Prognostic Performance Research Articles

Active Left Atrial Ejection Fraction as a Non-Invasive Marker in Pulmonary Hypertension Secondary to Heart Failure

BackgroundPatients with pulmonary hypertension (PH) secondary to left heart failure (HF) exhibit a complex pathophysiological profile and poor prognosis. Left atrial function is pivotal in the progression of this disease, yet its predictive significance remains exclusive. This study aimed to explore the predictive capability of LA metrics in this population and compare them with other common predictors. MethodsIn this retrospective study, consecutive patients with PH secondary to HF who underwent cardiac magnetic resonance (CMR) imaging between December 2010 and December 2021 were enrolled. The composite endpoint was defined as all-cause death, heart-lung transplantation, or left ventricular assist device implantation. Survival analyses were performed using Kaplan–Meier curves and Cox regression analyses. ResultsA total of 174 patients with PH secondary to HF, with a mean age of 53.2 ± 14.9 years, including 90 men, were included in the final analysis. During a median follow-up of 31.9 months, 58 patients (33.3%) with PH reached the endpoints. There was a fair correlation between active left atrial ejection fraction (LAEF) and pulmonary artery wedge pressure (r = -0.397, p = 0.044). Active LAEF had a strong correlation with oxygen consumption at anaerobic threshold (r = 0.769, p < 0.001) and peak oxygen consumption (r = 0.754, p < 0.001). Active LAEF demonstrated comparable prognostic performance to other variables measured by echocardiography or CMR. After adjusting for clinical variables and left ventricular ejection fraction, active LAEF was still an independent predictor for adverse events (C-statistic: 0.784). Subgroup analysis among HF patients with preserved ejection fraction demonstrated that those with active LAEF ≤ 8.6% had a 7.05-fold higher risk of experiencing the composite endpoint compared to those with active LAEF > 8.6%. ConclusionAlthough active LAEF does not demonstrate statistical improvement in outcome discrimination compared to established metrics, it may still merit consideration for assessing disease severity and prognosis in patients with PH secondary to HF. The integration of active LAEF and HF subtypes may stratify individuals at different levels of risk.

Machine learning-based cell death marker for predicting prognosis and identifying tumor immune microenvironment in prostate cancer

BackgroundProstate cancer (PCa) incidence and mortality rates are rising, necessitating precise prognostic tools to guide personalized treatment. Dysregulation of programmed cell death pathways in tumor suppression and cancer development has garnered increasing attention, providing a new research direction for identifying biomarkers and potential therapeutic targets. MethodsIntegrating multiple database resources, we constructed and optimized a prognostic signature based on the expression of programmed cell death-related genes (PCDRG) using ten machine learning algorithms. Model performance and prognostic effects were further evaluated. We analyzed the relationships between signature and clinicopathological features, somatic mutations, drug sensitivity, and the tumor immune microenvironment, and constructed a nomogram. The expression level of PCDRGs were evaluated and compared. ResultsOf 1560 PCDRGs, 149 were differentially expressed in PCa, with 34 associated with biochemical recurrence. The PCDRG-derived index (PCDI), constructed using the random forest algorithm, exhibited optimal prognostic performance, successfully stratifying PCa patients into two groups with significant prognostic differences. Patients with high PCDI scores exhibited poorer survival and lower immunotherapy benefit. PCDI was closely associated with the infiltration of specific immune cells, particularly positive correlations with macrophages and T helper cells, and negative correlations with neutrophils, suggesting that PCDI may influence the tumor immune microenvironment, thereby affecting patient prognosis and treatment response. PCDI was associated with age, pathological stage, somatic mutations, and drug sensitivity. The PCDI-based nomogram demonstrated excellent performance in predicting biochemical recurrence in PCa patients. Finally, the differential expression of these PCDRGs was verified based on cell lines and PCa patient expression profile data. ConclusionThis study developed an effective prognostic indicator for prostate cancer, PCDI, using machine learning approaches. PCDI reflects the link between aberrant programmed cell death pathways and disease progression and treatment response.

Distributed non-disclosive validation of predictive models by a modified ROC-GLM

BackgroundDistributed statistical analyses provide a promising approach for privacy protection when analyzing data distributed over several databases. Instead of directly operating on data, the analyst receives anonymous summary statistics, which are combined into an aggregated result. Further, in discrimination model (prognosis, diagnosis, etc.) development, it is key to evaluate a trained model w.r.t. to its prognostic or predictive performance on new independent data. For binary classification, quantifying discrimination uses the receiver operating characteristics (ROC) and its area under the curve (AUC) as aggregation measure. We are interested to calculate both as well as basic indicators of calibration-in-the-large for a binary classification task using a distributed and privacy-preserving approach.MethodsWe employ DataSHIELD as the technology to carry out distributed analyses, and we use a newly developed algorithm to validate the prediction score by conducting distributed and privacy-preserving ROC analysis. Calibration curves are constructed from mean values over sites. The determination of ROC and its AUC is based on a generalized linear model (GLM) approximation of the true ROC curve, the ROC-GLM, as well as on ideas of differential privacy (DP). DP adds noise (quantified by the ℓ2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ell _2$$\\end{document} sensitivity Δ2(f^)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\Delta _2(\\hat{f})$$\\end{document}) to the data and enables a global handling of placement numbers. The impact of DP parameters was studied by simulations.ResultsIn our simulation scenario, the true and distributed AUC measures differ by ΔAUC<0.01\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\Delta \ ext {AUC} < 0.01$$\\end{document} depending heavily on the choice of the differential privacy parameters. It is recommended to check the accuracy of the distributed AUC estimator in specific simulation scenarios along with a reasonable choice of DP parameters. Here, the accuracy of the distributed AUC estimator may be impaired by too much artificial noise added from DP.ConclusionsThe applicability of our algorithms depends on the ℓ2\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\ell _2$$\\end{document} sensitivity Δ2(f^)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\Delta _2(\\hat{f})$$\\end{document} of the underlying statistical/predictive model. The simulations carried out have shown that the approximation error is acceptable for the majority of simulated cases. For models with high Δ2(f^)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\Delta _2(\\hat{f})$$\\end{document}, the privacy parameters must be set accordingly higher to ensure sufficient privacy protection, which affects the approximation error. This work shows that complex measures, as the AUC, are applicable for validation in distributed setups while preserving an individual’s privacy.

The Importance of Being Grade 3: A Plea for a Three-tier Hybrid Classification System for Grade in Primary Non–muscle-invasive Bladder Cancer

Grade is an important determinant of progression in non–muscle-invasive bladder cancer. Although the World Health Organization (WHO) 2004/2016 grading system is recommended, other systems such as WHO1973 and WHO1999 are still widely used. Recently, a hybrid (three-tier) system was proposed, separating WHO2004/2016 high grade (HG) into HG/grade 2 (G2) and HG/G3 while maintaining low grade. We assessed the prognostic performance of HG/G3 and HG/G2. Three independent cohorts with 9712 primary (first diagnosis) Ta-T1 bladder tumors were analyzed. Time to progression was analyzed with cumulative incidence functions and Cox regression models. Harrell’s C-index was used to assess discrimination. Time to progression was significantly shorter for HG/G3 than for HG/G2 in multivariable analyses (cohort 1: hazard ratio [HR] = 1.92; cohort 2: HR = 2.51, and cohort 3: HR = 1.69). Corresponding progression risks at 5 yr were 18%, 20%, and 18% for HG/G3 versus 7.3%, 7.5%, and 9.3% for HG/G2, respectively. Cox models using hybrid grade performed better than models with WHO2004/2016 (all cohorts; p < 0.001). For the three cohorts, C-indices for WHO2004/2016 were 0.69, 0.62, and 0.75, while, for hybrid grade, C-indices were 0.74, 0.68, and 0.78, respectively. Subdividing the HG category into HG/G2 and HG/G3 stratifies time to progression and supports the recommendation to adopt the hybrid grading system for Ta/T1 bladder cancers.

Prognostic performance of FIGO 2023 endometrial carcinoma staging: a comparison to FIGO 2009 staging in the setting of known and unknown molecular classification.

The 2023 FIGO staging criteria for endometrial cancer (EC) introduced marked changes from the 2009 version. The full implication of these changes for patient diagnosis and treatment is unknown. We evaluate the differences in staging and prognostication between the two systems, with and without inclusion of molecular classification. We assigned (1) FIGO 2009, (2) 2023 molecular-agnostic and (3) 2023 molecular-informed stages to 404 fully staged and molecularly classified patients with EC. Disease-specific and progression/relapse-free survival were analysed via the Kaplan-Meier method and compared with log-rank testing; 118 of 252 (47%) FIGO 2009 stage I patients were upstaged based on histopathological findings alone. Stage I/II subgroup survival distribution analysis showed a worse prognosis in FIGO 2023 IIB and IIC patients. In the molecular-informed FIGO 2023 system, three of 15 (20%) POLE-mutated stage I/II cases were downstaged from FIGO 2009 and eight (53%) were downstaged from molecular-agnostic FIGO 2023. Fifty-one of 60 (85%) p53-abnormal tumours were upstaged from the FIGO 2009, whereas 13 of 60 (22%) were upstaged from the 2023 molecular-agnostic stage. Molecular classification improved prognostic stratification for both 2009 and 2023 FIGO systems. Downstaging based on POLE mutation more accurately represents patient outcomes. However, in the absence of known POLE status, applying molecular-agnostic FIGO 2023 criteria for stage I/II disease should be conducted with caution. For aggressive histotypes, additionally reporting FIGO 2009 stage should be considered. Upstaging based on substantial lymphovascular space invasion, aggressive histotype with any myometrial invasion and abnormal p53 improves prognostic discernment. Further subdivisions within stage I/II provide minimal additional prognostic information.

The prognostic utility of Lactate/Albumin*Age score in septic patient with normal lactate level

BackgroundA previous study has shown that the lactate/albumin*age (LAA) score is useful for predicting mortality in patients with sepsis admitted to the ICU. We aimed to evaluate the clinical significance of the LAA score in patients with sepsis who presented to the emergency department (ED) MethodsThis retrospective observational study used data from the Korean Shock Society Registry collected between January 2017 and December 2021. The prognostic performance of the LAA score for predicting the 28-day mortality was evaluated. Lactate and albumin levels were measured immediately after arrival to the ED. ResultsOf the 5346 patients with sepsis, data from 3240 were analyzed. The area under the receiver operating characteristic curve (AUROC) of the LAA score (0.737, 95 % confidence interval (CI) 0.716–0.757), was higher than that of lactate (0.699, 95 % CI 0.677–0.720, p < 0.001), lactate/albumin (LA) ratio (0.730, 95 % CI 0.709–0.751, p = 0.016), and Sequential Organ Failure Assessment (SOFA) score (0.698, 95 % confidence interval 0.676–0.720, p = 0. 004), and Acute Physiology and Chronic Health Evaluation (APACHE) II scores (0.672; 95 % confidence interval 0.649–0.694, p < 0.001). The optimal cut-off value for the LAA score was 119.9. In the Kaplan-Meier analysis according to the optimal cutoff value, the 28-day mortality rates were higher in the high LAA score group (log-rank test, p < 0.001). The LAA score was independently associated with 28-day mortality in the multivariate Cox proportional hazards model (adjusted hazard ratio 2.07, 95 % CI 1.76–2.43, p < 0.001). In the normal (<2 mmol/L) lactate group, the AUROC value for LAA score was higher than LA ratio (normal group 0.674 vs 0.634, p < 0.004). In patients over 65 years old, LAA score (0.731) showed a higher AUROC value than LA ratio (0.725). (p < 0.001) ConclusionThe LAA score may be used as an independent predictor of mortality in patients with sepsis in the emergency department. Our results show that it performs better than serum lactate alone, LA ratio, and SOFA and APACHE II scores. While this suggests that the LAA could provide clinicians with a useful tool for timely early intervention and care planning in patients with a poor prognosis, further validation in large multicenter prospective studies are necessary to confirm its reliability and practicality as a readily available and objective biomarker.

NINJ1: A Novel Sepsis Severity and Mortality Biomarker.

Multiple cell death modalities are implicated in sepsis pathobiology. However, the clinical relevance of NINJ1, a key mediator of plasma membrane rupture during lytic cell death, in sepsis progression and outcomes has remained poorly explored. Circulating NINJ1 levels were measured in 116 septic ICU patients, 16 non-septic ICU controls, and 16 healthy controls. Comparative analysis of serum NINJ1 across these groups was performed. Correlations between NINJ1 and clinical disease severity scores (SOFA, APACHE II) as well as laboratory parameters were examined in the sepsis cohort. Furthermore, we assessed the prognostic performance of NINJ1 for predicting 28-day mortality in septic patients using receiver operating characteristic (ROC) analyses. Circulating NINJ1 levels were elevated in septic patients and positively correlated with sepsis severity scores. NINJ1 also showed positive correlations with liver injury markers (AST/ALT) and coagulation parameters (D-dimer, APTT, PT, TT) in sepsis. Further analysis using the ISTH overt DIC scoring system revealed an association between NINJ1 and sepsis-induced coagulopathy.ROC analysis demonstrated NINJ1 outperformed traditional inflammatory biomarkers PCT and CRP in predicting 28-day sepsis mortality, although its prognostic accuracy was lower than SOFA and APACHE II scores. Combining NINJ1 with SOFA improved mortality prediction from an AUC of 0.6843 to 0.773. Circulating NINJ1 serves as a novel sepsis biomarker indicative of disease severity, coagulopathy and mortality risk, and its integration with SOFA and APACHE II scores substantially enhances prognostic risk stratification. These findings highlight the prospective clinical utility of NINJ1 for sepsis prognostication and monitoring, warranting further validation studies to facilitate implementation.

Longitudinal Changes of CT-radiomic and Systemic Inflammatory Features Predict Survival in Advanced Non-Small Cell Lung Cancer Patients Treated With Immune Checkpoint Inhibitors.

This study aims to determine whether longitudinal changes in CT radiomic features (RFs) and systemic inflammatory indices outperform single-time-point assessment in predicting survival in advanced non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors (ICIs). We retrospectively acquired pretreatment (T0) and first disease assessment (T1) RFs and systemic inflammatory indices from a single-center cohort of stage IV NSCLC patients and computed their delta (Δ) variation as [(T1-T0)/T0]. RFs from the primary tumor were selected for building baseline-radiomic (RAD) and Δ-RAD scores using the linear combination of standardized predictors detected by LASSO Cox regression models. Cox models were generated using clinical features alone or combined with baseline and Δ blood parameters and integrated with baseline-RAD and Δ-RAD. All models were 3-fold cross-validated. A prognostic index (PI) of each model was tested to stratify overall survival (OS) through Kaplan-Meier analysis. We included 90 ICI-treated NSCLC patients (median age 70y [IQR=42 to 85], 63 males). Δ-RAD outperformed baseline-RAD for predicting OS [c-index: 0.632 (95%CI: 0.628 to 0.636) vs. 0.605 (95%CI: 0.601 to 0.608) in the test splits]. Integrating longitudinal changes of systemic inflammatory indices and Δ-RAD with clinical data led to the best model performance [Integrated-Δ model, c-index: 0.750 (95% CI: 0.749 to 0.751) in training and 0.718 (95% CI: 0.715 to 0.721) in testing splits]. PI enabled significant OS stratification within all the models (P-value <0.01), reaching the greatest discriminative ability in Δ models (high-risk group HR up to 7.37, 95% CI: 3.9 to 13.94, P<0.01). Δ-RAD improved OS prediction compared with single-time-point radiomic in advanced ICI-treated NSCLC. Integrating Δ-RAD with a longitudinal assessment of clinical and laboratory data further improved the prognostic performance.

CT-based multimodal deep learning for non-invasive overall survival prediction in advanced hepatocellular carcinoma patients treated with immunotherapy

ObjectivesTo develop a deep learning model combining CT scans and clinical information to predict overall survival in advanced hepatocellular carcinoma (HCC).MethodsThis retrospective study included immunotherapy-treated advanced HCC patients from 52 multi-national in-house centers between 2018 and 2022. A multi-modal prognostic model using baseline and the first follow-up CT images and 7 clinical variables was proposed. A convolutional-recurrent neural network (CRNN) was developed to extract spatial-temporal information from automatically selected representative 2D CT slices to provide a radiological score, then fused with a Cox-based clinical score to provide the survival risk. The model’s effectiveness was assessed using a time-dependent area under the receiver operating curve (AUC), and risk group stratification using the log-rank test. Prognostic performances of multi-modal inputs were compared to models of missing modality, and the size-based RECIST criteria.ResultsTwo-hundred seven patients (mean age, 61 years ± 12 [SD], 180 men) were included. The multi-modal CRNN model reached the AUC of 0.777 and 0.704 of 1-year overall survival predictions in the validation and test sets. The model achieved significant risk stratification in validation (hazard ratio [HR] = 3.330, p = 0.008), and test sets (HR = 2.024, p = 0.047) based on the median risk score of the training set. Models with missing modalities (the single-modal imaging-based model and the model incorporating only baseline scans) can still achieve favorable risk stratification performance (all p < 0.05, except for one, p = 0.053). Moreover, results proved the superiority of the deep learning-based model to the RECIST criteria.ConclusionDeep learning analysis of CT scans and clinical data can offer significant prognostic insights for patients with advanced HCC.Critical relevance statementThe established model can help monitor patients’ disease statuses and identify those with poor prognosis at the time of first follow-up, helping clinicians make informed treatment decisions, as well as early and timely interventions.Key PointsAn AI-based prognostic model was developed for advanced HCC using multi-national patients.The model extracts spatial-temporal information from CT scans and integrates it with clinical variables to prognosticate.The model demonstrated superior prognostic ability compared to the conventional size-based RECIST method.Graphical

Validation of the cutoff values for the number of metastatic lymph nodes for esophageal cancer staging: a multi-institutional analysis of 655 patients in Japan.

The number of metastatic lymph nodes (LNs) is an important prognostic factor for esophageal cancer, and N staging is important for prognostic stratification. The optimal cutoff values for clinical (cN) and pathologic N (pN) staging should be reconsidered following advances in neoadjuvant therapy. The study included 655 patients who underwent esophagectomy between January 2014 and December 2016 in four high-volume centers in Japan. Optimal cutoff values for the number of metastatic LNs in cN and pN staging were examined using X-tile, and their prognostic performance was validated using the Kaplan-Meier method. The cutoff values were 1, 2, and 3 for cN staging and 1, 3, and 7 for pN staging. Prognosis was significantly better in patients with cN0 than in those with modified (m)-cN1 (p = 0.0211). However, prognosis was not significantly different among the patients with m-cN1, m-cN2, and m-cN3 disease. Prognosis was significantly different among the patients with pN0, pN1, pN2, and pN3 disease (pN0 vs pN1, p < 0.0001; pN1 vs pN2, p < 0.0001; pN2 vs pN3, p < 0.0001). In patients who received preoperative neoadjuvant therapy, prognosis, which was not significantly different among the patients with cN0, m-cN1, m-cN2, and m-cN3 disease (cN0 vs m-cN1, p = 0.5675; m-cN1 vs m-cN2, p = 0.4425; m-cN2 vs m-cN3, p = 0.7111), was significantly different among the patients with pN0, pN1, pN2, and pN3 disease (pN0 vs pN1, p = 0.0025; pN1 vs pN2, p = 0.0046; pN2 vs pN3, p = 0.0104). cN has no prognostic impact in patients who underwent preoperative treatment followed by esophagectomy, despite the optimization of cN classification. The conventional TNM8th pN classification is useful for predicting prognosis even for patients who have undergone preoperative treatment. The conventional cutoffs for metastatic LNs in the International Union against Cancer tumor node metastasis staging system are valid and can be effectively used in clinical practice.

Predictive role of hematological indices in patients with acute coronary syndrome in Ethiopia: Intrahospital outcomes

BackgroundApart from troponins, various additional biomarkers that indicate myocardial injury, inflammation, thrombosis, and other routes are being studied to improve the treatment of acute coronary syndrome (ACS). Myeloid activity has been found to be elevated in ACS, and this has sparked a great deal of interest in hematological parameters since they might offer independent insights into pathophysiology and risk assessment. ObjectiveThe purpose of this study was to evaluate the hematological markers' prognostic ability for all intrahospital causes of mortality in individuals with an ACS diagnosis. MethodsA long-term cohort study based at an institution was done. At Jimma Medical Center, patients with an ACS diagnosis were progressively brought in between May 1, 2022, and October 31, 2023. Complete blood counts (CBC) and biochemical analysis were carried out. Multilevel mixed effect logistic regression was computed to evaluate the predictive competence of hematological indices on intrahospital mortality. Prognostic performance of hematological parameters was done using the ROC curve analysis. ResultA total of 110 patients were included, of which 99 (90 %) were diagnosed ST-elevation myocardial infarction, and 74 (67.3 %) were men. The mean age was 56 (±11) years. RDW, platelet count, and MCV were independently associated with intrahospital mortality (AOR = 1.20 with P < 0.001, AOR = 0.995 with P < 0.03, and AOR = 0.897 with P < 0.025, respectively). The predictive power of RDW-SD for intrahospital mortality was evaluated by ROC analysis, the AUC value were 0.737 (95 % CI 0.669–0.805). ConclusionThis study found that red cell distribution width, mean corpuscular volume, and platelets were predictive factors for intrahospital death in patients with ACS. Thus, it is possible to predict the prognosis of an ACS patient using hematological data.

Comprehensive characterization of cytokines in patients under extracorporeal membrane oxygenation: Evidence from integrated bulk and single-cell RNA sequencing data using multiple machine learning approaches.

ECMO (extracorporeal membrane oxygenation) is an effective technique for providing short-term mechanical support to the heart, lungs, or both. During ECMO treatment, the inflammatory response, particularly involving cytokines, plays a crucial role in pathophysiology. However, the potential effects of cytokines on patients receiving ECMO are not comprehensively understood. We acquired three ECMO support datasets, namely two bulk and one single-cell RNA sequencing (RNA-seq), from the GEO (Gene Expression Omnibus) combined with hospital cohorts to investigate the expression pattern and potential biological processes of cytokine-related genes (CRGs) in patients under ECMO. Subsequently, machine learning approaches, including support vector machine (SVM), random forest (RF), modified Lasso penalized regression, extreme gradient boosting (XGBoost), and artificial neural network (ANN), were applied to identify hub CRGs, thus developing a prediction model called CRG classifier. The predictive and prognostic performance of the model was comprehensively evaluated in GEO and hospital cohorts. Finally, we mechanistically analyzed the relationship between hub cytokines, immune cells, and pivotal molecular pathways. Analyzing bulk and single-cell RNA-seq data revealed that most CRGs were significantly differentially expressed, the enrichment scores of cytokine and cytokine cytokine receptor (CCR) interaction were significantly higher during ECMO. Based on multiple machine learning algorithms, nine key CRGs (CCL2, CCL4, IFNG, IL1R2, IL20RB, IL31RA, IL4, IL7, and IL7R) were used to develop the CRG classifier. The CRG classifier exhibited excellent prognostic values (AUC > 0.85), serving as an independent risk factor. It performed better in predicting mortality and yielded a larger net benefit than other clinical features in GEO and hospital cohorts. Additionally, IL1R2, CCL4, and IL7R were predominantly expressed in monocytes, NK cells, and T cells, respectively. Their expression was significantly positively correlated with the relative abundance of corresponding immune cells. Gene set variation analysis (GSVA) revealed that parainflammation, complement and coagulation cascades, and IL6/JAK/STAT3 signaling were significantly enriched in the subgroup that died after receiving ECMO. Spearman correlation analyses and Mantel tests revealed that the expression of hub cytokines (IL1R2, CCL4, and IL7R) and pivotal molecular pathways scores (complement and coagulation cascades, IL6/JAK/STAT3 signaling, and parainflammation) were closely related. A predictive model (CRG classifier) comprising nine CRGs based on multiple machine learning algorithms was constructed, potentially assisting clinicians in guiding individualized ECMO treatment. Additionally, elucidating the underlying mechanistic pathways of cytokines during ECMO will provide new insights into its treatment.

Value of Cerebroplacental Ratio in Predicting Adverse Perinatal Outcome in Term Pregnancies Complicated by Obesity

AbstractTo evaluate the performance of cerebroplacental ratio (CPR) in predicting composite adverse perinatal outcome (CAPO) in women with obesity compared to non-obese women at term.This is a retrospective cohort study in a single tertiary referral centre over a 3-year period. All singleton pregnancies with CPR measurements ≥ 37 + 0 weeks and estimated fetal weight ≥ 10th centile and attempted vaginal delivery were included and divided into two groups defined by pre-pregnancy body mass index (BMI) &lt;/≥ 30 kg/m2. The presence of at least one of the following outcome parameters was defined as CAPO: operative delivery (OD) due to intrapartum fetal compromise (IFC), admission to the neonatal intensive care unit, umbilical cord arterial pH ≤ 7.15, 5 min Apgar &lt; 7. The prognostic performance of CPR MoM was evaluated using receiver operating characteristic (ROC) analysis.The study cohort included 1207 pregnancies, of which 112 were women with a BMI ≥ 30 kg/m2. In obese women, CAPO occurred in 21 cases (18.8%) compared to 247 (22.6%) cases in women with BMI &lt; 30 kg/m2 (p = 0.404). In the entire study cohort, CPR MoM was significantly lower in the CAPO and OD for IFC group. ROC analyses revealed a significant predictive value of low CPR MoM for CAPO in obese women (AUC = 0.64, p = 0.024). Furthermore, CPR was predictive for OD for IFC not only in obese (AUC = 0.72, p = 0.023) but also in non-obese (AUC = 0.61, p = 0.003) women.Low CPR MoM was predictive for CAPO and OD for IFC in obese women without additional risk factors. However, the overall predictive performance of CPR for CAPO in obese women was poor.

Turning to immunosuppressive tumors: Deciphering the immunosenescence-related microenvironment and prognostic characteristics in pancreatic cancer, in which GLUT1 contributes to gemcitabine resistance

Increasing evidence indicates that the remodeling of immune microenvironment heterogeneity influences pancreatic cancer development, as well as sensitivity to chemotherapy and immunotherapy. However, a gap remains in the exploration of the immunosenescence microenvironment in pancreatic cancer. In this study, we identified two immunosenescence-associated isoforms (IMSP1 and IMSP2), with consequential differences in prognosis and immune cell infiltration. We constructed the MLIRS score, a hazard score system with robust prognostic performance (area under the curve, AUC = 0.91), based on multiple machine learning algorithms (101 cross-validation methods). Patients in the high MLIRS score group had worse prognosis (P < 0.0001) and lower abundance of immune cell infiltration. Conversely, the low MLIRS score group showed better sensitivity to chemotherapy and immunotherapy. Additionally, our MLIRS system outperformed 68 other published signatures. We identified the immunosenescence microenvironmental windsock GLUT1 with certain co-expression properties with immunosenescence markers. We further demonstrated its positive modulation ability of proliferation, migration, and gemcitabine resistance in pancreatic cancer cells. To conclude, our study focused on training of composite machine learning algorithms in multiple datasets to develop a robust machine learning modeling system based on immunosenescence and to identify an immunosenescence-related microenvironment windsock, providing direction and guidance for clinical prediction and application.

Microbiome of esophageal endoscopic wash samples is associated with resident flora in the esophagus and incidence of cancer

Change in mucosal microbiome is associated with various types of cancer in digestive tract. We hypothesized that microbial communities in the esophageal endoscopic wash fluids reflects resident flora in esophageal mucosa that is associated with esophageal carcinoma (EC) risk and/or directly correlates microbiome derived from EC tumor tissue. Studying microbial communities in esophageal endoscopic wash samples would be therefore useful to predict the incidence or risk of EC. We examined microbial communities of the endoscopic wash samples from 45 primary EC and 20 respective non-EC controls using 16S rRNA V3-V4 amplicon sequencing. The result was also compared with microbial communities in matched endoscopic biopsies from EC and non-cancerous esophageal mucosa. Compared with non-EC controls, 6 discriminative bacterial genera were detected in EC patients. Among them, relative abundance ratio of Prevotella and Shuttlewarthia, as well as decrease of genus Prevotella presented good prognostic performance to discriminate EC from controls (area under curve, 0.86, 0.82, respectively). Multivariate analysis showed occurrence of EC was an independent factor associated with decrease of this bacteria. Abundance of genus Prevotella in the esophageal endoscopic wash samples was significantly correlated with the abundance of this bacteria in the matched endoscopic biopsies from non-cancerous esophageal mucosa but not in the EC tissues. Our findings suggest that microbiome composition in the esophageal endoscopic wash samples reflects resident flora in the esophagus and significantly correlates with the incidence of EC.

A novel prognostic model based on pyroptosis signature in AML

Acute myeloid leukemia (AML), a highly heterogeneous myeloid malignancy, remains a challenge in terms of proper risk stratification. In this study, we developed a novel pyroptosis prognostic model based on pyroptosis-related gene pairs, which exhibited excellent prognostic performance across multiple cohorts (N = 1506) and accurately predicted both adult and pediatric AML prognosis. Additionally, we integrated the pyroptosis risk model with other clinical risk factors to construct a highly operational nomogram. Moreover, our findings indicate a significant correlation between elevated pyroptosis risk scores and increased stemness of AML. Using CIBERSORT immune analysis, we found a decreased proportion of resting NK cells and activated mast cells in the high-risk group. Through analyzing the correlation between chemotherapy drug response sensitivity and risk scores, we found that AZD1332 and BPD-0008900 were extremely sensitive in the high-risk group.

Predicting 1, 2 and 3 year emergent referable diabetic retinopathy and maculopathy using deep learning

BackgroundPredicting diabetic retinopathy (DR) progression could enable individualised screening with prompt referral for high-risk individuals for sight-saving treatment, whilst reducing screening burden for low-risk individuals. We developed and validated deep learning systems (DLS) that predict 1, 2 and 3 year emergent referable DR and maculopathy using risk factor characteristics (tabular DLS), colour fundal photographs (image DLS) or both (multimodal DLS).MethodsFrom 162,339 development-set eyes from south-east London (UK) diabetic eye screening programme (DESP), 110,837 had eligible longitudinal data, with the remaining 51,502 used for pretraining. Internal and external (Birmingham DESP, UK) test datasets included 27,996, and 6928 eyes respectively.ResultsInternal multimodal DLS emergent referable DR, maculopathy or either area-under-the receiver operating characteristic (AUROC) were 0.95 (95% CI: 0.92–0.98), 0.84 (0.82–0.86), 0.85 (0.83–0.87) for 1 year, 0.92 (0.87–0.96), 0.84 (0.82–0.87), 0.85 (0.82–0.87) for 2 years, and 0.85 (0.80–0.90), 0.79 (0.76–0.82), 0.79 (0.76–0.82) for 3 years. External multimodal DLS emergent referable DR, maculopathy or either AUROC were 0.93 (0.88–0.97), 0.85 (0.80–0.89), 0.85 (0.76–0.85) for 1 year, 0.93 (0.89–0.97), 0.79 (0.74–0.84), 0.80 (0.76–0.85) for 2 years, and 0.91 (0.84–0.98), 0.79 (0.74–0.83), 0.79 (0.74–0.84) for 3 years.ConclusionsMultimodal and image DLS performance is significantly better than tabular DLS at all intervals. DLS accurately predict 1, 2 and 3 year emergent referable DR and referable maculopathy using colour fundal photographs, with additional risk factor characteristics conferring improvements in prognostic performance. Proposed DLS are a step towards individualised risk-based screening, whereby AI-assistance allows high-risk individuals to be closely monitored while reducing screening burden for low-risk individuals.

Short- and long-term prognostic performance of exercise ECG and myocardial perfusion SPECT.

Myocardial perfusion SPECT(MPS) and exercise electrocardiography(Ex-ECG) results are of prognostic importance for short-term follow up duration. However, the value of MPS or Ex-ECG findings for long-term risk assessment is less evident as underlying risk factors for ischemic heart disease (IHD) gain in importance. To assess the short- and long-term prognostic value of MPS and Ex-ECG in relation to known risk factors. An observational study of 908 patients (age 63 years, 49% male, 45% prior IHD) referred for MPS and Ex-ECG. Follow-up was divided into two periods (short-term: <5 years and long-term: >5 years). Cardiac events were defined as a composite of acute myocardial infarction, unstable angina, unplanned revascularization and cardiovascular death. The composite endpoint occurred in 95 patients (short-term follow up) and in 94 patients (long-term follow up). In multivariable models stress testing had a strong predictive value for short-term follow up (HR for MPS = 2.9, CI = 1.9-4.5, p < 0.001 and HR for Ex-ECG = 2.1, CI 1.3-3.3, p = 0.002), but no predictive value for long-term follow up (HR for MPS = 0.9, CI = 0.5-1.5, p = 0.70 and HR for Ex-ECG = 1.0, CI = 0.6-1.6, p = 0.92). Male sex and prior IHD were significant predictors regardless of follow up duration. Age, diabetes and decreased exercise capacity were risk factors for long-term follow up. The prognostic value of MPS and Ex-ECG results are strong for short-term follow up but diminish over time and do not contribute significantly in multivariable models after 5 years. Long-term prognosis is primarily governed by underlying risk factors and exercise capacity.

Breast Cancer Index in Premenopausal Women With Early-Stage Hormone Receptor–Positive Breast Cancer

Adjuvant ovarian function suppression (OFS) with oral endocrine therapy improves outcomes for premenopausal patients with hormone receptor-positive (HR+) breast cancer but adds adverse effects. A genomic biomarker for selecting patients most likely to benefit from OFS-based treatment is lacking. To assess the predictive and prognostic performance of the Breast Cancer Index (BCI) for OFS benefit in premenopausal women with HR+ breast cancer. This prospective-retrospective translational study used all available tumor tissue samples from female patients from the Suppression of Ovarian Function Trial (SOFT). These individuals were randomized to receive 5 years of adjuvant tamoxifen alone, tamoxifen plus OFS, or exemestane plus OFS. BCI testing was performed blinded to clinical data and outcome. The a priori hypothesis was that BCI HOXB13/IL17BR ratio (BCI[H/I])-high tumors would benefit more from OFS and high BCI portended poorer prognosis in this population. Settings spanned multiple centers internationally. Participants included premenopausal female patients with HR+ early breast cancer with specimens in the International Breast Cancer Study Group tumor repository available for RNA extraction. Data were collected from December 2003 to April 2021 and were analyzed from May 2022 to October 2022. Primary end points were breast cancer-free interval (BCFI) for the predictive analysis and distant recurrence-free interval (DRFI) for the prognostic analyses. Tumor specimens were available for 1718 of the 3047 female patients in the SOFT intention-to-treat population. The 1687 patients (98.2%) who had specimens that yielded sufficient RNA for BCI testing represented the parent trial population. The median (IQR) follow-up time was 12 (10.5-13.4) years, and 512 patients (30.3%) were younger than 40 years. Tumors were BCI(H/I)-low for 972 patients (57.6%) and BCI(H/I)-high for 715 patients (42.4%). Patients with tumors classified as BCI(H/I)-low exhibited a 12-year absolute benefit in BCFI of 11.6% from exemestane plus OFS (hazard ratio [HR], 0.48 [95% CI, 0.33-0.71]) and an absolute benefit of 7.3% from tamoxifen plus OFS (HR, 0.69 [95% CI, 0.48-0.97]) relative to tamoxifen alone. In contrast, patients with BCI(H/I)-high tumors did not benefit from either exemestane plus OFS (absolute benefit, -0.4%; HR, 1.03 [95% CI, 0.70-1.53]; P for interaction = .006) or tamoxifen plus OFS (absolute benefit, -1.2%; HR, 1.05 [95% CI, 0.72-1.54]; P for interaction = .11) compared with tamoxifen alone. BCI continuous index was significantly prognostic in the N0 subgroup for DRFI (n = 1110; P = .004), with 12-year DRFI of 95.9%, 90.8%, and 86.3% in BCI low-risk, intermediate-risk, and high-risk N0 cancers, respectively. In this prospective-retrospective translational study of patients enrolled in SOFT, BCI was confirmed as prognostic in premenopausal women with HR+ breast cancer. The benefit from OFS-containing adjuvant endocrine therapy was greater for patients with BCI(H/I)-low tumors than BCI(H/I)-high tumors. BCI(H/I)-low status may identify premenopausal patients who are likely to benefit from this more intensive endocrine therapy.

Validation of an AI-based solution for breast cancer risk stratification using routine digital histopathology images

BackgroundStratipath Breast is a CE-IVD marked artificial intelligence-based solution for prognostic risk stratification of breast cancer patients into high- and low-risk groups, using haematoxylin and eosin (H&E)-stained histopathology whole slide images (WSIs). In this validation study, we assessed the prognostic performance of Stratipath Breast in two independent breast cancer cohorts.MethodsThis retrospective multi-site validation study included 2719 patients with primary breast cancer from two Swedish hospitals. The Stratipath Breast tool was applied to stratify patients based on digitised WSIs of the diagnostic H&E-stained tissue sections from surgically resected tumours. The prognostic performance was evaluated using time-to-event analysis by multivariable Cox Proportional Hazards analysis with progression-free survival (PFS) as the primary endpoint.ResultsIn the clinically relevant oestrogen receptor (ER)-positive/human epidermal growth factor receptor 2 (HER2)-negative patient subgroup, the estimated hazard ratio (HR) associated with PFS between low- and high-risk groups was 2.76 (95% CI: 1.63–4.66, p-value < 0.001) after adjusting for established risk factors. In the ER+/HER2- Nottingham histological grade (NHG) 2 subgroup, the HR was 2.20 (95% CI: 1.22–3.98, p-value = 0.009) between low- and high-risk groups.ConclusionThe results indicate an independent prognostic value of Stratipath Breast among all breast cancer patients, as well as in the clinically relevant ER+/HER2- subgroup and the NHG2/ER+/HER2- subgroup. Improved risk stratification of intermediate-risk ER+/HER2- breast cancers provides information relevant for treatment decisions of adjuvant chemotherapy and has the potential to reduce both under- and overtreatment. Image-based risk stratification provides the added benefit of short lead times and substantially lower cost compared to molecular diagnostics and therefore has the potential to reach broader patient groups.