Whole Transcriptome Sequencing In Refractory T-Cell Acute Lymphoblastic Leukemia

T-cell acute lymphoblastic leukemia (T-ALL) is a malignancy of the lymphoblasts committed to the T-cell lineage. Despite the therapeutic improvements witnessed over the years, ∼25% of children and ∼50% of adults still show a poor long-term outcome. While many recurrent oncogenic lesions have been identified through the characterization of chromosomal aberrations and candidate gene sequencing, several observations indicate that additional genetic alterations, not evident by conventional cytogenetics, might influence leukemogenesis and treatment outcome. Improvement of our knowledge in the identification and characterization of new oncogenic genome variations is expected to lead to a better prognostic classification and should also allow the design of tailored therapeutic strategies.

To get further insights into the molecular pathogenesis of T-ALL and to identify novel markers for risk stratification and treatment improvement, we performed whole transcriptome sequencing (RNA-seq) on 18 refractory T-ALL cases sampled at diagnosis (median age 37.5 years, range 11-55). A pool of normal thymus cells was used as negative control. Next generation sequencing libraries were constructed from the mRNA fraction, followed by paired-end sequencing on a HiSeq2000 (Illumina). Sequence reads were aligned to the reference genome and were processed to identify gene expression levels, gene fusion transcripts and single nucleotide variations (SNVs).

We first determined accurate gene expression levels from the RNA-seq data and used them to classify patients into T-ALL subtypes. Next, we applied the deFuse algorithm to detect fusion transcripts. Fusion transcripts detected also in normal thymus cells were filtered out, as well as fusions involving ribosomal genes. After applying these filters, we obtained 407 fusion transcripts (average: 22.6/sample, range: 0-84) predominantly involving genes localized on the same chromosome and mostly generated by deletion (306/407). Novel candidate fusion transcripts were confirmed by RT-PCR and Sanger sequencing. The SET-NUP214 fusion was identified in 2 cases, as well as 2 novel fusion transcripts involving the T-cell receptor (TCR) genes and not detected by conventional cytogenetics: the first fusion resulted in a chromosomal rearrangement between HOXA-AS4 and TRBC2 (also accompanied by overexpression of the HOXA genes) and the second between TRAC and SOX8 (associated with SOX8 overexpression). Interestingly, we also found out-of-frame fusion transcripts leading to the potential inactivation of tumor suppressor genes, such as PTEN-FAS and MAST3-C19orf10.

Finally, we performed SNV calling on our dataset. After removing the most common polymorphisms, we obtained 1,483 protein-altering SNVs (missense, nonsense mutations and mutations affecting splicing), ranging between 30 and 131 per sample, with 85 genes that contain a protein-altering mutation in at least 3 of the 18 samples (i.e. 16% of cohort). Members or modulators of NOTCH and JAK/STAT pathways were the most recurrently mutated, each accounting for ∼38% of cases. In particular, 7/18 samples showed previously reported lesions in the NOTCH1 (n=5) and FBXW7 (n=1) genes but also in novel candidates as NOTCH2 (n=1), NOTCH3 (n=1) and SPEN (n=1). Interestingly, 1 patient showed 2 different mutations in the exon 26 of NOTCH1, while in 2 samples NOTCH1 mutation was associated with mutations in NOTCH2 or NOTCH3. Similarly, the JAK/STAT pathway was affected in 7/18 samples, including JAK1 (n=2), JAK3 (n=5), TYK2 (n=1) but also the novel candidates STAT5A (n=2) and STAT6 (n=1). Four of the 5 JAK3-positive patients showed also a mutation in another gene of the same pathway, such as JAK1 (n=1), STAT5A (n=2) and STAT6 (n=1). Thus, mutational screening of both the NOTCH and JAK/STAT pathway shows that mutations can occur simultaneously and suggests that more than one lesion is required for leukemic transformation.

In conclusion, RNA-seq appears as a promising tool to dissect the heterogeneity of T-ALL and to identify targets that might be useful for tailored therapeutic interventions. Further investigations are ongoing to determine the recurrence and specificity of these lesions, and their potential in inducing a refractory phenotype. Finally, in vitro experiments will be carried out to investigate the transforming capability of specific lesions and the targettability of the recurrently impaired pathways.