Find the right PDX model for your research

There is a desperate need to improve the success of drug development in the field of oncology. The rate at which preclinical agents fail to demonstrate sufficient safety or efficiency in oncology clinical trials has been pegged between 85% 1 to 95% 2. Furthermore, assuming a drug doesn't fail, it can still cost more than $400M to get it through to Phase III trails 3. Therefore, there is a need to reduce these costs and attrition rates.

Chemotherapy Vials, image from the National Cancer Institute

One of the major reasons given for this high failure rate is the lack of models that are capable of recapitulating the heterogeneity and physiological conditions found in cancer patients 4. Our ability to translate findings from cancer cell lines grown in culture is restricted by the variable results gained from these models.

PDX models to the rescue

Patient-derived xenograft (PDX) models, however, offer the potential to transform translational oncology 5. PDX models are created by transplanting fresh tumour tissues from human patients directly into mice.

How PDX models are made and characterised

PDX tumours have been shown to preserve key features of a specific cancer and these models have been shown to be predictive of clinical outcomes – unlike cell culture models. Furthermore, PDX models have been used to predict biomarkers of drug susceptibility and drug resistance, which is crucial for clinical trial phases of development where multiple drugs fail. They are becoming the preferred preclinical tool to improve the drug development process.

But it's not that simple

Every PDX model - like every cancer - has a unique genetic make-up. Therefore, it can be very hard for researchers to find the right PDX model for their specific research question. Researchers across many labs, including the PDX providers, are using a variety of techniques to map the genomic characteristics of each model, including genotyping, gene expression and whole-genome sequencing. However, getting access to this data, which is crucial in deciding which PDX model to use, is not easy.

"The current process for finding the right PDX model is like finding a needle in a haystack".

It is a repetitive labour-intensive process, where researchers must go to each PDX vendor in turn to see if a relevant PDX model exists. This can involve at best using multiple different query interfaces, at worst having to generate new genotyping data from PDX tissue samples. Internal projects to characterise and catalogue the available PDX models are often duplicating efforts and may be incomplete or out-of-date.

Cancer researcher, University of Cambridge:

"At Cambridge we really are lucky, we have an encyclopaedia 6 of breast cancer PDX models - but this took 7 years to build! For other researchers, searching for the right PDX model is a complicated and time-consuming process. Mostly you are just contacting other researchers, hoping someone has what you are looking for".

What is Repositive doing?

In collaboration with AstraZeneca, Repositive is launching a PDX consortium to develop a collaborative, pre-competitive PDX resource. The objective of the project is to provide a resource for streamlined discovery and queries on molecular data from PDX models across multiple sources to suit specific needs in oncology research. Read more about the Repositive PDX consortium.

The Repositive PDX resource will be a catalogue of genomic data resources for PDX data, allowing researchers to directly find data for their PDX model of interest. See what PDX data we currently have on our public Repositive platform.

By maximising availability and reuse of existing PDX data, the oncology research units will shorten their time for data access and speed up their decision process for conducting PDX experiments.

The Repositive PDX resource workflow

PDX models currently on Repositive

Horizon Discovery

Horizon Discovery has an extensive, well-characterised collection of PDX models (Browse). For breast cancer models, they have the WHIM (Washington University Human-in-Mouse) collection, and for melanoma models they have partnered with The Wistar Institute of Anatomy and Biology in Philadelphia to offer a subset of their collection. Horizon Discovery offers their collections as ungrafted PDX suspensions or as comprehensive in vivo efficacy studies.

Jackson Labs Mouse Tumor Biology Database

The MTB database (Browse) was initiated by The Jackson Laboratory in 1997, with funding from the National Cancer Institute (USA) 7. It aims to provide web-based access to data on the pathobiology of cancer in genetically defined strains of laboratory mice. The MTB database does not exclusively host PDX model data, however it does contain data from over 400 PDX models. All of which are listed on the Repositive platform - Browse MTB JAX models.


In collaboration with AstraZeneca, Repositive is launching a PDX consortium to develop a collaborative, pre-competitive PDX resource with easy queries for assessing molecular data from PDX models across multiple data sources. Read more about the Repositive PDX consortium.


  1. High-throughput screening using patient-derived tumor xenografts to predict clinical trial drug response. Hui Gao et al., Nature Medicine. October 2015 | doi:10.1038/nm.3954



  4. Patient-derived tumour xenografts as models for oncology drug development. John J. Tentler et al., Nat Rev Clin Oncol. 2012 | doi:10.1038/nrclinonc.2012.6

  5. High-throughput screening using patient-derived tumor xenografts to predict clinical trial drug response. Hui Gao et al., Nature Medicine 21, 1318–1325 (2015) | doi:10.1038/nm.3954

  6. A Biobank of Breast Cancer Explants with Preserved Intra-tumor Heterogeneity to Screen Anticancer Compounds. Alejandra Bruna., et al. Cell 167, 260–274. September 22, 2016 |

  7. The Mouse Tumor Biology database. Debra M. Krupke et al., Nat Rev Cancer. 2008 June | doi:10.1038/nrc2390

Related Blog Posts

Repositive launches consortium with AstraZeneca to improve oncology research

Read more posts by Charlotte Whicher