The tissue and circulating cell-free DNA-derived genetic landscape of pre-malignant colorectal lesions and its application for early diagnosis of colorectal cancer

This study is led by Prof. Luo Huiyan and Prof. Xu Ruihua (Department of Medical Oncology, Sun Yat Sen University Cancer Center). Colorectal adenomas (CRAs) represent precancerous lesions that precede the development of colorectal cancer (CRC). Regular monitoring of CRAs can hinder the progression into carcinoma.

To explore the utility of tissue DNA and circulating cell-free DNA (cfDNA) in early diagnosis of CRC, authors retrospectively sequenced paired tissue and plasma samples from 85 patients with conventional CRAs. The genetic alterations identified in these samples were compared with those from 78 stage-I CRC patients (CRC-I) in the ChangKang project. 

The findings reveal that a small proportion of samples (5/85, 5.9%) successfully identified tissue-origin variants within the matched cfDNA. Moreover, authors explored potential biological factors affecting the detectability of cfDNA mutations in CRA. They observed that the genomic burden of CNVs in both tissue and plasma samples intensifies with malignancy, suggesting that CNV features could offer valuable insights for the early detection applications of cfDNA.

Additionally, this study delves into the genomic characteristics and the progression of cancer progression from CRA to CRC-I. Within the CRA cohort, authors pinpointed 12 genes, notably APC, KRAS and SOX9, that exhibited significant mutated rates in tissue. Patients harboring mutations in KMT2C and KMT2D displayed persistent adenoma. By comparing with the mutational profiles of metastatic CRC plasma samples, authors found that ZNF717 was exclusively mutated in CRAs, while KMT2C and KMT2D mutations were detected in both CRA and CRC. The presence of cfDNA mutations in plasma was validated through polymerase chain reaction, enhancing the feasibility of using cfDNA mutations for early CRC screening. Compared to CRC-I, CRAs exhibited a reduced frequency of TP53 and PIK3CA somatic mutations and underwent non-neutral evolution more often.

Furthermore, authors established a random forest model based on 15 characteristic genes to distinguish CRA and CRC, achieving an area under the curve of 0.89. Through this endeavor, they identified two novel genes, CNTNAP5 and GATA6, implicated in CRC carcinogenesis.