Mathilde Wagner
Mathilde Wagner has a total of 138 co-authors. Here are the top co-authors by number of shared publications:
- Olivier Lucidarme
Pitié-Salpêtrière Hospital (France)
3 shared publications - Quentin Vanderbecq
Department of Radiology, AP-HP.Sorbonne, Saint Antoine Hospital, Paris, France
3 shared publications - Jean-Baptiste Bachet
Department of Gastroenterology, Pitié-Salpêtrière Hospital—Sorbonne Université, Paris, France
2 shared publications - Pierre-Antoine Soret
Sorbonne Université, Inserm, Centre de Recherche Saint-Antoine, CRSA, F-75012 Paris, France
2 shared publications - Sabrina Sidali
AP-HP, Hôpital Beaujon, Service d'Hépatologie, DMU DIGEST, Centre de Référence des Maladies Vasculaires du Foie, FILFOIE, ERN RARE-LIVER
2 shared publications - Thibault Voron
Department of General and Digestive Surgery, Sorbonne Université, Saint-Antoine Hospital, AP-HP, Hopital Saint-Antoine, Paris, France
2 shared publications
Latest Publications
- Open Access
- 2 November 2026
- Published by JMIR Publications Inc. in JMIR Medical Education
- Vol. 12, e89445
Use of Large Language Models for Automated Short-Answer Question Grading in Second-Cycle Medical Education: Performance and Consistency Evaluation Study
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.
- 2 November 2026
- Published by JMIR Publications Inc. in JMIR Medical Education
- Vol. 12, e89445
Open AccessUse of Large Language Models for Automated Short-Answer Question Grading in Second-Cycle Medical Education: Performance and Consistency Evaluation Study
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
- 11 June 2026
- Published by Wiley in Liver International
- Vol. 46 (7) , e70691
Pre‐ TIPS Liver and Spleen Volumetry Are Not Associated With Liver‐Related Outcomes After TIPS Placement for Refractory Ascites
Background & Aims: Transjugular intrahepatic portosystemic shunt (TIPS) improves survival in refractory ascites. A careful patient's selection is mandatory as TIPS can lead to complications. Liver volumetry is predictive of outcomes before hepatic surgery, but data on its role before TIPS placement are scarce. We aimed to evaluate whether liver and spleen volume measurements are associated with prognosis after TIPS placement in patients with ascites. Methods: We analysed data from three French centers, treated with TIPS between 2017 and February 2022. Inclusion criteria encompassed a TIPS placement for refractory or recurrent ascites and availability of cross‐sectional imaging. Exclusion criteria included non‐cirrhotic portal hypertension, other indications for TIPS, hepatocellular carcinoma beyond Milan criteria, and extrahepatic malignancy. Liver and spleen volumes were measured using pre‐TIPS CT or MRI scans. The primary endpoint was 1‐year transplant‐free survival (TFS). Secondary endpoints were overt hepatic encephalopathy (HE), recurrence of ascites, acute variceal bleeding, and jaundice. Results: The 160 patients were included (median age 60 years, male gender 83.8%, alcohol‐related cirrhosis 58.8%, with active alcohol consumption in 25.6%, Child‐Pugh B cirrhosis in 81.2%, median MELD score was 12). The 1‐year TFS was 60.1%. Multivariate analysis identified serum creatinine (HR = 1.01 95% CI [1.00–1.01], p = 0.04), total bilirubin (HR = 1.02 95% CI [1.02–1.04], p = 0.004), and portal pressure gradient (HR = 1.09 95% CI [1.01–1.18], p = 0.03) as independent factors associated with TFS. Neither liver‐to‐spleen volume ratios (LSVR) ( p = 0.36) nor liver volume index ( p = 0.92) were significantly associated with death or LT. Overall, 38.1% of patients developed overt HE after TIPS, with lower platelet count (HR = 1.01 95% CI [1.00–1.01], p = 0.04) emerging as an independent predictor. No radiological characteristics were associated with the recurrence of ascites. Conclusions: In this multicenter study, liver and spleen volumes were not associated with transplant‐free survival or liver‐related outcomes in patients undergoing TIPS for ascites. These findings suggest that liver volumetry should not be a determining factor in patient selection for TIPS placement.
- 11 June 2026
- Published by Wiley in Liver International
- Vol. 46 (7) , e70691
Open AccessPre‐ TIPS Liver and Spleen Volumetry Are Not Associated With Liver‐Related Outcomes After TIPS Placement for Refractory Ascites
Background & Aims: Transjugular intrahepatic portosystemic shunt (TIPS) improves survival in refractory ascites. A careful patient's selection is mandatory as TIPS can lead to complications. Liver volumetry is predictive of outcomes before hepatic surgery, but data on its role before TIPS placement are scarce. We aimed to evaluate whether liver and spleen volume measurements are associated with prognosis after TIPS placement in patients with ascites. Methods: We analysed data from three French centers, treated with TIPS between 2017 and February 2022. Inclusion criteria encompassed a TIPS placement for refractory or recurrent ascites and availability of cross‐sectional imaging. Exclusion criteria included non‐cirrhotic portal hypertension, other indications for TIPS, hepatocellular carcinoma beyond Milan criteria, and extrahepatic malignancy. Liver and spleen volumes were measured using pre‐TIPS CT or MRI scans. The primary endpoint was 1‐year transplant‐free survival (TFS). Secondary endpoints were overt hepatic encephalopathy (HE), recurrence of ascites, acute variceal bleeding, and jaundice. Results: The 160 patients were included (median age 60 years, male gender 83.8%, alcohol‐related cirrhosis 58.8%, with active alcohol consumption in 25.6%, Child‐Pugh B cirrhosis in 81.2%, median MELD score was 12). The 1‐year TFS was 60.1%. Multivariate analysis identified serum creatinine (HR = 1.01 95% CI [1.00–1.01], p = 0.04), total bilirubin (HR = 1.02 95% CI [1.02–1.04], p = 0.004), and portal pressure gradient (HR = 1.09 95% CI [1.01–1.18], p = 0.03) as independent factors associated with TFS. Neither liver‐to‐spleen volume ratios (LSVR) ( p = 0.36) nor liver volume index ( p = 0.92) were significantly associated with death or LT. Overall, 38.1% of patients developed overt HE after TIPS, with lower platelet count (HR = 1.01 95% CI [1.00–1.01], p = 0.04) emerging as an independent predictor. No radiological characteristics were associated with the recurrence of ascites. Conclusions: In this multicenter study, liver and spleen volumes were not associated with transplant‐free survival or liver‐related outcomes in patients undergoing TIPS for ascites. These findings suggest that liver volumetry should not be a determining factor in patient selection for TIPS placement.
- Open Access
- 15 December 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Use of Large Language Models for Automated Short-Answer Question Grading in Second-Cycle Medical Education: Performance and Consistency Evaluation Study (Preprint)
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.
- 15 December 2025
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessUse of Large Language Models for Automated Short-Answer Question Grading in Second-Cycle Medical Education: Performance and Consistency Evaluation Study (Preprint)
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
Radiologist-AI Collaboration for Ischemia Diagnosis in Small Bowel Obstruction: Multicentric Development and External Validation of a Multimodal Deep Learning Model
Purpose To develop and externally validate a multimodal AI model for detecting ischaemia complicating small-bowel obstruction (SBO). Methods We combined 3D CT data with routine laboratory markers (C-reactive protein, neutrophil count) and, optionally, radiology report text. From two centers, 1,350 CT examinations were curated; 771 confirmed SBO scans were used for model development with patient-level splits. Ischemia labels were defined by surgical confirmation within 24 hours of imaging. Models (MViT, ResNet-101, DaViT) were trained as unimodal and multimodal variants. External testing was used for 66 independent cases from a third center. Two radiologists (attending, resident) read the test set with and without AI assistance. Performance was assessed using AUC, sensitivity, specificity, and 95% bootstrap confidence intervals; predictions included a confidence score. Results The image-plus-laboratory model performed best on external testing (AUC 0.69 [0.59–0.79], sensitivity 0.89 [0.76–1.00], and specificity 0.44 [0.35–0.54]). Adding report text improved internal validation but did not generalize externally; image+text and full multimodal variants did not exceed image+laboratory performance. Without AI, the attending outperformed the resident (AUC 0.745 [0.617–0.845] vs 0.706 [0.581–0.818]); with AI, both improved, attending 0.752 [0.637–0.853] and resident 0.752 [0.629–0.867], rising to 0.750 [0.631–0.839] and 0.773 [0.657–0.867] with confidence display; differences were not statistically significant. Conclusion A multimodal AI that combines CT images with routine laboratory markers outperforms single-modality approaches and boosts radiologist readers’ performance notably junior, supporting earlier, more consistent decisions within the first 24 hours. Key Points A multimodal artificial intelligence (AI) model that combines CT images with laboratory markers detected ischemia in small-bowel obstruction with AUC 0.69 (95% CI 0.59–0.79) and sensitivity 0.89 (0.76–1.00) on external testing, outperforming single-modality models. Adding report text did not generalize across sites: the image+text model fell from AUC 0.82 (internal) to 0.53 (external), and adding text to image+biology left external AUC unchanged (0.69) with similar specificity (0.43–0.44). With AI assistance both junior and senior readers improved; the junior’s AUC rose from 0.71 to 0.77, reaching senior-level performance. Summary Statement A multicentric AI model combining CT and routine laboratory data (CRP and neutrophilia) improved radiologists’ detection of ischemia in small-bowel obstruction. This tool supports earlier decision-making within the first 24 hours.
Open AccessRadiologist-AI Collaboration for Ischemia Diagnosis in Small Bowel Obstruction: Multicentric Development and External Validation of a Multimodal Deep Learning Model
Purpose To develop and externally validate a multimodal AI model for detecting ischaemia complicating small-bowel obstruction (SBO). Methods We combined 3D CT data with routine laboratory markers (C-reactive protein, neutrophil count) and, optionally, radiology report text. From two centers, 1,350 CT examinations were curated; 771 confirmed SBO scans were used for model development with patient-level splits. Ischemia labels were defined by surgical confirmation within 24 hours of imaging. Models (MViT, ResNet-101, DaViT) were trained as unimodal and multimodal variants. External testing was used for 66 independent cases from a third center. Two radiologists (attending, resident) read the test set with and without AI assistance. Performance was assessed using AUC, sensitivity, specificity, and 95% bootstrap confidence intervals; predictions included a confidence score. Results The image-plus-laboratory model performed best on external testing (AUC 0.69 [0.59–0.79], sensitivity 0.89 [0.76–1.00], and specificity 0.44 [0.35–0.54]). Adding report text improved internal validation but did not generalize externally; image+text and full multimodal variants did not exceed image+laboratory performance. Without AI, the attending outperformed the resident (AUC 0.745 [0.617–0.845] vs 0.706 [0.581–0.818]); with AI, both improved, attending 0.752 [0.637–0.853] and resident 0.752 [0.629–0.867], rising to 0.750 [0.631–0.839] and 0.773 [0.657–0.867] with confidence display; differences were not statistically significant. Conclusion A multimodal AI that combines CT images with routine laboratory markers outperforms single-modality approaches and boosts radiologist readers’ performance notably junior, supporting earlier, more consistent decisions within the first 24 hours. Key Points A multimodal artificial intelligence (AI) model that combines CT images with laboratory markers detected ischemia in small-bowel obstruction with AUC 0.69 (95% CI 0.59–0.79) and sensitivity 0.89 (0.76–1.00) on external testing, outperforming single-modality models. Adding report text did not generalize across sites: the image+text model fell from AUC 0.82 (internal) to 0.53 (external), and adding text to image+biology left external AUC unchanged (0.69) with similar specificity (0.43–0.44). With AI assistance both junior and senior readers improved; the junior’s AUC rose from 0.71 to 0.77, reaching senior-level performance. Summary Statement A multicentric AI model combining CT and routine laboratory data (CRP and neutrophilia) improved radiologists’ detection of ischemia in small-bowel obstruction. This tool supports earlier decision-making within the first 24 hours.
- Open Access
- 4 October 2023
- Published by Elsevier in Contemporary Clinical Trials
- Vol. 134, 107352
Liver Investigation: Testing Marker Utility in Steatohepatitis (LITMUS): Assessment & validation of imaging modality performance across the NAFLD spectrum in a prospectively recruited cohort study (the LITMUS imaging study): Study protocol
- 4 October 2023
- Published by Elsevier in Contemporary Clinical Trials
- Vol. 134, 107352
Open AccessLiver Investigation: Testing Marker Utility in Steatohepatitis (LITMUS): Assessment & validation of imaging modality performance across the NAFLD spectrum in a prospectively recruited cohort study (the LITMUS imaging study): Study protocol
- 26 June 2021
- Published by Springer Nature in Obesity Surgery
- Vol. 31 (11) , 5051-5053
Chylous Ascites After Gastric Bypass Surgery: a Feature of Mesenteric Torsion; Comment on: “Chyloperitoneum Due to Small Bowel Obstruction Following Laparoscopic Roux-en-Y Gastric Bypass: a Case Report and Review of Literature”
- 26 June 2021
- Published by Springer Nature in Obesity Surgery
- Vol. 31 (11) , 5051-5053
Chylous Ascites After Gastric Bypass Surgery: a Feature of Mesenteric Torsion; Comment on: “Chyloperitoneum Due to Small Bowel Obstruction Following Laparoscopic Roux-en-Y Gastric Bypass: a Case Report and Review of Literature”
- Open Access
- 30 January 2020
- Published by Springer Nature in BMC Cancer
- Vol. 20 (1) , 1-11
Treatment intensification with hepatic arterial infusion chemotherapy in patients with liver-only colorectal metastases still unresectable after systemic induction chemotherapy – a randomized phase II study -- SULTAN UCGI 30/PRODIGE 53 (NCT03164655)- study protocol
Background: Approximately 40% of colorectal cancer patients will develop colorectal liver metastases (CRLM). The most effective approach to increase long-term survival is CRLM complete resection. Unfortunately, only 10–15% of CRLM are initially considered resectable. The objective response rates (ORR) after current first-line systemic chemotherapy (sys-CT) regimens range from 40 to 80% and complete resection rates (CRR) range from 25 to 50% in patients with initially unresectable CRLM. When CRLM patients are not amenable to complete resection after induction of sys-CT, ORRs obtained with second-line sys-CT are much lower (between 10 and 30%) and consequently CRRs are also low (< 10%). Hepatic arterial infusion (HAI) oxaliplatin may represent a salvage therapy in patients with CRLM unresectable after one or more sys-CT regimens with ORRs and CRRs up to 60 and 30%, respectively. This study is designed to evaluate the efficacy of an intensification strategy based on HAI oxaliplatin combined with sys-CT as a salvage treatment in patients with CRLM unresectable after at least 2 months of first-line induction sys-CT. Objectives and endpoints of the phase II study: Our main objective is to investigate the efficacy, in term of CRR (R0-R1), of treatment intensification in patients with liver-only CRLM not amenable to curative-intent resection (and/or ablation) after at least 2 months of induction sys-CT. Patients will receive either HAI oxaliplatin plus systemic FOLFIRI plus targeted therapy (i.e. anti-EGFR antibody or bevacizumab) or conventional sys-CT plus targeted therapy (i.e. anti-EGFR or antiangiogenic antibody). Secondary objectives are to compare: progression-free survival, overall survival, objective response rate, depth of response, feasibility of delivering HAI oxaliplatin including HAI catheter-related complications, and toxicity (NCI-CTCAE v4.0). Methods: This study is a multicenter, randomized, comparative phase II trial (power, 80%; two-sided alpha-risk, 5%). Patients will be randomly assigned in a 1:1 ratio to receive HAI oxaliplatin combined with systemic FOLFIRI plus targeted therapy (experimental arm) or the best sys-CT plus targeted therapy on the basis of their first-line prior sys-CT history and current guidelines (control arm). One hundred forty patients are required to account for non-evaluable patients. Trial registration: ClinicalTrials.gov, (NCT03164655). Trial registration date: 11th May 2017.
- 30 January 2020
- Published by Springer Nature in BMC Cancer
- Vol. 20 (1) , 1-11
Open AccessTreatment intensification with hepatic arterial infusion chemotherapy in patients with liver-only colorectal metastases still unresectable after systemic induction chemotherapy – a randomized phase II study -- SULTAN UCGI 30/PRODIGE 53 (NCT03164655)- study protocol
Background: Approximately 40% of colorectal cancer patients will develop colorectal liver metastases (CRLM). The most effective approach to increase long-term survival is CRLM complete resection. Unfortunately, only 10–15% of CRLM are initially considered resectable. The objective response rates (ORR) after current first-line systemic chemotherapy (sys-CT) regimens range from 40 to 80% and complete resection rates (CRR) range from 25 to 50% in patients with initially unresectable CRLM. When CRLM patients are not amenable to complete resection after induction of sys-CT, ORRs obtained with second-line sys-CT are much lower (between 10 and 30%) and consequently CRRs are also low (< 10%). Hepatic arterial infusion (HAI) oxaliplatin may represent a salvage therapy in patients with CRLM unresectable after one or more sys-CT regimens with ORRs and CRRs up to 60 and 30%, respectively. This study is designed to evaluate the efficacy of an intensification strategy based on HAI oxaliplatin combined with sys-CT as a salvage treatment in patients with CRLM unresectable after at least 2 months of first-line induction sys-CT. Objectives and endpoints of the phase II study: Our main objective is to investigate the efficacy, in term of CRR (R0-R1), of treatment intensification in patients with liver-only CRLM not amenable to curative-intent resection (and/or ablation) after at least 2 months of induction sys-CT. Patients will receive either HAI oxaliplatin plus systemic FOLFIRI plus targeted therapy (i.e. anti-EGFR antibody or bevacizumab) or conventional sys-CT plus targeted therapy (i.e. anti-EGFR or antiangiogenic antibody). Secondary objectives are to compare: progression-free survival, overall survival, objective response rate, depth of response, feasibility of delivering HAI oxaliplatin including HAI catheter-related complications, and toxicity (NCI-CTCAE v4.0). Methods: This study is a multicenter, randomized, comparative phase II trial (power, 80%; two-sided alpha-risk, 5%). Patients will be randomly assigned in a 1:1 ratio to receive HAI oxaliplatin combined with systemic FOLFIRI plus targeted therapy (experimental arm) or the best sys-CT plus targeted therapy on the basis of their first-line prior sys-CT history and current guidelines (control arm). One hundred forty patients are required to account for non-evaluable patients. Trial registration: ClinicalTrials.gov, (NCT03164655). Trial registration date: 11th May 2017.
- Open Access
- 9 December 2019
- Published by Springer Nature in Cancer Imaging
- Vol. 19 (1) , 1-8
RECIST and CHOI criteria in the evaluation of tumor response in patients with metastatic colorectal cancer treated with regorafenib, a prospective multicenter study
Background: To evaluate the objective response rate (ORR) at 2 months of treatment with regorafenib according to RECIST 1.1, Choi, and modified Choi (mChoi) criteria in patients with metastatic colorectal cancer (mCRC). Methods: Baseline and 2-month contrast-enhanced computed-tomography (CECT) scans of 55 patients with mCRC, prospectively enrolled in phase II TEXCAN trial, were centrally assessed. The primary endpoint was 2-month ORR by RECIST 1.1, Choi, and mChoi criteria. Final outcome was overall survival (OS). Results: Of 55 patients included in this study (Intention-to-treat [ITT1] population), 35 had CECT at 2 months (ITT2 population). According to RECIST 1.1 criteria, 20 (57%) patients were SD and 15 were PD (43%) in the ITT2 population. According to Choi criteria, 18 (51%) patients were responders and 17 (48%) were non-responders. Median OS was 5.3 months (95% CI 3.7–8.6) in the ITT1 population and 8.9 months (95% CI 5.1–12.6) in the ITT2 population. In the ITT2 population, median OS was 16 months (95% CI 6.6–17.5) in SD patients (n = 20) and 4.6 months (95% CI 3.3–5.8) in PD patients (n = 15), according to RECIST 1.1 criteria (HR = 6.48). Median OS was 7.9 months (95% CI 4.2–17.5) in responders (n = 18) and 9.9 months (95% CI 3.7-NA) in non-responders (n = 17) according to Choi criteria (HR = 1.06). All patients except one were classified as non-responders with mChoi criteria. Conclusion: At 2 months, unlike RECIST 1.1, Choi and mChoi criteria could not identify mCRC patients with regorafenib survival benefit. Trial registration: ClinicalTrials.gov Identifier: NCT02699073.Registered March 4, 2016, Retrospectively registered.
- 9 December 2019
- Published by Springer Nature in Cancer Imaging
- Vol. 19 (1) , 1-8
Open AccessRECIST and CHOI criteria in the evaluation of tumor response in patients with metastatic colorectal cancer treated with regorafenib, a prospective multicenter study
Background: To evaluate the objective response rate (ORR) at 2 months of treatment with regorafenib according to RECIST 1.1, Choi, and modified Choi (mChoi) criteria in patients with metastatic colorectal cancer (mCRC). Methods: Baseline and 2-month contrast-enhanced computed-tomography (CECT) scans of 55 patients with mCRC, prospectively enrolled in phase II TEXCAN trial, were centrally assessed. The primary endpoint was 2-month ORR by RECIST 1.1, Choi, and mChoi criteria. Final outcome was overall survival (OS). Results: Of 55 patients included in this study (Intention-to-treat [ITT1] population), 35 had CECT at 2 months (ITT2 population). According to RECIST 1.1 criteria, 20 (57%) patients were SD and 15 were PD (43%) in the ITT2 population. According to Choi criteria, 18 (51%) patients were responders and 17 (48%) were non-responders. Median OS was 5.3 months (95% CI 3.7–8.6) in the ITT1 population and 8.9 months (95% CI 5.1–12.6) in the ITT2 population. In the ITT2 population, median OS was 16 months (95% CI 6.6–17.5) in SD patients (n = 20) and 4.6 months (95% CI 3.3–5.8) in PD patients (n = 15), according to RECIST 1.1 criteria (HR = 6.48). Median OS was 7.9 months (95% CI 4.2–17.5) in responders (n = 18) and 9.9 months (95% CI 3.7-NA) in non-responders (n = 17) according to Choi criteria (HR = 1.06). All patients except one were classified as non-responders with mChoi criteria. Conclusion: At 2 months, unlike RECIST 1.1, Choi and mChoi criteria could not identify mCRC patients with regorafenib survival benefit. Trial registration: ClinicalTrials.gov Identifier: NCT02699073.Registered March 4, 2016, Retrospectively registered.
- 11 September 2019
- Published by Wiley in Journal of Magnetic Resonance Imaging
- Vol. 51 (4) , 1065-1074
Comparison of pulsed and oscillating gradient diffusion‐weighted MRI for characterizing hepatocellular nodules in liver cirrhosis: ex vivo study in a rat model
Background: In contrast to classical pulsed gradient diffusion‐weighted MRI, oscillating gradient diffusion‐weighted MR imaging (DWI) is sensitive to short distance diffusion changes at the intracellular level.Purpose: To compare the diagnostic performance of pulsed and oscillating DWI for characterizing hepatocellular nodules in a rat model of hepatic cirrhosis.Study Type: Prospective, experimental study.Animal Model: Cirrhosis was induced by weekly intraperitoneal injection of diethylnitrosamine in Wistar rats.Field Strength/Sequence: Ex vivo liver MRI was performed at 7T with T1‐weighted, T2‐weighted, pulsed, and oscillating gradient diffusion‐weighted sequences.Assessment: Apparent diffusion coefficient from pulsed (ADCpulsed) and oscillating gradient (ADCoscillating) sequences was calculated in 82 nodules identified on the T1/T2‐weighted images and on pathological examination. Two pathologists classified the nodules in three categories: benign (regenerative and low‐grade dysplastic nodules), with intermediate malignancy (high‐grade dysplastic nodules and early hepatocellular carcinomas) and overtly malignant (progressed hepatocellular carcinomas).Statistical Tests: Differences between groups were assessed with Kruskal–Wallis and Mann–Whitney tests.Results: ADC, mainly ADCoscillating, increased in the group of nodules with intermediate malignancy (ADCpulsed: 0.75 ± 0.25 × 10‐3 mm2/s vs. 0.64 ± 0.07 × 10‐3 mm2/s in benign nodules, P = 0.025; ADCoscillating: 0.81 ± 0.20 × 10‐3 mm2/s vs. 0.65 ± 0.13 × 10‐3 mm2/s, P = 0.0008) and ADCpulsed decreased in the group of progressed hepatocellular carcinomas (ADCpulsed: 0.60 ± 0.08 × 10‐3 mm2/s, P = 0.042; ADCoscillating: 0.68 ± 0.08 × 10‐3 mm2/s, P = 0.1).Data Conclusion: ADC during hepatocarcinogenesis in rats increased in nodules with intermediate malignancy and decreased in progressed hepatocellular carcinomas. Our results suggest that oscillating gradient DWI is more sensitive to the early steps of hepatocarcinogenesis and might be useful for differentiating between high‐grade dysplastic nodules / early hepatocellular carcinomas and regenerating nodules / low‐grade dysplastic nodules.Level of Evidence: 2Technical Efficacy: Stage 1J. Magn. Reson. Imaging 2020;51:1065–1074.
- 11 September 2019
- Published by Wiley in Journal of Magnetic Resonance Imaging
- Vol. 51 (4) , 1065-1074
Comparison of pulsed and oscillating gradient diffusion‐weighted MRI for characterizing hepatocellular nodules in liver cirrhosis: ex vivo study in a rat model
Background: In contrast to classical pulsed gradient diffusion‐weighted MRI, oscillating gradient diffusion‐weighted MR imaging (DWI) is sensitive to short distance diffusion changes at the intracellular level.Purpose: To compare the diagnostic performance of pulsed and oscillating DWI for characterizing hepatocellular nodules in a rat model of hepatic cirrhosis.Study Type: Prospective, experimental study.Animal Model: Cirrhosis was induced by weekly intraperitoneal injection of diethylnitrosamine in Wistar rats.Field Strength/Sequence: Ex vivo liver MRI was performed at 7T with T1‐weighted, T2‐weighted, pulsed, and oscillating gradient diffusion‐weighted sequences.Assessment: Apparent diffusion coefficient from pulsed (ADCpulsed) and oscillating gradient (ADCoscillating) sequences was calculated in 82 nodules identified on the T1/T2‐weighted images and on pathological examination. Two pathologists classified the nodules in three categories: benign (regenerative and low‐grade dysplastic nodules), with intermediate malignancy (high‐grade dysplastic nodules and early hepatocellular carcinomas) and overtly malignant (progressed hepatocellular carcinomas).Statistical Tests: Differences between groups were assessed with Kruskal–Wallis and Mann–Whitney tests.Results: ADC, mainly ADCoscillating, increased in the group of nodules with intermediate malignancy (ADCpulsed: 0.75 ± 0.25 × 10‐3 mm2/s vs. 0.64 ± 0.07 × 10‐3 mm2/s in benign nodules, P = 0.025; ADCoscillating: 0.81 ± 0.20 × 10‐3 mm2/s vs. 0.65 ± 0.13 × 10‐3 mm2/s, P = 0.0008) and ADCpulsed decreased in the group of progressed hepatocellular carcinomas (ADCpulsed: 0.60 ± 0.08 × 10‐3 mm2/s, P = 0.042; ADCoscillating: 0.68 ± 0.08 × 10‐3 mm2/s, P = 0.1).Data Conclusion: ADC during hepatocarcinogenesis in rats increased in nodules with intermediate malignancy and decreased in progressed hepatocellular carcinomas. Our results suggest that oscillating gradient DWI is more sensitive to the early steps of hepatocarcinogenesis and might be useful for differentiating between high‐grade dysplastic nodules / early hepatocellular carcinomas and regenerating nodules / low‐grade dysplastic nodules.Level of Evidence: 2Technical Efficacy: Stage 1J. Magn. Reson. Imaging 2020;51:1065–1074.
- 1 June 2019
- Published by Elsevier in Diagnostic and Interventional Imaging
- Vol. 100 (6) , 363-370
Do regions of interest location and type influence liver stiffness measurement using magnetic resonance elastography?
- 1 June 2019
- Published by Elsevier in Diagnostic and Interventional Imaging
- Vol. 100 (6) , 363-370
Do regions of interest location and type influence liver stiffness measurement using magnetic resonance elastography?