Editor's Note: Biliary tract cancer (BTC) is the second most common primary liver malignancy after hepatocellular carcinoma (HCC), and its incidence has been rising globally over the past few decades. Intrahepatic cholangiocarcinoma accounts for 20% of all BTCs, with the remainder consisting of extrahepatic cholangiocarcinoma (including hilar and distal cholangiocarcinoma) and gallbladder cancer. A recent study published in Hepatology International explored the molecular subtyping of BTC patients based on 19 PANoptosis-related gene (PRG) signatures, providing new insights for personalized treatment strategies in BTC.Recent research has highlighted the crucial role of the tumor microenvironment in BTC progression and invasiveness. The primary goal of cancer therapy is to disrupt tumor cells, and a key strategy involves inducing cancer cell death. Programmed cell death (PCD) is an active mechanism that maintains the development and survival of organisms. The most studied PCD pathways include apoptosis, pyroptosis, and necroptosis, all of which play critical roles in the tumor microenvironment. A recently discovered cell death process, known as PANoptosis, includes apoptosis, pyroptosis, and necroptosis. While studies on PANoptosis have largely focused on solid malignancies such as HCC, pancreatic cancer, lung cancer, and kidney cancer, its role in BTC pathogenesis remains unclear. This study investigated the prognostic value of PRGs and established markers to predict BTC patients’ prognosis and immune/chemotherapy responses, demonstrating a significant correlation between PRGs and the tumor immune microenvironment.
Researchers performed consensus clustering analysis and categorized patients into two subtypes, A and B. Based on previous studies that identified 19 PRGs, the researchers conducted LASSO regression analysis on these 19 gene signatures, constructing and validating a 9-gene risk score prognostic model that accurately predicts overall survival (OS) in different BTC patients. Additionally, a nomogram was developed to showcase the clinical utility and robustness of the model.
Prognostic Model Construction
The researchers calculated coefficient values at various penalty levels. First, they determined the optimal λ value through ten-fold cross-validation, selecting two best-fit values (λmin and λ1se). They constructed the LASSO model by minimizing the mean square error and selected two gene sets (a 9-gene set for λmin and a 4-gene set for λ1se). To enhance the model’s prognostic predictive performance, they ultimately chose the 9-gene risk prediction model for further study. The ROC curve analysis of the prediction model showed AUC values of 0.818 (5-year), 0.848 (3-year), and 0.778 (1-year), indicating good performance in predicting OS.
Through multivariate Cox regression analysis, the risk scores derived from the coefficients of these nine candidate genes were used to classify patients into low-risk (n=66) and high-risk (n=49) groups, with a threshold of 0.92. The Kaplan-Meier survival curve showed that OS was significantly lower in the high-risk group compared to the low-risk group (P<0.0001).
The results indicated that BTC patients in group B were more likely to have a high-risk score and poorer prognosis. The researchers compared the expression levels of the 19 PRGs between the high-risk and low-risk groups, finding significant differences in the expression levels of CASP7, TAB3, MLKL, RIPK3, and ZBP1 between the two groups.
Model Validation and Development of Prognostic Nomogram
External validation was conducted to clarify the generalizability of the 9-gene risk score prognostic model. The researchers divided the TCGA BTC cohort into high-risk and low-risk groups based on the risk scores (threshold: 0.2). The results showed that the high-risk group had poorer OS (P=0.0091). In the TCGA validation cohort, the difference in RIPK1 expression between the two groups was nearly statistically significant (P=0.061), a phenomenon possibly related to the small sample size of the validation group.
The AUC values of the 9-gene risk score prognostic model for predicting 1-year, 3-year, and 5-year OS in BTC patients demonstrated good efficacy. Independent prognostic analysis, nomogram development, and clinical applicability analysis confirmed that the 9-gene risk score prognostic model can independently and accurately predict OS in BTC patients. Subsequently, researchers selected TH23 patients from the GSE89749 database for randomized validation and found significant differences in prediction accuracy with and without the risk score as a nomogram factor.
Immune Landscape and Prediction of Immunotherapy and Chemotherapy Response
The researchers assessed the relationship between the risk score and immune cell infiltration. Analysis showed no statistically significant difference in ImmuCellAI infiltration scores between the high-risk and low-risk groups. They then explored the different immune cell abundances of cell types in the first and second layers between the high-risk and low-risk groups. In the high-risk group, the numbers of dendritic cells (DCs), monocytes, macrophages, and NKT cells were higher. In the low-risk group, γ-δ cells, Th1 cells, and cytotoxic cells were more abundant.
The researchers further grouped samples by high or low expression levels of the nine genes and analyzed the abundance of various immune cell infiltrations at the subgroup level, finding that each risk gene was associated with different immune cell subtypes.
Further correlation analysis between the expression levels of each risk gene and immune cell infiltration scores revealed a significant correlation between MUC16 and immune cell infiltration scores, indicating that MUC16 plays a critical role in the 9-gene risk score model.
Additionally, the researchers explored the relationship between chemotherapy drugs and the risk score model. The results showed statistically significant differences in the effects of docetaxel and vinblastine between the two groups. In future BTC treatments, low-risk patients may benefit from combined immunotherapy with docetaxel and vinblastine.
The researchers also examined the expression of the nine genes in each cell, noting that ALKAL2 expression was not detected in single-cell data, leaving eight genes expressed. Among these, four genes were significantly expressed in immune cells, SLC20A1 and KRT17 were predominantly expressed in malignant tumor cells, and ITGA1 was more expressed in stromal cells.
The researchers concluded that PRGs can accurately predict the prognosis and immunotherapy response in BTC patients, thus providing guidance for personalized treatment of BTC. This study deepens our understanding of PANoptosis in BTC and offers more strategies for the clinical treatment of this malignancy.
