ESTRO 38 Abstract book

S498 ESTRO 38



Homogeneity Index [2] (CTV – GTV only): HI = (D 2% - D 98% ) / D 50% , Conformation Number [3] (GTV only): CN = GTV 95% /GTV * GTV 95% /V 95% . Plans with above-median QF and HI and below-median CN were selected to train a linear regression model to predict D 98% in the GTV using a 5-fold cross validation scheme. Training was performed on the first 3 principle components (PCs) per Overlap Volume Histogram (OVH) of PTV, GTV, bladder and rectum (in total 10 OVHs). Subsequently, the model was applied to the remaining plans to identify treatment plans with more than 10 Gy difference between planned and predicted dose in the GTV. Results Dose-volume parameters are shown in Table 1. D 50% and D 98% were above the threshold of 82.4 Gy in 99 and 71% of the GTVs, respectively. We observed comparable bladder and rectum dose-volume parameters between both arms of the trial. Median QF and HI of all dose-escalated plans were 5.4% and 26.9%, median CN of all GTVs was 0.25. Seventy-eight out of 374 GTVs were selected to train both models. For D 98% the best predictive model included 5 PCs. Eleven GTVs were identified with a planned D 98% more than 10 Gy lower than predicted, corresponding with 9 treatment plans. Figure 1 displays the predicted versus planned D 98% in the GTVs of both training and test data. 

Conclusion For skull-base meningiomas IMPT allows for a considerable dose reduction in the hippocampi and normal brain compared to both IMRT and VMAT, which may lead to a clinically relevant reduction of late neurocognitive side effects. PO-0928 Treatment plan quality assessment for focal dose escalation in prostate cancer M. Van Schie 1 , T. Janssen 1 , D. Eekhout 1 , I. Walraven 1 , A. Kotte 2 , L. Kerkmeijer 2 , C. Draulans 3 , K. Haustermans 3 , R.J. Smeenk 4 , U. Van der Heide 1 1 Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands ; 2 University Medical Center Utrecht, Radiation Oncology, Utrecht, The Netherlands ; 3 University Hospitals Leuven, Radiation Oncology, Leuven, Belgium ; 4 Radboud University Medical Center, Radiation Oncology, Nijmegen, The Netherlands Purpose or Objective In a multicenter randomized trial of focal prostate dose escalation up to 95 Gy was prescribed to the tumor in the dose escalation arm, with 77 Gy to the entire prostate in both arms. However, as the dose constraints to normal tissue surrounding the target volume were leading, we investigated what extent of dose escalation to the tumor was achieved. In clinical practice, it can be challenging to assess whether a treatment plan is optimal in terms of achieved boost dose and dose to organs at risk (OARs). Therefore we developed a predictive model to identify treatment plans in which a higher boost dose could be achieved for the specific patient’s anatomy. Material and Methods We analyzed planning CT scans and clinical delineations together with co-registered planned dose distributions of 531 patients from 4 institutions. We compared dose characteristics for the standard and dose-escalated arms. For the latter we determined the percentage of plans with an D 50% and D 98% above 82.4 Gy (107 % of 77 Gy), and quantified the correspondence with prescribed dose using the following parameters:  Quality Factor [1] : QF = 1/n * ∑ |1- D planned /D prescribed |,

Conclusion In this study we observed that a relevant dose escalation above 82.4 Gy was realized in the majority of patients. Meanwhile we showed no differences in OAR dose between both arms of the trial. Although limited dose escalation to a large extent can be assigned to unfavorable anatomy, we developed a model that identified treatment plans where nonetheless a higher boost dose was predicted to be achievable. [1] B. Vanderstraeten et al, PMB 2006 [2] Q. Wu et al, IJROBP 2003 [3] A. van ‘t Riet et al, IJROBP 1997