Overall survival advantage with partial nephrectomy: A bias of observational data?
Article first published online: 14 MAY 2013
Copyright © 2013 American Cancer Society
Volume 119, Issue 16, pages 2981–2989, 15 August 2013
How to Cite
Shuch, B., Hanley, J., Lai, J., Vourganti, S., Kim, S. P., Setodji, C. M., Dick, A. W., Chow, W.-H., Saigal, C. and the Urologic Diseases in America Project (2013), Overall survival advantage with partial nephrectomy: A bias of observational data?. Cancer, 119: 2981–2989. doi: 10.1002/cncr.28141
- Issue published online: 2 AUG 2013
- Article first published online: 14 MAY 2013
- Manuscript Accepted: 1 APR 2013
- Manuscript Revised: 28 MAR 2013
- Manuscript Received: 21 FEB 2013
- partial nephrectomy;
- radical nephrectomy;
- survival advantage;
- observational data;
- selection bias
Partial nephrectomy (PN) and radical nephrectomy (RN) are standard treatments for a small renal mass. Retrospective studies suggest an overall survival (OS) advantage, however a randomized phase 3 trial suggests otherwise. The effects of both surgical modalities on OS were evaluated compared with controls.
A matched cohort study was performed using the Surveillance, Epidemiology, and End Results (SEER)-Medicare dataset. Individuals treated with PN or RN for localized renal cell carcinoma (RCC) measuring ≤4 cm were compared with 2 control groups (non–muscle-invasive bladder cancer (BCC) and noncancer controls (NCC). Using a greedy algorithm, RCC groups were matched with controls by demographics and comorbidities. OS for surgical groups and controls were compared. The cause of death was evaluated for cancer groups when differences in OS were noted.
Patients undergoing PN and RN were matched with controls. All cancer groups had >95% 10-year cancer-specific survival (CSS). Median OS was similar between RN (9.05 years) and BCC (8.67 years; P = .067) and NCC (8.77 years; P = .49). Median OS was improved for PN (10.45 years) compared with BCC (8.75 years; P<.001) and NCC controls (8.76 years; P<.001). A multivariate Cox hazards model demonstrated that PN improved OS compared with NCC (hazard ratio, 1.257; P<.001) and BCC (hazard ratio, 1.364; P<.001).
RN patients had similar OS compared with controls, suggesting that this treatment modality does not compromise survival. Patients undergoing PN had improved OS compared with controls, suggesting possible selection bias. The apparent survival advantage conferred by PN in SEER-Medicare case series is likely the result of selection bias involving unmeasured confounders. Cancer 2013;119:2981—2989. © 2013 American Cancer Society.