Bing Chen

Chromosome karyotyping is a critical way to diagnose various hematological malignancies and genetic diseases, of which chromosome detection in raw metaphase cell images is the most critical and challenging step. In this work, focusing on the joint optimization of chromosome localization and classification, we propose ChromTR to accurately detect and classify 24 classes of chromosomes in raw metaphase cell images. ChromTR incorporates semantic feature learning and class distribution learning into a unified DETR-based detection framework. Specifically, we first propose a Semantic Feature Learning Network (SFLN) for semantic feature extraction and chromosome foreground region segmentation with object-wise supervision. Next, we construct a Semantic-Aware Transformer (SAT) with two parallel encoders and a Semantic-Aware decoder to integrate global visual and semantic features. To provide a prediction with a precise chromosome number and category distribution, a Category Distribution Reasoning Module (CDRM) is built for foreground-background objects and chromosome class distribution reasoning. We evaluate ChromTR on 1404 newly collected R-band metaphase images and the public G-band dataset AutoKary2022. Our proposed ChromTR outperforms all previous chromosome detection methods with an average precision of 92.56% in R-band chromosome detection, surpassing the baseline method by 3.02%. In a clinical test, ChromTR is also confident in tackling normal and numerically abnormal karyotypes. When extended to the chromosome enumeration task, ChromTR also demonstrates state-of-the-art performances on R-band and G-band two metaphase image datasets. Given these superior performances to other methods, our proposed method has been applied to assist clinical karyotype diagnosis.

Chromosome recognition is a critical and time-consuming process in karyotyping, especially for R-band chromosomes with poor visualization quality. In this paper, we propose an end-to-end grouping guided R-band chromosome recognition method GRC-Net. GRC-Net serves the chromosome recognition task as the main task and takes the chromosome length grouping task and centromere position grouping task as auxiliary tasks. Two auxiliary modules, Chromosome Length Grouping Module (CLGM) and Centromere Position Grouping Model (CPGM), are designed to extract the task-specific feature and refine the feature map of the main task. A large-scale R-band chromosome dataset with 1735 cases was collected. Experiment results on the 423 testing cases show that the proposed GRC-Net gets the highest accuracy of 96.87%, outperforming the baseline by 2.24%. With grouping task-guided feature extraction, GRC-Net reduces approximately 50% of the inter-group misclassification. The proposed GRC-Net can serve as a general framework for incorporating the domain knowledge into the process of feature learning, meanwhile, a powerful tool to assist clinical chromosome karyotyping.

Understanding the molecular pathogenesis of acute myeloid leukemia (AML) with well-defined genomic abnormalities has facilitated the development of targeted therapeutics. Patients with t(8;21) AML frequently harbor a fusion gene RUNX1-RUNX1T1 and KIT mutations as “secondary hit”, making the disease one of the ideal models for exploring targeted treatment options in AML. In this study we investigated the combination therapy of agents targeting RUNX1-RUNX1T1 and KIT in the treatment of t(8;21) AML with KIT mutations. We showed that the combination of eriocalyxin B (EriB) and homoharringtonine (HHT) exerted synergistic therapeutic effects by dual inhibition of RUNX1-RUNX1T1 and KIT proteins in Kasumi-1 and SKNO-1 cells in vitro. In Kasumi-1 cells, the combination of EriB and HHT could perturb the RUNX1-RUNX1T1-responsible transcriptional network by destabilizing RUNX1-RUNX1T1 transcription factor complex (AETFC), forcing RUNX1-RUNX1T1 leaving from the chromatin, triggering cell cycle arrest and apoptosis. Meanwhile, EriB combined with HHT activated JNK signaling, resulting in the eventual degradation of RUNX1-RUNX1T1 by caspase-3. In addition, HHT and EriB inhibited NF-κB pathway through blocking p65 nuclear translocation in two different manners, to synergistically interfere with the transcription of KIT. In mice co-expressing RUNX1-RUNX1T1 and KIT^N822K, co-administration of EriB and HHT significantly prolonged survival of the mice by targeting CD34⁺CD38⁻ leukemic cells. The synergistic effects of the two drugs were also observed in bone marrow mononuclear cells (BMMCs) of t(8;21) AML patients. Collectively, this study reveals the synergistic mechanism of the combination regimen of EriB and HHT in t(8;21) AML, providing new insight into optimizing targeted treatment of AML.

Context.—: Conventional karyotype analysis, which provides comprehensive cytogenetic information, plays a significant role in the diagnosis and risk stratification of hematologic neoplasms. The main limitations of this approach include long turnaround time and laboriousness. Therefore, we developed an integral R-banded karyotype analysis system for bone marrow metaphases, based on deep learning. Objective.—: To evaluate the performance of the internal models and the entire karyotype analysis system for R-banded bone marrow metaphase. Design.—: A total of 4442 sets of R-banded normal bone marrow metaphases and karyograms were collected. Accordingly, 4 deep learning–based models for different analytic stages of karyotyping, including denoising, segmentation, classification, and polarity recognition, were developed and integrated as an R-banded bone marrow karyotype analysis system. Five-fold cross validation was performed on each model. The whole system was implemented by 2 strategies of automatic and semiautomatic workflows. A test set of 885 metaphases was used to assess the entire system. Results.—: The denoising model achieved an intersection-over-union (IoU) of 99.20% and a Dice similarity coefficient (DSC) of 99.58% for metaphase acquisition. The segmentation model achieved an IoU of 91.95% and a DSC of 95.79% for chromosome segmentation. The accuracies of the segmentation, classification, and polarity recognition models were 96.77%, 98.77%, and 99.93%, respectively. The whole system achieved an accuracy of 93.33% with the automatic strategy and an accuracy of 99.06% with the semiautomatic strategy. Conclusions.—: The performance of both the internal models and the entire system is desirable. This deep learning–based karyotype analysis system has potential in a clinical application.

Background: Colon cancer is characterized by a sophisticated tumor microenvironment (TME). Cancer-associated fibroblasts (CAFs), which make up the majority of the stromal cells in TME, participate in tumor development and immune regulation. Further investigations of CAFs would facilitate an in-depth understanding of its role in colon cancer TME. Methods: In this study, we estimated CAF abundance based on The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO) databases using the Microenvironment Cell Populations-counter (MCP-counter) algorithm. CAF-related genes were identified by differential gene expression analysis combined with weighted gene coexpression network analysis. For further selection, the least absolute shrinkage and selection operator (LASSO)-Cox regression was used, and the prognostic value of the selected gene was confirmed in numerous external cohorts. The function enrichment, immunological characteristics, tumor mutation signature, immunotherapy response, and drug sensitivity of the selected gene were subsequently explored. The bioinformatics analysis results were validated using immunohistochemistry on clinical samples from our institution. Results: According to our findings, cartilage oligomeric matrix protein (COMP) was uncovered as a candidate CAFs-driven biomarker in colon cancer and plays an important role in predicting prognosis in colon cancer. COMP upregulation was associated with enhanced stromal and immune activation, and immune cell infiltration, especially M2 macrophages. Genes that mutated differently between the high- and low-COMP expression subgroups may be correlated with TME change. Following verification, COMP reliably predicted the immunotherapy response and drug response. In addition, our experimental validation demonstrated that COMP overexpression is associated with colon cancer carcinogenesis and is strongly associated with CAFs and M2 macrophage infiltration. Conclusion: Our study uncovered that COMP was a key CAFs-driven gene associated with M2 macrophage infiltration and acted as a convincing predictor for prognosis and immunotherapy response in colon cancer patients.

Chromosomal translocations and generating fusion genes are closely associated with disease initiation and progression in acute myeloid leukemia (AML). In this study, we identified a novel t(X;17)(q28;q21) chromosomal rearrangement in a patient with acute monocytic leukemia. Using RNA-sequencing, we identified a KANSL1-MTCP1 and a KANSL1-CMC4 fusion gene. 5 '-UTR sequences of the KANSL1 gene were found to become fused upstream of the coding sequence region of the MTCP1 and CMC4 genes, respectively, resulting in an aberrantly high expression of these genes. Functional studies revealed that overexpression of the MTCP1 gene induced an increased cell proliferation and partial blockage of cell differentiation, suggesting that the aberrant expression of MTCP1 is of critical importance in leukemogenesis.

Based on the leukemia‐associated immunophenotypes (LAIPs), minimal residual disease (MRD) related to the outcome can be detected by multiparameter flow cytometry in acute myeloid leukemia (AML) patients. Although 0.1% was commonly used as a cutoff value, measurable MRD or MRD level below 0.1% has also been associated with prognostic significance and more sensitive thresholds (−5 to 5.71 × 10⁻⁴. The baseline level of each LAIP was considered as the individual threshold for MRD assessment. MRD statuses stratified by 0.1% and individual threshold were termed as 0.1%‐MRD and individual‐MRD, respectively. The patients of individual‐MRD^neg showed significantly better survival compared with those of 0.1%‐MRD^neg/individual‐MRD^pos or 0.1%‐MRD^pos. Multivariate analysis showed that when time points of complete remission, post the first and second consolidation courses, were considered, only individual‐MRD post second consolidation presented independent prognostic value. Notably, in patients of cytogenetic/molecular low‐risk (LR) or intermediate‐risk (IR), the individual‐MRD status could identify the patients with significantly different outcomes, while 0.1%‐MRD status could not. Furthermore, among the patients of the LR or IR group which received chemotherapy only, those with individual‐MRD^neg status presented favorable survival, which was comparable with that of the patients accepted allogeneic hematopoietic stem cell transplantation (ASCT). This approach is useful in the selection of an ASCT strategy for clinical practice.