Impact of Local Recurrence on Cause-Specific Death After Stereotactic Body Radiotherapy for Early-Stage Non-Small Cell Lung Cancer: Dynamic Prediction Using Landmark Model.

Purpose: The purpose of this study was to assess the impact of local recurrence (LR) on cause-specific death (CSD) in patients with non-small cell lung cancer (NSCLC) treated with stereotactic body radiotherapy (SBRT). A dynamic prediction model that incorporated LR as a time-dependent covariate was used.
Methods and Materials: This study included 535 stage I (cT1-T2aN0M0) NSCLC patients treated with SBRT from two institutions. We developed a landmark dynamic prediction model to estimate the probability of a CSD. This model determined the probability of surviving for an additional 3 years at different prediction time points during follow-up, given the history of recurrence status. The baseline covariates included in the model were age, sex, T stage, and histology, while the time-dependent covariates were LR and regional and/or distant recurrence (RDR) status.
Results: Overall, 137 patients (25.6%) died of lung cancer within a median follow-up of 4.1 years. Of the 195 patients who developed recurrence, 28, 125, and 42 patients had LR only, RDR only, and both, respectively. The landmark model showed that older age, advanced T stage, LR, and RDR were significantly associated with an increased risk of subsequent CSD. Among these covariates, LR (odds ratio [OR], 8.5; 95% confidence interval [CI], 6.0-12.0; P < .001) and RDR (OR, 11.6; 95% CI, 9.1-14.9; P < .001) demonstrated strong effects on CSD within 3 years after the prediction time points. The dynamic prediction provided information on the probability of future CSD according to individual recurrence status during follow-up.
Conclusions: Dynamic prediction using the landmark model showed that LR had a substantial impact on subsequent CSD, which was comparable to that of RDR. This result supports the notion that strategies to improve local control are reasonable.
(Copyright © 2021 Elsevier Inc. All rights reserved.)