ALBERT

Description: RUNX1 is mutated in ∼10% of adult acute myeloid leukemia (AML). Although most RUNX1 mutations in this disease are believed to be acquired, they can also be germline. Indeed, germline RUNX1 mutations result in the well-described autosomal-dominant familial platelet disorder with predisposition to hematologic malignancies (RUNX1-FPD, FPD/AML, FPDMM); ∼44% of affected individuals progress to AML or myelodysplastic syndromes. Using the Leucegene RUNX1 AML patient group, we sought to investigate the proportion of germline vs acquired RUNX1 mutations in this cohort. Our results showed that 30% of RUNX1 mutations in our AML cohort are germline. Molecular profiling revealed higher frequencies of NRAS mutations and other mutations known to activate various signaling pathways in these patients with RUNX1 germline–mutated AML. Moreover, 2 patients (mother and son) had co-occurrence of RUNX1 and CEBPA germline mutations, with variable AML disease onset at 59 and 27 years, respectively. Together, these data suggest a higher than anticipated frequency of germline RUNX1 mutations in the Leucegene cohort and further highlight the importance of testing for RUNX1 mutations in instances in which allogeneic stem cell transplantation using a related donor is envisioned.

Description: Purpose Differential counting of blood cells is the basis of diagnostic hematology. In many circumstances, identification of cells in bone marrow smears is the golden standard for diagnosis. Presently, methods for automatic differential counting of peripheral blood are readily available commercially. However, morphological assessment and differential counting of bone marrow smears are still performed manually. This procedure is tedious, time-consuming and laden with high inter-operator variation. In recent years, deep neural networks have proven useful in many medical image recognition tasks, such as diagnosis of diabetic retinopathy, and detection of cancer metastasis in lymph nodes. However, there has been no published work on using deep neural networks for complete differential counting of entire bone marrow smear. In this work, we present the results of using deep convolutional neural network for automatic differential counting of bone marrow nucleated cells. Materials & Methods The bone marrow smears from patients with either benign or malignant disorders in National Taiwan University Hospital were recruited in this study. The bone marrow smears are stained with Liu's stain, a modified Romanowsky stain. Digital images of the bone marrow smears were taken using 1000x oil immersion lens and 20MP color CCD camera on a single microscope with standard illumination and white-balance settings. The contour of each nucleated cell was artificially defined. These cells were then divided into a training/validation set and a test set. Each cell was then classified into 1 of the 11 categories (blast, promyelocyte, neutrophilic myelocyte, neutrophilic metamyelocyte, neutrophils, eosinophils and precursors, basophil, monocyte and precursors, lymphocyte, erythroid lineage cells, and invalid cell). In training/validation set, the classification of each cell was annotated once by experienced medical technician or hematologist. The annotated dataset was used to train a Path-Aggregation Network for instance segmentation task. In test set, cell classification was annotated by three medical technicians or hematologists; only over 2/3 consensus was regarded as valid. After the neural network model was fully trained, the ability of the model to classify and detect bone marrow nucleated cells was evaluated in terms of precision, recall and accuracy. During the model training, we used group normalization and stochastic gradient descent optimizer for training. Random noise, Gaussian blur, rotation, contrast and color shift were also used as means for data augmentation. Results The digital images of 150 bone marrow aspirate smears were taken for this study. They included 61 for acute leukemia, 39 for lymphoma, 2 for myelodysplastic syndrome (MDS), 2 for myeloproliferative neoplasm (MPN), 10 for MDS/MPN, 12 for multiple myeloma, 4 for hemolytic anemia, 9 for aplastic anemia, 8 for infectious etiology and 3 for solid cancers. The final data contained 5927 images and 187730 nucleated bone marrow cells, which were divided into 2 sets: 5630 images containing 170966 cells as the training/validation set, and 297 images containing 16764 cells as the test set. Among the 16764 cells annotated in test set, 15676 cells (93.6 %) reached over 2/3 consensus. The trained neural network achieved 0.832 recall and 0.736 precision for cell detection task, 0.79 mean intersection over union (IOU) for cell segmentation task, mean average precision of 0.659 and accuracy of 0.801 for cell classification. For individual cell categories, the model performs the best with "erythroid-lineage-cells" (0.971 recall, 0.935 precision) and the worst with "monocyte-and-precursors" (0.825 recall, 0.337 precision). Conclusions We have created the largest and the most comprehensive annotated bone marrow smear image dataset for deep neural network training. Compared with previous works, our approach is more practical for clinical application because it is able to take in an entire field of smear and generate differential counts without any other preprocessing steps. Current results are highly encouraging. With continued expansion of dataset, our model would be more precise and clinically useful. Figure Disclosures Yeh: aether AI: Other: CEO and co-founder. Yang:aether AI: Employment. Tien:Novartis: Honoraria; Daiichi Sankyo: Honoraria; Celgene: Research Funding; Roche: Honoraria; Johnson &Johnson: Honoraria; Alexion: Honoraria; BMS: Honoraria; Roche: Research Funding; Celgene: Honoraria; Pfizer: Honoraria; Abbvie: Honoraria. Hsu:aether AI: Employment.

Description: Background: Myelodysplastic syndromes (MDS) are highly heterogeneous regarding pathogenesis and clinical outcome. Traditionally, the revised international prognostic scoring system (IPSS-R)incorporating clinical features and cytogenetic abnormalities has been the standard for prognostication, while in recent years, recurrent somatic mutations are becoming increasingly important for this purpose. Additionally, gene expression profiling of coding genes and microRNAs has also emerged as a powerful tool to separate patients into distinct prognostic categories. Long non-coding RNAs (lncRNAs) are transcripts longer than 200 nucleotides without protein-coding capacity. lncRNAs not only participate in normal hematopoiesis, but perturbation of their expressions may contribute to development of acute leukemia. However, the impact of lncRNA expression profiling on the prognosis of MDS patients has not yet been explored to date. Aims: This study was aimed to evaluate lncRNAs expression profiling in MDS and to find out lncRNAs whose expression levels were associated with clinical outcomes. A scoring system was constructed to better risk-stratify MDS patients. We also tried to seek clues to the functionality of these lncRNAs. Materials & Methods: By using the Affymetrix Human Transcriptome Array 2.0 platform, we obtained the global expression profiles of 24120 lncRNAs in 176 adult patients with de novo MDS diagnosed according to the 2008 WHO classification. Through mathematical modeling, we identified six lncRNAs whose expression levels were significantly associated with overall survival (OS). We then constructed a risk scoring system with the weighted sum of each of these six lncRNAs, and evaluated the correlation of the scores with clinical features,cytogenetic abnormalities, gene mutations, andtreatment outcomesof the MDS patients.The reliability of our lncRNA scoring system was assured based on the coefficient of variance obtained from a permutation test, and further validated by using a five-fold cross-validation, in which 80% of the patients were used as the training samples to develop the scoring model, whose prediction performances were evaluated by using the rest of the 20% of the patients. Analysis of mutations in 21 genes was performed by conventional Sanger sequencing technique. Results: Higher lncRNA scores were positively associated with refractory anemia with excess of blasts (RAEB)-1 or RAEB-2 subtypes, complex cytogenetic changes, and IPSS-R high and very high risks, but inversely associated with RCUD, RARS, RCMD subtypes, a normal karyotype, and IPSS-R low risk. Patients with higher lncRNA scores had more frequent RUNX1, ASXL1, TP53, EZH2, SRSF2 and ZRSR2 mutations, shorter OS (median 14.0 months vs. 77.3 months, P

Description: Leukemic stem cells (LSCs) possess biological properties shared with normal hematopoietic stem cells. They are responsible for chemoresistance and relapse in acute myeloid leukemia (AML). Although myelodysplastic syndrome (MDS) has traditionally been regarded as a "hematopoietic stem cell disorder", the clinical and biological impact of LSCs on MDS patients are not well defined. To address this question, we used the Affymetrix HTA 2.0 microarray platform to profile 16 out of the 17 recently reported stemness genes (one of them, the ARHGAP22 gene was not included in our array) in our 160 adult primary MDS patients. The diagnoses were based on the 2016 World Health Organization (WHO) classification. Patients with antecedent chemotherapy or hematologic malignancies were excluded. Forty-one (25.6%) patients had MDS with single lineage dysplasia (MDS-SLD), 20 (12.5%) had MDS with ring sideroblasts (MDS-RS), 31 (19.4%) had MDS with multilineage dysplasia (MDS-MLD), 32 (20%) had MDS with excess blasts-1 (MDS-EB1) and 36 (22.5%), MDS-EB2. The risk distribution of the cohort was very-high risk, 15.3%; high risk, 21.3%; intermediate risk, 24.7%; low risk 34.7%; and very-low risk 4% according to the revised international prognosis scoring system (IPSS-R). We identified 4 genes (LAPTM4B, NGFRAP1, NYNRIN, and EMP1) whose expression levels were significantly correlated with overall survival (OS). An LSC4 score (0.731 x LAPTM4B - 1.259 x NGFRAP1 + 0.304 x NYNRIN + 0.231 x EMP1) was constructed based on the weighted sums derived from Cox regression analysis. Higher LSC4 scores were associated with higher IPSS-R scores, complex cytogenetics, and higher incidences of mutations of RUNX1, ASXL1, TP53, SRSF2 and ZRSR2. High-score patients had significantly higher 3-year AML transformation rate (36.3% vs 11.3%, P