Tumor necrosis factor alpha-induced protein 3: A key biomarker for response to ursodeoxycholic acid in primary biliary cholangitis

Abstract

BACKGROUND

The pathogenesis of primary biliary cholangitis (PBC) remains unclear. Ursodeoxycholic acid (UDCA) is the only first-line clinical treatment, but approximately 40% of patients exhibit a poor response.

AIM

To identify novel biomarkers for PBC to predict the efficacy of UDCA and enhance treatment.

METHODS

Microarray expression profiling datasets were downloaded from the Gene Expression Omnibus and analyzed to identify differentially expressed genes between PBC patients and healthy controls. Immunohistochemistry was performed to validate key genes in liver tissues of the participants. Logistic regression was employed to evaluate prognostic risk factors, receiver operating characteristic curves were used to assess predictive performance, and correlations between key genes and clinicopathological characteristics were analyzed.

RESULTS

By bioinformatic analysis, 13 genes primarily associated with the progression of PBC were identified, and tumor necrosis factor alpha-induced protein 3 (TNFAIP3) was selected for further investigation. Then expression of TNFAIP3 in PBC patients was significantly elevated compared to healthy controls on immunohistochemistry (P < 0.0001). Multivariate Cox regression analysis indicated that both TNFAIP3 and fatigue were independent risk factors for response to UDCA in PBC patients (P < 0.05). The area under the curve for TNFAIP3 and fatigue were 0.691 and 0.704, respectively, while their combination showed a significantly higher area under the curve of 0.848. The expression of TNFAIP3 was also correlated with age, albumin, total bilirubin, alkaline phosphatase and splenomegaly (P < 0.05).

CONCLUSION

TNFAIP3 and fatigue are independent risk factors for response to UDCA in Chinese patients with PBC. TNFAIP3 may be a potential biomarker or therapeutic target for PBC. These findings offer new insights into the pathogenesis of PBC.

Key Words: Primary biliary cholangitis; Bioinformatics analysis; Tumor necrosis factor alpha-induced protein 3; Biomarkers; Predictor; Therapeutic target

Core Tip: By bioinformatics analysis, we identified 13 differentially expressed genes linked to primary biliary cholangitis progression. Tumor necrosis factor alpha-induced protein 3 (TNFAIP3) was selected for validation via immunohistochemistry and confirmed as a promising biomarker for ursodeoxycholic acid response. Clinical validation further proved its significant correlation with treatment outcomes. Additionally, fatigue was identified as an independent predictor of ursodeoxycholic acid efficacy, and the combination of TNFAIP3 expression and fatigue exhibited superior predictive performance. The expression of TNFAIP3 was also associated with key clinicopathological features, including age, albumin levels, total bilirubin, alkaline phosphatase, and splenomegaly, providing new insights into disease progression.

INTRODUCTION

Primary biliary cholangitis (PBC) is a rare autoimmune and cholestatic liver disease characterized by chronic inflammation and fibrotic destruction of interlobular bile ducts, leading to cholangitis, fibrosis, and eventually cirrhosis[1,2]. The etiology and specific pathogenesis of PBC is still unclear[3]. The global incidence of PBC was estimated at 1.76 per 100000 individuals per year, while the prevalence reaches approximately 14.6 per 100000, based on recent pooled data[4]. Of note, the median survival of symptomatic patients was only 7.5 years[5].

For almost two decades, ursodeoxycholic acid (UDCA) has remained the first-line treatment for PBC, helping to delay disease advancement and extend the duration before liver transplantation becomes necessary[6-8]. However, up to 40% of patients with PBC still have inadequate response to UDCA[9]. Recently, obeticholic acid has been indicated for the treatment of PBC in combination with UDCA in adults who have inadequate response to UDCA[10]. In addition, several studies have found the benefits of UDCA in combination with fibrates[11,12]. The latest guidelines recommend early initiation of second-line combination therapy for UDCA-unresponsive PBC. However, the current time to evaluate the efficacy of UDCA is between six months and two years, during which time the disease is likely to develop further due to the lack of effective early intervention[10]. Therefore, it is imperative to identify PBC patients with an incomplete response to UDCA as early as possible, promptly initiate second-line drug therapy, delay histological progression, and improve prognosis.

Microarray-based high-throughput bioinformatics approaches have been extensively applied to identify differentially expressed genes (DEGs) and elucidate key molecular pathways associated with PBC pathogenesis[13-15]. Leveraging PBC-related datasets from the Gene Expression Omnibus (GEO), it is possible to comprehensively investigate key genes linked to disease development. Accordingly, this study aimed to identify the key biomarkers in the progression of PBC, clarify the factors affecting the efficacy of UDCA, and enhance treatment strategies at an early stage.

MATERIALS AND METHODS

The workflow of this study is shown in Figure 1.

Figure 1 Patient selection and study design. GEO: Gene Expression Omnibus; HCs: Healthy controls; PBC: Primary biliary cholangitis; UDCA: Ursodeoxycholic acid; TNFAIP3: Tumor necrosis factor alpha-induced protein 3.

Study cohort

A total of 197 PBC patients and 71 healthy controls (HCs) were recruited from the Department of Rheumatology and Immunology at the Affiliated Hospital of Qingdao University between January 2018 and April 2023. The diagnosis of PBC was based on the following criteria: (1) Biochemical evidence of cholestasis determined by elevated alkaline phosphatase (ALP) in serum; (2) Presence of anti-mitochondrial antibodies (AMA) or other PBC-specific autoantibodies if AMA was negative, including anti-sp100 or anti-glycoprotein 210 antibodies (anti-gp210); and (3) Histological evidence of non-suppurative destructive cholangitis and interlobular bile duct destruction in liver biopsy. Patients who met at least two of the three criteria were included in the study[5]. The exclusion criteria included fatty liver, autoimmune hepatitis, hepatitis B virus infection, drug-induced liver injury, hepatocellular carcinoma, myelodysplastic syndrome, primary myelofibrosis and if percutaneous liver biopsy was not performed.

Liver tissue was obtained from 71 PBC patients via percutaneous liver biopsy, and normal liver tissues of 71 HCs were collected from age- and sex-matched non-tumor liver tissue adjacent to hemangiomas resected during surgery. The demographic and clinical details of each patient were obtained from electronic medical records, and clinical outcomes were obtained from outpatient and inpatient records or telephone follow-up. The follow-up commenced at the time of PBC diagnosis, with the study endpoint defined as 1-year post-UDCA treatment initiation. The 71 PBC patients were divided into UDCA responders and non-responders according to the Paris criteria[16,17]. All patients received standardized treatment with UDCA (13-15 mg/kg/day). The study was conducted in accordance with the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the Affiliated Hospital of Qingdao University, approval No. QYFY WZLL 29650, and written informed consent was obtained from all participants for sample collection and study participation.

Baseline data collection

Demographic and clinical characteristics of the study participants were collected, including gender, age, clinical symptoms (fatigue, pruritus), previous medical history, laboratory parameters [platelets, albumin, total bilirubin (TBIL), alanine aminotransferase, aspartate aminotransferase (AST), ALP, gamma-glutamyl transpeptidase, total bile acid, AMA, AMA-M2, anti-gp210, anti-sp100, anti-centromere antibodies and immunoglobulin M], imaging parameters (splenomegaly, osteoporosis), histological examination, and treatment status.

Histological analysis

Ultrasound-guided percutaneous liver biopsies were performed in 71 PBC patients. Liver tissue specimens (1-2 cm) were fixed in 4% formaldehyde solution (G1101, Servicebio, China), and embedded in paraffin. Serial sections were stained with hematoxylin and eosin (HE, G1076, Servicebio, China). Pathological staging of PBC was assessed using the Scheuer classification (stage I: Portal inflammation; stage II: Periportal inflammation; stage III: Bridging fibrosis; stage IV: Cirrhosis) by experienced liver pathologists[18]. In this study, PBC in Scheuer stages I−II was categorized as early-histologic-stage, while stages III-IV were categorized as advanced-histologic-stage.

Definition of biochemical response to UDCA

The biochemical response to UDCA treatment was defined using the Paris criteria. In early-histologic-stage PBC patients, responders were defined according to the Paris II criteria[16]: ALP and AST ≤ 1.5 × upper limit of normal (ULN), with a normal bilirubin level. The Paris I criteria[17] were applied to define responders in advanced-histologic-stage PBC: ALP < 3 ULN, or AST < 2 ULN, or bilirubin ≤ 1 mg/dL. All definitions were applied after 1 year of UDCA treatment.

Data included

Following a systematic review, three gene expression datasets (GSE93170, GSE119600 and GSE159676) were selected from the National Center for Biotechnology Information GEO database. All the datasets were microarray data. The microarray data from GSE93170 were based on the GPL14550 platform and consisted of 6 PBC patients and 6 HCs[13]. The GSE119600 database, utilizing the GPL10558 platform, included 78 PBC patients and 38 HCs[14]. The GSE159676 database of the GPL6244 platform included 3 PBC patients and 6 HCs[15]. All data were available online and free of charge. Batch correction was performed on the three datasets using the Limma and surrogate variable analysis packages (Supplementary Figure 1).

Identification of DEGs

The GEO2R online analysis tool was utilized to identify the DEGs between PBC patients and HCs. Upon accessing the relevant GEO data website, the built-in GEO2R link was selected, followed by clicking on “Define groups” for grouping configuration. Subsequently, all PBC samples within the dataset were assigned to the experimental group, while all control samples were designated as the control group. After keeping the other settings at their default configurations on the website, an online analysis was conducted, and all analysis results were exported in Excel format. In the dataset of the results, genes that met the cutoff criteria of an adjusted P < 0.05 and |log2 fold-change| ≥ 0.2 were considered DEGs. The DEGs from each dataset were exported as separate Excel files. To analyze the overlap of DEGs among the three datasets, a Venn diagram web tool was employed (https://bioinfogp.cnb.csic.es/tools/venny/, accessed on 11 January 2024).

Immunohistochemistry

Immunohistochemistry was used to assess the expression of tumor necrosis factor alpha-induced protein 3 (TNFAIP3) in liver tissues. Paraffin-embedded sections were deparaffinized and immersed in antigen repair solution (citric acid repair liquid, G1202, Servicebio, China) and microwaved for 8 minutes, repeated three times. After cooling in a ventilated area, the slides were blocked by 3% bovine serum albumin (GC305010, Servicebio, China) for 30 minutes at room temperature and washed 3 times with phosphate buffer saline (PBS). Subsequently, the sections were incubated with TNFAIP3 monoclonal antibody (1:1000, GB112182-100, Servicebio, China) overnight at 4 °C. After incubation, the slides were washed 3 times with PBS and incubated with goat anti-rabbit IgG (horseradish peroxidase, 1:200; GB23303, Servicebio, China) for 50 minutes at room temperature. Then the slides were washed 3 times with PBS and incubated with DAB color developing solution (G1212, Servicebio, China). Pannoramic MIDI (3DHistech, Hungary) was used for image acquisition and the percent area of positive staining was analyzed by Image J software.

Statistical analysis

Continuous data are presented as mean ± SD or median (interquartile range), while categorical data are expressed as number (percentage). The t test or the Mann-Whitney U test was used to analyze continuous data whereas the χ² test or Fisher’s exact test was employed to estimate categorical data. Odds ratios (ORs) were calculated using a logistic regression model in univariate and multivariate analyses. Pearson’s correlation coefficient was employed to assess the correlation of data. All parameters that exhibited strong correlations (P < 0.05) in the univariate analysis were included as covariates in the multivariate logistic regression analysis, which was conducted to detect independent relationships. Receiver operating characteristic (ROC) curves were used to compare the predictive values. The optimal cutoff value based on the ROC curve was used as the positive threshold. SPSS version 25 (IBM, Armonk, NY, United States) was used for the statistical analyses. All analyses were two-sided, and differences were defined as statistically significant when P < 0.05.

RESULTS

Identification of DEGs

In this work, we examined three microarray expression datasets (GSE93170, GSE119600, and GSE159676) to explore DEGs between HCs and PBC patients. Upon differential expression analysis according to predetermined thresholds (adjusted P < 0.05 and |log2 fold-change) ≥ 0.2), 2347 DEGs comprising 1335 upregulated genes and 1012 downregulated genes in GSE93170, 1952 DEGs including 1108 upregulated genes and 844 downregulated genes in GSE119600, and 622 DEGs with 375 upregulated genes and 247 downregulated genes in GSE159676 were identified. The DEGs were illustrated using a volcano diagram (Figure 2A), and hierarchical clustering of the top 100 genes per dataset was displayed as heatmaps (Figure 2B). Shared DEGs across the three datasets were identified and presented using a Venn diagram analysis. The results revealed that 13 DEGs shared differential genes (MYOF, NEDD9, CELF2, SSH2, SEMA4D, TNFAIP3, TAGAP, BTLA, ARRDC2, FAHD1, PNP, SUMO1, UTY) (Figure 3A).

Figure 2 Volcano plots and heatmaps of differentially expressed genes in primary biliary cholangitis datasets. A: Volcano plots of differentially expressed genes (DEGs) identified in GSE93170, GSE119600, and GSE159676 datasets. A total of 1335 upregulated and 1012 downregulated genes were identified in GSE93170, 1108 upregulated and 844 downregulated genes in GSE119600, and 375 upregulated and 247 downregulated genes in GSE159676 (adjusted P < 0.05; |log2 fold change| ≥ 0.2). Orange and purple points represent upregulated and downregulated genes, respectively; B: Heatmaps of the top 100 DEGs from each dataset, clustered by expression levels. Red and blue colors indicate higher and lower expression levels, respectively. PBC: Primary biliary cholangitis; HCs: Healthy controls.

Figure 3 Identification of common differentially expressed genes and investigation of tumor necrosis factor alpha-induced protein 3 expression in primary biliary cholangitis. A: Venn diagram illustrated the common differentially expressed genes among three datasets (GSE159676, GSE93170, and GSE119600). A total of 13 differentially expressed genes were identified to be shared across all three datasets, including MYOF, NEDD9, CELF2, SSH2, SEMA4D, tumor necrosis factor alpha-induced protein 3 (TNFAIP3), TAGAP, BTLA, ARRDC2, FAHD1, PNP, SUMO1, and UTY; B: Immunohistochemistry of TNFAIP3 expression in the liver tissues from healthy controls (HCs) and primary biliary cholangitis (PBC) patients. Representative images showed the distinct staining pattern in both groups. Scale bars: 100 μm (upper panel), 20 μm (lower panel); C: Quantification of TNFAIP3 expression in the liver tissue from HCs and PBC patients. The relative TNFAIP3 expression level in the liver tissue of PBC was significantly higher than in HCs (^aP < 0.0001). PBC: Primary biliary cholangitis; HCs: Healthy controls; TNFAIP3: Tumor necrosis factor alpha-induced protein 3.

Gene ontology enrichment analysis of these 13 shared DEGs identified significant molecular functions, including protease binding. Among the 13 genes, TNFAIP3 and SUMO1 were enriched for protease binding activity (Supplementary Figure 2). This finding suggests a potential role of protease interaction pathways in PBC pathogenesis, particularly involving TNFAIP3, a key regulator of inflammatory signaling.

The ubiquitin-editing enzyme TNFAIP3 (also known as A20) is a negative regulator of nuclear factor kappa beta (NF-κB), and regulates immune cells such as T cells, B cells, macrophages and dendritic cells, thus participating in the occurrence and development of various autoimmune or inflammatory diseases[19]. Protein-protein interaction (PPI) analysis revealed TNFAIP3 as a hub gene interacting with key NF-κB pathway components (Supplementary Figure 3). Gene ontology enrichment further highlighted its central role in regulating NF-κB signaling, supporting its mechanistic involvement in PBC pathogenesis (Supplementary Table 1). To investigate the involvement of TNFAIP3 in PBC, we examined its expression in the liver tissue of HCs and PBC patients (Figure 3B and C). The hepatic TNFAIP3 protein levels were considerably increased in PBC patients compared to HCs.

Clinical features of PBC patients

This study enrolled 197 PBC patients. After excluding those with fatty liver disease, autoimmune hepatitis, viral hepatitis, liver cancer, hematologic disorders, or without liver biopsy, 71 PBC patients treated with UDCA were included (Figure 1). To identify predictive factors for poor response to UDCA, baseline clinical characteristics and TNFAIP3 expression were analyzed in UDCA responders and non-responders. The UDCA responders included 46 patients [aged 51 years (42-58), 1 male], while the UDCA non-responders included 25 patients [aged 48 years (43-57), 1 male] (Table 1). Compared to UDCA responders, non-responders showed a significantly higher prevalence of fatigue (56.0% vs 15.2%, P = 0.001), elevated levels of ALP (239.9 ± 38.7 vs 124.0 ± 12.2, P = 0.008), gamma-glutamyl transpeptidase (283.6 ± 76.5 vs 109.4 ± 16.5, P = 0.035), and TBIL (47.1 ± 11.8 vs 20.7 ± 3.7, P = 0.041), as well as higher positivity of anti-gp210 antibody (52.0% vs 28.3%, P = 0.047) and expression of TNFAIP3 (12.3 ± 1.0 vs 9.0 ± 0.6, P = 0.003). Conversely, the positivity of anti-centromere antibodies was less frequent in non-responders (4.0% vs 28.3%, P = 0.014). Variables with statistical significance in univariate analysis were included in the multivariable logistic regression analysis. As shown in the forest plots, fatigue (OR = 5.572; 95%CI: 1.346-23.062, P = 0.018) and TNFAIP3 (OR = 0.851; 95%CI: 0.733-0.988, P = 0.035) were both independent risk factors for response to UDCA in PBC patients (Figure 4A). To investigate the predictive value of TNFAIP3 and fatigue, we performed ROC curve analysis (Figure 4B). The area under the curve values for TNFAIP3 (0.691) and fatigue (0.704) were similar, whereas their combination significantly improved the predictive performance (area under the curve = 0.848). Based on the independent predictors identified, a nomogram model was developed and validated to predict non-response to UDCA, demonstrating good calibration and clinical utility (Supplementary Figure 4).

Figure 4 Risk factors and predictive performance of tumor necrosis factor alpha-induced protein 3 and fatigue for response to ursodeoxycholic acid in primary biliary cholangitis patients. A: Forest plot showed independent risk factors for response to ursodeoxycholic acid (UDCA) in primary biliary cholangitis patients. Fatigue (odds ratio = 5.572; 95%CI: 1.346-23.062, P = 0.018) and tumor necrosis factor alpha-induced protein 3 (TNFAIP3) expression (odds ratio = 0.851; 95%CI: 0.733-0.988, P = 0.035) were both identified as risk factors associated with poor response to UDCA. Variables with statistical significance are marked in orange, while non-significant variables are marked in blue; B: Receiver operating characteristic curve analysis of TNFAIP3 and fatigue for response to UDCA. The area under the curve values are 0691 for TNFAIP3, 0.704 for fatigue, and 0.848 for their combination. The orange, blue, and purple lines represent the combination, TNFAIP3 alone, and fatigue alone, respectively. OR: Odds ratio; ROC: Receiver operating characteristic; CI: Confidence interval; AUC: Area under the curve; TNFAIP3: Tumor necrosis factor alpha-induced protein 3; TBIL: Total bilirubin; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transpeptidase; ACA: Anti-centromere antibodies.

Table 1 Clinical, pathological, and tumor necrosis factor alpha-induced protein 3 expression characteristics of primary biliary cholangitis patients with different response to ursodeoxycholic acid, mean ± SD, n (%).

Characteristics	All patients (n = 71)	UDCA-responders (n = 46)	UDCA-non-responders (n = 25)	P value
Gender				1.000
Female	69 (97.2)	45 (97.8)	24 (96.0)
Male	2 (2.8)	1 (2.2)	1 (4.0)
Age (year)				0.479
Median (IQR)	51 (42-57)	51 (42-58)	48 (43-57)	-
Fatigue	21 (29.6)	7 (15.2)	14 (56.0)	0.001
Pruritus	11 (15.5)	6 (13.0)	5 (20.0)	0.439
PLT (109/L)	207.7 ± 8.8	217.6 ± 11.8	189.4 ± 12.0	0.127
ALT (U/L)	61.1 ± 8.4	55.2 ± 11.0	72.1 ± 12.6	0.341
AST (U/L)	52.9 ± 9.4	48.6 ± 13.3	60.8 ± 10.9	0.541
ALP (U/L)	164.8 ± 16.9	124.0 ± 12.2	239.9 ± 38.7	0.008
GGT (U/L)	172.4 ± 30.7	109.4 ± 16.5	283.6 ± 76.5	0.035
TBIL (μmol/L)	30.0 ± 5.0	20.7 ± 3.7	47.1 ± 11.8	0.041
TBA (μmol/L)	27.9 ± 6.2	21.9 ± 5.2	38.8 ± 14.8	0.194
Albumin (g/L)	39.3 ± 0.9	40.0 ± 1.2	38.0 ± 1.1	0.258
AMA	49 (69.0)	33 (71.7)	16 (64.0)	0.501
AMA-M2	47 (66.2)	32 (69.6)	15 (60.0)	0.416
Anti-sp100	17 (23.9)	11 (23.9)	6 (24.0)	0.993
Anti-gp210	26 (36.6)	13 (28.3)	13 (52.0)	0.047
ACA	14 (19.7)	13 (28.3)	1 (4.0)	0.014
IgM (g/L)	4.1 ± 0.8	3.1 ± 0.4	5.8 ± 2.0	0.092
Splenomegaly	14 (19.7)	7 (15.2)	7 (28.0)	0.196
Osteoporosis	18 (25.4)	11 (23.9)	7 (28.0)	0.705
Liver Fibrosis				0.106
I	45 (63.4)	29 (63.0)	16 (64.0)
II	21 (29.6)	15 (32.6)	6 (24.0)
III	5 (7.0)	2 (4.3)	3 (12.0)
IV	0 (0)	0 (0)	0 (0)
Ductopenia	19 (26.8)	10 (21.7)	9 (36.0)	0.195
TNFAIP3 (IOD)	10.1 ± 0.5	9.0 ± 0.6	12.3 ± 1.0	0.003

IQR: Interquartile range; PLT: Platelets; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transpeptidase; TBIL: Total bilirubin; TBA: Total bile acid; AMA: Anti-mitochondrial antibodies; ACA: Anti-centromere antibodies; IgM: Immunoglobulin M; IOD: Integrated optical density; TNFAIP3: Tumor necrosis factor alpha-induced protein 3; UDCA: Ursodeoxycholic acid.

Relationship between the expression of TNFAIP3 and clinicopathological features in PBC patients

Based on expression levels of TNFAIP3 (optimal cutoff value: 11.455), PBC patients were divided into TNFAIP3-high (n = 28) and TNFAIP3-low (n = 43) groups to investigate the role of TNFAIP3. Clinical parameters, including demographic characteristics, clinical symptoms, biochemical parameters, immunological markers and pathological features, were compared between the two groups (Table 2). The results indicated that, compared to the TNFAIP3-Low group, the TNFAIP3-high group demonstrated a younger age at onset [46 years (38-51) vs 53 years (44-60), P = 0.004], a significantly higher level of ALP (215.5 ± 35.4 vs 131.8 ± 14.1, P = 0.034), and a higher incidence of splenomegaly (32.1% vs 11.6%, P = 0.034).

Table 2 Clinical and pathological features of primary biliary cholangitis patients with different tumor necrosis factor alpha-induced protein 3expressions, mean ± SD, n (%).

Characteristics	TNFAIP3-high (n = 28)	TNFAIP3-low (n = 43)	P value
Gender			0.515
Female	28 (100)	41 (95.3)
Male	0 (0)	2 (4.7)
Age (year)			0.004
Median (IQR)	46 (38-51)	53 (44-60)	-
Fatigue	11 (39.3)	10 (23.3)	0.148
Pruritus	3 (10.7)	8 (18.6)	0.508
PLT (109/L)	197.6 ± 12.8	214.2 ± 12.0	0.362
ALT (U/L)	78.3 ± 17.2	50.0 ± 7.9	0.100
AST (U/L)	69.3 ± 21.6	42.2 ± 6.4	0.161
ALP (U/L)	215.5 ± 35.4	131.8 ± 14.1	0.034
GGT (U/L)	193.7 ± 40.4	155.8 ± 42.8	0.545
TBIL (μmol/L)	41.9 ± 11.38	22.3 ± 3.2	0.107
TBA (μmol/L)	40.2 ± 14.7	19.8 ± 3.5	0.187
Albumin (g/L)	38.2 ± 1.0	40.0 ± 1.3	0.321
AMA	19 (67.9)	30 (69.8)	0.865
AMA-M2	19 (67.9)	28 (65.1)	0.811
Anti-sp100	7 (25.0)	10 (23.3)	0.866
Anti-gp210	11 (39.3)	15 (34.9)	0.707
ACA	3 (10.7)	11 (25.6)	0.143
IgM (g/L)	5.5 ± 1.8	3.1 ± 0.4	0.139
Splenomegaly	9 (32.1)	5 (11.6)	0.034
Osteoporosis	8 (28.6)	10 (23.3)	0.615
Liver fibrosis			0.698
I	19 (67.9)	26 (60.5)
II	6 (21.4)	15 (34.9)
III	3 (10.7)	2 (4.7)
IV	0 (0)	0 (0)
Ductopenia	10 (35.7)	9 (20.9)	0.169

IQR: Interquartile range; PLT: Platelets; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; ALP: Alkaline phosphatase; GGT: Gamma-glutamyl transpeptidase; TBA: Total bile acid; AMA: Anti-mitochondrial antibodies; ACA: Anti-centromere antibodies; IgM: Immunoglobulin M; TNFAIP3: Tumor necrosis factor alpha-induced protein 3.

A correlation heatmap was constructed to explore the relationship between TNFAIP3 expression and clinical indicators in PBC (Figure 5). The level of TNFAIP3 showed significant correlations with age, albumin, TBIL, ALP and splenomegaly (P < 0.05). Specifically, TNFAIP3 was positively correlated with TBIL (r = 0.26), ALP (r = 0.28) and splenomegaly (r = 0.24), whereas inverse trends were observed for age (r = -0.31) and albumin (r = -0.28). The expression of TNFAIP3 showed a weak negative correlation with liver fibrosis (r = -0.14), which was not statistically significant.

Figure 5 Correlation heatmap. The heatmap illustrated the correlations between expression of tumor necrosis factor alpha-induced protein 3 and various clinical parameters in primary biliary cholangitis patients. The color of the circles indicates the status of the correlation, with red representing positive correlations and blue representing negative correlations. The size and color depth of the circles represent the strength of the correlation, with larger and darker circles indicating stronger correlations. Asterisks denote statistically significant correlations (^aP < 0.05, ^bP < 0.01, ^cP < 0.001). TNFAIP3: Tumor necrosis factor alpha-induced protein 3; PLT: Platelets; ACA: Anti-centromere antibodies; AMA: Anti-mitochondrial antibodies; IgM: Immunoglobulin M; TBA: Total bile acid; TBIL: Total bilirubin; ALP: Alkaline phosphatase; AST: Aspartate aminotransferase; GGT: Gamma-glutamyl transpeptidase; ALT: Alanine aminotransferase.

DISCUSSION

PBC is a chronic inflammatory cholestatic liver disease that progresses to end-stage biliary cirrhosis if not effectively treated. UDCA is currently the first-line therapy for PBC. In early-stage patients, UDCA treatment improves survival rates[20]. The median survival for advanced-stage PBC is 6-10 years[21]. Patients with an incomplete response or intolerance to UDCA, who tend to have a more severe disease burden and progression rate, require prompt second-line treatment, such as obeticholic acid[1]. Therefore, there is an urgent need to identify novel biomarkers for predicting response to UDCA or to discover new therapeutic targets.

In this study, the DEGs and hub genes in CD4⁺ T cells, blood samples, and liver tissues from PBC patients and HCs were analyzed. Bioinformatics analysis identified 13 key genes - MYOF, NEDD9, CELF2, SSH2, SEMA4D, TNFAIP3, TAGAP, BTLA, ARRDC2, FAHD1, PNP, SUMO1, and UTY - as critical in PBC progression. Subsequently, TNFAIP3 was selected for further investigation.

TNFAIP3/A20 is a ubiquitin-editing enzyme that regulates inflammation primarily by terminating NF-κB activation. It contains an N-terminal cysteine protease/ovarian tumor domain, which is essential for its deubiquitylating activity, and a C-terminal zinc finger domain that provides E3 ubiquitin ligase activity[22]. Although the expression of TNFAIP3 is low in most cell types, it is rapidly induced upon NF-κB activation, where it serves as a negative feedback regulator to suppress further NF-κB signaling[23]. The critical role of TNFAIP3 in human disease is supported by both TNFAIP3-deficient murine models and human genetic studies[24-27]. Notably, previous work has shown overactivation of the NF-κB pathway in PBC[28]. Our research confirmed significantly elevated TNFAIP3 expression in the liver tissues of PBC patients, suggesting that this molecule may regulate the pathogenesis of PBC through modulation of the NF-κB pathway.

In recent years, numerous clinical trials have been conducted globally to evaluate the efficacy of UDCA in treating PBC patients, and the clinical benefits are well-established[29-31]. However, current response criteria primarily rely on biochemical markers assessed after 1 or 2 years of treatment. Studies indicate that, without early intervention, most PBC patients experience histological progression and deterioration within the first 1-2 years[32]. Thus, accurately assessing early drug response has become a critical focus in PBC research. Previous studies have identified demographic characteristics such as age and sex[33,34], serum biochemical markers[17], autoantibodies[35], clinical manifestations like fatigue and pruritus[36], and histological findings[37] as predictors of response to UDCA. In this study, we analyzed clinical parameters to identify risk factors for poor response to UDCA in PBC patients. The results showed that expression of TNFAIP3 and fatigue were independent risk factors of response to UDCA. Moreover, the combination of TNFAIP3 and fatigue achieved a good predictive value. Correlations were observed between TNFAIP3 and age, albumin, ALP, and splenomegaly. Thus, TNFAIP3 may serve as a promising biomarker for the outcome of PBC.

This study had some limitations. First, the microarray data were obtained from public datasets and not generated by the authors, and this study focused solely on transcriptional profiling using microarray data, lacking integration with genomic, proteomic, or metabolomic data. Second, the sample size for analysis and verification was small, which may limit the statistical power of the study, making it difficult to detect subtle differences or associations. This could affect the robustness and generalizability of our conclusions, as small samples often lead to overestimation of effects or an increased risk of Type I and Type II errors. Third, the levels of TNFAIP3 in the blood samples of patients with PBC were not confirmed. Finally, the function and molecular mechanisms involving TNFAIP3 in the development and progression of PBC remain unclear. In the future, in-depth analysis using in vivo and in vitro experiments is warranted.

CONCLUSION

This study identifies TNFAIP3 as a promising biomarker for the prognosis of PBC patients through bioinformatics analysis and clinical validation. TNFAIP3 not only predicts response to UDCA but also holds potential as a novel therapeutic target. Additional studies are warranted to clarify the mechanistic involvement of TNFAIP3 in PBC pathogenesis.

ACKNOWLEDGEMENTS

We would like to thank the patients and their families for providing their consent to publish this report.