LAUSR

TO THE EDITOR Nucleophosmin (NPM1)-mutated acute myeloid leukemia (AML) is a recognized entity related to distinctive biological, clinical, and prognostic features. Nearly all NPM1 variants are frame-shift mutations located in exon 11, inducing the insertion of a new nuclear export signal (NES) at the C-terminus end of the NPM1 protein, leading to its aberrant accumulation in the cytoplasm [1]. Although 90% of NPM1 mutations correspond to the three most frequent subtypes A, B, and D, more than 50 different mutations have already been described, but their impact on leukemia pathogenesis, patient prognosis, and therapeutic monitoring remains unclear. The detection and quantification of NPM1-mutated (NPM1m) transcripts represents a specific marker for the molecular monitoring of measurable residual disease (MRD) in AML patients. An increase in NPM1m expression level between two time points is highly predictive of impending hematological relapse [2]. In addition, poor NPM1m-based MRD clearance after induction [3], consolidation [4], or hematopoietic stem cell transplantation [5] has been reported to be associated with a higher risk of relapse. Thus, NPM1m has become a decisive parameter for patient stratification and a guide for the development of new targeted therapies based on NPM1 status in AML [6]. However, in most laboratories MRD detection is restricted to the most frequent type A, B, and D mutations. Considering that approximately 10% of NPM1-mutated AML patients harbor a “rare” (non-A/B/D) variant [1], alternative approaches have been developed, mostly based on home-made allele-specific oligonucleotide (ASO)-real time quantitative-polymerase chain reaction (ASO-RT-qPCR) and droplet digital PCR (ddPCR) [2]. In this study, conducted on behalf of the French GBMHM (Groupe des Biologistes Moléculaires des Hémopathies Malignes) study group, we first provided a comprehensive description of rare NPM1 variants that have been published to date, as well as unpublished variants observed in our laboratories, hence providing an exhaustive list useful for routine diagnostic work-up. An inventory of all published NPM1 variants was performed using the PubMed database, that identified 89 mutations of NPM1 that were not A, B, or D type (called “non-A/B/D” in this study): 80 in exon 11, three in exon 10, two in exon 9 and four in exon 5 (Table 1). In our retrospective cohort of 1 561 AML patients with NPM1mutation diagnosed in four French university clinical centers, screening for NPM1 mutation was performed using high resolution melting or fragment analysis and Sanger sequencing as part of patient care. We identified 202/1 561 (13%) patients harboring non-A/B/D NPM1m variants, corresponding to 79 variants (Supplementary Fig. 1), including 57 new (never described) variants; regarding the latter, 55 mutations were localized in exon 11, one in exon 10, and one in exon 5. All newly identified mutations affecting exon 11 corresponded to a frameshift due to a 4-nucleotide insertion, as usually observed. The unique new mutation in exon 10 (c.805delinsCGTTCACTT) resulted in an 8-nucleotide shift, as for to the two other variants previously described in exon 10 [7, 8]. The new variant identified in exon 5 was an in-frame insertion of 27 nucleotides. The majority of the variants were identified in only one patient, but some variants were harbored by several patients (up to 28 patients for the c.863_864insCTTG (type I/Pm/4/H†), see Table 1). These recurrent variants have already been described in the scientific literature, suggesting that they correspond to more frequent events. We then analyzed the C-terminus end of the putative NPM1 proteins derived from the 146 non-A/B/D variants and investigated for the presence of an NES motif. 69 (47%) variants displayed a classical Lxxx-V-xx-V-x-L NES motif and 69 (47%) a variant motif (ψ-XXX-ψ-XXψ-X other than L-xxx-V-xx-V-x-L), corresponding in our cohort to 166 (82%) and 35 (17%) non-A/B/D NPM1 variant patients,