Comprehensive Profiling of Somatic Mutations Which Define Primary Disease and Relapse in Various Acute Leukemia Subtypes

Abstract 783

Acute leukemia comprises several distinct subtypes. Apart from certain specific cytogenetic abnormalities associated with each of these subtypes, acquisition of additional somatic mutations is crucial to disease development. The present study is a comprehensive exploration of somatic mutations associated with several leukemia subtypes to catalog pattern of genetic changes associated with each category of leukemia. Moreover, the focus of the study is on identifying relapse specific mutations, thereby enabling us to understand the evolutionary relationship between clones present in primary leukemia and relapse.

Whole exome sequencing was performed on 5–10 matched diagnosis, relapse and remission samples from each AML subclass characterized by either t(8;21), t(15;17), FLT3-ITD, MLL-PTD or normal cytogenetics (without FLT3-ITD and MLL-PTD) AML. Identification of somatic mutations and small indels was carried out using an automatic analysis pipeline which included 1) genome mapping using BWA, 2) calling for variant sites using SAMTOOLS, 3) annotation using ANNOVAR, and 4) filtering using Fisher exact test and sequence quality information. Candidate mutations were validated using Sanger sequencing and 5–10 nonsynonymous mutations per sample were found in each subtype, except the normal cytogenetics samples where 25–30 mutations were found, signifying a higher mutational load. In all subtypes, excluding APL, a large proportion of the mutations present at diagnosis were retained in relapse, suggesting a similar pattern of disease progression. In t(8;21) AML, we sequenced 10 paired diagnosis and relapse cases and identified 47 mutations shared between the two stages, while 10 mutations were diagnosis-specific and 35 were relapse-specific. Recurrent mutations were observed in KIT, TET2, TTN, and MGA genes in this subtype. In FLT3-ITD AML, 60 mutations in 50 different genes were found in 10 relapse and diagnosis cases. Amongst 50 mutations present at diagnosis, 38 were shared with the relapse samples. We observe recurrent mutations in several known genes including DNMT3A, WT1, RUNX1 and IDH1. In 5 cases of MLL-PTD AML, 42 genes were mutated, 24 of which were shared at both diagnosis and relapse. Similarly, a high prevalence of primary disease-associated mutations was also detected at relapse in normal cytogenetics AML, indicating persistence of disease clones during chemotherapy. In 8 cases of APL, 63 mutations in 61 genes occurred, with mutations in WT1 and PML appearing twice. Interestingly, while 4 cases shared at least one mutation between diagnosis and relapse stages, in the remaining 4 cases, the relapse resulted from independent clones.

In acute lymphocytic leukemia exomes from 7 matched diagnosis and relapse pediatric cases, which should have done well but did poorly (normal cytogenetic or hyperdiploid cases), a similar rate of somatic mutations occurred as observed in myeloid malignancies. Interestingly, no non-synonymous mutations were detected in primary disease samples in 4 out of 7 cases, but each accumulated multiple mutations at relapse.

The frequency of mutations in the newly identified mutant genes is currently being examined in a large cohort of diagnosis/relapse samples using custom target capture probes. We are integrating the exome sequencing data with gene expression and SNP array data for further selection of potential driver mutations. Pathway analysis is also being performed using the validated gene list from each leukemia subtype to explore affected biological processes. Finally, clonal evolution analysis using density-based clustering algorithm (DBSCAN) is being carried out to study the emergence of disease clones and its progression from primary disease to relapse stage. To our knowledge, this is the first large scale analysis of relapse leukemia in a variety of subtypes.