Introduction Circulating tumor DNA (ctDNA) which is tumor‐specific DNA sequences found in blood, has been considered an important new strategy that will aid in the treatment of non‐Hodgkin's lymphoma (NHL) because ctDNA could be used in various aspects of NHL management such as diagnosis prior to therapy and early prediction of treatment failure as well as disease relapse. High‐throughput DNA sequencing of ctDNA encoding tumor‐specific mutations also could serve as a surrogate for the entire tumor genome and overcome the barrier of spatial and temporal tumor heterogeneity. However, the role of ctDNA in blood is still not determined in primary central nervous system lymphoma (PCNSL) because it is not clear how blood-brain barrier affects the level of ctDNA in blood of patients with PCNSL. Thus, we performed this pilot study to explore clinical relevance of ctDNA in patients who were consecutively diagnosed with PCNSL. Methods The study population was from our prospective cohort study for lymphoma patients (NCT03117036) that has been enrolling patients since January 2017. All patients were enrolled after obtaining written informed consents, and plasma samples for the isolation of ctDNA and targeted sequencing were collected at diagnosis and after the interim and final response evaluation. Additional samples were also collected at the time of relapse. Using DNA samples extracted from plasma, high-depth targeted sequencing consisting of 66 genes was performed. Out of 853 patients enrolled in this cohort, we selected 42 patients who were diagnosed with PCNSL after excluding secondary CNS lymphoma. Results Their median age was 52 years (range, 19-85 years) and, male (n=22) was a slightly more than female (n=20). All patients included in the study were administrated with combination systemic chemotherapy consist of rituximab, high dose methotrexate, vincristine and procarbazine as the first line. By confirming the somatic genetic mutation of tissue samples, we have proven that the detected ctDNA originated from the parent brain tumor. In plasma analysis, targeted sequencing was performed for 41 patients except a patient as a DNA quality control failure. Among 41 patients, the ctDNA amount at the diagnosis was 13.6ng/mL (Range 3.3-121.6). Fifteen patients (36.58%) only carried at least one mutation within the target regions of the panel. BCOR, BTG2 and IDH2 were highly ranked mutated genes in our study, followed by BTG1, DNMT3A, DUSP2, HIST1H1E, PIK3CA, PIK3CD, and PIM. Unlike previously reported studies, we could not observe MYD88 mutation, but more wide range profiles of ctDNA obtained. Compared to ctDNA amount and brain MRI imaging at baseline, we measured the long axis of the largest brain mass and analyzed the association with ctDNA amount. We could figure out the correlation between the amount of ctDNA and brain mass size. The median follow-up duration was 8.9 months (range 1.1-71.2) and median overall survival was 45.8 months (NR). The patients detected ctDNA at the diagnosis did not show any particularly inferior survival outcome than patients without ctDNA (NR vs.45.8months, p-value=0.92). Conclusion In the results of evaluating only the baseline samples, the detection rate in our study was 37% in PCNSL. In addition, the amount of plasma ctDNA showed the possibility of clinical application to predict disease burden at the time of diagnosis. In the future, we would analyze serial ctDNA amount, after then investigate the correlation with chemotherapy response and residual disease monitoring. In conclusion, although more research and time are required for the settlement of ctDNA in clinical practice, it will be possible to realize precision medicine with the advantage of non-invasive procedure. Disclosures Kim: F. Hoffmann-La Roche Ltd: Research Funding; Celltrion: Research Funding; Novartis: Research Funding; Donga: Research Funding; Kyowa-Kirin: Research Funding; Novartis: Research Funding; J + J: Research Funding.