Background: Preoperative malnutrition is a significant factor in patients with pancreatic tumors undergoing pancreaticoduodenectomy. The aim of this study was to assess the association between preoperative malnutrition and delayed discharge within a ten-day timeframe and potential correlations between preoperative malnutrition and postoperative surgical complications. Methods: A retrospective cohort study was conducted, recruiting a final sample of 79 patients with benign or malignant cephalic pancreatic tumors from 2015 to 2022. The risk of malnutrition was assessed using the Malnutrition Universal Screening Tool, while length of hospital stay and relevant clinical data were extracted from clinical documentation. Results: The preoperative malnutrition risk was high in 21.52% of the sample, moderate in 36.71%, and low in 41.77%. Body mass index (BMI) (p = 0.007) and postoperative surgical complications (p < 0.001) were significantly correlated with delayed discharge. No statistically significant differences were found between levels of malnutrition risk and delayed discharge (p = 0.122), or postoperative surgical complications (p = 0.874). Conclusions: Postoperative complications and BMI emerge as significant risk factors. The limited sample size may have compromised the collection of homogeneous and significant data. Future studies should evaluate the implementation of personalized nutritional screening tools, nutritional assessment plans, and the involvement of specialized health professionals.

The clinical management of Pancreatic Neuroendocrine Neoplasms (Pan-NENs) is complicated by the disease’s intrinsic variability, which creates significant hurdles for accurate risk profiling and the standardization of treatment protocols. Recently, Artificial Intelligence (AI) has offered a promising avenue to address these challenges. By integrating and processing high-dimensional multimodal datasets (encompassing clinical history, radiomics, and pathology), these computational tools can refine survival forecasts and support the development of personalized medicine. However, the transition from experimental success to routine clinical use is currently obstructed by reliance on limited, retrospective cohorts that lack external validation, alongside unresolved concerns regarding algorithmic transparency and ethical governance. This review evaluates the current landscape of AI-driven prognostic modeling for Pan-NENs and critically examines the pathway towards their reliable integration into clinical practice.

Background: Robotic liver resection (RLR) has seen remarkable advancements in recent years, overcoming many limitations of laparoscopic liver resection (LLR). RLR has evolved to include increasingly complex procedures, offering enhanced precision, reduced blood loss, and lower complication rates. Materials and Methods: A total of 150 consecutive RLRs, performed at the Department of General Surgery II and HPB Unit of Santa Chiara Hospital (Trento, Italy), between January 2013 and June 2024 were retrospectively reviewed. Collected data included demographics, disease etiology, operative parameters, oncologic margins, and perioperative outcomes. Results: Indications were malignant disease in 83% of cases while benign disease accounted for 17%. Minor resections accounted for 91%. Cirrhosis was present in 49% of patients (Child–Pugh A 91%; B 9%; mean MELD 9). According to the Iwate difficulty score, resections were low difficulty in 38% of cases, intermediate in 50%, advanced in 7%, expert in 5%. Conversion rate was 12%, mainly for bleeding or adhesions. Mean blood loss was 159 mL (66% <100 mL); Pringle maneuver was used in 3%; drains omitted in 45%; ICG fluorescence used in 81%. Mean operative time was 250 min (console time 184 min). Mean lesion size was 34 mm; R0 margin rate was 82%. Overall mortality was 1.3%; morbidity 24% (Clavien–Dindo ≥ III in 10%). Mean hospital stay was 7 days (median 5; range 2–46). Conclusions: RLR is a safe and effective alternative to laparoscopy, providing comparable or superior perioperative outcomes. Medium-volume centers can achieve high-quality results with RLR. Continued technological advancements will further expand its applications to increasingly complex liver procedures.

The incidence of gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs) is increasing, but population registries seldom capture detailed clinical data. The Italian Association for Neuroendocrine Tumours (Itanet) established a nationwide prospective database to describe presentation, diagnostic pathways, management, and outcomes of newly diagnosed GEP-NENs in Italy. This multicentre prospective observational study enrolled 2138 consecutive patients with newly diagnosed GEP-NENs across 38 Italian centres (2019–2024). Clinical, pathological, imaging, and treatment data were prospectively collected and centrally validated. Descriptive and survival analyses were performed; Ki-67 was modelled as a continuous variable. Median age was 60.6 years, and 55.9% (1195/2138) were male. Tumours were well-differentiated NETs in 90.8% of patients (1942/2138), mainly of pancreatic (41.4%, 886/2138) or ileal (19.7%, 422/2138) origin. Median Ki-67 was 2%. An incidental diagnosis occurred in 58.6% (1254/2138) of cases. Among symptomatic patients, the mean diagnostic delay was 197 days (224 for pancreatic vs 184 for small bowel; p = 0.039). ⁶⁸Ga-DOTA-peptide PET showed higher diagnostic yield than CT or MRI in small-bowel primaries (93.7% vs 83.1% vs 67.4%), whereas performance was similar in pancreatic tumours. Data on first-line treatment were available for 2050 patients. Initial management included surgery in 36.4% (746/2050), watchful waiting in 19.7% (404/2050), endoscopic resection in 7.7% (158/2050). Overall, 30.6% (627/2050) of patients received systemic therapy, most commonly somatostatin analogues in 23.7% (487/2050). Over a median follow-up of 271 days (IQR 121–530), 62 deaths were observed (event rate 4.9%). Overall survival differed markedly according to metastatic status and tumour grade. Ki-67 was prognostic when modelled continuously (p < 0.001), and a 15% cutoff identified poorer outcomes. This nationwide prospective study delineates real-world diagnostic and therapeutic patterns of GEP-NENs in Italy, confirms Ki-67 as a continuous prognostic biomarker, and identifies a 15% threshold associated with worse survival, providing a benchmark for outcome assessment and future clinical research. None.

The clinical management of Pancreatic Neuroendocrine Neoplasms (Pan-NENs) is complicated by the disease's intrinsic variability, which creates significant hurdles for accurate risk profiling and the standardization of treatment protocols. Recently, Artificial Intelligence (AI) has offered a promising avenue to address these challenges. By integrating and processing high-dimensional multimodal datasets (encompassing clinical history, radiomics, and pathology), these computational tools can refine survival forecasts and support the development of personalized medicine. However, the transition from experimental success to routine clinical use is currently obstructed by reliance on limited, retrospective cohorts that lack external validation, alongside unresolved concerns regarding algorithmic transparency and ethical governance. This review evaluates the current landscape of AI-driven prognostic modeling for Pan-NENs and critically examines the pathway towards their reliable integration into clinical practice.