Evaluation of HMGB1 Expression as a Clinical Biomarker for Cholangiocarcinoma

Abstract

Background/Aim

Cholangiocarcinoma (CCA) is an epithelial malignancy that is most prevalent in Southeast Asia, particularly in the northeast of Thailand. Identifying and establishing specific biomarkers of CCA is crucial for ensuring accurate prognosis and enabling effective treatment. High-mobility group box 1 (HMGB1) is a damage-associated molecular pattern (DAMP) molecule that can be released by dead or injured cells and is associated with tumor progression. This study aimed to investigate the expression levels of HMGB1 in CCA.

Materials and Methods

The clinical significance of HMGB1 levels was assessed by examining their correlation with patients’ clinicopathological data. A bioinformatics analysis was conducted to examine HMGB1 mRNA expression and perform survival analysis. The expression levels of 137 tissue cases were evaluated using the immunohistochemical technique, whereas the serum levels of 31 cases were evaluated using indirect ELISA.

Results

The GEPIA analysis demonstrated that HMGB1 exhibited elevated mRNA expression in CCA compared to the normal group. Immunohistochemical staining revealed that HMGB1 expression was primarily localized in the nucleus. High HMGB1 expression was observed in 57.6% of tissue samples, while low expression was detected in 42.4%. There was a significant positive correlation between high HMGB1 expression and the extrahepatic type of CCA as well as lymph node metastasis. The measurement of HMGB1 levels were assessed using indirect ELISA in 31 CCA serum samples, where 51.6% exhibited elevated concentrations of HMGB1. Elevated serum HMGB1 levels were significantly associated with advanced tumor stages and high levels of bilirubin levels.

Conclusion

HMGB1 in both tissue biopsies and blood serum shows potential as a predictive biomarker in CCA patients. These biomarkers could form the basis for facilitating more effective treatment planning.

Cholangiocarcinoma (CCA) is an epithelial malignancy that develops along the biliary tract and within the liver parenchyma, characterized by cholangiocyte differentiation. CCA is categorized based on its primary location as either intrahepatic CCA (iCCA) or extrahepatic CCA (eCCA). eCCAs are further divided into perihilar CCA (pCCA) and distal CCA (dCCA) (1). Fluke-related CCA, caused by parasitic infection, is a significant risk factor of CCA in Southeast Asia. Chronic inflammation is mainly caused by a parasitic infection that affects cholangiocyte differentiation leading to CCA. Previous evidence has indicated that liver flukes in the Opisthorchiidae family are a major risk factor for CCA in Southeast Asia. The prevalence of Opisthorchis viverrini (OV) is highest in northeast Thailand. A study in Khon Kaen revealed a significant positive association between OV infection and the incidence of CCA (2). The early stages of CCA lack specific signs for diagnosis, whereas late stages are more aggressive. Most patients present at an advanced stage, and many pass away before surgery can be performed. The survival rate for surgical patients is 20.6% after five years, while the survival rate for patients diagnosed within two years is only 8.1% (3,4). However, the course of therapy varies for each CCA patient. Therefore, identifying biomarkers to predict the prognosis of CCA patients is essential. Investigating and establishing specific biomarkers of CCA will be crucial in ensuring accurate prognoses and enabling more effective treatment.

High-mobility group box 1 (HMGB1) is a damage-associated molecular pattern (DAMP) molecule which can be released by dead or injured cells. HMGB1 is mainly located in the nucleus, but can be translocated to the cytosol, plasma membrane, and extracellular space to respond to various stresses including cell elimination and cancer progression. HMGB1/DAMP is released in ferroptosis by inflammation response. Advanced glycosylation end-product specific receptor (AGER/RAGE) is required for HMGB1 release (5). Xu et al. (6) determined the HMGB1 expression level to predict prognosis in iCCA. The result revealed that overexpression of HMGB1 in a tissue microarray of iCCA is associated with lymphatic microvessel density. Knockdown of HMGB1 can inhibit transforming growth factor-β-induced epithelial mesenchymal transition (EMT) by increasing E-cadherin and decreasing N-cadherin in RBE cells. The results demonstrated that overexpression of HMGB1 can predict prognosis and progression in iCCA patients.

This study aimed to determine the expression levels of HMGB1 in tissue and serum from CCA patients. The study analyzed the pattern of HMGB1 expression along with laboratory results and clinicopathological data, to determine their potential use as predictive or prognostic biomarkers in CCA for treatment planning.

Materials and Methods

Bioinformatical analysis of HMGB1 genes. HMGB1 gene expression profiles for gastrointestinal (GI) cancer were analyzed using the GEPIA2 online database. GI cancers as follows: cholangiocarcinoma (CCA or CHOL), colon adenocarcinoma (COAD), esophageal carcinoma (ESCA), liver hepatocellular carcinoma (LIHC), pancreatic adenocarcinoma (PAAD), rectum adenocarcinoma (READ), and stomach adenocarcinoma (STAD) were obtained from TCGA project and GTEx project online databases.

Human CCA tissue microarray and CCA serum. CCA tissue microarrays (n=137) were obtained from patients who were diagnosed with CCA by the pathologist. The CCA tissues were collected from patients who underwent surgery from 2005 to 2017 at the Srinagarind Hospital, Khon Kaen University, Thailand. Sera of CCA patients (n=31) were collected from Srinagarind Hospital. Sera of CCA patients were divided into two groups based on staging of CCA, early CCA group (stage I-II) and advanced CCA (stage III-IV).

Samples were kept at Cholangiocarcinoma Research Institute (CARI), Khon Kaen University, Thailand. This study has been reviewed by the Khon Kaen University Ethics Committee for Human Research based on the Declaration of Helsinki and the ICH Good Clinical Practice Guidelines (HE641654).

Clinicopathological and laboratory data. The clinicopathological characteristics and data of CCA patients, including sex, age, tumor location (iCCA, eCCA), tumor growth type (intraductal, mass forming, and mixed type), lymph node metastasis, distant metastasis, tumor staging was defined according to the American Joint Committee on Cancer Staging Manual, 8th edition. History of OV infection, chemotherapy treatment, liver function test, and patient survival rate were obtained from the Cholangiocarcinoma Research Institute (CARI).

Immunohistochemistry. CCA tissue microarray was performed using a 2.0 mm diameter needle with 70 cores per block. The tissue in the paraffin blocks was sectioned by a microtome. The sections were baked for 30 min at 60˚C before xylene deparaffinized and subsequent rehydration through graded ethanol (100%, 95%, 80%, and 70%). Antigen retrieval for HMGB1 was performed by heating the sections in 1X Tris EDTA buffer (pH 9.0) for 3 min using a pressure cooker, followed by cooling and washing with PBS. Subsequently, 0.3% hydrogen peroxide was used to block endogenous peroxidase activity followed by a 5% skim milk for 30 min to prevent non-specific binding. Each section was incubated with primary antibody HMGB1 (1:100, Invitrogen, Carlsbad, CA, USA, #PA1-16926) at 4˚C overnight. Then, the sections were subjected to three washes with 1X PBS+0.1% tween20 and were incubated with horseradish peroxidase-conjugated secondary antibodies (Rabbit, EnVision™, DAKO, Carpinteria, CA, USA) for 1 h. The sections were washed again with 1X PBS+0.1% tween20. After that, peroxidase activity was revealed by 3,3’-Diaminobenzidine substrates (DAB; Vector Laboratories, Inc., Newark, CA, USA), and the sections were counterstained with Mayer hematoxylin (DAKO, S3309). The sections were dehydrated with serial ethanol dilution (70%, 80%, 90%, and 100%) and were mounted by using Biomount HM. Slides were then examined under a light microscope.

Immunohistochemistry scoring. The immunohistochemical results were assessed using a Histoscore (H-score) (7). The expression level of HMGB1 proteins in cancer cells was quantified as the percentage of positively stained protein in the nucleus (0-100%), multiplied by the intensity of the color stains. The color stains were divided into four intensity types, 0 (no staining), 1 (weak), 2 (moderate), and 3 (strong). The possible scores were from 0 to 300. The expression level was categorized as either low or high based on the cut-off by a median value of the H-score for each protein.

Indirect ELISA. An indirect ELISA was performed to evaluate the diagnostic potential of circulating proteins including HMGB1 as biomarkers in CCA serum. A total of 31 serum samples from CCA patients were analyzed. The serum was diluted in coating buffer with a pH of 9.6 at a dilution of 1:100 and was then added to a 96-well MaxiSorp flat bottom plate, with 100 μl of the diluted serum applied to each duplicate well. The plate was incubated overnight and was washed using a solution of 1X PBS+0.05 tween20 and then discarded. Non-specific binding was prevented by blocking with 5% skim milk in the coating buffer for 1 hour at 37˚C. Then, the plate was washed by using 1X PBS+0.05 tween20. The primary antibodies HMGB1 (1:100, Invitrogen, #PA1-16926) was diluted in incubation buffer (pH 7.4) and added at a volume of 100 μl per well. The plate was then incubated at 37˚C for 1 h. Afterwards, it was washed using 1X PBS+0.05% tween20. Following the wash, horseradish peroxidase-conjugated secondary antibodies (anti-rabbit IgG, Invitrogen, #G-21234) were added. The plate was washed once more, Orthophenylene diamine hydrochloride (OPD) substrate was added, and the plate was then incubated in darkness for 30 min. To stop the reaction, 100 μl/well of 4M H₂SO₄ was added. The activity was observed by using an ELISA reader at the optical density (OD) of 492 nm. The OD value was calculated and a definite cut-off value by a median. The result was provided as a group of low and high-expression groups.

Statistical analysis. Statistical analyses were performed using IBM SPSS version 27.0 software (SPSS Inc., Chicago, IL, USA) and GraphPad Prism V.8.0 (GraphPad Software, San Diego, CA, USA). The continuous data were presented as means±SD. The categories data were reported in percentages. The association between each protein expression and clinicopathological and laboratory data of CCA patients was evaluated by using the chi-square test and z-test. Kaplan-Meier survival analysis methods were used to evaluate overall survival. The Log-rank test was used to compare survival between low and high expression. The univariate and multivariate were used to determine variables affecting overall survival. For the continuous data as OD. result of protein expression in sera, Mann-Whitney test was employed. A p-value <0.05 was considered statistically significant.

Ethics statement. All human specimens and the protocols in this study were approved by the Human Ethics Committee of Khon Kaen University, based on the ethics of human specimen experimentation of the National Research Council of Thailand (HE641654) and informed consent was obtained from each subject before surgery.

Results

Patient characteristics. Complete data for 137 CCA tissue microarrays were obtained from patients who were diagnosed with CCA by the pathologist. The mean age of the patients was 60.29 years (range, 39-82 years). Out of the total 137 patients, 86 (62.8%) were male and 52 (37.2%) were female. The reference value of laboratory results based on Srinagarind Hospital, Khon Kaen University, Thailand was obtained for analysis.

Distribution of HMGB1 mRNA expression and its correlation with clinicopathological data. The mRNA expression levels of HMGB1 in GI cancers were assessed using the GEPIA2 online analysis tool. HMGB1 exhibited elevated expressions in the tumor group compared to the normal group, particularly in the CCA, colon adenocarcinoma, esophageal carcinoma, liver hepatocellular carcinoma, pancreatic adenocarcinoma, rectum adenocarcinoma, and stomach adenocarcinoma subtypes, as demonstrated in Figure 1A. The Kaplan-Meier survival analysis demonstrated that patients with high expression of HMGB1 tended to associate with lower overall survival (OS) (p=0.3) and had significantly associated with disease-free survival (DFS) compared to those with low expression of HMGB1 (p<0.01) in GI cancers. In only CCA, the expression of HMGB1 did not demonstrate a significant difference in both overall survival and disease-free survival as illustrated in as shown in Figure 1B.

Figure 1.

The mRNA expression levels of HMGB1 in GI tract cancer and survival analysis and disease-free survival along with mRNA expression level of HMGB1 in GI cancer via GEPIA2 analysis. (A) HMGB1 mRNA expression levels in GI tract cancer compared to normal group. Red color refers to cancer group, and green color refers to normal group. (B) Overall survival and disease-free survival analysis of HMGB1 by the Kaplan-Meier method with a log-rank test from the GEPIA2 database in GI cancer. (C) Overall survival and disease-free survival analysis of HMGB1 by Kaplan-Meier method with a log-rank test from the GEPIA2 database in CCA only. HMGB1: High-mobility group box 1; GI: Gastrointestinal.

HMGB1 protein expression level in CCA tissues. Immunohistochemical analysis was performed to evaluate the HMGB1 protein expression. The immunoreactivity of HMGB1 was predominantly found in nucleus staining and found in cytoplasm of all CCA tissues. The levels of HMGB1 were detected in normal bile ducts, however their expression was found to be low.

HMGB1 expression was low in 57.6% (79/137) of cases, while high expression of HMGB1 was reported in 42.4% (58/137) of cases, as shown in Figure 2. The high expression levels of HMGB1 were significantly correlated with extrahepatic type of CCA (p<0.01), lymph node metastasis (p=0.01). Additionally, extrahepatic type was associated significantly with CCA patients who survived for two years and exhibited elevated HMGB1 expression (p=0.02). High HMGB1 tends to associate with lymph node metastasis (p=0.05) as shown in Table I.

Figure 2.

Immunohistochemical staining for HMGB1 protein expression in CCA tissues. (A) HMGB1 expression in normal bile duct and CCA area. (B) Low ACSL4 expression in CCA area. (C) High HMGB1 expression in CCA area. HMGB1: High-mobility group box 1; CCA: cholangiocarcinoma; scale bar: 50 μm; magnification ×40.

Table I. Correlation between HMGB1 expression and patient clinicopathological data.

HMGB1: High-mobility group box 1; TNM: Tumor Node Metastasis; OV: Opisthorchis viverrine. Statistically significant p-values are shown in bold.

Alteration of HMGB1 protein and its correlation with clinicopathological data and laboratory results. The correlation between overall survival of CCA patients and expression of HMGB1 was analyzed using Kaplan-Meier survival curve analysis. There was no statistically significant correlation between HMGB1 expression and survival time in all CCA patients (p=0.93), as shown in Figure 3A. When focusing on the association between CCA patients and their survival time, specifically on a 2-year survival analysis, there was no significant correlation between low HMGB1 or high HMGB1 expression and their survival time with a median survival of 10.5 months and 8.8 months, respectively (p=0.48) as shown in Figure 3B.

Figure 3.

Overall survival analysis of HMGB1 expression in all CCA patients and 2-year survival analysis. (A) Survival analysis of HMGB1 expression in CCA patients. (B) Survival analysis of HMGB1 expression in CCA 2-year survival. All analyzed cut-off by median of H-score. HMGB1: High-mobility group box 1; CCA: cholangiocarcinoma.

Alteration of HMGB1 protein expression level and the correlation with clinicopathological data and laboratory results in CCA sera. The levels of HMGB1, secreted proteins, were measured in the serum of two CCA groups: early CCA and advanced CCA. The advanced CCA group exhibited a significantly elevated level of HMGB1 in comparison to early CCA groups. The mean value of each early CCA group and advanced CCA group were 0.55 and 0.82, respectively as depicted in Figure 4A.

Figure 4.

HMGB1 expression levels and overall survival analysis in advanced CCA compared to early CCA in serum. (A) HMGB1 expression was found to be significantly higher in advanced CCA sera than in early CCA sera. (B) The AUC curve in different group based on HMGB1 expression (C) Survival analysis of HMGB1 expression in CCA sera patients. All analyzed cut-off by YI. HMGB1: High-mobility group box 1; CCA: cholangiocarcinoma; AUC: area under the curve; YI: Youden Index; OD: optical density.

HMGB1 is capable of distinguishing advanced CCA from early CCA (AUC=0.775, p<0.01), as depicted in Figure 4B. HMGB1 level can predict the severity of CCA with a cut-off value of 0.64, a Youden Index of 0.48, a sensitivity of 75.0%, and a specificity of 73.3%. HMGB1 was categorized with high expression and low expression with OD cut-off at 0.64 as shown in Table II.

Table II. Predictive values of serum HMGB1 levels for CCA prognosis, based on the optimal cut-off derived from ROC analysis and Youden index calculation.

HMGB1: High-mobility group box 1; CCA: cholangiocarcinoma; AUC: area under the curve; ROC: receiver operating characteristic; LR: likelihood ratio; YI: Youden Index; Statistically significant p-values are shown in bold.

There was a strong association between elevated levels of both HMGB1 and advanced TMN stage (stage III-IV), as well as high total bilirubin levels, direct bilirubin level and higher level of CA19-9 (p<0.01, p=0.01, p=0.03, p<0.01, respectively). Elevated HMGB1 levels showed a trend to significant correlation with increased levels of AST (p=0.07) as shown in Table III. There is a trend to link between the high level of HMGB1 and shorter survival time (p=0.07) as demonstrated in Figure 4C. The HMGB1 low-expression group had a median survival time of 28.7 months, while the HMGB1 high-expression group had a median survival time of 6.7 months.

Table III. Correlation between HMGB1 expression and clinicopathological data in CCA sera.

HMGB1: High-mobility group box 1; CCA: cholangiocarcinoma; TNM: Tumor Node Metastasis; AST: aspartate transaminase; ALP: alkaline phosphatase; CA19-9: carbohydrate antigen 19-9; CEA: cancer embryonic antigen; Statistically significant p-values are shown in bold.

Discussion

HMGB1, which is characteristic of a prototypical nuclear DAMP, is secreted by ferroptotic cancer cells. The dual function of HMGB1 presents as follows: immunogenic cell death (ICD) in early ferroptotic cells and tumor progression in cancer cells. When aberrant CD8+ T cell release interferon gamma, it binds to the JAK/STAT receptor on the surface of early ferroptotic cell. This triggers the release HMGB1, which binds to its receptors RAGE/AGER and TLR2,4 in inflammatory cells. This affects the immune response leading to elimination of cells. On the other hand, ferroptotic cancer cells release HMGB1 into the extracellular space. HMGB1 acts as a proinflammatory cytokine and chemokines by binding to its receptor RAGE/AGER and TLR2,4 on inflammatory cells. This binding activates ERK1/2 and NF-kB, resulting in the release of proinflammatory mediator, such as IL-1B, IL-6, TNFα. Consequently, this leads to an increase in cell proliferation, migration and invasion (5,8,9).

A previous study reported the correlation between HMGB1 expression and poorer prognosis in many cancers. Nuclear HMGB1 interacts with and enhances the activities of several factors implicated in cancer development (10). In the present study, the HMGB1 expression levels were assessed in CCA tissues and CCA sera. Our results demonstrate that, high expression levels of HMGB1 were significantly associated with eCCA and lymph node metastasis, indicating poorer prognosis in CCA tissues. Similarly, Xu et al. showed that HMGB1 was highly express in eCCA cells and associated with VEGFR2 expression leading to angiogenesis and poor prognosis in eCCA (11). Moreover, Suwannakul et al. (12) reported that overexpression of HMGB1 was significantly associated with metastasis stage in CCA patient with Clonorchis sinensis infection. Elevated HMGB1 expression was significantly associated with tumor grade and stage in bladder cancer tissues (13). Knockdown of HMGB1 downregulated the expression of p-AKT, Ki-67 and MMP-2, and inhibited the proliferative activities and metastatic potential in liver cancer (14). Alternatively, HMGB1 expression in CCA sera was assessed by indirect ELISA technique. The results of the present study revealed that the advanced CCA group exhibited significantly elevated levels of HMGB1 in CCA sera compared to the early CCA group. HMGB1 were capable of distinguishing advanced CCA from early CCA. Elevated HMGB1 levels in CCA sera were associated with a higher staging especially the metastasis stage. In the case of advanced CCA, the tumor invades into bile ducts and causes bile duct obstruction leading to high levels of bilirubin. Similarly, our result showed that elevated HMGB1 was significantly associated with high levels of bilirubin which indicated a poorer prognosis in patients with CCA. A study has reported that high levels of HMGB1 tend to be associated with tumor progression in gastric cancer (15). The elevated level of HMGB1 was significantly associated with lymph node metastasis and staging of pancreatic ductal adenocarcinoma (PDAC). Moreover, elevated HMGB1 was significantly correlated with shorter survival in PDAC patients (16). Similarly, with our results, high a level of HMGB1 was significantly associated with advanced TMN stage, elevated levels of bilirubin, and higher level of CA19-9. According to our results, HMGB1 levels may serve as prognosis markers in CCA patients by using CCA tissues and especially by using CCA sera which is a less invasive technique. As the present study was limited by the small sample size of CCA sera, further validation on more CCA sera samples should be conducted to confirm the HMGB1 levels in CCA sera.

Conclusion

Our study demonstrated a significant association between high HMGB1 expression and extrahepatic CCA and lymph node metastasis. The study suggests that HMGB1 expression could be used as a biomarker for predicting prognosis parameters in CCA tissues. Importantly, we found that high HMGB1 levels in CCA sera were both significantly associated with advanced tumor stages and high bilirubin levels. This study revealed that HMGB1 expression might be a significant prognostic marker in predicting prognosis for improved therapeutic planning and management in patients with CCA.

Funding

The research on “Ferroptosis: the new biomarker and targeting therapy in Cholangiocarcinoma” by Khon Kaen University has received funding support from National Sciences Research and Innovation Fund (NSRF).

Conflicts to Interest

The Authors declare no potential conflicts of interest.

Authors’ Contributions

Supakan Amontailak: Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing-original draft. Attapol Titapun, Apinya Jusakul, Raynoo Thanan: Conceptualization, Supervision, Validation, Writing-review & editing. Phongsaran Kimawaha, Wassana Jamnongkan, Papitchaya Sirithawat: Formal analysis, Writing-review & editing. Malinee Thanee, Songpol Haohan: Resources, Writing-review & editing. Anchalee Techasen: Conceptualization, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing-review & editing. All Authors contributed to the final manuscript.

Acknowledgements

We would like to acknowledge Prof. Ross H Andrews for editing the MS via Publication Clinic KKU, Thailand.