PSPierre-Antoine Soret

Pierre-Antoine Soret

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Pierre-Antoine Soret has a total of 55 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Use of Large Language Models for Automated Short-Answer Question Grading in Second-Cycle Medical Education: Performance and Consistency Evaluation Study

    by Sabrina Sidali, Pierre-Antoine Soret, Mathilde Wagner, Thibault Voron, Quentin Vanderbecq

    Background: Short-answer questions (SAQs) effectively assess higher-order reasoning skills in medical education but face challenges in scalability due to time-intensive manual grading and concerns about interrater reliability. Recent advances in large language models (LLMs) may offer a solution through automated assessment. Objective: This study evaluated the performance and consistency of LLMs—ChatGPT, Claude Sonnet 4, and Mistral Pro—compared with those of human graders for SAQ grading in hepatology and gastroenterology. Methods: We retrospectively analyzed SAQs from hepatology and gastroenterology examinations at Sorbonne University, Paris, France (2023-2025). Three human evaluators (2 hepatogastroenterologists and 1 radiologist) independently graded all responses twice, with the majority consensus score serving as the reference standard. Six LLMs were evaluated: ChatGPT (GPT-4o, GPT-4.5, GPT-4o3, and GPT-5), Claude Sonnet 4, and Mistral Pro. Each LLM graded all responses twice using standardized prompts, including questions, model answers, grading guidelines, and scoring rubrics. Agreement was assessed using quadratic weighted kappa (QWK), Cohen κ, the intraclass correlation coefficient (ICC), and the mean absolute error (MAE) for interhuman, intrahuman, LLM-versus-human, and intramodel comparisons. Results: A total of 902 student responses from 12 SAQs were included and analyzed. Interhuman grader agreement was almost perfect (QWK 0.90-0.92; ICC 0.90-0.92; MAE 0.05-0.07). Claude Sonnet 4 and GPT-4o3 demonstrated near-human–level performance when compared with the majority vote of human evaluators (Claude Sonnet 4: QWK 0.88-0.90; ICC 0.88-0.90; MAE 0.06-0.07 and GPT-4o3: QWK 0.82-0.89; ICC 0.82-0.89; MAE 0.07-0.09). GPT-5 showed substantial agreement (QWK 0.77-0.79). Mistral Pro and older ChatGPT versions performed poorly (QWK <0.63). Intramodel consistency was the highest for Claude Sonnet 4 (QWK 0.93), approaching the best human intrarater reliability. Agreement did not correlate with prompt design choices (number of examples or minimum word count). Conclusions: Advanced LLMs, particularly Claude Sonnet 4 and GPT-4o3, demonstrated near-human–level grading performance, supporting their potential use as assistive tools in SAQ grading. However, performance varied by model, and agreement alone does not ensure educational validity. Until broader multicenter validation is conducted across diverse domains, stakes, and question formats—along with safeguards for bias, privacy, and model drift—LLMs should augment rather than replace human faculty graders.


  • Open Access

    Orchestrated cell death pathways in cholangiopathies

    by Maria Eugenia Guicciardi, Valeria Pistorio, Pierre-Antoine Soret, Virginie Steunou, Gregory J Gores, Jérémie Gautheron

    Cholangiopathies comprise a heterogeneous group of chronic liver diseases characterised by progressive bile duct injury, ductular reaction and fibrotic remodelling. Despite distinct aetiologies, these disorders share common cellular stressors that ultimately converge on cholangiocyte dysfunction and loss. In this context, programmed cell death has emerged as a central mechanism linking epithelial stress, immune activation and the development of biliary fibrosis. A range of regulated cell death pathways, including apoptosis, necroptosis, pyroptosis, ferroptosis and autophagy-dependent cell death, can be triggered in cholangiocytes in a disease-dependent and context-dependent manner. Growing evidence suggests that these pathways rarely act in isolation. Instead, they intersect through shared molecular nodes, forming integrated cell death programmes such as PANoptosis. Importantly, it is becoming clear that not only overt cell death but also sublethal activation of death-related signalling can perpetuate cholangiocyte dysfunction and fuel chronic disease progression. Current therapeutic strategies, such as ursodeoxycholic acid and peroxisome proliferator-activated receptor agonists, mainly aim to modify bile composition and improve biochemical parameters. However, they do not directly target the molecular networks governing cholangiocyte death, which may partly explain their limited impact on fibrosis and long-term outcomes. In this review, we synthesise current knowledge on regulated cell death mechanisms in cholangiopathies, highlight key conceptual and technical challenges that complicate their interpretation and therapeutic targeting, and discuss how sublethal death signalling may connect with senescence to sustain cholangiopathy progression and shape therapeutic perspectives.

  • Étude de l’inhibition pharmacologique de RIPK1 dans la cholangite sclérosante primitive

    by Pierre-Antoine Soret

    La cholangite sclérosante primitive (CSP) est une maladie cholestatique chronique caractérisée par des sténoses fibro-inflammatoires des canaux biliaires. Aucun traitement efficace n'est disponible pour prévenir la survenue des complications hépatiques de la CSP. La mort des cholangiocytes contribue au développement de la réaction inflammatoire et de la fibrose hépatique dans la CSP, et constitue donc une cible thérapeutique d'intérêt. La nécroptose, une voie de mort cellulaire programmée proinflammatoire, est impliquée dans la physiopathologie de plusieurs maladies hépatiques chroniques. Les objectifs de ce travail de recherche étaient : 1/ évaluer la contribution de la nécroptose dans la mort des cholangiocytes en contexte de CSP ; 2/ tester l'efficacité de l'inhibition pharmacologique de la protéine serine/threonine-protein kinase 1 (RIPK1), principal régulateur de la nécroptose, dans un modèle murin de CSP. L'analyse par immunofluorescence multiplexe de biopsies de patients atteints de CSP a montré que la nécroptose était fortement activée dans les cholangiocytes. L'étude in vitro de la lignée de cholangiocytes humains H69 a permis de démontrer que l'invalidation de RIPK1 prévenait la mort programmée des cholangiocytes. In vivo, l'inhibition pharmacologique de RIPK1 par RIPA-56 et Necrostatin-1 permettait de réduire les lésions de cholangite fibro-inflammatoire et de fibrose hépatique associées au modèle murin de CSP Abcb4-/-, et aboutissait à la reprogrammation de nombreuses voies de signalisation intra-cellulaire impliquées dans la prolifération cellulaire, l'inflammation et la production de la matrice extra-cellulaire. En conclusion, la stratégie d'inhibition de RIPK1 semble efficace pour traiter les lésions de cholangite sclérosante, et constitue un espoir pour la prise en charge des patients atteints de CSP.

  • Open Access

    RIPK1 inhibition reduces biliary injury and fibrosis in primary sclerosing cholangitis

    by Pierre-Antoine Soret, Virginie Steunou, Julien Hedou, Juliette Tokgozoglu, Valeria Pistorio, Amine Majdi, Nadia Darwane, Jean-Louis Delaunay, Siham Benyahia, Laetitia Dinard, et al.

    Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease with no effective curative therapies. Necroptosis, a regulated necrotic cell death pathway controlled by receptor-interacting protein kinase 1 (RIPK1), has emerged as a potential driver of inflammation and fibrosis in chronic liver disorders. We investigated the role of necroptosis in PSC and whether RIPK1 inhibition could modify disease course and progression. Spatial profiling of human PSC biopsies revealed that necroptosis primarily affects cholangiocytes, while apoptosis was more frequent in hepatocytes and nonbiliary cells. In vitro, necroptosis inhibition protected cholangiocytes, and RIPK1 deletion conferred resistance to TNF-mediated cytotoxicity. In a murine model of PSC, pharmacological RIPK1 inhibition reduced cholestatic injury, hepatic inflammation, and biliary fibrosis. Multiomic analyses comprehensively demonstrated reprogramming toward wild-type–like profiles following treatment. These findings identify necroptosis as a critical effector in PSC and highlight RIPK1 inhibition as a promising disease-modifying approach, opening the door to targeted necroptosis-based therapies for this otherwise untreatable disease.


  • Open Access

    Use of Large Language Models for Automated Short-Answer Question Grading in Second-Cycle Medical Education: Performance and Consistency Evaluation Study (Preprint)

    by Sabrina Sidali, Pierre-Antoine Soret, Mathilde Wagner, Thibault Voron, Quentin Vanderbecq

    BACKGROUND Short-answer questions (SAQs) effectively assess higher-order reasoning skills in medical education but face challenges in scalability due to time-intensive manual grading and concerns about interrater reliability. Recent advances in large language models (LLMs) may offer a solution through automated assessment. OBJECTIVE This study evaluated the performance and consistency of LLMs—ChatGPT, Claude Sonnet 4, and Mistral Pro—compared with those of human graders for SAQ grading in hepatology and gastroenterology. METHODS We retrospectively analyzed SAQs from hepatology and gastroenterology examinations at Sorbonne University, Paris, France (2023-2025). Three human evaluators (2 hepatogastroenterologists and 1 radiologist) independently graded all responses twice, with the majority consensus score serving as the reference standard. Six LLMs were evaluated: ChatGPT (GPT-4o, GPT-4.5, GPT-4o3, and GPT-5), Claude Sonnet 4, and Mistral Pro. Each LLM graded all responses twice using standardized prompts, including questions, model answers, grading guidelines, and scoring rubrics. Agreement was assessed using quadratic weighted kappa (QWK), Cohen κ, the intraclass correlation coefficient (ICC), and the mean absolute error (MAE) for interhuman, intrahuman, LLM-versus-human, and intramodel comparisons. RESULTS A total of 902 student responses from 12 SAQs were included and analyzed. Interhuman grader agreement was almost perfect (QWK 0.90-0.92; ICC 0.90-0.92; MAE 0.05-0.07). Claude Sonnet 4 and GPT-4o3 demonstrated near-human–level performance when compared with the majority vote of human evaluators (Claude Sonnet 4: QWK 0.88-0.90; ICC 0.88-0.90; MAE 0.06-0.07 and GPT-4o3: QWK 0.82-0.89; ICC 0.82-0.89; MAE 0.07-0.09). GPT-5 showed substantial agreement (QWK 0.77-0.79). Mistral Pro and older ChatGPT versions performed poorly (QWK <0.63). Intramodel consistency was the highest for Claude Sonnet 4 (QWK 0.93), approaching the best human intrarater reliability. Agreement did not correlate with prompt design choices (number of examples or minimum word count). CONCLUSIONS Advanced LLMs, particularly Claude Sonnet 4 and GPT-4o3, demonstrated near-human–level grading performance, supporting their potential use as assistive tools in SAQ grading. However, performance varied by model, and agreement alone does not ensure educational validity. Until broader multicenter validation is conducted across diverse domains, stakes, and question formats—along with safeguards for bias, privacy, and model drift—LLMs should augment rather than replace human faculty graders.

  • Open Access

    Epidemiology, comorbidities, treatments and outcomes of autoimmune liver diseases: A French nationwide study

    by Christophe Corpechot, Pierre Hornus, Mallory Cals, Pierre Rinder, Théo Marcille, Amina Malek, Karima Ben Belkacem, Farid Gaouar, Yasmina Chabane, Pierre-Antoine Corret, et al.


  • PBC-HOPE: A Randomized Controlled Trial of Hypnosis and Psychoeducation in Women With Primary Biliary Cholangitis and Fatigue

    by Aurélie Untas, Cécile Goffette, Cécile Flahault, Christel Vioulac, Olivier Chazouillères, Pierre-Antoine Soret, Karima Ben Belkacem, Farid Gaouar, Nathalie Bernard, Alexandra Rousseau, et al.

    INTRODUCTION: Fatigue is the main symptom of primary biliary cholangitis (PBC) but has not yet been improved by any therapeutic intervention. This study evaluated the efficacy and safety of hypnosis and psychoeducation in improving fatigue associated with PBC (PBC-HOPE ClinicalTrial.gov number, NCT03630718). METHODS: Fifty-five women with PBC and significant fatigue, defined by a PBC-40 fatigue score ≥ 33, were randomly assigned to standard care (SC) alone (n = 18), SC plus hypnosis (n = 18), and SC plus psychoeducation (n = 19), with 4 weekly sessions for the intervention groups. Self-report questionnaires, including the PBC-40, were completed at inclusion (D0) and week 12. The first 8 patients in each group were interviewed at both times. The primary outcome was the difference in PBC-40 fatigue score between D0 and week 12. The secondary and exploratory outcomes were the psychometric scores and interview findings. RESULTS: The primary outcome was not achieved, with a median (interquartile range) difference in PBC-40 fatigue score of −3.0 (−10.0 to 1.0), −6.0 (−8.0 to −4.0), and −6.0 (−11.5 to −4.8) for SC, SC-hypnosis, and SC-psychoeducation, respectively. The quantitative secondary outcomes were consistent with this result. The qualitative exploratory outcomes indicated that both interventions positively modified patients' perceptions of fatigue, underlining the appropriation of the intervention. No serious adverse events occurred. DISCUSSION: At 12 weeks, hypnosis and psychoeducation interventions were not associated with a significant reduction in quantitative measures of fatigue associated with PBC. However, the qualitative changes in perceived fatigue associated with these interventions suggest that maintenance sessions could be beneficial in the longer term.