Researchers writing in JAMA Network Open report that .
The finding, which was derived using instrumental variable analysis to adjust for bias due to unmeasured variables, is in contradiction to findings from prior observational data sets.
“These observational studies did not account for selection bias related to unmeasured confounding by surgical indication, and as such, their results may not accurately reflect the effectiveness of the intervention,” wrote the authors, led by Nicholas H. Chakiryan, MD, H. Lee Moffitt Cancer Center and Research Institute, Tampa, Fla.
The primary outcome analysis using conventional adjustments for selection bias in the final study population of 12,766 patients (median age 63 years, 68% male, 88% White) found cytoreductive nephrectomy performed in 5,005 patients (39%) to be associated with a significant overall survival benefit (multivariable Cox proportional hazards regression: hazard ratio [HR], 0.49; 95% confidence interval [CI], 0.47-0.51; propensity score matching: HR, 0.48; 95%CI, 0.46-0.50). Analysis using instrumental variable estimates, however, did not demonstrate an association between cytoreductive nephrectomy and overall survival (HR, 0.92; 95%CI, 0.78-1.09). “This discrepancy likely reflects the fact that surgical indication for cytoreductive nephrectomy is primarily driven by factors that are not commonly measured or available in observational data sets,” wrote Dr. Chakiryan and colleagues.
For metastatic clear cell renal cell carcinoma (ccRCC) surgical candidates lacking poor-risk disease, cytoreductive nephrectomy has been a clinical standard for decades. Several large observational studies conducted during the current postcytokine, tyrosine kinase–inhibiting targeted therapy era have demonstrated that cytoreductive nephrectomy continues to offer substantial overall survival benefit. These studies did not, however, account for selection bias related to unmeasured confounding by surgical indication.
The researchers identified 12,766 cases of ccRCC from the National Cancer Database, which includes more than 70% of incident cancer cases diagnosed in the United States, from Jan. 1, 2006, to Dec. 31, 2016. Their primary objective was to assess the effect of cytoreductive nephrectomy on overall survival for patients with metastatic ccRCC using instrumental variable analysis to adjust for unmeasured confounding and to compare these results with those generated by conventional adjustments for selection bias.
Instrumental variables are used to control for confounding and measurement error in observational studies. In this study, increasing distance to the treating facility was a significant instrumental variable (P < .001), with an increasing proportion of patients undergoing cytoreductive nephrectomy as distance to a facility increased. “It is worth reinforcing that instrumental variable estimates reflect the outcomes of marginal patients in the sample,” the researchers noted. “In this instance, marginal patients are those whose cytoreductive nephrectomy status was primarily associated with their distance to the treating facility. Increasing distance to the treating facility was significantly associated with receipt of a cytoreductive nephrectomy, presumably because patients are more willing to travel to referral centers for complex surgical care with a limited number of visits, as opposed to receipt of systemic therapy that requires frequent visits for an indefinite period and can be effectively administered locally.”
“Consistent with contemporary level 1 evidence, instrumental variable analysis demonstrated that cytoreductive nephrectomy was not associated with improved overall survival for patients with metastatic clear cell renal cell carcinoma,” the authors concluded.
Among limitations of the analysis, they noted that instrumental variable analyses functionally compare marginal patient populations within the overall cohort, potentially limiting the generalizability of the results.