E!114899: Automated high-throughput oncology drug screening in zebrafish using robotics and AI

Cancer is the most burdensome disease in the world with an estimated 18.1 million new cases and 9.6 million deaths in 2018. Drug candidates are assessed in living organisms quite late in the preclinical drug development process. Consequently, only 9% of cancer drugs entering phase 1 clinical trials complete phase 3 successfully. Furthermore, an average of 10,000 compounds need to be screened for each approved drug. Therefore, high-throughput screening methods that provide biologically relevant data are needed to 1) rapidly assess the large number of compounds generated in drug discovery programs and 2) to increase the number of drugs that make it successfully through clinical trials. The goal of this project was to develop robots that can aid cancer drug discovery screens using the zebrafish as a model organism. Using robotic and AI technology, the ROBO-FISH consortium aimed to significantly increase throughput for cancer drug screens. We achieved this through optimizing speed and efficiency for tumor cell transplantations and monitoring of tumor growth and survival using robotic microinjection and imaging in zebrafish fry (larvae). Our objective was to deliver an injection robot and two types of imaging robots (high/low resolution) to create a fast, accurate, easy to use, high-throughput workflow that could revolutionize cancer drug screening. Zebrafish xenograft (cell transplantation) models have been increasingly recognized for their ability to predict patient responses to cancer therapeutics, suggesting their potential as diagnostic tools in clinical settings. However, these models require the precise microinjection of cancer cell suspensions in many small and fragile zebrafish fry (larvae). Manual injections are so challenging that, even after months of training, variability in experimental results persists among researchers. This limits the uptake and deployment of zebrafish xenograft models for clinical use and drug discovery. To address this challenge, the consortium partners designed, built, and validated automated microinjection and imaging robots. For the injection robot, combined results of injections from the project consortium partners into the vasculature, perivitelline space, and hindbrain demonstrated an average injection success rate of approximately 60%, with a larval survival rate exceeding 70%, comparable to manual injections using a traditional micromanipulator. Notably, the fully automated mode of the robot was twice as fast as manual injections. This automation of the microinjection process significantly reduces the need for extensive personnel training while it enhances reproducibility, efficiency, and accuracy, paving the way for more extensive use of zebrafish xenograft models in drug discovery and patient diagnostics. We are now testing and optimizing the imaging robots, so that this can be coupled to the injection robot to create a single robot workflow. Finally, we have shown proof of principle and demonstrated for the first time, successful xenografting of patient-derived paediatric glioma cells into developing zebrafish larvae. These new fish "avatars" for gliomas, are now being used for the systematic testing of single and combination drug treatments to identify potential new therapies for a highly aggressive pediatric brain tumor type that currently results in very poor prognosis. This work has now forged new collaborations with international expert laboratories that combine the advantages of the zebrafish model organism with human computational biology and mammalian tumor models.

The successful generation of tumour cell transplantation and high-resolution confocal microscopy robots within the ROBOFISH project, as well as the implementation of image recognition machine learning models, resulted in a new international collaboration. The aim of the collaboration is to establish human-derived xenograft transplantation zebrafish models (avatars) for highly aggressive paediatric brain tumours and use these avatars for combinatorial cancer drug discovery. The initial findings of this collaboration has stimulated interest from other research groups to apply the same strategy towards other tumour types. We are also in discussion with radiology experts to use the zebrafish avatars to expand combination treatment options for cancer patients. The societal impact of the tumour transplantation and imaging robots built and optimised in the ROBOFISH project will aid in expediting both oncology drug discovery and basic medical research, towards a deeper understanding of the causes of cancer. Furthermore, the same robots can also be used for other types of cell transplantation studies, pathogen challenge studies, drug discovery, and toxicological testing of drugs and environmental pollutants - thus expanding the repertoire of applications and possibilities for national and international research collaborations.

The project goal is to develop robots for oncology drug screening in zebrafish. Cancer is the most burdensome disease in the world with an estimated 18.1 million new cases and 9.6 million deaths in 2018. Drug candidates are assessed in living organisms late in the preclinical development process. Consequently, only 9% of oncology drugs entering phase 1 clinical trials, complete phase 3 successfully. Furthermore, 10,000 compounds are screened for each approved drug. High-throughput screening that provides biologically relevant data is needed to 1) rapidly assess the large number of compounds generated in drug discovery and 2) to increase the number of drugs that make it through clinical trials. Zebrafish injected with patient-derived tumour cells (PDX) mimic in vivo responses and could be used in the early drug screening workflow. Their small size, easy and inexpensive maintenance and abundant availability of genetic tools make zebrafish highly suitable for oncology drug screening. Consortium partners and zebrafish CRO market leaders ZeClinics (ZC) and BioReperia (BR) are experiencing strong market demand for zebrafish PDX drug screening, but are critically impeded by their current low throughput, manual, labour intensive workflow. Through deep learning image recognition, advanced robotics and automated imaging, Life Science Methods (LSM) and Confocal.nl will develop the ROBO-FISH robots that can inject 10 times faster than trained scientists and can provide high-throughput image analysis for zebrafish. The Centre for Molecular Medicine Norway (NCMM) and CROs ZC and BR will validate the robots in their expert zebrafish labs to generate feedback on performance, applicability and reproducibility. The ROBO-FISH consortium will deliver an injection robot and two types of imaging robots (high/low resolution) to create a fast, accurate, easy to use, high-throughput workflow that could revolutionize oncology drug screening.