Integrative genomic analysis of pediatric T-cell lymphoblastic lymphoma reveals candidates of clinical significance

Abstract

T-cell lymphoblastic lymphoma (T-LBL) is a heterogeneous malignancy of lymphoblasts committed to T-cell lineage. The dismal outcomes (15%-30%) after T-LBL relapse warrant establishing risk-based treatment. To our knowledge, this study presents the first comprehensive, systematic, integrated, genome-wide analysis including relapsed cases that identifies molecular markers of prognostic relevance for T-LBL. NOTCH1 was identified as the putative driver for T-LBL. An activated NOTCH/PI3K-AKT signaling axis and alterations in cell cycle regulators constitute the core oncogenic program for T-LBL. Mutated KMT2D was identified as a prognostic marker. The cumulative incidence of relapse was 47% ± 17% in patients with KMT2D mutations, compared with 14% ± 3% in wild-type KMT2D. Structural analysis of the mutated domains of KMT2D revealed a plausible impact on structure and functional consequences. These findings provide new insights into the pathogenesis of T-LBL, including high translational potential. The ongoing LBL 2018 trial (www.clinicaltrials.gov #NCT04043494) allows for prospective validation and subsequent fine tuning of the stratification criteria for T-LBL risk groups to improve survival of pediatric patients.

© 2021 by The American Society of Hematology.

Figures

Graphical abstract

Figure 1.

Genome-wide screening for recurrent genetic and epigenetic alterations in T-LBL. (A) Overview of the top 50 mutated genes identified by WES in 15 pediatric T-LBL samples including 5 corresponding relapse samples. The relative mutation frequencies (sample-wise) are indicated to the left and the pathways to the far left. Gene names and types of mutation are to the right. The number of mutations is displayed at the top in a bar plot. (B) The top 55 differentially methylated genes enriched by significant CpGs in primary samples in comparison with their corresponding germline samples (TG vs TP 1e−4). (C) The top 55 differentially methylated genes enriched by significant CpGs in relapse samples in comparison with their corresponding germline samples (TG vs TR 5e−7). Array used: Infinium MethylationEPIC BeadChip 850K. (D) The frequency of CNAs detected in 22 T-LBL samples including select primary and matched relapse samples. CNAs in the analyzed samples are displayed against the chromosomal numbers and position. The CNA profiles for primary relapse− samples (TP1-10), for primary relapse+ samples (TP11-16), and matched relapse samples (TR11-16) are presented separately. Array used: Infinium Omni2.5Exome-8 v1.3. Results validated independently by MLPA for specific regions in Chr 1p, 4q, 6q, 9p, 10q, 11p, and 17q. Red: amplifications; blue: deletions; yellow: loss of heterozygosity.

Figure 2.

Mutational spectrum of T-LBL. An overview of recurrently mutated genes (VAF cutoff, >10%) and alterations in T-LBL. The samples are sorted into 3 sections: primary samples from relapse− patients, primary samples from relapse+ patients, and relapsed cases, as indicated at the bottom of the panel. The frequency of mutations (Mut. Freq.) and name of the pathways is indicated to the left. The names of the genes and the type of mutations are indicated on the right. The number of mutations identified by targeted sequencing is displayed at the top of the plot as bar plots. TRG rearrangements (ABD), PTEN deletions (PTENdel), and LOH6q alterations are displayed in a subpanel below. *BCL11b was not part of the targeted panel, but was analyzed by Sanger sequencing. Samples that failed in targeted sequencing are represented in dark gray boxes.

Figure 3.

Schematic display of localization and frequencies of mutations in KMT2D and prognostic value of the most relevant candidate genes. Schematic display of localization and frequencies of mutations identified for KMT2D in total T-LBL samples (VAF cutoff, >1%) (A), KMT2D in relapse+ samples (B). The data for mutations (A-B) in pediatric patients with T-ALL were imported from the St Jude Pediatric Cancer Genomic data portal and are displayed in the lower part of the schematic structure, and the mutations identified in the current T-LBL project are displayed in the upper part.

Figure 4.

Five-year cumulative incidence of relapse.KMT2D mutational status (A), KMT2D and PTEN mutational status (B), KMT2D and PTEN mutational status on an N/Fwt background (C) and KMT2D and PTEN mutational status on an N/Fmut background (D). A VAF cutoff of >10% and only nonsynonymous mutations were included for analysis. In case of N/Fmut, only mutations from the hotspots (exons 26, 27, and 34 for NOTCH1 and exons 9, 10, and 12 for FBXW7) were included.

Figure 5.

Modeling of KMT2D domain structures and comparison with the mutated domain structures. (A) Schematic overview of the predictive value of intrinsically disordered regions in KMT2D. The KMT2D gene is aligned with the PONDR (Predictor of Naturally Disordered Regions) chart to visualize mutations that lie in the predicted disordered region. (B) The domain structure of PHD7 of KMT2Dwt was generated by Phyre2. The schematic display of Arg residues (cyan) are represented in the stick structures near PHD7 of KMT2Dmut at the respective positions R5153P and R5154Q (cyan), the latter having been identified in T-ALL (red). (C) The crystal structure of the FYR domain, superimposition of Phrye2 depiction of the KMT2D (magenta), FYRN (pink), and FYRC (orange) motifs with the crystal structure of the TBRG1 motif (PDB: 2WZO). The residues forming the core hydrophobic pocket (yellow) are shown with stick structures and display of the Ileu5232 (cyan) in the FYRN motif. (D) The mutation Ileu5232Val (cyan) leads to loss of CH3, which could result in destabilizing of the domain. (E) Crystal structure of KMT2D SET domain (PDB: 4Z4P). The mutation Ser5476 (cyan) is located near the SAM binding cleft (SAH is modeled) and makes direct contact with a water molecule that supports Cys5532 (yellow, stick structure) located in the post-SET loop involved in zinc binding (purple). (F) The Ser5476Pro mutation impairs water binding, which could destabilize the Cys-zinc interaction required for proper folding.