Résumé
Background
Liquid biopsy holds great promise for transforming patient care, yet current biomarkers such as circulating tumor DNA and circulating tumor cells lack sensitivity in early-stage disease. This highlights the need for novel biomarkers capable of detecting subtle molecular alterations preceding overt tumor development. In this context, epitranscriptomics, the study of RNA chemical modifications, emerges as a new frontier. Over 100 RNA modifications have been identified in mammals, forming a dynamic regulatory layer that links environmental cues to gene expression, disease onset and progression. Notably, modified nucleosides resulting from RNA degradation cannot be recycled and are excreted into biofluids, making them attractive non-invasive biomarkers.
Methods
We developed an integrated platform combining targeted mass spectrometry and machine learning to sensitively quantify and profile circulating modified nucleosides from minimal plasma volumes (<10 μL), using tailored internal standards for robust normalization.
Results
Using this platform, we identified a minimal circulating epitranscriptomic signature composed of five nucleosides for colorectal cancer (CRC), based on a training cohort (n = 130 CRC patients, stages I–IV; n = 30 healthy donors). A supervised machine learning model demonstrated that this signature discriminates CRC samples with high accuracy, achieving a cross-validation F1-score ≥ 0.95, including in early-stage (stage I) disease. Validation in an independent cohort (n = 47 CRC patients, stages I–IV; n = 20 healthy donors) confirmed the robustness of this signature, with precision, recall, and F1-score all ≥ 0.95, reinforcing its strong potential for early CRC detection. Notably, this signature-based discrimination was lost following tumor resection (n = 20 patients with resectable stage I–II CRC).
Conclusions
Collectively, our findings establish circulating RNA modification profiling as a powerful and non-invasive approach for early gastrointestinal cancer detection and management. This work paves the way for a new generation of biomarkers and has the potential to significantly advance precision oncology by enabling earlier diagnosis and improved patient stratification.