Abstract
OBJECTIVE: To emulate two target clinical trials of radical nephrectomy (RN) with lymph node dissection (LND) vs radical nephrectomy alone. METHODS: Using the National Cancer Database, we separately emulated an index trial of patients with cT1-3cN0cM0 renal cell carcinoma (RCC), designed to resemble EORTC 30881 ("index trial emulation"), and a hypothetical trial of patients at increased risk for lymph node metastases with cT1-4cN0-1cM0 RCC ("high-risk trial emulation"). A propensity score for LND was estimated using preoperative features (Model 1) or preoperative and pathologic features (Model 2). The associations of LND with overall survival (OS) were estimated using Cox regression with stabilized inverse probability weights. RESULTS: A total of 67,388 patients were included in the index trial emulation. Median follow-up was 49.2 (interquartile range 27.2-74.3) months. LND was not associated with improved OS when adjusting using either Model 1 (hazard ratio [HR] 1.26; 95% confidence interval [CI] 1.20-1.33; P <.0001) or Model 2 (HR 1.13; 95% CI 1.07-1.20; P <.0001). A total of 69,477 patients were included in the high-risk trial emulation. Median follow-up was 48.6 (interquartile range 26.6-73.8) months. LND was not associated with improved OS when adjusting using either Model 1 (HR 1.24; 95% CI 1.18-1.30; P <.0001) or Model 2 (HR 1.09; 95% CI 1.04-1.16; P = .001). In sensitivity analyses, LND was not associated with improved OS across cN stage, pT stage, tumor grade, histologic subtype, or probability of pN1 disease. CONCLUSION: In observational analyses, that emulate target trials representing EORTC 30881 and a trial of LND in high-risk RCC, LND was not associated with improved OS.