Comprehensive Genomic Classification of Pediatric Mixed-Phenotype Acute Leukemia

Mixed-phenotype acute leukemia (MPAL) is defined by a blast population that expresses a specific combination of lineage-defining myeloid antigens and lineage-defining T- or B-lymphoid antigens.^1-3  The 2008 World Health Organization’s (WHO) classification further categorizes MPAL by specific chromosome rearrangements involving BCR-ABL1 fusion and KMNT2A (also known as MLL), but since the genetic basis of most cases of MPAL is unknown, most cases are classified as one of the following lineage mixes: B/myeloid MPAL, T/myeloid MPAL, or MPAL NOS.³ 

MPAL, which accounts for approximately 3 percent of acute leukemia cases in children and adults, is associated with a greater propensity for intrinsic resistance to chemotherapy and thus, has a poor prognosis.^4,5  Because most patients with MPAL are excluded from all acute myeloid leukemia (AML) and acute lymphocytic leukemia (ALL) frontline clinical trials, the optimal treatment for MPAL remains uncertain. Retrospective analyses indicate that the outcomes for patients with MPAL are better when ALL-based chemotherapy regimens are used, but inferior to outcomes for equivalent pediatric or adult ALL and AML cohorts.^6-8  Allogeneic hematopoietic stem cell transplantation (HSCT) in first remission improves outcomes and is recommended for patients with MPAL.⁹  Recent efforts have applied comprehensive genomic tools to MPAL to better understand its biology and provide insight as to why current therapies are less effective, as well as to develop better therapies for this group of poorly responsive patients with acute leukemia.

To this end, four reports published in 2018 have sought to characterize the genetics underlying MPAL.^10-13  An international collaboration of pediatric MPAL cases stands out and is reported by Dr. Thomas B. Alexander and colleagues in Nature.¹⁰  The authors present the results of their comprehensive genomic analyses of pediatric MPAL and compare its “genomic landscape” to other pediatric acute leukemia subtypes. Additionally, they used cell sorting coupled with targeted sequencing to characterize the cell of origin for pediatric MPAL.

The authors analyzed 115 samples from patients with MPAL (T/M MPAL [n=49], B/M MPAL [n=35], KMT2Ar MPAL [n=16], BCR-ABL1 MPAL [n=2], and MPAL NOS [n=8], acute undifferentiated leukemia [n=5]) by whole exome or whole genome sequencing, transcriptome sequencing, single-nucleotide polymorphism array analysis, and methylation array analysis. The comparison cohorts included pediatric patients with AML, early T-cell precursor (ETP)-ALL, non-ETP T-ALL, and B-ALL.

They identified 158 recurrently altered genes with 81 genes identified in at least three cases. The most commonly mutated genes identified in pediatric MPAL are also commonly mutated in AML (FLT3, RUNX1, and CEBPA) and ALL (CDKN2A, CDKN2B, ETV6, and VPREB1), or both (WT1, KMT2A). Similar to studies of AML and ALL, KMT2A cases demonstrated the lowest mutation burden, and the mutation burden for T/M MPAL and B/M MPAL was similar to that of AML and ALL. Alterations in genes encoding transcriptional regulators were detected in the majority of T/M MPAL and B/M MPAL cases, with alterations in WT1, RUNX1, and CEBPA mainly occurring in T/M MPAL compared to ZNF384, VPREB1, TCF3, and PAX5 in B/M MPAL. Alterations in signaling pathway genes were common in all subsets of MPAL with mutations in FLT3 being the most common signaling mutation and occurring in all subsets, but more frequently in T/M MPAL. Other mutations in signaling pathways such as NRAS and KRAS were distributed across all MPAL subsets. Mutations in genes encoding epigenetic regulators occurred in 69 percent of T/M MPAL cases and 63 percent of B/M MPAL cases. For KMT2A cases, in addition to translocation involving KMT2A, mutations in MLLT3 and AFF1 were identified in more than one case.

A recurrent translocation involving the entire coding region of ZNF384 was present in nearly half the cases of B/M MPAL and in one KMT2A MPAL case. The genomic landscape and the gene expression profiles of ZNF384-associated B/M MPAL were similar to that of ZNF384-associated B-ALL except for KDM6A alterations, which were only identified in the MPAL cases. The authors propose that ZNF384 rearrangements define a distinct subtype of MPAL for consideration in future studies.

Genomic comparison studies were performed on cases of T-cell ALL, T/M MPAL, ETP-ALL, and AML. The authors found that core transcription factors known to drive T-ALL (TAL1, TAL2, TLX1, TLX3, LMO1, LMO2, NKX21, HOXA10, and LYL1) were less frequently altered in T/M MPAL and ETP-ALL, while genetic alterations MYB, LEF1, CDKN2A, CDKN2B, and NOTCH1 common to T-ALL were rare in T/M MPAL and ETP-ALL. Conversely, WT1 alterations were common in T/M MPAL and ETP-ALL, but uncommon in T-ALL. Gene expression profiling of T/M MPAL and ETP-ALL revealed similar expression patterns. The authors concluded that the extensive overlap of genomic alterations in T/M MPAL and ETP-ALL suggests a common progenitor cell of origin.

The co-expression of surface antigens associated with both myeloid and lymphoid lineages, as well as an enrichment in stem cell signatures in some cases of MPAL, raises the question of the cell of origin for MPAL. The authors examined the mutational burden of highly enriched hematopoietic stem and progenitor cell compartments in primary samples from patients with MPAL. In the cases examined, the authors consistently demonstrated a uniform distribution of case-specific somatic mutations between the malignant lymphoid and myeloid compartments of the leukemia. These findings are consistent with a model in which distinct combinations of genetic drivers target the primitive hematopoietic stem and progenitor compartments, giving rise to the clinical entities of MPAL, T/M MPAL, and B/M MPAL.