Larson, Ellen L.

BACKGROUND: The Ventral Rectopexy International Expert Panel recently published a consensus update on ventral rectopexy. The ability of large language models to synthesize the literature on ventral rectopexy without an explicit knowledge base was studied before the publication of the consensus. OBJECTIVE: To compare the responses and citations of different large language models on ventral rectopexy using the expert panel consensus as the reference standard. DESIGN: ChatGPT-4o, Gemini 1.5 Pro, and OpenEvidence were compared on content appropriateness (1 = inappropriate to 5 = appropriate), readability (Flesch reading ease), response length, citation fabrication, and citation quality per Oxford Levels of Evidence. The most content-appropriate chatbot response selected by the expert panel was de-identified and presented alongside the consensus text to 15 colorectal surgeons who attempted to identify the chatbot-generated text. SETTINGS: Questions were submitted from September 18 to 19, 2024. Analysis was performed before the online publication of the consensus on January 30, 2025. MAIN OUTCOME MEASURES: The main outcome measures included content appropriateness, fabricated citation rate, citation quality, and the accuracy of identifying human versus chatbot text by colorectal surgeons. RESULTS: OpenEvidence ranked highest for content appropriateness (mean 3.5/5), above Gemini (3.0/5) and ChatGPT (2.8/5; p < 0.001). ChatGPT was most verbose with the highest readability scores ( p = 0.021). ChatGPT fabricated 53% of citations; Gemini fabricated 12%; OpenEvidence fabricated 0% ( p < 0.001). All OpenEvidence citations were peer reviewed, with 40 of 117 (34%) citing level I to III studies compared with 46 of 94 (49%) of the references cited in the consensus ( p = 0.043). Chatbot-generated responses were identified with 28 of 51 (55%) accuracy. LIMITATIONS: Reproducibility may be affected due to the nature of large language models and the availability of the consensus after publication. CONCLUSIONS: OpenEvidence outperformed Gemini 1.5 Pro and ChatGPT-4o in content appropriateness and peer-reviewed citation quantity and quality. Chatbot-generated text was indistinguishable from expert-authored consensus to subject matter experts. Large language models, as an early-stage research tool, may be viable for future consensus working groups, provided transparent disclosure and rigorous oversight. See Video Abstract . ¿PUEDEN LOS MODELOS LINGÜÍSTICOS DE GRAN TAMAÑO SER UNA HERRAMIENTA VIABLE PARA LOS GRUPOS DE TRABAJO DE CONSENSO? LA EXPERIENCIA DEL GRUPO DE CONSENSO DE EXPERTOS EN RECTOPEXIA VENTRAL ANTECEDENTES: El Panel Internacional de Expertos en Rectopexia Ventral publicó recientemente una actualización consensuada sobre la rectopexia ventral. Antes de la publicación del consenso, se estudió la capacidad de los grandes modelos lingüísticos para sintetizar la bibliografía sobre la rectopexia ventral sin una base de conocimientos explícita. Objetivo: Comparar las respuestas y citas de diferentes grandes modelos lingüísticos sobre la rectopexia ventral utilizando el consenso del panel de expertos como estándar de referencia. DISEÑO: Se compararon ChatGPT-4o, Gemini 1.5 Pro y OpenEvidence en cuanto a la idoneidad del contenido (1 = inadecuado a 5 = adecuado), la legibilidad (índice de legibilidad de Flesch), la longitud de la respuesta, la fabricación de citas y la calidad de las citas según los niveles de evidencia de Oxford. La respuesta del chatbot más adecuada en cuanto al contenido seleccionada por el panel de expertos se anonimizó y se presentó junto con el texto del consenso a 15 cirujanos colorrectales que intentaron identificar el texto generado por el chatbot. ENTORNO: Las preguntas se enviaron los días 18 y 19 de septiembre de 2024. El análisis se realizó antes de la publicación del consenso en línea el 30 de enero de 2025. PRINCIPALES MEDIDAS DE RESULTADOS: Adecuación del contenido, tasa de citas inventadas, calidad de las citas, precisión en la identificación del texto humano frente al del chatbot por parte de los cirujanos colorrectales. RESULTADOS: OpenEvidence obtuvo la puntuación más alta en cuanto a la adecuación del contenido (media de 3,5/5), por encima de Gemini (3,0/5) y ChatGPT (2,8/5) (p<0,001). ChatGPT fue el más prolijo, con las puntuaciones más altas en legibilidad (p = 0,021). ChatGPT inventó el 53% de las citas; Gemini inventó el 12 %; OpenEvidence inventó el 0 % (p < 0,001). Todas las citas de OpenEvidence fueron revisadas por pares, con 40/117 (34 %) que citaban estudios de nivel I-III frente a 46/94 (49 %) de las referencias citadas en el consenso (p = 0,043). Las respuestas generadas por el chatbot se identificaron con una precisión del 28/51 (55 %). LIMITACIONES: Reproducibilidad debido a la naturaleza de los modelos de lenguaje grandes y a la disponibilidad del consenso tras la publicación. CONCLUSIONES: OpenEvidence superó a Gemini 1.5 Pro y ChatGPT-4o en cuanto a la idoneidad del contenido y la cantidad y calidad de las citas revisadas por pares. El texto generado por el chatbot era indistinguible del consenso redactado por expertos para los expertos en la materia. Los modelos de lenguaje grandes como herramienta de investigación en fase inicial pueden ser viables para futuros grupos de trabajo de consenso, siempre que se garantice una divulgación transparente y una supervisión rigurosa. ( AI-generated translation )

Background and Aims: Cholangiocarcinoma (CCA) is one of the most lethal cancers, characterized by molecular heterogeneity and treatment resistance. To uncover new biological signals and therapeutic opportunities, we employed multiomic characterization combined with machine learning. Approach and Results: We profiled all anatomical CCA subtypes using whole exome sequencing, mRNA sequencing, and proteome/phosphoproteome analysis. Integrative dimensional reduction revealed RNA, protein, and phosphoprotein features driving tumor heterogeneity, enabling clustering. Machine learning algorithms identified molecular features for each cluster and mapped external datasets and patient-derived xenograft (PDX) models onto these clusters. Kinase enrichment analysis highlighted targetable kinases active in each cluster. In vivo validation was performed in cluster-specific PDX models using the selective TNK1 inhibitor, TP-5801. We identified 3 molecular clusters with distinct pathway characterization: immunomodulatory (cluster 1), metabolic (cluster 2), and gene regulation/cellular fate (cluster 3). Cluster assignment was independent of anatomic subtype but correlated with overall survival following curative-intent resection. We also identified multiomic features and pathways linked to overall survival and lymph node metastases, crucial for patient treatment selection. Kinase enrichment analysis pinpointed TNK1 as a highly active kinase in the metabolic cluster. Treatment with TP-5801 significantly reduced tumor growth in a metabolic PDX model, but not in models representing the other clusters. Combining internal data with publicly available datasets, we identified the immunomodulatory cluster as most responsive to gemcitabine/cisplatin therapy, confirmed in vivo using cluster-specific PDX models. Conclusions: Integrated multiomic characterization provides translational insights by defining unique molecular subtypes associated both with therapeutic response and overall clinical outcomes. This approach identified TNK1 as a previously unrecognized therapeutic target in a defined subset of CCA tumors.

Abstract Background: The standard first line therapy for cholangiocarcinoma (CCA), gemcitabine and cisplatin +/- immunotherapy, only induces disease control in approximately 75% of patients (KEYNOTE-966 Trial). While several candidate single genes that predict gemcitabine and cisplatin resistance have been proposed, the overall molecular landscape that leads to resistance is not well understood. Predicting response based on molecular characteristics would benefit the clinical decision-making process when selecting appropriate chemotherapy regimens for CCA patients. Methods: Fifty-five Mayo Clinic CCA patients who underwent whole transcriptome RNA sequencing were included. Forty-eight patients had TempusXR sequencing and 7 patients used institutional labs. Normalized transcript-level data was converted to KEGG pathway-level data using overrepresentation ranking. Patient oncology notes were manually reviewed for response to gemcitabine plus cisplatin therapy, with or without concurrent pembrolizumab or durvalumab. Oncologists used a mixture of formal RECIST criteria and expert opinion to determine whether the response to gemcitabine and cisplatin +/- immunotherapy was progressive disease (PD), stable disease (SD), partial response (PR), or complete response (CR). The binary outcome of PD vs SD/PR/CR based on KEGG pathway overrepresentation was modelled using a Tabular Prior-data Fitted Network (TabPFN). Validation was performed using leave-one-out cross-validation. External model testing was performed on RNA sequencing data from 11 patient-derived xenografts that were treated with gemcitabine and cisplatin and underwent ultrasound measurements of tumor size. Results: There were 34 patients with SD/PR/CR and 21 patients with PD as the best response while receiving gemcitabine and cisplatin +/- immunotherapy. TabPFN cross-validation accuracy was 0.73, recall 0.94, precision 0.71, and F1 score 0.81. External validation accuracy was 0.70, with one xenograft exhibiting mixed response. In contrast, the best trained model using a 14-gene subset previously published for gemcitabine resistance prediction in pancreatic adenocarcinoma had accuracy 0.58. Conclusions: Machine learning models like TabPFN can capture relationships in high-dimensional transcriptomic information in novel ways that enhance our ability to predict chemotherapy response. Integrating machine learning to understand the complex molecular landscape of CCA will be a valuable tool for therapeutic regimen selection. Citation Format: Ellen L. Larson, Erik Jessen, Dong-Gi Mun, Amro Abdelrahman, Jennifer Tomlinson, Danielle Carlson, Hojjat Salehinejad, Caitlin Conboy, Rory Smoot. Machine learning models of RNA expression landscapes help predict overall tumor response to chemotherapy in cholangiocarcinoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr A020.

Background: Cholangiocarcinoma is a molecularly heterogenous cancer arising from the biliary epithelium. Individual omic techniques have identified clinically relevant, targetable subtypes but most tumors have no targetable alterations based on exome or transcriptomic evaluation. Integrated multiomics approaches incorporating transcriptomic, proteomic, and phosphoproteomic characterization provide deeper understanding and identify non-mutated, activated pathways that can be targeted. To optimize the potential clinical utility of these approaches, minimal classifier features must be defined. Methods: Approximately 60,000 RNA, protein, and phosphoprotein features were extracted from 35 cholangiocarcinomas treated at the primary US study site. 206 patients treated and profiled at an institution in China were also included. Using the Multiomics Factor Analysis clustering method, patients were sorted into 3 distinct molecular profiles. Important molecular features were selected using competing feature selection (FS) methods: Boruta, RreliefF, variable selection using random forests (RF), recursive factor elimination (RFE), median decrease accuracy in RF, and Concrete Autoencoders. Machine learning classification prediction models including Extreme Gradient Boosting, gradient-boosted trees, multinomial logit, and support vector machine (SVM) were tested on each FS. Each combination of FS and predictive model was tested with 10 replicates of 10-fold cross-validation of a randomly-selected balanced test set and an oversampled, balanced training subset. Accuracy, F1 score, AUC, precision, and recall were assessed for each FS and predictive model combination. A custom function of physical assay cost and classification accuracy was optimized to find the best assay. KEGG pathway overrepresentation analysis was used to analyze feature subsets. Results: A consistent set of 50 important RNA and protein features was identified, including previously reported prognostic factors, like SpryD4 protein expression, as well as previously unreported pathway surrogates, such as Slc2A2 or Pipox protein. Pathway overrepresentation analysis found insulin signaling and resistance and amino acid biosynthesis/metabolism pathways were overrepresented in the discriminative feature set. The gradient-boosted trees algorithm combined with the RFE FS method had the best performance, with a multiclass macro average (±sd) cross-validation accuracy 89% (±6%), F1 score 89% (±8%), AUC 0.98 (±0.02), precision 90% (±10%), recall 89% (±6%). Conclusions: Feature selection methods and predictive machine learning models work synchronously to develop accurate molecular subtype assays based on just 50 protein and RNA levels. This strategy has the potential to make molecular subtyping for cholangiocarcinoma and other molecularly complex cancers fast and widely available. Citation Format: Ellen Larson, Erik Jessen, Dong-Gi Mun, Jennifer Tomlinson, Amro Abdelrahman, Danielle Carlson, Hojjat Salehinejad, Rory Smoot. Feature selection and machine learning strategies optimize an affordable molecular assay for cholangiocarcinoma subtype [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1107.

Abstract Background: Cholangiocarcinoma (CCA) is a diverse type of cancer. For several years, platinum-based chemotherapy (Gemcitabine and Cisplatin) has been the standard treatment and serves as the foundation for current immunotherapy combinations. The typical response rates are generally between 20% and 30%. In this study, we present the results of preclinical efficacy trials conducted on CCA patient-derived xenograft (PDX) models, correlating them with the baseline multiomics profiling of these models to identify molecular characteristics linked to treatment response. Methods: Tumor samples from twenty-eight CCA PDX were used to isolate RNA and protein to be included in multiomics analysis (Phosphoproteomics, Proteomics, and Transcriptomics). A subset of PDX models were treated with various doses of Gemcitabine and Cisplatin. The Median Efficacy Index (MEI) was calculated as the ratio of the difference in median of delta in tumor volume at the last day of the study between the treatment and the control arms divided by the median of delta in tumor volume of the control at the last day of the treatment (positive= growth, negative=inhibition, and 0=no efficacy). A Hierarchical All-Against-All algorithm was utilized to compare multiomics features to the MEI and pathway analysis completed for identified molecular features. Pathway enrichment was calculated using the KEGG database, and random-forest variable selection chose the pathways most influential in predicting sensitivity or resistance. Results: After conducting preclinical efficacy trials, the 11 tested CCA PDX models were classified into sensitive, moderately sensitive, and resistant models based on the MEI on each treatment arm. The top up-regulated genes were identified for Gemcitabine response (SRPK2, GLK2, and RIOK2) and Cisplatin response (NTRK1, PRPF4B, and SRPK2). The top up-regulated genes related to Gemcitabine resistance were NTRK1, PIP5K1C, and CSNK2A2, while those related to Cisplatin resistance were SRPK1, PRPF4B, and CSNK2A1. Downregulation of the anti-folate resistance pathway is a strong predictor of gemcitabine sensitivity, and the lysine degradation pathway is a strong predictor of cisplatin sensitivity. Conclusion: Multiomics signatures of cholangiocarcinoma can forecast responses to standard chemotherapy in preclinical models. New markers of resistance to Gemcitabine and Cisplatin were discovered, linked to NTRK1, SRPK1, and SRPK2. These resistance profiles will be tested using suitable targeted therapies on the resistant CCA patient-derived xenograft (PDX) models. Citation Format: Amro M. Abdelrahman, Danielle M. Carlson, Ellen L. Larson, Erik Jessen, Dong-Gi Mun, Isaac Lynch, Alessandro Fogliati, Stella Konadu Adjei Antwi, Aushinie Abeynayake, Akhilesh Pandey, Mark J. Truty, Gregory J. Gores, Rory L. Smoot. Baseline multiomics and chemotherapy efficacy in preclinical trials using patient-derived xenograft models of cholangiocarcinoma can reveal distinct signatures of chemotherapy response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6762.

The mammalian liver's regenerative ability has led researchers to engineer animals as incubators for expansion of human hepatocytes. The expansion properties of human hepatocytes in immunodeficient mice are well known. However, little has been reported about larger animals that are more scalable and practical for clinical purposes. Therefore, we engineered immunodeficient swine to support expansion of human hepatocytes and identify barriers to their clinical application. Immunodeficient swine were engineered by knockout of the recombinase-activating gene 2 ( RAG2 ) and fumarylacetoacetate hydrolase (FAH). Immature human hepatocytes (ihHCs) were injected into fetal swine by intrauterine cell transplantation (IUCT) at day 40 of gestation. Human albumin was measured as a marker of engraftment. Cytotoxicity against ihHCs was measured in transplanted piglets and control swine. We initially detected higher levels of human albumin in cord blood of newborn FAH/ RAG2 -deficient (FR) pigs compared with immunocompetent controls (196.26 ng/dL vs. 39.29 ng/dL, p = 0.008), indicating successful engraftment of ihHCs after IUCT and adaptive immunity in the fetus. Although rare hepatocytes staining positive for human albumin were observed, levels of human albumin did not rise after birth, but declined, suggesting rejection of xenografted ihHCs. Cytotoxicity against ihHCs increased after birth by 3.8% (95% CI: [2.1%–5.4%], p < 0.001) and inversely correlated with declining levels of human albumin ( p = 2.1 × 10 −5 , R 2 = 0.17). Circulating numbers of T cells and B cells were negligible in FR pigs. However, circulating natural killer (NK) cells exerted cytotoxicity against ihHCs. NK cell activity was lower in immunodeficient piglets after IUCT than in naive controls (30.4% vs. 40.1%, p = 0.011, 95% CI for difference [2.7%–16.7%]). In conclusion, ihHCs were successfully engrafted in FR swine after IUCT. NK cells were a significant barrier to expansion of hepatocytes. New approaches are needed to overcome this hurdle and allow large-scale expansion of human hepatocytes in immunodeficient swine. Impact statement There is currently a need for robust expansion of human hepatocytes. We describe an immunodeficient swine model into which we engrafted immature human hepatocytes (ihHCs). We identified the mechanism of the eventual graft rejection by the intact NK cell population, which has not been previously shown to have a significant role in xenograft rejection. By both improving engraftment and reducing NK cell-mediated cytotoxicity toward the graft through intrauterine cell transfer, we confirmed the presence of residual adaptive immunity in this model of immunodeficiency and the ability to induce hyposensitization in the NK cell population by taking advantage of the fetal microenvironment.

Cell therapies using immune cells or non-parenchymal cells of the liver have emerged as potential treatments to facilitate immunosuppression withdrawal and to induce operational tolerance in liver transplant (LT) recipients. Recent pre-clinical and clinical trials of cellular therapies including regulatory T cells, regulatory dendritic cells, and mesenchymal cells have shown promising results. Here we briefly summarize current concepts of cellular therapy for induction of operational tolerance in LT recipients.

Abstract Optimal treatment of patients with various types of liver tumors or certain liver diseases frequently demands major liver resection, which remains a clinical challenge especially in children. Eighty seven consecutive pediatric liver resections including 51 (59%) major resections (resection of 3 or more hepatic segments) and 36 (41%) minor resections (resection of 1 or 2 segments) were analyzed. All patients were treated between January 2010 and March 2018. Perioperative outcomes were compared between major and minor hepatic resections. The male to female ratio was 1.72:1. The median age at operation was 20 months (range, 0.33–150 months). There was no significant difference in demographics including age, weight, ASA class, and underlying pathology. The surgical management included functional assessment of the future liver remnant, critical perioperative management, enhanced understanding of hepatic segmental anatomy, and bleeding control, as well as refined surgical techniques. The median estimated blood loss was 40 ml in the minor liver resection group, and 90 ml in major liver resection group (P < .001). Children undergoing major liver resection had a significantly longer median operative time (80 vs 140 minutes), anesthesia time (140 vs 205 minutes), as well as higher median intraoperative total fluid input (255 vs 450 ml) (P < .001 for all). Fourteen (16.1%) patients had postoperative complications. By Clavien-Dindo classification, there were 8 grade I, 4 grade II, and 2 grade III-a complications. There were no significant differences in complication rates between groups (P = .902). Time to clear liquid diet (P = .381) and general diet (P = .473) was not significantly different. There was no difference in hospital length of stay (7 vs 7 days, P = .450). There were no 90-day readmissions or mortalities. Major liver resection in children is not associated with an increased incidence of postoperative complications or prolonged postoperative hospital stay compared to minor liver resection. Techniques employed in this study offered good perioperative outcomes for children undergoing major liver resections.