Prediction and Clinically Important Factors of Acute Kidney Injury Non-recovery

ObjectiveThis study aimed to identify phenotypic clinical features associated with acute kidney injury (AKI) to predict non-recovery from AKI at hospital discharge using electronic health record data.MethodsData for hospitalized patients in the AKI Recovery Evaluation Study were derived from a large healthcare delivery system in Taiwan between January 2011 and December 2017. Living patients with AKI non-recovery were used to derive and validate multiple predictive models. In total, 64 candidates variables, such as demographic characteristics, comorbidities, healthcare services utilization, laboratory values, and nephrotoxic medication use, were measured within 1 year before the index admission and during hospitalization for AKI.ResultsAmong the top 20 important features in the predictive model, 8 features had a positive effect on AKI non-recovery prediction: AKI during hospitalization, serum creatinine (SCr) level at admission, receipt of dialysis during hospitalization, baseline comorbidity of cancer, AKI at admission, baseline lymphocyte count, baseline potassium, and low-density lipoprotein cholesterol levels. The predicted AKI non-recovery risk model using the eXtreme Gradient Boosting (XGBoost) algorithm achieved an area under the receiver operating characteristic (AUROC) curve statistic of 0.807, discrimination with a sensitivity of 0.724, and a specificity of 0.738 in the temporal validation cohort.ConclusionThe machine learning model approach can accurately predict AKI non-recovery using routinely collected health data in clinical practice. These results suggest that multifactorial risk factors are involved in AKI non-recovery, requiring patient-centered risk assessments and promotion of post-discharge AKI care to prevent AKI complications.

Novel Immune-Related Gene-Based Signature Characterizing an Inflamed Microenvironment Predicts Prognosis and Radiotherapy Efficacy in Glioblastoma

Effective treatment of glioblastoma (GBM) remains an open challenge. Given the critical role of the immune microenvironment in the progression of cancers, we aimed to develop an immune-related gene (IRG) signature for predicting prognosis and improving the current treatment paradigm of GBM. Multi-omics data were collected, and various bioinformatics methods, as well as machine learning algorithms, were employed to construct and validate the IRG-based signature and to explore the characteristics of the immune microenvironment of GBM. A five-gene signature (ARPC1B, FCGR2B, NCF2, PLAUR, and S100A11) was identified based on the expression of IRGs, and an effective prognostic risk model was developed. The IRG-based risk model had superior time-dependent prognostic performance compared to well-studied molecular pathology markers. Besides, we found prominent inflamed features in the microenvironment of the high-risk group, including neutrophil infiltration, immune checkpoint expression, and activation of the adaptive immune response, which may be associated with increased hypoxia, epidermal growth factor receptor (EGFR) wild type, and necrosis. Notably, the IRG-based risk model had the potential to predict the effectiveness of radiotherapy. Together, our study offers insights into the immune microenvironment of GBM and provides useful information for clinical management of this desperate disease.

A Predictor Combining Clinical and Genetic Factors for AML1-ETO Leukemia Patients

Core Binding Factor (CBF)-AML is one of the most common somatic mutations in acute myeloid leukemia (AML). t(8;21)/AML1-ETO-positive acute myeloid leukemia accounts for 5-10% of all AMLs. In this study, we consecutively included 254 AML1-ETO patients diagnosed and treated at our institute from December 2009 to March 2020, and evaluated molecular mutations by 185-gene NGS platform to explore genetic co-occurrences with clinical outcomes. Our results showed that high KIT VAF(≥15%) correlated with shortened overall survival compared to other cases with no KIT mutation (3-year OS rate 26.6% vs 59.0% vs 69.6%, HR 1.50, 95%CI 0.78-2.89, P=0.0005). However, no difference was found in patients’ OS whether they have KIT mutation in two or three sites. Additionally, we constructed a risk model by combining clinical and molecular factors; this model was validated in other independent cohorts. In summary, our study showed that c-kit other than any other mutations would influence the OS in AML1-ETO patients. A proposed predictor combining both clinical and genetic factors is applicable to prognostic prediction in AML1-ETO patients.

Re-wiring and gene expression changes of AC025034.1 and ATP2B1 play complex roles in early-to-late breast cancer progression

Background Elucidating the dynamic topological changes across different stages of breast cancer, called stage re-wiring, could lead to identifying key latent regulatory signatures involved in cancer progression. Such dynamic regulators and their functions are mostly unknown. Here, we reconstructed differential co-expression networks for four stages of breast cancer to assess the dynamic patterns of cancer progression. A new computational approach was applied to identify stage-specific subnetworks for each stage. Next, prognostic traits of genes and the efficiency of stage-related groups were evaluated and validated, using the Log-Rank test, SVM classifier, and sample clustering. Furthermore, by conducting the stepwise VIF-feature selection method, a Cox-PH model was developed to predict patients’ risk. Finally, the re-wiring network for prognostic signatures was reconstructed and assessed across stages to detect gain/loss, positive/negative interactions as well as rewired-hub nodes contributing to dynamic cancer progression. Results After having implemented our new approach, we could identify four stage-specific core biological pathways. We could also detect an essential non-coding RNA, AC025034.1, which is not the only antisense to ATP2B1 (cell proliferation regulator), but also revealed a statistically significant stage-descending pattern; Moreover, AC025034.1 revealed both a dynamic topological pattern across stages and prognostic trait. We also identified a high-performance Overall-Survival-Risk model, including 12 re-wired genes to predict patients’ risk (c-index = 0.89). Finally, breast cancer-specific prognostic biomarkers of LINC01612, AC092142.1, and AC008969.1 were identified. Conclusions In summary new scoring method highlighted stage-specific core pathways for early-to-late progressions. Moreover, detecting the significant re-wired hub nodes indicated stage-associated traits, which reflects the importance of such regulators from different perspectives.

10 -Years Coronary Artery Disease Incidence and its Predictors in Overweight and Obese Adults Using A Competing Risk Model: A Longitudinal Study of Yazd Healthy Heart Cohort (YHHC)

We applied competing risk model to identify the predictors for Coronary Artery Disease (CAD) among 866 overweight and obese participants aged 20-74 years using their registered medical records in the first and second phase of Healthy Heart Cohort (YHHC) conducted in Yazd. These participants were free of coronary heart disease in the first phase of study. CAD was considered as the primary event and all other noncardiac deaths were considered as a competing event. The cumulative incidence of any CAD at the 5-year and 10-year follow-ups was approximately 6.8% and 10.6%, respectively, and approximately 4.6% and 8.5%, respectively, for all other noncardiac deaths. In both cause-specific and Fine-Gray models of risk factor diabetes type II, hypertriglyceridemia, university level of education (reversely), uric acid, age, systolic blood pressure and female gender (reversely) were associated with the increase risk of CAD. In addition to other traditional cardio metabolic risk factor we found that uric acid increased the risk of CAD in overweight and obese adults. It seems that lifestyle modification can reduce the risk of CAD. Also, high level of education had a protective effect on the risk of CAD. Both cause-specific and fine-gray models predicted similarly 10-years of CAD. The use of competing risk models in the presence of competing events is emphasized when interpreting survival studies.

Machine learning prediction model of acute kidney injury after percutaneous coronary intervention

Acute kidney injury (AKI) after percutaneous coronary intervention (PCI) is associated with a significant risk of morbidity and mortality. The traditional risk model provided by the National Cardiovascular Data Registry (NCDR) is useful for predicting the preprocedural risk of AKI, although the scoring system requires a number of clinical contents. We sought to examine whether machine learning (ML) techniques could predict AKI with fewer NCDR-AKI risk model variables within a comparable PCI database in Japan. We evaluated 19,222 consecutive patients undergoing PCI between 2008 and 2019 in a Japanese multicenter registry. AKI was defined as an absolute or a relative increase in serum creatinine of 0.3 mg/dL or 50%. The data were split into training (N = 16,644; 2008–2017) and testing datasets (N = 2578; 2017–2019). The area under the curve (AUC) was calculated using the light gradient boosting model (GBM) with selected variables by Lasso and SHapley Additive exPlanations (SHAP) methods among 12 traditional variables, excluding the use of an intra-aortic balloon pump, since its use was considered operator-dependent. The incidence of AKI was 9.4% in the cohort. Lasso and SHAP methods demonstrated that seven variables (age, eGFR, preprocedural hemoglobin, ST-elevation myocardial infarction, non-ST-elevation myocardial infarction/unstable angina, heart failure symptoms, and cardiogenic shock) were pertinent. AUC calculated by the light GBM with seven variables had a performance similar to that of the conventional logistic regression prediction model that included 12 variables (light GBM, AUC [training/testing datasets]: 0.779/0.772; logistic regression, AUC [training/testing datasets]: 0.797/0.755). The AKI risk model after PCI using ML enabled adequate risk quantification with fewer variables. ML techniques may aid in enhancing the international use of validated risk models.

Identification and Comprehensive Prognostic Analysis of a Novel Chemokine-Related lncRNA Signature and Immune Landscape in Gastric Cancer

Gastric cancer (GC) is a malignant tumor with poor survival outcomes. Immunotherapy can improve the prognosis of many cancers, including GC. However, in clinical practice, not all cancer patients are sensitive to immunotherapy. Therefore, it is essential to identify effective biomarkers for predicting the prognosis and immunotherapy sensitivity of GC. In recent years, chemokines have been widely reported to regulate the tumor microenvironment, especially the immune landscape. However, whether chemokine-related lncRNAs are associated with the prognosis and immune landscape of GC remains unclear. In this study, we first constructed a novel chemokine-related lncRNA risk model to predict the prognosis and immune landscape of GC patients. By using various algorithms, we identified 10 chemokine-related lncRNAs to construct the risk model. Then, we determined the prognostic efficiency and accuracy of the risk model. The effectiveness and accuracy of the risk model were further validated in the testing set and the entire set. In addition, our risk model exerted a crucial role in predicting the infiltration of immune cells, immune checkpoint genes expression, immunotherapy scores and tumor mutation burden of GC patients. In conclusion, our risk model has preferable prognostic performance and may provide crucial clues to formulate immunotherapy strategies for GC.

RNA N6-Methyladenosine Regulators Contribute to Tumor Immune Microenvironment and Have Clinical Prognostic Impact in Breast Cancer

Purpose: This study aims to reveal the relationship between RNA N6-methyladenosine (m6A) regulators and tumor immune microenvironment (TME) in breast cancer, and to establish a risk model for predicting the occurrence and development of tumors.Patients and methods: In the present study, we respectively downloaded the transcriptome dataset of breast cancer from Gene Expression Omnibus (GEO) database and The Cancer Genome Atlas (TCGA) database to analyze the mutation characteristics of m6A regulators and their expression profile in different clinicopathological groups. Then we used the weighted correlation network analysis (WGCNA), the least absolute shrinkage and selection operator (LASSO), and cox regression to construct a risk prediction model based on m6A-associated hub genes. In addition, Immune infiltration analysis and gene set enrichment analysis (GSEA) was used to evaluate the immune cell context and the enriched gene sets among the subgroups.Results: Compared with adjacent normal tissue, differentially expressed 24 m6A regulators were identified in breast cancer. According to the expression features of m6A regulators above, we established two subgroups of breast cancer, which were also surprisingly distinguished by the feature of the immune microenvironment. The Model based on modification patterns of m6A regulators could predict the patient’s T stage and evaluate their prognosis. Besides, the low m6aRiskscore group presents an immune-activated phenotype as well as a lower tumor mutation load, and its 5-years survival rate was 90.5%, while that of the high m6ariskscore group was only 74.1%. Finally, the cohort confirmed that age (p < 0.001) and m6aRiskscore (p < 0.001) are both risk factors for breast cancer in the multivariate regression.Conclusion: The m6A regulators play an important role in the regulation of breast tumor immune microenvironment and is helpful to provide guidance for clinical immunotherapy.

Inference for cumulative risk model under step-stress experiments and its application in nanocrystalline data

With the popularity of step-stress accelerated life testing, researchers are exploring more possibilities for models that relate the life distributions under different stress levels. Cumulative risk model assumes that the effects of stress changes have a lag period before they are fully observed, which guarantees the continuity of the hazard rate function. This paper studies the cumulative risk model for Lomax distribution with step-stress experiments. For maximum likelihood estimation, Newton-Rapson method is adopted to get point estimates. Meanwhile, the asymptotic normality of the maximum likelihood estimator is used to obtain asymptotic confidence intervals. For Bayesian estimation, point estimates and highest posterior density credible intervals under squared error loss function with informative prior and non-informative prior are derived using Metropolis-Hastings method and Metropolis-Hastings within Gibbs algorithm. To evaluate the effects of stress change time and the length of lag period, as well as the performance of different methods, numerical simulations are conducted. Then a real nanocrystalline data set is analyzed.