Background The association between body mass index (BMI) and outcomes after liver resection remains controversial. We examined this association using a large real-world electronic health record cohort. Methods Patients undergoing liver resection for cancer from 2017 to 2025 were identified from Epic Cosmos. BMI was categorized according to World Health Organization criteria, with normal weight as the reference. Multivariable regression, restricted cubic splines, subgroup analyses, and sensitivity analyses were performed. Results Among 16,022 patients, class II/III obesity was associated with higher odds of any 30-day complication (aOR 1.22, 95%CI 1.09–1.36), cardiac events (aOR 1.32, 95%CI 1.16–1.51), and renal events (aOR 1.77, 95%CI 1.41–2.22). After additional adjustment for cirrhosis, ALBI grade, and malnutrition, the association with any 30-day complication persisted (aOR 1.39, 95%CI 1.23–1.57). Higher BMI was associated with shorter length of stay in the primary analysis (aIRR 0.89–0.94). Inverse associations with 90-day mortality and hepatobiliary complications were attenuated after additional adjustment. Discussion Severe obesity was associated with higher postoperative morbidity, particularly cardiac and renal complications. BMI should be interpreted alongside nutritional status and hepatic reserve in perioperative risk assessment. The attenuated inverse associations should not be interpreted as evidence that higher BMI is protective.
BACKGROUND:Intrahepatic cholangiocarcinoma (iCCA) is an aggressive primary liver cancer for which surgical resection with lymphadenectomy remains the only potentially curative treatment. Although minimally invasive liver surgery (MILS) provides benefits over open surgery, whether robotic liver resection (RLR) offers advantages over laparoscopic resection (LLR) in oncological adequacy and perioperative outcomes remains unclear. METHODS:We retrospectively analyzed consecutive patients with iCCA who underwent MILS in 13 Italian hepatobiliary centers. Patients were stratified by surgical approach (LLR vs RLR) and matched 2:1 by propensity score matching (PSM). Primary endpoints were oncological outcomes and adequacy of lymph node dissection (LND). Secondary endpoints included intraoperative and short-term postoperative outcomes. RESULTS:Overall, 645 patients (468 LLR, 177 RLR) were included; after PSM, 511 patients (334 LLR, 177 RLR) were analyzed. RLR achieved higher R0 resection rates (p = 0.002) and a greater mean lymph node yield (p = 0.001). Overall survival (OS) and disease-free survival (DFS) were comparable between groups. RLR was also associated with lower conversion rates, fewer postoperative complications, and shorter hospital stay. CONCLUSION:RLR for iCCA is safe and provides advantages in margin status, lymph node dissection, and perioperative outcomes, while long-term oncological outcomes (OS and DFS) are comparable to LLR.
BACKGROUND:The integration of Artificial Intelligence (AI) into clinical decision-making is accelerating, yet its reliability compared to Multidisciplinary Team (MDT) meetings in Hepato-Pancreato-Biliary (HPB) surgery remains underexplored. This study evaluated the concordance between AI recommendations and MDT consensus in real-world HPB oncology. METHODS:A retrospective comparative analysis of all HPB cases discussed at the MDT meetings of the 1st Department of Surgery, National and Kapodistrian University of Athens, between July 1st and December 31st 2025, was conducted. Anonymized text-only clinical summaries were processed using a large language model (Gemini 3.1 Pro) to generate treatment recommendations. The primary endpoint was the agreement rate between the AI and the MDT decision. RESULTS:An overall agreement rate of 91.7% (k = 0.841; p < 0.01) was observed. Concordance was exceptionally high in straightforward cases (96.7%; k = 0.938; p < 0.01), mirroring standard oncologic workflows. Conversely, in complex scenarios agreement dropped significantly to 62.5% (k = 0.304; p = 0.09). In these instances, AI recommendations diverged from the individualized, clinical judgment applied by the MDT. CONCLUSIONS:AI platforms demonstrate high concordance with MDTs in routine HPB scenarios, suggesting potential for streamlining standard workloads. Nevertheless, human multidisciplinary expertise remains indispensable for complex cases requiring nuanced clinical intuition.
BACKGROUND:Molecular subtyping of pancreatic ductal adenocarcinoma (PDAC) may inform prognosis and treatment selection, and its clinical application remains under investigation. Although PurIST is a validated transcriptomic classifier, immunohistochemical (IHC) markers such as GATA6 and CK5 have been proposed as practical surrogates. We evaluated whether GATA6 and CK5 IHC could substitute for transcriptomic PDAC subtyping. METHODS:RNA-seq and GATA6/CK5 IHC were performed on 130 matched tumor samples. Most patients received neoadjuvant therapy. PurIST subtype assignment served as the reference. Concordance was assessed using Cohen's kappa and overall survival using log-rank tests. RESULTS:PurIST classified 8 tumors (6%) as basal-like. GATA6 alone showed poor concordance with PurIST (kappa=0.18, 95% CI -0.06 to 0.42) and was not prognostic of survival (P=0.98). CK5 showed better concordance (kappa=0.39, 95% CI 0.07 to 0.70) but was not significantly prognostic (P=0.084). Combined GATA6/CK5 improved concordance modestly (kappa=0.54, 95% CI 0.18 to 0.90) and stratified OS in the survival cohort (P=0.041), but generated categories that did not align cleanly with established subtypes. GATA6 IHC correlated weakly with RNA-seq expression (rho=0.178, P=0.0429). DISCUSSION:In predominantly post-neoadjuvant PDAC specimens, GATA6 and CK5 capture aspects of subtype biology but are not reliable substitutes for transcriptomic classification.
BACKGROUND:Pancreatectomy carries significant risk of morbidity and mortality, particularly in the geriatric population; the risk of postoperative loss of independence (LOI) remains underexplored. The aim was to identify perioperative factors associated with LOI following pancreatectomy in this population. METHODS:The NSQIP database was queried to identify adults ≥75 years who underwent pancreatectomy from 2021 to 2024. LOI was the primary outcome, defined as decline in functional status and/or non-home discharge and/or home discharge services. Multivariable logistic regression was performed to identify predictors of LOI. RESULTS:Of 6227 patients, 1731(27.8%) experienced LOI at discharge. Factors associated with LOI included failure to wean from mechanical ventilation >48 h (OR 4.14,95%CI:2.19-7.85), cognitive impairment (OR 2.16,95%CI:1.50-3.10), and history of falls (OR 1.96,95%CI:1.56-2.48). Additional predictors included age >80 years (OR 1.4,95%CI:1.23-1.58), female sex (OR:1.26,95%CI:1.1-1.43), and COPD (OR 1.49,95%CI:1.13-1.95). The presence of home support prior to surgery (OR 0.81,95%CI:0.7-0.94) and early discharge (≤5 days) (OR 0.41,95%CI:0.33-0.51) were demonstrated to be protective. DISCUSSION:The understanding of LOI in this population can facilitate preoperative counseling and expectation setting. Care teams should work to mitigate LOI and/or obtain necessary support in the perioperative period.
BACKGROUND:Small colorectal liver metastases (CRLMs) commonly disappear following neoadjuvant chemotherapy. Some authors have suggested upfront surgery, fiducial placement or blind resection of these Disappearing Liver Lesions (DLLs). This retrospective review defines the natural history, recurrence patterns, and survival of patients with DLLs. METHODS:Patients who underwent liver directed surgery for small CRLMs (≤3 cm) were identified from a prospectively collected database. DLLs were defined as tumors not seen on pre-operative MRI or intra-operative ultrasound following neoadjuvant chemotherapy. Statistical analysis was conducted using Python 3.11(Python software foundation, 2023). RESULTS:The study included 72 patients with a median follow-up of 55 months. Overall survival was 73 months, and disease recurrence occurred in 46 patients (64%). After neoadjuvant chemotherapy, 29 patients (40%) had at least one DLL. Among patients with DLLs, distant hepatic or systemic recurrence occurred in 21 (72%). However, regrowth of DLL occurred in only 3 patients (10%). Two of these were treated with median survival of 97 months and the third had synchronous extrahepatic disease that precluded additional surgery. DISCUSSION:DLLs are common, but only rarely regrow (10%). Pre-emptive treatment or fiducial placement for small CRLMs is likely unnecessary and may lead to overtreatment.
BACKGROUND:In laparoscopic cholecystectomy, the gallbladder is removed via the umbilical or epigastric/subxiphoid port. AIMS:We aim to compare the umbilical and epigastric ports for gallbladder extraction in laparoscopic cholecystectomy in terms of postoperative pain, surgical site infection, postoperative hernia, operating time, and gallbladder retrieval time. METHODS:Systematic review and meta-analysis of randomized controlled trials from PubMed, Scopus,WOS, CENTRAL, and Clinicaltrials.gov from inception to 18 June 2025. PROSPERO ID (CRD420251143624). RESULTS:Analysis of 20 RCTs showed that gallbladder extraction via the umbilical port was associated with significant reduction in postoperative pain at 1 h (SMD:-1.31, 95%CI:-2.14; -0.49, p = 0.0018), 6 h (SMD:-3.21, 95%CI:-5.71; -0.70, p = 0.0121), 12 h (SMD:-3.06, 95%CI:-5.40; -0.71, p = 0.0107), 24 h (SMD:-1.96, 95%CI:-3.71; -0.21, p = 0.0283), and 36 h (SMD:-0.98, 95%CI:-1.28; -0.69, p < 0.0001). PIs crossed the null for time points except 36 h. No significant differences were observed at 48, 72 h, 7, or 30 days. Umbilical port was associated with longer hospital stay (MD:1.79 days, 95%CI:0.54; 3.04, p = 0.0050, PI:-0.24; 3.82). No differences were found for surgical site infection, hernia, operative time, extraction time, or retrieval difficulty, PIs frequently crossed the null. CONCLUSION:Extraction via the umbilical port was associated with statistically significant reduction in early postoperative pain. Wide PIs crossing the null indicate substantial uncertainty. Neither port site demonstrates definitive advantage; selection should be guided by surgeon preference and patient-specific factors.
BACKGROUND:Pancreatoduodenectomy (PD) is a morbid procedure often associated with high rates of postoperative infectious complications, contributing to worse prognosis. We sought to investigate the optimal surgical prophylaxis antibiotic for PD patients using the ACS-NSQIP. STUDY DESIGN:The ACS-NSQIP (2016-2023) was queried for elective PDs and their prophylactic antibiotics. Patients were categorized by antibiotic type: first-generation cephalosporin (FC), second/third-generation cephalosporin (STC), and broad-spectrum (BS - e.g., piperacillin-tazobactam, cefoxitin, and fourth-generation cephalosporins). The primary outcomes were surgical site infection (SSI) and Clostridium difficile colitis. Inverse probability weighting using multinomial propensity scores was applied to balance groups and estimate adjusted probabilities and pairwise risk differences. RESULTS:Among 23,258 PD patients, adjusted SSI probabilities were 22.7% (21.7-23.8%), 22.3% (21.3-23.3%), and 20.0% (19.2-20.8%) for FC, STC, and BS, respectively (p < 0.001). BS antibiotics consistently reduced SSI risk without significant differences in other outcomes, including pancreatic fistula, readmission, reoperation, mortality, or major complications. Infection rates differed by biliary drainage status (p < 0.001). C. difficile rates were low and comparable across groups (1.1% vs. 1.9% vs. 1.7%). CONCLUSION:Broader prophylactic coverage in elective PD reduces postoperative infections without increasing C. difficile risk, especially benefiting patients with preoperative biliary instrumentation.
