Comprehensive genomic analysis reveals molecular heterogeneity in pediatric ALK-positive anaplastic large cell lymphoma

Abstract

Anaplastic large cell lymphoma (ALCL) is a mature T-cell lymphoma that accounts for 10-15% of childhood lymphomas. Despite the observation that more than 90% of pediatric cases harbor the anaplastic lymphoma kinase (ALK) rearrangement resulting in aberrant ALK kinase expression, there is significant clinical, morphologic, and biological heterogeneity. To gain insights into the genomic aberrations and molecular heterogeneity within ALK-positive ALCL (ALK+ ALCL), we analyzed 46 pediatric ALK+ ALCLs by whole-exome sequencing, RNA sequencing, and DNA methylation profiling. Whole-exome sequencing found on average 25 SNV/Indel events per sample with recurring genetic events in regulators of DNA damage (TP53, MDM4), transcription (JUNB), and epigenetic regulators (TET1, KMT2B, KMT2A, KMT2C, KMT2E). Gene expression and methylation profiling consistently subclassified ALK+ ALCLs into two groups characterized by differential ALK expression levels. The ALK-low group showed enrichment of pathways associated with immune response, cytokine signaling, and a hypermethylated predominant pattern compared to the ALK-high group, which had more frequent copy number changes and was enriched with pathways associated with cell growth, proliferation, and metabolism. Altogether, these findings suggest that there is molecular heterogeneity within pediatric ALK+ ALCL, predicting distinct biological mechanisms that may provide novel insights into disease pathogenesis and represent prognostic markers.

Fig. 1. Genomic alterations in pediatric ALCL.

Heatmap demonstrates the somatic mutation profile and copy number (CN) alterations identified in the cohort of pediatric ALK+ ALCL samples by WES, separated by gene functional groups. Only genetic alterations with presumed functional consequences are shown. Split cell for the PRDM1 and PREX2 indicates more than one genomic alteration (mutations and copy number changes). (*) indicates mutations validated by targeted sequencing or RNAseq. LH lymphohistiocytic, SCV small cell variant, NA not available, F female, M male.

Fig. 2. Gene expression profile classifies pediatric ALCL into two groups.

A Unsupervised hierarchical clustering reveals two distinct groups. The top 100 differentially expressed genes are shown as a heatmap. *Indicates samples with an EEF1G-ALK fusion. B Boxplot of differential ALK expression levels in ALCL samples detected by RNA seq analysis (P = 0.026). FPKM, fragments per kilobase of exon per million mapped fragments. C, D Pathway analysis in ALK-low and ALK-high groups (KEGG and MSigDB), p value < 0.05 and FDR < 0.05. E Differential expression represented as a volcano plot. ALK-low group shows up-regulation of cytokine and immune-related markers. ALK-high group shows up-regulation of ALK, MYC, and other proliferative markers. F xCell analysis shows differential Immune Scores in ALCL samples.

Fig. 3. Genome-wide DNA methylation classifies pediatric ALCL into two groups.

A Unsupervised clustering with bootstrapping identified two separate clusters. Each probe was then summarized to a single value for each associated genomic region. Each methylation value was annotated as Hyper-methylation or Hypo-methylation. Differential methylation was performed using Wilcoxon rank-sum test. FPKM, fragments per kilobase of exon per million mapped fragments. B Sankey diagram representation of the unsupervised hierarchical clustering result. The diagram depicts the group clustering based on RNA expression and methylation profiling. C Histogram of the methylation probe intensity (Beta value) in the ALK-high and ALK-low group. D Density scatter plot of methylation probes in ALK-Low and ALK-High samples, showing the increased methylation status toward the ALK-Low group. Density heat is shown in log count.

Fig. 4. Pathway analysis of methylated probes correlated (negative and positive correlation) with ALK expression.

Correlation analysis is based on Spearman Rank with positive correlation shown in (A) and negative correlation shown in (B). Both are filtered based on an absolute rho value cutoff of 0.75. Bar plots show enriched pathways from EnrichR’s ChEA_2016 sorted by their log-transformed p value score.

Fig. 5. Relapsed signature score in pediatric ALK+ ALCL patients.

A Differential gene expression analysis between relapse and diagnosis samples from ALCL patients. Genes differentially upregulated in relapsed samples are highlighted in green. B The relapse signature score is calculated based on single-sample GSEA. Samples were ordered based on the calculated relapse signature score. Relapse samples are colored in red. Diagnosis samples of patients who eventually relapse are colored in blue. The remaining diagnosis samples are colored in gray. The median cutoff for the diagnosis samples is indicated by the arrow. C Kaplan–Meier graph showing progression-free survival (PFS) according to the relapse signature score. The relapse signature score was categorized into HIGH and LOW based on the median level. A statistically significant difference in the Kaplan–Meier survival curves between the high and low malignancy-risk groups was determined by the two-sided log-rank test. The number of patients at risk is listed below the survival curves.

Fig. 6. Overview of clinical findings and copy number changes associated with the two ALCL groups and impact on outcome.

A Heatmap with the distribution of clinical findings and copy number alterations in the diagnostic ALCL samples; cases are assigned to ALK-low and ALK-high group based on the expression and methylation profile. Five cases were classified differently based on RNA expression and DNA methylation analysis (designated as unclassifiable). B Kaplan–Meier curve showing progression-free survival (PFS) in ALCL according to the presence of copy number changes.