P066 | FACTORS INFLUENCING MOLECULAR RELAPSE IN FAVORABLE-RISK, INTENSIVELY-TREATED, NPM1-MUTATED ACUTE MYELOID LEUKEMIA

As per 2021 ELN guidelines, for NPM1-mutated AML patients only the end-of-treatment (EOT) MRD is considered a clinically relevant biomarker, with thresholds of qPCR NPM1^mut/ABL1% >2% (hereafter defined as NPM1%) and logarithmic reduction relative to diagnosis <3 (hereafter defined as LogR). However, substantial uncertainty remains regarding those thresholds and time-points in this AML category. We conducted a retrospective analysis on 30 favorable-risk, intensively treated, NPM1-mutated AML patients diagnosed at our institution between 2017 and 2024, collecting clinical (age, WBC count, GO vs noGO) and biological (post-induction, EOT and long term follow-up bone marrow MRD values) variables aiming to identify a correlation with molecular relapse (molR) risk and molecular relapse-free survival (molRFS) during follow-up. Median EOT NPM1% and LogR were 0.037% and 4.9, respectively, with only 1/30 (3.3%) patients meeting the threshold of NPM1>2% or LogR<3. Overall, 17/30 (56.6%) patients relapsed, with 15/17 (88.2%) within the first 6 months. No correlation between molR and age (p=0.36), WBC count (p=0.97), use of GO (p=0.89), NPM1% at diagnosis (p=0.38) and EOT NPM1>2% or LogR<3 (p=0.17) was observed. Using ROC curves we identified EOT NPM1% and LogR thresholds of 0.01% and 4.75, respectively, both in 10/30 (33.3%) patients; 9/10 (90%) relapsed all within 5 months (OR 13.5, 95%CI 1.4-128.4, p=0.01, χ²= 8.3, p=0.009. Median molRFS NR vs 1.5mo, HR 5.17, 95%CI 1.9-13.9, p=0.001) (Fig. 1A). When applied in the post-induction setting, ROC curves identified NPM1% and LogR thresholds of 2.1% and 2.95. Post-induction NPM1>2.1% was observed in 11/30 (36.6%) patients, with 10/11 (90.9%) relapses (OR 17.1, 95%CI 1.8-163.8, p=0.006, χ²=8.3, p=0.0039. Median molRFS NR vs 3mo, HR 3.38, 95%CI 1.3-9.0, p=0.015) (Fig. 1B), while post-induction LogR<2.95 was observed in 14/30 (46.6%) patients, with 12/14 (85.7%) relapses (OR 13.2, 95%CI 2.1-82.6, p=0.003, χ²=9, p=0.0026. Median molRFS NR vs 3mo, HR 5.03, 95%CI 1.7-14.6, p=0.003). In our cohort, ELN-recommended MRD cut-offs proved ineffective in predicting molecular relapse as well as other clinical and biological factors. Other statistically relevant EOT MRD thresholds were identified. Interestingly, post-induction thresholds similar to the ELN-recommended EOT MRD cut-offs were identified and proved significant in this setting. More data are warranted due to the small sample size, highlighting the need for larger studies.