Assessing the impact of underlying liver disease on 30-day postoperative outcomes after pancreatectomy: an ACS NSQIP® analysis

Abstract

Background: Pancreatectomy remains associated with significant postoperative morbidity despite recent advances. The Model for End-Stage Liver Disease (MELD), derived from readily available serum chemistries, is primarily applied towards liver transplantation recipient selection. It has become increasingly utilized to predict complications in non-transplant surgical populations. Liver disease has been linked to adverse outcomes following gastrointestinal surgery, however studies examining its impact on outcomes following pancreatectomy are limited. Employing the American College of Surgeons National Surgical Quality Improvement Program® (ACS NSQIP®) Participant Use Data File (PUF), we aimed to compare 30-day postoperative outcomes of patients undergoing elective pancreatectomy and stratified by MELD score. Methods: Elective Pancreatoduodenectomies (PDs) and Distal Pancreatectomies (DPs) were identified from the 2014–2016 Procedure Targeted Pancreatectomy PUFs. MELD scores were calculated using 90-day preoperative serum creatinine, total bilirubin, INR, and serum sodium. Examined 30-day outcomes included mortality, cardiopulmonary complications, prolonged postoperative length-of-stay (defined as >14 days for PDs and >7 days for DPs), discharge not-to-home, intraoperative/postoperative transfusion, any complication, and serious complication. Outcomes were compared between clinically relevant MELD score strata (<11 vs. ≥11) as established by the United Network for Organ Sharing (UNOS). Multiple logistic regression was used to adjust for potential confounders, including patient sex, age, BMI, relevant comorbidities, ASA classification, and histology. Comparisons yielding p-values <0.05 were considered statistically significant. Results: A total of 7,580 elective PDs and 3,295 elective DPs with evaluable MELD scores were analyzed. Of these, 1,701 (22.4%) PDs and 223 (6.8%) DPs had a MELD score ≥11. PD patients with MELD ≥11 exhibited significantly higher risks for mortality [OR = 2.075 (95% CI: 1.429, 3.012), p < 0.001], discharge not-to-home [OR = 1.256 (95% CI: 1.073, 1.471), p = 0.005], and transfusion [OR = 1.699 (95% CI: 1.490, 1.936), p < 0.001]. DP patients with MELD ≥11 exhibited significantly higher risks for prolonged LOS [OR = 1.754 (95% CI: 1.308, 2.354), p < 0.001], discharge not-to-home [OR = 1.829 (95% CI: 1.158, 2.889), p = 0.010], and transfusion [OR = 2.780 (95% CI: 2.023, 3.819), p<0.001]. In PD, following risk adjustment, MELD ≥11 was significantly predictive of mortality [OR = 1.690 (95% CI: 1.154, 2.475), p = 0.007] and transfusion [OR = 1.553 (95% CI: 1.356, 1.779), p < 0.001]. In DP, following risk adjustment, MELD ≥11 was significantly predictive of prolonged LOS [OR = 1.422 (95% CI: 1.043, 1.940), p = 0.026] and transfusion [OR = 2.299 (95% CI: 1.643, 3.217), p < 0.001]. Conclusion: A MELD score ≥11 is associated with a nearly-two fold increase in the odds of PD mortality despite its low prevalence. These findings underscore the importance of preoperative liver disease assessment in stratifying risks of 30-day postoperative complications following pancreatectomy.