Predictive Value of the ALBI Grade for Short-Term Morbidity and Mortality After Pancreatoduodenectomy

Background: Pancreatoduodenectomy (PD) is a complex surgical procedure with substantial postoperative morbidity and mortality. Postoperative outcomes following PD are typically determined by measuring preoperative liver function in the form of serum albumin and bilirubin levels. In this study, the prognostic value of the preoperative ALBI grade was compared with that of albumin and bilirubin alone in predicting postoperative complications and short-term mortality after PD. Methods: This retrospective cohort study included 1149 patients who underwent partial or total PD between 2014 and 2017 at a high-volume surgical center. The preoperative ALBI grade (I-III) was calculated in these patients as well as the preoperative levels of bilirubin and albumin. The associations between ALBI grade, albumin and bilirubin levels with postoperative complications and 30- and 90-day mortality were analyzed using univariate and multivariable logistic regression models. The predictive performance of the ALBI grade and albumin and bilirubin levels was assessed using receiver operating characteristic (ROC) curves and area under the curve (AUC) analysis. Results: Higher ALBI grades were significantly associated with worse perioperative outcomes, including increased incidence of liver perfusion failure (42.4%) and postpancreatectomy hemorrhage (15.2%). In a pre-specified reduced multivariable model adjusted for BMI, ASA physical status as well as surgical complexity, ALBI grade III remained associated with 30- and 90-day mortality (OR 6.21; 95% CI, 1.62-23.78; p = 0.008 and OR 8.97; 95% CI, 2.91-27.62; p < 0.001) but was attenuated compared to univariate analyses. Seven of the 8 patients with ASA grade IV physical status were in ALBI grade III and accounted for 5 of the 6 30-day and 6 of the 8 90-day deaths in this group; in a sensitivity analysis excluding these patients, the association between ALBI grade III and 30-day mortality was no longer statistically significant, and the univariable association with 90-day mortality was also attenuated to non-significance, although the BMI-adjusted 90-day estimate remained significant (OR 5.02; 95% CI, 1.07-23.54; p = 0.041). ROC analysis showed that the ALBI score (AUC, 0.74 and 0.74) did not outperform albumin alone (AUC, 0.69 and 0.70) in predicting 30- and 90-day mortality (DeLong p = 0.21-0.24), although both outperformed bilirubin alone (AUC, 0.54-0.56). Conclusions: Preoperative ALBI grade III is associated with short-term morbidity and mortality after PD, but this association is confounded in part by baseline surgical risk (ASA physical status), and the ALBI score does not outperform albumin alone as a predictor of mortality. As a simple, readily available composite marker, the ALBI grade may still help flag patients who warrant closer perioperative surveillance, although this retrospective study does not establish that correcting albumin or bilirubin levels before surgery improves outcomes.