An Estimate of Local Failure in the TARGIT-A Trial of Pre-pathology Intraoperative Radiation Therapy for Early Breast Cancer

Abstract

TARGIT-A was a pragmatic randomized noninferiority trial including women with early-stage breast cancer treated postlumpectomy with either external beam radiation therapy (EBRT) or 50 kV x-rays delivered intraoperatively with or without EBRT, as indicated. The long-term update of the pre-pathology cohort did not include a 10-year estimate of the primary endpoint of local failure (although tabular 5-year data was provided). Here, we used the data from the pre-pathology manuscript to estimate the cumulative incidence of local failure. Using digitizer software and the published survival curves, we extracted the Kaplan-Meier rate of local recurrence-free survival and overall survival. The extracted data were calibrated to the published point-estimates to within ±0.5%. The data were then fit to parametric survival models, and overall survival and local recurrence-free survival curves were subtracted to give the estimate of local failure in the presence of the competing risk of death. Bootstrap resampling was used to assess for parameter uncertainty in the modeling process. Our analysis estimated that the risk of local failure at 10 years in the TARGIT-A pre-pathology cohort is approximately 1.7% with EBRT (95% confidence interval [CI], 0%-4.3%) and 5.5% in the pragmatic risk-adapted TARGIT strategy (95% CI, 2.9%-8.0%). A weighted average estimate suggests that the risk of local failure in low-risk women treated with TARGIT alone is approximately 6.6% at 10 years (95% CI, 3.3%-10.0%), with an estimated difference of 4.9% (95% CI, 0.6%-9.2%) compared with EBRT. These data allow for contextualization and informed decisions when considering megavoltage EBRT, kilovoltage intraoperatively, or omission of radiation therapy entirely.