Search

Xinrui Yang,1 Jinlong Shi,2 Xinpei Zhang,1 Gaoqi Zhang,1 Jilei Zhang,1 Siyuan Yang,1 Jing Wang,1 Xiaoyan Ke,1 Lin Fu1 1Department of Hematology and Lymphoma Research Center, Peking University, Third Hospital, Beijing 100191, China; 2Department of Biomedical Engineering, Chinese PLA General Hospital, Beijing 100853, China Purpose: DNMT3A and NPM1 mutations are known to impact the prognosis of acute myeloid leukemia (AML). DNMT3A mutations are negative prognostic factors, while NPM1 mutations are low-risk factors and inclined to concurrently appear with DNMT3A mutations. In this study, we aimed to find out how NPM1 mutations affect patients’ outcomes in the background of DNMT3A mutations. Patients and methods: We screened The Cancer Genome Atlas (TCGA) database and found 51 AML patients with DNMT3A mutations. Of them, 28 patients had a combination of NPM1 mutations. Results: In all, NPM1 had the highest mutation frequency (n=28, 54.9%). DNMT3Amut/NPM1mut patients had higher bone marrow (BM) blasts (P=0.015), higher FLT3-ITD/TKD rate (P=0.004), and lower IDH2 mutation rate (P=0.014) than the DNMT3Amut/NPM1wild patients, while their prognoses were the same as the DNMT3Amut/NPM1wild patients (P>0.1). All 51 patients benefited from hematopoietic stem cell transplantation (HSCT) treatment (P=0.005 and 0.001 for event-free survival [EFS] and overall survival [OS], respectively). In the 23 patients with DNMT3Amut/NPM1wild, those who received HSCT had prolonged EFS and OS (P=0.043 and 0.008, respectively), while HSCT treatment did not produce a positive impact on EFS and OS in the remaining 28 patients with DNMT3Amut/NPM1mut (P=0.056 and 0.053, respectively). Conclusion: Our study found that NPM1 mutations influenced BM blasts’ percentage, FLT3-ITD/TKD rate, and IDH2 mutation rate in AML patients with DNMT3A mutations but made little difference to the overall prognosis. While HSCT treatments benefited all DNMT3Amut patients, it was better for DNMT3Amut/NPM1wild patients to extend their EFS and OS. Keywords: AML, DNA methyltransferases 3A, nucleophosmin 1, next-generation sequencing, prognosis

Objective: To combine medical knowledge and medical data to interpretably predict the risk of disease. Methods: We formulated the disease prediction task as a random walk along a knowledge graph (KG). Specifically, we build a KG to record relationships between diseases and risk factors according to validated medical knowledge. Then, a mathematical object walks along the KG. It starts walking at a patient entity, which connects the KG based on the patient current diseases or risk factors and stops at a disease entity, which represents the predicted disease. The trajectory generated by the object represents an interpretable disease progression path of the given patient. The dynamics of the object are controlled by a policy-based reinforcement learning (RL) module, which is trained by electronic health records (EHRs). Experiments: We utilized two real-world EHR datasets to evaluate the performance of our model. In the disease prediction task, our model achieves 0.743 and 0.639 in terms of macro area under the curve (AUC) in predicting 53 circulation system diseases in the two datasets, respectively. This performance is comparable to the commonly used machine learning (ML) models in medical research. In qualitative analysis, our clinical collaborator reviewed the disease progression paths generated by our model and advocated their interpretability and reliability. Conclusion: Experimental results validate the proposed model in interpretably evaluating and optimizing disease prediction. Significance: Our work contributes to leveraging the potential of medical knowledge and medical data jointly for interpretable prediction tasks., Comment: 10 pages, 5 figures

Xinpei Zhang,1 Jinlong Shi,2 Jilei Zhang,1 Xinrui Yang,1 Gaoqi Zhang,1 Siyuan Yang,1 Jing Wang,1 Xiaoyan Ke,1 Lin Fu1 1Department of Hematology and Lymphoma Research Center, Peking University, Third Hospital, Beijing 100191, China; 2Department of Biomedical Engineering, Chinese PLA General Hospital, Beijing 100853, China Purpose: The incidence of DNMT3A mutations in acute myeloid leukemia (AML) is quite high and often confers a poorer prognosis. Another common gene involved in AML is IDH1/2. However, the influence of IDH1/2 mutations on outcomes in DNMT3A-mutated patients remains unknown. This study aims to determine the effect of IDH1/2mut on the prognosis in patients with DNMT3A-mutated AML. Patients and methods: We screened patients from The Cancer Genome Atlas database and selected 51 patients with AML and the DNMT3A mutation, among which 16 patients (31.4%) had both DNMT3A and IDH1/2mut. Results: Among our sample, 11 cases had the IDH1 mutation (21.7%), and 5 cases had the IDH2 mutation (9.8%). Patients in the DNMT3AmutIDH1/2wild group showed a greater number of NPM1 mutation (P=0.022), and higher event-free survival (EFS) and overall survival (OS) after hematopoietic stem cell transplantation (HSCT) (P=0.010 and P=0.007, respectively). Patients in the DNMT3AmutIDH1/2mut group showed no increase in EFS or OS after HSCT or chemotherapy. Other factors, like white blood cells, bone marrow blasts, peripheral blood blasts, and mutated recurrent gene numbers had no significant influence on EFS and OS. Conclusion: The IDH1/2 gene had little influence on the prognosis of patients with the DNMT3A mutation. For patients in the DNMT3AmutIDH1/2wild group, HSCT had a more favorable therapeutic effect. For patients with DNMT3A and IDH1/2mut, chemotherapy and HSCT appeared to have similar efficacy. Keywords: acute myeloid leukemia, molecular mutations, prognosis, next-generation sequencing

MicroRNA-9-1(miR-9-1) plays an important role in the mechanism that regulates the lineage fate of differentiating hematopoietic cells. Recent studies have shown that miR-9-1 is downregulated in t (8; 21) AML. However, the pathogenic mechanisms underlying miR-9-1 downregulation and the RUNX1-RUNX1T1 fusion protein, generated from the translocation of t (8; 21) in AML, remain unclear. RUNX1-RUNX1T1 can induce leukemogenesis through resides in and functions as a stable RUNX1-RUNX1T1-containing transcription factor complex. In this study, we demonstrate that miR-9-1 expression increases significantly after the treatment of RUNX1-RUNX1T1 (+) AML cell lines with decitabine (a DNMT inhibitor) and trichostatin A (an HDAC inhibitor). In addition, we show that RUNX1-RUNX1T1 triggers the heterochromatic silencing of miR-9-1 by binding to RUNX1-binding sites in the promoter region of miR-9-1 and recruiting chromatin-remodeling enzymes, DNMTs, and HDACs, contributing to hypermethylation of miR-9-1 in t (8; 21) AML. Furthermore, because RUNX1, RUNX1T1, and RUNX1-RUNX1T1 are all regulated by miR-9-1, the silencing of miR-9-1 enhances the oncogenic activity of these genes. Besides, overexpression of miR-9-1 induces differentiation and inhibits proliferation in t (8; 21) AML cell lines. In conclusion, our results indicate a feedback circuitry involving miR-9-1 and RUNX1-RUNX1T1, contributing to leukemogenesis in RUNX1-RUNX1T1 (+) AML cell lines.

Cytogenetically normal acute myeloid leukemia (CN-AML) is the largest and most heterogeneous AML subgroup. It lacks sensitive and specific biomarkers. Emerging evidences have suggested that microRNAs are involved in the pathogenesis of various leukemias. This paper evaluated the association between microRNA expression and prognostic outcome for CN-AML, based on the RNA/microRNA sequencing data of CN-AML patients. High let-7a-2-3p expression and low miR-188-5p expression were identified to be significantly associated with longer overall survival (OS) and event free survival (EFS) for CN-AML, independently or in a combined way. Their prognostic values were further confirmed in European Leukemia Net (ELN) genetic categories. Also, in multivariable analysis with other known risk factors, high let-7a-2-3p and low miR-188-5p expression remained to be associated with longer OS and EFS. In addition, the prognostic value of these two microRNAs was confirmed in patients who received hematopoietic stem cell transplantation (HSCT). To gain more biological insights of the underlying mechanisms, we derived the genome-wide differential gene/microRNA signatures associated with the expression of let-7a-2-3p and miR-188-5p. Several common microRNA signatures indicating favorable outcome in previous studies were up-regulated in both high let-7a-2-3p expressers and low miR-188-5p expressers, including miR-135a, miR-335 and miR-125b and all members of miR-181 family. Additionally, common up-regulated genes included FOSB, IGJ, SNORD50A and ZNF502, and FOSB was a known favorable signature in AML. These common signatures further confirmed the underlying common mechanisms for these two microRNAs value as favorable prognostic biomarkers. We concluded that high let-7a-2-3p and low miR-188-5p expression could be potentially used as favorably prognostic biomarkers independently or in a combined way in CN-AML patients, whether they received HSCT or not.

Computational methods for miRNA target prediction have already provided a large amount of potential targets, but many of them suffer from high false positive rates, therefore a post-processing filter is needed for the predicted targets. This paper proposed a statistical method based on first-order non-homogeneous Markov chain to explore the base-pairing complementarity characteristics between miRNAs and their targets. The parameters of the model were estimated by the data set of the known miRNA-mRNA duplexes. By using our model, the true miRNA:mRNA interaction pairs can be well distinguished from the permutation pairs. It showed that our method can be an efficient post-processing filter for the predicted miRNA targets.