Machine learning to predict completion of treatment for pancreatic cancer.

BACKGROUND: Chemotherapy enhances survival rates for pancreatic cancer (PC) patients postsurgery, yet less than 60% complete adjuvant therapy, with a smaller fraction undergoing neoadjuvant treatment. Our study aimed to predict which patients would complete pre- or postoperative chemotherapy through machine learning (ML). METHODS: Patients with resectable PC identified in our institutional pancreas database were grouped into two categories: those who completed all intended treatments (i.e., surgery plus either neoadjuvant or adjuvant chemotherapy), and those who did not. We applied logistic regression with lasso penalization and an extreme gradient boosting model for prediction, and further examined it through bootstrapping for sensitivity. RESULTS: Among 208 patients, the median age was 69, with 49.5% female and 62% white participants. Most had an Eastern Cooperative Oncology Group (ECOG) performance status of ≤2. The PC predominantly affected the pancreatic head. Neoadjuvant and adjuvant chemotherapies were received by 26% and 47.1%, respectively, but only 49% completed all treatments. Incomplete therapy was correlated with older age and lower ECOG status. Negative prognostic factors included worsening diabetes, age, congestive heart failure, high body mass index, family history of PC, initial bilirubin levels, and tumor location in the pancreatic head. The models also flagged other factors, such as jaundice and specific cancer markers, impacting treatment completion. The predictive accuracy (area under the receiver operating characteristic curve) was 0.67 for both models, with performance expected to improve with larger datasets. CONCLUSIONS: Our findings underscore the potential of ML to forecast PC treatment completion, highlighting the importance of specific preoperative factors. Increasing data volumes may enhance predictive accuracy, offering valuable insights for personalized patient strategies.

Celiac artery stenosis and its impact on postoperative outcomes following pancreaticoduodenectomy: An institutional analysis.

BACKGROUND: The detection of a stenotic celiac artery (CA) typically mandates intraoperative revascularization during pancreaticoduodenectomy (PD) to preserve liver perfusion. The impact of CA stenosis on postoperative outcomes is unclear. This study evaluates whether CA stenosis (CAS) is associated with increased postoperative complications. METHODS: We conducted a retrospective analysis of PD patients from February 2014 to February 2022. Preoperative imaging assessed the CA lumen, categorizing it as patent, <50%, or ≥50% stenosis. Patients with narrowed SMA were excluded. Complications were categorized using the Clavien-Dindo system, and statistical analyses identified outcome differences. RESULTS: We included 427 patients in the study. Of these, 52 had CAS, and 311 had no-vessel stenosis (NVS). The median age of the CAS and NVS groups was 68 and 65 years, respectively. Postoperatively, 17.6% of patients with CAS exhibited delayed gastric emptying (DGE) versus 25.3% in the NVS group. Postoperative pancreatic fistula (POPF) was found in 13.5% of patients with CAS, compared with 23.7% without stenosis. The median length of hospital stay was shorter for patients with CAS (9 days) than for those with CAS (12 days). Severity-based classifications indicated higher complications in the no stenosis group and a 33.0% readmission rate within 30 days compared with 21.2% in CAS patients. However, none of these differences were statistically significant. CONCLUSIONS: Critical stenosis of the CA does not significantly affect postoperative outcomes following PD, suggesting preoperative correction of the narrowed CA may not be necessary. Further research is needed to confirm these findings.

Outcomes of minimally invasive vs. open pancreatoduodenectomies in pancreatic adenocarcinoma: analysis of ACS-NSQIP data.

INTRODUCTION: Pancreatoduodenectomy (PD) for pancreatic adenocarcinoma (PDAC) presents a significant challenge owing to its aggressive nature. Traditionally performed as open surgery, the advent of minimally invasive surgery (MIS) including laparoscopic and robotic techniques, offers a potential alternative. This study assessed the use and outcomes of MIS and open PD for PDAC treatment. METHODS: We analyzed ACS-NSQIP data (2015-2021) using regression models to compare patient outcomes across open PD, MIS PD, and conversions from MIS to open (MIS-O). RESULTS: Of 19,812 PDAC patients, 1,293 (6.53%) underwent MIS, 18,116 (91.44%) underwent open PD, and 403 (2.03%) underwent MIS converted to open PD (MIS-O). The MIS rate increased from 6.1% to 9.2%. Black patients had a higher MIS-O rate (RR, 1.55; p = 0.025). Open PD was associated with more severe conditions (ASA ≥ III, malnutrition) and prior radiation therapy. MIS patients more often had neoadjuvant chemotherapy. Complex procedures, such as vein resection, favored open PD. Need for arterial resection was associated with MIS-O (RR, 2.11; p = 0.012), and operative time was significantly associated with MIS (OR: 4.32, 95% CI: 3.43-5.43, p-value: < 0.001) No differences in the overall morbidity or 30-day mortality were observed. MIS led to shorter stays but higher risks of reoperation and pulmonary embolism. MIS-O increased the delayed gastric emptying rate (RR, 1.79; p < 0.001). CONCLUSION: During 2015-2021, an increasing number of patients with PDAC are undergoing MIS PD. Morbidity and mortality did not differ between open and MIS PD. MIS was performed more frequently in patients with better nutritional status and lower ASA, or when vascular resection was not anticipated. In well selected patients, short-term outcomes of MIS and open PD seem similar.

Impact of postoperative pancreatic fistula on outcomes in pancreatoduodenectomy: a comprehensive analysis of American College of Surgeons National Surgical Quality Improvement Program data.

BACKGROUND: Pancreatoduodenectomy (PD) is a major surgical procedure associated with significant risks, particularly postoperative pancreatic fistula (POPF). Studies have highlighted the importance of certain risk factors for POPF, which are crucial for surgical decision-making and the management of high-risk patients undergoing PD. This study aimed to assess the surgical outcomes of patients undergoing PD who met the International Study Group of Pancreatic Surgery - Class D (ISGPS-D) criteria. METHODS: This study analyzed American College of Surgeons National Surgical Quality Improvement Program data (2014-2021) for patients undergoing ISGPS-D PD, classified as having a soft pancreatic texture and a pancreatic duct of ≤3 mm. This study focused on mortality rates and the correlation between several factors and POPF (ISGPS grade B/C). RESULTS: From 5964 patients who underwent PD and met the ISGPS-D criteria, the 30-day mortality rate was 1.98%. Males had a higher incidence of POPF than females (57.42% vs 47.35%, respectively; P < .001). Patients with POPF experienced significantly higher rates of major postoperative complications (Clavien-Dindo grade ≥ IIIa), including thrombosis, pneumonia, sepsis, delayed gastric emptying, wound disruption, infections, and acute renal failure. There was a marked increase in the 30-day readmission and mortality rates in patients with POPF (30.0% vs 17.6% and 3.2% vs 1.4%, respectively; all P < .001). Multivariate analysis highlighted female sex as a protective factor against mortality (odds ratio [OR], 0.47; P < .001) and extended hospital stay (>10 days) as a predictor of increased mortality risk (OR, 2.37; P < .001). CONCLUSION: This study underscored the significant association between POPF and increased postoperative morbidity and mortality rates. Future efforts should concentrate on refining surgical techniques and improving preoperative assessments to mitigate the risks associated with POPF in patients undergoing PD.

Minorities Face Delays to Pancreatic Cancer Treatment Regardless of Diagnosis Setting.

INTRODUCTION: Our analysis was designed to characterize the demographics and disparities between the diagnosis of pancreas cancer during emergency presentation (EP) and the outpatient setting (OP) and to see the impact of our institutions pancreatic multidisciplinary clinic (PMDC) on these disparities. METHODS: Institutional review board-approved retrospective review of our institutional cancer registry and PMDC databases identified patients diagnosed/treated for pancreatic ductal adenocarcinoma between 2014 and 2022. Chi-square tests were used for categorical variables, and one-way ANOVA with a Bonferroni correction was used for continuous variables. Statistical significance was set at p < 0.05. RESULTS: A total of 286 patients met inclusion criteria. Eighty-nine patients (31.1%) were underrepresented minorities (URM). Fifty-seven (64.0%) URMs presented during an EP versus 100 (50.8%) non-URMs (p = 0.037). Forty-one (46.1%) URMs were reviewed at PMDC versus 71 (36.0%) non-URMs (p = 0.10). No differences in clinical and pathologic stage between the cohorts (p = 0.28) were present. URMs took 22 days longer on average to receive treatment (66.5 days vs. 44.8 days, p = 0.003) in the EP cohort and 18 days longer in OP cohort (58.0 days vs. 40.5 days, p < 0.001) compared with non-URMs. Pancreatic Multidisciplinary Clinic enrollment in EP cohort eliminated the difference in time to treatment between cohorts (48.3 days vs. 37.0 days; p = 0.151). RESULTS: Underrepresented minorities were more likely to be diagnosed via EP and showed delayed times to treatment compared with non-URM counterparts. Our PMDC alleviated some of these observed disparities. Future studies are required to elucidate the specific factors that resulted in these findings and to identify solutions.

Potential Role for Observation in Small Solid Pseudopapillary Neoplasm (SPN).

BACKGROUND: Solid pseudopapillary neoplasms (SPN) are rare tumors of the pancreas, typically affecting young women. Resection is the mainstay of treatment but is associated with significant morbidity and potential mortality. We explore the idea that small, localized SPN could be safely observed. METHODS: This retrospective review of the Pancreas National Cancer Database from 2004 to 2018 identified SPN via histology code 8452. RESULTS: A total of 994 SPNs were identified. Mean age was 36.8 ± 0.5 years, 84.9% (n = 844) were female, and most had a Charlson-Deyo Comorbidity Coefficient (CDCC) of 0-1 (96.6%, n = 960). Patients were most often staged clinically as cT2 (69.5%, n = 457) followed by cT3 (17.6%, n = 116), cT1 (11.2%, n = 74), and cT4 (1.7%, n = 11). Clinical lymph node and distant metastasis rates were 3.0 and 4.0%, respectively. Surgical resection was performed in 96.6% of patients (n = 960), most commonly partial pancreatectomy (44.3%) followed by pancreatoduodenectomy (31.3%) and total pancreatectomy (8.1%). In patients clinically staged as node (N0) and distant metastasis (M0) negative, occult pathologic lymph node involvement was found in 0% (n = 28) of patients with stage cT1 and 0.5% (n = 185) of patients with cT2 disease. The risk of occult nodal metastasis significantly increased to 8.9% (n = 61) for patients with cT3 disease. The risk further increased to 50% (n = 2) in patients with cT4 disease. CONCLUSIONS: Herein, the specificity of excluding nodal involvement clinically is 99.5% in tumors ≤ 4 cm and 100% in tumors ≤ 2 cm. Therefore, there may be a role for close observation in patients with cT1N0 lesions to mitigate morbidity from major pancreatic resection.

Pancreatic Cancer Patient-derived Organoids Can Predict Response to Neoadjuvant Chemotherapy.

OBJECTIVE: To evaluate if patient-derived organoids (PDOs) may predict response to neoadjuvant (NAT) chemotherapy in patients with pancreatic adenocarcinoma. BACKGROUND: PDOs have been explored as a biomarker of therapy response and for personalized therapeutics in patients with pancreatic cancer. METHODS: During 2017-2021, patients were enrolled into an IRB-approved protocol and PDO cultures were established. PDOs of interest were analyzed through a translational pipeline incorporating molecular profiling and drug sensitivity testing. RESULTS: One hundred thirty-six samples, including both surgical resections and fine needle aspiration/biopsy from 117 patients with pancreatic cancer were collected. This biobank included diversity in stage, sex, age, and race, with minority populations representing 1/3 of collected cases (16% Black, 9% Asian, 7% Hispanic/Latino). Among surgical specimens, PDO generation was successful in 71% (15 of 21) of patients who had received NAT prior to sample collection and in 76% (39 of 51) of patients who were untreated with chemotherapy or radiation at the time of collection. Pathological response to NAT correlated with PDO chemotherapy response, particularly oxaliplatin. We demonstrated the feasibility of a rapid PDO drug screen and generated data within 7 days of tissue resection. CONCLUSION: Herein we report a large single-institution organoid biobank, including ethnic minority samples. The ability to establish PDOs from chemotherapy-naive and post-NAT tissue enables longitudinal PDO generation to assess dynamic chemotherapy sensitivity profiling. PDOs can be rapidly screened and further development of rapid screening may aid in the initial stratification of patients to the most active NAT regimen.

Ovarian Metastasis from Pancreatic Ductal Adenocarcinoma.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) has a high propensity for systemic dissemination. Ovarian metastases are rare and poorly described. METHODS: We identified PDAC cases with ovarian metastasis from a prospectively maintained registry. We reported on the association between outcomes and clinicopathologic factors. Recurrence-free (RFS) and overall survival (OS) were calculated using Kaplan-Meier analysis. RESULTS: Twelve patients with PDAC and synchronous or metachronous ovarian metastases were identified. Nine patients (75%) underwent pancreatectomy for localized PDAC and developed metachronous ovarian recurrence. The median OS for all patients was 25.4 (IQR:15.4-82.9) months. For the nine patients with metachronous ovarian metastasis, the median RFS and OS were 14.2 (IQR:7.2-58.3) and 44.6 (IQR:18.6-82.9) months, respectively. Nodal disease, poor grade, vascular invasion in the pancreatic primary, and bilateral ovarian disease tended to confer worse outcomes. CONCLUSION: Patients with resected PDAC and ovarian recurrence tend to have a comparable disease course to more common patterns of recurrence. Primaries with nodal disease, poorer grade, vascular invasion, and bilateral ovarian disease were indicative of more aggressive disease biology. The ideal management remains largely unknown, and future collaborative efforts should optimize therapeutic strategies.