Impact of variant topography of <i>NPM1</i> mutations in acute myeloid leukemia.

Abstract

6536 Background: NPM1 is a frequent driver mutation in AML and conveys favorable prognosis in the absence of FLT3 mutations. Currently, NPM1 subtypes (Type A, B, D and non-ABD) are approached uniformly. A granular analysis evaluating NPM1 subtypes (ST) is essential to improve molecular risk stratification in AML. We seek to evaluate the molecular intricacies of NPM1 variants and genomic architecture to determine how these influence outcomes in AML. Methods: The cohort had patient data from Karmanos Cancer Institute, meta-analytic cohorts (Awada et al., 2021; Kewan et al., 2023), cBioPortal, and AACR GENIE (v17) (Cerami et al., 2012; Gao et al., 2013; de Bruijn et al., 2023). Clinical and genomic characteristics of NPM1 MT AML were reviewed. Insertions between nucleotides 863 and 864 determined the ST classification: TCTG (A), CATG (B), CCTG (D) and others (non-ABD). Stratification of NPM1 by VAF was performed by quartiles from the median. Clonal hierarchy was defined as a difference in VAF ≥5% and differences &lt;5% were designated as codominant. Results: Of 7653 patients with AML, 5342 (84.2%) had de novo AML. NPM1 MT was present in 1570. Type A was most common (73%), followed by non-ABD (18%), B (5%) and D (4%). Median age was 66 years (IQR 53.8-73.7) and was similar across groups. Females more frequently had non-ABD than B or D (67% vs 45% and 53%, p=0.0008 and 0.04, respectively). Non-ABD had the least proliferative bone marrow compared to A, B or D (blast percentage 54 vs 79, 75.5, 70, p&lt;0.0001, p=0.0015, p=0.02). Type A had lower VAF than other ST (29% vs 34%, p&lt;0.0001), while B and D had higher VAF than other ST (38% vs 30%, p&lt;0.0001 and 33% vs 30% p=0.0004). The most common co-mutation with NPM1 was DNMT3A (48%). Type A was enriched for DNMT3A compared to other ST (54% vs 34%, p&lt;0.00001). Type A and D had more IDH1 , IDH2 , and NRAS mutations than B and non-ABD (17% vs 10%, p=0.01; 23% vs 9%, p&lt;0.00001; 16% vs 10%, p=0.03, respectively). B and non-ABD were more enriched in WT1 compared to A and D (12% vs 4%, p=0.00002). Non-ABD was more enriched for FLT3 mutations than other ST (39% vs 24%, p=0.00012). DNMT3A R882 was enriched in high VAF (HV) compared to low VAF (LV) NPM1 (54% vs 46%, p=0.01). FLT3 was more common in HV NPM1 (34% vs 17%, p&lt;0.0001). NPM1 was often a subclone (82%), but was more likely to be dominant in B (29%, p=0.001). The median OS of NPM1 MT AML was 19 mo, similar across ST (A: 18 vs B: 14 mo, p=0.3896). OS was numerically higher in D (26 mo) and non-ABD (25 mo). LV NPM1 (Q1 ≤ 22.8%) had significantly better OS than HV (Q3 ≥ 38%): 21 vs 10 mo (p&lt;0.0001). This trend was most pronounced in D and non-ABD (D: 58 vs 10 mo, p=0.009; non-ABD: 29 vs 7 mo, p=0.005). Conclusions: As survival differences between NPM1 mutation subtypes were numerically different, enrichment of Type A and D for IDH1/2 and NRAS , non-ABD for FLT3 , B and D for WT1 probes the question of the impact of these co-mutations and VAF on NPM1 outcomes. Further work will evaluate the impact of NPM1 ST on de novo and therapy-related AML.