Early vs. delayed primary local therapy and survival in de novo metastatic prostate cancer: target trial emulation

Primary-directed local therapy (PDLT) in de novo metastatic prostate cancer remains controversial. Evidence from randomized trials supports benefit in low-volume disease, but the influence of metastatic phenotype and treatment timing is unclear. We conducted a population-based target trial emulation to evaluate the effect of PDLT timing on overall survival and identify factors associated with early therapy.

Using SEER 18 registries (2010-2022), we identified 16,740 men with de novo metastatic prostate adenocarcinoma and complete metastatic site data. Patients were classified by metastatic phenotype (bone-only, lung-only, liver ± others, brain ± others, polymetastatic). We emulated a target trial comparing early PDLT (≤ 6 months), deferred PDLT (6-12 months), and no PDLT, using a clone-censor-weight approach with stabilized inverse probability of censoring weights. Weighted Cox models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for overall survival. Subgroup analyses evaluated effect modification by metastatic phenotype and by treatment modality, with additional sensitivity analyses stratified by diagnosis era. Logistic regression identified predictors of early PDLT receipt.

Overall, 31.0% received PDLT within 12 months (early: 29.3%, deferred: 0.3%). Early and deferred PDLT were associated with statistically significant but very small relative survival differences (HR 0.96, 95% CI 0.93-1.00; HR 0.97, 95% CI 0.94-0.99); the absolute difference at 36 months was 0.16 months (≈ 5 days) for early PDLT. Benefits were confined to bone-only disease (early HR 0.93, 95% CI 0.90-0.97), whereas early PDLT was associated with worse outcomes in lung-only (HR 1.45, 95% CI 1.10-1.91) and liver ± others (HR 1.14, 95% CI 1.02-1.28). The 36-month RMST gain attributed to deferred PDLT (1.72 months) is statistically significant but should be considered exploratory given the very small subgroup size (n = 52). Younger age, lower PSA, brain involvement, and more recent diagnosis predicted early PDLT.

PDLT in bone-only metastatic prostate cancer was associated with a small relative survival difference of uncertain clinical relevance. Associations with worse outcomes in visceral metastatic phenotypes likely reflect residual confounding by indication and should not be interpreted as causal harm. Metastatic phenotype may add to existing volume-based stratification rather than replace it. Findings concerning the deferred-PDLT subgroup are exploratory and require prospective validation.