The hidden genomic landscape of acute myeloid leukemia: subclonal structure revealed by undetected mutations

Abstract

The analyses carried out using 2 different bioinformatics pipelines (SomaticSniper and MuTect) on the same set of genomic data from 133 acute myeloid leukemia (AML) patients, sequenced inside the Cancer Genome Atlas project, gave discrepant results. We subsequently tested these 2 variant-calling pipelines on 20 leukemia samples from our series (19 primary AMLs and 1 secondary AML). By validating many of the predicted somatic variants (variant allele frequencies ranging from 100% to 5%), we observed significantly different calling efficiencies. In particular, despite relatively high specificity, sensitivity was poor in both pipelines resulting in a high rate of false negatives. Our findings raise the possibility that landscapes of AML genomes might be more complex than previously reported and characterized by the presence of hundreds of genes mutated at low variant allele frequency, suggesting that the application of genome sequencing to the clinic requires a careful and critical evaluation. We think that improvements in technology and workflow standardization, through the generation of clear experimental and bioinformatics guidelines, are fundamental to translate the use of next-generation sequencing from research to the clinic and to transform genomic information into better diagnosis and outcomes for the patient.

Figure 1

Comparison between the mutational analyses performed with the MuTect and SomaticSniper bioinformatics pipelines in 133 AMLs.^, (A) The same 133 AML samples were analyzed by WES and subsequently by MuTect or SomaticSniper bioinformatics pipelines: for every patient, the number of mutations identified with the 2 methods is reported on the x (SomaticSniper) and y (MuTect) axes; red points correspond to outliers (patients having a very high number of mutations in 1 or both methods). The Pearson correlation coefficient (r) was calculated for all samples (r = 0.08) or after removal of the outliers (r = 0.3). Both r values indicate a significant discordance between the 2 methods. The black dashed trend line indicates the expected number of identified mutations assuming that the analysis with the 2 pipelines gives exactly the same results. (B) For both methods, the percentage of mutations for every possible base change is reported. We considered both the complete set of patients and the set after outlier removal (no Outliers). The different distributions of the various types of mutations indicate that the 2 methods of analysis describe 2 different mutational landscapes for the same 133 AML patients.

Figure 2

Comparison of variant base frequency in the SNVs and corresponding positions in the matching normal samples. (A) Common. (B) MuTect only. (C) SomaticSniper only. Each single dot corresponds to the variant base frequency in the tumor and normal samples for: (A) SNVs identified in AMLs by both pipelines (SomaticSniper and MuTect) and corresponding positions in the normal samples (“common”; B-C), and SNVs identified exclusively by MuTect or SomaticSniper and corresponding positions in the normal samples (“MuTect only”; “SomaticSniper only”). Black dots correspond to validated SNVs, gray dots with black circle to not-validated SNVs, and gray dots are not tested SNVs. We noticed that the majority of not-validated SNVs in MuTect appear at very low frequencies. Two not-validated SNVs in SomaticSniper hit known driver genes of AML (TET2 and DNMT3A); we think that the presence of these variants in the normal sample can be due to the contamination with tumor DNA given by the minimal residual disease.