Differential lung tissue response and patient outcomes between protons and photons receiving regional nodal irradiation for breast cancer: A Prospective Correlative Study within a Randomized Trial.

Abstract

To quantify radiation-induced lung density changes in breast cancer patients treated with proton therapy and compare the effect to photon therapy, assessing both temporal dynamics and spatial distribution of these changes. We prospectively collected serial non-contrast CT scans on a cohort of 91 patients at 6- and 12-month post-RT. All patients were co-enrolled on a randomized clinical trial and a correlative substudy at [Institution] between 2017 and 2023. Deformably registered planning and follow-up CT scans were analyzed using voxel-wise Hounsfield unit (HU) changes binned by dose (5 Gy(RBE) increments) within the ipsilateral lung. Dose-response relationships were modeled using the Lyman-Kutcher-Burman (LKB) model to estimate the dose associated with 50% of the observed maximum lung density changes in HU (TD₅₀) and slope (m) parameters as comparative metrics between modalities. Bootstrap analysis with 10,000 iterations was used to determine confidence intervals and assess statistical significance. Relative biological effectiveness (RBE) was calculated as the ratio of photon to proton TD₅₀ values per voxel. Demographic and clinical variables were tested to rule out confounding. A total of 71 patients had complete follow-up imaging (36 protons, 35 photons). Proton therapy demonstrated significantly lower TD₅₀ values than photon therapy at both follow-up (FU) visits, indicating greater radiation sensitivity. The right-sided free-breathing cohorts showed TD₅₀ of 33.4 Gy(RBE) for protons versus 45.1 Gy for photons at 8.5 months, with RBE of 1.35 [95% CI: 1.18-1.44], decreasing to 1.27 [1.14-1.36] at FU2. The overall population showed TD₅₀ of 34.5 Gy(RBE) for protons versus 43.6 Gy for photons at FU1 with an RBE of 1.27 [1.14-1.33], decreasing to 1.24 [1.12-1.33] at FU2. No significant confounders were identified. Temporal analysis revealed an earlier onset of HU changes with protons, but there were no significant associations between patient pulmonary toxicities and TD₅₀ values. No patients developed clinically relevant pneumonitis or fibrosis. Cough and dyspnea rates were similar between groups (p>0.40), with no correlations between HU changes and symptoms. Despite equivalent clinical pulmonary outcomes, proton therapy demonstrates greater subclinical lung tissue changes than photon therapy. These findings support ongoing evaluation of potentially LET- and RBE-based optimization strategies in proton therapy planning and highlight the need for personalized approaches accounting for RBE variability.