A Risk Score Combined Clinical and Molecular Profiles Identifies a High-Risk Subgroup Within AML1-ETO-Positive Acute Myeloid Leukemia
Min Yang,Yi Zhang,Jinghan Wang,Lixia Liu,Chengcheng Wang,Feng Lou,Shanbo Cao,Jie Jin
DOI: https://doi.org/10.1182/blood-2020-142100
IF: 20.3
2020-01-01
Blood
Abstract:Background: AML1-ETO-positive acute myeloid leukemia, is classified as a favorable leukemia subtype according to the European Leukemia Net (ELN) risk stratification. Nevertheless, studies show the biology and prognosis within the AML1-ETO-positive AML are highly different, which suggests that more prognostic factors are needed to be identified. Aims: This study mainly revealed the genomic mutation characteristics and explored more factors which affect the prognosis of Chinese AML1-ETO-positive AML patients. Methods: A total of 167 AML1-ETO-positive patients who diagnosed and treated in Zhejiang Institute of Hematology had cryopreserved DNA for deep target 185-gene regional sequencing. Variants were detected with a variant allele frequency (VAF) cutoff of 0.5%. We used a LASSO Cox regression model to build risk score for predicting overall survival. A nomogram was constructed to display the risk of death in individuals. The discrimination of the risk score was measured by the concordance index (C-index) and areas under time-dependent receiver-operating characteristics (ROC) curves (AUCs), and the calibration of the risk score was explored graphically by calibration plots. Patients (n=75) from other hospital were used as a validation cohort. Results: The median age in analyzed patients was 42(6-78) years. The most common recurrent mutations occurred in KIT(n=84,50%), ASXL2(n=46,28%), NRAS(n=37,22%), FLT3-ITD(n=35,21%) and TET2(n=30,18%). We observed that high KIT mutant allele burden predicts for poor outcome in t(8:21) AML. High KIT VAF(≥15%) correlated with shortened overall survival compared to the other KIT mutated cases including low VAF and wild-type KIT (3-year OS 26.6% vs 59.0% vs 69.6%, HR 1.50, 95%CI 0.78-2.89, P=0.0005). In addition, we also identified some other mutated genes influence the prognosis of patients with t (8;21), such as FLT3-ITD high mutation burden(VAF≥44% vs other cases, 3-year OS 30.0% vs 56.2%, HR 2.94, 95%CI 0.43-20.18, P=0.056), TET2 high mutation burden (VAF≥43% vs other cases, 3-year OS 33.3% vs 56.5%, HR 2.87, 95%CI 0.66-12.46, P=0.018) and DHX15 high mutation burden (VAF≥22% vs other cases, 3-year OS 15.0% vs 58.3%, HR 2.65, 95%CI 0.81-8.73, P=0.011). In univariate analyses for OS, age>42 (3-year OS 46.3% vs 64.4%, HR 1.91, 95%CI 1.14-3.14, P=0.012), WBC>27.1×109/L(3-year OS 34.3% vs 60.0%, HR 2.59, 95%CI 1.13-5.9, P=0.001), BM blast>20% (3-year OS 52.2% vs 92.8%, HR 6.36, 95%CI 2.7-14.97, P =0.035), LDH > 504U/L (3-year OS 44.1% vs 67.1%, HR 2.62, 95%CI 1.50-4.59, P=0.0007), PLT≤28×109/L (3-year OS 47.1% vs 66.9%, HR 1.89, 95%CI 1.13-3.17, P=0.019), HB≤87g/L (3-year OS 49.4% vs 73.8%, HR 2.20, 95%CI 1.27-3.84, P=0.019) were significantly associated with poor OS. Six variables were incorporated in our scoring model by LASSO, including age, WBC, PLT, KIT mutation, FLT3-ITD mutation and TET2 mutation. A risk scoring model was developed incorporating the weighted coefficients of these variables. The risk score grouped AML1-ETO AML patients into two subgroup: low risk (LR, n=68) and high risk (HR, n=86) groups. The 3-year OS for LR and HR groups were 72.7% and 43.0% (P<0.0001, Figure A). The similar results were also observed in validation cohort (3-year OS 79.1% vs 49.5%, P= 0.01; Figure B). Concordance index [train: 0.708, 95% CI (0.680, 0.736), validation: 0.722, 95% CI (0.666, 0.778)] demonstrated well discrimination power and calibration plots showed that the nomograms did well compared with an ideal model. Conclusion: In this study, our findings indicate that the prognostic effect of gene mutation in de novo t(8:21) AML may be influenced by the relative abundance of the mutated allele. A novel scoring model was developed and validated that incorporated molecular and clinical profiles. According to our score model, AML1-ETO AML patients could be further stratified into two subgroups with distinct clinical outcomes. Our data can serve as a basis for guided and risk-adapted treatment strategies for CBF-AML patients. The results are needed to be validated in other independent cohorts and prospective studies before implementation into clinics. Disclosures No relevant conflicts of interest to declare.