Comprehensive Characterization of Metastatic Patterns and a Redefined Prognostic Framework in Metastatic Prostate Cancer

BACKGROUND: Metastatic prostate cancer (mPCa) exhibits highly complex metastatic patterns. Comprehensive evaluation of distinct metastatic patterns and their integrated prognostic impact remains lacking. METHODS: Patients with mPCa were identified from the Surveillance, Epidemiology, and End Results (SEER) database, and metastatic patterns were systematically characterized. Cancer-specific survival (CSS) and overall survival (OS) were estimated using the Kaplan-Meier method and compared using multivariable Cox regression. Based on metastatic characteristics, patients were classified into seven pattern-based categories and subsequently consolidated into four risk groups (low, intermediate, high, and very high). Temporal robustness was assessed across diagnostic periods, and prognostic discrimination was compared with the conventional M classification using Harrell's C-index and time-dependent area under the receiver operating characteristic curve (AUC). RESULTS: A total of 13,325 patients with mPCa were identified. Compared with M1b disease, liver-only and multiple visceral metastases were associated with significantly poorer survival (all HR > 1, p < 0.05), whereas lung-only and other single visceral metastases showed comparable outcomes (all p > 0.05). Concomitant distant lymph node involvement adversely affected survival primarily in M1b disease (all HR > 1, p < 0.05), while concomitant bone involvement worsened outcomes in most subgroups except those with multiple visceral sites. Integrating these findings, the four-tier risk stratification model demonstrated clear stepwise discrimination for both CSS (Low risk as reference; Intermediate: HR = 1.63, 95% CI, 1.45-1.84, p < 0.001; High: HR = 2.04, 95% CI, 1.76-2.37, p < 0.001; Very high: HR = 3.99, 95% CI, 3.39-4.71, p < 0.001) and OS (Low risk as reference; Intermediate: HR = 1.53, 95% CI, 1.38-1.70, p < 0.001; High: HR = 1.93, 95% CI, 1.70-2.19, p < 0.001; Very high: HR = 3.48, 95% CI, 2.99-4.06, p < 0.001). Temporal robustness analysis showed consistent stepwise survival gradients across both diagnostic periods. Compared with the conventional M classification, the four-tier risk model showed higher C-index values for CSS (0.695 vs 0.638) and OS (0.696 vs 0.639), with consistently higher time-dependent AUCs at 12, 36, and 60 months. CONCLUSIONS: Survival varies substantially across metastatic patterns in mPCa. A simplified four-tier risk stratification model integrating multidimensional metastatic features may provide incremental prognostic information beyond conventional M staging, although further external validation is required before clinical implementation.