Frontiers in Oncology (May 2024)

Guiding post-pancreaticoduodenectomy interventions for pancreatic cancer patients utilizing decision tree models

  • Haixin Wang,
  • Bo Shen,
  • Peiheng Jia,
  • Hao Li,
  • Xuemei Bai,
  • Yaru Li,
  • Kang Xu,
  • Pengzhen Hu,
  • Pengzhen Hu,
  • Li Ding,
  • Na Xu,
  • Xiaoxiao Xia,
  • Yong Fang,
  • Hebing Chen,
  • Yan Zhang,
  • Shutong Yue

DOI
https://doi.org/10.3389/fonc.2024.1399297
Journal volume & issue
Vol. 14

Abstract

Read online

BackgroundPancreatic ductal adenocarcinoma (PDAC) is frequently diagnosed in advanced stages, necessitating pancreaticoduodenectomy (PD) as a primary therapeutic approach. However, PD surgery can engender intricate complications. Thus, understanding the factors influencing postoperative complications documented in electronic medical records and their impact on survival rates is crucial for improving overall patient outcomes.MethodsA total of 749 patients were divided into two groups: 598 (79.84%) chose the RPD (Robotic pancreaticoduodenectomy) procedure and 151 (20.16%) chose the LPD (Laparoscopic pancreaticoduodenectomy) procedure. We used correlation analysis, survival analysis, and decision tree models to find the similarities and differences about postoperative complications and prognostic survival.ResultsPancreatic cancer, known for its aggressiveness, often requires pancreaticoduodenectomy as an effective treatment. In predictive models, both BMI and surgery duration weigh heavily. Lower BMI correlates with longer survival, while patients with heart disease and diabetes have lower survival rates. Complications like delayed gastric emptying, pancreatic fistula, and infection are closely linked post-surgery, prompting conjectures about their causal mechanisms. Interestingly, we found no significant correlation between nasogastric tube removal timing and delayed gastric emptying, suggesting its prompt removal post-decompression.ConclusionThis study aimed to explore predictive factors for postoperative complications and survival in PD patients. Effective predictive models enable early identification of high-risk individuals, allowing timely interventions. Higher BMI, heart disease, or diabetes significantly reduce survival rates in pancreatic cancer patients post-PD. Additionally, there’s no significant correlation between DGE incidence and postoperative extubation time, necessitating further investigation into its interaction with pancreatic fistula and infection.

Keywords