Circular eccDNA shows potential to predict early-stage LUAD recurrence

LUAD is the most common histological subtype of lung cancer. Although surgery can be curative for patients diagnosed at an early stage, approximately 30%-50% develop recurrence or aggressive metastasis within five years. More effective molecular markers are therefore needed to improve postoperative risk assessment and follow-up. eccDNA is a class of circular double-stranded DNA molecules that exists independently of chromosomes and can be detected in both tumor tissue and plasma. Its involvement in tumor evolution, heterogeneity, metastasis, and treatment resistance has attracted growing interest, but the characteristics of eccDNA associated with early-stage LUAD recurrence remain incompletely understood.

Published in Precision Clinical Medicine on 6 July 2026 (DOI: 10.1093/pcmedi/pbag017), the study was conducted by researchers from Southwest Jiaotong University, The Third People's Hospital of Chengdu, Chengdu University of Traditional Chinese Medicine, and Deyang People's Hospital. The study enrolled 90 treatment-naïve patients with early-stage LUAD and collected tumor tissue, matched adjacent non-tumor tissue, and plasma samples. Circle-seq and public multi-omics datasets were subsequently integrated to investigate the genomic distribution, transcriptional activity, chromatin characteristics, and potential clinical relevance of eccDNA.

The results showed that eccDNAs from recurrent tumors had higher GC content and stronger split-read signals than those from non-recurrent tumors. They were also preferentially derived from transcriptionally active genomic regions, including exons, CpG islands, and chromatin A compartments, and were associated with active histone modifications. Recurrence-associated eccDNAs carried several genes involved in cancer-related pathways, including mTOR, Notch, and Ras signaling. Differences were also observed in the expression of eccDNA-associated immune factors and their receptors, suggesting a possible relationship between eccDNA and altered immune signaling in recurrent LUAD.

In plasma, the researchers identified 2,387 eccDNAs that were commonly upregulated in lung cancer and recurrence-associated samples. Genes linked to these plasma eccDNA signals were integrated with public transcriptomic and survival data, leading to the selection of seven genes—AFAP1L2, LHX8, IL20RB, SLC12A8, EGLN3, CDH3, and PLTP—for construction of a recurrence risk model. The model consistently distinguished patients with different DFS(disease-free survival) outcomes in both training and validation cohorts. Together, these findings provide a multi-omics view of recurrence-associated eccDNA in LUAD and support its potential as a source of prognostic biomarkers. Further validation in larger prospective cohorts may facilitate the development of eccDNA-based approaches for individualized postoperative risk assessment and monitoring.