Prognostic factors in patients receiving surgery and radiation therapy for retroperitoneal sarcoma: A machine-learning analysis.

BACKGROUND: The addition of radiation therapy to surgery for retroperitoneal sarcoma remains controversial. Improved patient selection may help identify optimal candidates for multimodality treatment. The aim of this analysis was to define prognostic factors among patients who receive radiation therapy and surgery to aid in patient selection for multimodal therapy. METHODS: Patients who received radiation therapy and underwent curative-intent resection for retroperitoneal sarcoma between 2004 and 2016 were identified from a national cohort in the United States (National Cancer Database). A machine-based classification and regression tree model was used to generate similar groups of patients relative to overall survival based on preoperative factors. RESULTS: A total of 1,443 patients received radiation therapy in addition to surgery. Median age was 61 years old and 55.0% were female. Most patients (66%) received care at an academic or integrated network cancer program. With a median follow-up of 84 months, receipt of radiation therapy was not associated with improved overall survival (P = .81). Classification and regression tree analysis revealed a significant association between overall survival and American Joint Committee on Cancer stage group, age, tumor histology, and Charlson comorbidity score. Application of these parameters via machine learning stratified patients into 5 cohorts with distinct survival outcomes. In the most favorable cohort (Cohort 1: American Joint Committee on Cancer stage group ≤II, age ≤61, histology including fibrosarcoma, well differentiated liposarcoma, myxoid liposarcoma, and leiomyosarcoma), the 5-year overall survival was 81.7% and median overall survival was not reached; in the least favorable cohort (Cohort 6: American Joint Committee on Cancer stage group >II, age >68) where the 5-year survival was 41.3% and median overall survival was 45.2 months (P < .001 versus Cohort 1). CONCLUSION: In the absence of a defined survival benefit, patients with advanced American Joint Committee on Cancer stage group, older age, and medical comorbidities have relatively unfavorable overall survival after combined modality therapy and therefore stand the least to gain from the addition of radiation therapy to surgery. In contrast, younger patients with good performance status and retroperitoneal sarcoma histologies with a higher propensity for local recurrence may have the greatest opportunity to benefit from radiation therapy.

Optimal Pancreatic Surgery: Are We Making Progress in North America?

OBJECTIVE: Our aims were to assess North American trends in the management of patients undergoing pancreatoduodenectomy (PD) and distal pancreatectomy (DP), and to quantify the delivery of optimal pancreatic surgery. BACKGROUND: Morbidity after pancreatectomy remains unacceptably high. Recent literature suggests that composite measures may more accurately define surgical quality. METHODS: The 2013 to 2017 American College of Surgeons National Surgical Quality Improvement Program Participant Use Files were queried to identify patients undergoing PD (N = 16,222) and DP (N = 7946). Patient, process, procedure, and 30-day postoperative outcome variables were analyzed over time. Optimal pancreatic surgery was defined as the absence of postoperative mortality, serious morbidity, percutaneous drainage, and reoperation while achieving a length of stay equal to or less than the 75th percentile (12 days for PD and 7 days for DP) with no readmissions. Risk-adjusted time-trend analyses were performed using logistic regression, and the threshold for statistical significance was P ≤ 0.05. RESULTS: The use of minimally invasive PD did not change over time, but robotic PD increased (2.5 to 4.2%; P < 0.001) and laparoscopic PD decreased (5.8% to 4.3%; P < 0.02). Operative times decreased (P < 0.05) and fewer transfusions were administered (P < 0.001). The percentage of patients with a drain fluid amylase checked on postoperative day 1 increased (P < 0.001), and a greater percentage of surgical drains were removed by postoperative day 3 (P < 0.001). Overall morbidity (P < 0.02), mortality (P < 0.05), and postoperative length of stay (P = 0.002) decreased. Finally, the rate of optimal pancreatic surgery increased for PD (53.7% to 56.9%; P < 0.01) and DP (53.3% to 58.5%; P < 0.001), and alspo for patients with pancreatic cancer (P < 0.01). CONCLUSIONS: From 2013 to 2017, pre, intra, and perioperative pancreatectomy processes have evolved, and multiple postoperative outcomes have improved. Thus, in 4 years, optimal pancreatic surgery in North America has increased by 3% to 5%.

500 Minimally Invasive Robotic Pancreatoduodenectomies: One Decade of Optimizing Performance.

OBJECTIVES: This study aims to present the outcomes of our decade-long experience of robotic pancreatoduodenectomy and provide insights into successful program implementation. BACKGROUND: Despite significant improvement in mortality over the past 30 years, morbidity following open pancreatoduodenectomy remains high. We implemented a minimally invasive pancreatic surgery program based on the robotic platform as one potential method of improving outcomes for this operation. METHODS: A retrospective review of a prospectively maintained institutional database was performed to identify patients who underwent robotic pancreatoduodenectomy (RPD) between 2008 and 2017 at the University of Pittsburgh. RESULTS: In total, 500 consecutive RPDs were included. Operative time, conversion to open, blood loss, and clinically relevant postoperative pancreatic fistula improved early in the experience and have remained low despite increasing complexity of case selection as reflected by increasing number of patients with pancreatic cancer, vascular resections, and higher Charlson Comorbidity scores (all P<0.05). Operating room time plateaued after 240 cases at a median time of 391 minutes (interquartile rang 340-477). Major complications (Clavien >2) occurred in less than 24%, clinically relevant postoperative pancreatic fistula in 7.8%, 30- and 90-day mortality were 1.4% and 3.1% respectively, and median length of stay was 8 days. Outcomes were not impacted by integration of trainees or expansion of selection criteria. CONCLUSIONS: Structured implementation of robotic pancreatoduodenectomy can be associated with excellent outcomes. In the largest series of RPD, we establish benchmarks for the surgical community to consider when adopting this approach.

Robotic pancreatoduodenectomy with vascular resection: Outcomes and learning curve.

INTRODUCTION: The safety, efficacy, and learning curve for robotic pancreatoduodenecomy has been reported; however, the outcomes and learning curve of robotic pancreatoduodenecomy with vascular resections remain unknown. Our aim was to evaluate the outcomes of robotic pancreatoduodenecomy with vascular resections compared with robotic pancreatoduodenecomy without vascular resection and to identify the learning curve and benchmarks for improved performance during robotic pancreatoduodenecomy with vascular resections. METHODS: A retrospective review of consecutive patients who underwent robotic pancreatoduodenecomy with vascular resections and robotic pancreatoduodenecomy between 2011 and 2017. Patients were analyzed consecutively, and a cumulative sum analysis was performed to detect improvements in performance over time. RESULTS: Of 380 consecutive robotic pancreatoduodenecomy patients, 50 (13%) underwent robotic pancreatoduodenecomy with vascular resections. Compared with robotic pancreatoduodenecomy, robotic pancreatoduodenecomy with vascular resections were more likely to have had pancreatic adenocarcinoma (84% vs 42%) and had received neoadjuvant therapy (35% vs 65%, P < .01). Robotic pancreatoduodenecomy with vascular resections operative time revealed a steady, significant decrease (Rho = -0.38, p = .006) with marked initial improvement after the first 8 cases and maturation of the learning curve after 35 cases. A significant decrease in duration of the hospital stay was observed throughout the experience (Rho = -0.528, P < .0001), whereas margin status, pancreatic fistula, major morbidity, and mortality remained constant and comparable to robotic pancreatoduodenecomy alone. CONCLUSION: Robotic pancreatoduodenectomy with vascular resections is safe and feasible. For surgeons who have surpassed the learning curve of robotic pancreatoduodenectomy, it appears that improvements in performance of robotic pancreatoduodenecomy with vascular resections can be observed after 35 cases.

The Beneficial Effects of Minimizing Blood Loss in Pancreatoduodenectomy.

OBJECTIVE: The aim of this study was to elucidate the impact of intraoperative blood loss on outcomes following pancreatoduodenectomy (PD). BACKGROUND: The negative impact of intraoperative blood loss on outcomes in PD has long been suspected but not well characterized, particularly those factors that may be within surgeons' control. METHODS: From 2001 to 2015, 5323 PDs were performed by 62 surgeons from 17 institutions. Estimated blood loss (EBL) was discretized (0 to 300, 301 to 750, 751 to 1300, and >1300 mL) using optimal scaling methodology. Multivariable regression, adjusted for patient, surgeon, and institutional variables, was used to identify associations between EBL and perioperative outcomes. Factors associated with both increased and decreased EBL were elucidated. The relative impact of surgeon-modifiable contributors was estimated through beta coefficient standardization. RESULTS: The median EBL of the series was 400 mL [interquartile range (IQR) 250 to 600]. Intra-, post-, and perioperative transfusion rates were 15.8%, 24.8%, and 37.2%, respectively. Progressive EBL zones correlated with intra- but not postoperative transfusion in a dose-dependent fashion (P < 0.001), with a key threshold of 750 mL EBL (8.14% vs 40.9%; P < 0.001). Increasing blood loss significantly correlated with poor perioperative outcomes. Factors associated with increased EBL were trans-anastomotic stent placement, neoadjuvant chemotherapy, pancreaticogastrostomy reconstruction, multiorgan or vascular resection, and elevated operative time, of which 38.7% of the relative impact was "potentially modifiable" by the surgeon. Conversely, female sex, small duct, soft gland, minimally invasive approach, pylorus-preservation, biological sealant use, and institutional volume (≥67/year) were associated with decreased EBL, of which 13.6% was potentially under the surgeon's influence. CONCLUSION: Minimizing blood loss contributes to fewer intraoperative transfusions and better perioperative outcomes for PD. Improvements might be achieved by targeting modifiable factors that influence EBL.

Identification of an Optimal Cut-off for Drain Fluid Amylase on Postoperative Day 1 for Predicting Clinically Relevant Fistula After Distal Pancreatectomy: A Multi-institutional Analysis and External Validation.

OBJECTIVE: The aim of this study was to investigate the relationship between drain fluid amylase value on the first postoperative day (DFA1) and clinically relevant fistula (CR-POPF) after distal pancreatectomy (DP), and to identify the cut-off of DFA1 that optimizes CR-POPF prediction. BACKGROUND: DFA1 is a well-recognized predictor of CR-POPF after pancreatoduodenectomy, but its role in DP is largely unexplored. METHODS: DFA1 levels were correlated with CR-POPF in 2 independent multi-institutional sets of DP patients: developmental (n = 338; years 2012 to 2017) and validation cohort (n = 166; years 2006 to 2016). Cut-off choice was based on Youden index calculation, and its ability to predict CR-POPF occurrence was tested in a multivariable regression model adjusted for clinical, demographic, operative, and pathological variables. RESULTS: In the developmental set, median DFA1 was 1745 U/L and the CR-POPF rate was 21.9%. DFA1 correlated with CR-POPF with an area under the curve of 0.737 (P < 0.001). A DFA1 of 2000 U/L had the highest Youden index, with 74.3% sensitivity and 62.1% specificity. Patients in the validation cohort displayed different demographic and operative characteristics, lower values of DFA1 (784.5 U/L, P < 0.001), and reduced CR-POPF rate (10.2%, P < 0.001). However, a DFA1 of 2000 U/L had the highest Youden index in this cohort as well, with 64.7% sensitivity and 75.8% specificity. At multivariable analysis, DFA1 ≥2000 U/L was the only factor significantly associated with CR-POPF in both cohorts. CONCLUSION: A DFA1 of 2000 U/L optimizes CR-POPF prediction after DP. These results provide the substrate to define best practices and improve outcomes for patients receiving DP.

Risk Factors and Mitigation Strategies for Pancreatic Fistula After Distal Pancreatectomy: Analysis of 2026 Resections From the International, Multi-institutional Distal Pancreatectomy Study Group.

OBJECTIVE: To identify a clinical fistula risk score following distal pancreatectomy. BACKGROUND: Clinically relevant pancreatic fistula (CR-POPF) following distal pancreatectomy (DP) is a dominant contributor to procedural morbidity, yet risk factors attributable to CR-POPF and effective practices to reduce its occurrence remain elusive. METHODS: This multinational, retrospective study of 2026 DPs involved 52 surgeons at 10 institutions (2001-2016). CR-POPFs were defined by 2016 International Study Group criteria, and risk models generated using stepwise logistic regression analysis were evaluated by c-statistic. Mitigation strategies were assessed by regression modeling while controlling for identified risk factors and treating institution. RESULTS: CR-POPF occurred following 306 (15.1%) DPs. Risk factors independently associated with CR-POPF included: age (<60 yrs: OR 1.42, 95% CI 1.05-1.82), obesity (OR 1.54, 95% CI 1.19-2.12), hypoalbuminenia (OR 1.63, 95% CI 1.06-2.51), the absence of epidural anesthesia (OR 1.59, 95% CI 1.17-2.16), neuroendocrine or nonmalignant pathology (OR 1.56, 95% CI 1.18-2.06), concomitant splenectomy (OR 1.99, 95% CI 1.25-3.17), and vascular resection (OR 2.29, 95% CI 1.25-3.17). After adjusting for inherent risk between cases by multivariable regression, the following were not independently associated with CR-POPF: method of transection, suture ligation of the pancreatic duct, staple size, the use of staple line reinforcement, tissue patches, biologic sealants, or prophylactic octreotide. Intraoperative drainage was associated with a greater fistula rate (OR 2.09, 95% CI 1.51-3.78) but reduced fistula severity (P < 0.001). CONCLUSIONS: From this large analysis of pancreatic fistula following DP, CR-POPF occurrence cannot be reliably predicted. Opportunities for developing a risk score model are limited for performing risk-adjusted analyses of mitigation strategies and surgeon performance.

Characterization and Optimal Management of High-risk Pancreatic Anastomoses During Pancreatoduodenectomy.

OBJECTIVE: The aim of this study was to identify the optimal fistula mitigation strategy following pancreaticoduodenectomy. BACKGROUND: The utility of technical strategies to prevent clinically relevant postoperative pancreatic fistula (CR-POPF) following pancreatoduodenectomy (PD) may vary by the circumstances of the anastomosis. The Fistula Risk Score (FRS) identifies a distinct high-risk cohort (FRS 7 to 10) that demonstrates substantially worse clinical outcomes. The value of various fistula mitigation strategies in these particular high-stakes cases has not been previously explored. METHODS: This multinational study included 5323 PDs performed by 62 surgeons at 17 institutions. Mitigation strategies, including both technique related (ie, pancreatogastrostomy reconstruction; dunking; tissue patches) and the use of adjuvant strategies (ie, intraperitoneal drains; anastomotic stents; prophylactic octreotide; tissue sealants), were evaluated using multivariable regression analysis and propensity score matching. RESULTS: A total of 522 (9.8%) PDs met high-risk FRS criteria, with an observed CR-POPF rate of 29.1%. Pancreatogastrostomy, prophylactic octreotide, and omission of externalized stents were each associated with an increased rate of CR-POPF (all P < 0.001). In a multivariable model accounting for patient, surgeon, and institutional characteristics, the use of external stents [odds ratio (OR) 0.45, 95% confidence interval (95% CI) 0.25-0.81] and the omission of prophylactic octreotide (OR 0.49, 95% CI 0.30-0.78) were independently associated with decreased CR-POPF occurrence. In the propensity score matched cohort, an "optimal" mitigation strategy (ie, externalized stent and no prophylactic octreotide) was associated with a reduced rate of CR-POPF (13.2% vs 33.5%, P < 0.001). CONCLUSIONS: The scenarios identified by the high-risk FRS zone represent challenging anastomoses associated with markedly elevated rates of fistula. Externalized stents and omission of prophylactic octreotide, in the setting of intraperitoneal drainage and pancreaticojejunostomy reconstruction, provides optimal outcomes.