The Role of Endoscopic Ultrasonography (EUS) in Metastatic Tumors in the Pancreas: 10 Years of Experience from a Single High-Volume Center.

BACKGROUND: Metastatic pancreatic lesions (MPLs) are relatively uncommon, constituting 2 to 5% of all pancreatic tumors. They often manifest as solitary lesions without distinct clinical symptoms, usually identified incidentally during radiologic imaging for the surveillance of prior malignancies. Differentiating these lesions from primary pancreatic tumors presents a significant challenge due to their nonspecific presentation. METHODS: We aimed to prospectively assess the effectiveness of endoscopic ultrasound (EUS) and EUS-guided fine needle aspiration/biopsy (EUS-FNA/B) in diagnosing MPLs in a carefully selected cohort of patients presenting with pancreatic masses. Additionally, we sought to examine the relevance of specific EUS findings in supporting the initial diagnosis of MPLs and their agreement with the definitive cytological diagnosis. This study retrospectively analyzed data from 41 patients diagnosed with MPLs between 2013 and 2023, focusing on their clinical and pathological characteristics, the echogenic features of the pancreatic lesions, and the techniques used for tissue acquisition. RESULTS: The incidence of MPLs in our cohort was 3.53%, with the most frequent primary tumors originating in the kidney (43.90%), colorectum (9.76%), lung (9.76%), lymphoma (9.76%), and breast (4.88%). MPLs typically presented as hypoechoic, oval-shaped lesions with well-defined borders and were predominantly hypervascular. Interestingly, 68.29% of the cases were discovered incidentally during follow-up of the primary tumors, while the involvement of the common bile duct was uncommon (19.51%). CONCLUSIONS: EUS and EUS-FNA/B have been validated as valuable diagnostic tools for identifying MPLs. While our findings are promising, further multicenter studies are necessary to corroborate these results and elucidate the predictive value of specific EUS characteristics in determining the metastatic origin of pancreatic lesions.

Optimizing large language models in digestive disease: strategies and challenges to improve clinical outcomes.

Large Language Models (LLMs) are transformer-based neural networks with billions of parameters trained on very large text corpora from diverse sources. LLMs have the potential to improve healthcare due to their capability to parse complex concepts and generate context-based responses. The interest in LLMs has not spared digestive disease academics, who have mainly investigated foundational LLM accuracy, which ranges from 25% to 90% and is influenced by the lack of standardized rules to report methodologies and results for LLM-oriented research. In addition, a critical issue is the absence of a universally accepted definition of accuracy, varying from binary to scalar interpretations, often tied to grader expertise without reference to clinical guidelines. We address strategies and challenges to increase accuracy. In particular, LLMs can be infused with domain knowledge using Retrieval Augmented Generation (RAG) or Supervised Fine-Tuning (SFT) with reinforcement learning from human feedback (RLHF). RAG faces challenges with in-context window limits and accurate information retrieval from the provided context. SFT, a deeper adaptation method, is computationally demanding and requires specialized knowledge. LLMs may increase patient quality of care across the field of digestive diseases, where physicians are often engaged in screening, treatment and surveillance for a broad range of pathologies for which in-context learning or SFT with RLHF could improve clinical decision-making and patient outcomes. However, despite their potential, the safe deployment of LLMs in healthcare still needs to overcome hurdles in accuracy, suggesting a need for strategies that integrate human feedback with advanced model training.

Systematic review: The use of large language models as medical chatbots in digestive diseases.

BACKGROUND: Interest in large language models (LLMs), such as OpenAI's ChatGPT, across multiple specialties has grown as a source of patient-facing medical advice and provider-facing clinical decision support. The accuracy of LLM responses for gastroenterology and hepatology-related questions is unknown. AIMS: To evaluate the accuracy and potential safety implications for LLMs for the diagnosis, management and treatment of questions related to gastroenterology and hepatology. METHODS: We conducted a systematic literature search including Cochrane Library, Google Scholar, Ovid Embase, Ovid MEDLINE, PubMed, Scopus and the Web of Science Core Collection to identify relevant articles published from inception until January 28, 2024, using a combination of keywords and controlled vocabulary for LLMs and gastroenterology or hepatology. Accuracy was defined as the percentage of entirely correct answers. RESULTS: Among the 1671 reports screened, we identified 33 full-text articles on using LLMs in gastroenterology and hepatology and included 18 in the final analysis. The accuracy of question-responding varied across different model versions. For example, accuracy ranged from 6.4% to 45.5% with ChatGPT-3.5 and was between 40% and 91.4% with ChatGPT-4. In addition, the absence of standardised methodology and reporting metrics for studies involving LLMs places all the studies at a high risk of bias and does not allow for the generalisation of single-study results. CONCLUSIONS: Current general-purpose LLMs have unacceptably low accuracy on clinical gastroenterology and hepatology tasks, which may lead to adverse patient safety events through incorrect information or triage recommendations, which might overburden healthcare systems or delay necessary care.

Optimization of hepatological clinical guidelines interpretation by large language models: a retrieval augmented generation-based framework.

Large language models (LLMs) can potentially transform healthcare, particularly in providing the right information to the right provider at the right time in the hospital workflow. This study investigates the integration of LLMs into healthcare, specifically focusing on improving clinical decision support systems (CDSSs) through accurate interpretation of medical guidelines for chronic Hepatitis C Virus infection management. Utilizing OpenAI's GPT-4 Turbo model, we developed a customized LLM framework that incorporates retrieval augmented generation (RAG) and prompt engineering. Our framework involved guideline conversion into the best-structured format that can be efficiently processed by LLMs to provide the most accurate output. An ablation study was conducted to evaluate the impact of different formatting and learning strategies on the LLM's answer generation accuracy. The baseline GPT-4 Turbo model's performance was compared against five experimental setups with increasing levels of complexity: inclusion of in-context guidelines, guideline reformatting, and implementation of few-shot learning. Our primary outcome was the qualitative assessment of accuracy based on expert review, while secondary outcomes included the quantitative measurement of similarity of LLM-generated responses to expert-provided answers using text-similarity scores. The results showed a significant improvement in accuracy from 43 to 99% (p < 0.001), when guidelines were provided as context in a coherent corpus of text and non-text sources were converted into text. In addition, few-shot learning did not seem to improve overall accuracy. The study highlights that structured guideline reformatting and advanced prompt engineering (data quality vs. data quantity) can enhance the efficacy of LLM integrations to CDSSs for guideline delivery.

Predicting response to non-selective beta-blockers with liver-spleen stiffness and heart rate in patients with liver cirrhosis and high-risk varices.

INTRODUCTION: Non-selective beta-blockers (NSBB) are used for primary prophylaxis in patients with liver cirrhosis and high-risk varices (HRVs). Assessing therapeutic response is challenging due to the invasive nature of hepatic venous pressure gradient (HVPG) measurement. This study aims to define a noninvasive machine-learning based approach to determine response to NSBB in patients with liver cirrhosis and HRVs. METHODS: We conducted a prospective study on a cohort of cirrhotic patients with documented HRVs receiving NSBB treatment. Patients were followed-up with clinical and elastography appointments at 3, 6, and 12 months after NSBB treatment initiation. NSBB response was defined as stationary or downstaging variceal grading at the 12-month esophagogastroduodenoscopy (EGD). In contrast, non-response was defined as upstaging variceal grading at the 12-month EGD or at least one variceal hemorrhage episode during the 12-month follow-up. We chose cut-off values for univariate and multivariate model with 100% specificity. RESULTS: According to least absolute shrinkage and selection operator (LASSO) regression, spleen stiffness (SS) and liver stiffness (LS) percentual decrease, along with changes in heart rate (HR) at 3 months were the most significant predictors of NSBB response. A decrease > 11.5% in SS, > 16.8% in LS, and > 25.3% in HR was associated with better prediction of clinical response to NSBB. SS percentual decrease showed the highest accuracy (86.4%) with high sensitivity (78.8%) when compared to LS and HR. The multivariate model incorporating SS, LS, and HR showed the highest discrimination and calibration metrics (AUROC = 0.96), with the optimal cut-off of 0.90 (sensitivity 94.2%, specificity 100%, PPV 95.7%, NPV 100%, accuracy 97.5%).

The Metabolic Impact of Nonalcoholic Fatty Liver Disease on Cognitive Dysfunction: A Comprehensive Clinical and Pathophysiological Review.

Nonalcoholic fatty liver disease (NAFLD) exponentially affects the global healthcare burden, and it is currently gaining increasing interest in relation to its potential impact on central nervous system (CNS) diseases, especially concerning cognitive deterioration and dementias. Overall, scientific research nowadays extends to different levels, exploring NAFLD's putative proinflammatory mechanism of such dysmetabolic conditions, spreading out from the liver to a multisystemic involvement. The aim of this review is to analyze the most recent scientific literature on cognitive involvement in NAFLD, as well as understand its underlying potential background processes, i.e., neuroinflammation, the role of microbiota in the brain-liver-gut axis, hyperammonemia neurotoxicity, insulin resistance, free fatty acids, and vitamins.

Clostridioides difficile infection: history, epidemiology, risk factors, prevention, clinical manifestations, treatment, and future options.

SUMMARYClostridioides difficile infection (CDI) is one of the major issues in nosocomial infections. This bacterium is constantly evolving and poses complex challenges for clinicians, often encountered in real-life scenarios. In the face of CDI, we are increasingly equipped with new therapeutic strategies, such as monoclonal antibodies and live biotherapeutic products, which need to be thoroughly understood to fully harness their benefits. Moreover, interesting options are currently under study for the future, including bacteriophages, vaccines, and antibiotic inhibitors. Surveillance and prevention strategies continue to play a pivotal role in limiting the spread of the infection. In this review, we aim to provide the reader with a comprehensive overview of epidemiological aspects, predisposing factors, clinical manifestations, diagnostic tools, and current and future prophylactic and therapeutic options for C. difficile infection.

The Gut-Liver-Brain Axis: From the Head to the Feet.

The gut-liver-brain axis, a multifaceted network of communication, intricately connects the enteric, hepatic, and central nervous systems [...].

Harnessing the power of synthetic data in healthcare: innovation, application, and privacy.

Data-driven decision-making in modern healthcare underpins innovation and predictive analytics in public health and clinical research. Synthetic data has shown promise in finance and economics to improve risk assessment, portfolio optimization, and algorithmic trading. However, higher stakes, potential liabilities, and healthcare practitioner distrust make clinical use of synthetic data difficult. This paper explores the potential benefits and limitations of synthetic data in the healthcare analytics context. We begin with real-world healthcare applications of synthetic data that informs government policy, enhance data privacy, and augment datasets for predictive analytics. We then preview future applications of synthetic data in the emergent field of digital twin technology. We explore the issues of data quality and data bias in synthetic data, which can limit applicability across different applications in the clinical context, and privacy concerns stemming from data misuse and risk of re-identification. Finally, we evaluate the role of regulatory agencies in promoting transparency and accountability and propose strategies for risk mitigation such as Differential Privacy (DP) and a dataset chain of custody to maintain data integrity, traceability, and accountability. Synthetic data can improve healthcare, but measures to protect patient well-being and maintain ethical standards are key to promote responsible use.

Bilirubin-Induced Transcriptomic Imprinting in Neonatal Hyperbilirubinemia.

Recent findings indicated aberrant epigenetic control of the central nervous system (CNS) development in hyperbilirubinemic Gunn rats as an additional cause of cerebellar hypoplasia, the landmark of bilirubin neurotoxicity in rodents. Because the symptoms in severely hyperbilirubinemic human neonates suggest other regions as privileged targets of bilirubin neurotoxicity, we expanded the study of the potential impact of bilirubin on the control of postnatal brain development to regions correlating with human symptoms. Histology, transcriptomic, gene correlation, and behavioral studies were performed. The histology revealed widespread perturbation 9 days after birth, restoring in adulthood. At the genetic level, regional differences were noticed. Bilirubin affected synaptogenesis, repair, differentiation, energy, extracellular matrix development, etc., with transient alterations in the hippocampus (memory, learning, and cognition) and inferior colliculi (auditory functions) but permanent changes in the parietal cortex. Behavioral tests confirmed the presence of a permanent motor disability. The data correlate well both with the clinic description of neonatal bilirubin-induced neurotoxicity, as well as with the neurologic syndromes reported in adults that suffered neonatal hyperbilirubinemia. The results pave the way for better deciphering the neurotoxic features of bilirubin and evaluating deeply the efficacy of new therapeutic approaches against the acute and long-lasting sequels of bilirubin neurotoxicity.

Gut Microbes Meet Machine Learning: The Next Step towards Advancing Our Understanding of the Gut Microbiome in Health and Disease.

The human gut microbiome plays a crucial role in human health and has been a focus of increasing research in recent years. Omics-based methods, such as metagenomics, metatranscriptomics, and metabolomics, are commonly used to study the gut microbiome because they provide high-throughput and high-resolution data. The vast amount of data generated by these methods has led to the development of computational methods for data processing and analysis, with machine learning becoming a powerful and widely used tool in this field. Despite the promising results of machine learning-based approaches for analyzing the association between microbiota and disease, there are several unmet challenges. Small sample sizes, disproportionate label distribution, inconsistent experimental protocols, or a lack of access to relevant metadata can all contribute to a lack of reproducibility and translational application into everyday clinical practice. These pitfalls can lead to false models, resulting in misinterpretation biases for microbe-disease correlations. Recent efforts to address these challenges include the construction of human gut microbiota data repositories, improved data transparency guidelines, and more accessible machine learning frameworks; implementation of these efforts has facilitated a shift in the field from observational association studies to experimental causal inference and clinical intervention.

Diverticular Disease and Rifaximin: An Evidence-Based Review.

There have been considerable advances in the treatment of diverticular disease in recent years. Antibiotics are frequently used to treat symptoms and prevent complications. Rifaximin, a non-absorbable antibiotic, is a common therapeutic choice for symptomatic diverticular disease in various countries, including Italy. Because of its low systemic absorption and high concentration in stools, it is an excellent medicine for targeting the gastrointestinal tract, where it has a beneficial effect in addition to its antibacterial properties. Current evidence shows that cyclical rifaximin usage in conjunction with a high-fiber diet is safe and effective for treating symptomatic uncomplicated diverticular disease, while the cost-effectiveness of long-term treatment is unknown. The use of rifaximin to prevent recurrent diverticulitis is promising, but further studies are needed to confirm its therapeutic benefit. Unfortunately, there is no available evidence on the efficacy of rifaximin treatment for acute uncomplicated diverticulitis.

Pylephlebitis: A Systematic Review on Etiology, Diagnosis, and Treatment of Infective Portal Vein Thrombosis.

Pylephlebitis, defined as infective thrombophlebitis of the portal vein, is a rare condition with an incidence of 0.37-2.7 cases per 100,000 person-years, which can virtually complicate any intra-abdominal or pelvic infections that develop within areas drained by the portal venous circulation. The current systematic review aimed to investigate the etiology behind pylephlebitis in terms of pathogens involved and causative infective processes, and to report the most common symptoms at clinical presentation. We included 220 individuals derived from published cases between 1971 and 2022. Of these, 155 (70.5%) were male with a median age of 50 years. There were 27 (12.3%) patients under 18 years of age, 6 (2.7%) individuals younger than one year, and the youngest reported case was only 20 days old. The most frequently reported symptoms on admission were fever (75.5%) and abdominal pain (66.4%), with diverticulitis (26.5%) and acute appendicitis (22%) being the two most common causes. Pylephlebitis was caused by a single pathogen in 94 (42.8%) cases and polymicrobial in 60 (27.2%) cases. However, the responsible pathogen was not identified or not reported in 30% of the included patients. The most frequently isolated bacteria were Escherichia coli (25%), Bacteroides spp. (17%), and Streptococcus spp. (15%). The treatment of pylephlebitis consists initially of broad-spectrum antibiotics that should be tailored upon bacterial identification and continued for at least four to six weeks after symptom presentation. There is no recommendation for prescribing anticoagulants to all patients with pylephlebitis. However, they should be administered in patients with thrombosis progression on repeat imaging or persistent fever despite proper antibiotic therapy to increase the rates of thrombus resolution or decrease the overall mortality, which is approximately 14%.

Celiac Disease and Neurological Manifestations: From Gluten to Neuroinflammation.

Celiac disease (CD) is a complex multi-organ disease with a high prevalence of extra-intestinal involvement, including neurological and psychiatric manifestations, such as cerebellar ataxia, peripheral neuropathy, epilepsy, headache, cognitive impairment, and depression. However, the mechanisms behind the neurological involvement in CD remain controversial. Recent evidence shows these can be related to gluten-mediated pathogenesis, including antibody cross-reaction, deposition of immune-complex, direct neurotoxicity, and in severe cases, vitamins or nutrients deficiency. Here, we have summarized new evidence related to gut microbiota and the so-called "gut-liver-brain axis" involved in CD-related neurological manifestations. Additionally, there has yet to be an agreement on whether serological or neurophysiological findings can effectively early diagnose and properly monitor CD-associated neurological involvement; notably, most of them can revert to normal with a rigorous gluten-free diet. Moving from a molecular level to a symptom-based approach, clinical, serological, and neurophysiology data might help to disentangle the many-faceted interactions between the gut and brain in CD. Eventually, the identification of multimodal biomarkers might help diagnose, monitor, and improve the quality of life of patients with "neuroCD".

Incisional hernia by liver cysts.

Autosomal dominant polycystic kidney disease (ADPKD) is the most commonly inherited kidney disease and is associated with cystic manifestation in the liver. Patients with ADPKD are at higher risk for hernias, here we present an image of an incisional hernia full of multiple liver cysts.

Predictors of Hepatocellular Carcinoma Early Recurrence in Patients Treated with Surgical Resection or Ablation Treatment: A Single-Center Experience.

Introduction: Hepatocellular carcinoma (HCC) is the sixth most diagnosed malignancy and the fourth leading cause of cancer-related death worldwide, with poor overall survival despite available curative treatments. One of the most crucial factors influencing survival in HCC is recurrence. The current study aims to determine factors associated with early recurrence of HCC in patients with BCLC Stage 0 or Stage A treated with surgical resection or local ablation. Materials and Methods: We retrospectively enrolled 58 consecutive patients diagnosed with HCC within BCLC Stage 0 or Stage A and treated either by surgical resection or local ablation with maximum nodule diameter < 50 mm. In the first year of follow-up after treatment, imaging was performed regularly one month after treatment and then every three months. Each case was discussed collectively by the Liver Multidisciplinary Group to decide diagnosis, treatment, follow-up, and disease recurrence. Variables resulting in statistically significant difference were then studied by Cox regression analysis; univariately and then multivariately based on forward stepwise Cox regression. Results are represented in hazard ratio (H.R.) with 95% confidence interval (C.I.). Results: There was no statistically significant difference in recurrence rates (34.8 vs. 45.7%, log-rank test, p = 0.274) between patients undergoing surgical resection and local ablation, respectively. Early recurrence was associated with male gender (HR 2.5, 95% C.I. 1.9−3.1), nodule diameter > 20 mm (HR 4.5, 95% C.I. 3.9−5.1), platelet count < 125 × 103 cell/mm3 (HR 1.6, 95% C.I. 1.2−1.9), platelet-lymphocyte ratio < 95 (HR 2.1, 95% C.I. 1.7−2.6), lymphocyte-monocyte ratio < 2.5 (HR 1.9, 95% C.I. 1.4−2.5), and neutrophil-lymphocyte ratio > 2 (HR 2.7, 95% C.I. 2.2−3.3). Discussion and Conclusions: Our results are in line with the current literature. Male gender and tumor nodule dimension are the main risk factors associated with early HCC recurrence. Platelet count and other combined scores can be used as predictive tools for early HCC recurrence, although more studies are needed to define cut-offs.