Adjuvant Therapy for Resected Gallbladder Cancer: Analysis of the National Cancer Data Base

Abstract

Background: Management of resected gallbladder cancer relies on single-arm trials and retrospective observations. Our objective was to evaluate adjuvant therapy in a nationwide data set using causal inference methods to address sources of bias.

Methods: We studied patients with T2-3 or node-positive, nonmetastatic gallbladder cancer, resected with grossly negative margins and reported to the National Cancer Data Base between 2004 and 2011. We defined adjuvant therapy as any chemotherapy within 90 days of surgery, and upfront concurrent chemoradiation as radiation within 14 days of first chemotherapy. After adjusting for missing data and guarantee-time bias, and using propensity score analysis to minimize indication bias, we compared overall survival of patients receiving adjuvant therapies with untreated case subjects.

Results: Adjuvant chemotherapy was administered to 28.8% of 4775 patients, and upfront chemoradiation to 13.5%. Treatment was less frequent among patients who were older, patients with comorbidities, and among white Hispanic women. T3 or node-positive disease, microscopically positive margins, or extended resection increased the likelihood of adjuvant therapy. Overall survival at three years was 39.9% (95% confidence interval [CI] = 38.4% to 41.4%) and was unaffected by adjuvant therapy after adjusting for multiple confounders (hazard ratio = 1.01, 95% CI = 0.92 to 1.10). Patients with T3 or node-positive tumors treated with upfront adjuvant chemoradiation had a modest early survival advantage (absolute difference at two years = 6.8%, 95% CI = 1.1% to 12.6%), but survival curves converged after five years of follow-up.

Conclusions: The curative potential of current adjuvant therapy in gallbladder cancer is questionable, justifying placebo-controlled investigation of novel chemotherapy combinations or alternative approaches. Chemoradiation may provide a short-term benefit in locally advanced tumors.