Delivered dose can be a better predictor of rectal toxicity than planned dose in prostate radiotherapy

Abstract

Background and purposeFor the first time, delivered dose to the rectum has been calculated and accumulated throughout the course of prostate radiotherapy using megavoltage computed tomography (MVCT) image guidance scans. Dosimetric parameters were linked with toxicity to test the hypothesis that delivered dose is a stronger predictor of toxicity than planned dose.Material and methodsDose–surface maps (DSMs) of the rectal wall were automatically generated from daily MVCT scans for 109 patients within the VoxTox research programme. Accumulated-DSMs, representing total delivered dose, and planned-DSMs, from planning CT data, were parametrised using Equivalent Uniform Dose (EUD) and ‘DSM dose-width’, the lateral dimension of an ellipse fitted to a discrete isodose cluster. Associations with 6 toxicity endpoints were assessed using receiver operator characteristic curve analysis.ResultsFor rectal bleeding, the area under the curve (AUC) was greater for accumulated dose than planned dose for DSM dose-widths up to 70 Gy. Accumulated 65 Gy DSM dose-width produced the strongest spatial correlation (AUC 0.664), while accumulated EUD generated the largest AUC overall (0.682). For proctitis, accumulated EUD was the only reportable predictor (AUC 0.673). Accumulated EUD was systematically lower than planned EUD.ConclusionsDosimetric parameters extracted from accumulated DSMs have demonstrated stronger correlations with rectal bleeding and proctitis, than planned DSMs.

Patterns in ano-rectal dose maps and the risk of late toxicity after prostate IMRT

“…Scaife et al [39] showed that there can be large differences in DSM pixel values when using planned and delivered dose, respectively. As shown convincingly in [14], models based on planned dose have a lower predictive power due to the lower accuracy of the input data for model fitting. However, the improvement in predictivity from using delivered dose was rather limited and the relative importance of different spatial parameters was generally preserved.…”

“…Many investigators have studied the risk of rectal complications from radiotherapy in terms of dose-volume histogram (DVH) data [1]. However, a limitation of a DVH is that the spatial distribution of dose is disregarded; therefore, several studies have recently considered spatial features [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] of the three-dimensional (3D) dose distribution in order to improve the performance of predictive models. The development of ever more elaborate normal-tissue complication probability (NTCP) models is driven by the concern that a poor characterization of the dose-response relationship may unnecessarily limit the effectiveness of the treatment; a simplistic NTCP model could imply conservative dose-volume constraints to the organs-at-risk (OAR), or cause avoidable suffering through a suboptimal distribution of the dose to OAR for a given tumor dose (e.g., Ref.…”

“…The tubular structure of the rectum allows the delivered dose to be easily represented by dose-surface maps (DSMs), by virtually unfolding the rectum and visualizing the resulting dose distribution in two dimensions (2D). To search for significant dose patterns, several investigators co-register the DSMs or 3D patient volumes and derive the 'mean dose distribution' for the patients with and without toxicity, respectively [7,8,11,14,17,21,22], i.e., a dose distribution where each voxel contains the mean value of that voxel for all patients. Then, a voxel-wise t-test can visualize the areas/volumes where the difference in dose is significant.…”

“…The largest study to date [5], including 361 patients treated with 3D-conformal radiotherapy (3D-CRT), showed that late rectal bleeding correlated the strongest with the lateral extent of the 60-Gy isodose (in 32 or 37 fractions). The risk of late rectal bleeding and proctitis as a function of the lateral extent of selected isodoses has also been studied for 109 patients treated with tomotherapy [14]. Importantly, their model was based on the accumulated dose rather than planned dose.…”

Refinement & validation of rectal wall dose volume objectives for prostate hypofractionation in 20 fractions

“…Therefore, our results are to be used in the context of intensity modulation with daily IGRT [12]. This also implies that the deterioration of the dose distribution during the course of treatment is not accounted for, though the improvement in the accuracy to predict LRB between the planned and delivered dose in prostate radiotherapy seems quite modest [41]. Moreover, we have shown in a previous study on a group of similar patients treated with conventional fractionation (otherwise identical in target volume and RW delineation as well as PTV margins) that rectal wall DVHs randomly change during treatment, but this does not have a significant impact on the probability of side effects [42].…”