Session Item

Gated irradiation during mechanically-induced deep inspiration breath-holds (DIBH) is a recent solution allowing to accurately treat lung, liver or breast tumors. In addition to smaller safety margins for mobile tumors, the larger lung inflation with mechanically-assisted and non-invasive ventilation (MANIV) also contributes to improve lung sparing. In parallel, lung functional avoidance radiotherapy (LFAR) is an advanced planning technique that reduces the dose to the most ventilated regions. Combining MANIV and LFAR would be a further step forward in protecting lung function. However, it is not known whether the increased lung volume with MANIV is related to homogenous or heterogenous hyperfunction of functional zones and/or to recruitment of poorly ventilated areas. In the latter two cases, the LFAR would be biased by giving inadequate weight to some functional regions and sparing artificially ventilated areas that contribute marginally to normal respiratory exchanges, respectively. The objective of this work was therefore to compare the pulmonary ventilation pattern with and without mechanical ventilation and to discuss the possible consequences for LFAR.

This study examined 60 patients eligible for adjuvant breast radiotherapy who were included in a randomized trial (NCT04457102 on ClinicalTrials.gov). Each patient was simulated with a first 3D fast free breathing CT and a second CT during a voluntary DIBH (vDIBH arm, n = 30) or a mechanically-induced DIBH (MANIV-DIBH arm, n = 28) depending on the treatment arm.  As detailed in Figure 1, the lung functional map was computed with the Jacobian method based on non-rigid registration between DIBH and free-breathing CT. A voxel-based analysis was then performed: the functional maps were spatially normalized to a reference patient. The voxel-wise statistical analysis was carried out using a non-parametric multiple comparisons permutation test (1000 permutations). Functional lung subregions were defined for both arms on the mean functional image as voxels that showed no more than 15% ventilatory defects.

As shown in Figure 2, the islets of voxels showing a statistically significant difference in ventilation were mostly located in functional areas. Indeed, only 6.67 % of the non-functional voxels with voluntary DIBH became functional under mechanical ventilation. The increase in lung volume with MANIV-DIBH resulted in a homogeneous scaling of the vDIBH functional map according to a regression line which minimized the variations of the less functional zones. Dice similarity index between functional subregions from both arms was 93,26 %.

The ventilation pattern under mechanical ventilation consists of a uniform increase in the function of spontaneously ventilated areas with minor recruitment of non-ventilated lung alveoli. Functional volumes are therefore barely changed and the impact on LFAR should be minimal.