Multicenter Validation Research Articles

Abstract PO5-06-08: Real-world analysis of efficacy, adverse events and predictive biomarkers for advanced triple negative breast cancer (TNBC) treated with immune check point Inhibitors (ICI): A single center experience

Abstract Introduction KEYNOTE-355 investigated the addition of pembrolizumab to chemotherapy in advanced TNBC patients (pts) with PD-L1 positive tumors showing median overall survival (OS) of 23.0 months (mos) and median progression-free survival (PFS) of 16.1 mos. Given significant OS benefit with pembrolizumab, it is the current standard of care while the indication of atezolizumab has been withdrawn due to lack of OS benefit in IMPASSION-130. Clinical efficacy and adverse events in pts on this treatment regimen in clinical practice is unknown and understanding the real-world outcomes of this regimen is critical. Methods We conducted a retrospective, single-center observational study among TNBC pts treated with ICI (Pembrolizumab or Atezolizumab) at Roswell Park Comprehensive Cancer Center from January 2017 to May 2023. Demographics and clinicopathological variables were collected including comorbidities, laboratory data, sites of metastases (mets), treatment received, immune related adverse events (irAEs), and clinical outcomes. Adverse events were reported using the Common Terminology Criteria for Adverse Events v5.0. Serial CT scans were reviewed and response rate was determined using RECIST v1.1. Patient demographic, clinical and outcome characteristics were summarized by treatment and survival outcomes were obtained using standard Kaplan-Meier methods. Associations between survival outcomes and baseline or treatment characteristics were evaluated using Cox regression models using Firth’s method. All analyses were conducted in SAS v9.4 at a significance level of 0.05. Results A total of 44 pts with advanced TNBC treated with ICI were included (23 received pembrolizumab and 21 received Atezolizumab). The study population consisted of all female pts with stage IV disease, median age 53.7 years (IQR 30.4-85.2), 68.2 % (30/44) White, 25.0% (11/44) Black, 34.1% (15/44) obese (BMI≥30). 27.3% (12/44) pts developed any grade irAEs which included myocarditis (4.5%), rash (9.1%), hyperthyroidism (6.8%), hypothyroidism (4.5%), adrenal insufficiency (2.3%), diabetes mellitus (2.4%), colitis (2.3%), and transaminitis (2.3%). Distribution of reported irAE was 25% grade 1 (3/12), 66.7% grade 2 (8/12) and 8.3% grade 3 (1/12). 75% pts (33/44) received standard treatment (ICI+ chemotherapy) of which 72.7 % (24/33) received it in the first-line setting. Median OS in pts treated in first line was 16.2 mos (95% CI, 10.7-NR) and median PFS was 4.4 mos (95% CI, 2.7-8.4). Overall response rate (ORR) was 29.1% (7/24) of which 8.3% (2/24) had complete response and 20.8% (5/24) had partial response. ORR was 22.2 % (2/9) in pembrolizumab cohort and 33.3 % (5/15) in atezolizumab cohort. Among pts with first line treatment, obese pts were found to have improved PFS compared to non-obese (11.5 vs 3.8 mos, univariate hazard ratio (HR) 0.46, 95% CI 0.23-0.93, p= 0.031), and this difference was maintained in multivariable analysis even after adjusting for age (adjusted HR (aHR) 0.40, 95% CI 0.17-0.98, p= 0.044). Pts with brain mets had poor extracranial PFS compared to those without brain mets (2.8 vs. 5.2 mos, HR 2.25, 95% CI 1.08-4.68, p= 0.036), however this difference was not observed when further adjusted for age (aHR 2.27, 95% CI= 0.91-5.68, p= 0.08). Conclusion Clinical outcomes in our study were inferior to KEYNOTE-355 where median PFS was 16.1 mos and OS was 23.0 mos and ORR was 52.7% for PD-L1 positive population. This may reflect a more heterogeneous population of pts treated in routine clinical practice who are typically less fit than pts on clinical trials. Our study found improved outcomes among obese patients, similar to data reported in other disease settings. These data warrant multi-center validation with larger number of patients. Citation Format: Archit Patel, Arya Mariam Roy, Malak Alharbi, Kristopher Attwood, Chi-Chen Hong, Song Yao, Thaer Khoury, Amy Early, Tracey O'Connor, Ellis Levine, Shipra Gandhi. Real-world analysis of efficacy, adverse events and predictive biomarkers for advanced triple negative breast cancer (TNBC) treated with immune check point Inhibitors (ICI): A single center experience [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-06-08.

One-stop detection of anterior cruciate ligament injuries on magnetic resonance imaging using deep learning with multicenter validation.

Anterior cruciate ligament (ACL) injuries are closely associated with knee osteoarthritis (OA). However, diagnosing ACL injuries based on knee magnetic resonance imaging (MRI) has been subjective and time-consuming for clinical doctors. Therefore, we aimed to devise a deep learning (DL) model leveraging MRI to enable a comprehensive and automated approach for the detection of ACL injuries. A retrospective study was performed extracting data from the Osteoarthritis Initiative (OAI). A total of 1,589 knees (comprising 1,443 intact, 90 with partial tears, and 56 with full tears) were enrolled to construct the classification model. This one-stop detection pipeline was developed using a tailored YOLOv5m architecture and a ResNet-18 convolutional neural network (CNN) to facilitate tasks based on sagittal 2-dimensional (2D) intermediate-weighted fast spin-echo sequence at 3.0T. To ensure the reliability and robustness of the classification system, it was subjected to external validation across 3 distinct datasets. The accuracy, sensitivity, specificity, and the mean average precision (mAP) were utilized as the evaluation metric for the model performance by employing a 5-fold cross-validation approach. The radiologist's interpretations were employed as the reference for conducting the evaluation. The localization model demonstrated an accuracy of 0.89 and a sensitivity of 0.93, achieving a mAP score of 0.96. The classification model demonstrated strong performance in detecting intact, partial tears, and full tears at the optimal threshold on the internal dataset, with sensitivities of 0.941, 0.833, and 0.929, specificities of 0.925, 0.947, and 0.991, and accuracies of 0.940, 0.941, and 0.989, respectively. In comparison, on a subset consisting of 171 randomly selected knees from the OAI, the radiologists demonstrated a sensitivity ranging between 0.660 and 1.000, specificity ranging between 0.691 and 1.000, and accuracy ranging between 0.689 and 1.000. On a subset consisting of 170 randomly selected knees from the Chinese dataset, the radiologists exhibited a sensitivity ranging between 0.711 and 0.948, specificity ranging between 0.768 and 0.977, and accuracy ranging between 0.683 and 0.917. After retraining, the model achieved sensitivities ranging between 0.630 and 0.961, specificities ranging between 0.860 and 0.961, and accuracies ranging between 0.832 and 0.951, respectively, on the external validation dataset. The proposed model utilizing knee MRI showcases robust performance in the domains of ACL localization and classification.

PECARN prediction rules for CT imaging of children presenting to the emergency department with blunt abdominal or minor head trauma: a multicentre prospective validation study

The intra-abdominal injury and traumatic brain injury prediction rules derived by the Pediatric Emergency Care Applied Research Network (PECARN) were designed to reduce inappropriate use of CT in children with abdominal and head trauma, respectively. We aimed to validate these prediction rules for children presenting to emergency departments with blunt abdominal or minor head trauma. For this prospective validation study, we enrolled children and adolescents younger than 18 years presenting to six emergency departments in Sacramento (CA), Dallas (TX), Houston (TX), San Diego (CA), Los Angeles (CA), and Oakland (CA), USA between Dec 27, 2016, and Sept 1, 2021. We excluded patients who were pregnant or had pre-existing neurological disorders preventing examination, penetrating trauma, injuries more than 24 h before arrival, CT or MRI before transfer, or high suspicion of non-accidental trauma. Children presenting with blunt abdominal trauma were enrolled into an abdominal trauma cohort, and children with minor head trauma were enrolled into one of two age-segregated minor head trauma cohorts (younger than 2 years vs aged 2 years and older). Enrolled children were clinically examined in the emergency department, and CT scans were obtained at the attending clinician's discretion. All enrolled children were evaluated against the variables of the pertinent PECARN prediction rule before CT results were seen. The primary outcome of interest in the abdominal trauma cohort was intra-abdominal injury undergoing acute intervention (therapeutic laparotomy, angiographic embolisation, blood transfusion, intravenous fluid for ≥2 days for pancreatic or gastrointestinal injuries, or death from intra-abdominal injury). In the age-segregated minor head trauma cohorts, the primary outcome of interest was clinically important traumatic brain injury (neurosurgery, intubation for >24 h for traumatic brain injury, or hospital admission ≥2 nights for ongoing symptoms and CT-confirmed traumatic brain injury; or death from traumatic brain injury). 7542 children with blunt abdominal trauma and 19 999 children with minor head trauma were enrolled. The intra-abdominal injury rule had a sensitivity of 100·0% (95% CI 98·0-100·0; correct test for 145 of 145 patients with intra-abdominal injury undergoing acute intervention) and a negative predictive value (NPV) of 100·0% (95% CI 99·9-100·0; correct test for 3488 of 3488 patients without intra-abdominal injuries undergoing acute intervention). The traumatic brain injury rule for children younger than 2 years had a sensitivity of 100·0% (93·1-100·0; 42 of 42) for clinically important traumatic brain injuries and an NPV of 100·0%; 99·9-100·0; 2940 of 2940), whereas the traumatic brain injury rule for children aged 2 years and older had a sensitivity of 98·8% (95·8-99·9; 168 of 170) and an NPV of 100·0% (99·9-100·0; 6015 of 6017). The two children who were misclassified by the traumatic brain injury rule were admitted to hospital for observation but did not need neurosurgery. The PECARN intra-abdominal injury and traumatic brain injury rules were validated with a high degree of accuracy. Their implementation in paediatric emergency departments can therefore be considered a safe strategy to minimise inappropriate CT use in children needing high-quality care for abdominal or head trauma. The Eunice Kennedy Shriver National Institute of Child Health and Human Development.

Cooperative AI training for cardiothoracic segmentation in computed tomography: An iterative multi-center annotation approach

PurposeRadiological reporting is transitioning to quantitative analysis, requiring large-scale multi-center validation of biomarkers. A major prerequisite and bottleneck for this task is the voxelwise annotation of image data, which is time-consuming for large cohorts. In this study, we propose an iterative training workflow to support and facilitate such segmentation tasks, specifically for high-resolution thoracic CT data. MethodsOur study included 132 thoracic CT scans from clinical practice, annotated by 13 radiologists. In three iterative training experiments, we aimed to improve and accelerate segmentation of the heart and mediastinum. Each experiment started with manual segmentation of 5–25 CT scans, which served as training data for a nnU-Net. Further iterations incorporated AI pre-segmentation and human correction to improve accuracy, accelerate the annotation process, and reduce human involvement over time. ResultsResults showed consistent improvement in AI model quality with each iteration. Resampled datasets improved the Dice similarity coefficients for both the heart (DCS 0.91 [0.88; 0.92]) and the mediastinum (DCS 0.95 [0.94; 0.95]). Our AI models reduced human interaction time by 50 % for heart and 70 % for mediastinum segmentation in the most potent iteration. A model trained on only five datasets achieved satisfactory results (DCS > 0.90). ConclusionsThe iterative training workflow provides an efficient method for training AI-based segmentation models in multi-center studies, improving accuracy over time and simultaneously reducing human intervention. Future work will explore the use of fewer initial datasets and additional pre-processing methods to enhance model quality.

Artificial Intelligence in Emergency Trauma Care: A Preliminary Scoping Review.

This study aimed to analyze the use of generative artificial intelligence in the emergency trauma care setting through a brief scoping review of literature published between 2014 and 2024. An exploration of the NCBI repository was performed using a search string of selected keywords that returned N=87 results; articles that met the inclusion criteria (n=28) were reviewed and analyzed. Heterogeneity sources were explored and identified by a significance threshold of P < 0.10 or an I2 value exceeding 50%. If applicable, articles were categorized within three primary domains: triage, diagnostics, or treatment. Findings suggest that CNNs demonstrate strong diagnostic performance for diverse traumatic injuries, but generalized integration requires expanded prospective multi-center validation. Injury scoring models currently experience calibration gaps in mortality quantification and lesion localization that can undermine clinical utility by permitting false negatives. Triage predictive models now confront transparency, explainability, and healthcare ecosystem integration barriers limiting real-world translation. The most significant literature gap centers on treatment-oriented generative AI applications that provide real-time guidance for urgent trauma interventions rather than just analytical support.

Are Radiomic Spleen Features Useful for Assessing the Response to Infliximab in Patients With Crohn's Disease? A Multicenter Study.

To develop and validate a radiomics nomogram for assessing the response of patients with Crohn's disease (CD) to infliximab. Radiomics features of the spleen were extracted from computed tomography enterography images of each patient's arterial phase. The feature selection process was performed using the least absolute shrinkage and selection operator algorithm, and a radiomics score was calculated based on the radiomics signature formula. Subsequently, the radiomic model and the clinical risk factor model were separately established based on the radiomics score and clinically significant features, respectively. The performance of both models was evaluated using receiver operating characteristic curves, decision curve analysis curves, and clinical impact curves. Among the 175 patients with CD, 105 exhibited a clinical response, and 60 exhibited clinical remission after receiving infliximab treatment. Our radiomic model, comprising 20 relevant features, demonstrated excellent predictive performance. The radiomic nomogram for predicting clinical response showed good calibration and discrimination in the training cohort (area under the curve [AUC] 0.909, 95% confidence interval [CI] 0.840-0.978), the validation cohort (AUC 0.954, 95% CI 0.889-1), and the external cohort (AUC = 0.902, 95% CI 0.83-0.974). Accordingly, the nomogram was also suitable for predicting clinical remission. Decision curve analysis and clinical impact curves highlighted the clinical utility of our nomogram. Our radiomics nomogram is a noninvasive predictive tool constructed from radiomic features of the spleen. It also demonstrated good predictive accuracy in evaluating CD patients' response to infliximab treatment. Multicenter validation provided high-level evidence for its clinical application.

Validation of a clinical assay for botulinum neurotoxins through mass spectrometric detection.

Botulism is a paralytic disease due to the inhibition of acetylcholine exocytosis at the neuromuscular junction, which can be lethal if left untreated. Botulinum neurotoxins (BoNTs) are produced by some spore-forming Clostridium bacteria. The current confirmatory assay to test for BoNTs in clinical specimens is the gold-standard mouse bioassay. However, an Endopep-MS assay method has been developed to detect BoNTs in clinical samples using benchtop mass spectrometric detection. This work demonstrates the validation of the Endopep-MS method for clinical specimens with the intent of method distribution in public health laboratories. The Endopep-MS assay was validated by assessing the sensitivity, robustness, selectivity, specificity, and reproducibility. The limit of detection was found to be equivalent to or more sensitive than the mouse bioassay. Specificity studies determined no cross-reactivity between the different serotypes and no false positives from an exclusivity panel of culture supernatants of enteric disease organisms and non-toxigenic strains of Clostridium. Inter-serotype specificity testing with 19 BoNT subtypes was 100% concordant with the expected results, accurately determining the presence of the correct serotype and the absence of incorrect serotypes. Additionally, a panel of potential interfering substances was used to test selectivity. Finally, clinical studies included clinical specimen stability and reproducibility, which was found to be 99.9% from a multicenter evaluation study. The multicenter validation study also included a clinical validation study, which yielded a 99.4% correct determination rate. Use of the Endopep-MS method will improve the capacity and response time for laboratory confirmation of botulism in public health laboratories.

Realizing the Promise of Artificial Intelligence in Hepatocellular Carcinoma through Opportunities and Recommendations for Responsible Translation

This study aims to provide an overview of the current state-of-the-art applications of artificial intelligence (AI) and machine learning in the management of hepatocellular carcinoma (HCC), and to explore future directions for continued progress in this emerging field. This study is a comprehensive literature review that synthesizes recent findings and advancements in the application of AI and machine learning techniques across various aspects of HCC care, including screening and early detection, diagnosis and staging, prognostic modeling, treatment planning, interventional guidance, and monitoring of treatment response. The review draws upon a wide range of published research studies, focusing on the integration of AI and machine learning with diverse data sources, such as medical imaging, clinical data, genomics, and other multimodal information. The results demonstrate that AI-based systems have shown promise in improving the accuracy and efficiency of HCC screening, diagnosis, and tumor characterization compared to traditional methods. Machine learning models integrating clinical, imaging, and genomic data have outperformed conventional staging systems in predicting survival and recurrence risk. AI-based recommendation systems have the potential to optimize personalized therapy selection, while augmented reality techniques can guide interventional procedures in real-time. Moreover, longitudinal application of AI may enhance the assessment of treatment response and recurrence monitoring. Despite these promising findings, the review highlights the need for rigorous multicenter prospective validation studies, standardized multimodal datasets, and thoughtful consideration of ethical implications before widespread clinical implementation of AI technologies in HCC management.

Multicenter validation study of a treatment selection MAP for pancreatic neuroendocrine tumors.

Somatostatin analogs, molecular-targeted agents and cytotoxic anticancer agents are available as therapeutic agents for the systemic treatment of pancreatic neuroendocrine tumors, and we have developed a first-line treatment selection MAP to enable selection of the optimal treatment strategy for pancreatic neuroendocrine tumors. The purpose of this study was to validate the usefulness of the treatment selection MAP. Patients who had received systemic therapy for a pancreatic neuroendocrine tumor between January 2017 and December 2020 were compared according to whether they had been treated as recommended by the MAP (matched patients) or not (unmatched patients) to determine whether better outcomes were achieved by the matched patients. The primary endpoint was progression-free survival of the matched group and unmatched groups in the somatostatin analog, molecular-targeted agent and cytotoxic anticancer agents areas of the MAP. There were 41 (55%) MAP-matched patients in all areas among the 74 patients registered at seven hospitals. The MAP-matched rates were 100, 77 and 38% in the somatostatin analog area, molecular-targeted agent area and cytotoxic anticancer agents area, respectively. All of the unmatched patients had been selected for less intensive treatment. The median progression-free survival in the matched group and unmatched group in the molecular-targeted agent area of the MAP were 46.6 and 15.4months, respectively, and a multivariate analysis identified MAP-matched (hazard ratio 0.18 [95% confidence interval: 0.04-0.87], P=0.032) as the only significant independent favorable predictive factor. The usefulness of the MAP for treatment selection was validated in the molecular-targeted agent area of the MAP.

Evaluation of a machine-learning model based on laboratory parameters for the prediction of acute leukaemia subtypes: a multicentre model development and validation study in France

Acute leukaemias are life-threatening haematological cancers characterised by the infiltration of transformed immature haematopoietic cells in the blood and bone marrow. Prompt and accurate diagnosis of the three main acute leukaemia subtypes (ie acute lymphocytic leukaemia [ALL], acute myeloid leukaemia [AML], and acute promyelocytic leukaemia [APL]) is of utmost importance to guide initial treatment and prevent early mortality but requires cytological expertise that is not always available. We aimed to benchmark different machine-learning strategies using a custom variable selection algorithm to propose an extreme gradient boosting model to predict leukaemia subtypes on the basis of routine laboratory parameters. This multicentre model development and validation study was conducted with data from six independent French university hospital databases. Patients aged 18 years or older diagnosed with AML, APL, or ALL in any one of these six hospital databases between March 1, 2012, and Dec 31, 2021, were recruited. 22 routine parameters were collected at the time of initial disease evaluation; variables with more than 25% of missing values in two datasets were not used for model training, leading to the final inclusion of 19 parameters. The performances of the final model were evaluated on internal testing and external validation sets with area under the receiver operating characteristic curves (AUCs), and clinically relevant cutoffs were chosen to guide clinical decision making. The final tool, Artificial Intelligence Prediction of Acute Leukemia (AI-PAL), was developed from this model. 1410 patients diagnosed with AML, APL, or ALL were included. Data quality control showed few missing values for each cohort, with the exception of uric acid and lactate dehydrogenase for the cohort from Hôpital Cochin. 679 patients from Hôpital Lyon Sud and Centre Hospitalier Universitaire de Clermont-Ferrand were split into the training (n=477) and internal testing (n=202) sets. 731 patients from the four other cohorts were used for external validation. Overall AUCs across all validation cohorts were 0·97 (95% CI 0·95-0·99) for APL, 0·90 (0·83-0·97) for ALL, and 0·89 (0·82-0·95) for AML. Cutoffs were then established on the overall cohort of 1410 patients to guide clinical decisions. Confident cutoffs showed two (0·14%) wrong predictions for ALL, four (0·28%) wrong predictions for APL, and three (0·21%) wrong predictions for AML. Use of the overall cutoff greatly reduced the number of missing predictions; diagnosis was proposed for 1375 (97·5%) of 1410 patients for each category, with only a slight increase in wrong predictions. The final model evaluation across both the internal testing and external validation sets showed accuracy of 99·5% for ALL diagnosis, 98·8% for AML diagnosis, and 99·7% for APL diagnosis in the confident model and accuracy of 87·9% for ALL diagnosis, 86·3% for AML diagnosis, and 96·1% for APL diagnosis in the overall model. AI-PAL allowed for accurate diagnosis of the three main acute leukaemia subtypes. Based on ten simple laboratory parameters, its broad availability could help guide initial therapies in a context where cytological expertise is lacking, such as in low-income countries. None.

Prognostic and predictive value of tumor deposits in advanced signet ring cell colorectal cancer: SEER database analysis and multicenter validation.

Colorectal signet-ring cell carcinoma (SRCC) is a rare cancer with a bleak prognosis. The relationship between its clinicopathological features and survival remains incompletely elucidated. Tumor deposits (TD) have been utilized to guide the N staging in the 8th edition of American Joint Committee on Cancer (AJCC) staging manual, but their prognostic significance remains to be established in colorectal SRCC. The subjects of this study were patients with stage III/IV colorectal SRCC who underwent surgical treatment. The research comprised two cohorts: a training cohort and a validation cohort. The training cohort consisted of 631 qualified patients from the SEER database, while the validation cohort included 135 eligible patients from four independent hospitals in China. The study assessed the impact of TD on Cancer-Specific Survival (CSS) and Overall Survival (OS) using Kaplan-Meier survival curves and Cox regression models. Additionally, a prognostic nomogram model was constructed for further evaluation. In both cohorts, TD-positive patients were typically in the stage IV and exhibited the presence of perineural invasion (PNI) (P < 0.05). Compared to the TD-negative group, the TD-positive group showed significantly poorer CSS (the training cohort: HR, 1.87; 95% CI, 1.52-2.31; the validation cohort: HR, 2.43; 95% CI, 1.55-3.81; all P values < 0.001). This association was significant in stage III but not in stage IV. In the multivariate model, after adjusting for covariates, TD maintained an independent prognostic value (P < 0.05). A nomogram model including TD, N stage, T stage, TNM stage, CEA, and chemotherapy was constructed. Through internal and external validation, the model demonstrated good calibration and accuracy. Further survival curve analysis based on individual scores from the model showed good discrimination. TD positivity is an independent factor of poor prognosis in colorectal SRCC patients, and it is more effective to predict the prognosis of colorectal SRCC by building a model with TD and other clinically related variables.

Predicting contralateral extraprostatic extension in unilateral high-risk prostate cancer: a multicentric external validation study.

Accurate prediction of extraprostatic extension (EPE) is crucial for decision-making in radical prostatectomy (RP), especially in nerve-sparing strategies. Martini et al. introduced a three-tier algorithm for predicting contralateral EPE in unilateral high-risk prostate cancer (PCa). The aim of the study is to externally validate this model in a multicentric European cohort of patients. The data from 208 unilateral high-risk PCa patients diagnosed through magnetic resonance imaging (MRI)-targeted and systematic biopsies, treated with RP between January 2016 and November 2021 at eight referral centers were collected. The evaluation of model performance involved measures such as discrimination (AUC), calibration, and decision-curve analysis (DCA) following TRIPOD guidelines. In addition, a comparison was made with two established multivariable logistic regression models predicting the risk of side specific EPE for assessment purposes. Overall, 38%, 48%, and 14% of patients were categorized as low, intermediate, and high-risk groups according to Martini et al.'s model, respectively. At final pathology, EPE on the contralateral prostatic lobe occurred in 6.3%, 12%, and 34% of patients in the respective risk groups. The algorithm demonstrated acceptable discrimination (AUC 0.68), comparable to other multivariable logistic regression models (p = 0.3), adequate calibration and the highest net benefit in DCA. The limitations include the modest sample size, retrospective design, and lack of central revision. Our findings endorse the algorithm's commendable performance, supporting its utility in guiding treatment decisions for unilateral high-risk PCa patients.

A multi-center validation of the electronic health record admission source and discharge location fields against the clinical notes for identifying inpatients with long-term care facility exposure.

Residence or recent stay in a long-term care facility (LTCF) is an important risk factor for antibiotic-resistant bacterial colonization. However, absent dedicated intake questionnaires or resource-intensive chart review, ascertaining LTCF exposure in inpatients is challenging. We aimed to validate the electronic health record (EHR) admission and discharge location fields against the clinical notes for identifying LTCF-exposed inpatients. We conducted a retrospective study of 1020 randomly sampled adult admissions between 2016 and 2021 across 12 University of Maryland Medical System hospitals. Using study-developed guidelines, we categorized the following data for LTCF exposure: each admission’s history & physical (H&P) note, each admission’s EHR-extracted “Admission Source,” and (3) the EHR-extracted admission and discharge locations for previous admissions (≤90 days). We estimated sensitivities, with 95% CIs, of H&P notes and of EHR admission/discharge location fields for detecting “current” and “any recent” (≤90 days, including current) LTCF exposure. For detecting current LTCF exposure, the sensitivity of the index admission’s EHR-extracted “Admission Source” was 46% (95% CI: 35%–58%) and of the H&P note was 92% (83%–97%). For detecting any recent LTCF exposure, the sensitivity of “Admission Source” across the index and previous admissions was 32% (24%–41%), “Discharge Location” across previous admission(s) was 57% (47%–66%), and of the H&P note was 68% (59%–76%). The combined sensitivity of admission source and discharge location for detecting any recent LTCF exposure was 76% (67%–83%). The EHR-obtained admission source and discharge location fields identified 76% of LTCF-exposed patients compared to chart review but disproportionately missed currently exposed patients.

Identifying risk for status epilepticus with the ADAN scale: a prospective multicenter validation study.

Status epilepticus (SE) is a serious event associated with high mortality. This study aims to validate the recently developed ADAN (Abnormal speech, ocular Deviation, Automatisms, and Number of motor epileptic seizures) scale for detecting high risk for SE. Prospective, multicenter, observational study in adults with suspected epileptic seizures. Consecutive recruitment took place over a 27-month period in 4 hospital emergency departments (EDs). The main endpoint was the proportion of patients with criteria for SE based on the collection and analysis of clinical characteristics and the ADAN scale criteria on arrival at the ED. Of the 527 patients recruited, 203 (38.5%) fulfilled the criteria that predicted SE. Multiple regression analysis demonstrated that the 4 ADAN criteria were the only variables independently associated with a final diagnosis of SE (P .001). The predictive power of the scale was 90.9% (95% CI, 88.4%-93.4%) for a final SE diagnosis. We established 3 risk groups based on ADAN scores: low (score, 0-1: 8.7%), moderate (2, 46.6%), and high (> 2, 92.6%). A cut point of more than 1 had a sensitivity of 88.2% for predicting SE, specificity of 77.8%, positive predictive value of 71.3%, and negative predictive value of 91.3%. The ADAN scale is a prospectively validated, simple clinical tool for identifying patients in the ED who are at high risk for SE.

313 Development and Validation of a Scoring System for Hemorrhage Risk in Brain Arteriovenous Malformations

INTRODUCTION: The dilemma between natural rupture risk and adverse outcomes of intervention is of major concern for patients with unruptured arteriovenous malformations (AVMs). The existing risk score for AVM rupture includes factors that are controversial and lacks prospective validation. METHODS: This prognostic study developed a prediction model derived from a single-center cohort (derivation cohort) and validated in a multicenter external validation cohort and a conservative treatment validation cohort. Patients were recruited from a nationwide multicenter prospective collaboration registry in China. A total of 4135 patients were enrolled. 3962 patients were included (3585 in the derivation cohort and 377 in the multicenter external validation cohort); 1028 patients from the derivation cohort who had time-to-event data and prerupture imaging results were included in the conservative treatment validation cohort. RESULTS: Four risk factors were used to develop the scoring system: ventricular system involvement, venous aneurysm, deep location, and exclusively deep drainage (VALE). The VALE scoring system performed well in all 3 cohorts, with areas under the receiver operating characteristic curve of 0.77 (95% CI, 0.75-0.78) in the derivation cohort, 0.85 (95% CI, 0.81-0.89) in the multicenter external validation cohort, and 0.73 (95% CI, 0.65-0.81) in the conservative treatment validation cohort. The 10-year hemorrhage-free rate was 95.5% (95% CI, 87.1%-100%) in the low-risk group, 92.8% (95% CI, 88.8%-97.0%) in the moderate-risk group, and 75.8% (95% CI, 65.1%-88.3%) in the high-risk group; the model discrimination was significant when comparing these rates between the high-risk group and the low- and moderate-risk groups (p &lt; .001 for both comparisons). CONCLUSIONS: In this prognostic study, the VALE scoring system was developed to distinguish rupture risk among patients with AVMs. The stratification of unruptured AVMs may enable patients with low risk of rupture to avoid unnecessary interventions.

Prediction of histopathologic grades of bladder cancer with radiomics based on MRI: Comparison with traditional MRI

PurposeTo compare biparametric magnetic resonance imaging (bp-MRI) radiomics signatures and traditional MRI model for the preoperative prediction of bladder cancer (BCa) grade. Materials and methodsThis retrospective study included 255 consecutive patients with pathologically confirmed 113 low-grade and 142 high-grade BCa. The traditional MRI nomogram model was developed using univariate and multivariate logistic regression by the mean apparent diffusion coefficient (ADC), vesical imaging reporting and data system, tumor size, and the number of tumors. Volumes of interest were manually drawn on T2-weighted imaging (T2WI) and ADC maps by 2 radiologists. Using one-way analysis of variance, correlation, and least absolute shrinkage and selection operator methods to select features. Then, a logistic regression classifier was used to develop the radiomics signatures. Receiver operating characteristic (ROC) analysis was used to compare the diagnostic abilities of the radiomics and traditional MRI models by the DeLong test. Finally, decision curve analysis was performed by estimating the clinical usefulness of the 2 models. ResultsThe area under the ROC curves (AUCs) of the traditional MRI model were 0.841 in the training cohort and 0.806 in the validation cohort. The AUCs of the 3 groups of radiomics model [ADC, T2WI, bp-MRI (ADC and T2WI)] were 0.888, 0.875, and 0.899 in the training cohort and 0.863, 0.805, and 0.867 in the validation cohort, respectively. The combined radiomics model achieved higher AUCs than the traditional MRI model. decision curve analysis indicated that the radiomics model had higher net benefits than the traditional MRI model. ConclusionThe bp-MRI radiomics model may help distinguish high-grade and low-grade BCa and outperforming the traditional MRI model. Multicenter validation is needed to acquire high-level evidence for its clinical application.

A commentary on "Predicting recurrence and survival in patients with non-metastatic renal-cell carcinoma after nephrectomy: a prospective population-based study with multicenter validation".

A commentary on "Predicting recurrence and survival in patients with non-metastatic renal-cell carcinoma after nephrectomy: a prospective population-based study with multicenter validation".

The Key Role of Tumor Budding in Predicting the Status of Lymph Node Involvement in Early Gastric Cancer Patients: A Clinical Multicenter Validation in China.

Accurate preoperative prediction of lymph node (LN) involvement is essential for the management of early gastric cancer (EGC). Our objective was to formulate a potent nomogram for predicting LN involvement in EGC by leveraging an innovative predictor of tumor budding. We assembled a cohort of EGC patients who underwent radical surgery at two tertiary cancer centers. Tumor budding was stratified by using an optimal cutoff value and integrated with other clinicopathological variables to ascertain the risk factors associated with LN involvement. A nomogram was developed and its predictive performance was assessed by using receiver operating characteristic (ROC) curves and calibration plots. In addition, we conducted decision curve analysis to evaluate its clinical utility. Finally, an external validation was conducted by using an independent cohort. Finally, 307 eligible patients (215 in the primary cohort and 92 in the validation cohort) were included. Tumor budding, categorized by a count of two, exhibited a robust association with LN involvement (OR 14.12, p = 0.012). Other significant risk factors include lymphovascular invasion, depth of tumor invasion, ulceration, and tumor differentiation. Notably, the nomogram demonstrated exceptional discriminative power (area under the ROC curve, 0.872 in the primary cohort and 0.885 in the validation cohort) and precise predictive capabilities. Furthermore, the nomogram showed notable clinical applicability through decision curve analysis, particularly in endoscopic curability C-2, by mitigating the risk of overtreatment. Tumor budding is a robust predictor of LN involvement in EGC. The incorporation of tumor budding into a nomogram is an effective strategy, thereby informing and enhancing clinical decision-making.

Multicentre validation of CT grey-level co-occurrence matrix features for overall survival in primary oesophageal adenocarcinoma

BackgroundPersonalising management of primary oesophageal adenocarcinoma requires better risk stratification. Lack of independent validation of proposed imaging biomarkers has hampered clinical translation. We aimed to prospectively validate previously identified prognostic grey-level co-occurrence matrix (GLCM) CT features for 3-year overall survival.MethodsFollowing ethical approval, clinical and contrast-enhanced CT data were acquired from participants from five institutions. Data from three institutions were used for training and two for testing. Survival classifiers were modelled on prespecified variables (‘Clinical’ model: age, clinical T-stage, clinical N-stage; ‘ClinVol’ model: clinical features + CT tumour volume; ‘ClinRad’ model: ClinVol features + GLCM_Correlation and GLCM_Contrast). To reflect current clinical practice, baseline stage was also modelled as a univariate predictor (‘Stage’). Discrimination was assessed by area under the receiver operating curve (AUC) analysis; calibration by Brier scores; and clinical relevance by thresholding risk scores to achieve 90% sensitivity for 3-year mortality.ResultsA total of 162 participants were included (144 male; median 67 years [IQR 59, 72]; training, 95 participants; testing, 67 participants). Median survival was 998 days [IQR 486, 1594]. The ClinRad model yielded the greatest test discrimination (AUC, 0.68 [95% CI 0.54, 0.81]) that outperformed Stage (ΔAUC, 0.12 [95% CI 0.01, 0.23]; p = .04). The Clinical and ClinVol models yielded comparable test discrimination (AUC, 0.66 [95% CI 0.51, 0.80] vs. 0.65 [95% CI 0.50, 0.79]; p > .05). Test sensitivity of 90% was achieved by ClinRad and Stage models only.ConclusionsCompared to Stage, multivariable models of prespecified clinical and radiomic variables yielded improved prediction of 3-year overall survival.Clinical relevance statementPreviously identified radiomic features are prognostic but may not substantially improve risk stratification on their own.Key Points• Better risk stratification is needed in primary oesophageal cancer to personalise management.• Previously identified CT features—GLCM_Correlation and GLCM_Contrast—contain incremental prognostic information to age and clinical stage.• Compared to staging, multivariable clinicoradiomic models improve discrimination of 3-year overall survival.

A novel preoperative score to predict severe acute cholecystitis.

In a large multicenter trial, The Parkland Grading Scale (PGS) for acute cholecystitis outperformed other grading scales and has a positive correlation with complications but is limited in its inability to preoperatively predict high-grade cholecystitis. We sought to identify preoperative variables predictive of high-grade cholecystitis (PGS 4 or 5). In a six-month period, patients undergoing cholecystectomy at a single institution with prospectively graded PGS were analyzed. Stepwise logistic regression models were constructed to predict high-grade cholecystitis. The relative weight of the variables was used to derive a novel score, the Severe Acute Cholecystitis Score (SACS). This score was compared with the Emergency Surgery Acuity Score(ESS), American Association for the Surgery of Trauma (AAST) preoperative score and Tokyo Guidelines (TG) for their ability to predict high-grade cholecystitis. Severe Acute Cholecystitis Score was then validated using the database from the AAST multicenter validation of the grading scale for acute cholecystitis. Of the 575 patients that underwent cholecystectomy, 172 (29.9%) were classified as high-grade. The stepwise logistic regression modeling identified seven independent predictors of high-grade cholecystitis. From these variables, the SACS was derived. Scores ranged from 0 to 9 points with a C statistic of 0.76, outperforming the ESS ( C statistic of 0.60), AAST (0.53), and TG (0.70) ( p < 0.001). Using a cutoff of 4 or more on the SACS correctly identifies 76.2% of cases with a specificity of 91.3% and a sensitivity of 40.7%. In the multicenter database, there were 464 patients with a prospectively collected PGS. The C statistic for SACS was 0.74. Using the same cutoff of 4, SACS correctly identifies 71.6% of cases with a specificity of 83.8% and a sensitivity of 52.2%. The Severe Acute Cholecystitis Score can preoperatively predict high-grade cholecystitis and may be useful for counseling patients and assisting in surgical decision making. Diagnostic Test/Criteria; Level III.