Analysis of predictors for postoperative complications after pancreatectomy--what is new after establishing the definition of postpancreatectomy acute pancreatitis (PPAP)?

Machine learning model-based prediction of postpancreatectomy acute pancreatitis following pancreaticoduodenectomy: A retrospective cohort study

BACKGROUND

The International Study Group of Pancreatic Surgery has established the definition and grading system for postpancreatectomy acute pancreatitis (PPAP). There are no established machine learning models for predicting PPAP following pancreaticoduodenectomy (PD).

AIM

To explore the predictive model of PPAP, and test its predictive efficacy to guide the clinical work.

METHODS

Clinical data from consecutive patients who underwent PD between 2016 and 2024 were retrospectively collected. An analysis of PPAP risk factors was performed, various machine learning algorithms [logistic regression, random forest, gradient boosting decision tree, extreme gradient boosting, light gradient boosting machine, and category boosting (CatBoost)] were utilized to develop predictive models. Recursive feature elimination was employed to select several variables to achieve the optimal machine algorithm.

RESULTS

The study included 381 patients, of whom 88 (23.09%) developed PPAP. PPAP patients exhibited a significantly higher incidence of postoperative pancreatic fistula (55.68% vs 14.68%, P < 0.001), grade C postoperative pancreatic fistula (9.09% vs 1.37%, P = 0.001). The CatBoost algorithm outperformed other algorithms with a mean area under the receiver operating characteristic curve of 0.859 [95% confidence interval (CI): 0.814-0.905] in the training cohort and 0.822 (95%CI: 0.717-0.927) in the testing cohort. According to shapley additive explanations analysis, pancreatic texture, main pancreatic duct diameter, body mass index, estimated blood loss, and surgery time were the most important variables based on recursive feature elimination. The CatBoost algorithm based on selected variables demonstrated superior performance, with an area under the receiver operating characteristic curve of 0.837 (95%CI: 0.788-0.886) in the training cohort and 0.812 (95%CI: 0.697-0.927) in the testing cohort.

CONCLUSION

We developed the first machine learning-based predictive model for PPAP following PD. This predictive model can assist surgeons in anticipating and managing this complication proactively.

Textbook outcome after pancreatoduodenectomy and distal pancreatectomy with postoperative hyperamylasemia-a propensity score matching analysis

PURPOSE

Postoperative serum hyperamylasemia (POH) is a part of the new, increasingly highlighted, definition for postpancreatectomy pancreatitis (PPAP). This study aimed to analyze whether the biochemical changes of PPAP are differently associated with postoperative complications after distal pancreatectomy (DP) compared with pancreatoduodenectomy (PD). The textbook outcome (TO) was used as a summary measure to capture real-world data.

METHODS

The data were retrospectively extracted from a prospective clinical database. Patients with POH, defined as levels above our institution's upper limit of normal on postoperative day 1, after DP and the corresponding propensity score-matched cohort after PD were evaluated on postoperative complications by using logistic regression analyses.

RESULTS

We analyzed 723 patients who underwent PD and DP over a period of 9 years. After propensity score matching, 384 patients (192 patients in each group) remained. POH was observed in 78 (41.1%) and 74 (39.4%) after PD and DP correspondingly. There was a significant increase of postoperative complications in the PD group: Clavien-Dindo classification system ≥3 (P < .01 vs P = .71), clinically relevant postoperative pancreatic fistula (P < .001 vs P = .2), postpancreatectomy hemorrhage (P < .001 vs P = .11), and length of hospital stay (P < .001 vs P = .69) if POH occurred compared with in the DP group. TO was significantly unlikely in cases with POH after PD compared with DP (P > .001 vs P = .41). Furthermore, POH was found to be an independent predictor for missing TO after PD (odds ratio [OR], 0.29; 95% CI, 0.14-0.60; P < .001), whereas this was not observed in patients after DP (OR, 0.53; 95% CI, 0.21-1.33; P = .18).

CONCLUSION

As a part of the definition for PPAP, POH is a predictive indicator associated with postoperative complications after PD but not after DP.

Association of Postoperative Hyperamylasemia With Clinically Relevant Postoperative Pancreatic Fistula in Pancreatoduodenectomy

Background In this study, we aimed to determine the association between postoperative hyperamylasemia (POH) and clinically relevant postoperative pancreatic fistula (CR-POPF) after pancreatoduodenectomy (PD). Methodology A prospective observational study of 140 consecutive PDs between March 2020 and March 2022 was conducted. POH was defined as an elevation in serum pancreatic amylase levels above the institutional upper limit of normal on postoperative day (POD) 1 (>100 U/L). CR-POPF was defined as the International Study Group of Pancreatic Surgery Grade B or C POPF. The primary outcome was the rate of CR-POPF in the study population. The trial was prospectively registered with Clinicaltrials.gov (NCT04514198). Results In our study, 93 (66.42%) patients had POH (serum amylase >100 U/L). CR-POPF developed in 48 (34.28%) patients: 40 type B and 8 type C. CR-POPF rate was 43.01% (40/93) in patients with POH compared to 17.02% (8/47) in patients without POH (p = 0.0022). Patients with POH had a mean serum amylase of 422.7 ± 358.21 U/L on POD1 compared to 47.2 ± 20.19 U/L in those without POH (p < 0.001). Serum amylase >100 U/L on POD1 was strongly associated with developing CR-POPF (odds ratio = 3.71; 95% confidence interval = 1.31-10.37) on logistic regression, with a sensitivity and specificity of 83.3% and 42.4%, respectively. Blood loss >350 mL, pancreatic duct size <3 mm, and elevated POD1 serum amylase >100 U/L were predictive of CR-POPF on multivariate analysis (p < 0.001). Conclusions An elevated serum amylase on POD1 may help identify patients at risk for developing POPF following PD.

Risk Factors of Postoperative Acute Pancreatitis and Its Impact on the Postoperative Course after Pancreaticoduodenectomy-10 Years of Single-Center Experience

BACKGROUND

Clinically relevant acute postoperative pancreatitis (CR-PPAP) after pancreaticoduodenectomy (PD) is a complication that may lead to the development of local and systemic consequences. The study aimed to identify risk factors for CR-PPAP and assess the impact of CR-PPAP on the postoperative course after PD.

METHODS

The study retrospectively analyzed data from 428 consecutive patients who underwent PD at a single center between January 2013 and December 2022. The presence of increased amylase activity in plasma, above the upper limit of normal 48 h after surgery, was checked. CR-PPAP was diagnosed when accompanied by disturbing radiological features and/or symptoms requiring treatment. We investigated the relationship between the occurrence of CR-PPAP and the development of postoperative complications after PD, and possible predictors of CR-PPAP.

RESULTS

The postoperative follow-up period was 90 days. Of the 428 patients, 18.2% (n = 78) had CR-PPAP. It was associated with increased rates of CR-POPF, delayed gastric emptying, occurrence of intra-abdominal collections, postoperative hemorrhage, peritonitis, and septic shock. Patients who developed CR-PPAP were more often reoperated (37.17% vs. 6.9%, p < 0.0001)) and had increased postoperative mortality (14.1% vs. 5.74%, p < 0.0001). Soft pancreatic parenchyma, intraoperative blood loss, small diameter of the pancreatic duct, and diagnosis of adenocarcinoma papillae Vateri were independent risk factors for CR-PPAP and showed the best performance in predicting CR-PPAP.

CONCLUSIONS

CR-PPAP is associated with an increased incidence of postoperative complications after PD, worse treatment outcomes, and an increased risk of reoperation and mortality. Pancreatic consistency, intraoperative blood loss, width of the duct of Wirsung, and histopathological diagnosis can be used to assess the risk of CR-PPAP. Amylase activity 48 h after surgery > 161 U/L is highly specific in the diagnosis of CR-PPAP.