Somatic and Germline Mutational Heterogeneity in High Hyperdiploid Acute Lymphoblastic Leukemia

High hyperdiploidy (HD), the most common cytogenetic subtype of acute lymphoblastic leukemia (ALL), is characterized by a nonrandom gain of chromosomes and is thought to arise from a single abnormal mitosis. However, the causes of this leukemia-initiating event remain unknown. A recognized enrichment of HD-ALL among children with RASopathies and with germline ETV6 mutations suggests that germline predisposition underlies a component of HD-ALL risk, in addition to the increased risk of the HD subtype with ALL-associated heritable risk variants in PIP4K2A, ARID5B, and CEBPE. Though cure rates of HD-ALL are high, the significant treatment-related morbidities and mortality warrant more etiologic investigations which may reveal molecularly-targeted therapies for this disease.

We carried out deep-sequencing of 538 cancer-related genes using the UCSF500 Cancer Gene Panel in 57 HD-ALL tumors from California Childhood Leukemia Study patients. Selected patients lacked overt KRAS and NRAS hotspot mutations (assessed by Sanger sequencing) and common ALL deletions (assessed by MLPA), to enrich for discovery of novel driver genes. A Combined Annotation Dependent Depletion (CADD) Phred score ≥20 was used to filter predicted damaging mutations. To remove polymorphisms, we retained only mutations with allele frequency <0.01% in the Exome Aggregation Consortium (ExAC). We adjusted the mutant allele fraction (MAF) of each mutation in relation to chromosome copy-number, as determined using the CNVkit tool. Sanger sequencing of remission DNA was used to validate a subset of predicted germline mutations (adjusted MAF≥0.45) of interest, including in known ALL predisposition genes.

Novel damaging somatic mutations were discovered in epigenetic regulatory genes, including DOT1L (n=4), with 33% of patients harboring mutations in this pathway. Somatic mutations in the receptor tyrosine kinase (RTK)/Ras/MAPK signaling pathway were found in two thirds of patients, including mutations in ROS1, which mediates phosphorylation of the PTPN11-encoded protein SHP2. An extraordinary level of tumor heterogeneity was detected, with microclonal (mutant allele fraction <10%) hotspot mutations in KRAS, NRAS, FLT3 or PTPN11 identified in 31/57 (54.4%) patients. Multiple microclonal mutations at KRAS and NRAS codons 12 and 13 significantly co-occurred within tumor samples (P=4.8x10^-4), suggesting ongoing formation of, and selection for, Ras mutation. Moreover, 7 patients had multiple microclonal mutations at the same Ras hotspot locus, in adjacent codon 12/13 nucleotides or in adjacent codons. The adjacent mutations occurred on different sequencing reads in all 7 patients (P=0.016), indicating they were part of distinct tumor subclones.

We also detected an unexpectedly high frequency of putatively causal germline mutations, which were validated in remission DNA samples by Sanger sequencing. At least 25% of HD-ALL patients carried one or more rare (<0.01% allele frequency in ExAC) and predicted-damaging germline mutations in known ALL predisposition genes, DNA repair genes, or within known hotspot mutation loci that had previously been reported mutated only in tumor genomes.

Future work is required to investigate whether tumor microheterogeneity should impact therapeutic regimens and to elucidate the biologic function of epigenetic dysregulation in development of HD-ALL. Whole-exome sequencing of more patients and functional analysis of novel mutations are required to understand the contribution of germline predisposition to HD-ALL etiology, which may be much larger than previously realized.