Analysis of Clonal Evolution in Chemorefractory Acute Myeloid Leukemia from Diagnosis to Relapse

Introduction

Acute myeloid leukemia (AML) is a genetically heterogeneous disease characterized by significant clonal evolution during disease progression, explaining why adult AML remains a disease that is often difficult to cure. It is necessary to therapeutically target all relevant AML cell subclones in a patient to prevent future relapses. Hence, understanding of the clonal genetic evolution of cancer cells in each AML patient is critical to develop curative therapeutic approaches.

In our individualized systems medicine (ISM) initiative, we have performed ex-vivodrug sensitivity and resistance testing (DSRT) with a comprehensive set of 306 cancer drugs on primary cells from 34 AML patients (updated, Pemovska et al, Cancer Discovery, 2013). Based on the DSRT as well as sequencing data, selected patients were treated with novel targeted therapies and clinical responses were observed in a significant proportion of patients. Serial samples were obtained from patients during diagnosis, resistance after chemotherapy and relapse after novel targeted treatments. Here, we applied next-generation sequencing from 67 such samples from 23 AML patients in order to a) follow clonal progression of adult AML in patients during treatment with novel targeted drugs, b) identify if clones emerging at relapse were present as small subpopulations before therapy.

Methods

Clonal composition at a given time point was estimated using kernel density estimates of somatic single nucleotide variants (SNV) and copy number alterations (CNA) identified using exome sequencing. CNAs (amplifications, deletions and loss of heterozygosity variants) were integrated with the SNVs for the estimation of sub-clonal populations. With the availability of serial samples, subclones were followed up during relapse and subsequent targeted therapy.

In order to identify, if relapse-specific clones pre-existed as a minor subclone at diagnosis or emerged as a novel clone, high confidence mutations detected in post-treatment and relapse samples (on average 10-15 variants per patient) were validated in the diagnostic and pre-treatment samples using ultra-deep amplicon sequencing assay (coverage of 10000-100,000x). This assay enables detection of low-frequency subclones with very high sensitivity and less bias.

Results

Integrated data from somatic single nucleotide and copy number alterations indicated significant clonal heterogeneity in relapsed AML and sub-clonal evolution during disease progression. We saw two trends in subclonal evolution following relapse, i) a preexisting clone was enriched during treatment or subsequent to drug resistance and became the driving clone, or ii) a new subclone not detectable at diagnosis emerged at relapse. For example, treatment of an AML patient with a combination of dasatinib-sunitinib-temsirolimus led to the selection of an already existing low-frequency subclone carrying ETV6-NTRK3fusion. In another patient case, a new subclone consisting of BRCA1, PTPN11 and MLH1 mutations emerged as a driving clone at relapse. This clone could not be detected in the pre-treatment samples even after validation.

We observed higher number of CNAs per sample in the chemorefractory AML in comparison to primary AML samples. Among the diagnostic samples only 2/8 showed significant CNAs, with one sample having a focal deletion of TET2and another sample showed amplification of chromosome 8. In contrast, in the chemorefractory cases, many additional CNAs were acquired following relapse and drug resistance (in comparison to the matching diagnostic and pre-treatment samples), which may imply a role for genomic instability in the development of resistant disease state.

Conclusions

Availability of serial samples before and after treatment enabled us to follow clonal evolution and to differentiate subclones that emerged specifically at relapse from subclones present already at diagnosis. It is critical to ascertain the genomic clonal composition during AML progression in each patient as well as to determine the effect of drug treatments at subclone level. Curative therapies that block the growth of the multiple AML subclones need to be tailored for each patient.