Daily online contour adaptation by radiation therapists 45 significant reduction of contour adaptation times by improving the accuracy of the propagated contours, potentially removing the need for contour adaptation altogether. We found a median relative delineated volume difference of 9.5% between the first and subsequent fractions in the RTT group and adaptations by Observer 1 and 2 did not have great impact on these volume differences. Most patients (22/30) showed only increased contour volumes in fraction 2 to 5 compared to the first fraction. However, we could not identify a trend in the prostate volume over the course of treatment. Firstly, it could be that there is a tendency to enlarge the propagated contour instead of ‘shaving-off’ parts where the contour is too large, with the aim of not missing any part of the prostate. Secondly, there could be an actual increase in prostate volume over the course of radiotherapy treatment. Several reports have been published with varying results.16–19 While some studies showed an overall reduction of prostate volume at the end of treatment, these patients were treated with ³ 38 fractions and a maximum fractional dose of 2.0 Gy.16–18 Both King et al.16 and Nichol et al.17 reported an initial increase early in the course, the latter reporting a volume increase up to 34%. Gunnlaugsson et al.19 reported on volume changes in patients treated with ultrahypofractionated radiotherapy (7 x 6.1 Gy). They showed a mean increase of 14% mid-treatment and 9% at the end of treatment. These results could support our findings of increased volumes for all fractions compared to the baseline volume for the majority of patients. Still, we cannot conclude that the volume changes we observed are completely due to actual prostate volume changes, instead of (in part) interobserver variability. Visual inspection of the contours showed some variation mainly in the apex and base region of the prostate and this was confirmed by the adaptations by Observer 1 and 2. While adjustments were made in about half of the fractions, these adjustments generally consisted of adjusting, adding, and/or removing one to three contour slices. Both the apex and base of the prostate are sometimes poorly visible on the T2-weighted MR scans that are currently used in our clinic for daily imaging and contour variability could be reduced with enhanced image quality. As stated before, these are the same areas that have been characterised in literature as being prone to interobserver variability.9,10 Although no statistical testing was performed, the recalculated DSC clearly reflect that the adjustments did not affect DSC in a significant way. Most interobserver DSC values were > 0.95, which is comparable to interobserver variability as discussed earlier.14 For the fraction with lowest interobserver DSC (0.91 for patient 4, fraction 4, as visualised in Figure 3), the seminal vesicles were partly missed in the CTV delineation. The high overall interobserver DSC can be explained by the relatively small (volume) changes that have been made. The question remains whether or not these adaptations have clinical consequences. In case of the current 5 mm PTV margin, one can argue that these adaptations mostly fall well within these margins. Thus, especially when taking the dosegradient of external beam radiotherapy into account, these minor adaptations are not likely to influence the target coverage in a significant way. This view might change when smaller margins are used. In seven fractions (5.8%), contour adaptations by Observer 1 and 2 were larger and potentially clinically relevant, as judged by Observer 3. To estimate the effect on the CTV dose for these seven fractions, we calculated DVHs for all CTV contours using the RTT contour-based online dose distribution (Figure 5). Also, we calculated D99% for the CTV (Table S1, Supplementary B). For 2
RkJQdWJsaXNoZXIy MTk4NDMw