for steroid therapy, had a significantly lower IgG4 ratio (P = .033).
AST, Bilirubin and alkaline phosphatase were significantly higher, and albumin was significantly lower in IgG4-PSC group (see Table 1).   Similarly, results regarding prognostic scores within the subgroup of patients with IgG4 measurement within a more limited time-frame showed significantly higher scores in PSC-IgG4 (see Table S1).

| Elevated IgG4 is associated with inferior transplant-free survival
To analyse the impact of different IgG4-levels on the combined endpoint liver transplantation or death, Kaplan-Meier estimates of transplant-free survival were compared by log-rank test. Patients with IgG4-serum values above the respective ULN had a significant inferior transplant-free survival compared to PSC patients with normal IgG4levels (P = .001) (see Figure 3). Results were still significant when excluding patients with IgG4-measurement later than 2 years after PSC diagnosis (see Table S1). 17 Table 2). AIH-phenotype, small-duct PSC, concomitant IBD or dominant stricture (P = .1) did not have an impact on transplant-free survival rates (see Table 2). A trend towards inferior transplant-free survival for patients with extrahepatic dominant strictures could be observed (P = .078). Steroid administration at time of first IgG4-measurement did not influence transplant-free survival rates, neither did steroid therapy during any time in course of disease (see Table 2).

| D ISCUSS I ON
Elevated Serum-IgG4 has been described in patients with PSC before. 21 Here, we report in a large and representative multicenter cohort of 289 patients from three German liver transplantation centres IgG4-levels above the normal range in 14.5% of the PSC patients (PSC-IgG4). IgG4-values were only moderately elevated (median at 1.3-fold) in our cohort as opposed to highly increased IgG4 that is typical for IAC. 30 Accordingly, the distribution of IgG4concentrations in the overall cohort displayed two peak concentrations. The separating line between those two peaks was determined at 0.9-fold of the ULN, thus suggesting that a slightly lower cut-off value might be more suitable to distinguish the subgroup of PSC-IgG4 patients from classical PSC patients.
IBD is typically associated with PSC, and also a risk factor for disease progression or recurrence after transplantation. 18,31 It was therefore noteworthy that we could not confirm previous observations on a reduced prevalence of IBD in PSC-IgG4 compared to classical PSC. 21,22 Navaneethan et al. did also not observe higher incidence of IBD but different UC patterns and higher inflammatory bowel disease activity, resulting in shorter colectomy-free time. 32 Unfortunately, in our cohort activity of IBD was not assessed.
We observed a significant lower frequency of elevated IgG4levels in patients with the autoimmune hepatitis variant phenotype of PSC. Recent reports showed that IgG4-expression can be associated with AIH as well, 33,34 leading to speculations whether AIH might be the hepatic manifestation of IgG4-related diseases. 34 Pathophysiologically, it therefore seemed unlikely for PSC/AIHphenotype to function as a protective factor for IgG4-elevation.
When further investigating this subgroup, it was apparent that patients with PSC/AIH-phenotype received systemic steroid therapy at the time of IgG4-measurement significantly more often (75%) than PSC patients without concomitant AIH (11.4%). On the other hand, only one patient treated with steroids had increased IgG4-values. It is well-known that in IgG4-related diseases 35 and moreover in PSC with high IgG4 36 systemic corticosteroids can effectively reduce serum-IgG4-levels and induce a treatment response. 37 We therefore hypothesize that systemically administered budesonide in patients with AIH is the main reason for the observed lower IgG4-levels in  40 We therefore assume that the impact of IgG4 on transplant-free survival is somewhat exaggerated in these highrisk cohorts, while in population-based studies it only manifests statistically in much larger cohorts.
Our data did not show any evidence that malignancy could be the main reason of shorter transplant-free survival in PSC-IgG4, as the incidence of neither colorectal nor hepatobiliary malignancies was significantly associated with increased IgG4-levels. According to previous reports, IAC is not strongly associated with a higher risk for CCA 39 although two studies suggested that patients with IgG4 related disease might have a higher risk to develop pancreaticobiliary malignancies. 19,20 To further investigate inferior survival rates in this subgroup we analysed the baseline biochemical parameters and found that PSC-IgG4 patients presented with significantly higher levels for bilirubin and ALP. These findings are mostly in line with the previous studies. 21