Cell-free DNA fragmentomes for treatment response monitoring in patients with metastatic colorectal cancer: The Dolphin study

Introduction: Accurate monitoring of tumor response in patients with metastatic colorectal cancer (mCRC) undergoing systemic therapy is essential to enable timely and informed treatment decisions. Although CT imaging is currently the standard-modality, it has limitations including limited accuracy for detecting small lesions, as well as inter-reader variability. A reliable blood-based biomarker may enable a more precise and dynamic assessment of treatment response. Circulating tumor DNA (ctDNA) is indicative of cancer cell burden and may complement CT imaging for evaluating treatment responses. DELFI tumor fraction (DELFI-TF) is a mutation- and tumor-independent ctDNA assay which has demonstrated potential for longitudinal monitoring of treatment response. The DOLPHIN study aims to evaluate the clinical added value of DELFI-TF to conventional imaging-based response monitoring in patients with mCRC. Method: DOLPHIN is a prospective, observational substudy of the Prospective Dutch Colorectal Cancer Cohort, in which blood samples are collected longitudinally in conjunction with routine CT imaging in mCRC patients. ctDNA levels are longitudinally assessed using the DELFI-TF ctDNA assay, a locked and validated machine learning model that quantifies tumor burden using cell-free DNA (cfDNA) fragmentomic data derived from low-coverage whole genome sequencing. Simultaneously, ddPCR analysis is performed on samples from patients with a tumor tissue-confirmed RAS or BRAF mutation. The primary endpoint of this study is the association between changes in ctDNA levels and clinically determined treatment response. Secondary endpoints include the correlation between ctDNA dynamics and clinical, biochemical, and radiological responses at multiple timepoints during systemic treatment (I), lead time of ctDNA testing compared with CT imaging for identifying progressive disease (II), prognostic value of longitudinal ctDNA testing (III) and the cost-effectiveness of various monitoring approaches (IV). Results: Between March 16 2023 and October 16 2025, 504 patients were enrolled before starting the second line of treatment. Nine patients were excluded due to incorrect entry criteria. A comprehensive overview of patient demographics and clinical characteristics will be presented at the conference. In total 2214 blood samples and 2468 CT scans have been collected, with collection planned to continue until March 2026. To date, 2186 samples (86.8%) have been successfully matched to a corresponding CT scan, and 1881 (86.0%) of these were obtained within 14 days of imaging. Conclusion: In the ongoing DOLPHIN study, we successfully aligned a majority of blood samples with corresponding CT scans. This study will determine whether DELFI-TF can complement treatment response monitoring in mCRC patients and potentially enable parts of follow-up to be shifted to the home setting.