Comprehensive Analysis of Mixed Phenotype Acute Leukemia (MPAL) By Whole Genome Sequencing

Abstract

Background: Mixed phenotype acute leukemias (MPAL) comprise a rare subgroup of acute leukemias with significant expression of lineage-specific antigens of more than one cell lineage not allowing assignment to a single cell lineage. While two MPAL subgroups are genetically defined according to the WHO 2016 classification, i.e. MPAL with BCR-ABL1 and MPAL with KMT2A rearrangement, the majority of cases are grouped as MPAL, B/myeloid, NOS or MPAL, T/myeloid, NOS and lack any specific or characteristic genetics so far.Aims: Molecular characterization of MPAL by whole genome sequencing (WGS).Patient cohort and methods: Diagnostic peripheral blood or bone marrow samples of 12 MPAL patients (median age 64 years, range 38-96; 5 male, 7 female) were selected for WGS with 90x coverage. One had MPAL with BCR-ABL1, 5 had MPAL, B/myeloid, NOS and 6 had MPAL, T/myeloid, NOS. Library preparation was performed using Truseq DNA PCR-Free HT sample preparation kit (Illumina, San Diego, CA). Sequencing was performed on NovaSeq instruments (Illumina). Tumor-only data was aligned by BaseSpace (BS) WGS app with default parameters and (structural) variant calling was performed with the Tumor/Normal app using an unmatched reference DNA (Promega, Fitchburg, WI). Data was subsequently loaded into BS Variant Interpreter to filter and prioritize variants of interest. For the structural variants we looked at all high confidence variant calls passing filters with at least 5 paired reads and 3 split reads with a somatic QScore of >30. In addition, all calls were eliminated with a population frequency of >1% according to the DGV (Database of Genomic variants). We performed mutation signature analysis by looking at the distribution of base substitutions across the sample with a 3mer context (including the base before and after the substitution) to decompose somatic mutation patterns and compared the results to the COSMIC signature database. For mutation analysis variants with a read frequency of >0.2% and ExAC population frequency of <1% were considered. For the first set of analysis we filtered down the variant list to the coding region of 65 genes described mutated in hematologic malignancies. For the second set of analysis we considered also genes beyond these 65 genes but required that variants had an entry in the COSMIC database and were classified with a “damaging” score by PolyPhen-2 and SIFT data bases.Results: Regarding the first set of analysis we found a median of 6 (range 2-7) mutations per patient, 63 overall, involving 35 of the 65 investigated genes. The majority were missense mutations (n=31), frameshift mutations (n=14) and splice-site mutations (n=13). The genes listed in table 1 were found most frequently mutated. Except for RUNX1 none of the frequently encountered mutations affected the case with MPAL with BCR-ABL1 while there was no restriction of mutations to either of the two other MPAL subgroups. 11 of these 63 mutations had been previously analyzed by routine diagnostics applying next generation sequencing (NGS) yielding identical results.Regarding the second set of analysis without limitation on the above mentioned 65 genes we found a median of 83 (range 63-129) mutations per patient, 1079 overall, involving 736 genes. Specifically, four candidate genes were found mutated in at least five patients (table 2). NBPF1 has been described as tumor suppressor, PRAMEF2 belongs to the PRAME family of genes associated with various malignancies and as a prognostic factor in B-ALL, MUC12 is a highly polymorphous gene involved in a variety of cancer types and described to be epigenetically regulated and thus a marker for early cancer detection, ARSD has been described overexpressed and prognostic in CLL. Again, mutations were not restricted to one MPAL subgroup.We subsequently conducted a context-sensitive mutation signature analysis with the SNV data, using 96 substitution patterns rather than only 6 to increase the complexity. In all instances we found a highly similar pattern across all 12 samples with a strong bias towards C>T substitutions (figure 1).Conclusion: WGS is capable of reproducing routine NGS diagnostics and of identifying mutations not yet described for MPAL which might be used to better understand and categorize this disease. [Display omitted] DisclosuresKern:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Nadarajah:MLL Munich Leukemia Laboratory: Employment. Meggendorfer:MLL Munich Leukemia Laboratory: Employment. Walter:MLL Munich Leukemia Laboratory: Employment. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership.