Costa, BrunaEstrada, Marta F.Gomes, AntónioFernandez, Laura M.Azevedo, José M.Póvoa, VandaFontes, MárciaAlves, AntónioGalzerano, AntónioCastillo-Martin, MireiaHerrando, IgnacioBrandão, ShermannCarneiro, CarlaNunes, VítorCarvalho, CarlosParvaiz, AmjadMarreiros, AnaFior, Rita2024-10-032024-10-032024-06-052041-1723http://hdl.handle.net/10400.1/25991Cancer patients often undergo rounds of trial-and-error to find the most effective treatment because there is no test in the clinical practice for predicting therapy response. Here, we conduct a clinical study to validate the zebrafish patient-derived xenograft model (zAvatar) as a fast predictive platform for personalized treatment in colorectal cancer. zAvatars are generated with patient tumor cells, treated exactly with the same therapy as their corresponding patient and analyzed at single-cell resolution. By individually comparing the clinical responses of 55 patients with their zAvatar-test, we develop a decision tree model integrating tumor stage, zAvatar-apoptosis, and zAvatar-metastatic potential. This model accurately forecasts patient progression with 91% accuracy. Importantly, patients with a sensitive zAvatar-test exhibit longer progression-free survival compared to those with a resistant test. We propose the zAvatar-test as a rapid approach to guide clinical decisions, optimizing treatment options and improving the survival of cancer patients.engjournal article10.1038/s41467-024-49051-0