ESTRO 2021 Abstract Book

S859

ESTRO 2021

management of small volume brain metastases. SRS can provide local control rates between 70-90% with a reduction in cognitive side effects compared to WBRT. We carried out a single institution retrospective analysis to evaluate local control (LC) and overall survival (OS) in patients treated with SRS for brain metastases from a range of primary tumour sites. Materials and Methods We conducted a retrospective review of all patients treated with SRS between January 2017 and February 2020 at our institution. Collected data included patient demographics, primary malignancy, volume of metastases, whether patients received WBRT or underwent neurosurgery, Recursive Partition Analysis (RPA) class and systemic treatment (SACT). The treatment planning system used was BrainLab iPlan for delivery of Dynamic Conformal Arcs using a Varian TruebeamTM STx. Dose/fractionation delivered was 15-25Gy in 1-5#, selected in line with institutional protocol. Outcome data analysis was conducted using R survminer package. Results A total of 125 lesions from 77 patients treated with SRS were included in this analysis. The most common primary sites were NSCLC 34%, melanoma 29%, renal 16% and breast 15%. The majority of lesions (74%) received 21Gy/1# prescribed to 80% isodose. 17 lesions (14%) were treated with a fractionated schedule. 14 (18%) patients also received WBRT, 4 prior to SRS. Most patients had a single lesion treated (58%) with a maximum of four lesions treated in one episode. Median volume of treated lesions was 1.8cc (range 0.03 – 20.2cc). Most patients were RPA Class 2 (71%) followed by RPA Class 1 (26%). Over half of patients proceeded to SACT following SRS, at a median interval of 6.5 months (excluding ‘adjuvant’ treatment). Breakdown as follows: 26% checkpoint inhibitors, 18% chemotherapy, 9% tyrosine kinase inhibitors. Radionecrosis was pathologically confirmed in two lesions (1.6%) and suspected based on MRI in a further three (2.4%). Median OS was 15 months. 12 month LC was 73% (fig. 1) with OS 52%. We did not observe a difference in LC by primary tumour site (p=0.58), or with fractionation (median 39mo vs 39mo, p=0.58). Median duration of LC was 26 months for melanoma, 31 months for lung and not reached in breast and renal cancers. No improvement in OS was noted in patients who also had WBRT (p=0.86). Larger tumour volume (> 2cc) was associated with significantly worse LC (median 18mo vs 44mo, p< 0.01) and OS (median 9mo vs 30mo, p<0.01) (fig. 2). BED10>60Gy (achievable with smaller tumour volumes) and surgery were both associated with improved OS (p=0.01, p=0.02).