The Low-Dose Bath Paradox: Do Spatial Irradiation Patterns Play a Role in the Incidence of Radiation Pneumonitis Following PSPT or IMRT?

Abstract

Radiation pneumonitis (RP) is a common side effect of thoracic irradiation, whose incidence has been associated with several dose-volume histogram (DVH) predictors, including the lung volume receiving at least 20 Gy (V20), the mean lung dose and V5 [Pinnix et al. IJROBP 2015; 92(1):175]. Nonetheless, a randomized trial of IMRT vs passively scattered proton therapy (PSPT) for non-small cell lung cancer (NSCLC) [Liao et al. J Clin Oncol 2018; 36(18):1813] did not show any difference in RP incidence, despite PSPT significantly spared healthy lung tissues, with a p-value of the null hypothesis on V5 of 10-10. Following the recent findings on regional dose-sensitivity of the lungs [Palma et al. IJROBP 2016; 96(1):127], this study aims to explore whether the spatial location of dose delivery in PSPT and IMRT patients could account for the observed mismatch between DVH predictors and RP outcome in the trial. We analyzed 178 patients prospectively treated with PSPT (64) or IMRT (114) for NSCLC to a prescribed dose of 66 or 74 Gy(RBE) in conventional daily fractionation with concurrent chemotherapy. 32 (28%) of IMRT patients and 23 (36%) of PSPT patients developed any CTCAE-grade RP, while 22 (19%) of IMRT patients and 18 (28%) of PSPT patients developed clinically symptomatic RP. A Voxel-Based Analysis (VBA) of local dose differences in healthy tissues was performed according to a non-parametric permutation test accounting for multiple comparisons and possible non-dosimetric confounding factors. From the obtained 3D significance maps, we derived the clusters of voxels that exhibit significant dose differences between groups. The VBA highlighted that:1.significant dose differences between patients with and without RP are found in the lower part of the lungs and in the heart, independently from the RP severity threshold used for patients’ grouping;2.PSPT significantly reduced dose to healthy tissue in the medial-anterior and upper parts of the thorax;3.the regions spared by PSPT and the clusters with doses significantly correlated to RP development are disjoint. In particular, the mean dose differences between PSPT and IMRT in the clusters of voxel associated with the RP development are not significant (p=0.39 and p=0.51 for clusters derived for any CTCAE-grade RP and symptomatic RP, respectively). The analyzed trial data provided a valuable opportunity to substantiate previously reported hypotheses on the prominent role of the lower parts of the lungs in the development of RP. Most importantly, the VBA of dose distribution associated to treatment modality showed that the tissues significantly spared by PSPT seemed not strongly sensitive for RP development. This could solve the seeming inconsistency between V5 distribution and RP incidence in the studied trial. The relationship between RP and thoracic regional radio-sensitivity should be considered in clinical practice and in the design of future trials.