Are mean lung dose and changes in respiration during RT predictive for pulmonary function changes after RT?

Purpose/Objective: Radiotherapy (RT) of lung cancer patients could be improved if patient specific dose tolerances can be estimated during the first weeks of a fractionated treatment course. Such tolerances may be estimated from the delivered dose distribution and changes in ventilation and respiration patterns extracted from frequently recorded 4DCone Beam CT. Materials and Methods: This is a study of 140 non-small-cell lung cancer (NSCLC) patients, which were treated with 60-66 Gy in 30-33 fractions. Measures of ventilation were generated by the freeware tool elastix (http://elastix.isi.uu.nl/), which was used for deformable registrations between inspiration phases and expiratory phases as well as calculation of Jacobian determinant images of Elekta XVI 4D-CBCT reconstructions. Approximately 3000 recorded XVI scans were available for this study group. The mean of the Jacobian determinant (MJD) within the healthy lung region was used as a measure of the overall lung ventilation during the treatment course. Amsterdam Shrouds of the 4D-CBCT projections were generated by RTK-software, and the respiratory period (ReP) was estimated from the extracted respiratory signal. For each patient the slope of linear fits of MJD and ReP values vs. fraction number for the first 10 fractions as well as for the full number of fractions were used as measures for ventilation and respiration changes during RT. Relative changes (after/before-1) in forced expiratory volume in 1 second (FEV1) and forced expiratory vital capacity (FVC) before and after the treatment course were chosen as clinical endpoints (DELTAFEV1 REL and DELTAFVCREL). A multivariable analysis was performed on DELTAFEV1 REL and DELTAFVCREL respectively as dependent variables with the covariables MJD-slope, ReP-slope and mean lung dose (MLD). Results: The deformable registrations and calculating MJDs as well as estimating RePs from Amsterdam Shrouds were carried out successfully for all the 140 patients. In the multivariable analysis only MLD was significant in relation to DELTAFEV1 REL and DELTAFVCREL (DELTAFEV1 REL=-0.006MLD+0.078; pMLD=0.009 and DELTAFVCREL=- 0.008MLD+0.115; pMLD=0.002, respectively). Plots of DELTAFEV1 REL and DELTAFVCREL as a function of MLD are shown in the figure. Both plots show a declining tendency, indicating that higher MLD corresponds to larger decrease in clinically measured lung function tests. Spearman correlation coefficient between DELTAFEV1 REL and DELTAFVCREL versus mean lung dose were - 0.249 (p=0.003) and -0.252 (p=0.003), respectively. Conclusions: In this study it was possible to carry out the deformable registrations on the 4D-CBCT-scans, but for the current 4D-CBCTimage quality, it was not possible to show an association between changes in mean Jacobian or respiration period during RT with changes in FEV1 and FVC.