Objective Pancreatic ductal adenocarcinoma(PDAC)is a deadly cancer,and harborsextensive genomic alterations with clinical implications.However,the inaccessibilityof tumor specimens with adequate cellularity constitutes a major impediment inimplementation of genomics-driven precision medicine,in particular for metastaticlesions.Circulating cell-free DNA(cfDNA)offers an alternative option for tissuebiopsy in genomic characterization.Here we sought to evaluate the copy-numberalterations(CNAs)of metastatic pancreatic cancer by cfDNA profiling and determineits clinical significance.Methods Our study included plasma and tumor samples from 72 individuals with pathology-proven PDAC.Among them,67 patients had metastatic disease and other 5 subjects were diagnosed as localized pancreatic cancer.The majority of patients were treatment-naive at the first blood drawing,and underwent chemotherapy as first-line treatment.Other six patients had received surgical resection and blood samples were obtained when liver metastasis were found postoperatively.Low-coverage whole-genome sequencing(5x)was performed,and copy number status and tumor fraction(TFx)were estimated using ichorCNA method.A bin-size of 1M was adopted for copy number estimation.We compared the CNAs between primary PDAC tissues derived from TCGA database and cfDNA.Serial plasma samples were obtained from a subset of patients,and genomic evolution could thus be analyzed.Results Somatic CNAs were observed in half of patients(36/72)through cfDNA analysis,with recurrent aberrations identified like loss of TP53,CDKN2A,and SMAD4.The sequencing depth did not affect the outcome of copy number profiles.Significantly more patients with liver metastasis had detectable CNAs than those with metastasis in other sites(lung or peritoneum)or without metastasis.Copy number profiles were largely concordant between local(primary tumor in TCGA database)and metastatic(cfDNA in our cohort)settings.However,genomic regions with altered copy number were much broader in cfDNA,and several PDAC-related alterations including KRAS and MYC amplifications were more frequent in the metastatic setting.These were confirmed in 5 subjects who had concomitant tissue and plasma samples available.Notably,there was much more subclonal alterations in cfDNA than tissues.TFx were inferred from cfDNA analysis and ranged from 0 to 78%.TFx correlated with the presence of liver metastasis,higher tumor burden measured on imaging evaluation,increased level of tumor markers including CA19-9,and shortened overall survival.In 14 patients with serial plasma analysis,CNAs were highly consistent in each patient with few newly occurring alterations following chemotherapy.Interestingly,patients had increased level of copy number gains of KRAS when resistance to chemotherapy developed.At the same time,TFx dynamics was in line with the kinetics of allele fraction of mutated tumor-related genes(KRAS,TP53,among others),therapy response and change of tumor burden.Conclusions Our findings suggest remarkably altered copy number landscape in metastatic PDAC exclusively through cfDNA analysis.Certain CNAs and TFx are clinicalrelevant and prognostic,with implications for metastasis,chemotherapy resistance,and novel therapeutic approaches.CNAs are mostly the early genetic event during oncogenesis,and show conserved characteristics following chemotherapy in the context of metastasis.KRAS amplification might be associated with resistance to chemotherapy.The use of cfDNA represents a novel and effective approach to tailor treatment in the era of personalized medicine.