P519: risk stratification by nomogram scoring system of npm1-mutated cytogenetically normal acute myeloid leukemialeukaemia based on a multi-gene panel

Background: Approximately 50% of adult cytogenetically normal acute myeloid leukaemia (CN-AML) patients have nucleophosmin 1 (NPM1) mutations, which frequently accompany other driver mutations that may affect the clinical course and prognosis. Aims: This study was aimed at elucidating the prognosis and best therapy for NPM1 mutated CN-AML patients, on which currently consensus is lacking. Methods: Overall, 150 CN-AML patients with NPM1 mutations were enrolled for targeted region sequencing using a 273-gene panel. The patients were randomised to training (N=90) and validation cohorts (N=60). A nomogram model was constructed by combining molecular and clinical data; risk stratification was performed using a scoring system based on prognostic factors. Results: Besides NPM1, 1747 mutations were detected in 273 genes; more than five mutations were detected in 96.7% of the patients (median: 11 [range: 4–24] mutations/patient). Multivariable analyses showed the following factors independently correlated with worse leukaemia-free survival (LFS): high white blood cell (WBC) count (reference: ≥27.63×109/L; hazard ratio [HR]: 2.911, P=0.004) and mutations in DNMT3A (HR: 2.825, P = 0.005), FLT3-ITD (HR: 2.211, P=0.043), DDX11 (HR: 6.895, P<0.001), and LILRB3 (HR: 9.319, P<0.001). A nomogram model was constructed using the above factors (Figure 1 A), the model showed good calibration between the predicted and observed values of the 5-year LFS in both training and validation cohorts (Figure 1 B-C). On the basis of the risk coefficients of the five factors in the model, we obtained the following formula: Score = WBC×1.068 + DNMT3A×1.039 + FLT3-ITD×0.794 + DDX11×1.931 + LILRB3×2.232. ROC curve analysis showed that the AUC of the prognostic model for predicting relapse was 0.742 (P<0.001); the corresponding score of the maximum Youden index was 2.17. Patients were assigned to the low-, intermediate-, and high-risk subgroups per the score (corresponding scores: 0, <2.17, and ≥2.17). The 5-year LFS rates of the low-, intermediate-, and high-risk subgroups in the training and validation cohorts decreased significantly in turn (P<0.001 and P=0.004, respectively) (Figure 1 D-E). Post-remission therapy analysis per the risk model showed that compared to chemotherapy, allogeneic hematopoietic stem cell transplantation (allo-HSCT) improved the 5-year LFS (high-risk subgroup, 62.22% vs. 14.42%, P=0.039; intermediate-risk subgroup, 77.73% vs. 41.20%, P=0.035) and cumulative incidence of relapse (CIR) (high-risk subgroup, 2-year CIR: 0.00% vs. 61.54%, P<0.001; intermediate-risk subgroup, 5-year CIR: 11.74% vs. 52.54%, P=0.006) in the high- and intermediate-risk subgroups, but not in low-risk subgroups (Figure 1 F-K). Summary/Conclusion: Our study indicated that NPM1-mutated CN-AML could be further refined into three risk subgroups and allo-HSCT should be recommended as the first-line post-remission therapy for intermediate- and high-risk patients.Figure 1. (A) Nomogram model. For WBC, “< 27.63 or >= 27.63” represents white blood cell count < or ≥ 27.63×109/L at diagnosis; for genes mutation, “YES” represents mutation. (B-C) Calibration plots of the training (B) and validation (C) cohorts. (D) Leukemia-free survival analyses by risk stratification of training cohort. (E) Leukemia-free survival analyses by risk stratification of validation cohort. (F-K) Prognosis of post-remission therapies by risk stratification. Leukemia-free survival and cumulative incidence of relapse in high-risk subgroup (F-G), intermediate-risk subgroup (H-I) and low-risk subgroup (J-K). Keywords: Risk factor, Allo-SCT, AML, Gene mutation