MRD-seq: A sensitive next generation sequencing-based approach for measurable residual disease detection in acute myeloid leukaemia

Abstract Acute myeloid leukaemia (AML) is an aggressive and genetically heterogeneous malignancy of the haematopoietic system, characterised by poor prognosis and high relapse rates. Although complete remission can often be achieved through initial treatment, a significant proportion of patients relapse due to the persistence of measurable residual disease (MRD). Quantification of MRD, particularly at high sensitivity, is essential for evaluating therapeutic efficacy, predicting relapse risk, and guiding treatment decisions. Notably, patients who achieve deep molecular remission (MRD negative), have markedly improved outcomes compared to those with shallower responses (MRD positive) (Short et al., 2020). Thus, the ability to sensitively and quantitatively monitor residual leukemic cells during and after therapy is of high clinical importance. Current MRD assessment methods in AML are largely based on quantitative PCR (qPCR) targeting specific leukaemia-associated mutations. However, given the genetic diversity of AML, with over 200 recurringly mutated genes and highly individual mutation profiles, qPCR-based MRD assays are currently only available for approximately one-third of patients. Each qPCR assay requires laborious development, optimisation, and calibration, making routine and broad implementation impractical. To overcome these limitations, we developed MRD-Seq, a next-generation sequencing (NGS)-based method for sensitive, mutation-specific MRD detection. MRD-Seq involves PCR amplification of genomic regions harbouring known somatic mutations using proof-reading polymerases, followed by deep sequencing. Unique molecular identifiers (UMIs) are used to collapse duplicate reads and detect sequencing errors, enabling accurate quantification of mutant allele frequency. An in-house R script was developed to determine the proportion of mutation-bearing versus wild-type reads. To validate the approach, MRD-Seq assays were developed for two recurring AML-associated mutations: NPM1 Type A and IDH1R132H. Assays were tested on serial dilution series prepared from cell lines with these mutations: OCI-AML3 for NPM1 Type A and U-87 IDH1-mutant for IDH1R132H. The NPM1 Type A MRD-Seq assay achieved a sensitivity of 1 × 10⁻⁵, while the IDH1R132H assay demonstrated sensitivity of 1 × 10⁻³. MRD-Seq demonstrated high sensitivity comparable to that of qPCR-based assays. Importantly, MRD-Seq can be rapidly adapted to any somatic point mutation with minimal assay reconfiguration, simply by designing a PCR primer pair flanking the mutation of interest. In cases such as NPM1 mutations, where multiple variants cluster within a small genomic region, MRD-Seq can detect and quantify multiple mutation types simultaneously using a single assay, without the need for mutation-specific calibration curves. MRD-Seq offers a practical and highly sensitive platform for MRD monitoring that can be broadly applied to the majority of AML patients, addressing a critical unmet need in clinical disease management. The simplicity of its design and adaptability to virtually any somatic mutation make it well-suited not only for AML but also for other malignancies where tumour-specific mutations can be identified. We aim to further increase the sensitivity of the assay, particularly for detecting point mutations. Additionally, we will apply MRD-Seq to AML patient samples to validate its performance in real-world settings and assess its potential as a robust tool for personalised disease monitoring and risk stratification.