Genetic evidence for the causal influence of inflammatory factors on intrahepatic cholangiocarcinoma risk

Abstract

BACKGROUND

Intrahepatic cholangiocarcinoma (ICC) is a highly malignant liver cancer subtype with limited effective treatment options. Emerging evidence suggests that inflammatory factors play a critical role in ICC progression within the tumor microenvironment (TME). However, causal relationships between specific inflammatory factors and ICC risk remain unclear.

AIM

To investigate the causal relationship between inflammatory factors and ICC.

METHODS

This study used Mendelian randomization (MR) and Bayesian weighted MR (BWMR) analyses to investigate the causal impact of inflammatory factors on ICC risk. Genetic data from genome-wide association studies were utilized to identify and validate instrumental variables for 91 inflammatory factors, followed by sensitivity analyses to ensure robustness.

RESULTS

MR analysis identified significant associations between elevated levels of artemin and matrix metalloproteinase (MMP)-10 and increased ICC risk. BWMR and meta-MR analysis results validated these associations. Sensitivity analyses confirmed the stability of these findings, indicating that specific inflammatory factors may contribute causally to ICC development.

CONCLUSION

This study provides evidence that certain inflammatory factors, particularly artemin and MMP-10, are causally linked to ICC risk, identifying them as potential risk factors and therapeutic targeting. These findings enhance the understanding of the inflammatory components of the TME in ICC, supporting the development of targeted intervention strategies.

Key Words: Intrahepatic cholangiocarcinoma; Mendelian randomization; Bayesian weighted Mendelian randomization; Inflammatory factors; Tumor microenvironment

Core Tip: This study investigates the causal role of inflammatory factors in the risk of intrahepatic cholangiocarcinoma (ICC) using Mendelian randomization and Bayesian weighted Mendelian randomization analyses. Elevated levels of artemin and matrix metalloproteinase-10 were significantly associated with increased ICC risk, suggesting their potential as risk factors and therapeutic targeting. These results highlight the crucial role of inflammatory factors within the tumor microenvironment in ICC development, offering new insights for targeted interventions.

INTRODUCTION

Liver cancer is the seventh most commonly diagnosed cancer globally and the third leading cause of cancer-related deaths[1]. Histologically, primary liver cancer is primarily categorized into two subtypes: Hepatocellular carcinoma (HCC), which accounts for 75%-80% of cases, and intrahepatic cholangiocarcinoma (ICC), comprising 10%-15%[1,2]. ICC is a malignant tumor originating from the intrahepatic secondary bile ducts and their branches. Over recent decades, the incidence of ICC has steadily increased[3]. While it may be associated with conditions such as chronic hepatitis, cirrhosis, biliary tract inflammation, and hepatobiliary fluke infections, many cases arise de novo without any identifiable risk factors[4,5]. Despite advances in available therapies, surgical resection remains the cornerstone of ICC treatment; however, only about 20% of patients are eligible due to advanced-stage diagnosis[6]. Even among those who undergo surgery, approximately 80% experience recurrence, with 5-year survival rates ranging from 17% to 35%[6,7]. Current treatment strategies are often insufficient, underscoring the urgent need to identify novel risk factors and clarify the mechanisms driving ICC pathogenesis. Inflammation, while serving as a protective immune response against harmful stimuli, can become detrimental when chronic or uncontrolled, leading to tissue damage, carcinogenesis, and even mortality. Given this, it is plausible that inflammatory mediators play a role in ICC development[8,9]. A deeper understanding of their involvement is crucial for improving preventive strategies and therapeutic outcomes in patients with ICC.

Liver cancer - particularly ICC, which is strongly associated with inflammation - demonstrates that treatment efficacy is closely influenced by the tumor microenvironment (TME). The TME is a dynamic and complex system composed of various immune cell subsets, extracellular matrix (ECM) components, and a wide array of cytokines; all of which play essential roles in tumor initiation, progression, and therapeutic response[10-13]. Inflammation is a multifaceted physiological response to injury, infection, or irritation, triggering immune activation, the release of chemical mediators, and recruitment of immune cells to eliminate pathogens, remove damaged tissue, and promote repair[14,15]. Leukocytes and mast cells secrete proinflammatory factors such as interleukin (IL)-6, IL-1, nuclear factor-κB, tumor necrosis factor (TNF)-α, signal transducer and activator of transcription (STAT)3, and transforming growth factor-β; all of which contribute significantly to the persistence of chronic inflammation and tumorigenesis[16]. ICC is particularly linked to chronic liver conditions such as hepatitis, where sustained inflammation in the hepatic microenvironment fosters cirrhosis and cancer progression[17]. Hepatic stellate cells and tumor-associated macrophages promote fibrosis through ECM deposition, simultaneously supporting tumor growth and angiogenesis - key processes driven by inflammatory signaling[18]. Inflammatory cells such as neutrophils, macrophages, and lymphocytes facilitate tumor development by releasing growth factors and proteolytic enzymes that enhance tumor proliferation and invasion[19]. Extensive research across various malignancies has shown correlations between inflammation levels and clinical outcomes[20,21]. However, the precise relationship between the diversity of inflammatory cell populations and the risk of ICC remains unclear. This gap highlights the urgent need for comprehensive studies to establish causal links between specific inflammatory cell subsets and ICC development.

Previous studies have suggested associations between inflammatory mediators and ICC; however, these findings were primarily correlative and limited by potential confounders or reverse causation. Therefore, distinguishing true causal relationships remains a critical gap. Mendelian randomization (MR), which uses genetic variants as instrumental variables (IVs) for exposures, provides a robust method to infer causality by minimizing confounding factors and avoiding reverse causation, thereby offering more reliable insights into the etiological role of inflammatory factors in ICC. Current research predominantly focuses on the general impact of inflammatory factors on ICC, yet there is a scarcity of studies exploring the correlation between specific inflammatory factors and ICC. The present study utilizes MR analysis to explore the causal relationships between specific human inflammatory factors and the risk of developing ICC. We used meta-MR techniques to validate the crucial role of specific inflammatory factors in the risk of developing ICC.

MATERIALS AND METHODS

Study design

This study used MR analysis to explore potential causal links between specific inflammatory factors and the risk of ICC. As outlined in Figure 1A, three critical assumptions underpin MR studies: (1) A significant association exists between genetic variants and specific inflammatory factors; (2) These genetic variants have no association with confounding factors; and (3) The genetic variants influence ICC risk exclusively through their impact on specific inflammatory factors. This study leveraged desensitized data from publicly available genome-wide association studies (GWASs) to explore the relationships between specific inflammatory factors and ICC in European populations, ensuring no personal privacy or identifiable information was involved. Figure 1B illustrates the comprehensive workflow of the study. As the research did not include any personally identifiable data and lacked institutional consent authorization, it was exempt from ethical review.

Figure 1 An overview of the current research process and the foundational assumptions of Mendelian randomization analyses. A: Diagram of the three core assumptions in Mendelian randomization; B: Overall flowchart of the study process. SNPs: Single nucleotide polymorphisms; MR: Mendelian randomization; BWMR: Bayesian weighted Mendelian randomization.

GWAS data acquisition

Inflammatory factors samples used for exposure were sourced from published articles[22]. The GWAS catalog, covering entries GCST90274758 to GCST90274848, provides comprehensive GWAS statistics for each inflammatory factor trait, with data from 14736 European individuals (Supplementary Table 1). GWAS summary statistics for ICC in a European ancestry dataset, comprising 832 cases and 475259 controls (total sample size: 476091), were analyzed with 24196592 genetic variants. The specifics are detailed in Table 1.

Table 1 Details of the genome-wide association studies and datasets used in our analyses.

Trait	Cases	Control	Ancestry	Ref.	GWAS ID
Inflammatory factors	14736	/	European	Zhao et al[22]	GCST90274758 to GCST90274848
ICC (training cohort)	832	475259	European	Sakaue et al[48]	ebi-a-GCST90018803
ICC (validation cohort)	104	456244	European	Jiang et al[23]	GCST90043859

ICC: Intrahepatic cholangiocarcinoma; GWAS: Genome-wide association study.

Identification of IVs

To establish the causal link between specific inflammatory factor and ICC risk, we carefully selected appropriate single nucleotide polymorphisms (SNPs) as IVs through several quality control steps. Initially, IVs demonstrating strong correlations with specific inflammatory factor were chosen. When the initial stringent threshold (P < 5 × 10^-8) did not provide enough IVs, we adjusted to a more relaxed threshold (P < 5 × 10^-6) to gather a sufficient number for robust analysis. To address linkage disequilibrium (LD), we set the LD correlation coefficient to r² < 0.001 and a clumping window of > 10000 kb. Furthermore, palindromic SNPs were excluded from the IV selection. Finally, to assess the risk of weak IVs bias, we calculated the F-statistic for the IVs, considering an F-statistic > 10 as indicative of no significant weak instrumental bias in our MR analyses. Finally, MR analysis also requires the removal of confounders from IVs, which can be facilitated using the LDtrait Tool (https://ldlink.nih.gov/?tab=ldtrait). Subsequently, when conducting reverse MR analysis with ICC as the exposure and specific inflammatory factor as the outcome, the selection of IVs must also meet the previously mentioned four criteria.

MR and Bayesian weighted MR

To assess the causal relationship between specific inflammatory factor and ICC, we utilized MR analysis employing five distinct statistical approaches: Inverse variance weighted (IVW), weighted median, MR-Egger, simple mode, and weighted mode. IVW, which presupposes that all IVs adhere to MR assumptions, offers a dependable causality result and is frequently the main analytical method. To minimize the risk of false positives from multiple MR comparisons, we applied the Benjamini-Hochberg procedure to adjust the P values and control the false discovery rate (FDR). This correction was performed post hoc across the results generated by multiple statistical methods in the MR analyses. To enhance the reliability and practical significance of our study, we used specific inflammatory factor as the outcome and ICC as the exposure to conduct reverse MR validation. Bayesian weighted MR (BWMR) analysis complemented the two-sample MR analysis. This approach not only accounted for the polygenic structure and pleiotropy of diseases or traits, which are not considered in traditional MR analysis, but also enhanced the stability and reliability of the final results.

Validation cohort and MR meta-analysis

To confirm the accuracy of our positive MR results, we validated these findings using an external dataset. The GWAS summary statistics for the ICC validation cohort, based on a European ancestry dataset, were sourced from the GWAS catalog and include 104 cases and 456244 controls (total sample size: 456348) (Table 1)[23]. We combined the same positive inflammatory factors analyzed through the IVW-MR and BWMR methods across both discovery and validation cohorts. Meta-merging MR results from multiple databases helps to reduce database-specific biases, enhance confidence in study findings, and reconcile potential inconsistencies across studies. This approach also improved the accuracy of effect estimates, providing a more robust foundation for interpreting the causal impact of inflammatory factors on ICC risk.

Heterogeneity, pleiotropy, and sensitivity analyses

To evaluate the robustness of our findings, various sensitivity analyses were conducted. The Cochrane Q test (used with IVW) and the Rucker Q test (applied in MR-Egger) were utilized to check for heterogeneity, with P > 0.05 indicating the absence of heterogeneity. The mr_pleiotropy_test function within the Rstudio TwoSampleMR package was used to assess the pleiotropy of our effect estimates through the MR-Egger intercept method, with P > 0.05 suggesting no presence of horizontal pleiotropy. In the final stages, we used MR-PRESSO analysis to adjust causal association estimates for outliers, involving the removal of pleiotropic outlier SNPs before conducting a MR reanalysis. Furthermore, the leave-one-out method was utilized to determine whether any significant association was influenced predominantly by a single SNP. Conclusively, a reverse MR analysis was carried out to investigate the possibility of an inverse causal relationship between ICC and significantly identified inflammatory factor.

All statistical analyses were performed using the TwosampleMR package within the RStudio software (version 4.2.2). P < 0.05 was deemed to provide statistically significant evidence for a causal relationship.

RESULTS

IVs selection

According to IV selection criteria, 1817 SNPs were identified as IVs for 91 types of inflammatory factors. Each SNP had an F statistic > 10, indicating no weak instrument bias. Detailed information on the IVs across different inflammatory factor categories is provided in Supplementary Table 2.

Exploring the causal effect of specific inflammatory factors on ICC using MR analysis

We carried out two-sample MR analysis on the causal relationship between 91 inflammatory factors types and ICC (Figure 2). Using the IVW method, we found suggestive evidence for a causal association between increased risk of ICC and genetically predicted increases in artemin [standard error (SE) = 0.153, odds ratio (OR) = 1.587, 95%CI: 1.177-2.142, P = 0.002, FDR = 0.012]; IL-13 (SE = 0.200, OR = 1.763, 95%CI: 1.192-2.607, P = 0.005, FDR = 0.023); fms-related tyrosine kinase 3 ligand (FLT3 LG) (SE = 0.107, OR = 1.333, 95%CI: 1.082-1.642, P = 0.007, FDR = 0.035); eukaryotic translation initiation factor 4E-binding protein 1 (4E-BP1) (SE = 0.194, OR = 1.524, 95%CI: 1.042-2.227, P = 0.030, FDR = 0.074); and matrix metalloproteinase (MMP)-10 (SE = 0.112, OR = 1.253, 95%CI: 1.007-1.560, P = 0.043, FDR = 0.054; Figure 3A). In contrast, genetically predicted abundance of T-cell surface glycoprotein CD5 (SE = 0.139, OR = 0.716, 95%CI: 0.545-0.940, P = 0.016, FDR = 0.081); STAM binding protein (SE = 0.205, OR = 0.619, 95%CI: 0.414-0.927, P = 0.020, FDR = 0.099); neurturin (SE = 0.205, OR = 0.655, 95%CI: 0.438-0.977, P = 0.038, FDR = 0.191); and IL-5 (SE = 0.241, OR = 0.613, 95%CI: 0.382-0.983, P = 0.042, FDR = 0.211) were negatively related to ICC risk (Figure 3A). Although some P values fell below the conventional significance threshold of 0.05, the corresponding FDR values exceeded this cutoff. Our decision to interpret results primarily based on P values was grounded in the specific context of this study. While we acknowledge the importance of FDR correction for addressing multiple comparisons, we believe that, in this instance, P values remain a valid and informative indicator, particularly given the need for a more conservative assessment of statistical significance. The weighted median, simple mode, and weighted mode showed the same directional impact as the IVW method, although their P values did not consistently reach statistical significance (Supplementary Table 3).

Figure 2  A circular heatmap illustrating the causal impact of inflammatory factors on intrahepatic cholangiocarcinoma risk, analyzed using five statistical methods: Inverse variance weighted, weighted median, Mendelian randomization-Egger, Simple mode, and weighted mode.

Figure 3 Forest plots illustrating the causal effects of specific inflammatory factors on intrahepatic cholangiocarcinoma risk. A: Mendelian randomization (MR) analysis results using the Inverse-Variance weighted method, showing odds ratios with corresponding 95%CI and P values for each inflammatory factor; B: Bayesian weighted MR analysis results, validating the causal associations identified by MR. BWMR: Bayesian weighted Mendelian randomization; IVW: Inverse-Variance weighted method; OR: Odds ratio; Nsnp: Number of SNPs.

Exploring the causal effect of specific inflammatory factors on ICC using BWMR analysis

We conducted further BWMR analysis to examine ICC outcomes, selecting nine inflammatory factors previously identified as having genetic causal links to ICC. The BWMR findings supported prior MR results, demonstrating significant associations between ICC risk and the following factors: Artemin (SE = 0.153, OR = 1.599, 95%CI: 1.165-2.194, P = 0.004); 4E-BP1 (SE = 0.054, OR = 1.545, 95%CI: 1.055-2.263, P = 0.025); FLT3 LG (SE = 0.058, OR = 1.306, 95%CI: 1.060-1.610, P = 0.012); IL-13 (SE = 0.107, OR = 1.720, 95%CI: 1.112-2.659, P = 0.015); IL-5 (SE = 0.973, OR = 0.597, 95%CI: 0.378-0.944, P = 0.027); MMP-10 (SE = 0.005, OR = 1.248, 95%CI: 1.005-1.550, P = 0.045); neurturin (SE = 0.833, OR = 0.638, 95%CI: 0.435-0.937, P = 0.022); STAM binding protein (SE = 0.929, OR = 0.606, 95%CI: 0.395-0.930, P = 0.022); and T-cell surface glycoprotein CD5 (SE = 0.660, OR = 0.685, 95%CI: 0.517-0.909, P = 0.009; Figure 3B).

Reverse MR analysis

A reverse MR analysis was utilized through the five statistical methods to explore the potential causal association between ICC, and 91 specific inflammatory factors (Supplementary Figure 1). The results indicated significant causal relationships between ICC and programmed cell death ligand 1 (IVW: SE = 0.016, OR = 0.957, 95%CI: 0.937-0.999, P = 0.043; Figure 4A). Moreover, the MR-Egger, weighted median, simple mode, and weighted mode methods exhibited consistent directional effects with the IVW approach, although their P values did not always achieve statistical significance (Figure 4B). The leave-one-out sensitivity analysis confirmed the robustness of our MR findings, indicating that no single SNP exerted a substantial influence on the results (Figure 4C). However, there was no significant reverse causal association found between ICC and other inflammatory factors.

Figure 4 Results of reverse Mendelian randomization analysis examining intrahepatic cholangiocarcinoma as the exposure and specific inflammatory factors as the outcomes. A: A forest plot displays the significant causal effects of intrahepatic cholangiocarcinoma (ICC) on programmed cell death 1 ligand 1 levels; B: Scatter plots were employed to illustrate the causal relationship between ICC and programmed cell death ligand 1 levels, using five different Mendelian randomization methods; C: Forest plots from "leave-one-out" sensitivity analyses, illustrating the influence of individual single nucleotide polymorphisms on the overall results. ICC: Intrahepatic cholangiocarcinoma; MR: Mendelian randomization; IVW: Inverse-Variance weighted method; OR: Odds ratio; SNPs: Single nucleotide polymorphisms; Nsnp: Number of SNPs.

Validation cohort results and MR meta-analysis

When the nine positive inflammatory factors identified in the discovery cohort were validated in the validation cohort, only artemin (SE = 0.518, OR = 3.752, 95%CI: 1.361-10.347, P = 0.011) and MMP-10 (SE = 0.313, OR = 1.871, 95%CI: 1.013-3.456, P = 0.045) showed consistent, significant associations with ICC across both cohorts (Figure 5A). The MR-Egger, weighted median, simple mode, and weighted mode methods exhibited consistent directional effects with the IVW approach, although their P values did not always achieve statistical significance (Figure 5B and C). However, no significant causal relationship was found between ICC and the remaining seven specific inflammatory factors in the validation cohort. The MR-positive results for the two inflammatory factors significantly associated with ICC in both the discovery and validation cohorts were combined using the IVW and BWMR methods. After MR-meta merging, the P values for both factors were < 0.05, further strengthening the confidence in our findings (Figure 6).

Figure 5 Validation of causal associations between specific inflammatory factors and intrahepatic cholangiocarcinoma risk in an independent cohort. A: Forest plot showing the odds ratios with 95%CI for inflammatory factors previously identified in the discovery cohort, evaluated in the validation cohort using the Inverse-Variance weighted (IVW) Mendelian randomization (MR) method; B: Scatter plot illustrating the causal association between genetically predicted artemin levels and ICC risk, analyzed using five MR methods (IVW, weighted median, MR-Egger, simple mode, and weighted mode); C: Scatter plot illustrating the causal association between genetically predicted matrix metalloproteinase-10 levels and ICC risk, using the same five MR methods. ICC: Intrahepatic cholangiocarcinoma; MR: Mendelian randomization; IVW: Inverse-Variance weighted method; OR: Odds ratio; SNPs: Single nucleotide polymorphisms; Nsnp: Number of SNPs.

Figure 6 The meta-Mendelian randomization analysis was conducted to examine the association between artemin levels and matrix metalloproteinase-10 levels using data from both the training and validation sets. A: Meta-Mendelian randomization (MR) analysis of artemin levels was conducted based on intrahepatic cholangiocarcinoma (ICC) data from both the training and validation sets; B: Meta-MR analysis of matrix metalloproteinase-10 levels was conducted based on ICC data from both the training and validation sets. ICC: Intrahepatic cholangiocarcinoma; OR: Odds ratio; GWAS: Genome-wide association studies; Nsnp: Number of SNPs.

Heterogeneity and pleiotropy

To bolster the reliability of our MR findings, we rigorously assessed for heterogeneity and pleiotropy. These steps were crucial to identify any genetic variations that could impact multiple traits or uncover confounding factors that might have skewed the observed relationships, thus enhancing the overall robustness of our conclusions. The MR-Egger Intercept test indicated no significant horizontal pleiotropy (P > 0.05; Table 2). Additionally, the MR-Egger and IVW heterogeneity tests showed that there was no significant heterogeneity among the genetic variants used in our analysis (P > 0.05; Table 2). To confirm the stability of our MR findings, we performed a sensitivity analysis using the leave-one-out method. By sequentially excluding each SNP and comparing the causal effects of the remaining SNPs against the full MR analysis, we established that the MR results were highly consistent and stable (Supplementary Figure 2A and B). The MR-PRESSO analysis did not detect any outliers that significantly affected the study outcomes (Table 2). The symmetry observed in the funnel and forest plots further substantiates the reliability of our results (Supplementary Figure 2C-F).

Table 2 Sensitive analysis of the causal association between two circulating inflammatory factors and risk of intrahepatic cholangiocarcinoma.

Exposure	Outcome	Method	Heterogeneity		Pleiotropy		MR-PRESSO
			Q	P value	Egger intercept	P value	Outliers	P value
Artemin levels	ICC	MRE	12.173	0.878	-0.038	0.303	0	0.325
		IVW	13.294	0.864
MMP-10 levels	ICC	MRE	3.911	0.996	-0.032	0.265	0	0.695
		IVW	5.260	0.990

ICC: Intrahepatic cholangiocarcinoma; MRE: Mendelian randomization-Egger; MMP: Matrix metalloproteinase; IVW: Inverse-Variance weighted; MR: Mendelian randomization.

DISCUSSION

ICC is a prevalent and deadly malignancy within the digestive system, whose global burden has increased markedly from 1990 to 2019, posing significant threats to human life, health, and the global economy[24]. Despite ongoing efforts in diagnosis and treatment, most ICC patients are diagnosed at advanced stages, making them unsuitable for surgical resection and leading to poor prognosis[25,26]. Therefore, it is imperative to elucidate the risk factors and mechanisms contributing to the pathogenesis of ICC to develop innovative approaches for its prevention and treatment. The TME plays a crucial role in the formation, survival, and metastasis of tumor tissues[27,28]. Beyond cancer cells, the TME comprises a complex mix of innate and adaptive immune cells, inflammatory factors, stromal cells, endothelial cells, and cancer-associated fibroblasts[28]. The liver is a key immune organ enriched with various immunocompetent cells, including Kupffer cells, liver sinusoidal endothelial cells, hepatic stellate cells, pit cells, and lymphocytes such as natural killer cells, γδ T cells, and dendritic cells[27]. Inflammatory factors, cytokines and chemokines, such as IL-1α, TNF-α, and IL-6, are essential for liver regeneration after injury and play a crucial role in regulating metabolism, as well as the formation and development of tumors[29,30]. It has been reported in previous fundamental research that inflammatory factors are associated with ICC[31,32]. However, there is uncertainty regarding the causality of this association. Hence, it is important to understand the underlying interplay between inflammatory factors and ICC in the TME.

Our study leverages both individual and aggregated GWAS datasets on a large scale to systematically explore the genetic role of inflammatory factors in influencing the risk of ICC. To our knowledge, this is the first MR analysis to investigate causal relationships between multiple inflammatory factors and the risk of ICC. Using SNPs as IVs and applying various two-sample MR methods, we identified significant associations between specific inflammatory factors - namely artemin, IL-13, FLT3 LG, 4E-BP1, MMP-10, T-cell surface glycoprotein CD5, STAM binding protein, neurturin, and IL-5 - and the risk of ICC. Additionally, we enhanced the depth of our study and the reliability of our findings by incorporating reverse MR, BWMR and meta-MR analyses. Our comprehensive genetic analysis using GWAS summary data suggests that certain inflammatory factors, including artemin and MMP-10, contribute to ICC risk. It is noteworthy that only artemin and MMP-10 maintained significant associations with ICC risk in the validation cohort. This may be attributed to the smaller sample size of the validation cohort, leading to limited statistical power for detecting modest effects. Additionally, biological heterogeneity between cohorts, such as differences in genetic background, environmental exposures, or disease characteristics, could have influenced the replicability of causal signals. These factors highlight the inherent challenges in genetic epidemiological studies and underscore the need for further validation in larger, more diverse populations.

Tumor cells reside in a TME that consists of signaling molecules and stromal components, including vasculature, immune cells, fibroblasts, and the ECM[28,33]. Research indicates that the TME is highly complex, with inflammatory factors playing a crucial role in influencing tumor development, including in liver cancer[32,34]. ICC tumors are marked by a dense connective tissue layer infiltrated with immune and inflammatory cells[35]. As ICC progresses, tumor cells engage in reciprocal communication with surrounding stromal and immune cells, indicating a functional link between chronic inflammation and ICC development[35,36]. Inflammatory cytokines activate inducible nitric oxide synthase, producing high levels of nitric oxide that lead to single-strand, double-strand, and oxidative DNA damage, along with suppression of DNA repair enzymes[37]. IL-6, an inflammatory mediator produced by cholangiocarcinoma (CCA) and stromal inflammatory cells, operates via autocrine or paracrine signaling to enhance cell survival and stimulate cell growth[38]. In CCA, IL-6 promotes tumor progression by activating the STAT3 pathway, with this effect further amplified by KRAS mutations and epidermal growth factor receptor signaling, creating a protumorigenic feedback loop within the TME[38]. Inflammatory signaling pathways contribute to CCA development by inducing DNA damage and inhibiting apoptosis that would normally follow such damage. These pathways also stimulate cell proliferation, and the combined effects of DNA damage, apoptosis evasion, and increased proliferation are key steps in cellular transformation.

Artemin, part of the glial cell line-derived neurotrophic factor family, is essential for enhancing cell survival, migration, and invasion - critical processes in cancer metastasis[39]. In our current study, we found a significant association between elevated artemin levels and an increased risk of ICC, with an OR of 1.587, indicating a strong link between higher artemin levels and ICC development. Studies have demonstrated that artemin supports tumor survival by activating signaling pathways such as AKT/mTORC1 and MAPK; both of which are implicated in cancer cell proliferation and evasion of apoptosis[40,41]. In CCA specifically, these pathways might contribute to the aggressive nature and high mortality rate associated with ICC[42,43]. Elevated artemin levels could therefore be promoting tumor growth and metastasis, making it a potential biomarker for early detection and a target for therapeutic interventions. MMP-10 is a protease that breaks down components of the ECM, promoting tumor invasion and metastasis by enabling cancer cells to infiltrate surrounding tissues and access blood vessels for spread[44]. In our study, the association between elevated MMP-10 Levels and an increased risk of ICC, with an OR of 1.253, indicates a significant but comparatively modest link relative to artemin. This finding suggests that, although MMP-10 may not be as potent a risk factor as artemin, its role in ECM degradation and tissue remodeling remains relevant to ICC progression. Previous research supports the notion that MMPs are overexpressed in many cancers, where they contribute to metastasis[45,46]. In ICC, increased MMP-10 levels might support the aggressive nature of the disease by promoting stromal remodeling, enabling tumor cells to invade the hepatic tissue and form secondary sites[47]. Artemin and MMP-10, identified as causally related to ICC risk, hold potential as biomarkers for early screening and as targets for novel therapeutic strategies aimed at modulating inflammatory signaling pathways in ICC patients.

Our study had several limitations. Firstly, the reliance on self-reported cases from various datasets could have introduced potential biases. Secondly, the conclusions regarding the impact of inflammatory factors on ICC are constrained by the limited number of genetic instruments that achieved the significance threshold of P < 5 × 10^-6. Lastly, given the higher prevalence of ICC in men compared to women, there could be gender-specific differences in the relationship between ICC and inflammatory factors. Unfortunately, we lack sex-segregated data to explore this possibility.

CONCLUSION

Through MR and BWMR analyses, this study establishes a causal association between specific inflammatory factors, including artemin and MMP-10, and increased ICC risk. Our findings demonstrate that elevated levels of these factors significantly correlate with higher ICC susceptibility, suggesting their potential as risk factors and as therapeutic targets. These insights deepen our understanding of the role of inflammatory components within the TME, providing a foundation for ICC intervention strategies. Despite limitations such as potential dataset bias and the absence of sex-specific data, further research is needed to validate these findings and enhance their applicability across diverse populations.