Rare Types of NPM1 Mutations Share a Similar Molecular Background to the Common NPM1 Variant and Have No Impact on Survival in Adults with Acute Myeloid Leukemia

Background: NPM1 mutated (NPM1^MT) Acute Myeloid Leukemia (AML) accounts for around one third of all de novo AML. Virtually all mutations cluster on the exon 12 of the NPM1 gene and result in a frameshift due to a four base pairs insertion. Different types of NPM1 mutations have been described: the most common, “Type A”, accounts for around 80% of all the mutations. Rarer NPM1 mutations show diverse insertions and represent the remaining 20%; these are named alphabetically in the order of their identification (ex “B”, “D”, etc.). While the prognostic impact of NPM1 type A has been largely explored, the other mutations are less characterized mostly because of their relative rarity.

Methods: We analyzed a large, retrospective cohort of adult patients (pts) with NPM1^MT AML. The cohort was generated using two sources. Real world cases were collected from two Italian Hematology Units (IRCCS San Gerardo dei Tintori, Monza and Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan). The second group was gathered using publicly available datasets of adults with AML as follow: 1- Bottomly D. et al., Cancer Cell 2022, 2- GENIE version 14.1 (AACR Project Genie Consortium, et al., Cancer discovery 2017), 3- Ptashkin R. et al., Nature Communications 2023 implementing the MSK-IMPACT Heme gene panel, 4- TCGA (Cancer Genome Atlas Research Network. New England Journal of Medicine 2013), 5- Welch S. et al., New England Journal of Medicine 2016. Inclusion criteria were the presence of a well annotated subtype of NPM1 mutation, information on FLT3 status (i.e., mutated- FLT3^MT versus wild type- FLT3^WT) and overall survival (OS); availability of data on companion mutations was desirable.

Results: We collected a cohort of 523 pts with NPM1^MT AML, of which 470 had complete annotations. Distribution of NPM1 mutation subtypes was: type A 78.7%, type B 6.2%, type D 11.1%. We observed a set of other mutations which individually represented <1%; these were grouped as NPM1 “others” and accounted for a total of 4% of the entire cohort.

Survival analysis showed similar OS across the NPM1 mutation subtypes with significant differences based on the FLT3 status (p. 0.0014). For FLT3^WT cases, median OS from initial diagnosis was 19.7 months (mo) for NPM1 type A (n=214), 33.6 mo for type B (n=13), 14.5 mo for type D (n=29) and 10.8 mo for type “others” (n=8), respectively (p. 0.9). For FLT3^MT cases, median OS was 11.4 mo for NPM1 type A (n=156), 9.4 mo for type B (n=16), 11.2 mo for type D (n=23) and 12.1 mo for type “others” (n=11) respectively (p. 0.63). Significance was also reached when comparing NPM1 type A for FLT3^WT versus FLT3^MT pts (p. 0.0001) and NPM1 type D for FLT3^WT versus FLT3^MT pts (p. 0.0047). For NPM1 type B and NPM1 type “others”, significance could not be reached likely because of limited number of patients within these two groups (p. 0.11 and p. 0.5, respectively).

Looking at companion mutations, we found a similar pattern of co-mutations across the different NPM1 mutations subtypes. The most commonly mutated gene was FLT3 (type A: 42%, type B: 50%, type D: 44%, type “others”: 50%). Following were mutations affecting DNMT3A (type A: 45%, type B: 23%, type D: 38%, type “others”: 25%) and TET2 (type A: 18%, type B: 19%, type D: 18%, type “others”: 6%).

The DNMT3A mutational status alone did not have a prognostic impact, with comparable OS throughout the different NPM1 mutation subtypes (p. 0.65). However, when considering the pattern of the three genes (NPM1, FLT3 and DNMT3A) a significantly different OS was observed: in the overall NPM1^MT cohort, the more dismal OS was seen in the group of triply mutated (TM) NPM1^MT/FLT3^MT/DNMT3A^MT AML (n=109, median OS: 17.4 mo) compared to the group of NPM1^MT/FLT3^MT/DNMT3A^WT (n=97, median OS: 19.8 mo), NPM1^MT/FLT3^WT/DNMT3A^WT (n=166, median OS: 32.3 mo) and NPM1^MT/FLT3^WT/DNMT3A^MT (n=98, median OS: 35.3) (p. <0.001). This result was replicated among the different subtypes of NPM1 mutations. On a multivariate analysis considering age, FLT3 status and comutations, only the FLT3 status resulted strongly associated with OS (p. <0.001) among the four NPM1 subtypes of mutations.

Conclusion: The rare NPM1 mutation subtypes share a similar pattern of co-mutations with the common NPM1 type A, are similarly affected by the FLT3 status and by the presence of a TM-phenotype and feature similar OS. Publicly available datasets can be exploited to collect information on rare entities which would be otherwise poorly characterized.