Researchers have developed a platform of pan-cancer tumour organoids and a neural-network-based drug assay, that has the potential to advance precision medicine research in oncology.
Barriers to precision medicine
Precision medicine has recently emerged as a highly effective and durable approach for treating disease. It involves the identification and administration of therapies that target the unique biology of a patient’s disease. For cancer patients, targeted therapies have been developed against several molecular alterations present within tumours, such as oncogenic mutations. However, patients respond variably to these targeted treatments and there are no known therapies for many other alterations. Overall, poor knowledge surrounding cellular and genetic heterogeneity in cancer has limited progress in precision medicine.
Recently, patient-derived tumour organoid technology, or cellular models of diverse cancer types, has advanced oncological research. It has also facilitated therapeutic development and precision medicine studies. However, there has been a lack of scalable and reproducible methods to develop and profile tumour organoids. Furthermore, therapeutic profiling assays across these models have not been generalisable across different tumour types.
Pan-cancer tumour organoids
Recently, researchers from Tempus Labs in Chicago established a platform of pan-cancer tumour organoids for use in precision medicine. The platform can be applied to high-throughput translational research, such as biobanking, molecular profiling and drug screening. Their work has been published in Cell Reports.
The researchers generated tumour organoids by culturing metastatic tumours derived from 1,298 North American patients with major types of carcinomas. They further determined minimal culture conditions for initiating and propagating each type of tumour organoid. Except for certain gastric and pancreatic cancers, the growth factors Noggin and epidermal growth factor were sufficient to successfully culture organoids.
Critically, these cultured tumour organoids accurately represented the molecular features of patient tumours. DNA sequencing and whole-transcriptome analysis showed that organoids recapitulated the genomic and transcriptomic signatures of their corresponding source tumours. This was verified with an independent cohort of representative cancer patients.
Predicting patient-specific drug response
To use these organoids for drug development and precision medicine, the researchers developed a label-free, high-throughput organoid drug screening assay. Cell viability assays are generally used to measure a tumour’s response to drugs. However, variability in organoid generation has meant that technical replicates display high levels of variation.
Instead, the proposed assay monitors drug response by observing cells in real time under light microscopy. This approach is highly reproducible, minimises tumour processing and enables for high-throughput applications. From here, the researchers developed a neural network-based model to predict drug response from light microscopy.
The network revealed differential responses to clinically relevant drugs within tumour types, reflecting intratumoural heterogeneity. Variability in drug response may reveal mechanisms of therapeutic resistance, and thus provide predictive and prognostic biomarkers for individual response to treatment. Moreover, such temporal profiling of drug response may also allow for the quantification of transient responses, optimisation of appropriate time points for endpoint studies and characterisation of adaptive responses.
Implications
These findings hold the potential to greatly advance the utility of tumour organoids. The pan-cancer platform uses optimised, defined minimal media to successfully initiate and propagate tumour organoids. Importantly, the organoids retain genomic and transcriptomic fidelity to source tumours. Future organoid studies may thus be conducted with less complex media formulations, thereby reducing time, costs and user errors.
Meanwhile, the neural network-based drug assay enables high-throughput in vitro profiling of potential therapeutics. Importantly, it is capable of predicting patient-specific variability in drug responses across tumour types.
The clinical applicability of these approaches still requires validation. Nevertheless, they represent potentially useful analytical tools for studying tumour organoid development, drug screening and monitoring patient response to therapy. Ultimately, this work brings the field of oncology one step closer towards achieving the goal of precision medicine.
Image credit: Harvey Parvin – canva
