e21042 Background: Secreted protein acidic and cysteine-rich (SPARC) was highly expressed in non-small cell lung cancer (NSCLC) tissue. Studies have indicated the association between SPARC expression and immune-cell infiltration and the prognosis of lung cancer. Despite the optimal first-line treatment approach in advanced NSCLC that lacks a driver mutation includes immunotherapy with or without chemotherapy, a well-defined tool or test for predicting response has yet to be created. We aimed to elucidate the influence of SPARC expression terms of clinicopathology, pembrolizumab response and prognosis in metastatic NSCLC patients. Methods: Thirty-six patients diagnosed with advanced-stage NSCLC without actionable driver mutation and recieved pembrolizumab in the first line setting were included in this study. Programmed death ligand-1(PD-L1) and SPARC expression were assessed by means of PD-L1 IHC 22C3 pharmDx and BenchMark ULTRA assay(Ventana Medical Systems, Tucson, AZ) in paraffin-embedded blocks. The expressions were categorised according to the tumor proportion score. Staining intensity of SPARC was graded according to the following criteria: 1+ (weak), 2+ (moderate), and 3+(strong). Results:The median age was 62 (range 45-81). One patient was female, and four patients did not have a smoking history.Twenty-seven of tumor were adenocarcinoma, and six were squamous cell carcinoma. PDL-1 status was < 1% in 4 (11.4%), 1-49% in 12 (34.3%), and > 50% in 20 (54.3%) patients.There was no significant correlation between SPARC expression and smoking status, histologic type of tumour, T and N status, and liver and bone metastasis. Central nervous system (CNS) metastasis and progression was seen in 8 (57.1%), 2 (12.5%), and 1 (16.7%) and in 6 (16.7%), 0, and 6 (37.5%) patients with SPARC 1+,2+, and 3 +, respectively. Controversially, higher SPARC expression was significantly correlated with lower rates of brain metastasis but higher rates of CNS progression (p = 0.022 and p = 0.011, respectively). While PD-L1 and SPARC expression were not significantly correlated, a trend toward low SPARC expression was seen in patients with PD-L1 > 50%. The objective response rate (ORR) trended to be higher in the SPARC 1+ group (85.7%, 43.8%, and 50.0% in SPARC 1+, 2+ and 3+ group, respectively, p = 0.052). Univariate analysis revealed that SPARC expression was not a significant prognostic factor for both PFS (p = 0.7) and OS (p = 0.07). But, low SPARC expression was a significant predictive factor for the first-line pembrolizumab treatment response (p = 0.04, OR:0.11, 95%CI 0.01-0.92) in NSLC patients. Conclusions: Our study lightened a new area in the literature that SPARC expression might predict the pembrolizumab response and might be a marker for CNS progression in NSLC patients. These results should be explored in further studies to address the potential therapeutic implications of SPARC expression.