Prognostic value of postoperative dynamic PNI trajectories for biochemical recurrence in patients with high-risk localized prostate cancer: a longitudinal trajectory clustering study

Background The prognostic nutritional index (PNI) reflects nutritional-immune status, but the value of a single baseline PNI for predicting biochemical recurrence (BCR) after radical prostatectomy in high-risk localized prostate cancer remains unclear. We aimed to identify postoperative PNI trajectories and compare their BCR predictive value with baseline PNI. Methods We retrospectively included 436 patients. PNI was measured preoperatively (T0) and at 3 (T1) and 12 (T2) months post-surgery. Longitudinal trajectory clustering identified three PNI trajectory classes: persistently high (PHigh, n = 169, 46.4%), persistently low (PLow, n = 146, 40.1%), and declining (Declining, n = 49, 13.5%). Cox regression assessed associations with BCR; a Cox-based time-to-event prediction model was evaluated using time-dependent AUC, IPCW Brier score, and calibration at 3-year and 5-year horizons. A 12-month landmark analysis was performed to address potential guarantee-time bias. Results During median 38.2-month follow-up, 122 (28.0%) developed BCR. Baseline PNI was not independently associated with BCR (continuous HR = 0.97, p = 0.320; dichotomized HR = 1.23, p = 0.347). Three trajectories emerged: persistently high (PHigh, 46.4%), persistently low (PLow, 40.1%), and declining (Declining, 13.5%). Relative to PHigh, PLow (HR = 2.95) and Declining (HR = 5.80) both had significantly higher BCR risk ( p < 0.001). A 12-month landmark analysis confirmed these findings (PLow HR = 2.72, p = 0.002; Declining HR = 4.79, p = 1.24 × 10 −6 ). The multivariable Cox model combining trajectory, log PSA, surgical margin, and nodal metastasis achieved a 3-year time-dependent AUC of 0.823 (optimism-corrected 0.815) and a 5-year AUC of 0.790 (optimism-corrected 0.781). The Brier score was 0.117 at 3 years (25.8% improvement over null) and 0.166 at 5 years; the calibration slope was 1.057 (optimism-corrected 0.989) at 3 years and 1.001 (optimism-corrected 0.940) at 5 years. Conclusion Postoperative PNI follows three distinct change-pattern subgroups. Dynamic PNI trajectories are strongly associated with BCR risk, whereas a single baseline PNI is not. Patients with persistently low or declining PNI are at high risk, supporting routine PNI monitoring to guide individualized follow-up. These observational findings require prospective multicenter validation.