Error-corrected sequencing strategies enable comprehensive detection of leukemic mutations relevant for diagnosis and minimal residual disease monitoring

Abstract

Background: Pediatric leukemias have a diverse genomic landscape associated with complex structural variants, including gene fusions, insertions and deletions, and single nucleotide variants. Routine karyotype and fluorescence in situ hybridization (FISH) techniques lack sensitivity for smaller genomic alternations. Next-generation sequencing (NGS) assays are being increasingly utilized for assessment of these various lesions. However, standard NGS lacks quantitative sensitivity for minimal residual disease (MRD) surveillance due to an inherently high error rate.

Methods: Primary bone marrow samples from pediatric leukemia (n = 32) and adult leukemia subjects (n = 5), cell line MV4-11, and an umbilical cord sample were utilized for this study. Samples were sequenced using molecular barcoding with targeted DNA and RNA library enrichment techniques based on anchored multiplexed PCR (AMP®) technology, amplicon based error-corrected sequencing (ECS) or a human cancer transcriptome assay. Computational analyses were performed to quantitatively assess limit of detection (LOD) for various DNA and RNA lesions, which could be systematically used for MRD assays.

Results: Matched leukemia patient samples were analyzed at three time points; diagnosis, end of induction (EOI), and relapse. Similar to flow cytometry for ALL MRD, the LOD for point mutations by these sequencing strategies was ≥0.001. For DNA structural variants, FLT3 internal tandem duplication (ITD) positive cell line and patient samples showed a LOD of ≥0.001 in addition to previously unknown copy number losses in leukemia genes. ECS in RNA identified multiple novel gene fusions, including a SPANT-ABL gene fusion in an ALL patient, which could have been used to alter therapy. Collectively, ECS for RNA demonstrated a quantitative and complex landscape of RNA molecules with 12% of the molecules representing gene fusions, 12% exon duplications, 8% exon deletions, and 68% with retained introns. Droplet digital PCR validation of ECS-RNA confirmed results to single mRNA molecule quantities.

Conclusions: Collectively, these assays enable a highly sensitive, comprehensive, and simultaneous analysis of various clonal leukemic mutations, which can be tracked across disease states (diagnosis, EOI, and relapse) with a high degree of sensitivity. The approaches and results presented here highlight the ability to use NGS for MRD tracking.

Keywords: Computational biology; Error-corrected sequencing; Minimal residual disease; Next generation sequencing; Pediatric leukemia.

Fig. 1

Variant tracking at diagnosis, end of induction (EOI), and relapse in a single pediatric AML subject across multiple regions in the genome. The top panel represents the analysis between the diagnostic and end of induction sample. Left axis and blue lines represent the variant allele frequency (VAF), and the orange lines and right x-axis represent the delta of VAF between the diagnostic and end of induction sample. The bottom panel of the Figure represents the analysis between endo of induction and relapse with the same x- and y- axis as the top panel

Fig. 2

Error-corrected sequencing noise reduction for low allelic variants. Top panel represents all variants (blue dots) identified in the genomic region of a known somatic mutation located in PTPN11 (red dot). Bottom panel demonstrates noise reduction with the application of error-corrected sequencing

Fig. 3

Development of a specific and sensitive NGS panel for FLT3-ITD detection and disease monitoring. a Serial dilution results for MV4–11 cell line (30 bp ITD in FLT3). The x-axis is allele fraction and y-axis is the number of ITD supporting reads / 10 M sequencing reads. b Nine leukemia samples were analyzed with varying ITD sizes (right y-axis, grey bars) and allele fraction (left y-axis, blue bars). Subjects are across the x-axis, with D = diagnosis, EOI = end of induction, and R = relapse sample

Fig. 4

Summary of RNA-ECS results for pediatric leukemia diagnostic samples. a Distribution of allelic specific single nucleotide variants and gene counts. Pie chart represents the distribution across all leukemia samples. b The bar graph represents the counts per gene. c Distribution of RNA StVs and gene counts. Pie chart represents the distribution across all leukemia samples for SNVs. d The bar graph represents the counts per gene

Fig. 5

Novel RNA variants identified at time of diagnosis. a An aberrant RNA molecule in ZCCHC7 was identified in several of the B-ALL samples. b Novel cryptic gene fusion in SPTAN1-ABL1 was identified in a T-ALL subject and confirmed via Sanger sequencing. ExPASy translation of the fusion product sequence revealed the in-frame fusion of SPTAN1 (amino acids in red) and ABL1 (amino acids in black)