Personalized cell-free DNA fragmentation dynamics in early-stage NSCLC in TRACERx

Background: Plasma cell-free DNA (cfDNA) fragmentation reflects underlying chromatin organization and can be disrupted in cancer. Circulating tumor DNA (ctDNA) fragments differ from non-tumor cfDNA in size, genomic distribution, and end-motif characteristics. Recent studies have shown that fragmentomic profiles can distinguish individuals with and without lung cancer, and fragment sizes from targeted sequencing can be leveraged for cancer detection. In the TRACERx study, personalized ctDNA sequencing has previously been used to monitor individuals with stage I-III non-small cell lung cancer (NSCLC) who underwent curative-intent surgery, enabling prognostication at baseline and post-operative landmark timepoints. Here, we investigate the dynamics of ctDNA and cfDNA fragmentation in response to surgery, relapse and systemic therapies. Methods: We characterised cell-free DNA (cfDNA) fragmentation in 1,069 longitudinal plasma samples from 197 individuals using Anchored multiplex PCR and targeted sequencing. We developed a computational pipeline to extract fragment size distributions and end-motif features from tumor-derived and non-tumor cfDNA fragments. Results: Distinct cfDNA fragmentation dynamics were observed across clinical timepoints, and mutant fragments were shorter than non-mutant fragments. Treatment effects on fragmentation were observed: a transient release of non-tumor cfDNA was observed post-surgery, plus systemic therapies were associated with alterations in fragmentation pattern. Associations between fragmentation metrics, clinicopathologic features, and biological variables such as extrachromosomal DNA, were explored. Conclusion: Targeted plasma sequencing within TRACERx enables integrated evaluation of tumor- and non-tumor-derived cfDNA. Fragmentation dynamics reflect treatment and disease status, suggesting their potential as a complementary non-invasive marker for monitoring and mechanistic insights in early-stage NSCLC.