Margriet Kwint

Chapter 2 30 Treatment outcome LF and RF were defined as an in-field failure within the PTV of the primary tumour (LF) and the involved lymph nodes (RF). In both cases, pathologic confirmation or an increase in tumor diameter of at least 20% compared to the previous CT-scan was scored as a failure (according to RECIST) and sometimes confirmed by PET. LF and RF were calculated from date of diagnosis until first date of failure, last date of FU or death. OS was calculated from the date of pathologically proven diagnosis until last date of FU or death. Progression-free survival (PFS) was calculated from the date of pathologically proven NSCLC until the date of first failure (local, regional, distant), last date of FU or death. LF, RF and PFS were classified based on FU-records, imaging reports of tumour progression and repeat CT-scans. Statistical analysis Patient and tumour characteristics at baseline are presented as the mean (+standard deviation (SD) or the median (+ interquartile range, IQR) and proportions in case of a categorical variable. The independent samples T-test was performed to compare characteristics in case of a normal distribution (age andOAR dose volume parameters). The Mann-Whitney U-test was used to compare continuous variables in non-normal distributions (volume of the primary tumour and lymph nodes). The Pearson’s chi- squared test was performed to compare the patient and tumour characteristics as well as the toxicity and failure rates in case of binary, nominal or ordinal variables and a non-normal distribution. Kaplan-Meier survival curves for the reference-cohort and reduction-cohort were plotted for each endpoint. Log-rank tests were performed to assess differences in late toxicity, LF, RF, OS and PFS between the reference-cohort and reduction-cohort. Proportional hazards assumptions for each model were tested by interpretation of the survival plots. Cox proportional hazards analyses were performed to assess the independent effect of dose-reduction on each endpoint. First, a univariate model was constructed. Then, we subsequently adjusted for possible confounding or mediating variables. The variables included are known to be associated with dose-reduction and/or are closely related to outcome. To assess improvement of the model, percentage changes in the HR ≥10% were tested [20]. Since the toxicity profile of cCRT is different compared to sCRT and RT-only (sCRT/RT), a subgroup analysis was performed between cCRT and sCRT/RT [21]. P-values <0.05 were considered statistically significant. The data were analysed using SPSS software, version 25.0, for Windows (IBM).