BackgroundProstate cancer (PCa) is among the leading causes of cancer mortality. Dicycloplatin is a newer generation platinum-based drug that has less side effects than cisplatin and carboplatin. However, its effects in PCa is mixed due to lack of appropriate stratifying biomarkers. Aiming to search for such biomarkers, here, we analyze a group of PCa patients with different responses to dicycloplatin.MethodsWe carried out whole-exome sequencing on cell-free DNA (cfDNA) and matched leukocyte DNA from 16 PCa patients before treatment with dicycloplatin. We then compared the clinical characteristics, somatic mutations, copy number variants (CNVs), and mutational signatures between the dicycloplatin-sensitive (nine patients) and dicycloplatin-resistant (seven patients) groups and tested the identified mutations, CNV, and their combinations as marker of dicycloplatin response.ResultsThe mutation frequency of seven genes (SP8, HNRNPCL1, FRG1, RBM25, MUC16, ASTE1, and TMBIM4) and CNV rate of four genes (CTAGE4, GAGE2E, GAGE2C, and HORMAD1) were higher in the resistant group than in the sensitive group, while the CNV rate in six genes (CDSN, DPCR1, MUC22, TMSB4Y, VARS, and HISTCH2AC) were lower in the resistant group than in the sensitive group. A combination of simultaneous mutation in two genes (SP8/HNRNPCL1 or SP8/FRG1) and deletion of GAGE2C together were found capable to predict dicycloplatin resistance with 100% sensitivity and 100% specificity.ConclusionWe successfully used cfDNA to monitor mutational profiles of PCa and designed an effective composite marker to select patients for dicycloplatin treatment based on their mutational profile.