Abstract Book

S327

ESTRO 37

Conclusion Our newly developed technique is significantly more robust against setup errors and patient motion for plans with multiple isocenters such as craniospinal Irradiation.

OC-0618 Tumor dose inhomogeneity and outcome after stereotactic radiotherapy for brain metastases F. Lucia 1 , S. Key 1 , L. Olivier 1 , G. Dissaux 1 , G. Goasduff 1 , A.S. Lucia 1 , O. Pradier 1 , U. Schick 1 1 CHRU Brest- Hôpital Morvan, Radiation oncology, Brest, France Purpose or Objective To analyse the impact of inhomogeneous versus homogeneous dose distribution on local control (LC), freedom from distant brain metastases (FFDBM), overall survival (OS), and radionecrosis (RN) in patients treated with fractionated stereotactic radiotherapy (SRT) for newly-diagnosed brain metastases. Material and Methods From 2014 to 2017, 134 patients with a median age of 61 years (range, 26-86) underwent SRT for brain metastases (114 with ≤2 and 20 with 3-6 metastases) from various primaries (breast, pulmonary, colo-rectal, renal, melanoma) at our institution and were included in this analysis. Treatment was delivered using volumetric modulated arc therapy (VMAT) on a flattening-filter-free (FFF) linear accelerator equipped with a 2.5 mm multi- leaf collimator (MLC). Ninety-one consecutive patients with 136 metastases were irradiated at a dose of 23.1Gy in 3 fractions of 7.7Gy delivered homogeneously (99% of the dose had to cover 99% of the PTV) (group 1). After modification of the SRT protocol, an inhomogeneous dose of 11Gy prescribed to the isocenter with the 70% isodose line covering the PTV was delivered in the following 43 patients (72 metastases) (group 2). Variables analyzed for LC, FFDBM, OS, and RN included dose distribution, age (<58 vs ≥58), gender, histology (breast cancer vs others), diagnosis-specific Graded Prognostic Assessment (DS-GPA) score (≤2.5 vs >2.5), number of brain metastases (≤2 vs >2), presence of extracranial metastases, volume of GTV (≤2.1 vs >2.1 cc) and PTV (≤2.3 vs >2.3 cc). Results After a median follow-up of 8.5 months (range, 2.5-30.0), the 6-months LC, FFDBM, OS and RN were 95%, 81%, 86%, 3%, respectively. The 6-month LC was 93% (106/136) in group 1 and 100% (72/72) in group 2 (p=0.01). The 6- months FFDBM (p=0.13), OS (p=0.13) and RN (p=0.11) showed no association with SRT planning. In multivariate analysis, improved LC was significantly correlated with SRT dose distribution (p=0.03), age (p=0.04) and number of brain metastases (p=0.04), while PTV volume (p=0.001) and number of brain metastases (p=0.002) were associated with FFDBM. Histology (p=0.0003), DS-GPA score (p=0.0004), PTV volume (p=0.003) and number of brain metastases (p=0.004) were associated with OS. RN was significantly increased in case of large PTV volume (p=0.04) and in the presence of 3 or more metastases.

Conclusion SRT delivered with inhomogeneous dose distribution resulted in better LC than with homogeneous distribution. FFDBM, OS and radionecrosis were not significantly different. OC-0619 Robust optimization for head and neck cancer patients with SIB-VMAT: Planning and simulated delivery S. Cucchiaro 1 , C. Ernst 1 , P. Piret 1 , X. Werenne 1 , P. Coucke 1 , V. Baart 1 , D. Dechambre 1 , A. Gulyban 2 1 C.H.U. - Sart Tilman, Radiotherapy department, Liège, Belgium 2 Europe Hospitals Brussels, Radiotherapy department, Brussels, Belgium Purpose or Objective Robust optimization during the treatment planning could account for setup uncertainties, thus providing an alternative to clinical (CTV) to planning target volume (PTV) margins. In our current investigation, we aimed to evaluate 1) the organs at risk (OAR) dose difference between robust (ROB) vs. classical (PTV) optimization and 2) the risk of reduced CTV coverage by simulated treatment delivery using either external landmark-based (LM) or extended no-action level (eNAL) verification protocol for patient with head and neck cancer using volumetric arc therapy (VMAT) in a simultaneously integrated boost (SIB) approach. Material and Methods Between Jan and May 2017 twenty consecutive H&N patients with bilateral elective lymph node irradiation were selected for this study. All plans consisted of a SIB dual VMAT in 30 fractions for two or three target/dose levels (total dose ranged between 54-70 Gy) using Raystation (version 5.0.2, RaySearch, Stockholm, Sweden). For PTV plans 5 mm uniform CTV-PTV margin were used to achieve clinically acceptable plans. For ROB plans all PTV objectives were altered to CTV with 5 mm uniform robustness before an additional optimization. For delivery simulations random setup error samples from historical H&N data (N=6725) were generated and used for perturbed dose calculation (imitated isocenter shifts) with LM (= no correction) and eNAL (first 5 fraction, followed by weekly correction) verification protocols. OAR specific generalized equivalent uniform (gEUD) doses were used and compared using paired t-tests with p<0.05 significance level. Average target coverages were evaluated for each CTVs with their corresponding dose coverage (V95%>95%) based on the plans, LM and eNAL simulations. Results A total of 120 DVHs were analyzed. The setup errors ranged between (-1.34 and 0.72/-0.92 and 0.84 /-1.06 and 0.96 cm) for lateral, vertical and longitudinal directions. All optimization resulted in a clinically acceptable plans with lower OAR doses for ROB reaching