Introduction: Only 30% of patients with liver metastases (LM) may be suitable for surgery due to unfavourable location, comorbidities or extrahepatic disease burden. Patients treated with stereotactic radiotherapy (SBRT) reported excellent local control (LC) and low toxicity rate although data on global disease control are missing. The aim of this preliminary analysis was to assess patterns of failure in a cohort of patients treated with SBRT for LM. Method: Data from patients treated between 2018 and 2020 at our Institution with SBRT to LM receiving at least an EQD2 of 50 Gy (α/β=10) as per ESTRO consensus were collected. Failure patterns following SBRT and rates of local control (LC), intrahepatic relapse (excluding treated site, IHR), extrahepatic relapse (EHR), and overall survival (OS) were evaluated. Results: Forty-three patients received liver SBRT due to oligometastatic (20,46%) and oligoprogressive (23,54%) disease. Most common primary tumors were breast (n=18,42%) and colon (n=10,23%) cancer. SBRT was performed using Cyberknife real-time tumor tracking(n=30,70%) or abdominal compression-assisted VMAT (n=13,30%) delivering 35-60 Gy in 3-5 fractions, corresponding to median EQD2 of 94 (50-150) Gy. Twelve (28%) patients were chemotherapy-naïve, while the remaining patients received 1(20,46%), 2(5,12%) or ≥3(6,14%) chemotherapy lines. Median follow-up was 12 months. Patterns of failure are reported in Table 1. One-year LC, IHR, EHR and OS were 80%,51%,49% and 87% respectively. At multivariate analysis LC was significantly correlated with EQD2≥94Gy(p=0.009) and ≥3 chemotherapy lines(p=0.04). IHR and EHR were significantly associated with local failure (p=0.0013) and intrahepatic progression (p=0.03), respectively. A significant correlation between OS and local relapse was shown (p=0.026). Conclusion: In our experience, improved LC using high BED in non-heavily pretreated patients was correlated to reduced risk of IHR and to improved OS. IHR was the dominant mode of failure in patients treated with SBRT for LM,and was correlated to further progression at extrahepatic sites. Our findings suggest that IHR may result from uncontrolled macroscopic LM rather than ubiquitous micrometastatic dissemination, and preceed further systemic spread at distant sites. Our findings support the use of SBRT as an efficient tool to block stepwise metastatic spread from uncontrolled isolated LM to liver, and from liver to distant site, thus extending global disease control. [ABSTRACT FROM AUTHOR]