Background: Anti-EGFR antibody is recommended as one of the standard treatments in patients with RAS wild-type advanced colorectal cancer (CRC). Few reports have evaluated low-frequency subclones of alterations in EGFR signaling pathway genes in plasma cell-free DNA (cfDNA) during anti-EGFR treatment in prospective cohort of advanced CRC patients. Methods: Key inclusion criteria of this study was as follows: 20 years old and over, performance status 0 to 2, metastatic or recurrent CRC (adenocarcinoma), RAS wild-type CRC diagnosed by PCR-rSSO using tumor tissue, no prior use of anti-EGFR antibody. Plasma samples were prospectively collected at pre-treatment, at 4, 10 weeks, every 10 weeks thereafter, and at the end of the anti-EGFR treatment. With QX200 Droplet Digital PCR platform, we analyzed hotspot mutations of KRAS, NRAS, BRAF, and PIK3CA and copy number (CN) variant of HER2 and MET using plasma samples of pre- and post-treatment as well as during treatment (limited to patients with any gene alterations at pre- and post-treatment). Results: Total 50 patients were registered in this study, but 3 patients were excluded because inclusion criteria were not met. Finally, 47 patients were included in full analysis set. In 15 patients (31.9%), genetic alterations including KRAS (G12G13: 10.1%, A59Q61: 2.2%), NRAS (G12G13: 2.2%, Q61: 2.1%), BRAF (V600: 6.4%), PIK3CA (E545: 4.3%, H1047: 4.3%), and HER2 (amplification: 6.5%) were detected in pre-treatment samples. Analysis of pre-treatment samples showed that response rates of patients with any gene alterations (N = 13) and that without any gene alterations (N = 29) were 30.8% and 58.6%, respectively (p = 0.095). When the cut-off of variant allele frequency (VAF) was set as 0.5% in patients with any gene mutations, the response rates for cases with VAF ≥0.5% (N = 6) and those with VAF < 0.5% (N = 7) were 0% and 57.1%, respectively (p = 0.049). Among 41 patients with detectable target lesions by CT scan, there was significant difference in percentage of best tumor shrinkage between patients with gene alterations and those without gene alterations (32.1% vs 10.5%, p = 0.037), when cut-off VAF of any gene mutations was set as 0.5% and cut-off CN of HER2 or MET was set as 4 at pre-treatment. Due to dynamics of each gene alteration in cfDNA during treatment, mutations of KRAS A59Q61, NRAS Q61, and MET amplification were characterized as acquired alterations which were newly detected after the administration of anti-EGFR antibodies. Conclusions: Our study demonstrated the clinical relevance of minor mutant subclones in EGFR signaling pathway genes in plasma for predicting response to anti-EGFR treatment. Although sample size of this study was small, preferable threshold for distinguishing responders from non-responders may be 0.5% as VAF of these gene mutations. Clinical trial information: 000034923.