NGS-based FLT3 mutation testing
Over the past two decades, NGS, particularly multiplex-targeted gene panel NGS, has become an indispensable tool in molecular laboratory testing, particularly for diagnosing genetic disorders and cancers. Although most clinical guidelines regarding FLT3-ITD are based on data derived from PCR-FLA, NGS-based mutation profiling has become routine in many clinical diagnostic laboratories. NGS platforms, such as Illumina and Ion Torrent, enable high-throughput sequencing of FLT3 alongside other clinically significant genes. Even when PCR-FLA is employed at the initial diagnosis of acute leukemias to facilitate early treatment decisions, targeted gene panel NGS for myeloid neoplasms, including FLT3, is typically performed to complete the initial workup.
Compared with PCR-FLA-based FLT3 mutation testing, NGS can detect FLT3 mutations beyond hotspot regions, including gain-of-function activating mutations in the N-lobe and activation loop, which may influence response to FLT3 inhibitors or confer treatment resistance. Additionally, NGS enables the identification of treatment resistance-associated mutations that impact therapeutic decision-making.11,59,85
Additionally, it enables the simultaneous detection of clinically significant co-mutations in NPM1, CEBPA, and IDH1/2, allowing for refined risk stratification and better treatment decisions. Although NGS theoretically identifies all types of mutations, its ability to detect large insertions or deletions (indels) is limited due to challenges in sequence alignment. Depending on the target enrichment method and the data analysis tools used, routine NGS is not always reliable for identifying insertions and deletions longer than 15 bp.86
Significant progress has been made in the bioinformatics analysis of NGS data to identify FLT3-ITD (reviewed by Yuan et al.)86 Detection tools are typically categorized into alignment-based and assembly-based approaches. Alignment-based tools, such as Pindel,87 ITDseek,88 getITD,89 ScanITD,90 and FLT3_ITD_ext,91 align raw reads to the reference sequence and extract discordant reads to detect FLT3-ITD. Assembly-based tools, including BreaKmer, ITDetector, and ITD assembler,92–94 reconstruct misaligned short reads using specialized algorithms before realigning the assembled contigs to the reference genome.86 Comparative studies have shown that FLT3_ITD_ext performs best for both qualitative and quantitative analysis of simulated FLT3-ITD with an average insertion length of 200 bp (±20 bp) and biological samples.86 These tools, developed in programming languages such as C++, Python, and Perl, require fine-tuning, and their integration into the NGS workflow necessitates bioinformatics expertise. Clinical laboratories that adopt proprietary NGS library preparation kits often utilize packaged tools designed specifically for those kits, integrating both wet-lab and bioinformatics validations to optimize performance in detecting various genetic alterations.
Targeted gene panel NGS is the preferred method in clinical laboratories for tumor sample sequencing, as it achieves high read depth while conserving sequencing resources. Two common approaches for enriching target genes during library preparation are hybrid capture-based and amplification-based methods. Hybrid capture-based enrichment employs sequence-specific capture probes that are complementary to regions of interest, whereas amplification-based methods use multiplex PCR to enrich target sequences while simultaneously tagging them with patient-specific indexes and sequencing platform adaptors.
Although hybridization-based methods are technically more demanding, they enable the capture of larger fragments and are less affected by mismatches and allele dropout. In contrast, amplicon-based methods require less hands-on time but are more susceptible to PCR bias and significantly influenced by primer design.91,95
The FLT3-ITD allelic ratio (AR), reflecting the mutant clone burden, has historically been an important prognostic factor.65 However, the 2022 European LeukemiaNet (ELN) recommendations for AML diagnosis and management no longer consider the arbitrary cutoff of 0.5 for FLT3-ITD AR in risk classification, citing challenges in standardizing measurements, the impact of FLT3 inhibitor-based treatments, and the increasing use of MRD testing to guide treatment decisions.49 Nonetheless, assessing FLT3-ITD allelic burden remains relevant in research and clinical management.48 The AR may still be meaningful when reporting NGS results for molecular diagnostic laboratories. Recent studies have demonstrated that by optimizing bioinformatic analyses, hybrid capture-based targeted panel NGS can achieve sufficient sequencing coverage of FLT3 to accurately detect ITDs beyond 200 bp and calculate AR that correlates well with PCR-FLA results.23,85,96 The AR is calculated using variant allele frequency (VAF) from NGS results as AR = FLT3-ITD VAF ÷ (1 − ITD VAF). A VAF of 0.33 aligns with the PCR-FLA AR cutoff of 0.5. However, this correlation is likely method-dependent and influenced by bioinformatics strategies.96,97 Targeted NGS assays with amplicon-based enrichment, particularly those using anchored multiplex PCR, which employs redundant gene-specific primers to cover FLT3 exons 14 and 15,91 have shown excellent sensitivity in detecting large ITDs up to 300 bp.91,98 Given the potential for large ITDs and mutations at primer sites that result in allele dropout, multiple primers are necessary to detect all ITDs. Larger insertions may be underestimated compared to PCR-FLA due to reduced PCR amplification efficiency and alignment challenges with long fragments during bioinformatics analysis.99 Refining informatics pipelines can improve AR calculations.91
By generating nucleotide reads, NGS provides precise location and the molecular architecture of FLT3-ITD.98 ITDs may start in exon 14 or intron 14–15 and extend into exon 15.85,91,98 Notably, external sequences up to 27 bp may also be inserted at ITD junctions. Among 105 ITDs studied by Ding et al.,98 only 42% were pure tandem duplications. ITDs extending into intronic regions can affect RNA splicing, meaning DNA sequencing alone may not accurately predict amino acid sequences.98 A comparative analysis of DNA versus cDNA sequencing revealed higher detection sensitivity and higher ARs with cDNA-based detection. The higher AR from cDNA also correlated with poor clinical outcomes when combined with longer ITDs (>48 bp),100 although the clinical outcomes were study-dependent. However, the lengths of ITD from DNA and RNA were identical. It appears that ITDs extending into the intronic region would not remove or add any splice sites. Given that the intronic sequence is inframe with exons 14 and 15 and there is no stop codon in the intronic region, the ITD sequence generated from DNA would be identical to the RNA sequence, and the nomenclature created from DNA sequencing would be correct.98
NGS methods also identify different variants harbored in subpopulations of neoplastic cells. Multiple mutations of up to seven different ITD variants have been documented,91 indicating clonal heterogeneity within the neoplastic cell population.98,99 More subclones may be detected in one patient when deep sequencing with better analytic sensitivity is achieved.101 Interestingly, multiple variants of FLT3-ITD also appear frequently in ALAL.46 Although the clinical significance of these findings has yet to be characterized,101,102 these results provide a deeper understanding of clonal architecture and allow for accurate follow-up of the evolution of different clones in the disease course.83 In addition to identifying FLT3-ITD using standard informatics pipelines, accurately naming FLT3-ITD variants using standardized HGVS nomenclature is challenging but crucial for inter-laboratory comparisons and tracking disease evolution.98,99 To address this, we have developed a Python script-based web application that standardizes FLT3-ITD nomenclature using assembled sequencing reads as input.98
As sequencing costs decline, some clinical laboratories have explored whole-exome sequencing and whole-genome sequencing for cancer mutation profiling, including FLT3-ITD detection.73 Emerging long-read sequencing technologies, such as Nanopore, show promise in efficient variant phasing, the analysis of GC-rich or repetitive regions, the characterization of genomic structural variants, and the identification of full-length transcripts and isoforms.103 As these technologies continue to advance, they may become mainstream clinical methods in the future.103,104