Beaujon Hospital

Purpose To evaluate the performance of a deep learning algorithm (DLA) for detecting liver metastases (LM) in patients with colorectal cancer (CRC) across diverse clinical contexts and compare its accuracy with that of radiologists. Materials and Methods This retrospective, bicentric study included patients with CRC who underwent contrast-enhanced abdominal CT between January 2019 and December 2021. The DLA accuracy was assessed at the per-nodule and per-patient levels and compared with that of a senior (R1) and an in-training (R2) radiologist blinded to each other's results. The LM detection and false detection rates and interreader agreement were determined. Results Among 181 patients with CRC (mean age, 64 years ± 13 [SD]; 102 male), 95 had LM and 86 had no LM. In the per-nodule analysis, the DLA LM detection rate was 81% (227 of 280; 95% CI: 76.1, 85.2), with no difference compared with R1 (79%; 222 of 280; 95% CI: 74.2, 83.6; P = .49) or R2 (76%; 214 of 280; 95% CI: 71.1, 81.0; P = .19). Detection rates of DLA increased with lesion size: less than 10 mm, 55% (51 of 93; 95% CI: 44.7, 64.6); 10-19 mm, 91% (96 of 106; 95% CI: 83.5, 94.8); and 20 mm or more, 99% (80 of 81; 95% CI: 93.3, 99.8). Detection of subcapsular LM was comparable across readers (DLA, 90% [113 of 125; 95% CI: 84.0, 94.4]; R1, 91% [114 of 125; 95% CI: 84.9, 95.0]; R2, 89% [111 of 125; 95% CI: 82.1, 93.2]). False detection rates were low (DLA, 22% [39 of 181; 95% CI: 16.2, 28.1]; R1, 20% [37 of 181; 95% CI: 15.2, 26.9]; R2, 26% [47 of 181; 95% CI: 20.1, 32.8]; DLA vs R1, P = .004; DLA vs R2, P = .01). DLA false positives were mainly biliary dilatations (n = 14) and diaphragmatic indentations (n = 12). Interreader agreement was moderate (κ = 0.63-0.75). Conclusion DLA performance did not differ from radiologists in detecting LM, with consistent results across lesion sizes and locations. Keywords: Imaging Modality, Abdomen, Gastrointestinal, Liver, Oncology, Comparative Studies, Segmentation, Diagnosis, Deep Learning Supplemental material is available for this article. © RSNA, 2026.

Despite growing clinical use of contrast-enhanced ultrasound (CEUS), inconsistency remains in the modality's role in clinical pathways for hepatocellular carcinoma (HCC) diagnosis and management. This AJR Expert Panel Narrative Review provides practical insights on the use of CEUS for the diagnosis of HCC across populations, including individuals at high risk for HCC, individuals with metabolic dysfunction-associated steatotic liver disease, and individuals not at high risk for HCC. Considerations addressed with respect to high-risk patients include CEUS diagnostic criteria for HCC, use of CEUS for differentiating HCC from non-HCC malignancy, use of CEUS for small (≤ 2 cm) lesions, use of CEUS for characterizing occult lesions on B-mode ultrasound, and use of CEUS for indeterminate lesions on CT or MRI. Representative literature addressing the use of CEUS for HCC diagnosis and gaps in knowledge requiring further investigation are highlighted. Throughout these discussions, the article distinguishes two broad types of ultrasound contrast agents used for liver imaging: pure blood-pool agents and a combined blood-pool and Kupffer-cell agent. Additional topics include the use of CEUS for treatment response assessment after nonradiation therapies and implications of artificial intelligence technologies. The article concludes with a series of consensus statements from the author panel.

Abstract Background Frailty, malnutrition, and sarcopenia are highly prevalent in digestive oncology and adversely affect treatment tolerance, toxicity, and survival. The integration of multimodal, multidisciplinary supportive care may mitigate these risks, but feasibility and clinical impact in routine practice remain uncertain. Methods We conducted two complementary single-center studies at a French cancer center. We developed a Nutritional Frailty Score to assess the risk of nutritional deterioration in patients receiving first-line chemotherapy for digestive cancers. ParcoursDig (NCT04478175) prospectively assessed the feasibility of systematic multimodal supportive care (dietitian, physiotherapist, palliative care) using risk stratification based on this score in patients with localized or advanced digestive cancers. PrePOP was a retrospective observational study of patients with pancreatic or esophagogastric cancers receiving perioperative chemotherapy and surgery. Outcomes included treatment completion, nutritional changes, toxicities, surgical complications, and survival. Results In ParcoursDig, only 44% of patients completed the full supportive care pathway (63% in sensitivity analysis), falling short of the predefined feasibility threshold. No significant improvements were seen in health-related quality of life or survival, highlighting real-world implementation challenges. In PrePOP, baseline malnutrition prevalence was 71% (31% severe). Malnutrition was clinically associated with higher rates of chemotherapy dose reduction but not with surgical complications or survival. Weight loss during systemic therapy was common, but early nutritional intervention appeared to prevent further deterioration for some patients. Muscle mass loss during follow-up correlated with poorer prognosis, emphasizing the need for longitudinal, muscle-specific assessment. Conclusions Low intensity interventions are not sufficient to improve patient outcomes. High-intensity modalities and individualized care trajectories with longitudinal follow-up warrant further evaluation in this population.

An average time to diagnosis of 7 years is commonly described for endometriosis; diagnostic delay for adenomyosis is unknown. This cross-sectional study aimed to describe diagnostic delay for endometriosis and adenomyosis and to investigate the factors associated with this delay. Community of Patients for Research (ComPaRe)-Endometriosis is a prospective e-cohort of women with endometriosis and/or adenomyosis in France. About 6,949 participants were included. We used linear regression modeling to assess factors associated with diagnostic delay, after adjustment for age, education level, body mass index, number of comorbidities, and family history of the disease or of chronic pelvic pain (CPP). The average diagnostic delay was 10 years for endometriosis and 11 years for adenomyosis. Factors associated with a significantly longer diagnostic delay were year of diagnosis (+0.34 years, p < 0.0001), unemployment (+0.7 years, p = 0.01), number of comorbidities (+0.3 years, p < 0.0001), family history of the disease or of CPP (+1.1 years, p < 0.0001), severe dysmenorrhea before diagnosis (numeric-rating scale: +0.8 years, p < 0.0001), number of health professionals consulted before diagnosis (+0.3 years, p < 0.0001), and presence of multiple symptoms leading to first consultation (+1.6 years, p < 0.0001). In contrast, factors associated with a shorter delay were a financial position perceived as comfortable (-1.4 years, p < 0.0001), age at menarche (-0.6 years, p < 0.0001), and age at first symptoms (-0.8 years, p < 0.0001). This study highlights several factors associated with diagnostic delay among women with self-reported endometriosis and/or adenomyosis and provides new insights to improve care by informing both clinicians and patients. Attention to these profiles may improve disease management.

ABSTRACTObjectiveWe sought to characterize the prognostic value of lymphocyte‐C‐reactive protein ratio (LCR) among patients undergoing liver resection (LR) for hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC).MethodsPatients who underwent curative‐intent LR for HCC and ICC between 2000 and 2023 were identified from a multiinstitutional database. The prognostic value of nine different inflammatory markers were evaluated relative to short‐ (i.e., postoperative morbidity) and long‐term (recurrence‐free survival [RFS] and overall survival [OS]) outcomes.ResultsAmong 715 patients, 499 (69.8%) and 216 (30.2%) individuals were included in the derivation and validation cohorts, respectively. Patients with advanced disease and poor tumor biology had lower median levels of LCR. An optimal LCR cutoff threshold of 6100 was identified in the derivation cohort. LCR demonstrated the highest accuracy to predict RFS and OS, with areas under the ROC curve of 0.724 and 0.716, respectively. After adjusting for relevant clinicodemographic factors, lower LCR remained associated with higher odds of postoperative complications (OR: 1.98 [95% CI: 1.27–3.10] and p = 0.003) and particularly, infectious complications (OR: 2.80 [95% CI: 1.57–5.01] and p < 0.001). A lower LCR was independently associated with worse RFS (HR: 2.43 [95% CI: 1.41–3.83] and p = 0.002) and OS (HR: 2.95 [95% CI: 2.10–4.16] and p < 0.001). The prognostic ability of LCR for short‐ and long‐term outcomes performed well in an independent validation cohort.ConclusionLCR was strongly associated with risk of postoperative morbidity as well as worse RFS and OS among patients undergoing LR for HCC and ICC. Preoperative LCR assessment can aid surgeons in the preoperative risk‐stratification of patients undergoing surgery for primary liver cancer.

Abstract Background and objectives – Absent cough reflex is associated with mortality intensive care unit (ICU) patients requiring deep sedation, suggesting that lower brainstem dysfunction contributed to adverse outcomes. We conducted a multicenter observational cohort study to confirm this hypothesis by assessing the peak latency (PL) of the lower brainstem-generated P14 evoked potential (EP), which is slightly increased by sedatives. We aimed to demonstrate that a P14-PL&gt; 16 ms is independently associated with day-28 mortality. Patients and methods - Mechanically ventilated adult patients, comatose or deeply sedated, brain-injured or not, were included. At day 3, EPs were performed in patients remaining unconscious. The Simplified Acute Physiological Score (SAPSII), initial Glasgow Coma Scale (GCS), sedation depth and brainstem reflexes were collected. The primary outcome was 28-day mortality. The secondary outcomes were delayed awakening and delirium after sedation discontinuation. Results - Between 2015 and 2019, 322 patients were included. EPs were performed in 264 (82%) patients, including 140 (53%) brain-injured and 251 (95%) deeply sedated patients. The median age, SAPSII and initial GCS were 62 years [50; 71], 49 [40; 62] and 11 [6; 15], respectively. A P14-PL &gt; 16ms was found in 76 (29%) patients and was associated with day-28 mortality (adjusted hazard ratio, 3.0; 95% confidence interval, [1.7-5.2]). Absent cough and pupillary light reflexes were associated with death. Only absent oculocephalogyre reflex was associated with delayed awakening (adjusted odds ratio, 2.1, 95%CI, [1.1 - 3.7]). Interpretation – Impaired neurological and neurophysiological lower brainstem responses are associated with mortality in deeply sedated patients. Funded by the French Ministry of Health; PRORETRO; n° P120915; ClinicalTrials.gov registry: NCT02395861; date: 24 March 2015