Role of cancer stem cell heterogeneity in intrahepatic cholangiocarcinoma

Abstract

Background: Intrahepatic cholangiocarcinoma (ICC) is a highly invasive bile duct cancer with poor prognosis due to frequent recurrence and limited effective treatments. Cancer stem cells (CSCs) contribute to ICC's therapeutic resistance and recurrence, driven by distinct cellular subpopulations with variable tumorigenic properties. Recent advances in single-cell RNA sequencing (scRNA-seq) have enabled a deeper exploration of cellular heterogeneity in tumors, offering insights into unique CSC subgroups that impact ICC progression and patient outcomes. This study aimed to investigate the effect of CSC heterogeneity on the prognosis of ICC.

Methods: The scRNA-seq dataset GSE142784 was retrieved from the Gene Expression Omnibus (GEO) database, and Bulk RNA-seq data were obtained from The Cancer Genome Atlas (TCGA) databases. Hallmarks and AUCell R package were adopted for analyzing the signaling pathway activity, CellChat for observing cell communication between subgroups, and SCENIC for analyzing transcription factors expression. The immune cell infiltration and drug sensitivity of the model were analyzed using the CIBERSORT algorithm and the "pRRophetic" R packages, respectively. And immunohistochemistry (IHC) tests were used to evaluate expression of transcription factors in ICC patients.

Results: Based on scRNA-seq data, five clusters (DLK⁺, CD13⁺, CD90⁺, CD133⁺, and other cholangiocarcinoma cells) were observed in ICC, which presented different signaling pathway activities, such as HSF1 and STAT1 were highly expressed in the CD133 cluster, and consistent with the results of IHC tests. Pathways like Notch and Wnt/β-catenin signaling transferred among above subgroups. Further, subgroups favored varied immune response and drug sensitivity, and CD133⁺ subgroup patients showed significantly shortened recurrence-free survival (RFS).

Conclusions: Configuring the subgroup of ICC is helpful for predicting the prognosis and drug resistance in ICC and can provide new strategies for cancer treatment.

Keywords: Intrahepatic cholangiocarcinoma (ICC); cancer stem cell heterogeneity (CSC heterogeneity); poor prognosis; therapeutic target.

Figure 1

Sorting of intrahepatic cholangiocarcinoma cell groups. (A) Cancer stem cells in intrahepatic cholangiocarcinoma patients were grouped by the t-SNE algorithm according to surface receptor markers; (B) pie chart for proportions of each cell group; (C) distribution and expression levels of EPCAM, KRT7 and KRT19 antigen positive cell groups; (D) surface antigen expression for cell groups. t-SNE, t-distributed stochastic neighbor embedding.

Figure 2

Sorting of intrahepatic cholangiocarcinoma stem cell groups. (A) Clustering of cancer stem cells in intrahepatic cholangiocarcinoma patients through the t-SNE algorithm based on surface receptor markers; (B) expression of tumor-associated genes in different cell groups; (C) pie chart showing the proportions of each cell group among all cells in tumor samples; (D) expression levels of different genes in different cell groups; (E) pseudotime analysis for paths of cell growth and development. CCA, cholangiocarcinoma; CSC, cancer stem cell; t-SNE, t-distributed stochastic neighbor embedding.

Figure 3

Signaling pathway analysis for different intrahepatic cholangiocarcinoma cell groups. (A) Pseudotime analysis of the expression and distribution of ANPEP, PROM1, THY1, CEBPD, HSF1 and STAT1. Pseudo-time reflects the relative progression of cells along a trajectory and is dimensionless. (B) Hallmark analysis of differentially enriched signaling pathways in cancer stem cells; (C) expression of related signaling pathways in different cell groups according to t-SNE. CCA, cholangiocarcinoma; CSC, cancer stem cell; NES, Normalized Enrichment Score; GSEA, Gene Set Enrichment Analysis; AUC, area under the curve; t-SNE, t-distributed stochastic neighbor embedding.

Figure 4

Interactions between intrahepatic cholangiocarcinoma cell groups and differences in transcription factors. (A) Cell-cell interaction; line segment thickness indicates the weight (upper) or number (lower) of interactions between groups, and loops indicate autocrine interactions. (B,C) The export (left) and input (right) signal patterns (B) and differences in transcription factors in different cell groups (C) shown in heatmaps. The bars represent relative signaling strength, which is dimensionless and derived from normalized interaction scores. (D) Violin plot showing the ex-pression of CEBPD, HSF1, and STAT1. (E) Representative IHC images for expression of CD13, CD133, CD90, HSF1 and STAT1 in ICC samples. For CK19, CD90, CD13 and CD133 staining, positive staining can be seen in the membrane of the tumor cells, while positive staining of HSF1 and STAT1 can be seen in the nuclear of the tumor cells. (Original magnification ×200; inset shows an enlarged area at ×400). CCA, cholangiocarcinoma; CSC, cancer stem cell; ICC, intrahepatic cholangiocarcinoma; IHC, immunohistochemistry.

Figure 5

Clinical prognostic ability of cancer stem cell subgroups. (A) Upregulated differentially expressed genes in different cancer stem cell subgroups. (B) Clustering and scoring of intra-hepatic cholangiocarcinoma-related data in the TCGA database were performed with the ssGSEA tool. (C,D) Kaplan-Meier curves showing relapse-free survival in patients with high and low CD13/CD90/CD133 expression (C) and in different patient clusters (D). (E,F) Pattern recognition analysis for the correlations of different patient clusters with sex, pathological tissue, disease duration, and recurrence. CCA, cholangiocarcinoma; CSC, cancer stem cell; FC, fold change; CDF, cumulative distribution function; TCGA, the Cancer Genome Atlas.

Figure 6

Correlations between cancer stem cell subgroups and immune features. (A) CIBERSORT analysis method for observing immune cells in cluster A/B/C; (B) ESTIMATE algorithm for calculating immune and stromal scores in the 3 groups of patients; (C) observation of the expression of genes related to the immunotherapy response in the three groups; (D) CIBERSORT for calculating the correlation of cholangiocarcinoma immune cells with the CD133, CD13, and CD90 subgroups in the TCGA cohort. The bars represent the number of samples. *, P<0.05; **, P<0.01; ns, not significant. PRG, prognostic-related genes; CSC, cancer stem cell; MDSC, myeloid-derived suppressor cells; TCGA, the Cancer Genome Atlas.

Figure 7

Drug sensitivity in cancer stem cell subgroups. The pRRophetic tool was applied to predict the IC₅₀ of different subgroups of CSCs for different drugs. (A-C) Sensitivity of the CD13-positive subgroup to etoposide, GDC0941, and pazopanib; (D,E) sensitivity of the CD133-positive subgroup to OSI.906 and lapatinib; (F-I) sensitivity of the CD90-positive subgroup to midostaurin, CMK, AZD7762, and BX.795. CSC, cancer stem cell; IC₅₀, half maximal inhibitory concentration.