Uncategorized
Rewriting the Oncology Playbook
Dr Mark Eccleston, Director of Inaphaea, discusses how patient-derived 3D models can bridge the translational divide in oncology.
Drug development is a complex, costly, and time-consuming process with a high failure rate. Effective cancer therapy development in particular, remains challenging with only 3.4% of drugs entering clinical trials between 2000 and 2015 reaching approval (1,2). Improved human relevant in-vitro and in silico models, applied in the earliest stages of drug development, should improve translational success of candidates as well as reduce unnecessary animal use whilst accelerating the development of novel therapeutics. This in turn would translate to reduced late-stage failures which can cost upwards of £1 billion per failed programme (3,4).
Recent regulatory changes in the UK and US have paved the way for New Approach Methodologies (NAMs), which utilise innovative strategies to enhance efficiency, reduce costs and improve patient outcomes. The FDA Modernisation Act 2.0 (2021-2022), set out a framework to remove a legal requirement for animal testing and using drug development with the groundbreaking announcement in April 2025 that animal testing would be replaced with more effective, human relevant models. Hot on the heels of this announcement, Qureator recently received an historic first FDA IND approval of an anti PD-1/L1 combination therapeutic based purely on a human relevant, vascularised organoid model which has been cited as a transformative moment for the drug development field (5).
The UK government followed suit in November 2025, publishing its strategy to replace animals in science and support development, validation and uptake of alternative technologies (6). This has generated substantial additional impetus for the development of advanced in silico and AI‑-enabled modelling platforms in parallel with organoid and other patient‑derived 3D culture systems. Converging these modalities is expected to enhance predictive fidelity across both domains, as increasingly patient‑–‑relevant multi-omics, high content‑ imaging, and quantitative phenotypic readouts from complex in vitro and ex‑ vivo systems provide richer, mechanistically anchored datasets for training, validating, and refining AI‑based computational models. This is particularly pertinent to the emerging patient ‑avatar paradigm, in which clinical datasets, molecular biomarker profiles, and therapeutic response information can be systematically digitised to generate computationally tractable patient representations. Preclinical systems – including conventional 2D cell lines, 3D spheroids, multicellular patient-derived cultures, and organoid platforms—can likewise be digitised, enabling high‑dimensional multi-omics, imaging and quantitative phenotypic outputs from these models to be integrated into avatar frameworks for in‑silico prediction of clinical outcomes. The resulting responder and nonresponder biomarker signatures can then be operationalised for patient stratification in real‑world clinical trials, increasing trial efficiency and success rates within a precision‑ medicine framework and potentially supporting rational indication‑ expansion.
New Approach Methodologies, including in-silico, ex-vivo and in-vitro models, once validated can be applied throughout the therapeutic development pipeline but the capabilities of the platforms must be matched to the stage of development and in particular, throughput requirements. In the earlier stages of hit identification, in-silico and high throughput in-vitro models are required and as the programme passes through to hit to lead, candidate selection and then preclinical validation models can become increasingly complex – 2D to 3D, single cell to multi-cellular, co culture with immune components for efficacy with organ on a chip and even whole, ex-vivo persufflated organ approaches for safety and toxicity screening.
The UK landscape review published in February 2026, identified shortcomings in the readiness of the UK’s academic ecosystem for human relevant clinical models in terms of translational readiness required for widespread industry adoption. These were, unsurprisingly, around robustness, validation, standardisation and scalability with recommendations to increase investment, standardisation and biobanking capacity. A plan was put in place to establish a pre-clinical translational models hub to assemble modelling capabilities and data generation with an unprecedented opportunity for industry/academic collaboration to close the translational gap (7). Investment through UK (Innovate) and EU (EIC) (8) grant funding initiatives to support NAMs development as well as deployment into clinical trials (9) is starting to come through to support industry involvement.
Commercial development of NAMs is well underway with a number of providers for organoids, spheroids and various platforms to support functionality including perfused chip formats, novel photolithographic approaches to incorporating vascularisation and media to support differentiated growth. Commercial development can bring in additional resources and frameworks for standardisation with companies used to providing services under ISO or GLP frameworks. Development of human tissue derived NAMs involves ethically sensitive activities, primarily the use of human-derived primary cells and tissues and the processing of anonymised human-associated data. All activities must comply fully with applicable ethical, legal, and regulatory frameworks, including EU Directive 2004/23/EC, EU Regulation 2016/679 (GDPR), Directive 2010/63/EU (3Rs, where applicable), and relevant UK legislation and will also requires appropriate commercial use of consent from the donors.
Patient-derived cell (PDC) models cultured from patient biopsies offer a useful platform for high throughput preclinical screening. They retain many tumour characteristics, including mixed stromal cell components and better represent patient variability including genetic mutations. Biopsy material is typically dissociated and passaged meaning it is no longer classified under The Human Tissue Act 2004 allowing storage and distribution. The cellular components can be characterised by flow cytometry to identify subtype and relative composition of cancer cells and stromal cells, including Cancer Associated Fibroblasts (CAFs). These cellular components can be separated by FACs or bead-based approaches and expanded, typically to low passage to retain their patient characteristics, and then recombined in defined ratios to generate reproducible models. These models therefore provide significantly more realistic conditions for testing effectiveness of drugs but are limited by availability and throughput relative to simple 2D cell lines (10). However, immortalised cell lines, even as 3D monoculture spheroid systems do not accurately predict in vivo drug performance, and whilst PDCs are more resource-intensive with lower-throughput they represent a more predictive, human-relevant in vitro systems which can be deployed during early preclinical stages compared to more complex organoid platforms. This approach is essential to improving translational success, reducing progression of non-viable candidates and reducing animal testing in oncology drug development.
The inclusion of CAFs within the co-culture supports association of the cells, 3D growth and development of tumour microenvironment (TME) characteristics including cell-cell communication. CAFs play a critical role in tumour growth and influence cancer cell proliferation and their survival. CAFs can limit drug penetration or modulate drug response by activating protective signalling pathways in cancer cells and represent legitimate therapeutic targets. The PDC models can be supplemented by co-culture with immune cells (heterotypic PBMCs or patient matched) bringing in an additional immune oncology dimension which is completely missing in standard xenograft approaches in immune compromised mouse models. Syngeneic and humanised mouse models can be employed but do not accurately recapitulate human clinical immune responses (11).
Patient-derived 3D cell models can therefore address a critical gap in preclinical modelling by introducing patient-relevant TME variability into a reproducible platform. They represent a shift from conventional 2D cell line screens, which are high throughput but oversimplified, toward models that more accurately predict in vivo drug efficacy, penetration, and combination therapy responses. Key characteristics include:
- Patient-specific variability enabling more relevant earlier assessment of drug responses.
- Bridging the preclinical gap between simple cell lines and expensive organoids platforms.
- Reduction and refinement of animal testing, by improving the predictive value of early in vitro screening and enabling better-informed progression decisions, thereby decreasing the number of compounds advanced unnecessarily into in vivo studies.
- Facilitation of complex therapy development, including multi-drug and combination treatments, by better recapitulating the TME earlier in the preclinical model whilst still enabling high throughput testing.
Drug development is approaching a pivotal shift: human relevant in vitro systems, ex vivo platforms, and AI driven in silico models are converging into a unified, predictive paradigm that promises faster, safer, and more precise therapeutic innovation. As regulatory momentum accelerates and these technologies mature, oncology R&D is poised to move beyond high-risk empirical testing toward data rich, patient aligned decision-making. The real impact will be felt where it matters most – delivering better therapies to patients sooner, with higher confidence and fewer failures. The organisations that embrace this transformation now will shape the future of translational medicine and redefine what is possible for patient benefit.
About the author
Dr Mark Eccleston is polymer chemist and biotechnology entrepreneur with over 30 years experience working in translation science in both drug and biomarker development. Eccleston is a former BBSRC Enterprise fellow and holds an MBA (Entrepreneurship).
References
1: https://www.nature.com/articles/nrd3078
3: https://cancertransfection.com/2d-and-3d-cell-line-models/?utm
4: https://pubmed.ncbi.nlm.nih.gov/40567279/
5: https://www.pcrm.org/news/innovative-science/fda-approves-new-cancer-drug-clinical-trials-based-nonanimal-data-only
8: https://ec.europa.eu/info/funding-tenders/opportunities/portal/screen/opportunities/topic-details/HORIZON-EIC-2026-AIC-02?order=DESC&pageNumber=1&pageSize=50&sortBy=startDate&isExactMatch=true&status=31094501,31094502,31094503&callIdentifier=HORIZON-EIC-2026-AIC
9: https://ec.europa.eu/info/funding-tenders/opportunities/portal/screen/opportunities/topic-details/HORIZON-HLTH-2026-01-TOOL-03?isExactMatch=true&status=31094501,31094502,31094503&callIdentifier=HORIZON-HLTH-2026-01&order=DESC&pageNumber=1&pageSize=50&sortBy=startDate
10: https://pmc.ncbi.nlm.nih.gov/articles/PMC9105149/
11: https://doi.org/10.1158/2326-6066.cir-24-1046
From DDW Volume 27 – Issue 2, Spring 2026 – Read the digital issue here
The post Rewriting the Oncology Playbook appeared first on Drug Discovery World (DDW).