Validation Cohort Research Articles

Updated risk prediction model for pancreaticoduodenectomy using data from the National Clinical Database in Japan

AbstractAimRisk prediction models for mortality, severe postoperative complications, and postoperative pancreatic fistula in patients undergoing pancreaticoduodenectomy were established using data from the National Clinical Database more than a decade ago, and the surgical outcomes of pancreaticoduodenectomy have improved over the years. The aim of this study is to update the risk prediction model for pancreaticoduodenectomy using National Clinical Database data.MethodsBetween 2019 and 2021, the data of 35 365 patients who underwent pancreaticoduodenectomy and who were registered in the National Clinical Database were retrospectively analyzed. According to the registration year, the patients were divided into two cohorts: the development cohort (2019–2020; n = 23 654) and the validation cohort (2021; n = 11 711). Logistic regression analyses were performed to create risk models for surgical mortality, severe postoperative complications, and grade C postoperative pancreatic fistula.ResultsOverall, the rates of surgical mortality, severe postoperative complications, and grade C postoperative pancreatic fistula were 1.8%, 2.2%, and 1.3%, respectively. Logistic regression analyses revealed 28, 28, and 14 risk factors for surgical mortality, severe postoperative complications, and grade C postoperative pancreatic fistula, respectively. The area under the receiver operating characteristic curve of the risk model in the development cohort was 0.759 for surgical mortality, 0.712 for severe complications, and 0.699 for postoperative pancreatic fistula, comparable to the validation cohort. The calibration plots were favorable in both cohorts.ConclusionThe updated risk model for pancreaticoduodenectomy will be useful to predict surgical mortality, severe postoperative complications, and grade C postoperative pancreatic fistula.

Screening high-risk individuals for primary gastric adenocarcinoma: evaluating progression-free survival probability score in the presence and absence of Rictor expression after gastrectomy

ObjectiveDeveloping nomogram-based risk stratification model to determine 3-year and 5-year progression-free survival (PFS) and to identify high-risk patients with gastric adenocarcinoma based on different Rictor statuses.Methods1366 individuals who underwent radical gastric surgery to treat gastric adenocarcinoma at Shanxi Cancer Hospital from May 2002 to December 2020 were analyzed. Cox regression analysis was employed to create the nomograms. The nomograms’ performance was assessed using C-index, time receiver operating characteristic (t-ROC) curves, calibration curves, and decision curve analysis (DCA) curves in training and validation cohorts. Subsequently, patients were categorized into high-risk and low-risk groups based on the nomogram’s risk scores.ResultsThe Rictor (-) nomogram for predicting PFS included variables such as age, number of positive lymph nodes, vascular invasion, maximum diameter of the tumor, omentum metastasis, and expression of MSH2. In the internal validation, the C-index of the Rictor (-) nomogram was 0.760 (95%CI: 0.720-0.799), which was superior to the C-index of the American Joint Committee on Cancer (AJCC) 8th edition TNM staging (0.683, 95%CI: 0.646-0.721). Similarly, the Rictor (+) nomogram for predicting PFS included variables such as gender, age, pT stage, number of positive lymph nodes, neural invasion, maximum diameter of the tumor, omentum metastasis, Clavien-Dindo classification for complications, and CGA expression. The C-index of the Rictor (+) nomogram was 0.795 (95%CI: 0.764-0.825), which outperformed the C-index of the AJCC 8th edition TNM staging (0.693, 95%CI: 0.662-0.723). The calibration curves, t-ROC curves, and decision curve analysis for both nomogram models demonstrated their excellent prediction ability.ConclusionThis study presents the first risk stratification for Rictor status in gastric adenocarcinoma. Our model identifies low-risk patients who may not require additional postoperative treatment, while high-risk patients should consider targeted therapies that specifically target Rictor-positive indicators.

SURG-21. DEVELOPMENT AND VALIDATION OF A CLINICAL RISK SCORE FOR POSTOPERATIVE OUTCOME IN NEWLY DIAGNOSED GLIOBLASTOMA: A REPORT OF THE RANORESECT GROUP

Abstract BACKGROUND Following resection or biopsy, patients with newly diagnosed glioblastoma frequently enter clinical trials. Nuanced risk assessment is warranted to reduce imbalances between study arms. Here, we aim to make use of our RANO classification for extent of resection (I) to analyze the interactive effects of residual tumor with clinical and molecular factors on outcome and (II) to define a postoperative risk assessment tool. METHODS The RANO resect group retrospectively compiled an international, seven-center training cohort of patients with newly diagnosed glioblastoma. The combined associations of residual tumor with molecular or clinical factors and survival were analyzed, and recursive partitioning analysis was performed for risk modeling. The resulting model was prognostically verified in a separate external validation cohort. RESULTS Our training cohort compromised 1003 patients with newly diagnosed IDH-wildtype glioblastoma, including 744 patients who underwent adjuvant radiochemotherapy (TMZ/RT→TMZ). Residual tumor, MGMT promotor methylation status, age, and KPS were prognostic of overall survival and forwarded into regression tree analysis. By combining eleven terminal nodes and individually weighting the prognostic factors, an additive scale (range, 0-9 points) integrating these four variables distinguished patients with low (0-2 points), intermediate (3-5 points), and high risk (6-9 points) for poor postoperative outcome (median overall survival: 24 vs. 16 vs. 6 months; p<0.01). The presence of the three risk groups was confirmed in the subgroups of patients treated with or without RT/TMZ→TMZ. The prognostic value of the risk score was verified in a external single-center validation cohort of newly diagnosed glioblastoma (n = 258, p<0.01). Compared to previously postulated models, goodness-of-fit measurements were superior for our score. CONCLUSIONS The novel RANO risk score serves as an easy-to-use, yet highly prognostic tool for postoperative patient stratification prior to further therapy. The score may serve to guide patient management and reduce imbalances between study arms in prospective trials.

Prospective Validation and Usability Evaluation of a Mobile Diagnostic App for Obstructive Sleep Apnea

Background/Objectives: Obstructive sleep apnea (OSA) classification relies on polysomnography (PSG) results. Current guidelines recommend the development of clinical prediction algorithms in screening prior to PSG. A recent intuitive and user-friendly tool (OSABayes), based on a Bayesian network model using six clinical variables, has been proposed to quantify the probability of OSA. Our aims are (1) to validate OSABayes prospectively, (2) to build a smartphone app based on the proposed model, and (3) to evaluate app usability. Methods: We prospectively included adult patients suspected of OSA, without suspicion of other sleep disorders, who underwent level I or III diagnostic PSG. Apnea–hypopnea index (AHI) and OSABayes probabilities were obtained and compared using the area under the ROC curve (AUC [95%CI]) for OSA diagnosis (AHI ≥ 5/h) and higher severity levels (AHI ≥ 15/h) prediction. We built the OSABayes app on ‘App Inventor 2’, and the usability was assessed with a cognitive walkthrough method and a general evaluation. Results: 216 subjects were included in the validation cohort, performing PSG levels I (34%) and III (66%). OSABayes presented an AUC of 83.6% [77.3–90.0%] for OSA diagnosis and 76.3% [69.9–82.7%] for moderate/severe OSA prediction, showing good response for both types of PSG. The OSABayes smartphone application allows one to calculate the probability of having OSA and consult information about OSA and the tool. In the usability evaluation, 96% of the proposed tasks were carried out. Conclusions: These results show the good discrimination power of OSABayes and validate its applicability in identifying patients with a high pre-test probability of OSA. The tool is available as an online form and as a smartphone app, allowing a quick and accessible calculation of OSA probability.

Development and validation of a nomogram for predicting the risk of developing gastric cancer based on a questionnaire: a cross–sectional study

BackgroundDetection of gastric cancer (GC) at early stages is an effective strategy for decreasing mortality. This study aimed to construct a prediction nomogram based on a questionnaire to assess the risk of developing GC.MethodsOur study comprised a total of 4379 participants (2326 participants from outpatient at Fengqing People’s Hospital were considered for model development and internal validation, and 2053 participants from outpatients at the endoscopy center at the Seventh Affiliated Hospital of Sun Yat-Sen University were considered for independent external validation) and gastric mucosa status was determined by endoscopy and biopsies. The eligible participants in development cohort from Fengqing people’s Hospital were randomly separated into a training cohort (n=1629, 70.0%) and an internal validation cohort (n=697, 30.0%). The relevant features were selected by a least absolute shrinkage and selection operator (LASSO), and the ensuing features were evaluated through multivariable logistic regression analysis. Subsequently, the variables were selected to construct a prediction nomogram. The discriminative ability and predictive accuracy of the nomogram were evaluated by the C-index and calibration plot, respectively. Decision curve analysis (DCA) curves were used for the assessment of clinical benefit of the model. This model was developed to estimate the risk of developing neoplastic lesions according to the “transparent reporting of a multivariable prediction model for individual prognosis or diagnosis” (TRIPOD) statement.ResultsSix variables, including age, sex, alcohol consumption, cigarette smoking, education level, and Hp infection status, were independent risk factors for the development of neoplastic lesions. Thus, these variables were incorporated into the final nomogram. The AUC of the nomogram were 0.701, 0.657 and 0.699 in the training, internal validation, and external validation cohorts, respectively. The calibration curve showed that the nomogram was in good agreement with the observed outcomes. Compared to treatment of all patients or none, our nomogram showed a notably higher clinical benefit.ConclusionThis nomogram proved to be a convenient, cost-effective tool to effectively predict an individual’s risk of developing neoplastic lesions, and it can act as a prescreening tool before gastroscopy.

Prognostic value of FDX1, the cuprotosis key gene, and its prediction models across imaging modalities and histology.

Cuprotosis has been identified as a novel way of cell death. The key regulator ferredoxin 1 (FDX1) was explored via pan-cancer analysis, and its prediction models were proposed across seven malignancies and two imaging modalities. The prognostic value of FDX1 was explored via 1654 cases of 33 types of cancer in the Cancer Genome Atlas database. The MRI cohort of hepatocellular carcinoma in the First Affiliated Hospital of Fujian Medical University, and CT and MRI images from the Cancer Imaging Archive, REMBRANDT and Duke databases were exploited to formulate radiomic models to predict FDX1 expression. After segmentation of volumes of interest and feature extraction, the recursive feature elimination algorithm was used to screen features, logistic regression was used to model features, immunohistochemistry staining with FDX1 antibody was performed to test the radiomic model. FDX1 was found to be prognostic in various types of cancer. The area under the receiver operating characteristic curve of radiomic models to predict FDX1 expression reached 0.825 (95% CI = 0.739-0.911). Cross-tissue compatibility was confirmed in pan-cancer validation and test cohorts. Mechanistically, the radiomic score was significantly correlated with various immunosuppressive genes and gene mutations. The radiomic score was also found to be an independent prognostic factor, making it a potentially actionable biomarker in the clinical setting. The expression of FDX1 could be non-invasively predicted via radiomics. The radiomic patterns with biological and clinical relevance across histology and modalities could have a broad impact on a larger population of patients.

Explainable machine learning model for predicting the risk of significant liver fibrosis in patients with diabetic retinopathy.

Diabetic retinopathy (DR), a prevalent complication in patients with type 2 diabetes, has attracted increasing attention. Recent studies have explored a plausible association between retinopathy and significant liver fibrosis. The aim of this investigation was to develop a sophisticated machine learning (ML) model, leveraging comprehensive clinical datasets, to forecast the likelihood of significant liver fibrosis in patients with retinopathy and to interpret the ML model by applying the SHapley Additive exPlanations (SHAP) method. This inquiry was based on data from the National Health and Nutrition Examination Survey 2005-2008 cohort. Utilizing the Fibrosis-4 index (FIB-4), liver fibrosis was stratified across a spectrum of grades (F0-F4). The severity of retinopathy was determined using retinal imaging and segmented into four discrete gradations. A ten-fold cross-validation approach was used to gauge the propensity towards liver fibrosis. Eight ML methodologies were used: Extreme Gradient Boosting, Random Forest, multilayer perceptron, Support Vector Machines, Logistic Regression (LR), Plain Bayes, Decision Tree, and k-nearest neighbors. The efficacy of these models was gauged using metrics, such as the area under the curve (AUC). The SHAP method was deployed to unravel the intricacies of feature importance and explicate the inner workings of the ML model. The analysis included 5,364 participants, of whom 2,116 (39.45%) exhibited notable liver fibrosis. Following random allocation, 3,754 individuals were assigned to the training set and 1,610 were allocated to the validation cohort. Nine variables were curated for integration into the ML model. Among the eight ML models scrutinized, the LR model attained zenith in both AUC (0.867, 95% CI: 0.855-0.878) and F1 score (0.749, 95% CI: 0.732-0.767). In internal validation, this model sustained its superiority, with an AUC of 0.850 and an F1 score of 0.736, surpassing all other ML models. The SHAP methodology unveils the foremost factors through importance ranking. Sophisticated ML models were crafted using clinical data to discern the propensity for significant liver fibrosis in patients with retinopathy and to intervene early. Improved early detection of liver fibrosis risk in retinopathy patients enhances clinical intervention outcomes.

Explainable machine learning for early prediction of sepsis in traumatic brain injury: A discovery and validation study.

People with traumatic brain injury (TBI) are at high risk for infection and sepsis. The aim of the study was to develop and validate an explainable machine learning(ML) model based on clinical features for early prediction of the risk of sepsis in TBI patients. We enrolled all patients with TBI in the Medical Information Mart for Intensive Care IV database from 2008 to 2019. All patients were randomly divided into a training set (70%) and a test set (30%). The univariate and multivariate regression analyses were used for feature selection. Six ML methods were applied to develop the model. The predictive performance of different models were determined based on the area under the curve (AUC) and calibration curves in the test cohort. In addition, we selected the eICU Collaborative Research Database version 1.2 as the external validation dataset. Finally, we used the Shapley additive interpretation to account for the effects of features attributed to the model. Of the 1555 patients enrolled in the final cohort, 834 (53.6%) patients developed sepsis after TBI. Six variables were associated with concomitant sepsis and were used to develop ML models. Of the 6 models constructed, the Extreme Gradient Boosting (XGB) model achieved the best performance with an AUC of 0.807 and an accuracy of 74.5% in the internal validation cohort, and an AUC of 0.762 for the external validation. Feature importance analysis revealed that use mechanical ventilation, SAPSII score, use intravenous pressors, blood transfusion on admission, history of diabetes, and presence of post-stroke sequelae were the top six most influential features of the XGB model. As shown in the study, the ML model could be used to predict the occurrence of sepsis in patients with TBI in the intensive care unit.

CTNI-13. 18-FET-PET ENHANCED DIAGNOSIS OF PSEUDOPROGRESSION IN GLIOBLASTOMA IS AN INDEPENDENT PROGNOSTIC FACTOR FOR OVERALL SURVIVAL

Abstract INTRODUCTION Detecting pseudoprogression (PSP) in glioblastoma during first-line treatment remains challenging and has significant clinical implications for patient management. The detection of PSP has been shown feasible using FET-PET. The prognostic impact of FET-PET diagnosed PSP on overall survival is still uncertain, as retrospective studies provide limited insights due to the substantial differences in the distribution of prognostic factors between tumor progression (TP) and PSP groups. This study aims to address these uncertainties by leveraging advanced analysis techniques. METHODS Patients with newly diagnosed glioblastoma (WHO 2021) who underwent chemoradiation and contrast-enhanced MRI suspected for TP and subsequent FET-PET scan at Essen University Hospital were retrospectively evaluated. The modified Response Assessment in Neuro-Oncology (RANO) criteria were used to diagnose PSP. FET-PET analysis included mean (TBRmean) and maximum tumor-to-brain ratios (TBRmax). The two groups (TP/PSP) were balanced using a propensity score analysis including age at diagnosis, extent of resection, Karnofsky-Performance-Status Scale (KPS) and MGMT methylation status. Overall survival was predicted using Kaplan-Meier and multivariate analysis as well as conventional FET-PET parameters, PET/MRI volumetry, and FET/MRI-derived radiomics. An independent cohort was used for validation. RESULTS Out of the 165 patients analysed in this study, 38 were diagnosed with PSP. Following balancing analysis, our study cohort consisted of 40 patients (20 TP/20 PSP). Patients with PSP had a significantly longer mean overall survival (mOS) compared to those with TP (mOS TP 20.24 months; mOS PSP 33.25 months; p=0.0023). The multivariate Cox-Regression analysis confirmed PSP as a significant prognostic marker for overall survival (p=0.0153). Confirmation of these results on an independent validation cohort is pending and will be presented at the meeting. CONCLUSION This study shows that PSP is an independent prognostic factor for overall survival in newly diagnosed glioblastoma patients.

PATH-06. TRANSLATION INTO CLINICAL PRACTICE OF THE G1-G7 MOLECULAR SUBGROUP CLASSIFICATION OF GLIOBLASTOMA

Abstract Glioblastoma is the most frequent and malignant primary brain neoplasm. Glioblastoma growth is promoted by alterations in mediators from five major pathways: MAPK and PI3K canonical growth pathways, telomere elongation/TERT, cell cycle G1-phase and the p53 pathway. In a recent breakthrough study on a prospective Discovery cohort, I proposed the first all-inclusive molecular classification of glioblastoma into seven subgroups, G1-G7, based on MAPK pathway activation. New data from a subsequent, prospective Validation cohort validated the G1-G7 molecular subgroup classification, showing significant demographic and molecular differences between subgroups. Beside the initial histologic diagnosis, the G1-G7 classification requires next generation sequencing (NGS) of formalin-fixed paraffin-embeded (FFPE) samples and genomic-transcriptomic correlations between DNA mutations, copy number variations (CNVs) and RNA expression. Thus, in the Combined cohort containing tumors from over 220 glioblastoma patients, aproximately 75% of cases fell into three major subgroups, G1/EGFR, G3/NF1 and G7/Other, whereas the rest of 25% cases made up the four minor subgroups, G2/FGFR3, G4/RAF, G5/PDGFRA and G6/Multi-RTK. The name of the mutated most upstream, non-redundant, MAPK pathway effector corresponds to each subgroup, except for the G7/Other subgroup in which MAPK activation is minimal and PI3K pathway activation is dominant. The effectors of the other four pathways may be redundant; however, genomic-transcriptomic correlations showed subgroup specificity, with opposing trends for the G1/EGFR and G7/Other molecular subgroups. This first, all-inclusive, molecular subgroup classification of glioblastoma in a large prospective patient cohort allowed integrated pathway analysis, risk stratification and demographic comparison between Caucasian/White and African-American/Black patients. Most importantly, its implementation into clinical practice opened the avenue to controlled therapeutic approaches.

The potential of ESCO2 as a prognostic and immunotherapeutic marker of pan-cancer and its role in anti-PD1 treatment of bladder cancer.

Background Establishment of sister chromatid cohesion N-acetyltransferase 2 (ESCO2), a member of the EFO2 family, is implicated in the pathogenesis and progression of various cancers. However, there has been limited comprehensive pan-cancer analysis conducted on ESCO2 thus far. Methods Publicly available databases, such as the UCSC Xena database, were utilized to examine differential expression patterns across various cancer types. In addition, variations in expression levels were investigated across distinct clinical stages. Univariate Cox regression and Kaplan-Meier survival analyses were conducted to evaluate the impact on overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) at a pan-cancer level. The correlation between ESCO2 expression and immune cell infiltration was examined to gain insight into the tumor microenvironment in different cancers. The results of the bioinformatic analysis were validated using immunotherapy clinical trials and pathological specimens. CCK-8 and transwell assay experiments were performed to investigate the biological function of ESCO2. Results ESCO2 expression was found to be up-regulated in most cancers, with a correlation to TNM stages. Prognostic analysis indicated that overexpression of ESCO2 was associated with poor prognosis in various cancers. Furthermore, the correlation between ESCO2 expression and immune cell infiltration suggested its potential as a predictor for immunotherapy efficacy. Notably, ESCO2 expression showed positive associations with immunoinhibitor, immnostimulator, MHC molecule, chemokine receptor, tumor mutation burden (TMB) and microsatellite instability (MSI) levels in bladder cancer (BLCA). The validation cohort for immunotherapy corroborated these findings and substantiated that ESCO2 could function as an autonomous prognostic biomarker and a promising target for cancer treatment via immunotherapy. In addition, in vitro experiments confirmed the role of ESCO2 in influcing the proliferation, invasion and migration of BLCA cells. Conclusion ESCO2 participates in regulating the immune infiltration and affecting the prognosis of patients in many cancers, especially in BLCA. ESCO2 may serve as a prognostic and immunotherapy biomarker in future treatment of human cancer.

The Risk Assessment Before Dose Tapering Among Methadone Maintenance Treatment Participants: Derivation and Validation of a Nomogram

ABSTRACT Many methadone maintenance treatment (MMT) participants experienced a tapering phase. The benefit of tapering is based on a balance between meeting the desire to reduce methadone dose and reduction in relapse. We aimed to develop and validate a nomogram to assess relapse risk after dose tapering. We developed and internally validated a nomogram for risk assessment before dose tapering in 432 participants with dose tapering in the non-Guangzhou region of Guangdong, China, and externally validated it in 117 participants with dose tapering in Guangzhou. Cox regression analysis showed that the taper start week (HR = 0.14, [0.08–0.22]) was an independent risk predictor of the relapse risk after tapering. The C-index of the nomogram was 0.76 (95%CI: 0.73–0.79) in the training cohort, 0.76 (95%CI: 0.72–0.80) in the testing cohort, and 0.84 (95%CI: 0.80–0.88) in the validation cohort. Decision curve analysis showed that the nomogram had better discriminative ability than other predictors. The nomogram was developed to assess the risk of relapse for MMT participants who volunteer a tapering phase and may help participants better make decisions about whether and how to reduce the dose to minimize the harm of relapse.

Clinical application of machine-based deep learning in patients with radiologically presumed adult-type diffuse glioma grades 2 or 3

Abstract BACKGROUND Radiologically presumed diffuse lower-grade glioma (dLGG) are typically non or minimal enhancing tumors, with hyperintensity in T2w-images. The aim of this study was to test the clinical usefulness of deep learning (DL) in IDH mutation prediction in patients with radiologically presumed dLGG. METHODS 314 patients were retrospectively recruited from six neurosurgical departments in Sweden, Norway, France, Austria, and the United States. Collected data included patients’ age, sex, tumor molecular characteristics (IDH, and 1p19q), and routine preoperative radiological images. A clinical model was built using multivariable logistic regression with the variables age and tumor location. DL models were built using MRI data only, and four DL architectures used in glioma research. In the final validation test, the clinical model and the best DL model were scored on an external validation cohort with 155 patients from the Erasmus Glioma Dataset. RESULTS The mean age in the recruited and external cohorts was 45.0 (SD 14.3) and 44.3 years (SD 14.6). The cohorts were rather similar, except for sex distribution (53.5% vs 64.5% males, p-value 0.03) and IDH status (30.9% vs 12.9% IDH wild-type, p-value <0.01). Overall, the area under the curve for the prediction of IDH mutations in the external validation cohort was 0.86, 0.82, and 0.87 for the clinical model, the DL model, and the model combining both models’ probabilities. CONCLUSIONS In their current state, when these complex models were applied to our clinical scenario, they did not seem to provide a net gain compared to our baseline clinical model.

Building a machine learning-based risk prediction model for second-trimester miscarriage

BackgroundSecond-trimester miscarriage is a common adverse pregnancy outcome that imposes substantial economic and psychological pressures on both the physical and mental well-being of patients and their families. Currently, there is a scarcity of research on predictive models for the risk of second-trimester miscarriage.MethodsClinical data were retrospectively collected from patients who were in the second trimester of pregnancy (between 14+0 and 27+6 weeks gestation), whose main diagnosis was “threatened abortion” and who were hospitalized at the Women and Children’s Hospital of Ningbo University from January 2020 to October 2023. Following preliminary data processing, the patient cohort was randomly stratified into a training cohort and a validation cohort at proportions of 70% and 30%, respectively. The Boruta algorithm and multifactor analysis were used to refine feature factors and determine the optimal features linked to second-trimester miscarriages. The imbalanced dataset from the training cohort was rectified by applying the SMOTE oversampling approach. Seven machine-learning models were built and subjected to a comprehensive analysis to validate and evaluate their predictive capabilities. Through this rigorous assessment, the optimal model was selected. Shapley additive explanations (SHAP) were generated to provide insights into the model’s predictions, and a visual representation of the predictive model was built.ResultsA total of 2006 patients were included in the study; 395 (19.69%) of them had second-trimester miscarriages. XGBoost was shown to be the optimal model after a comparison of seven different models utilizing metrics such as accuracy, precision, recall, the F1 score, precision-recall average precision, the receiver operating characteristic-area under the curve, decision curve analysis, and the calibration curve. The most significant feature was cervical length, and the top ten features of second-trimester miscarriage were found using the SHAP technique based on relevance rankings.ConclusionThe risk of a second-trimester miscarriage can be accurately predicted by the visual risk prediction model, which is based on the machine learning mentioned above.

Deep learning radiomics based on contrast enhanced MRI for preoperatively predicting early recurrence in hepatocellular carcinoma after curative resection

PurposeTo explore the role of deep learning (DL) and radiomics-based integrated approach based on contrast enhanced magnetic resonance imaging (CEMRI) for predicting early recurrence (ER) in hepatocellular carcinoma (HCC) patients after curative resection.MethodsTotal 165 HCC patients (ER, n = 96 vs. non-early recurrence (NER), n = 69) were retrospectively collected and divided into a training cohort (n = 132) and a validation cohort (n = 33). From pretreatment CEMR images, a total of 3111 radiomics features were extracted, and radiomics models were constructed using five machine learning classifiers (logistic regression, support vector machine, k-nearest neighbor, extreme gradient Boosting, and multilayer perceptron). DL models were established via three variations of ResNet architecture. The clinical-radiological (CR), radiomics combined with clinical-radiological (RCR), and deep learning combined with RCR (DLRCR) models were constructed. Model discrimination, calibration, and clinical utilities were evaluated by receiver operating characteristic curve, calibration curve, and decision curve analysis, respectively. The best-performing model was compared with the widely used staging systems and preoperative prognostic indexes.ResultsThe RCR model (area under the curve (AUC): 0.841 and 0.811) and the optimal radiomics model (AUC: 0.839 and 0.804) achieved better performance than the CR model (AUC: 0.662 and 0.752) in the training and validation cohorts, respectively. The optimal DL model (AUC: 0.870 and 0.826) outperformed the radiomics model in the both cohorts. The DL, radiomics, and CR predictors (aspartate aminotransferase (AST) and tumor diameter) were combined to construct the DLRCR model. The DLRCR model presented the best performance over any model, yielding an AUC, an accuracy, a sensitivity, a specificity of 0.917, 0.886, 0.889, and 0.882 in the training cohort and of 0.844, 0.818, 0.800, and 0.846 in the validation cohort, respectively. The DLRCR model achieved better clinical utility compared to the clinical staging systems and prognostic indexes.ConclusionBoth radiomics and DL models derived from CEMRI can predict HCC recurrence, and DL and radiomics-based integrated approach can provide a more effective tool for the precise prediction of ER for HCC patients undergoing resection.

Unsupervised Classification of the Host Response Identifies Dominant Pathobiological Signatures of Sepsis in Sub-Saharan Africa.

The global burden of sepsis is concentrated in sub-Saharan Africa, where inciting pathogens are diverse and HIV co-infection is a major driver of poor outcomes. Biological heterogeneity inherent to sepsis in this setting is poorly defined. To identify dominant pathobiological signatures of sepsis in sub-Saharan Africa and their relationship to clinical phenotypes, patient outcomes, and biological classifications of sepsis identified in high-income-countries (HICs). We analyzed two prospective cohorts of adults hospitalized with sepsis (severe infection with qSOFA score≥1) at disparate settings in Uganda (discovery cohort [Entebbe,urban], N=242; validation cohort [Tororo,rural], N=253). To identify pathobiological signatures in the discovery cohort, we applied unsupervised clustering to 173 soluble proteins reflecting key domains of the host response to severe infection. A random forest-derived classifier was used to predict signature assignment in the validation cohort. Two signatures (Uganda Sepsis Signature [USS]-1 and USS-2) were identified in the discovery cohort, distinguished by expression of proteins involved in myeloid cell and inflammasome activation, T cell co-stimulation and exhaustion, and endothelial barrier dysfunction. A five-protein classifier (AUROC 0.97) reproduced two signatures in the validation cohort with similar biological profiles. In both cohorts, USS-2 mapped to a more severe clinical phenotype associated with HIV and related immunosuppression, severe tuberculosis, and increased risk of 30-day mortality. Substantial biological overlap was observed between USS-2 and hyperinflammatory and reactive sepsis phenotypes identified in HICs. We identified prognostically-enriched pathobiological signatures among sepsis patients with diverse infections and high HIV prevalence in Uganda. Globally inclusive investigations are needed to define generalizable and context-specific mechanisms of sepsis pathobiology, with the goal of improving access to precision medicine treatment strategies.

Combined fibrinogen concentration and neutrophil-to-lymphocyte ratio, an integrative model of the inflammatory response and coagulation cascades, for predicting prognosis in patients with upper tract urothelial carcinoma.

Inflammation and coagulation cascades are closely correlated with cancer occurrence and progression. This study investigated the prognostic value of the combination of plasma fibrinogen level and neutrophil-to-lymphocyte ratio (F-NLR) in patients with upper tract urothelial carcinoma (UTUC). The predictive ability of the F-NLR for overall survival (OS), cancer-specific survival (CSS), and progression-free survival (PFS) was initially established and then further validated in patients who underwent radical nephroureterectomy for UTUC. As a result, patients were divided into three groups following the establishment of cut-off values for the NLR (≥2.53 vs. <2.53) and fibrinogen (≥4.55 vs. <4.55) through receiver operating characteristic (ROC) curve analysis: F-NLR score 0 (low fibrinogen and low NLR), 2 (high fibrinogen and high NLR), or 1 (remaining patients). The F-NLR score was then identified as an independent risk factor for OS, CSS, and PFS (all P-value <0.05) by multivariate regression analysis in both the training and validation cohorts. In addition, F-NLR-based nomograms for OS, CSS, and PFS were developed and evaluated using the concordance index (C-index) and calibration curves. The integration of the F-NLR into existing nomograms improved predictive accuracy compared to the use of nomograms without the F-NLR score. This suggests that the addition of F-NLR is beneficial for enhancing the accuracy of prognosis prediction in patients with UTUC. The F-NLR score may serve as a powerful predictor for patients with UTUC.

Soluble and EV-bound CD27 act as antagonistic biomarkers in patients with solid tumors undergoing immunotherapy

BackgroundThe major breakthrough in cancer therapy with immune checkpoint inhibitors (ICIs) has highlighted the important role of immune checkpoints in antitumoral immunity. However, most patients do not achieve durable responses, making biomarker research in this setting essential. CD27 is a well known costimulatory molecule, however the impact of its soluble form in ICI is poorly investigated. Therefore, we aimed at testing circulating concentrations of soluble CD27 (sCD27) and CD27 bound to extracellular vesicles (EVs) as potential biomarkers to predict response and overall survival (OS) in patients undergoing ICI.MethodsSerum and plasma levels of sCD27 were assessed by immunoassay in three patient cohorts (n = 187) with advanced solid malignancies including longitudinal samples (n = 126): a training (n = 84, 210 specimens, Aachen ICI) and validation cohort (n = 70, 70 specimens, Hamburg ICI), both treated with ICI therapy, and a second independent validation cohort (n = 33, 33 specimens, Hamburg non-ICI) undergoing systemic therapy without any ICI. In a subset (n = 36, 36 baseline and 108 longitudinal specimens), EV-bound CD27 from serum was measured, while EV characterization studies were conducted on a fourth cohort (n = 45).ResultsIn the Aachen and Hamburg ICI cohorts, patients with lower circulating sCD27 levels before and during ICI therapy had a significantly longer progression-free survival (PFS) and OS compared to patients with higher levels, a finding that was confirmed by multivariate analysis (MVA) (Aachen ICI: pPFS = 0.012, pOS = 0.001; Hamburg ICI: pPFS = 0.040, pOS = 0.004) and after randomly splitting both cohorts into training and validation. This phenomenon was not observed in the Hamburg non-ICI cohort, providing a rationale for the predictive biomarker role of sCD27 in immune checkpoint blockade. Remarkably, EV-bound CD27 baseline levels and dynamics during ICI therapy also emerged as potent predictive biomarkers, acting however antagonistically to soluble sCD27, i.e. higher levels were associated with PFS and OS benefit. Combining both molecules (“multi-CD27” score) enhanced the predictive ability (HRPFS: 17.21 with p < 0.001, HROS: 6.47 with p = 0.011).ConclusionSoluble and EV-bound CD27 appear to have opposing immunomodulatory functions and may represent easily measurable, non-invasive prognostic markers to predict response and survival in patients undergoing ICI therapy.

Modified FIB-4 Index in Type 2 Diabetes Mellitus with Steatosis: A Non-Linear Predictive Model for Advanced Hepatic Fibrosis

Background: The Fibrosis-4 (FIB-4) index is widely recommended as a first-tier method for screening advanced hepatic fibrosis; however, its diagnostic performance is known to be suboptimal in patients with Type 2 diabetes mellitus (T2DM). We aim to propose a modified FIB-4, using the parameters of the existing FIB-4, tailored specifically for diabetic patients with metabolic dysfunction-associated steatotic liver disease (MASLD). Methods: A total of 1503 patients who underwent liver biopsy were divided into T2DM (n = 517) and non-T2DM (n = 986) groups. The model was developed using multiple regression analysis in the derivation cohort and validated in the validation cohort. Diagnostic accuracy was evaluated using the area under the receiver operating characteristic (AUC) curves. Results: Among the 1503 individuals, those with T2DM were older, more likely to be male, and had a higher prevalence of advanced hepatic fibrosis (≥F3) compared to non-T2DM individuals. Independent risk factors for advanced fibrosis in T2DM included age, AST, AST/ALT ratio, albumin, triglycerides, and platelet count. The optimized FIB-4 model for T2DM with MASLD (Diabetes Fibrosis Index) demonstrated superior diagnostic accuracy (AUC 0.771) compared to the FIB-4 (AUC 0.735, p = 0.012). The model showed a higher negative predictive value than the original FIB-4 across all age groups in the diabetic group. Conclusions: The newly optimized FIB-4 model for T2DM with MASLD (Diabetes Fibrosis Index), incorporating a non-linear predictive model, improves diagnostic performance (AUC) and the negative predictive value in MASLD with T2DM.

Development and validation of a nomogram for predicting postpartum hemorrhage in women with preeclampsia: A retrospective case-control study.

This retrospective case-control study aimed to develop a nomogram for predicting postpartum hemorrhage in women with preeclampsia. This study was carried out at the Fujian Maternity and Child Health Hospital, involving 542 preeclampsia patients who underwent vaginal deliveries. The participants were split into 2 groups: a training cohort (85%, n = 460) and a validation cohort (15%, n = 82). Least absolute shrinkage and selection operator regression was applied to pinpoint relevant risk factors by selecting appropriate candidate variables. Subsequently, multivariate logistic regression analysis was conducted on the training set, leading to the creation of a nomogram as a visual risk prediction tool. The model's performance was tested and verified internally and externally by examining receiver operating characteristic curves and calibration curves. The correlation heatmap revealed collinearity among variables, necessitating the use of least absolute shrinkage and selection operator regression to select 4 candidate variables. Multivariate logistic regression analysis identified significant associations with the following outcomes: white blood cell count (odds ratio [OR]: 2.485, 95% confidence interval [CI]: 1.483-4.166), third stage of labor (OR: 1.382, 95% CI: 1.182-1.616), anemia (OR: 9.588, 95% CI: 4.022-22.854), and labor analgesia (OR: 0.187, 95% CI: 0.073-0.477). These variables were utilized to construct the nomogram. The receiver operating characteristic curves demonstrated good predictive performance (area under the curve train = 0.867, area under the curve test = 0.882), and the calibration curve yielded a C-index of 0.867. The nomogram created in this study has good sensitivity and specificity to assess risk and support clinical decision-making for postpartum hemorrhage in women with preeclampsia.