Explore the vast range of titles available.

Background Neoadjuvant therapy may improve survival of patients with pancreatic adenocarcinoma; however, determining response to therapy is difficult. Artificial intelligence allows for novel analysis of images. We hypothesized that a deep learning model can predict tumor response to NAC. Methods Patients with pancreatic cancer receiving neoadjuvant therapy prior to pancreatoduodenectomy were identified between November 2009 and January 2018. The College of American Pathologists Tumor Regression Grades 0-2 were defined as pathologic response (PR) and grade 3 as no response (NR). Axial images from preoperative computed tomography scans were used to create a 5-layer convolutional neural network and LeNet deep learning model to predict PRs. The hybrid model incorporated decrease in carbohydrate antigen 19-9 (CA19-9) of 10%. Accuracy was determined by area under the curve. Results A total of 81 patients were included in the study. Patients were divided between PR (333 images) and NR (443 images). The pure model had an area under the curve (AUC) of .738 (P < .001), whereas the hybrid model had an AUC of .785 (P < .001). CA19-9 decrease alone was a poor predictor of response with an AUC of .564 (P = .096). Conclusions A deep learning model can predict pathologic tumor response to neoadjuvant therapy for patients with pancreatic adenocarcinoma and the model is improved with the incorporation of decreases in serum CA19-9. Further model development is needed before clinical application.

Background As Enhanced Recovery After Surgery (ERAS®) programs expand across numerous subspecialties, growth and sustainability on a system level becomes increasingly important and may benefit from reporting multidisciplinary and financial data. However, the literature on multidisciplinary outcome analysis in ERAS is sparse. This study aims to demonstrate the impact of multidisciplinary ERAS auditing in a hospital system. Additionally, we describe developing a financial metric for use in gaining support for system-wide ERAS adoption and sustainability. Methods Data from HPB, colorectal and urology ERAS programs at a single institution were analyzed from a prospective ERAS Interactive Audit System (EIAS) database from September 2015 to June 2019. Clinical 30-day outcomes for the ERAS cohort ( n  = 1374) were compared to the EIAS pre-ERAS control ( n  = 311). Association between improved ERAS compliance and improved outcomes were also assessed for the ERAS cohort. The potential multidisciplinary financial impact was estimated from hospital bed charges. Results Multidisciplinary auditing demonstrated a significant reduction in postoperative length of stay (LOS) (1.5 days, p  < 0.001) for ERAS patients in aggregate and improved ERAS compliance was associated with reduced LOS (coefficient − 0.04, p  = 0.004). Improved ERAS compliance in aggregate also significantly associated with improved 30-day survival (odds ratio 1.04, p  = 0.001). Multidisciplinary analysis also demonstrated a potential financial impact of 44% savings ( p  < 0.001) by reducing hospital bed charges across all specialties. Conclusions Multidisciplinary auditing of ERAS programs may improve ERAS program support and expansion. Analysis across subspecialties demonstrated associations between improved ERAS compliance and postoperative LOS as well as 30-day survival, and further suggested a substantial combined financial impact.

Background Resection with trisectionectomy may necessitate liver molding for adequate future liver remnant (FLR), and subsequent complications can impact return to intended oncologic therapy (RIOT). This study evaluated whether a difference in RIOT exists with the use of molding and between liver molding techniques (associating liver partition and portal vein ligation for staged hepatectomy [ALPPS] and portal vein embolization [PVE]) with trisectionectomy. Methods A retrospective review evaluated trisectionectomies for malignancy. Outcomes were compared with and without molding, and RIOT was determined. Results Fifty-one patients underwent trisectionectomy: 11 ALPPS, 14 PVE, 26 without molding. 73% of ALPPS, 64% of PVE, and 58% without molding achieved RIOT (P = .971). There were no differences found in baseline characteristics, R0 rate, length of stay, readmission, complications, or mortality. Time to RIOT was significantly different (ALPPS: 3.3 months; PVE: 5.2 months; none: 2.4 months, P = .0203). There were no differences in recurrence or survival. Conclusions Liver molding should not cause apprehension as there are no differences in achieving RIOT. Although technique alters time to RIOT, this does not translate into improved outcomes, implicating disease biology, and regeneration stimulus.

The role of surgical intervention for necrotizing pancreatitis has evolved; however, no widely accepted algorithm has been established to guide timing and optimal modality in the minimally invasive era. This study aimed to retrospectively validate an established institutional timing- and physiologic-based algorithm constructed from evidence-based guidelines in a high-volume hepatopancreatobiliary center. Patients with necrotizing pancreatitis requiring early (≤six weeks from symptom onset) or delayed (>six weeks) surgical intervention were reviewed over a four-year period (n = 100). Early intervention was provided through laparoscopic drain-guided retroperitoneal debridement (n = 15) after failed percutaneous drainage unless they required an emergent laparotomy (due to abdominal compartment syndrome, bowel necrosis/perforation, or hemorrhage) after which conservative, sequential open necrosectomy was performed (n = 47). Robot-assisted (n = 16) versus laparoscopic (n = 22) transgastric cystgastrostomy for the delayed management of walled-off pancreatic necrosis was compared, including patient factors, operative characteristics, and 90-day clinical outcomes. Major complications after early debridement were similarly high (open 25% and drain-guided 27%), yet 90-day mortality was low (open 8.5% and drain-guided 7.1%). Patient and operative characteristics and 90-day outcomes were statistically similar for robotic versus laparoscopic transgastric cystogastrostomy. Our evidence-based algorithm provides a stepwise approach for the management of necrotizing pancreatitis, emphasizing minimally invasive early and late interventions when feasible with low morbidity and mortality. Robot-assisted transgastric cystogastrostomy is an acceptable alternative to a laparoscopic approach for the delayed treatment of walled-off pancreatic necrosis.

Management of pyogenic hepatic abscesses (PHA) varies among surgeons and institutions. Recent studies have advocated for first-line percutaneous drainage (PD) of all accessible hepatic abscesses, with surgery reserved as rescue only. Our study aimed to internally validate an established multimodal algorithm for PHA at a high-volume hepatopancreatobiliary center. Patients treated by the hepatopancreatobiliary service for PHA were retrospectively reviewed from 2008 through 2018. The algorithm defined intended first-line treatment as antibiotics for type I abscesses (<3 cm), PD for type II (≥3, unilocular), and surgical intervention (minimally invasive drainage or resection, when possible) for type III (≥3 cm, multilocular). Outcomes were compared between patients who received first-line treatment following the algorithm versus alternate therapy. Of 330 patients with PHA, 201 met inclusion criteria. Type III abscesses had significantly lower failure following algorithmic approach with surgery compared with PD (4% vs 28%, P = 0.018). Type II abscesses failed first-line PD in 27 per cent (13/48) with 11 patients requiring surgical rescue, whereas first-line surgery failed in only 13 per cent (2/15). No deaths occurred after any surgical intervention, and there was no statistical difference in major complications between first-line surgical intervention and PD for type II or III abscesses. These results support the algorithmic approach and demonstrate that minimally invasive surgical intervention is a safe and effective modality for large PHA. We recommend that select patients with large, complex abscesses should be considered for a first-line minimally invasive surgical approach depending on surgical experience and available resources.

Regionalization of complex surgical care has increased interhospital transfers to quaternary centers within large health-care systems. Risk-based patient selection is imperative to improve resource allocation without compromising care. This study aimed to develop predictive models for identifying low-risk patients for transfer to a fully integrated satellite hepatopancreatobiliary (HPB) service in the northeast region of the health-care system. HPB transfers to the quaternary center over 15 months from hospitals in proximity to the satellite HPB center. A predictive tool was developed based on simple pretransfer variables and outcomes for 30-day major complications (Clavien grade ≥ 3), readmission, and mortality. Thresholds for “low risk” were set at different SDs below mean for each model. Predictive models were developed from 51 eligible northeast region patient transfers for major complications (Brier score 0.1948, receiver operator characteristic (ROC) 0.7123, P = 0.0009), readmission (Brier score 0.0615, ROC 0.7368, P = 0.0020), and mortality (Brier score 0.0943, ROC 0.7989, P = 0.0023). Thresholds set from 2 SD below the mean for all models identified 2 as “low risk.” Adjusting the threshold for the serious complication model to only 1 SD below the mean increased the “low-risk” cohort to five patients. These models demonstrate an easy-to-use tool to assist surgeons in identifying low-risk patients for diversion to a fully integrated satellite center. Improved interhospital transfers within a region could begin a transition from centers of excellence toward health-care systems of excellence.

Postoperative laboratory testing is an underrecognized but substantial contributor to health-care costs. We aimed to develop and validate a clinically meaningful laboratory (CML) protocol with individual risk stratification using generalizable and institution-specific predictive analytics to reduce laboratory testing and maximize cost savings for low-risk patients. An institutionally based risk model was developed for pancreaticoduodenectomy and hepatectomy, and an ACS-NSQIP®–based model was developed for distal pancreatectomy. Patients were stratified in each model to the CML by individual risk of major complications, readmission, or death. Clinical outcomes and estimated cost savings were compared with those of a historical cohort with standard of care. Over 34 months, 394 patients stratified to the CML for pancreaticoduodenectomy or hepatectomy saved an estimated $803,391 (44.4%). Over 13 months, 52 patients stratified to the CML for distal pancreatectomy saved an estimated $81,259 (30.5%). Clinical outcomes for 30-day major complications, readmission, and mortality were unchanged after implementation of either model. Predictive analytics can target low-risk patients to reduce laboratory testing and improve cost savings, regardless of whether an institutional or a generalized risk model is implemented. Broader application is important in patient-centered health care and should transition from predictive to prescriptive analytics to guide individual care in real time.