RESUMO
OBJECTIVE: To evaluate whether postoperative serum hyperamylasemia (POH), with drain fluid amylase (DFA) and C-reactive protein (CRP), improves the Fistula Risk Score (FRS) accuracy in assessing the risk of a postoperative pancreatic fistula (POPF). SUMMARY BACKGROUND DATA: The FRS predicts POPF occurrence using intraoperative predictors with good accuracy but intrinsic limits. METHODS: Outcomes of patients who underwent pancreaticoduodenectomies between 2016 and 2021 were evaluated across FRS-risk zones and POH occurrence. POH consists of serum amylase activity greater than the upper limit of normal (52 U/l), persisting within the first 48 hours postoperatively (postoperative day -POD- 1 and 2). RESULTS: Out of 905 pancreaticoduodenectomies, some FRS elements, namely soft pancreatic texture (odds ratio (OR) 11.6), pancreatic duct diameter (OR 0.80), high-risk pathologic diagnosis (OR 1.54), but not higher blood loss (OR 0.99), were associated with POH. POH was an independent predictor of POPF, which occurred in 46.8% of POH cases ( P <0.001). Once POH occurs, POPF incidence rises from 3.8% to 42.9%, 22.9% to 41.7%, and 48.9% to 59.2% in patients intraoperatively classified at low, moderate and high FRS risk, respectively. The predictive ability of multivariable models adding POD 1 drain fluid amylase, POD 1-2 POH and POD 3 C-reactive protein to the FRS showed progressively and significantly higher accuracy (AUC FRS=0.82, AUC FRS-DFA=0.85, AUC FRS-DFA-POH=0.87, AUC FRS-DFA-POH-CRP=0.90, DeLong always P <0.05). CONCLUSIONS: POPF risk assessment should follow a dynamic process. The stepwise retrieval of early, postoperative biological markers improves clinical risk stratification by increasing the granularity of POPF risk estimates and affords a possible therapeutic window before the actual morbidity of POPF occurs.