Testing Repeat Radiation Therapy for Recurrent Non-small Cell Lung Cancer

Progression-free survival (PFS)

The primary analysis of PFS will be conducted in the intent-to-treat population, defined as all randomized patients. The distributions of PFS will be estimated by the Kaplan-Meier method. The primary comparison between arms will use a one-sided stratified log-rank test with the randomization stratification factors. A stratified Cox proportional hazards model will be used to estimate the hazard ratio and corresponding confidence interval. An unstratified log-rank test and unstratified Cox proportional hazards model will also be provided as sensitivity analyses. No post hoc modification or collapsing of stratification categories based on observed event counts will be performed. If the stratified Cox model is not estimable because of sparse data, the unstratified Cox model will be used to summarize the treatment hazard ratio and confidence interval, with the reason documented.

Overall survival

Will be summarized using Kaplan-Meier methods. Comparisons between arms will be summarized descriptively using hazard ratio estimates and corresponding confidence intervals.

Incidence of clinician-scored treatment-related adverse events

Clinician-scored adverse events will be summarized by study arm using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE). For each patient, the maximum severity reported for each adverse event will be used in the summaries. Adverse events will be summarized regardless of relationship to protocol treatment as assessed by the investigator. All adverse events, adverse events leading to withdrawal, interruption or modification of protocol treatment, Grade ≥ 3 adverse events, and serious adverse events will be summarized. Deaths and cause of death will be summarized. Event summaries will be presented by type, grade, and attribution and will be performed at the time of the primary endpoint analysis.

Incidence of patient-reported adverse events

A composite grading algorithm (Basch 2021) will be utilized to derive a single numerical grade for each adverse event scored at each timepoint using Patient-Reported Outcomes (PRO)-CTCAE. Patient-reported adverse events measured using PRO-CTCAE will also be summarized descriptively by arm and time point, together with compliance and missing-data patterns.

Site of first disease progression

Patterns of first failure will be summarized descriptively by arm using the categories specified in the protocol, including locoregional/intrathoracic progression, distant progression, or both. Formal hypothesis testing for these secondary endpoints is not planned unless otherwise specified in a future statistical analysis plan.

Clinical and dosimetric predictors of high-grade treatment-related toxicity

These analyses will be restricted to patients who initiate protocol reirradiation and will be descriptive and hypothesis-generating. Candidate predictors may include baseline clinical characteristics, prior treatment characteristics, and dosimetric parameters from the reirradiation and composite treatment plans. Associations with high-grade treatment-related toxicity will be summarized using appropriate descriptive and regression-based methods, as feasible given the number of observed events.

Radiotherapy modality toxicity and efficacy outcomes

Will include descriptive comparisons of toxicity and efficacy outcomes, including time-to-event outcomes and treatment-related adverse events, among patients treated with proton versus photon radiotherapy within Arm 2. Because radiotherapy modality is selected by the treating team and not randomized within Arm 2, these analyses will be considered descriptive and hypothesis-generating only.

Regional lung ventilation metrics

Among Arm 2 patients with available four-dimensional computed tomography planning scans, additional exploratory analyses may derive regional lung ventilation metrics and evaluate their association with pulmonary toxicity. These analyses are intended to identify potential imaging-based predictors of pulmonary toxicity and to generate hypotheses for future study.