Next Generation Genomic Analyses in T-ALL Patients Identify Recurrent and Novel Genomic Abnormalities

Background and Objective

15% of pediatric and 40% of adult T-cell Acute Lymphoblastic Leukemia (T-ALL) patients fail conventional therapy highlighting the need for novel therapeutic strategies based on genomic alterations of individuals. We interrogated genomic alterations in Australian T-ALL patients for patterns of mutation and druggable targets. A subset of samples were used to establish patient-derived xenograft (PDX) models to evaluate novel therapies.

Methods

T-ALL patients' samples underwent next generation genomic analyses (128 total samples including diagnosis, refractory and relapse timepoints from 118 patients of all age groups). mRNA sequencing (mRNAseq) identified gene fusions and structural variants and assessed gene expression (n=101 patients). Fusions were called when identified by 2/3 predictors (FusionCatcher, SOAPfuse, JAFFA). Variant calling utilized GATK HaplotypeCaller and underwent several filtering steps to eliminate possible germline alterations and common SNPs. DNA copy number variations (CNVs) were detected via Multiplex Ligation-dependent Probe Amplification (MLPA: P202, P335, P383; n=64 patients). Establishment of PDX models from patient material (bone marrow or peripheral blood) is ongoing. Recapitulation of human disease was confirmed by mRNAseq in a subset of xenografts.

Results

Genomic fusion genes were identified in 46/101 samples (46%) by mRNAseq; the most common fusion identified was STIL-TAL1 (n=6). Increased expression of LCK and/or LAT (encoded proteins are involved in T-cell receptor (TCR) signal transduction) was observed in 100% of patients with the STIL-TAL1 gene fusion indicating TCR signaling pathways may be perturbed in this sub-group. Other common gene fusions were MLLT10-DDX3X (n=5) and KMT2A-MLLT4/AFDN (n=4). We also observed the previously reported fusions SET-NUP214, KMT2A-MLLT1, PICALM-MLLT10, NUP214-ABL1, several fusions involving TCR subunits as well as novel fusions involving KMT2A, NOTCH1, LMO1, ZEB2.

Numerous nonsynonymous mutations were identified in 81/101 patients (80%) with mRNAseq data available (Figure 1). Broadly, the mutated genes encode proteins in the following categories: oncoproteins (NRAS, KRAS); tumour suppressors (TP53, CHEK2, BRCA1, BRCA2, PTEN), epigenetic regulators (EZH2, SETD5, DNMT3A); regulators of NOTCH signalling (NOTCH1, NOTCH2, FBXW7); transcription factors and regulators (IKZF1, KMT2A, EP400, SMARCD1, RUNX1, AFF1, AFF3); kinase and cytokine signal regulators (ATM, JAK2, JAK3, TYK2, FLT3, PTPN11). INDELS in clinically relevant genes were identified in 37/101 patients (37%) including alterations to: NOTCH1, PHF6, PTEN, STAT1, IL7R, CDKN2A, LYL1, WT1, JAK3, LEF1.

The most common copy number alterations identified in our patient cohort were CDKN2A/B deletions (30/64 patients, 47%), PHF6 duplication (20/64 patients, 31%) and MLLT3 deletion (13/64 patients, 20%). In patients with CDKN2A/B deletions and additional CNVs, PHF6 duplication (n=6) and MYB duplication (n=4) were mutually exclusive. However, one patient without CDKN2A/B deletions harbored both MYB and PHF6 duplications. MLLT3 deletion always co-occurred with CDKN2A/B deletions (13/13 patients with MLLT3 deletion), but was never observed with PTEN deletion (0/7 patients with PTEN deletion). Patients with either CDKN2A/B deletions or PHF6 duplications frequently harbored NOTCH1 abnormalities: 13/32 patients (41%) and 8/21 patients (38%), respectively.

PDX primagrafts investigated the engraftment latency and peripheral organ infiltration. Primagrafts established from patients harboring NUP214-SET1 or NUP214-ABL1 fusions engrafted at a slower rate (82 d and 91 d, respectively) than primagrafts from patients harboring a STIL-TAL1 fusion (45 d) or CDKN2A/B deletions (31 d and 48 d).

Conclusions

In our T-ALL cohort we demonstrate that the majority of cases harbor rearrangements, structural variations and duplication/deletion of genes associated with malignant transformation. We identified several co-occurring lesions as well as mutually exclusive genomic abnormalities. The top 20 mutated genes in our patient cohort differ to those reported for a pediatric cohort (Roberts et al 2019 Blood 134:649), indicating an association between patient age and genomic alteration. Secondary PDX models investigating novel targeted treatment strategies are ongoing.