104 Chapter 7 was the best predictor for pulmonary events. The mean lung V5 in patients with and without a pulmonary event was 55% vs 45%, respectively (p=0.02). When combining all cardiac events, no significant associations were found between DVH parameters and cardiac events. However, a significant association was found between new rhythm disorders and the mean dose to the left atrium (MLAD). The MLAD was 32.4 Gy among those with new rhythm disorders and 27.5 Gy in those without (p=0.03). Details on associations between radiation dose and events are presented in supplementary data 2. During treatment and follow up, both serum levels of HS-TNT (p<0.001) and NTProBNP (p<0.05) increased as compared to their baseline values. One NT-ProBNP outlier was excluded from the analyses (details in supplementary data 3). In the longitudinal analyses, the rise of HS-TNT over time was related to several DVH parameters of the heart (e.g., MHD regression coefficient 0.25; p=0.02). When cardiac event history was included as potential confounding factor, even more DVH parameters of the heart became statistically significant and the effect seemed more pronounced, as the regression coefficient increased (e.g., MHD regression coefficient 0.31; p<0.01). Moreover, cardiac history was also significantly associated with the course of HS-TNT in the longitudinal analyses (regression coefficient 8.1, p<0.01)(Table 2A). When looking more specifically at the different timepoints, we found a significant relation for the absolute values of HS-TNT at the end of treatment with several DVH parameter of the heart, including MHD, V25, V30 and V40 heart. Whereas, for relative values compared to baseline, only the V35 heart correlated significantly (regression coefficient 2.49; p=0.03). At later time points, these associations were not statistically significant anymore (Table 2B). The use of substructures of the heart instead of whole heart DVH parameters did not change these results. The levels NT-ProBNP fluctuated over time in many patients and were not normally distributed. For this reason, we analysed the change of NT-ProBNP compared to baseline at the different time points. In the longitudinal analyses, we did not find any significant associations with cardiac or pulmonary radiation dose parameters. When analysing NT-ProBNP at different timepoints, a significant association was observed with several radiation dose parameters to the lungs (e.g., mean lung dose, p<0.01)) at the end of treatment but not at later time points (Table 2C).
RkJQdWJsaXNoZXIy MTk4NDMw