ESTRO 2021 Abstract Book

S1158

ESTRO 2021

schedules were registered depending on tumor type, localization and stage. Most frequent schedule for BCC was 51Gy/17 fx; 5fx/wk. The majority (65%) could be treated with 70-100Kv RT. Different margins were reported but most frequent margin was 7-10 mm. For SCC only, 7% was reported treated with low-kilovoltage RT; 35% received electrons and the rest photon- based RT. Accordingly, a wide range of doses and fractions were used. As the database is only one year old, only few follow-up data are yet reported. Among those, who have had follow-up 6 month after end of RT, 161 (66%) were photo documented and 98% of the pts reported a cosmetic acceptable outcome. Conclusion After one year the national database structure works satisfactory and data are entered at an acceptable rate. Low-voltage RT is the predominant modality for BCC and moderate hypofractionation is frequently used with acceptable cosmetic results. PO-1409 Imaging and blood biomarkers in metastatic melanoma patients treated with immunotherapy. S. Burgermeister 1 , G. Hubert 2 , L. Basler 1 , S. Hogan 3 , M. Pavic 1 , M. Bogowicz 1 , D. Vuong 1 , S. Tanadini-Lang 1 , R. Förster 1 , M. Huellner 4 , R. Dummer 1 , M. Levesque 5 , M. Guckenberger 1 1 University Hospital and University of Zurich, Radiation Oncology, Zurich, Switzerland; 2 University Hospital and University of Zurich, Radio-oncology, Zurich, Switzerland; 3 University Hospital Zurich, University of Zurich, Dermatology, Zurich, Switzerland; 4 University Hospital and University of Zurich, Nuclear Medicine, Zurich, Switzerland; 5 University Hospital and University of Zurich, Dermatology, Zurich, Switzerland Purpose or Objective Since the introduction of immune checkpoint inhibitors, patients with metastatic melanoma often show long- term response. However, not all patients benefit from this treatment. Additionally, some patients exhibit temporary progression followed by regression or even remission. These highly variable response patterns underline the need for reliable patient response models. In this study, we explore imaging and blood biomarker contribution to survival prediction in a cohort of patients with metastatic melanoma treated with single or double checkpoint inhibition. Materials and Methods 102 metastatic melanoma patients treated in the University Hospital Zurich between the years 2013-2019 were included in this analysis. Patients were given anti-PD1 or a combination of anti-PD1 and anti-CTLA-4. PET/CT imaging was performed at three time-points (baseline: TP0, 3 months: TP1, and 6 months: TP2). All lesions equal or larger than 1 cc were segmented. Tumor burden (TB) at each time-point was measured as the sum of longest diameters. Additionally, the relative change of TB with respect to the baseline was also calculated. The patient response at TP2 was evaluated using iRECIST. Standard blood markers used in clinical practice were also analyzed. Cox proportional hazards model was used for survival analysis. Performance of the models was evaluated with hazard ratio (HR) and concordance index (CI) using 5-times-repeated 5-fold cross- validation. Risk groups were formed according to the log-partial hazard (log(PA)). Results iRECIST progressive status at TP2 was associated with significantly higher mortality, with HR=11.6 (3.1-22.8) and CI=0.72 (0.65-0.79). TB at TP1 and TP2 provided similar predictive power, with CI of 0.69 (0.61-0.78) and 0.74 (0.66-0.82), respectively. However, the relative change in TB showed even better scores with CI of 0.75 (0.67-0.82) and 0.78 (0.70-0.86), respectively. The number of new lesions appearing during follow-up was also a predictive factor (HR=1.5 (1.4-1.6) CI=0.77 (0.7-0.84)). High LDH levels translated into higher mortality: HR=1.7 (1.2-2.3) with CI=0.60 (0.51-0.69) at TP1 and HR=2.6 (1.9-3.5) with CI=0.72 (0.63-0.80) at TP2. LDH was only weakly correlated with TB (Pearson’s r = 0.4 (0.24-0.53); Fig. 2). A multivariate survival model including TB change, LDH, and appearance of new lesions showed superior predictive performance, with CI=0.85 (0.79-0.91).

Fig. 1: low-risk group log(PA)<0, high-risk group(log(PA)>=0)