Abstract Book

ESTRO 37

S385

delineation of observer number three the constraint was violated with more than 3Gy for both hippocampi (Fig.1). The D mean and D mean biological constraints were met for all delineations.

exist about the risk of tumor recurrences in the area of target reduction. Recent data reported no increased local relapse and low pulmonary toxicity so, waiting for the ongoing phase III trial investigating this issue (RTOG 1106- ACRIN 6697), the aims of this study is to compare the long term outcome for patients treated or not with the adaptive approach. Material and Methods Fifty LA-NSCLC patients enrolled in a prospective study where treatment was concomitant CRT and replanning in case of tumor shrinkage, have been compared with 41 patients without tumor shrinkage treated in the same period. Toxicity was evaluated with the RTOG/EORTC scale. The differences between groups were compared by Fisher’s exact test (two tail) or χ 2 when appropriate. The 'time to event” curve was calculated with the Kaplan- Meier method, and log-rank test was used to perform between-group comparisons. Results Patients in the non-adaptive (NAG) and adaptive groups (AG) have the following characteristics: ≤70 years old 46% and 44%; male 76% and 78%; adenocarcinoma 56% and 32%, squamous cell 37% and 56%; stage IIIA 63% and 58%; chemotherapy before chemoradiation 32% and 46%; platin at systemic dose concurrently to radiotherapy 51% and 62%, respectively. Patients in the AG were more likely to receive a total radiation dose equal or higher than 59.4Gy (58% vs 27%, p=0.003). No statistical differences were reported in local recurrences, even if in the AG were lower (31% and 48%). Acute ≥G2 esophageal and pulmonary toxicity was similar, but G3 acute lung toxicity was lower than a third in AG (2% vs 7%) and G3 chronic lung damage reduced by half (7.5% vs 4%). Median follow up for alive patients was 57.8 months. Median OS were 26,6 and 30,5 months and PFS 7,6 and 8.3 months between NAG and AG, respectively. Survival was affected by the rate of shrinking with better result for patients reducing 25-50% of the initial volume (median not reached) in comparison with no-reduction or ≤25% patients (median 25 months) (p=0.016). An apparent contra-intuitive result was the lower survival in case of reduction >50% (median 23 months). PFS reflected the same observation (0-25%: 7,5 m, 25-50% 13,8 m, and >50% 7,4 m, respectively p<0.05). Local control at 12 months was not different between groups (52%, 54% and 58%), while metastasis-free survival at 1 year was 56%, 73% and 23%, respectively (p=0.02) (Figure 1).

Conclusion The posterior and anterior-medial hippocampus borders have the largest delineation variability. Despite substantial delineation variability indicated by the low ICC and CI gen the hippocampus D mean and D mean biological constraints were not violated, fulfilling the trial protocol (NCT01780675) dose constraints. This is most probably due to the 5mm PRV margin. Some hippocampus delineations with larger volumes than the median surfaces violate the D 1% constraint. Hippocampus delineation might be improved by training, adjusting the RTOG outlining protocol, and by adding expertise from neuroradiology. A more accurate hippocampus delineation might allow reducing the PRV margin, which would increase the volume of brain receiving the prescription dose. PO-0749 Is tumor shrinking during chemoradiation for LA-NSCLC a biomarker for outcome? S. Ramella 1 , C. Greco 1 , E. Molfese 1 , A. Iurato 1 , M. Miele 1 , S. Silipigni 1 , R. D'Angelillo 1 , L. Trodella 1 1 Campus Biomedico University, Academic Department of Radiation Oncology, Roma, Italy Purpose or Objective Adaptation of dose distribution to tumor reduction during radiotherapy is an innovative approach. Even if the advantage for lung toxicity is intuitive, some concerns

Conclusion Waiting for randomized phase III results, adaptive approach confirms its role in escalating dose and reducing toxicity without compromise outcome. The worse outcome in patients with >50% reduction could be explained by high proliferating aggressive tumor behavior. The value of the shrinking rate as a biomarker for survival deserves to be investigated in future trials at the aim to intensify treatment in selected population.