Circulating microRNA analysis in a prospective co-clinical trial identifies MIR652-3p as a response biomarker and driver of regorafenib resistance mechanisms in colorectal cancer

Background: The multi-kinase inhibitor regorafenib has demonstrated efficacy in chemo-refractory metastatic colorectal cancer (mCRC) patients. However, lack of predictive biomarkers and concerns over significant toxicities hamper the use of regorafenib in clinical practice. Methods: Serial liquid biopsies were obtained at baseline and monthly until disease progression in chemo-refractory mCRC patients treated with regorafenib in a phase II clinical trial (PROSPECT-R n=40; NCT03010722) and in a multicentric validation cohort (n=241). Tissue biopsies collected at baseline, after 2 months and at progression in the PROSPECT-R trial were used to establish Patient-Derived Organoids (PDOs) and for molecular analyses. MicroRNA profiling was performed on baseline bloods using the NanoString nCounter platform and results were validated by digital droplet PCR and/or In Situ Hybridization in paired liquid and tissue biopsies. PDOs co-cultures and PDO-xenotransplants were generated for functional analyses. Results: Large-scale microRNA expression analysis in longitudinal matched liquid and tissue biopsies from the PROSPECT-R trial identified MIR652-3p as a biomarker of clinical benefit to regorafenib. These findings were confirmed in an independent validation cohort and in a "control" group of 100 patients treated with lonsurf. Using ex vivo co-culture assays paired with single-cell RNA-sequencing of PDO established pre- and post-treatment, we modelled regorafenib response observed in vivo and in patients, and showed that MIR652-3p controls resistance to regorafenib by impairing regorafenib-induced lethal autophagy and by orchestrating the switch from neo-angiogenesis to vessel co-option. Conclusions: Our results identify MIR652-3p as potential biomarker and as a driver of cell and non-cell autonomous mechanisms of resistance to regorafenib.