Table 2 Comparison of diagnostic methods for the KMT2A gene market

Immunophenotyping with NG2 antibody (CSPG4)

• Rapid detection of KMT2A-r in approximately 90% of cases.

• Useful for poor prognosis AML subtypes, especially FAB M5.

• Potential to identify KMT2A-r in both pediatric and adult AML cases.

• Limited to specific translocations (e.g., t(4;11), t(9;11)) that produce KMT2A-AF4 and KMT2A-AF9 fusions.

• Not specific to AML and may miss other KMT2A fusion genes.

• Potential variability in accuracy between AML and ALL samples.

Fluorescence in situ hybridization (FISH)

• Gold standard method

• Widely available and well-established technique.

• Can detect cryptic KMT2A-r when combined with KMT2A probe kits.

• Suitable for common KMT2A translocations.

• Limited in detecting rare or novel partner genes in KMT2A-r.

• May give false-negative results in minor breakpoint regions (e.g., KMT2A-USP2 fusion).

• Less effective for detecting cryptic translocations.

RT-PCR (Reverse Transcription Polymerase Chain Reaction)

• Effective for detecting known KMT2A fusions, especially in high-expression cases.

• Validated for identifying KMT2A-r in both pediatric and adult populations.

• Useful for identifying specific fusion genes and MRD.

• Limited to common fusion partners and not suitable for rare or cryptic fusions.

• Requires fresh/frozen samples or high-quality RNA.

• Not applicable for detecting novel fusion partners without specific primers.

Next-Generation Sequencing (NGS)

• High-throughput, sensitive, and can identify novel KMT2A fusion partners.

• Can detect cryptic translocations and multiple genetic alterations.

• Comprehensive approach for risk categorization and treatment selection.

• Costly and requires advanced bioinformatics capabilities.

• May struggle with detecting low-expression transcripts or rare fusions in some cases.

• High complexity and potentially overwhelming for routine clinical use.

Machine Learning (ML) Approaches

• Can predict and identify new biomarkers based on large datasets.

• Offers potential for improved diagnosis and risk stratification.

• Could enhance precision medicine and clinical decision-making.

• Limited annotated datasets and model interpretability.

• Requires large datasets and advanced computational tools.

• Still experimental, and further validation is required for clinical integration.