Association of early postoperative changes in the prognostic nutritional index with surgical site infection after oral cancer resection with flap reconstruction

The clinical significance of early postoperative changes in the prognostic nutritional index (PNI) remains unclear. We investigated whether early postoperative PNI changes were associated with subsequent surgical site infection (SSI) after oral cancer resection with flap reconstruction. This retrospective single-center study included 171 patients who underwent oral cancer resection with free or pedicled flap reconstruction between September 2012 and April 2026. PNI was calculated on postoperative days (PODs) 1, 3, and 5. Associations of the PNI POD3/POD1 and POD5/POD3 ratios with recipient-site SSI through POD30 were evaluated using landmark-based Cox regression analyses. Sensitivity analyses, bootstrap optimism correction of model discrimination, and paired comparisons with other postoperative markers were also performed. SSI occurred in 33 patients (19.3%), with the earliest diagnosis on POD6. Patients who developed SSI had a lower median PNI POD3/POD1 ratio than those who did not (0.87 vs. 0.94; P = 0.006). In the POD5-landmark model adjusted for the POD5/POD3 ratio, operative time, and diabetes mellitus, the hazard ratio for SSI per 0.1 increase in the POD3/POD1 ratio was 0.635 (95% confidence interval, 0.423–0.951; P = 0.027). In the model using only information available by POD3, the corresponding hazard ratio was 0.585 (95% confidence interval, 0.398–0.860; P = 0.006). The adjusted POD5-landmark model had an apparent C-index of 0.682 and an optimism-corrected C-index of 0.646. The POD3/POD1 ratio had an area under the curve of 0.655. At the cohort-derived cutoff of 0.913, the positive and negative predictive values were 30.1% and 90.9%, respectively. Discrimination did not differ detectably from that of the albumin POD3/POD1 ratio (area under the curve, 0.664; paired P = 0.822). A greater decline in PNI from POD1 to POD3 was associated with subsequent recipient-site SSI after limited covariate adjustment. Discrimination was modest, and superiority over the albumin ratio was not demonstrated. External validation is required before the PNI POD3/POD1 ratio or its cohort-derived cutoff can be used for clinical decision-making.