Noninvasive approach to indicate risk factors of nonalcoholic steatohepatitis overlapping autoimmune hepatitis based on peripheral lymphocyte pattern

From 2016 to 2019, percutaneous liver biopsy was performed on patients with clinically suspected NASH and/or AIH at the University of Tokyo Hospital. Among them, 115 patients provided written informed consent, and PBMCs were obtained from 88 patients (NASH:nonalcoholic fatty liver [NAFL] = 70:18) pathologically diagnosed as NAFLD, and from 27 patients diagnosed with AIH (acute:chronic = 12:15), not overlapping with NASH (Supplementary Fig. 1). The 70 NASH patients were divided into three groups: ANA negative, 36; ANA positive, 20; and AIH-overlap, 14. The patients met the following criteria: the levels of serum aspartate aminotransferase (AST) or alanine aminotransferase (ALT) above the upper limit of normal for ≥ 6 months; alcohol consumption < 30 g/day for men and < 20 g/day for women [8]; seronegativity for hepatitis B virus surface antigen and hepatitis C virus antibody and the absence of other liver diseases, such as primary biliary cholangitis (PBC), primary sclerosing cholangitis, and drug-induced liver injury (DILI); and no history of autoimmune disease except AIH, chronic inflammatory disease except hepatitis, and malignant tumor, because these comorbidities may affect peripheral lymphocyte patterns. Patients with NAFLD also met the criterion of enhanced hepatorenal contrast by abdominal ultrasonography (AUS). Although there were no specified guidelines or consensus on the diagnosis of AIH-overlap NASH [4, 7], in this study, patients suspected of overlapping AIH met at least one of the following criteria: serum ALT level 1.5 times higher than 6 months ago; ANA > 1:40; serum immunoglobulin (Ig) G level > upper limit of normal within 6 months [23]. Patients suspected of acute or chronic AIH alone also met at least one of the above criteria, and acute AIH was defined by the presence of acute-onset symptoms (e.g., jaundice, fatigue, right upper quadrant pain, and/or anorexia) for less than 6 months in combination with a total bilirubin level of > 5 mg/dL and/or a serum ALT level of greater than 10 times higher than the upper limit of normal [24]. The exclusion criteria were symptoms of viral or bacterial infection, malignant tumor or chronic inflammatory disease, and immunosuppressant use. Eighteen subjects who did not undergo liver biopsy were included as controls in the non-NASH group and met all of the following criteria: normal values of serum AST and ALT; no history of autoimmune disease, chronic inflammatory disease, or malignant tumor; and absence of NAFLD according to AUS. In total, 133 PBMC samples were separated from blood using a Ficoll Kit [12]. Regarding AIH-overlap NASH, PBMCs were obtained from 9 of 14 patients again after 1 year; these 9 patients, who could be followed up on steroid treatment in this study, underwent biochemical remission, compared before and after the treatment. Biochemical remission was defined as the complete normalization of aminotransferases and/or IgG [25]; all these 9 patients were free of steroid treatment after 1 year. This study was approved by the Ethics Committees of The University of Tokyo and was performed in accordance with the Declarations of Helsinki.

Comorbid diseases (immunological conditions in particular) and drug intake were recorded, and body mass index (BMI) was examined on the day of liver biopsy. Blood samples were obtained for biochemical tests for albumin, AST, ALT, γ-glutamyl transpeptidase (GGT), total bilirubin, high-density lipoprotein (HDL-C), low-density lipoprotein (LDL-C), triglycerides, creatinine, C-reactive protein (CRP), number of platelets, prothrombin time, fasting blood glucose (FBG), hemoglobin A1c (HbA1c), ANA, anti-mitochondrial antibody (AMA) -M2, IgA, IgG, IgM, and hyaluronic acid. ANA, AMA-M2, IgA, IgG, and IgM are serological hallmarks for diagnosis of autoimmune liver disease (AILD), such as AIH and PBC [26, 27]. Liver stiffness and controlled attenuation parameter (CAP) were also measured to assess fibrosis and steatosis, respectively, with transient liver elastography performed by FibroScan (FS) [14]. In control subjects clinically not suspected to have NAFLD and AIH, the ANA, AMA, IgA, IgG, IgM, hyaluronic acid, and liver stiffness levels and CAP values were not determined.

Liver biopsies were examined by trained, experienced hepatopathologists blinded to the clinical data and study design. Significant hepatic steatosis was considered present when ≥ 5% of hepatocytes exhibited fatty changes, and NASH liver histology was assessed according to the Matteoni classification [26, 28]. NASH progression was assessed by determining the NAFLD activity score (NAS) based on the degrees of steatosis, lobular inflammation (LI), and hepatocellular ballooning [29]. Pathological differences in NAS and fibrosis were assessed in patients with NAFLD. As diagnostic criteria of AIH, the revised International Autoimmune Hepatitis Group (IAIHG) scoring system was adopted [30], and especially, regarding liver histology, the cases with all pathological findings of interface hepatitis, lymphocytic infiltration, plasma cell infiltration, rosette formation, and emperipolesis were adopted in this study [31, 32]. Only 41 patients diagnosed with definite AIH (total IAIHG score ≥ 15) were enrolled, and all underwent pathological diagnosis by liver biopsy. Among these 41 patients, 14 with histological features of both NASH and AIH were defined as AIH-overlap NASH, although there were no established diagnostic criteria for AIH-overlap NASH [4, 7, 23]. NASH patients without distinct histological features of AIH were excluded from diagnosis of AIH-overlap, because severe portal inflammatory infiltrate in AIH-overlap NASH may obscure these features [33]. The NAFL patients had no distinct histological features of AIH, and patients not diagnosed with NAFLD and/or AIH were excluded.

PBMCs were processed as shown in the Supplementary Methods for surface staining of human peripheral lymphocytes based on two 11-color antibody cocktails (Supplementary Table 1). Subsequently, they were processed for FCM using fluorescence-activated cell sorting (FACS) Aria (BD Biosciences, San Jose, CA). FCM data were exported and analyzed using FlowJo version 10.8.1 (Treestar). FCM plots and representative gating strategies were used to identify lymphocytes as shown in the Supplementary Methods (Fig. 1). CD45-based gating was used to assess lymphocyte frequencies, because CD45 is a surface marker for total leukocytes [34]. CD45⁺ leukocytes were selected, followed by CD3⁺ (whole T cells), CD4⁺ CD8⁻ (helper T cells, Th), CD4⁻ CD8⁺ (cytotoxic T cells), CD3⁻ CD19⁺ (B cells), CD3⁻ CD56⁺ (NK cells), and CD3⁺ CD56⁺ (NKT cells) as shown in Fig. 1A and using Cocktail I, as summarized in Supplementary Table 1 [34‐36].

Th subsets were examined as contributors to the pathologies of autoimmune and inflammatory diseases [37]. Among Th subsets, CD25⁻ C–X–C chemokine receptor (CXCR) 3⁺ C–C chemokine receptor (CCR) 4⁻ CCR6⁻ Th1, CD25⁻ CXCR3⁻ CCR4⁻ CCR6⁺ Th2, CD25⁻ CXCR3⁻ CCR4⁺ CCR6⁺ Th17, and CD25⁺ CD127⁻ regulatory T cells (Tregs) were selected using the gating strategies in Fig. 1B and Cocktail II (Supplementary Table 1) [36, 38]. Tregs maintain immune homeostasis and prevent autoimmune diseases; i.e., they regulate immune tolerance [39]. Programmed cell death (PD) 1 and cytotoxic T-lymphocyte antigen (CTLA) 4 surface markers on CD4⁺ and CD8⁺ T cells have been implicated in oncology and autoimmune diseases [40, 41]. Indeed, PD1 expression can cause life-threatening autoimmune diseases by disrupting T-cell homeostasis [19, 20, 40]. CD4⁺ PD1⁺, CD4⁺ CTLA4⁺, CD8⁺ PD1⁺, and CD8⁺ CTLA4⁺ T cells were selected using the gating strategies described in Fig. 1C and Cocktail II (Supplementary Table 1).

CD45 can be used to calculate the frequencies of T cells, B cells, NK cells, and NKT cells; CD4 can be used to calculate those of Th subsets. Based on PD1 and CTLA4 expression, T lymphocyte frequencies were calculated by CD4- and CD8-based gating. Target lymphocytes in 133 PBMC samples were selected using the gating strategies shown in Fig. 1, and lymphocyte frequencies were calculated.

Statistical analysis is described in the Supplementary Methods.

By liver biopsy, 88 of 115 patients were diagnosed with NAFLD based on histologic findings; 18 subjects were examined as controls. Therefore, 106 subjects (NASH, 70; NAFL, 18; controls, 18) were enrolled; their baseline characteristics are listed in Table 1. There were no significant differences in age, sex ratio, or BMI between NASH and NAFL patients. Regarding laboratory parameters, AST, ALT, ANA, IgG, and hyaluronic acid were significantly increased in NASH compared to NAFL patients. Regarding liver pathological findings, the degrees of LI, ballooning, and liver fibrosis were significantly higher in NASH than NAFL patients. FibroScan^® (FS; Echosens™, Paris, France) values were also significantly higher in NASH patients. In NAFLD patients, BMI and the levels of AST, ALT, GGT, total bilirubin, HDL-C, triglycerides, and HbA1c were significantly increased compared to controls.

Several significant differences in lymphocyte frequencies were detected between NASH and NAFL (Fig. 2A–C). NASH patients had a significantly higher percentage of NKT cells than NAFL patients. They also had significantly higher percentages of Th2 and Th17 and a significantly lower percentage of CD8⁺ PD1⁺ T cells. Compared to controls, NAFLD patients had significantly higher percentages of CD3⁺ and CD4⁺ T cells and B cells, and a significantly lower percentage of Treg and CD4⁺ PD1⁺ T cells. These results implicate hepatic steatosis in the increased populations of peripheral T and B cells and the suppression of immune tolerance.

In univariate analyses, the levels of AST, ALT, ANA, IgG, and hyaluronic acid; liver stiffness value; and the frequencies of NKT cells, Th2, Th17, and CD8⁺ PD1⁺ T cells were significant determinants of NASH (Table 2). Multivariate analysis of these laboratory parameters and lymphocyte frequencies except ANA, IgG, and NKT cells suggested that the hyaluronic acid level, liver stiffness value, and the frequencies of Th17 and CD8⁺ PD1⁺ T cells are independent risk factors of NASH.

Seventy NASH patients (ANA negative, 36; positive, 20; AIH-overlap, 14) were enrolled; their baseline characteristics are listed in Table 3. According to the International Autoimmune Hepatitis Group (IAIHG) scores, 30–40% of ANA-negative NASH had probable AIH, all ANA-positive NASH patients had probable AIH, and all AIH-overlap NASH patients had definite AIH. Notably, two AIH-overlap NASH patients were ANA-negative. Therefore, irrespective of the presence of ANA, several NASH cases may show overlap with AIH. There were no significant differences in age and BMI between the patients with and without AIH-overlap NASH. Regarding laboratory parameters, the levels of AMA and IgM were significantly increased in AIH-overlap NASH patients. Regarding liver pathological findings, the degree of steatosis was significantly lower in AIH-overlap NASH patients. Furthermore, in ANA-positive and AIH-overlap NASH patients, the levels of ANA, AMA, and IgG were significantly increased compared to those who were ANA-negative, whereas the male sex ratio was significantly lower. Therefore, females are more likely to develop autoimmunity-related NASH compared to males.

Several significant differences in lymphocyte frequencies were observed between patients with and without AIH-overlap NASH (Fig. 3A–C). AIH-overlap NASH patients had significantly higher percentages of CD4⁺ T cells and Th17, and significantly lower percentages of Th1, CD4⁺ and CD8⁺ PD1⁺ T cells. Compared to ANA-negative NASH patients, those who were ANA-positive with AIH-overlap showed a significantly higher percentage of Tregs, suggesting that autoimmunity-related NASH may induce peripheral Tregs.

Furthermore, the changes in peripheral lymphocyte frequencies of 70 NASH patients according to the degree of NAS, fibrosis, and the levels of AST, ALT, and IgG were examined to show the associations between these lymphocytes and AIH-overlap NASH (Supplementary Table 2). Regarding liver damage, the frequency of CD4⁺ T cells was positively correlated with the levels of AST and ALT, while that of CD8⁺ T cells was negatively correlated. Notably, regarding liver fibrosis, the frequency of CD8⁺ PD1⁺ T cells was positively correlated with the degree of fibrosis, while that of NKT cells was negatively correlated; that of CD8⁺ PD1⁺ T cells was also negatively correlated with the IgG level. These results suggest that differential peripheral lymphocyte subsets are deeply associated with NASH; among them, CD8⁺ PD1⁺ T cells may be associated with liver fibrosis and autoimmunity in NASH.

In univariate analyses, male sex; the levels of ANA, IgG, and IgM; CAP value; and frequencies of CD4⁺ T cells, NK cells, Th1, Th17, CD4⁺ and CD8⁺ PD1⁺ T cells were significant determinants of AIH-overlap NASH (Table 4). Multivariate analysis of these laboratory parameters and lymphocyte frequencies except male sex, CD4⁺ T cells, and NK cells suggested that the frequency of CD8⁺ PD1⁺ T cells is an independent risk factor of AIH-overlap NASH. Next, we examined the changes in several laboratory parameters and peripheral lymphocyte frequencies, including CD8⁺ PD1⁺ T cells, before and after steroid therapy for 9 AIH-overlap NASH patients (Supplementary Table 3). The decreased frequency of CD8⁺ PD1⁺ T cells was significantly increased by steroid treatment, which may increase the usefulness of the frequency of CD8⁺ PD1⁺ T cells as an indicator of AIH-overlap NASH.

Furthermore, the target lymphocyte frequencies in 27 PBMC samples of AIH patients (acute:chronic = 12:15) were evaluated. These frequencies and baseline characteristics of total 41 patients together with 14 AIH-overlap NASH patients are listed in Supplementary Table 4. AIH-overlap NASH patients had a significantly lower percentage of CD4⁺ and CD8⁺ PD1⁺ T cells compared to those with acute and chronic AIH; there were no significant differences in autoimmunity-related parameters (ANA, AMA-M2, IgA, IgM, and Tregs). These results suggest that PD1 expression differs between AIH-overlap NASH and AIH, although these two diseases share several autoimmunity-related parameters.