Unsupervised clustering of RNA sequencing data identifies a novel XBP1-mutated subtype of B-cell acute lymphoblastic leukemia (B-ALL) Free

1. Research Objectives :With the application of RNA sequencing (RNA-seq) in B-ALL, the diagnosis of B-ALL has entered the era of precision molecular diagnostics. Despite almost 90% of newly diagnosed B-ALL cases can be diagnosed as the defined molecular subtypes with the integrated analysis of RNA seq, approximately 10% of patients still lack identifiable driver mutations due to the complex molecular pathogenesis of leukemia. Therefore, this study aims to:Using unsupervised clustering methods to find the potential novel subtypes by retrospectively analyzing RNA seq data in B-ALL,and identifing the potential molecular therapeutic targets.

2. Research Methods:This study included RNA seq data from 318 Ph-negative B-ALL​. We performed integrated data analysis of RNA-seq (including gene fusions, gene mutations, gene expression and B-ALL subtype prediction based on expression profiles)and classified B-ALL molecular subtypes according to the latest ICC guidelines.To further explore potential novel subtypes, we conducted unsupervised clustering analysis on the 318 RNA seq data.PCA and t-SNEwas applied for dimension reduction to. Random forest was used to screen the top 1000 most important genes.The DBSCAN algorithm was used for clustering.Differential expression genes(DEG) analysis and KEGG pathway enrichment analysis between novel subtype and healthy controls are performed to identify specific gene expression patterns and related biological pathways.

3. Research Results:(1) Clustering Results of 318 B-ALL Samples:The clustering results showed an Adjusted Rand Index (ARI) of 0.618 and a Silhouette Score of 0.517, indicating reasonable clustering quality.The unsupervised clustering analysis divided the 318 RNA-seq samples into 12 clusters described below:Cluster 0 (ZNF384-rearranged, 37/37),Cluster 1 (KMT2A-rearranged, 23/23),Cluster 3 (DUX4-rearranged, 28/28),Cluster 4-1(IDH1/2-mutated, 9/9),Cluster 4-2 (MEF2D-rearranged, 17/17),Cluster 5 (TCF3::PBX1-rearranged, 8/28),Cluster 6 (hyperdiploid subtype, 29/35),Cluster 7 (Ph-like, 19/20),Cluster 8(ETV6::RUNX1-rearranged, 30/33).Cluster 2 and Cluster 9 showed weak expression signatures, containing multiple subtypes (mainly PAX5-alt (32/81), hypodiploid(17/81),KMT2A-rearranged(11/81)).Cluster -1 (4/4) and Cluster 10 (4/4)were unclassifiable.Among them,Cluster -1 samples had non-distinctive expression features and were scattered across different clusters, suggesting biological heterogeneity.Interestingly, Cluster 10 samples formed a distinct cluster, indicating homogeneous biological characteristics, suggesting a potential novel subtype. (2) Identification of the XBP1-Mutated Subtype:Analysis of molecular alterations in unclassifiable Cluster 10 revealed that all 4 samples harbored frameshift mutations in the XBP1 gene, suggesting that XBP1 mutations may drive this novel subtype.(3) Molecular and Pathway Characterization of the XBP1-Mutated Subtype :DEG (Cluster10 vs Healthy controls)analysis showed that 7,056 genes were significantly downregulated and 3,326 genes were significantly upregulated, including: CD109, TMC7, LILRA3, INSL5, and ABHD4​ (key upregulated genes). LILRA3, INSL5 and ABHD4 are associated with immune regulation,glycometabolism, and lipid metabolism. CD109 and TMC7 have been reported in multiple cancers and play roles in tumor immune microenvironment regulation; current reseaches indicate that CD109 may be a potential therapeutic target. KEGG pathway enrichment analysis showed that neuroactive ligand-receptor interaction, hormone signaling, cytokine-cytokine receptor interaction, cAMP signaling pathway, and Stem cell pluripotency pathways​ were significantly enriched.These pathways suggest that immune microenvironment dysregulation and small-molecule metabolic alterations may drive B-ALL through cAMP pathway activation. cAMP signaling and CD109 could be a potential molecular therapeutic targets for this subtype.

4. Research Conclusions:Using the integrated RNA-seq analysis and unsupervised clustering, this study identified one potential novel B-ALL subtype characterized by XBP1 mutations. DEG and pathway enrichment analysis revealed the unique molecular and biological features of this subtype, providing a foundation for future exploration of the diagnostic biomarker and targeted therapy regarding to this novel subtype.