Application of interpretable machine learning algorithms to predict distant metastasis in ovarian clear cell carcinoma.

BACKGROUND: Ovarian clear cell carcinoma (OCCC) represents a subtype of ovarian epithelial carcinoma (OEC) known for its limited responsiveness to chemotherapy, and the onset of distant metastasis significantly impacts patient prognoses. This study aimed to identify potential risk factors contributing to the occurrence of distant metastasis in OCCC. METHODS: Utilizing the Surveillance, Epidemiology, and End Results (SEER) database, we identified patients diagnosed with OCCC between 2004 and 2015. The most influential factors were selected through the application of Gaussian Naive Bayes (GNB) and Adaboost machine learning algorithms, employing a Venn test for further refinement. Subsequently, six machine learning (ML) techniques, namely XGBoost, LightGBM, Random Forest (RF), Adaptive Boosting (Adaboost), Support Vector Machine (SVM), and Multilayer Perceptron (MLP), were employed to construct predictive models for distant metastasis. Shapley Additive Interpretation (SHAP) analysis facilitated a visual interpretation for individual patient. Model validity was assessed using accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and the area under the receiver operating characteristic curve (AUC). RESULTS: In the realm of predicting distant metastasis, the Random Forest (RF) model outperformed the other five machine learning algorithms. The RF model demonstrated accuracy, sensitivity, specificity, positive predictive value, negative predictive value, F1 score, and AUC (95% CI) values of 0.792 (0.762-0.823), 0.904 (0.835-0.973), 0.759 (0.731-0.787), 0.221 (0.186-0.256), 0.974 (0.967-0.982), 0.353 (0.306-0.399), and 0.834 (0.696-0.967), respectively, surpassing the performance of other models. Additionally, the calibration curve's Brier Score (95%) for the RF model reached the minimum value of 0.06256 (0.05753-0.06759). SHAP analysis provided independent explanations, reaffirming the critical clinical factors associated with the risk of metastasis in OCCC patients. CONCLUSIONS: This study successfully established a precise predictive model for OCCC patient metastasis using machine learning techniques, offering valuable support to clinicians in making informed clinical decisions.

A Novel Necroptosis-Related Signature Can Predict Prognosis and Chemotherapy Sensitivity in Multiple Myeloma.

BACKGROUND: Necroptosis is an inflammatory cell death mode, and its association with multiple myeloma (MM) remains unclear. METHODS: This prospective study first analyzed the association between necroptosis-related signature as well as prognosis and chemotherapy sensitivity in MM using the necroptosis score. Consensus clustering was used to identify necroptosis-related molecular clusters. Least absolute shrinkage and selection operator analysis and multivariate Cox regression analysis were performed to establish the prognostic model of necroptosis-related genes (NRGs). RESULTS: A high necroptosis score was associated with poor prognosis and abundant immune infiltration. Two molecular clusters (clusters A and B) significantly differed in terms of prognosis and tumor microenvironment. Cluster B had a worse prognosis and higher tumor marker pathway activity than cluster A. The risk score model based on four NRGs can accurately predict the prognosis of patients with MM, which was validated in two validation cohorts. Receiver operating characteristic curve analysis showed that the area under the curves of the risk score in predicting the 1-, 3-, and 5-year survival rates were 0.710, 0.758, and 0.834, respectively. Further, the activity of pathways related to proliferation and genetic regulation in the high-risk group significantly increased. The drug prediction results showed that the low-risk score group was more sensitive to bortezomib, cytarabine, and doxorubicin than the high-risk score group. Meanwhile, the high-risk score group was more sensitive to lenalidomide and vinblastine than the low-risk score group. Finally, the upregulation of model genes CHMP1A, FAS, JAK3, and HSP90AA1 in clinical samples collected from patients with MM was validated via real-time polymerase chain reaction. CONCLUSION: A systematic analysis of NRGs can help identify potential necroptosis-related mechanisms and provide novel biomarkers for MM prognosis prediction, tumor microenvironment evaluation, and personalized treatment planning.

Molecular subtypes predict therapeutic responses and identifying and validating diagnostic signatures based on machine learning in chronic myeloid leukemia.

Chronic myeloid leukemia (CML) is a hematological tumor derived from hematopoietic stem cells. The aim of this study is to analyze the biological characteristics and identify the diagnostic markers of CML. We obtained the expression profiles from the Gene Expression Omnibus (GEO) database and identified 210 differentially expressed genes (DEGs) between CML and normal samples. These DEGs are mainly enriched in immune-related pathways such as Th1 and Th2 cell differentiation, primary immunodeficiency, T cell receptor signaling pathway, antigen processing and presentation pathways. Based on these DEGs, we identified two molecular subtypes using a consensus clustering algorithm. Cluster A was an immunosuppressive phenotype with reduced immune cell infiltration and significant activation of metabolism-related pathways such as reactive oxygen species, glycolysis and mTORC1; Cluster B was an immune activating phenotype with increased infiltration of CD4 + and CD8 + T cells and NK cells, and increased activation of signaling pathways such as interferon gamma (IFN-γ) response, IL6-JAK-STAT3 and inflammatory response. Drug prediction results showed that patients in Cluster B had a higher therapeutic response to anti-PD-1 and anti-CTLA4 and were more sensitive to imatinib, nilotinib and dasatinib. Support Vector Machine Recursive Feature Elimination (SVM-RFE), Least Absolute Shrinkage Selection Operator (LASSO) and Random Forest (RF) algorithms identified 4 CML diagnostic genes (HDC, SMPDL3A, IRF4 and AQP3), and the risk score model constructed by these genes improved the diagnostic accuracy. We further validated the diagnostic value of the 4 genes and the risk score model in a clinical cohort, and the risk score can be used in the differential diagnosis of CML and other hematological malignancies. The risk score can also be used to identify molecular subtypes and predict response to imatinib treatment. These results reveal the characteristics of immunosuppression and metabolic reprogramming in CML patients, and the identification of molecular subtypes and biomarkers provides new ideas and insights for the clinical diagnosis and treatment.

Splicing factor-mediated regulation patterns reveals biological characteristics and aid in predicting prognosis in acute myeloid leukemia.

BACKGROUND: Alternative splicing (AS) of RNA is a fundamental biological process that shapes protein diversity. Many non-characteristic AS events are involved in the onset and development of acute myeloid leukemia (AML). Abnormal alterations in splicing factors (SFs), which regulate the onset of AS events, affect the process of splicing regulation. Hence, it is important to explore the relationship between SFs and the clinical features and biological processes of patients with AML. METHODS: This study focused on SFs of the classical heterogeneous nuclear ribonucleoprotein (hnRNP) family and arginine and serine/arginine-rich (SR) splicing factor family. We explored the relationship between the regulation patterns associated with the expression of SFs and clinicopathological factors and biological behaviors of AML based on a multi-omics approach. The biological functions of SRSF10 in AML were further analyzed using clinical samples and in vitro experiments. RESULTS: Most SFs were upregulated in AML samples and were associated with poor prognosis. The four splicing regulation patterns were characterized by differences in immune function, tumor mutation, signaling pathway activity, prognosis, and predicted response to chemotherapy and immunotherapy. A risk score model was constructed and validated as an independent prognostic factor for AML. Overall survival was significantly shorter in the high-risk score group. In addition, we confirmed that SRSF10 expression was significantly up-regulated in clinical samples of AML, and knockdown of SRSF10 inhibited the proliferation of AML cells and promoted apoptosis and G1 phase arrest during the cell cycle. CONCLUSION: The analysis of splicing regulation patterns can help us better understand the differences in the tumor microenvironment of patients with AML and guide clinical decision-making and prognosis prediction. SRSF10 can be a potential therapeutic target and biomarker for AML.

Ferroptosis-related molecular patterns reveal immune escape, inflammatory development and lipid metabolism characteristics of the tumor microenvironment in acute myeloid leukemia.

Background: An increasing number of studies have revealed the influencing factors of ferroptosis. The influence of immune cell infiltration, inflammation development and lipid metabolism in the tumor microenvironment (TME) on the ferroptosis of tumor cells requires further research and discussion. Methods: We explored the relationship between ferroptosis-related genes and acute myeloid leukemia (AML) from the perspective of large sample analysis and multiomics, used multiple groups to identify and verify ferroptosis-related molecular patterns, and analyzed the sensitivity to ferroptosis and the state of immune escape between different molecular pattern groups. The single-sample gene set enrichment analysis (ssGSEA) algorithm was used to quantify the phenotypes of ferroptosis-related molecular patterns in individual patients. HL-60 and THP-1 cells were treated with ferroptosis inducer RSL3 to verify the therapeutic value of targeted inhibition of GPX4. Results: Three ferroptosis-related molecular patterns and progressively worsening phenotypes including immune activation, immune exclusion and immunosuppression were found with the two different sequencing approaches. The FSscore we constructed can quantify the development of ferroptosis-related phenotypes in individual patients. The higher the FSscore is, the worse the patient's prognosis. The FSscore is also highly positively correlated with pathological conditions such as inflammation development, immune escape, lipid metabolism, immunotherapy resistance, and chemotherapy resistance and is negatively correlated with tumor mutation burden. Moreover, RSL3 can induce ferroptosis of AML cells by reducing the protein level of GPX4. Conclusions: This study revealed the characteristics of immunity, inflammation, and lipid metabolism in the TME of different AML patients and differences in the sensitivity of tumor cells to ferroptosis. The FSscore can be used as a biomarker to provide a reference for the clinical evaluation of the pathological characteristics of AML patients and the design of personalized treatment plans. And GPX4 is a potential target for AML treatment.

A novel fatty acid metabolism-related signature identifies features of the tumor microenvironment and predicts clinical outcome in acute myeloid leukemia.

BACKGROUND: Acute myeloid leukemia (AML) is the most common malignancy of the hematological system, and there are currently a number of studies regarding abnormal alterations in energy metabolism, but fewer reports related to fatty acid metabolism (FAM) in AML. We therefore analyze the association of FAM and AML tumor development to explore targets for clinical prognosis prediction and identify those with potential therapeutic value. METHODS: The identification of AML patients with different fatty acid metabolism characteristics was based on a consensus clustering algorithm. The CIBERSORT algorithm was used to calculate the proportion of infiltrating immune cells. We used Cox regression analysis and least absolute shrinkage and selection operator (LASSO) regression analysis to construct a signature for predicting the prognosis of AML patients. The Genomics of Drug Sensitivity in Cancer database was used to predict the sensitivity of patient samples in high- and low-risk score groups to different chemotherapy drugs. RESULTS: The consensus clustering approach identified three molecular subtypes of FAM that exhibited significant differences in genomic features such as immunity, metabolism, and inflammation, as well as patient prognosis. The risk-score model we constructed accurately predicted patient outcomes, with area under the receiver operating characteristic curve values of 0.870, 0.878, and 0.950 at 1, 3, and 5 years, respectively. The validation cohort also confirmed the prognostic evaluation performance of the risk score. In addition, higher risk scores were associated with stronger fatty acid metabolisms, significantly higher expression levels of immune checkpoints, and significantly increased infiltration of immunosuppressive cells. Immune functions, such as inflammation promotion, para-inflammation, and type I/II interferon responses, were also significantly activated. These results demonstrated that immunotherapy targeting immune checkpoints and immunosuppressive cells, such as myeloid-derived suppressor cells (MDSCs) and M2 macrophages, are more suitable for patients with high-risk scores. Finally, the prediction results of chemotherapeutic drugs showed that samples in the high-risk score group had greater treatment sensitivity to four chemotherapy drugs in vitro. CONCLUSIONS: The analysis of the molecular patterns of FAM effectively predicted patient prognosis and revealed various tumor microenvironment (TME) characteristics.

Inflammatory response mediates cross-talk with immune function and reveals clinical features in acute myeloid leukemia.

Accumulated genetic mutations are an important cause for the development of acute myeloid leukemia (AML), but abnormal changes in the inflammatory microenvironment also have regulatory effects on AML. Exploring the relationship between inflammatory response and pathological features of AML has implications for clinical diagnosis, treatment and prognosis evaluation. We analyzed the expression variation landscape of inflammatory response-related genes (IRRGs) and calculated an inflammatory response score for each sample using the gene set variation analysis (GSVA) algorithm. The differences in clinical- and immune-related characteristics between high- and low-inflammatory response groups were further analyzed. We found that most IRRGs were highly expressed in AML samples, and patients with high inflammatory response had poor prognosis and were accompanied with highly activated chemokine-, cytokine- and adhesion molecule-related signaling pathways, higher infiltration ratios of monocytes, neutrophils and M2 macrophages, high activity of type I/II interferon (IFN) response, and higher expression of immune checkpoints. We also used the Genomics of Drug Sensitivity in Cancer (GDSC) database to predict the sensitivity of AML samples with different inflammatory responses to common drugs, and found that AML samples with low inflammatory response were more sensitive to cytarabine, doxorubicin and midostaurin. SubMap algorithm also demonstrated that high-inflammatory response patients are more suitable for anti-PD-1 immunotherapy. Finally, we constructed a prognostic risk score model to predict the overall survival (OS) of AML patients. Patients with higher risk score had significantly shorter OS, which was confirmed in two validation cohorts. The analysis of inflammatory response patterns can help us better understand the differences in tumor microenvironment (TME) of AML patients, and guide clinical medication and prognosis prediction.

m(6)A Modification of lncRNA NEAT1 Regulates Chronic Myelocytic Leukemia Progression via miR-766-5p/CDKN1A Axis.

BACKGROUND: Chronic myeloid leukemia (CML) is an acquired hematopoietic stem malignant disease originating from the myeloid system. Long non-coding RNAs (lncRNAs) have been widely explored in cancer tumorigenesis. However, their roles in CML remain largely unclear. METHODS: The peripheral blood mononuclear cells (PBMCs) and CML cell lines (K562, KCL22, MEG01, BV173) were collected for in vitro research. Real-time quantitative polymerase chain reaction was used to determine the mRNA expression levels. Cell viability and apoptosis were analyzed by cell counting kit 8 and flow cytometry assays. The targeting relationships were predicted using Starbase and TargetScan and ulteriorly verified by RNA pull-down and luciferase reporter assays. Western blotting assay was performed to assess the protein expressions. N6-methyladenosine (m6A) modification sites were predicted by SRAMP and confirmed by Methylated RNA immunoprecipitation (MeRIP) assay. RESULTS: LncRNA nuclear-enriched abundant transcript 1 (NEAT1) expression levels were decreased in the CML cell lines and PBMCs of CML patients. Moreover, METTL3-mediated m6A modification induced the aberrant expression of NEAT1 in CML. Overexpression of NEAT1 inhibited cell viability and promoted the apoptosis of CML cells. Additionally, miR-766-5p was upregulated in CML PBMCs and abrogated the effects of NEAT1 on cell viability and apoptosis of the CML cells. Further, CDKN1A was proved to be the target gene of miR-766-5p and was downregulated in the CML PBMCs. Knockdown of CDKN1A reversed the effects of NEAT1. CONCLUSION: The current research elucidates a novel METTL3/NEAT1/miR-766-5p/CDKN1A axis which plays a critical role in the progression of CML.

Quantitative Proteomic Analysis of Plasma Exosomes to Identify the Candidate Biomarker of Imatinib Resistance in Chronic Myeloid Leukemia Patients.

BACKGROUND: Imatinib (IM), a tyrosine kinase inhibitor (TKI), has markedly improved the survival and life quality of chronic myeloid leukemia (CML) patients. However, the lack of specific biomarkers for IM resistance remains a serious clinical challenge. Recently, growing evidence has suggested that exosome-harbored proteins were involved in tumor drug resistance and could be novel biomarkers for the diagnosis and drug sensitivity prediction of cancer. Therefore, we aimed to investigate the proteomic profile of plasma exosomes derived from CML patients to identify ideal biomarkers for IM resistance. METHODS: We extracted exosomes from pooled plasma samples of 9 imatinib-resistant CML patients and 9 imatinib-sensitive CML patients by ultracentrifugation. Then, we identified the expression levels of exosomal proteins by liquid chromatography-tandem mass spectrometry (LC-MS/MS) based label free quantification. Bioinformatics analyses were used to analyze the proteomic data. Finally, the western blot (WB) and parallel reaction monitoring (PRM) analyses were applied to validate the candidate proteins. RESULTS: A total of 2812 proteins were identified in plasma exosomes from imatinib-resistant and imatinib-sensitive CML patients, including 279 differentially expressed proteins (DEPs) with restricted criteria (fold change≥1.5 or ≤0.667, p<0.05). Compared with imatinib-sensitive CML patients, 151 proteins were up-regulated and 128 proteins were down-regulated. Bioinformatics analyses revealed that the main function of the upregulated proteins was regulation of protein synthesis, while the downregulated proteins were mainly involved in lipid metabolism. The top 20 hub genes were obtained using STRING and Cytoscape, most of which were components of ribosomes. Moreover, we found that RPL13 and RPL14 exhibited exceptional upregulation in imatinib-resistant CML patients, which were further confirmed by PRM and WB. CONCLUSION: Proteomic analysis of plasma exosomes provides new ideas and important information for the study of IM resistance in CML. Especially the exosomal proteins (RPL13 and RPL14), which may have great potential as biomarkers of IM resistance.