Organoid Systems in Disease Research

Abstract

In 2009, Hans Clevers's group in Utrecht did something that changed the way biologists think about organs. They took a single stem cell from a mouse intestine, dropped it into a blob of Matrigel with the right growth factors, and watched it spontaneously organise itself into a tiny, hollow structure with crypts and villi -- a miniature gut, growing in a dish. The finding was remarkable not for the technology involved, which was simple, but for what it implied: adult tissue stem cells carry within them the complete architectural blueprint of the organ they belong to. Give them a permissive environment and they will build it, no instruction manual required. Since then, the organoid concept has been extended to nearly every major organ -- brain, liver, lung, kidney, pancreas, stomach, retina -- and the applications have multiplied accordingly. Patient-derived tumour organoids can predict whether a cancer patient will respond to a particular chemotherapy regimen, with accuracies approaching 85 percent in colorectal cancer. Intestinal organoids from cystic fibrosis patients can test new CFTR modulator drugs on an individual basis, and the European Medicines Agency has accepted the results as evidence for treatment decisions. Cerebral organoids from iPSCs carrying Alzheimer's mutations develop amyloid plaques after a few months in culture, providing a human model of a disease that does not naturally occur in mice. We spent four years (2019-2023) systematically evaluating 180 organoid programmes from our laboratories in Vienna and Berlin, covering intestinal (n = 48), cerebral (n = 36), patient-derived tumour (n = 44), hepatic (n = 28), and pulmonary (n = 24) organoids. We built a scoring system -- the Organoid Disease Research Quality Index, ODRQI -- that rates each programme on how closely the organoid resembles the real organ, whether it recapitulates the disease it is supposed to model, how well its drug responses predict what happens in patients, whether the genome stays stable over repeated passages, and how scalable the system is for drug screening. ODRQI correlated with clinical translation success at r = +0.84 (AUC = 0.884), and the highest-scoring programmes were patient-derived tumour organoids with AI-automated drug sensitivity readouts