Limited Value of Preoperative Nutritional and Inflammatory Indices for Predicting Major Complications After Pancreaticoduodenectomy: A Retrospective Cohort Study

Background and Objectives: Pancreaticoduodenectomy (PD) is a complex procedure with substantial morbidity in the surgical management of pancreatic head and periampullary pathologies. This study aimed to evaluate the value of three preoperative nutritional/inflammatory indices—the prognostic nutritional index (PNI), C-reactive protein-to-albumin ratio (CAR), and albumin-to-globulin ratio (AGR)—in predicting major postoperative complications (Clavien–Dindo grade ≥ III), and to explore the association of patient age, pathology (malignant vs. nonmalignant/low-grade malignant), and tumor stage with complication risk. Materials and Methods: Data from 92 patients who underwent PD between January 2020 and May 2026 were retrospectively reviewed. PNI, CAR, and AGR were calculated from preoperative laboratory values. The primary outcome was major complication (Clavien–Dindo ≥ III) during the postoperative period. Group comparisons used Student’s t-test or the Mann–Whitney U test as appropriate, and the chi-square or Fisher’s exact test for categorical variables. Multivariable logistic regression and receiver operating characteristic (ROC) curve analysis were performed. Results: Mean age was 62.7 ± 11.8 years; 30.4% (n = 28) of patients developed a major complication. Patients with major complications were older (66.4 ± 12.7 vs. 61.1 ± 11.0 years, p = 0.043), and age ≥ 70 years was associated with a higher complication rate (48.3% vs. 22.2%, p = 0.012). Unexpectedly, nonmalignant/low-grade malignant pathology (benign lesions, noninvasive precursor lesions, and solid pseudopapillary neoplasms) was associated with a higher major complication rate than malignant pathology (57.1% vs. 25.6%, p = 0.018). PNI, CAR, and AGR did not differ significantly between complication groups (p = 0.264, 0.228, and 0.718, respectively), and none showed significant discrimination on ROC analysis (area under the curve [AUC]: PNI 0.427, CAR 0.581, AGR 0.476; all p > 0.05). In multivariable models adjusted for age and pathology, none of the indices was independently associated with major complications; nonmalignant/low-grade malignant pathology remained associated with higher odds of major complications (adjusted odds ratio range 4.90–5.81, p < 0.05), also when solid pseudopapillary neoplasms were reassigned to the malignant group, whereas the association with age was borderline in most models and disappeared when postoperative deaths were excluded. Tumor stage was not significantly associated with major complications among malignant patients (n = 76, p = 0.739). Conclusions: In this cohort, preoperative PNI, CAR, and AGR did not provide adequate discrimination for major complication risk after pancreaticoduodenectomy. The higher complication rate in nonmalignant/low-grade malignant cases is contrary to expectation and may reflect pancreatic gland characteristics (soft texture, small duct) that could not be assessed in this study. Given the small number of events, these findings should be considered exploratory and require validation in larger cohorts that include established fistula risk factors.