Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Raffaele Pellegrino has a total of 474 co-authors. Here are the top co-authors by number of shared publications:

  • Antonietta Gerarda Gravina

    Department of Precision Medicine, Hepatogastroenterology and Digestive Endoscopy Unit, University of Campania “Luigi Vanvitelli”, Naples, Italy

    48 shared publications
  • Alessandro Federico

    Hepatogastroenterology Division, Department of Precision Medicine, University of Campania Luigi Vanvitelli, Piazza Miraglia 2, 80138 Naples, Italy

    27 shared publications
  • Antonietta Gerarda Gravina

    Department of Precision Medicine, University of Campania Luigi Vanvitelli

    18 shared publications
  • Alessandro Federico

    Hepatogastroenterology Division, Department of Precision Medicine, Luigi Vanvitelli University of Campania, Naples, Italy

    17 shared publications
  • Giovanna Palladino

    Hepatogastroenterology Division, Department of Precision Medicine, University of Campania Luigi Vanvitelli, Naples 80138, Italy

    14 shared publications
  • Giovanna Palladino

    Hepatogastroenterology Division, Department of Precision Medicine, University of Campania Luigi Vanvitelli, 80138 Naples, Italy

    11 shared publications

Latest Publications

  • Open Access

    Evaluation of the Mayo endoscopic score in ulcerative colitis using a multimodal large language model: a human-blinded accuracy study

    by Antonietta Gerarda Gravina, Raffaele Pellegrino, Giuseppe Imperio, Michele Izzo, Ilaria De Costanzo, Fabio Landa, Stefano Pellegrino, Alessandro Federico

    Background/Aims: To date, some large language models (LLMs), such as Chat Generative Pre-trained Transformer 4-omni (ChatGPT-4o), can process images through visual transformer patching. An LLM analysis was conducted to assess the ability of ChatGPT-4o to assign the Mayo endoscopic score (MES). Methods: A selection set of high-quality endoscopic frames was identified to compare 4 input models to select the most performant, confirmed in an extended set of 304 frames. Concordance with the evaluation by expert endoscopists was assessed. Results: Only one of the pre-tested models demonstrated significant concordance (κ = 0.232, 95% confidence interval [CI] = 0.167 to 0.296, P= 0.003; ρ = 0.36, P= 0.011), with a mean bias of –0.26 ± 1.192 (95% CI, –2.596 to 2.076). This was confirmed in the extended set (κ = 0.260, 95% CI = 0.195 to 0.324, P< 0.001; ρ = 0.288, P< 0.001). The absolute concordance for the selected model was 44% and 45.3% in the selection and extended sets, respectively. For the identification of moderately-to-severely active disease, a sensitivity of 73% (95% CI, 60% to 82%), specificity of 60% (95% CI, 54% to 66%), positive predictive value of 32% (95% CI, 25% to 40%), and negative predictive value of 90% (95% CI, 84% to 93%) were identified. Conclusions: ChatGPT-4o shows a mild potential in evaluating MES in endoscopic frames, but further refinements are mandatory.


  • Open Access

    Effectiveness and Safety of Risankizumab in Crohn Disease After Failure of a First Line of Advanced Therapy: A Long-Term, Real-Life Study

    by Antonio Tursi, Giammarco Mocci, Franco Scaldaferri, Daniele Napolitano, Federica Gaiani, Gianluigi De' Angelis, Davide Giuseppe Ribaldone, Eleonora Schillaci, Edoardo V. Savarino, Caterina D. E. Barba, et al.

    Background: Risankizumab (RZB) is an anti-interleukin 23 (anti-IL-23) approved for the treatment of Crohn disease (CD). Study question: We aimed to evaluate the effectiveness and safety of RZB in the treatment of CD in real-world settings. Study design: We performed a retrospective review of a multicentre consortium of patients with CD treated with RZB. Measures and Outcomes: Coprimary outcomes were clinical remission at week 12, 24, and 52 (Harvey–Bradshaw Index score of ≤4), and safety. Secondary outcomes included steroid-free clinical remission, clinical response, and endoscopic remission at 52 weeks. Results: A total of 487 patients were included. The median follow-up was 24 (interquartile range: 16–38) weeks. A total of 372 (76.4%) patients achieved clinical remission at maximal follow-up. According to the treatment line in which RZB was administered, clinical remission occurred in 115/134 (85.8%) patients on second-line therapy, 112/153 (73.2%) on third-line therapy, and 145/200 (72.5%) on fourth-line therapy, with a significant difference ( P = 0.010). Adverse events occurred in 60 patients (12.6%) during follow-up; most of them were mild (50/60, 83.3%). Regarding the secondary outcomes: steroid-free clinical remission was achieved in 342 (71.8%) patients, and clinical response was observed in 443 (91.0%) patients. Six (1.2%) patients underwent surgery during follow-up. Mucosal healing was achieved in 34/67 (50.7%) patients. Conclusions: In this extensive, real-world study, RZB was effective and well tolerated in an advanced-therapy–exposed population of patients with CD and associated with favorable clinical and endoscopic outcomes.

  • Open Access

    Editorial: The emerging role of large language model chatbots in gastroenterology and digestive endoscopy

    by Antonietta Gerarda Gravina, Raffaele Pellegrino, Rocco Maurizio Zagari

    Generative artificial intelligence (AI) has rapidly expanded in medicine in recent years. Large language models (LLMs) have predominated in this landscape and,recently have become multimodal enabling the processing of not only text but also files including images within their reasoning frameworks (1). This development, particularly in fields such as gastroenterology and digestive endoscopy, has provided a substantial impetus to test their applications and operational performance (2)(3)(4). In parallel with these insights into the potential of LLMs in medicine, attention has also been drawn to the ethical and medico-legal issues surrounding the application of these systems in clinical practice and medical research, prompting a cautious interpretation of LLM outputs, in particular when applied to patients, and encouraging the regulation of their use (1,5). Within this perspective, Irfan and Sirvent critically examined the role of LLMs from an ethical standpoint, recognising their potential in gastroenterology for diagnostic and therapeutic support. However, they underlined significant concerns, including the errors that such systems may generate, namely artificial hallucinations, the environmental costs associated with LLMs, and more technical issues, such as the potential tendency of the system to privilege certain sources over others, the absence of clear determinism in responses, limited transparency regarding sources, and challenges related to privacy and medico-legal responsibility. The authors concluded that LLMs may be integrated into practice only in presence of transparent data updating, equitable accessibility, environmental sustainability, and global epistemic inclusion under appropriate human oversight. Within this framework, this Research Topic serves as a catalyst for original studies, reviews, and perspectives on the application of LLMs in gastroenterology and digestive endoscopy.Omar and colleagues sought to address this question by conducting a systematic review of studies assessing the applications of both natural language processing and emerging LLMs in gastroenterology and hepatology. They showed a marked surge of research interest towards GPT-3 and GPT-4 models, in particular in the years 2023 to 2024, The authors highlighted how ChatGPT, which was the dominant LLM within this specific research setting, has been employed for the generation of scientifically sound medical information, for example in providing counselling on Helicobacter pylori infection or in responding to frequently asked questions from patients in the field of hepatology, and for the definition of management pathways, such as formulating the management of patients with colorectal cancer or identifying appropriate endoscopic surveillance intervals. Finally, it has also been utilised to assist medical students in competitive examinations by supporting them in answering examination-style questions during their preparation. El-Kassas et al. conducted a national study to assess the adoption of AI in hepatology practice and research in the Middle East and North Africa. The study included nearly 240 professional respondents and concluded that, although there is optimism regarding the use of AI, its actual utilisation remains limited with significant readiness gaps.Qin et al. proposed a multimodal framework for digestive endoscopy, the Deep Gastrointestinal Diagnosis Network, that integrated both endoscopic frames and textual clinical data through LLM, thereby creating an architecture that combines multi scale feature fusion, guided attention mechanisms, domain adaptation and uncertainty estimation, to enhance lesion detection, segmentation and classification, yielding promising results. On the other hands, Liu et al. developed two distinct machine learning models with the aim of predicting in-hospital mortality of patients admitted to intensive care for acute upper gastrointestinal bleeding. They identified a model with promising accuracy and highlighted prognostically relevant factors, including the use of vasoactive drugs, Glasgow Coma Scale score, AIMS65 score, use of anticoagulants, and peripheral oxygen saturation. Wang et al. evaluated two LLMs, ChatGPT-4o and DeepSeek-R1, in the perioperative pharmacological management of patients with gastric carcinoma by developing approximately 20 questions based on simulations and real clinical cases. They showed hat these tools may have good practical utility, notwithstanding substantial limitations concerning transparency, identification of the sources used by the LLM, and the completeness of their outputs. Finally, Ye et al. reassessed several LLMs, namely ChatGPT-3.5, ChatGPT-4 and ChatGPT-4o, using 30 questions on Helicobacter pylori infection based on international guidelines, with the outputs evaluated by three gastroenterologists. ChatGPT-4o was the most accurate model, although ChatGPT-3.5 was the most consistent moded in the responses.In response to the question of what unresolved limitations persist in the application of LLM in gastroenterology, hepatology, and digestive endoscopy, several assertions may be derived from the data produced within this Research Topic:• The risk of artificial hallucinations and inaccuracies, particularly in complex conditions such as gastrointestinal bleeding, decompensated chronic liver disease, gastrointestinal neoplasms, and inflammatory bowel diseases, becomes an absolutely significant limitation; • Variability of outputs with consequent reduced reliability, reproducibility, and absence of a deterministic input-output relationship; • LLMs are not always fully "explainable", which creates a lack of transparency regarding sources and hinders the mechanisms of clinical auditing; • Multimodal integration, particularly in the evaluation of endoscopic frames, remains largely underexplored and lacks of validation; LLM chatbots are extremely promising tools in gastroenterology and digestive...

  • Open Access

    Filgotinib effectiveness and safety as second or third-line therapy in patients with ulcerative colitis: data from a real-world study

    by Antonio Tursi, Giammarco Mocci, Francesco Costa, Linda Ceccarelli, Edoardo Savarino, Caterina De Barba, Caterina Mucherino, Elvira D’ANtonio, Laura Montesano, Manuela Marzo, et al.

    Background/Aims: Real-world data on the use of filgotinib (FILGO) in patients with ulcerative colitis (UC) are limited. This study aims to provide consistent results on the effectiveness and safety of FILGO in treating UC. Methods: A retrospective assessment of clinical and endoscopic activity was conducted in a cohort of patients with UC according to the full Mayo score. The primary co-endpoints of the study were the evaluation of the effectiveness and safety of FILGO. Results: We enrolled 102 patients with a median follow-up of 24 weeks (interquartile range, 8–24 weeks). At 8 weeks and the end of follow-up, clinical remission was achieved by 38 (37.2%) and 47 (46.1%) patients, respectively. Clinical remission was achieved in 13 of 18 patients (72.2%) receiving first-line therapy, 7 of 19 patients (36.8%) receiving second-line therapy, and 27 of 65 patients (41.5%) receiving third-line therapy (P= 0.002). Clinical remission at 8 weeks predicted clinical remission at the end of follow-up (P= 0.021). Age > 40 years (P= 0.046) and being on second- or third-line of treatment (P= 0.005) were negative predictors for clinical remission. Seventy-one patients (69.6%) achieved a clinical response. At endoscopic evaluation, mucosal healing was observed in 18 out of 30 patients (60.0%). Steroid-free remission was present in 38 out of 46 patients (82.6%). Five patients (4.9%) needed colectomy. Adverse events were recorded in 6 patients (5.8%): 2 cases (2%) were severe, requiring discontinuation of FILGO.Conclusions: Our real-world data confirms that FILGO is safe and effective for patients with UC. Its efficacy is significantly improved when used as a first-line treatment.


  • Open Access

    Management of Pregnancy in Women with Inflammatory Bowel Disease: Positioning Janus Kinase Inhibitors Within Current Evidence

    by Dario Colacurci, Raffaele Pellegrino, Alessia Lamart, Davide Staiano, Ilaria De Costanzo, Michele Izzo, Giuseppe Imperio, Fabio Landa, Giulia Scamardella, Enrica Di Lella, et al.

    Inflammatory bowel diseases (IBD) frequently affect women of reproductive age. Disease activity may arise during pregnancy, at times in severe forms, thereby generating complex clinical scenarios. Adequate control of disease activity throughout pregnancy and the achievement of a safe delivery with a healthy newborn, therefore, represent vital objectives in therapeutic management. In recent years, the therapeutic armamentarium for moderate to severe IBD has expanded exponentially, with the introduction of biological agents and small molecules. However, although these therapies have largely superseded conventional treatment in complex settings, they do not share the same safety profile in pregnancy. Concerns persist regarding potential transplacental transfer and possible teratogenic effects, which justify mandatory caution in their use during pregnancy. Nonetheless, clinicians may readily encounter scenarios of active IBD during pregnancy in patients who have previously experienced failure of the biological agents most extensively studied in this context, thus necessitating an evaluation of the safety of more novel therapeutic options. This review examines the available evidence on Janus kinase inhibitors. Current data, which are highly heterogeneous and of low quality, preclude any recommendation for the use of these small molecules during pregnancy. Prospective registries and large-scale observational studies are mandatory, pending the feasibility of dedicated trials, to better characterise these inhibitors, which could prove valuable, should the evidence ultimately support their use, in women with biologic multi-failure active IBD during pregnancy.


  • Open Access

    Concomitant 5‐Aminosalicylic Acid Does Not Affect the Efficacy of Janus Kinase Inhibitors in Ulcerative Colitis

    by Antonio Tursi, Andrea Pasta, Walter Elisei, Brigida Barberio, Giammarco Mocci, Giovanni Maconi, Antonietta Gerarda Gravina, Raffaele Pellegrino, Giorgia Bodini, Edoardo Vincenzo Savarino, et al.

    We evaluated whether concomitant 5‐aminosalicylic acid (5‐ASA) influences clinical remission in patients with ulcerative colitis (UC) receiving Janus kinase inhibitors (JAKi). In this retrospective, multicenter cohort study, UC patients receiving tofacitinib ( n = 181), upadacitinib ( n = 313), or filgotinib ( n = 139) were included. Time to clinical remission was analyzed using interval‐censored Cox models. Among the 633 patients treated with JAKi, 476 patients received 5‐ASA, and 151 did not. Cumulative probability of remission at week 48 was similar with and without 5‐ASA, at 81.3% and 77.0%, respectively. The adjusted hazard ratio for 5‐ASA vs. no 5‐ASA was 1.12 (95% CI 0.91–1.37, P = 0.519). We therefore found that concomitant 5‐ASA did not significantly affect the time to clinical remission in UC patients treated with JAKi.


  • Endoscopic calcium electroporation for unfit-for-surgery bleeding colorectal cancer: the dawn of a new treatment?

    by Raffaele Pellegrino, Valeria Nacca, Fernando Paragliola, Erika Martinelli, Alessandro Federico, Antonietta G. Gravina

    Article: Endoscopic calcium electroporation for unfit-for-surgery bleeding colorectal cancer: the dawn of a new treatment? - Minerva Medica 2024 Apr 10 - Minerva Medica - Journals


  • Open Access

    Prudence is never excessive: Caution warranted in 5-aminosalicylic acid derivatives withdrawal for ulcerative colitis

    by Antonietta Gerarda Gravina, Raffaele Pellegrino, Alessandro Federico

    This invited commentary discusses the recent study by Atay et al , which investigated relapse rates following the spontaneous withdrawal of maintenance 5-aminosalicylates in ulcerative colitis. The discussion focuses, in this patient setting, on the possible reasons that might prompt clinicians to pursue such exit strategies, and on the importance of exercising caution in these decisions, given the extremely narrow subsets of patients for whom international guidelines allow any degree of leeway.

  • Open Access

    Sex-Specific Real-Life Profiling in Vedolizumab, Ustekinumab, and Tofacitinib Effectiveness in Ulcerative Colitis

    by Antonio Tursi, Raffaele Pellegrino, Giammarco Mocci, Edoardo Vincenzo Savarino, Giovanni Maconi, Walter Elisei, Antonietta Gerarda Gravina, Italian Network For Inflammatory Bowel Diseases (In-Ibd)

    Background: This study aimed to explore whether differences exist between males and females in a cohort of bio-experienced UC patients treated with vedolizumab (VDZ), ustekinumab (UST), or tofacitinib (TOFA) in a 48-week retrospective study. Methods: We evaluated intra- and inter-treatment sex-specific differences regarding clinical response, remission, steroid-free remission, sustained clinical response, late remission, and changes in faecal calprotectin and inflammatory markers at 8, 24, and 48 weeks, as well as endoscopic response and remission at 48 weeks. Results: Among 602 patients (50.2% female), males treated with UST had higher rates of clinical (p = 0.029) and steroid-free clinical remission (p = 0.013) at 24 weeks. Conversely, females on TOFA showed higher clinical remission at 8 weeks (p = 0.043). In males, VDZ demonstrated a superior clinical response over time (p < 0.05), while TOFA showed the highest remission rate at 48 weeks. In females, TOFA was superior for clinical remission at 8 and 24 weeks (p < 0.05). Males had a higher late remission rate (p = 0.04) with an increased likelihood (aOR 1.958, 95%CI 1.088–3.524, p = 0.025). Endoscopic outcomes and faecal calprotectin levels showed no significant sex-specific differences. Conclusions: Sex-based profiling may guide individualised therapeutic strategies in UC patients in this setting.