Minimal Residual Disease Assessment of Common and Rare NPM1 Mutations Using a Single Massively Multiplex Digital PCR Assay

Introduction: Acute myelogenous leukemia (AML) is a fatal disease with dismal outcomes in which most patients relapse and ultimately succumb to their disease. The presence of minimal residual disease (MRD) in patients has been shown to be predictive of relapse and can thereby improve clinical decision-making. NPM1 exon 12 frameshift mutations (NPM1mut) are found in ~50% of cytogenetically normal AML and represent a well-established molecular MRD marker typically assessed by RQ-PCR. While assays are readily available for the most common NPM1mut, detection is complicated by hundreds of potential frameshift insertions, reports of NPM1mut type-switching, and a lack of sequence data in patients diagnosed for NPM1mut by capillary electrophoresis.

To simplify the deployment and increase the robustness of NPM1mut MRD detection, we developed a digital droplet PCR assay comprised of multiplex primer pools capable of detecting >95% of NPM1mut subtypes without requiring prior knowledge of NPM1 sequence. To assess its potential in patient care, we tested sensitivity and concordance with established type-specific assays and its robustness across different NPM1mut types. We further demonstrate its use in patients lacking NPM1mut sequence information.

Results: Detection of 200+ subtypes was evaluated. The assay demonstrated excellent concordance with subtype-specific assays (ρ_c = 0.97-0.99) as determined by testing in patient samples, cell lines, and synthetic cDNA bearing rare single and multiplex NPM1mut. OCI-AML3/MV-4-11 dilution experiments revealed sensitivity comparable with standard assays (1-2 x 10^-5) while retaining specificity for mutant NPM1. Multiplex primer pools produced lower coefficient of variation across digital PCR partitions relative to single primers containing deoxyinosine or 5-nitroindole.

To illustrate practical implementation where NPM1mut levels could not otherwise be monitored, we performed longitudinal retrospective MRD analysis of a patient who presented to our center without NPM1mut sequence data having been diagnosed for NPM1mut by capillary electrophoresis at an outside clinic. Monitoring was performed over the course of 356 days. To determine the actual NPM1 subtype, we performed ultra-deep targeted mRNA-seq during a transient spike in NPM1mut levels during remission at day 28. We identified 48/14,271 reads (0.34%) supporting the NPM1 subtype as type D (c.863_864insCCTG) with the incidental finding of IDH1-R132H mutation previously unknown in this patient. Subsequent re-evaluation with the subtype-specific assay (type D) demonstrated agreement with the massively multiplex assay. Steady increases in NPM1mut/10⁴ ABL1 ratios were observed to peak at day 196 when the patient presented 22% blasts in the bone marrow and >28,000 NPM1mut/10⁴ABL1 and re-entered remission with subsequent care. Targeted next generation sequencing at this time further corroborated the NPM1 type D and IDH1-R132H mutation.

Conclusions: The new assay demonstrates sensitive and robust quantification of MRD in a variety of NPM1+ AML. Patients lacking in NPM1 sequence information or who harbor rare subtypes with current standard method would greatly benefit from the assay. Moreover, deployment of the NPM1 MRD testing is simplified by covering >95% of NPM1 mutated patients in a single test without requiring plasmid standards or custom assays. Additionally, we show that MRD measurements with the multiplex assay can be used to guide the timing of deep sequencing to capture potential information about imminent relapse and to facilitate intervention at earlier stages in the disease.