Prognosis interfered with by clonal interference

In this issue of Blood, Itzykson et al report on deep sequencing of 6 receptor tyrosine kinase (RTK) genes (KIT, NRAS, KRAS, FLT3, JAK2, and CBL) in samples from a cohort of 445 patients (adult and pediatric) with core-binding factor (CBF) acute myeloid leukemia (AML) treated with intensive chemotherapy.¹ 

The authors report that the presence of clonal interference, as defined by the presence of ≥2 signaling clones in a single patient sample, and not variant allelic frequency (VAF), was prognostic of inferior event-free survival, higher cumulative incidence of failure, and worse overall survival and was independent of older age, high white cell count, and the presence of a t(8;21) cytogenetic abnormality (usually associated with poorer outcome within CBF AML).

Analysis of colonies derived from a single cell indicates that RTK mutations evolve in independent clones in AML,²  a process of clonal interference or parallel evolution, and thus the definition of clonal interference as used in the article by Itzykson et al is justified. Clonal interference was associated with older age and inv(16) cytogenetic abnormality but was not associated with lower remission rate and, more interestingly, was not associated with higher levels of minimal residual disease (MRD). Secondary mutations involving RTKs are common in CBF AML and have been reported to be prognostic alone (eg, KIT) or in combination (eg, KIT, RAS, FLT3).^3,4  Other articles associate inferior outcome not with presence of RTK mutations but with their allelic burden.⁵  Although epigenetic (eg, ASXL2), cohesin and other types of mutations have been associated with worse outcomes in CBF AML,⁶  the presence of these mutations did not erase the poorer impact of clonal interference in their analysis.

Clonal interference, initially implicated in asexually dividing populations, is defined as the simultaneous spread of multiple beneficial mutations in a population in contrast to the traditional view of rare beneficial mutations that survive drift and increase in frequency. Clonal interference indicates movement of multiple mutations as temporal clusters within a population thus escaping selection drift. One potential limitation of the analysis by Itzykson et al is that the presence of clonal interference is inferred on the basis of the coexistence of multiple RTK mutations and not single-cell sequencing.

Quantitative monitoring of MRD has been one of the most important prognostic parameters in CBF AML, because the presence of unique translocations make this subgroup of AML amenable to molecular monitoring of MRD.^7-9  Although MRD after 2 cycles of treatment was not impacted by presence of clonal interference in the analysis by Itzykson et al, MRD data were available in a smaller cohort (limited to after 2 cycles of therapy) and were included in a bivariate analysis only. Thus, the question of whether clonal interference is prognostic completely independent of serial MRD data has not been answered.

Effective treatment can overcome the impact of prognostic factors. In CBF AML, more intensive regimens have resulted in very impressive relapse-free survival but no overall survival.¹⁰  This raises the question, Can such regimens overcome the adverse impact of clonal interference in a genetically less complex disease such as CBF AML, which is considered sensitive to high-dose nucleoside analogs? The variability of number and intensity of consolidation therapies in the article by Itzykson et al limits the ability to answer this question.

Their article brings up more interesting biological questions. CBF AML is generally considered to have a less complex genomic architecture and have good outcomes with standard high-intensity chemotherapy. Thus, their article is important in highlighting the impact of intratumoral heterogeneity in a leukemia with a relatively simple genomic background. The higher risk of relapse with clonal interference raises 3 possibilities: high overall RTK mutation burden, a subset of mutations that are resistant to chemotherapy, and mutational instability. Although other articles associate higher allelic burden of RTK mutation with worse outcome, which implies the importance of the first possibility, the authors have tried to rule out the first possibility by showing that VAF does not matter and the second possibility by bootstrapping analysis, and they suggest that clonal interference is a likely proxy for mutational instability. Conceptually, these findings also pose the challenge of unearthing vulnerabilities of mutational architectures rather than single mutations. Single-cell sequencing of residual disease or serial sequencing of uniformly treated cohorts may answer some of these questions.