86 Background: Genomic analysis of ctDNA from cancer patients can allow identification of actionable alterations of tumor without an invasive biopsy. ctDNA analysis has been reported to be superior for global summary of tumor heterogeneity and a short turnaround time compared to tissue analysis,however, there are potential issues including the detection sensitivity, threshold to determine TMB-H, and clonal hematopoiesis of indeterminate potential (CHIP). In order to elucidate the impact of ctDNA analysis on personalized cancer treatment, we decided to examine the genetic landscape and clinical outcomes of ESCC cohort registered in the C-CAT database. Methods: We retrospectively analyzed data from 1059 cases of ESCC registered in the C-CAT from June 2019 to February 2024. Clinical characteristics, TMB status, pathological gene variants, and time to treatment failure (TTF) of first-line treatment were examined. Results: The panels applied to eligible patients consisted of 925 tumor tissue panels (87.3%) and 134 liquid panels (12.7%). Pathological gene mutations were detected in 924 cases (99.8%) in tissue panels and 133 cases (99.2%) in liquid panels. The top 10 detected genes were TP53 (93.4%), CDKN2A (57.8%), CDKN2B (45.5%), CCND1 (43.9%), FGF19 (41.0%), PIK3CA (29.0%), NFE2L2 (28.3%), BCL6 (16.5%), NOTCH1 (13.5%), and KMT2D (12.2%). The detection rate for these genes excluding NOTCH1 and KMT2D were significantly higher in tissue panels than in liquid panels (P≤0.01). The median TMB was 5.00 Muts/Mb for both. In first-line treatment, the TTF for immune checkpoint inhibitors (ICIs) and chemotherapy was significantly longer in TMB-H (≧10 Muts/Mb) compared to TMB-L (3.1M vs. 1.9M, P<0.01) in patients evaluated by tissue panels, but it was not proven in patients evaluated by liquid panels (P=0.38). Furthermore, the results suggest that patients with CDKN2A, CCND1, and FGF19 variants benefit more by adding ICIs to chemotherapy than patients without these variants and while the opposite was true for BCL6 and NOTCH1 and KMT2D. Conclusions: In this ESCC cohort, detection rates in the liquid panels were generally lower than in the tissue panels, suggesting that sufficient ctDNA could not be obtained in many cases. The appropriate patient selection for the liquid panels and the timing of blood collection should be discussed. Furthermore, the ideal threshold for determining TMB-H in the liquid panels needs to be further investigated.