Clinical Impact of the Quantitative Subclonal Architecture in Chronic Lymphocytic Leukemia

Introduction

Recent large scale genomic studies have disclosed the heterogeneity of the mutational landscape of chronic lymphocytic leukemia (CLL). The remarkable genomic plasticity of this disease has been further emphasized by the complex subclonal composition recognized in some tumors. Initial studies using high-coverage next generation sequencing (NGS) have revealed the prognostic impact of mutations at very low allelic frequency. The results of these studies have opened a new perspective where the proportion of cells carrying specific driver mutations rather than just the presence or absence of the alterations may be relevant to understand the evolution of this disease. However, the information generated has been limited to a small subset of CLL driver genes. The aims of this study were to define the deep mutational architecture of 28 frequently altered driver genes in CLL and determine the relevance of the subclonal quantitative composition in the progression of the disease.

Methods

Highly purified tumor samples from 406 untreated CLL patients were included in this study. Ultra-deep NGS of the 28 target genes was performed using the Acces-Array system (Fluidigm) (ATM, TP53, SF3B1, BIRC3, XPO1, RPS15, FBXW7, DDX3X, POT1, KLHL6, MGA, MYD88, IRF4, BRAF, NXF1, BCOR, ZNF292, NRAS, KRAS, CCND2, TRAF3, ZMYM3, MED12) or the Nextera-XT DNA library preparation kit (Illumina) (NOTCH1, NFKBIE, EGR2, PIM1, DTX1) before sequencing in a MiSeq (Illumina). A robust bioinformatic pipeline followed by an extensive verification process allowed the detection of mutations down to 0.3% of variant allele frequency (VAF). Copy number alterations were investigated by high density SNP-arrays in 376 cases. We calculated the cancer cell fraction (CCF) carrying each specific mutation using the PyClone algorithm. The prognostic impact of the mutations was evaluated for time to first treatment (TTFT) and overall survival from the time of sampling.

Results

The mutational frequency observed for virtually all genes was higher than in similar previous studies of population based CLL at diagnosis. We detected mutations with a VAF below the Sanger sequencing threshold (VAF <12%) in 24 (86%) of the genes, corresponding to 40% of all mutations identified (median per gene of 45%, range 7-68%). Most genes showed a continuous spectrum of mutated CCFs except MYD88, KLHL6, EGR2, NOTCH1, SF3B1, and FBXW7 that displayed a bimodal distribution with most of the cases carrying either small (CCF <20%) or large (CCF >80%) mutated clones. Overall, among the 260/406 (64%) cases carrying at least one mutation in any of the genes analyzed, a major mutated clone (CCF >80%) was only identified in half of the patients (127, 49%). Convergent mutational evolution, defined as the acquisition of independent genetic mutations in the same gene, was observed in 19 (68%) of the 28 genes analyzed, being present in 66/260 (25%) mutated cases. The number of cases with convergent evolution was directly related to the global mutational frequency of the gene. The clinical relevance of the mutations appeared to be gene specific and related to the quantitative magnitude of the different subclones. We identified three major patterns of specific gene CCF that influenced the prognosis of the patient: 1) CCF independent pattern in which the mere detection of a mutation at any CCF conferred an adverse prognosis (TP53, ATM, POT1, NFKBIE, XPO1, or RPS15 among others); 2) CCF gradual pattern in which the poor prognostic impact was a continuous variable directly related to the size of the mutated clone (SF3B1); and 3) CCF clonal pattern in which the prognostic impact of the mutations was a categorical variable defined by a certain threshold of the mutated clone (NOTCH1, BIRC3, EGR2, FBXW7). On the other hand, cases with convergent mutational evolution had a tendency to a shorter TTFT when compared to mutated cases without this phenomenon.

Conclusions

In conclusion, the emergence of subclonal mutations is a general and dynamic phenomenon in CLL that seems to involve virtually all driver genes and occurs at different time points of the disease. The clinical impact of the clonal architecture of the tumor is gene specific and related to the magnitude of the respective subclone. These findings provide new insights on the relevance of the subclonal mutational profile in CLL and the importance of quantitative mutational analyses for the management of the patients.