Preoperative CA 19-9 Predicts Disease Progression in Colorectal Peritoneal Metastases Treated with Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy: An Analysis from the US HIPEC Collaborative.

INTRODUCTION: Patients with colorectal peritoneal metastases (CRPM) are increasingly treated with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC). Unfortunately, data identifying preoperative risk factors for poor oncologic outcomes after this procedure are limited. We aimed to determine the prognostic value of preoperative CEA, CA 125, and CA 19-9 on disease progression after CRS/HIPEC. METHODS: Patients with CRPM treated with curative intent CRS/HIPEC from 12 participating sites in the United States from 2000 to 2017 were identified. Progression-free survival (PFS), defined as disease progression or recurrence, was the primary outcome. RESULTS: In 279 patients who met inclusion criteria, the rate of disease progression was 63.8%, with a median PFS of 11 months (interquartile range [IQR] 5-20). Elevated CA 19-9 was associated with dismal PFS at 2 years (8.9% elevated vs. 30% not elevated, p < 0.01). In 113 patients who underwent upfront CRS/HIPEC, CA 19-9 emerged as the sole tumor marker independently predictive of worse PFS (hazard ratio [HR] 2.88, p = 0.048). In the subgroup of patients who had received neoadjuvant therapy (NAT), no variable was independently predictive of PFS. CA 19-9 levels over 37 U/ml were highly specific for accelerated disease progression after CRS/HIPEC. Lastly, there was no association between PFS and elevated CEA or CA 125. CONCLUSIONS: Elevated CA 19-9 is associated with decreased PFS in patients with CRPM. While traditionally CEA is the main tumor marker assessed in colon cancer, we found that CA 19-9 may better inform preoperative risk stratification for poor oncologic outcomes in patients with CRPM. However, prospective studies are required to confirm this association.

Association of Neighborhood Disadvantage with Short- and Long-Term Outcomes After Pancreatectomy for Pancreatic Ductal Adenocarcinoma.

BACKGROUND: While lower socioeconomic status has been shown to correlate with worse outcomes in cancer care, data correlating neighborhood-level metrics with outcomes are scarce. We aim to explore the association between neighborhood disadvantage and both short- and long-term postoperative outcomes in patients undergoing pancreatectomy for pancreatic ductal adenocarcinoma (PDAC). PATIENTS AND METHODS: We retrospectively analyzed 243 patients who underwent resection for PDAC at a single institution between 1 January 2010 and 15 September 2021. To measure neighborhood disadvantage, the cohort was divided into tertiles by Area Deprivation Index (ADI). Short-term outcomes of interest were minor complications, major complications, unplanned readmission within 30 days, prolonged hospitalization, and delayed gastric emptying (DGE). The long-term outcome of interest was overall survival. Logistic regression was used to test short-term outcomes; Cox proportional hazards models and Kaplan-Meier method were used for long-term outcomes. RESULTS: The median ADI of the cohort was 49 (IQR 32-64.5). On adjusted analysis, the high-ADI group demonstrated greater odds of suffering a major complication (odds ratio [OR], 2.78; 95% confidence interval [CI], 1.26-6.40; p = 0.01) and of an unplanned readmission (OR, 3.09; 95% CI, 1.16-9.28; p = 0.03) compared with the low-ADI group. There were no significant differences between groups in the odds of minor complications, prolonged hospitalization, or DGE (all p > 0.05). High ADI did not confer an increased hazard of death (p = 0.63). CONCLUSIONS: We found that worse neighborhood disadvantage is associated with a higher risk of major complication and unplanned readmission after pancreatectomy for PDAC.

Multiomic Characterization Reveals a Distinct Molecular Landscape in Young-Onset Pancreatic Cancer.

PURPOSE: Using a real-world database with matched genomic-transcriptomic molecular data, we sought to characterize the distinct molecular correlates underlying clinical differences between patients with young-onset pancreatic cancer (YOPC; younger than 50 years) and patients with average-onset pancreatic cancer (AOPC; 70 years and older). METHODS: We analyzed matched whole-transcriptome and DNA sequencing data from 2,430 patient samples (YOPC, n = 292; AOPC, n = 2,138) from the Caris Life Sciences database (Phoenix, AZ). Immune deconvolution was performed using the quanTIseq pipeline. Overall survival (OS) data were obtained from insurance claims (n = 4,928); Kaplan-Meier estimates were calculated for age- and molecularly defined cohorts. Significance was determined as FDR-corrected P values (Q) < .05. RESULTS: Patients with YOPC had higher proportions of mismatch repair-deficient/microsatellite instability-high, BRCA2-mutant, and PALB2-mutant tumors compared with patients with AOPC, but fewer SMAD4-, RNF43-, CDKN2A-, and SF3B1-mutant tumors. Notably, patients with YOPC demonstrated significantly lower incidence of KRAS mutations compared with patients with AOPC (81.3% v 90.9%; Q = .004). In the KRAS wild-type subset (n = 227), YOPC tumors demonstrated fewer TP53 mutations and were more likely driven by NRG1 and MET fusions, whereas BRAF fusions were exclusively observed in patients with AOPC. Immune deconvolution revealed significant enrichment of natural killer cells, CD8+ T cells, monocytes, and M2 macrophages in patients with YOPC relative to patients with AOPC, which corresponded with lower rates of HLA-DPA1 homozygosity. There was an association with improved OS in patients with YOPC compared with patients with AOPC with KRAS wild-type tumors (median, 16.2 [YOPC-KRASWT] v 10.6 [AOPC-KRASWT] months; P = .008) but not KRAS-mutant tumors (P = .084). CONCLUSION: In this large, real-world multiomic characterization of age-stratified molecular differences in pancreatic ductal adenocarcinoma, YOPC is associated with a distinct molecular landscape that has prognostic and therapeutic implications.

Venous Thromboembolism Chemoprophylaxis Adherence Rates After Major Cancer Surgery.

Importance: Venous thromboembolism (VTE) represents a major source of preventable morbidity and mortality and is a leading cause of death in the US after cancer surgery. Previous research demonstrated variability in VTE chemoprophylaxis prescribing, although it is unknown how these rates compare with performance in the Veterans Health Administration (VHA). Objective: To determine VTE rates after cancer surgery, as well as rates of inpatient and outpatient (posthospital discharge) chemoprophylaxis adherence within the VHA. Design, Setting, and Participants: This retrospective cohort study within 101 hospitals of the VHA health system included patients aged 41 years or older without preexisting bleeding disorders or anticoagulation usage who underwent surgical treatment for cancer with general surgery, thoracic surgery, or urology between January 1, 2015, and December 31, 2022. The VHA Corporate Data Warehouse, Pharmacy Benefits Management database, and the Veterans Affairs Surgical Quality Improvement Program database were used to identify eligible patients. Data analysis was conducted between January 2022 and July 2023. Exposures: Inpatient surgery for cancer with general surgery, thoracic surgery, or urology. Main Outcomes and Measures: Rates of postoperative VTE events within 30 days of surgery and VTE chemoprophylaxis adherence were determined. Multivariable Poisson regression was used to determine incidence-rate ratios of inpatient and postdischarge chemoprophylaxis adherence by surgical specialty. Results: Overall, 30 039 veterans (median [IQR] age, 67 [62-71] years; 29 386 men [97.8%]; 7771 African American or Black patients [25.9%]) who underwent surgery for cancer and were at highest risk for VTE were included. The overall postoperative VTE rate was 1.3% (385 patients) with 199 patients (0.7%) receiving a diagnosis during inpatient hospitalization and 186 patients (0.6%) receiving a diagnosis postdischarge. Inpatient chemoprophylaxis was ordered for 24 139 patients (80.4%). Inpatient chemoprophylaxis ordering rates were highest for patients who underwent procedures with general surgery (10 102 of 10 301 patients [98.1%]) and lowest for patients who underwent procedures with urology (11 471 of 17 089 patients [67.1%]). Overall, 3142 patients (10.5%) received postdischarge chemoprophylaxis, with notable variation by specialty. Conclusions and Relevance: These findings indicate the overall VTE rate after cancer surgery within the VHA is low, VHA inpatient chemoprophylaxis rates are high, and postdischarge VTE chemoprophylaxis prescribing is similar to that of non-VHA health systems. Specialty and procedure variation exists for chemoprophylaxis and may be justified given the low risks of overall and postdischarge VTE.

Prediction of Early Recurrence Following CRS/HIPEC in Patients With Disseminated Appendiceal Cancer.

INTRODUCTION: In patients with disseminated appendiceal cancer (dAC) who underwent cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC), characterizing and predicting those who will develop early recurrence could provide a framework for personalizing follow-up. This study aims to: (1) characterize patients with dAC that are at risk for recurrence within 2 y following of CRS ± HIPEC (early recurrence; ER), (2) utilize automated machine learning (AutoML) to predict at-risk patients, and (3) identifying factors that are influential for prediction. METHODS: A 12-institution cohort of patients with dAC treated with CRS ± HIPEC between 2000 and 2017 was used to train predictive models using H2O.ai's AutoML. Patients with early recurrence (ER) were compared to those who did not have recurrence or presented with recurrence after 2 y (control; C). However, 75% of the data was used for training and 25% for validation, and models were 5-fold cross-validated. RESULTS: A total of 949 patients were included, with 337 ER patients (35.5%). Patients with ER had higher markers of inflammation, worse disease burden with poor response, and received greater intraoperative fluids/blood products. The highest performing AutoML model was a Stacked Ensemble (area under the curve = 0.78, area under the curve precision recall = 0.66, positive predictive value = 85%, and negative predictive value = 63%). Prediction was influenced by blood markers, operative course, and factors associated with worse disease burden. CONCLUSIONS: In this multi-institutional cohort of dAC patients that underwent CRS ± HIPEC, AutoML performed well in predicting patients with ER. Variables suggestive of poor tumor biology were the most influential for prediction. Our work provides a framework for identifying patients with ER that might benefit from shorter interval surveillance early after surgery.

NCCN Guidelines® Insights: Biliary Tract Cancers, Version 2.2023.

In 2023, the NCCN Guidelines for Hepatobiliary Cancers were divided into 2 separate guidelines: Hepatocellular Carcinoma and Biliary Tract Cancers. The NCCN Guidelines for Biliary Tract Cancers provide recommendations for the evaluation and comprehensive care of patients with gallbladder cancer, intrahepatic cholangiocarcinoma, and extrahepatic cholangiocarcinoma. The multidisciplinary panel of experts meets at least on an annual basis to review requests from internal and external entities as well as to evaluate new data on current and emerging therapies. These Guidelines Insights focus on some of the recent updates to the NCCN Guidelines for Biliary Tract Cancers as well as the newly published section on principles of molecular testing.

Validation of the AJCC 8th Edition Staging System for Disseminated Appendiceal Cancer Patients Treated with Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy: A Multi-institutional Analysis.

BACKGROUND: The AJCC 8th edition stratifies stage IV disseminated appendiceal cancer (dAC) patients based on grade and pathology. This study was designed to externally validate the staging system and to identify predictors of long-term survival. METHODS: A 12-institution cohort of dAC patients treated with CRS ± HIPEC was retrospectively analyzed. Overall survival (OS) and recurrence-free survival (RFS) were analyzed by using Kaplan-Meier and log-rank tests. Univariate and multivariate cox-regression was performed to assess factors associated with OS and RFS. RESULTS: Among 1009 patients, 708 had stage IVA and 301 had stage IVB disease. Median OS (120.4 mo vs. 47.2 mo) and RFS (79.3 mo vs. 19.8 mo) was significantly higher in stage IVA compared with IVB patients (p < 0.0001). RFS was greater among IVA-M1a (acellular mucin only) than IV M1b/G1 (well-differentiated cellular dissemination) patients (NR vs. 64 mo, p = 0.0004). Survival significantly differed between mucinous and nonmucinous tumors (OS 106.1 mo vs. 41.0 mo; RFS 46.7 mo vs. 21.2 mo, p < 0.05), and OS differed between well, moderate, and poorly differentiated (120.4 mo vs. 56.3 mo vs. 32.9 mo, p < 0.05). Both stage and grade were independent predictors of OS and RFS on multivariate analysis. Acellular mucin and mucinous histology were associated with better OS and RFS on univariate analysis only. CONCLUSIONS: AJCC 8th edition performed well in predicting outcomes in this large cohort of dAC patients treated with CRS ± HIPEC. Separation of stage IVA patients based on the presence of acellular mucin improved prognostication, which may inform treatment and long-term, follow-up strategies.

Multi-omic characterization reveals a distinct molecular landscape in young-onset pancreatic cancer.

Purpose: Using a real-world database with matched genomic-transcriptomic molecular data, we sought to characterize the distinct molecular correlates underlying clinical differences between young-onset pancreatic cancer (YOPC; <50-yrs.) and average-onset pancreatic cancer (AOPC; ≥70-yrs.) patients. Methods: We analyzed matched whole-transcriptome and DNA sequencing data from 2430 patient samples (YOPC, n=292; AOPC, n=2138) from the Caris Life Sciences database (Phoenix, AZ). Immune deconvolution was performed using the quanTIseq pipeline. Overall survival (OS) data was obtained from insurance claims (n=4928); Kaplan-Meier estimates were calculated for age-and molecularly-defined cohorts. Significance was determined as FDR-corrected P -values ( Q )<0.05. Results: YOPC patients had higher proportions of mismatch repair-deficient (dMMR)/microsatellite instability-high (MSI-H), BRCA2 -mutant, and PALB2 -mutant tumors compared with AOPC patients, but fewer SMAD4-, RNF43-, CDKN2A- , and SF3B1- mutant tumors. Notably, YOPC patients demonstrated significantly lower incidence of KRAS mutations compared with AOPC patients (81.3% vs. 90.9%; Q =0.004). In the KRAS- wildtype subset (n=227), YOPC tumors demonstrated fewer TP53 mutations and were more likely driven by NRG1 and MET fusions, while BRAF fusions were exclusively observed in AOPC patients. Immune deconvolution revealed significant enrichment of natural killer (NK) cells, CD8 + T cells, monocytes, and M2 macrophages in YOPC patients relative to AOPC patients, which corresponded with lower rates of HLA-DPA1 homozygosity. There was an association with improved OS in YOPC patients compared with AOPC patients with KRAS -wildtype tumors (median 16.2 [YOPC- KRAS WT ] vs. 10.6 [AOPC- KRAS WT ] months; P =0.008) but not KRAS -mutant tumors ( P =0.084). Conclusion: In this large, real-world multi-omic characterization of age-stratified molecular differences in PDAC, YOPC is associated with a distinct molecular landscape that has prognostic and therapeutic implications.

Conditional Survival Following Cytoreductive Surgery with Hyperthermic Intraperitoneal Chemotherapy for Peritoneal Surface Malignancies: An Analysis from the US HIPEC Collaborative.

INTRODUCTION: The long-term prognosis of patients who undergo cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) for peritoneal surface malignancies (PSM) varies considerably on the basis of histological and operative factors. While overall survival (OS) estimates are used to inform adjuvant therapy and surveillance strategies, conditional survival may provide more clinically relevant estimates of prognosis by accounting for disease-free time elapsed. PATIENTS AND METHODS: All patients from 12 academic institutions who underwent CRS ± HIPEC for PSM from 2000 to 2017 were retrospectively analyzed. OS and disease-free survival (DFS) rates were calculated using the Kaplan-Meier method while conditional overall (COS) and conditional disease-free survival (CDFS) rates were calculated at 1, 2, or 3 years from surgery for different tumor histologies. RESULTS: Overall, 1610 patients underwent CRS ± HIPEC. Among patients with benign appendiceal mucinous tumors (N = 460), 5-year OS and COS at 3 years were 92.1% and 96.3% (Δ4.2%), respectively. For patients with well-differentiated appendiceal cancers (N = 400), 5-year OS and COS at 3 years were 76.3% and 88.3% (Δ12.0%), respectively. For patients with high-grade appendiceal cancers (N = 258), 5-year OS and COS at 3 years were 43.8% and 75.4% (Δ31.6%), respectively. For patients with colorectal cancers (N = 362), 5-year OS and COS at 3 years were 31.8% and 67.3% (Δ35.5%), respectively. For patients with peritoneal mesothelioma (N = 130), 5-year OS and COS at 3 years were 67.6% and 89.7% (Δ22.1%), respectively. Similar trends were observed for DFS/CDFS. CONCLUSION: The conditional survival of patients undergoing CRS ± HIPEC for PSM is associated with tumor histology. COS and CDFS provide a more accurate, dynamic estimate of survival than OS and DFS, especially for patients with more aggressive histologies.

Influence of insurance status on the postoperative outcomes of cytoreductive surgery and HIPEC.

BACKGROUND: Cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) is increasingly performed for peritoneal surface malignancies but remains associated with significant morbidity. Scant research is available regarding the impact of insurance status on postoperative outcomes. METHODS: Patients undergoing CRS/HIPEC between 2000 and 2017 at 12 participating sites in the US HIPEC Collaborative were identified. Univariate and multivariate analyses were used to compare the baseline characteristics, operative variables, and postoperative outcomes of patients with government, private, or no insurance. RESULTS: Among 2268 patients, 699 (30.8%) had government insurance, 1453 (64.0%) had private, and 116 (5.1%) were uninsured. Patients with government insurance were older, more likely to be non-white, and comorbid (p < 0.05). Patients with government (OR: 2.25, CI: 1.50-3.36, p < 0.001) and private (OR: 1.69, CI: 1.15-2.49, p = 0.008) insurance had an increased risk of complications on univariate analysis. There was no independent relationship on multivariate analysis. An American Society of Anesthesiologists score of 3 or 4, peritoneal carcinomatosis index score >15, completeness of cytoreduction score >1, and nonhome discharge were factors independently associated with a postoperative complication. CONCLUSION: While there were differences in postoperative outcomes between the three insurance groups on univariate analysis, there was no independent association between insurance status and postoperative complications after CRS/HIPEC.

Do Lymph Node Metastases Matter in Appendiceal Cancer with Peritoneal Carcinomatosis? A US HIPEC Collaborative Study.

BACKGROUND: Whether formal regional lymph node (LN) evaluation is necessary for patients with appendiceal adenocarcinoma (AA) who have peritoneal metastases is unclear. The aim of this study was to evaluate the prognostic value of LN metastases on survival in patients treated with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS-HIPEC). METHODS: A retrospective analysis of the US HIPEC collaborative, a multi-institutional consortium comprising 12 high-volume centers, was performed to identify patients with AA who underwent CRS-HIPEC with adequate LN sampling (≥ 12 LNs). RESULTS: Two hundred-fifty patients with AA who underwent CRS-HIPEC were included. Outcomes were compared between LN - and LN + disease. Baseline patient characteristics between groups were similar, with most patients undergoing complete cytoreduction (0/1: 86.0% vs. 76.8%, p = 0.08), respectively. More adverse tumor factors were found in patients with LN + disease, including poor differentiation, signet ring cells, and lymphovascular invasion. Multivariate analysis of overall survival (OS) found LN + disease was independently associated with worse OS (HR: 2.82 95%CI: 1.25-6.34, p = 0.01), even after correction for receipt of systemic therapy. On Kaplan-Meier analysis, median OS was lower in patients with LN + disease (25.9 months vs. 91.4 months, p < 0.01). LN + disease remained associated with poor OS following propensity score matching (HR: 4.98 95%CI: 1.72-14.40, p < 0.01) and in patients with PCI ≥ 20 (HR: 3.68 95%CI: 1.54-8.80, p < 0.01). CONCLUSIONS: In this large multi-institutional study of patients with AA undergoing CRS-HIPEC, LN status remained associated with worse OS even in the setting of advanced peritoneal carcinomatosis. Formal LN evaluation should be performed for most patients with AA undergoing CRS-HIPEC.

Prognostic Utility of Systemic Immune-Inflammation Index After Resection of Extrahepatic Cholangiocarcinoma: Results from the U.S. Extrahepatic Biliary Malignancy Consortium.

BACKGROUND: We sought to define the association of the systemic immune inflammation index (SII) with prognosis and adjuvant therapy benefit among patients undergoing resection of extrahepatic cholangiocarcinoma (eCCA). METHODS: The impact of SII on overall (OS) and recurrence-free survival (RFS) following resection of eCCA was assessed and compared with other inflammatory markers and traditional prognostic factors. Propensity score matching (PSM) was used to determine the impact of adjuvant therapy (AT) on OS and RFS relative to low versus high SII. RESULTS: Patients with high versus low SII had worse 5-year OS (15.9% vs. 27.9%) and RFS (12.4% vs. 20.9%) (both p < 0.01). On multivariate analysis, high SII remained associated with worse OS (HR = 1.50, 95% CI 1.20-1.87) and RFS (HR = 1.46, 95% CI 1.18-1.81). Patients with T1/2 disease and a high-SII had worse 5-year OS versus individuals with T3/4 disease and low-SII (5-year OS: T1/2 & low-SII 35.6%, T1/2 & high-SII 16.4%, T3/4 & low-SII 22.1%, T3/4 & high-SII 15.6%, p < 0.01). Similarly, 5-year OS was comparable among individuals with N0 and high-SII versus N1 and low-SII (5-year OS: N0 & high-SII 23.2%, N1 and low-SII 19.8%, p = 0.95). On PSM, AT improved OS and RFS among patients with high SII (5-year OS: 22.5% vs. 12.3%, p < 0.01, 5-year RFS: 19.0% vs. 12.5%; p = 0.01) but not individuals with low SII (5-year OS: 22.9% vs. 26.9%; p = 0.98, 5-year RFS: 18.5% vs. 19.9%; p = 0.94). CONCLUSIONS: SII was independently associated with postoperative OS and RFS following curative-intent resection of eCCA. High SII up-staged patients relative T- and N-categories and identified patients with high SII as the most likely to benefit from AT.

ASO Author Reflections: An Opportunity to Provide Cost-Conscious Care for Patients Undergoing CRS-HIPEC.

The substantial cost of care for patients undergoing cytoreductive surgery with hyperthermic intraperitoneal chemotherapy can be augmented by changes in practice patterns. This report discusses the recent publication of the authors' group and their thoughts on cost-conscious cancer care.

Rethinking Routine Intensive Care After Cytoreductive Surgery With Heated Intraperitoneal Chemotherapy: The Fiscal Argument.

BACKGROUND: Patients undergoing cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC) are frequently admitted to the intensive care unit (ICU) for mitigation of potential complications, although ICU length of stay (LOS) is a significant driver of cost. This study asked whether a fiscal argument could be made for the selective avoidance of ICU admission after CRS/HIPEC. METHODS: Prospective data for select low-risk patients (e.g., lower peritoneal cancer index [PCI]) admitted to the intermediate care unit (IMC) instead of the ICU after CRS/HIPEC were matched with a historic cohort routinely admitted to the ICU. Cohort comparisons and the impact of the intervention on cost were assessed. RESULTS: The study matched 81 CRS/HIPEC procedures to form a cohort of 49 pre- and 15 post-intervention procedures for patients with similar disease burdens (mean PCI, 8 ± 6.7 vs. 7 ± 5.1). The pre-intervention patients stayed a median of 1 day longer in the ICU (1 day [IQR, 1-1 day] vs. 0 days [IQR, 0-0 days]) and had a longer LOS (8 days [IQR, 7-11 days] vs. 6 days [IQR, 5.5-9 days]). Complications and complication severity did not differ statistically. The median total hospital cost was lower after intervention ($30,845 [IQR, $30,181-$37,725] vs. $41,477 [IQR, $33,303-$51,838]), driven by decreased indirect fixed cost ($8984 [IQR, $8643-$11,286] vs. $14,314 [IQR, $12,206-$18,266]). In a weighted multiple variable linear regression analysis, the intervention was associated with a savings of $2208.68 per patient. CONCLUSIONS: Selective admission to the IMC after CRS/HIPEC was associated with $2208.68 in savings per patient without added risk. In this era of cost-conscious practice of medicine, these data highlight an opportunity to decrease cost by more than 5% for patients undergoing CRS/HIPEC.

Development and Validation of an Explainable Machine Learning Model for Major Complications After Cytoreductive Surgery.

Importance: Cytoreductive surgery (CRS) is one of the most complex operations in surgical oncology with significant morbidity, and improved risk prediction tools are critically needed. Machine learning models can potentially overcome the limitations of traditional multiple logistic regression (MLR) models and provide accurate risk estimates. Objective: To develop and validate an explainable machine learning model for predicting major postoperative complications in patients undergoing CRS. Design, Setting, and Participants: This prognostic study used patient data from tertiary care hospitals with expertise in CRS included in the US Hyperthermic Intraperitoneal Chemotherapy Collaborative Database between 1998 and 2018. Information from 147 variables was extracted to predict the risk of a major complication. An ensemble-based machine learning (gradient-boosting) model was optimized on 80% of the sample with subsequent validation on a 20% holdout data set. The machine learning model was compared with traditional MLR models. The artificial intelligence SHAP (Shapley additive explanations) method was used for interpretation of patient- and cohort-level risk estimates and interactions to define novel surgical risk phenotypes. Data were analyzed between November 2019 and August 2021. Exposures: Cytoreductive surgery. Main Outcomes and Measures: Area under the receiver operating characteristics (AUROC); area under the precision recall curve (AUPRC). Results: Data from a total 2372 patients were included in model development (mean age, 55 years [range, 11-95 years]; 1366 [57.6%] women). The optimized machine learning model achieved high discrimination (AUROC: mean cross-validation, 0.75 [range, 0.73-0.81]; test, 0.74) and precision (AUPRC: mean cross-validation, 0.50 [range, 0.46-0.58]; test, 0.42). Compared with the optimized machine learning model, the published MLR model performed worse (test AUROC and AUPRC: 0.54 and 0.18, respectively). Higher volume of estimated blood loss, having pelvic peritonectomy, and longer operative time were the top 3 contributors to the high likelihood of major complications. SHAP dependence plots demonstrated insightful nonlinear interactive associations between predictors and major complications. For instance, high estimated blood loss (ie, above 500 mL) was only detrimental when operative time exceeded 9 hours. Unsupervised clustering of patients based on similarity of sources of risk allowed identification of 6 distinct surgical risk phenotypes. Conclusions and Relevance: In this prognostic study using data from patients undergoing CRS, an optimized machine learning model demonstrated a superior ability to predict individual- and cohort-level risk of major complications vs traditional methods. Using the SHAP method, 6 distinct surgical phenotypes were identified based on sources of risk of major complications.

Intraoperative Pancreatic Neck Margin Assessment During Pancreaticoduodenectomy for Pancreatic Adenocarcinoma in the Era of Neoadjuvant Therapy: A Multi-institutional Analysis from the Central Pancreatic Consortium.

BACKGROUND: Data regarding the survival impact of converting frozen-section (FS):R1 pancreatic neck margins to permanent section (PS):R0 by additional resection (i.e., converted-R0) during upfront pancreaticoduodenectomy for pancreatic ductal adenocarcinoma (PDAC) are conflicting. The impact of neoadjuvant therapy on this practice and its relationship with overall survival (OS) is incompletely understood. METHODS: We reviewed PDAC patients (80% borderline resectable/locally advanced [BR/LA]) undergoing pancreaticoduodenectomy after neoadjuvant therapy at seven, academic, high-volume centers (2010-2018). Multivariable models examined the association of PS:R0, PS:R1, and converted-R0 margins with OS. RESULTS: Of 272 patients receiving at least 2 (median 4) cycles of neoadjuvant chemotherapy (71% mFOLFIRINOX or gemcitabine/nab-paclitaxel) and undergoing pancreaticoduodenectomy with intraoperative frozen-section assessment of the transected pancreatic neck margin, PS:R0 (n = 220, 80.9%) was observed in a majority of patients; 18 patients (6.6%) had converted-R0 margins following additional resection, whereas 34 patients (12.5%) had persistently positive PS:R1 margins. At a median follow-up of 42 months, PS:R0 resection was associated with improved OS compared with either converted-R0 or PS:R1 resection (median 25 vs. 14 vs. 16 months, respectively; p = 0.023), with no survival difference between the converted-R0 and PS:R1 groups (p = 0.9). On Cox regression, SMA margin positivity (hazard ratio 2.2, p = 0.012), but not neck margin positivity (hazard ratio 1.2, p = 0.65), was associated with worse OS. CONCLUSIONS: In this multi-institutional cohort of predominantly BR/LA PDAC patients undergoing pancreaticoduodenectomy following modern neoadjuvant therapy, pursuing a negative neck margin intraoperatively if the initial margin is positive does not appear to be associated with improved survival.