Independent Validation Cohort Research Articles

Serum Metabolic Fingerprints Characterize Systemic Lupus Erythematosus.

Metabolic fingerprints in serum characterize diverse diseases for diagnostics and biomarker discovery. The identification of systemic lupus erythematosus (SLE) by serum metabolic fingerprints (SMFs) will facilitate precision medicine in SLE in an early and designed manner. Here, a discovery cohort of 731 individuals including 357 SLE patients and 374 healthy controls (HCs), and a validation cohort of 184 individuals (SLE/HC, 91/93) are constructed. Each SMF is directly recorded by nano-assisted laser desorption/ionization mass spectrometry (LDI MS) within 1 minute using 1µL of native serum, which contains 908 mass to charge features. Sparse learning of SMFs achieves the SLE identification with sensitivity/specificity and area-under-the-curve (AUC) up to 86.0%/92.0% and 0.950 for the discovery cohort. For the independent validation cohort, it exhibits no performance loss by affording the sensitivity/specificity and AUC of 89.0%/100.0% and 0.992. Notably, a metabolic biomarker panel is screened out from the SMFs, demonstrating the unique metabolic pattern of SLE patients different from both HCs and rheumatoid arthritis patients. In conclusion, SMFs characterize SLE by revealing its unique metabolic pattern. Different regulation of small molecule metabolites contributes to the precise diagnosis of autoimmune disease and further exploration of the pathogenic mechanisms.

Microbe-derived uremic solutes enhance thrombosis potential in the host.

Alterations in gut microbial composition and function have been linked to numerous diseases. Identifying microbial pathways responsible for producing molecules that adversely impact the host is an important first step in the development of therapeutic interventions. Here, we first use large-scale clinical observations to link blood levels of defined microbial products to cardiovascular disease risks. Notably, the previously identified uremic toxins p-cresol sulfate and indoxyl sulfate were shown to predict 5-year mortality risks. After identifying the microbes and microbial enzymes involved in the generation of these uremic toxins, we used bioengineering technologies coupled with colonization of germ-free mice to show that the gut microbial genes that generate p-cresol and indole are sufficient to confer p-cresol sulfate and indoxyl sulfate formation, and a pro-thrombotic phenotype in vivo. The findings and tools developed serve as a critical step in both the study and targeting of these gut microbial pathways in vivo.

Development and Validation of Contrast-Enhanced CT-Based Deep Transfer Learning and Combined Clinical-Radiomics Model to Discriminate Thymomas and Thymic Cysts: A Multicenter Study

This study aims to evaluate the feasibility and effectiveness of deep transfer learning (DTL) and clinical-radiomics in differentiating thymoma from thymic cysts. Clinical and imaging data of 196 patients pathologically diagnosed with thymoma and thymic cysts were retrospectively collected from center 1. (training cohort: n=137; internal validation cohort: n=59). An independent external validation cohort comprised 68 thymoma and thymic cyst patients from center 2. Region of interest (ROI) delineation was performed on contrast-enhanced chest computed tomography (CT) images, and eight DTLmodelsincluding Densenet 169, Mobilenet V2, Resnet 101, Resnet 18, Resnet 34, Resnet 50, Vgg 13, Vgg 16 were constructed. Radiomics features were extracted from the ROI on the CT images of thymoma and thymic cyst patients, and feature selection was performed using intra-observer correlation coefficient (ICC), Spearman correlation analysis, and least absolute shrinkage and selection operator (LASSO) algorithm. Univariate analysis and multivariable logistic regression (LR) were used to select clinical-radiological features. Six machine learning classifiers, including LR, support vector machine (SVM), k-nearest neighbors (KNN), Light Gradient Boosting Machine (LightGBM), Adaptive Boosting (AdaBoost), and Multilayer Perceptron (MLP), were used to construct Radiomics and Clinico-radiologic models. The selected features from the Radiomics and Clinico-radiologic models were fused to build a Combined model. Receiver operating characteristic curve (ROC), calibration curve, and decision curve analysis (DCA) were used to evaluate the discrimination, calibration, and clinical utility of the models, respectively. The Delong test was used to compare the AUC between different models. K-means clustering was used to subdivide the lesions of thymomas or thymic cysts into subregions, and traditional radiomics methods were used to extract features and compare the ability of Radiomics and DTL models to reflect intratumoral heterogeneity using correlation analysis. The Densenet 169 based on DTL performed the best, with AUC of 0.933 (95% CI: 0.875-0.991) in the internal validation cohort and 0.962 (95% CI: 0.923-1.000) in the external validation cohort. The AdaBoost classifier achieved AUC of 0.965 (95% CI: 0.923-1.000) and 0.959 (95% CI: 0.919-1.000) in the internal and external validation cohorts, respectively, for the Radiomics model. The LightGBM classifier achieved AUC of 0.805 (95% CI: 0.690-0.920) and 0.839 (95% CI: 0.736-0.943) in the Clinico-radiologic model. The AUC of the Combined model in the internal and external validation cohorts was0.933 (95% CI: 0.866-1.000) and 0.945 (95% CI: 0.897-0.994), respectively. The results of the Delong test showed that the Radiomics model, DTL model, and Combined model outperformed the Clinico-radiologic model in both internal and external validation cohorts (p-values were 0.002, 0.004, and 0.033 in the internal validation cohort, while in the external validation cohort, the p-values were 0.014, 0.006, and 0.015, respectively). But there was no statistical difference in performance among the three models (all p-values <0.05). Correlation analysis showed that radiomics performed better than DTL in quantifying intratumoral heterogeneity differences between thymoma and thymic cysts. The developed DTL model and the Combined model based on radiomics and clinical-radiologic features achieved excellent diagnostic performance in differentiating thymic cysts from thymoma. They can serve as potential tools to assist clinical decision-making, particularly when endoscopic biopsy carries a high risk.

Novel nomogram for predicting survival in advanced non-small cell lung cancer receiving anti-PD-1 plus chemotherapy with or without antiangiogenic therapy.

This study aimed to develop and validate a novel nomogram to predict survival in advanced non-small cell lung cancer (NSCLC) receiving programmed cell death 1 (PD-1) inhibitor plus chemotherapy with or without antiangiogenic therapy. A total of 271 patients with advanced NSCLC who received anti-PD-1 plus chemotherapy with or without antiangiogenic therapy were enrolled in our center and randomized into the training cohort (n = 133) and the internal validation cohort (n = 138). Forty-five patients from another center were included as an independent external validation cohort. The nomogram was created based on the multivariate Cox regression analysis to predict overall survival (OS) and progression-free survival (PFS). The performance of the nomogram was assessed using the concordance index (C-index), the time-dependent area under the receiver operating (ROC) curves (AUCs), calibration curves, and decision curve analysis (DCA). Four factors significantly associated with OS were utilized to create a nomogram to predict OS: Eastern Cooperative Oncology Group performance status (ECOG PS), programmed cell death-ligand 1 (PD-L1) expression, chemotherapy cycle, and pretreatment lactate dehydrogenase-albumin ratio (LAR). Six variables significantly associated with PFS were incorporated into the development of a nomogram for predicting PFS: ECOG PS, histology, PD-L1 expression, chemotherapy cycle, pretreatment platelet to lymphocyte (PLR), and pretreatment LAR. The C-indexes of the nomogram for predicting OS and PFS were 0.750 and 0.747, respectively. The AUCs for predicting the 6-month, 12-month, and 18-month OS and PFS were 0.847, 0.791, and 0.776 and 0.810, 0.787, and 0.861, respectively. The calibration curves demonstrated a good agreement between predictions and actual observations. The DCA curves indicated that the nomograms had good net benefits. Furthermore, the nomogram model was well-validated in the internal and external cohorts. The novel nomogram for predicting the prognosis of advanced NSCLC receiving anti-PD-1 plus chemotherapy with or without antiangiogenic therapy may help guide clinical treatment decisions.

Differentiating adrenal metastases from benign lesions with multiphase CT imaging: Deep learning could play an active role in assisting radiologists

ObjectivesTo develop and externally validate multiphase CT-based deep learning (DL) models for differentiating adrenal metastases from benign lesions. Materials and methodsThis retrospective two-center study included 1146 adrenal lesions from 1059 patients who underwent multiphase CT scanning between January 2008 and March 2021. The study encompassed 564 surgically confirmed adenomas, along with 135 benign lesions and 447 metastases confirmed by observation. DL models based on multiphase CT images were developed, validated and tested. The diagnostic performance of the classification models, imaging phases and radiologists with or without DL were compared using accuracy (ACC) and receiver operating characteristic (ROC) curves. Integrated discrimination improvement (IDI) analysis and the DeLong test were used to compare the area under the curve (AUC) among models. Decision curve analysis (DCA) was used to assess the clinical usefulness of the predictive models. ResultsThe DL signature based on LASSO (DLSL) had a higher AUC than that of the other classification models (IDI > 0, P < 0.05). Furthermore, the precontrast phase (PCP)-based DLSL performed best in the independent external validation (AUC = 0.881, ACC = 82.9 %) and clinical test cohorts (AUC = 0.790, ACC = 70.4 %), outperforming DLSL based on the other single-phase or three-phase images (IDI > 0, P < 0.05). DCA demonstrated that PCP-based DLSL provided a higher net benefit (0.01–0.95). The diagnostic performance led to statistically significant improvements when radiologists incorporated the DL model, with the AUC improving by 0.056–0.159 and the ACC improving by 0.069–0.178 (P < 0.05). ConclusionThe DL model based on PCP CT images performed acceptably in differentiating adrenal metastases from benign lesions, and it may assist radiologists in accurate tumor staging for patients with a history of extra-adrenal malignancy.

An Artificial Intelligence Generated Automated Algorithm to Measure Total Kidney Volume in ADPKD

IntroductionAccurate tools to inform individual prognosis in patients with autosomal dominant polycystic kidney disease (ADPKD) are lacking. Here, we report an artificial intelligence (AI) generated method for routinely measuring total kidney volume (TKV). MethodsAn ensemble U-net algorithm was created using the nnUNet approach. The training and internal cross-validation cohort consisted of all 1.5T MRI data acquired using 5 different MRI scanners (454 kidneys, 227 scans) in the CYSTic consortium which was first manually segmented by a single human operator. As an independent validation cohort, we utilised 48 sequential clinical MRI scans with reference results of manual segmentation acquired by 6 individual analysts at a single centre. The tool was then implemented for clinical use and its performance analysed. ResultsThe training / internal validation cohort was younger (mean age 44.0 vs 51.5 years) and the female-male ratio higher (1.2 v 0.94) compared to the clinical validation cohort. The majority of CYSTic patients had PKD1 mutations (79%) and typical disease (Mayo Imaging Class 1, 86%). The median DICE score on the clinical validation dataset between the algorithm and human analysts was 0.96 for left and right kidneys with a median TKV error of -1.8%. The time taken to manually segment kidneys in the CYSTic dataset was 56 (±28) min whereas manual corrections of the algorithm output took 8.5 (±9.2) min per scan. ConclusionsOur AI-based algorithm demonstrates performance comparable to manual segmentation. Its rapidity and precision in real-world clinical cases demonstrate its suitability for clinical application.

Development of a Novel Prognostic Score in Patients with Waldenström Macroglobulinemia Prospectively Enrolled in the Observational NF10 Study of Fondazione Italiana Linfomi

Background. IPSSWM and revised-IPSSWM are the two currently prognostic scores used for survival prediction in symptomatic Waldenström Macroglobulinemia (WM) patients in need of treatment. In addition to age and serum beta-2 microglobulin levels, which are present in both scores, the IPSSWM includes anemia, thrombocytopenia and the concentration of the serum monoclonal component (MC), while revised-IPSSWM includes LDH and serum albumin levels. Both IPSSWM and revised-IPSSWM have been developed in retrospective series and apply only to symptomatic WM patients. Aim. To analyze WM patients prospectively enrolled in the NF10 observational study since 2010 with the aim to define a prognostic score applicable to all WM patients at diagnosis, either symptomatic or asymptomatic. Methods. The progression-free survival (PFS) and overall survival (OS) rate was estimated and plotted using the Kaplan-Meier method, with 95% confidence interval (95% CI), and the effect of covariables on PFS was estimated by means of the Cox proportional hazard (PH) regression. The association with PFS was expressed as hazard ratio (HR) with 95% CI, either in univariable or multivariable regression. All statistical tests were two-sided. The proportionality of risk of the multivariable Cox PH model was checked graphically by means of the scaled Schoenfeld residuals, and the model performance was analyzed using the likelihood displacement method. Starting from a model with 10 variables selected on the basis of clinical and statistical considerations, the final model was selected on considering the Akaike (AIC) and likelihood ratio test. The cut-offs were selected after modelling the covariate with restricted cubic spline in Cox regression. The final model was internally validated using bootstrap techniques to evaluate the discriminant power (C-index), slope shrinkage to check for overfitting, and calibration (comparison between predicted and actual survival). Results. We identified 303 eligible patients from the NF10 study, of whom 185 (60%) needed treatment immediately after diagnosis (42%) or later (18%). These were main patients characteristics: 59% were male, 46% were aged &amp;gt;70 years, 8% had constitutional symptoms, 55% had Hb &amp;lt;12 g/dL, 54% had an absolute lymphocyte count ≥2x10 9/L, 9% had platelets &amp;lt;100x10 9/L, 37% had a serum MC &amp;gt;2 g/dL, 21% had albumin &amp;lt;3.5 g/dL, 23% had LDH &amp;gt; upper limit of normal (ULN), 32% had beta-2 microglobulin &amp;gt;1.5xULN. With a median follow-up of 51 months, 5-year PFS for the entire cohort was 67% (95%CI 60-73%). In univariate analysis, age &amp;gt;70 years (P=0.043), constitutional symptoms (P=0.009), nodal sites (P=0.002), Hb &amp;lt;12 g/dL (P&amp;lt;0.001), absolute lymphocyte count ≥2x10 9/L (P=0.033), serum MC &amp;gt;2 g/dL (P=0.003), high beta-2 microglobulin (P&amp;lt;0.001) and LDH (P=0.001), albumin &amp;lt;3.5 g/dL (P&amp;lt;0.001), C-reactive protein &amp;gt;4 mg/dL (P=0.008) were associated with shorter PFS. In multivariate analysis, age, anemia, hypoalbuminemia and high LDH retained prognostic value and thus entered the final model. Patients were classified into 3 risk groups: low (0-1 factors, 62% of patients), intermediate (2 factors, 24%) or high (3-4 factors, 15%), with a 5-year PFS of 80% (95%CI 70-87%), 49% (95%CI 32-64%) and 25% (95%CI 11-42%), respectively, with C-Harrell=0.744 and robust internal validation and calibration. OS was respectively 96% (95%CI 90-98%), 77% (95%CI 62-87%) and 43% (95%CI 25-60%) for the low, intermediate and high risk group (Figure 1). Conclusions. We developed a prognostic model based on 4 parameters easily attainable in clinical practice, able to predict PFS and OS at diagnosis. This is the first prognostic score for WM developed in a prospective study and applicable to patients presenting either with symptomatic or asymptomatic disease. This prognostic model identifies a subset of high-risk patients with a dismal outcome and a median OS &amp;lt;5 years. External validation in independent cohorts is needed.

Metabolomics-Driven Systematic Research Reveals Different Metabolomic and Lipidomic Profiling of Bone Marrow Plasma in Patients with Multiple Myeloma

Myeloma cells resident in bone marrow (BM), therefore the microenvironment plays critical roles in the invasion, proliferation and migration. Obesity increases the morbidity and drug resistance in patient of multiple myeloma (MM) through changing the bone marrow environment. Extracellular environments such as the BM plasma likely have unique lipidomic metabolite profiles that predicts the prognosis in patients with multiple myeloma. However, the study of multiple omics on metabolic changes in BM of MM patients are still very limited. The present study utilized an untargeted metabolite and targeted complex lipid profiling (Liquid chromatography-mass spectrometry, LC-MS) analysis to identify significant differences in the metabolite levels of BM plasma between newly diagnosed MM patients with different BMI (≤24 kg/m 2, n = 10; ≥25 kg/m 2, n = 10). This was followed by verification of some of the metabolite differences of interest by targeted quantification of the metabolites using isotopic internal standards in the exploratory cohort as well as an independent validation cohort. Significant differences were noted in the amino acid profiles, lipid metabolism and amino sugar metabolism, such as increased L-Glutamate, L-Proline, D-Proline, palmitoleic acid and D-Fructose, etc between MM patients with different BMI ( Figure A-C). Glycerol phosphate shuttle, glycerolipid metabolism and de novo triacylglycerol biosynthesis were the top 3 enriched signaling pathways in BM of obese MM patients ( Figure D). Targeted complex lipid profiling analysis further showed that triglyceride (TG)，diglyceride (DG) were increased notably in obese BM plasma of MM patients and correlated significantly with BMI ( Figure E-G). Glycerolipids, triradylglycerols, diradylglycerols and fatty acyls metabolism were enriched in obese group by GSEA analysis ( Figure H). Furthermore, KEGG enrichment analysis showed differentiate expressed lipid metabolites in BM plasma of obese MM patients comparing to that of lean group, including regulation of lipolysis in adipocytes, glycerolipid metabolism, fat digestion and absorption, cholesterol metabolism and metabolic pathways, etc ( Figure I ). In summary, metabolite and complex lipid profiling of the BM plasma identifies differences in levels of metabolites and lipids between MM patients with different BMI. This may provide insight into the explanation that obesity increases the susceptibility to MM and drug resistance. Acknowledgement This project is supported by the Sun Yat-sen University Hundred Talents Program (PT19200101), Natural Science Foundation of Guangdong Province (2022A1515010290).

Genomic Tumor Correlates of Clinical Outcomes Following Organ-Sparing Chemoradiation Therapy for Bladder Cancer.

There is an urgent need for biomarkers of radiation response in organ-sparing therapies. Bladder preservation with trimodality therapy (TMT), consisting of transurethral tumor resection followed by chemoradiation, is an alternative to radical cystectomy for muscle-invasive bladder cancer (MIBC), but molecular determinants of response are poorly understood. We characterized genomic and transcriptomic features correlated with long-term response in a single institution cohort of patients with MIBC homogeneously treated with TMT. Pretreatment tumors from 76 patients with MIBC underwent whole-exome sequencing; 67 underwent matched transcriptomic profiling. Molecular features were correlated with clinical outcomes including modified bladder-intact event-free survival (mBI-EFS), a composite endpoint that reflects long-term cancer control with bladder preservation. With a median follow-up of 74.6 months in alive patients, 37 patients had favorable long-term response to TMT while 39 had unfavorable long-term response. Tumor mutational burden was not associated with outcomes after TMT. DNA damage response gene alterations were associated with improved locoregional control and mBI-EFS. Of these alterations, somatic ERCC2 mutations stood out as significantly associated with favorable long-term outcomes; patients with ERCC2 mutations had significantly improved mBI-EFS [HR, 0.15; 95% confidence interval (CI), 0.06-0.37; P = 0.030] and improved BI-EFS, an endpoint that includes all-cause mortality (HR, 0.33; 95% CI, 0.15-0.68; P = 0.044). ERCC2 mutant bladder cancer cell lines were significantly more sensitive to concurrent cisplatin and radiation treatment in vitro than isogenic ERCC2 wild-type cells. Our data identify ERCC2 mutation as a candidate biomarker associated with sensitivity and long-term response to chemoradiation in MIBC. These findings warrant validation in independent cohorts.

Identification and validation of M2 macrophage-related genes in endometriosis

AimsM2 macrophage is believed to play an important role in the development of endometriosis. This study aimed to identify several key genes related to the M2 macrophage in endometriosis. MethodDifferential expressed genes between endometriosis and non-endometriosis were identified based on three microarray datasets from the Gene Expression Omnibus database. Gene modules significantly associated with M2 macrophage were identified from the weighted gene co-expression network analysis. Furthermore, by intersecting the differential expressed genes and M2 macrophage-associated module genes, M2 macrophage-related genes in endometriosis were identified. Functional analyses of the Gene Ontology and Kyoto Encyclopedia of Genes and Genomes for these genes were then performed. Following, the least absolute shrinkage and selection operator, random forest, and receiver operating characteristic curves were further conducted to identify the key M2 macrophage-related genes in endometriosis. Finally, the expressions of key genes in endometriosis, as well as their correlations with M2 macrophages were verified in an independent validation cohort. ResultsTotally, 185 M2 macrophage-related genes were identified, and they were mainly enriched in functions associated with the cell cycle, oocyte maturation, and immune response. Following machine learning algorithms, eight key genes were selected in the endometriosis: PGR, OLFM4, PIP5K1B, CCNA1, BRIP1, CADM1, PRAME, and GCNT1. The eight key genes were confirmed to be negative with M2 macrophage infiltration levels. Furthermore, the expression levels of these genes were significantly lower in the middle secretory stage while relevantly higher in the proliferative stage. The validation analysis also showed similar outcomes with the above results. ConclusionEight M2 macrophage-related genes were identified as potential biomarkers of endometriosis, providing novel understanding of immune cells in the endometriosis.

Application of Machine Learning Models to Biomedical and Information System Signals From Critically Ill Adults

Machine learning (ML)-derived notifications for impending episodes of hemodynamic instability and respiratory failure events are interesting because they can alert physicians in time to intervene before these complications occur. Do ML alerts, telemedicine system (TS)-generated alerts, or biomedical monitors (BMs) have superior performance for predicting episodes of intubation or administration of vasopressors? An ML algorithm was trained to predict intubation and vasopressor initiation events among critically ill adults. Its performance was compared with BM alarms and TS alerts. ML notifications were substantially more accurate and precise, with 50-fold lower alarm burden than TS alerts for predicting vasopressor initiation and intubation events. ML notifications of internal validation cohorts demonstrated similar performance for independent academic medical center external validation and COVID-19 cohorts. Characteristics were also measured for a control group of recent patients that validated event detection methods and compared TS alert and BM alarm performance. The TS test characteristics were substantially better, with 10-fold less alarm burden than BM alarms. The accuracy of ML alerts (0.87-0.94) was in the range of other clinically actionable tests; the accuracy of TS (0.28-0.53) and BM (0.019-0.028) alerts were not. Overall test performance (F scores) for ML notifications were more than fivefold higher than for TS alerts, which were higher than those of BM alarms. ML-derived notifications for clinically actioned hemodynamic instability and respiratory failure events represent an advance because the magnitude of the differences of accuracy, precision, misclassification rate, and pre-event lead time is large enough to allow more proactive care and has markedly lower frequency and interruption of bedside physician work flows.

The Characteristics of Transcription Factors Regulating T Cell Exhaustion Were Analyzed to Predict the Prognosis and Therapeutic Effect in Patients with HCC.

Hepatocellular carcinoma (HCC) ranks as the third leading cause of cancer-related deaths, posing a significant threat to people in diverse regions. T-cell exhaustion (Tex) can hinder the efficacy of immunotherapy in patients with HCC, and the transcription factors that regulate Tex in HCC have not yet been fully elucidated. We used the single sample gene set enrichment analysis (ssGSEA) method to define the transcription factor pathway that regulates Tex and employed LASSO regression analysis to establish Tex related genes (TEXRS). To predict differences in immunotherapy efficacy between the two groups, we used the immunophenotype score and submap algorithm. RT-qPCR was used to detect the expression levels of the model genes in 21 pairs of HCC tissues. Finally, we assessed the cell communication strength and identified ligand receptors using the "CellChat" R package. Nine Tex transcription factors were identified as regulators of the HCC immune microenvironment, with Tex scores affecting patient survival. Patients with a high Tex Risk Score (TEXRS) had significantly worse overall survival compared to patients with low TEXRS. After adjusting for confounding factors, TEXRS remained an independent prognostic factor. Importantly, TEXRS performed well in multiple independent external validation cohorts. Various algorithms have shown that patients in the low-TEXRS group might benefit more from immunotherapy. Finally, RT-qPCR analysis of 21 HCC samples showed that C7, CD5L, and SDS were significantly downregulated in HCC tissues, consistent with the bioinformatics analysis results. TEXRS proved to be a valuable predictor of immunotherapy and transcatheter arterial chemoembolization efficacy in patients with HCC. This holds promise for enhancing the prognosis and treatment outcomes of patients with HCC.

Multiphase and multiparameter MRI-based radiomics for prediction of tumor response to neoadjuvant therapy in locally advanced rectal cancer

BackgroundTo develop and validate radiomics models for prediction of tumor response to neoadjuvant therapy (NAT) in patients with locally advanced rectal cancer (LARC) using both pre-NAT and post-NAT multiparameter magnetic resonance imaging (mpMRI).MethodsIn this multicenter study, a total of 563 patients were included from two independent centers. 453 patients from center 1 were split into training and testing cohorts, the remaining 110 from center 2 served as an external validation cohort. Pre-NAT and post-NAT mpMRI was collected for feature extraction. The radiomics models were constructed using machine learning from a training cohort. The accuracy of the models was verified in a testing cohort and an independent external validation cohort. Model performance was evaluated using area under the curve (AUC), sensitivity, specificity, positive predictive value, and negative predictive value.ResultsThe model constructed with pre-NAT mpMRI had favorable accuracy for prediction of non-response to NAT in the training cohort (AUC = 0.84), testing cohort (AUC = 0.81), and external validation cohort (AUC = 0.79). The model constructed with both pre-NAT and post-NAT mpMRI had powerful diagnostic value for pathologic complete response in the training cohort (AUC = 0.86), testing cohort (AUC = 0.87), and external validation cohort (AUC = 0.87).ConclusionsModels constructed with multiphase and multiparameter MRI were able to predict tumor response to NAT with high accuracy and robustness, which may assist in individualized management of LARC.

A CT-based radiomics classification model for the prediction of histological type and tumour grade in retroperitoneal sarcoma (RADSARC-R): a retrospective multicohort analysis

Retroperitoneal sarcomas are tumours with a poor prognosis. Upfront characterisation of the tumour is difficult, and under-grading is common. Radiomics has the potential to non-invasively characterise the so-called radiological phenotype of tumours. We aimed to develop and independently validate a CT-based radiomics classification model for the prediction of histological type and grade in retroperitoneal leiomyosarcoma and liposarcoma. A retrospective discovery cohort was collated at our centre (Royal Marsden Hospital, London, UK) and an independent validation cohort comprising patients recruited in the phase 3 STRASS study of neoadjuvant radiotherapy in retroperitoneal sarcoma. Patients aged older than 18 years with confirmed primary leiomyosarcoma or liposarcoma proceeding to surgical resection with available contrast-enhanced CT scans were included. Using the discovery dataset, a CT-based radiomics workflow was developed, including manual delineation, sub-segmentation, feature extraction, and predictive model building. Separate probabilistic classifiers for the prediction of histological type and low versus intermediate or high grade tumour types were built and tested. Independent validation was then performed. The primary objective of the study was to develop radiomic classification models for the prediction of retroperitoneal leiomyosarcoma and liposarcoma type and histological grade. 170 patients recruited between Oct 30, 2016, and Dec 23, 2020, were eligible in the discovery cohort and 89 patients recruited between Jan 18, 2012, and April 10, 2017, were eligible in the validation cohort. In the discovery cohort, the median age was 63 years (range 27-89), with 83 (49%) female and 87 (51%) male patients. In the validation cohort, median age was 59 years (range 33-77), with 46 (52%) female and 43 (48%) male patients. The highest performing model for the prediction of histological type had an area under the receiver operator curve (AUROC) of 0·928 on validation, based on a feature set of radiomics and approximate radiomic volume fraction. The highest performing model for the prediction of histological grade had an AUROC of 0·882 on validation, based on a radiomics feature set. Our validated radiomics model can predict the histological type and grade of retroperitoneal sarcomas with excellent performance. This could have important implications for improving diagnosis and risk stratification in retroperitoneal sarcomas. Wellcome Trust, European Organisation for Research and Treatment of Cancer-Soft Tissue and Bone Sarcoma Group, the National Institutes for Health, and the National Institute for Health and Care Research Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research.

Urinary peptide analysis to predict the response to blood pressure medication.

The risk of diabetic kidney disease (DKD) progression is significant despite treatment with renin-angiotensin system (RAS) blocking agents. Current clinical tools cannot predict whether or not patients will respond to treatment with RAS inhibitors (RASi). We aimed to investigate whether proteome analysis could identify urinary peptides as biomarkers that could predict the response to angiotensin-converting enzyme inhibitor and angiotensin-receptor blockers treatment to avoid DKD progression. Furthermore, we investigated the comparability of the estimated glomerular filtration rate (eGFR), calculated using four different GFR equations, for DKD progression. We evaluated urine samples from a discovery cohort of 199 diabetic patients treated with RASi. DKD progression was defined based on eGFR percentage slope results between visits (∼1 year) and for the entire period (∼3 years) based on the eGFR values of each GFR equation. Urine samples were analysed using capillary electrophoresis-coupled mass spectrometry. Statistical analysis was performed between the uncontrolled (patients who did not respond to RASi treatment) and controlled kidney function groups (patients who responded to the RASi treatment). Peptides were combined in a support vector machine-based model. The area under the receiver operating characteristic curve was used to evaluate the risk prediction models in two independent validation cohorts treated with RASi. The classification of patients into uncontrolled and controlled kidney function varies depending on the GFR equation used, despite the same sample set. We identified 227 peptides showing nominal significant difference and consistent fold changes between uncontrolled and controlled patients in at least three methods of eGFR calculation. These included fragments of collagens, alpha-1-antitrypsin, antithrombin-III, CD99 antigen and uromodulin. A model based on 189 of 227 peptides (DKDp189) showed a significant prediction of non-response to the treatment/DKD progression in two independent cohorts. The DKDp189 model demonstrates potential as a predictive tool for guiding treatment with RASi in diabetic patients.

Application of Machine Learning in Predicting Hepatic Metastasis or Primary Site in Gastroenteropancreatic Neuroendocrine Tumors.

Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) account for 80% of gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs). GEP-NETs are well-differentiated tumors, highly heterogeneous in biology and origin, and are often diagnosed at the metastatic stage. Diagnosis is commonly through clinical symptoms, histopathology, and PET-CT imaging, while molecular markers for metastasis and the primary site are unknown. Here, we report the identification of multi-gene signatures for hepatic metastasis and primary sites through analyses on RNA-SEQ datasets of pancreatic and small intestinal NETs tissue samples. Relevant gene features, identified from the normalized RNA-SEQ data using the mRMRe algorithm, were used to develop seven Machine Learning models (LDA, RF, CART, k-NN, SVM, XGBOOST, GBM). Two multi-gene random forest (RF) models classified primary and metastatic samples with 100% accuracy in training and test cohorts and >90% accuracy in an independent validation cohort. Similarly, three multi-gene RF models identified the pancreas or small intestine as the primary site with 100% accuracy in training and test cohorts, and >95% accuracy in an independent cohort. Multi-label models for concurrent prediction of hepatic metastasis and primary site returned >98.42% and >87.42% accuracies on training and test cohorts, respectively. A robust molecular signature to predict liver metastasis or the primary site for GEP-NETs is reported for the first time and could complement the clinical management of GEP-NETs.

Higher TP53BP2 expression is associated with HBsAg loss in peginterferon-α-treated patients with chronic hepatitis B

HBsAg loss is only observed in a small proportion of patients with chronic hepatitis B (CHB) who undergo interferon treatment. Investigating the host factors crucial for functional cure of CHB can aid in identifying individuals who would benefit from peginterferon-α (Peg-IFNα) therapy. We conducted a genome-wide association study (GWAS) by enrolling 48 patients with CHB who achieved HBsAg loss after Peg-IFNαtreatment and 47 patients who didn't. In the validation stage, we included 224 patients, of whom 90 had achieved HBsAg loss, to validate the identified significant single nucleotide polymorphisms. To verify the functional involvement of the candidate genes identified, we performed a series of invitro and invivo experiments. GWAS results indicated a significant association between the rs7519753C allele and serum HBsAg loss in patients with CHB after Peg-IFNα treatment (p= 4.85× 10-8, odds ratio= 14.47). This association was also observed in two independent validation cohorts. Expression quantitative trait locus analysis revealed higher hepatic TP53BP2 expression in individuals carrying the rs7519753C allele (p= 2.90× 10-6). RNA-sequencing of liver biopsies from patients with CHB after Peg-IFNα treatment revealed that hepatic TP53BP2 levels were significantly higher in the HBsAg loss group compared to the HBsAg persistence group (p= 0.035). Invitro and invivo experiments demonstrated that loss of TP53BP2 decreased interferon-stimulated gene levels and the anti-HBV effect of IFN-α. Mechanistically, TP53BP2 was found to downregulate SOCS2, thereby facilitating JAK/STAT signaling. The rs7519753C allele is associated with elevated hepatic TP53BP2 expression and an increased probability of serum HBsAg loss post-Peg-IFNα treatment in patients with CHB. TP53BP2 enhances the response of the hepatocyte to IFN-α by suppressing SOCS2 expression. Chronic hepatitis B (CHB) remains a global public health issue. Although current antiviral therapies are more effective in halting disease progression, only a few patients achieve functional cure for hepatitis B with HBsAg loss, highlighting the urgent need for a cure for CHB. This study revealed that the rs7519753C allele, which is associated with high expression of hepatic TP53BP2, significantly increases the likelihood of serum HBsAg loss in patients with CHB undergoing Peg-IFNα treatment. This finding not only provides a promising predictor for HBsAg loss but identifies a potential therapeutic target for Peg-IFNα treatment. We believe our results are of great interest to a wide range of stakeholders based on their potential clinical implications.

Development and validation of a nomogram predictive model for colorectal adenoma with low-grade intraepithelial neoplasia using routine laboratory tests: A single-center case-control study in China

BackgroundColorectal cancer (CRC) is the third most common cancer in the world and has a high mortality rate. Colorectal adenoma (CRA) is precancerous lesions of CRC. The purpose of the present study was to construct a nomogram predictive model for CRA with low-grade intraepithelial neoplasia (LGIN) in order to identify high-risk individuals, facilitating early diagnosis and treatment, and ultimately reducing the incidence of CRC. MethodsWe conducted a single-center case-control study. Based on the results of colonoscopy and pathology, 320 participants were divided into the CRA group and the control group, the demographic and laboratory test data were collected. A development cohort (n = 223) was used for identifying the risk factors for CRA with LGIN and to develop a predictive model, followed by an internal validation. An independent validation cohort (n = 97) was used for external validation. Receiver operating characteristic curve, calibration plot and decision curve analysis were used to evaluate discrimination ability, accuracy and clinical practicability of the model. ResultsFour predictors, namely sex, age, albumin and monocyte count, were included in the predictive model. In the development cohort, internal validation and external validation cohort, the area under the curve (AUC) of this risk predictive model were 0.946 (95%CI: 0.919–0.973), 0.909 (95 % CI: 0.869–0.940) and 0.928 (95%CI: 0.876–0.980), respectively, which demonstrated the model had a good discrimination ability. The calibration plots showed a good agreement and the decision curve analysis (DCA) suggested the predictive model had a high clinical net benefit. ConclusionThe nomogram model exhibited good performance in predicting CRA with LGIN, which can aid in the early detection of high-risk patients, improve early treatment, and ultimately reduce the incidence of CRC.

Prediction of clinically significant prostate cancer through urine metabolomic signatures: A large-scale validated study

PurposeCurrently, there are no accurate markers for predicting potentially lethal prostate cancer (PC) before biopsy. This study aimed to develop urine tests to predict clinically significant PC (sPC) in men at risk.MethodsUrine samples from 928 men, namely, 660 PC patients and 268 benign subjects, were analyzed by gas chromatography/quadrupole time-of-flight mass spectrophotometry (GC/Q-TOF MS) metabolomic profiling to construct four predictive models. Model I discriminated between PC and benign cases. Models II, III, and GS, respectively, predicted sPC in those classified as having favorable intermediate risk or higher, unfavorable intermediate risk or higher (according to the National Comprehensive Cancer Network risk groupings), and a Gleason sum (GS) of ≥ 7. Multivariable logistic regression was used to evaluate the area under the receiver operating characteristic curves (AUC).ResultsIn Models I, II, III, and GS, the best AUCs (0.94, 0.85, 0.82, and 0.80, respectively; training cohort, N = 603) involved 26, 24, 26, and 22 metabolites, respectively. The addition of five clinical risk factors (serum prostate-specific antigen, patient age, previous negative biopsy, digital rectal examination, and family history) significantly improved the AUCs of the models (0.95, 0.92, 0.92, and 0.87, respectively). At 90% sensitivity, 48%, 47%, 50%, and 36% of unnecessary biopsies could be avoided. These models were successfully validated against an independent validation cohort (N = 325). Decision curve analysis showed a significant clinical net benefit with each combined model at low threshold probabilities. Models II and III were more robust and clinically relevant than Model GS.ConclusionThis urine test, which combines urine metabolic markers and clinical factors, may be used to predict sPC and thereby inform the necessity of biopsy in men with an elevated PC risk.

Clinical prediction of HBV-associated cirrhosis using machine learning based on platelet and bile acids

ObjectivesThe present study was conducted to evaluate the performance of serum bile acids in the prediction of cirrhosis in chronic hepatitis B (CHB) population. MethodsDysregulated metabolites were explored using untargeted and targeted metabolomic analyses. A machine learning model based on platelet (PLT) and several bile acids was constructed using light gradient boosting machine (LightGBM), to differentiate HBV-associated cirrhosis (BAC) from CHB patients. ResultsSerum bile acids were dysregulated in BAC compared to CHB patients. The LightGBM model consisted of PLT, TUDCA, UDCA, TLCA, LCA and CA. The model demonstrated a strong discrimination ability in the internal test subset of the training cohort to diagnose BAC from CHB patients (AUC = 0.97). The high diagnostic accuracy of the model was further validated in an independent validation cohort. In addition, the model had high predictive efficacy in discriminating compensated BAC from CHB patients (AUC = 0.89). The performance of the model was better than AST/ALT ratio and the gradient boosting (GB)-based model reported in previous studies. ConclusionsOur study showed that this LightGBM model based on PLT and 5 bile acids has potential in clinical assessments of CHB progression and will be useful for early detection of cirrhosis in CHB patients.