New models for preclinical evaluation of drug efficacy in common solid tumours
Ongoing | IMI1 |
Prediction of cognitive properties of new drug candidates For neurodegenerative


Start Date
End Date
IMI1 - Call 2
Grant agreement number

IMI Funding
8 756 641
EFPIA in kind
9 661 201
2 602 918
Total Cost21 020 760


The PREDECT project will permit the emergence of faithful models for target validation and beyond. Traditional preclinical discovery methods, particularly for target validation, poorly predict drug efficacy, causing a high attrition rate in pharmaceutical research and development.

The PREDECT consortium will focus on complex but transferable next generation in vitro and in vivo models for breast, prostate and lung cancers. Models will be investigated for their improved potential to validate novel therapeutic targets. Known targets, in canonical pathways, will be interrogated for induction of phenotypic, proteomic and transcriptomic changes using inhibitors.

A strategy of seeking a ‘dynamic reciprocity’ of concordance between the steady and perturbed states of in vitro complex cultures, tissue slices and in vivo tumour models will be pursued by systems biology analyses. The project will develop and generate a repository of advanced complex models in 3 complementary areas:

  • in vitro 2D/3D organotypic (co-)cultures, stirred bioreactor aggregates and tissue slice systems
  • novel (orthotopic) grafts of human and mouse tumour samples
  • genetically-engineered and mosaic mouse models.

PREDECT aims to produce results which will shift paradigms in target validation and so leading to greater predictivity of drug efficacy in drug trials. PREDECT is coordinated by Servier and AstraZeneca and the managing entity of IMI JU funding is the University of Helsinki.

The team assembles world-class biologists, clinicians and computational scientists from 8 EU institutes, 3 SMEs and 7 EFPIA members who will work to develop and then critically assess new models for target validation.

Achievements & News

PREDECT’s new tumour models could lead to more effective treatments
December 2017

Laboratory tests are not sufficiently predictive of cancer drugs’ activity in humans. The rate of failure to take drugs from laboratory to registration for clinical use has been estimated to be as high as 90 %. In clinical trials, patients are thus exposed to drugs that don’t work and the cost to industry is enormous and unsustainable.### Real human cancers are complex: genetic changes within tumours over time cause cells in one cancer to differ from each other. To this complexity is added the cancer’s interaction with normal cells. IMI’s PREDECT project tackled the problem of complexity by creating laboratory models of the disease that better represent the complex characteristics of different cancers. More appropriate laboratory models for preclinical studies should increase the ability to predict drug efficacy. This will expand the number of treatments available for different types of cancer, potentially providing patients with a higher chance of survival.

‘We chose to focus on complex and sometimes difficult models of cancer,’ explains project coordinator John Hickman of the Servier research institute in France. ‘As an academic-industrial consortium we believed that although our models may not be rapid, nevertheless they better represent the complexity of cancer. These models provide a high content of information more relevant to human cancer. With the high rate of failure of current high-throughput and rapid models to predict clinical activity, we considered something had to change in the way industry looks for drugs.’

Capturing cancer’s complexity – PREDECT models show the way
July 2016

During the earlier stages of cancer drug development, researchers study cancer cells in the laboratory, for example in petri dishes. These two-dimensional models of cancer are relatively cheap and easy to use, so they are still widely used in research, yet they do not accurately replicate real tumours in the body.### This hinders the ability of researchers to study cancer in detail and develop new treatments. IMI’s PREDECT project has developed and analysed a number of more complex, three-dimensional models of prostate, breast and lung cancer that may more accurately mimic the behaviour of tumours in the body. According to Boehringer Ingelheim’s Ralph Graeser, there is no single perfect model that fits all purposes. To help scientists pick the right model for the right situation, the project team has written a paper in the journal Scientific Reports. The article sets out the strengths and weaknesses of the different models and provides detailed protocols for their use as well as advice on when and how to use them. ‘Although people have called for an end to use of 2D cultures, these models are still predominantly used, even though they poorly represent human tumours,’ explained Dr Graeser. ‘The robust protocols for the set-up and analysis of 3D cultures, as well as the cross-comparison of the platforms presented in this paper, should help scientists both in academia and industry to better incorporate these complex models in the drug discovery pipeline.’

PREDECT project delivers ‘potential game-changer’ for breast cancer research
April 2016

Scientists from IMI’s PREDECT project have developed the first animal model of a common form of breast cancer that faithfully replicates the human disease, opening up new avenues for studying breast cancer and developing and evaluating treatments. The findings are published in the prestigious journal Cancer Cell.### Some three quarters of breast cancer cases feature tumours that have a receptor for the hormone oestrogen. Efforts to study these ER+ cancers have been hampered by the fact that animals (like mice) used to study the disease do not accurately replicate how the disease behaves in human patients. Now, PREDECT researchers have created a mouse model that mimics human ER+ cancer better than any other existing model. ‘With this breakthrough, breast cancer disease, progression and metastasis now become amenable to study,’ said George Sflomos of EPFL, the first author of the paper. ‘We can now study crucial factors, such as the action of hormones and molecular responses to therapies, for the first time in a relevant context. But more importantly, this model opens up new opportunities not only for the development but also for the evaluation of breast cancer therapies.’ Meanwhile an accompanying editorial states: ‘This finding is a potential-game changer for breast cancer research, and we predict that it will likely translate into new therapeutic strategies for ER+ breast cancer in the near future’.

PREDECT takes lab-based cancer models to the next level
January 2016

IMI’s PREDECT project is making progress on its goal of improving lab-based cancer models to make them better, more reliable tools for use in medical research and drug development. Studying cancer in the laboratory is challenging; tumour samples in petri dishes simply do not behave in the same way as tumours in the body.### PREDECT is working to improve these over-simplistic models by designing complex three-dimensional models that also include cells from the body’s connective tissues which interact with the tumour. Analyses of the models reveal that these models’ behaviour is much closer to that of cancer in the body. Scientists in the project are now using these models in their research. Another group of researchers in the project is looking at improving ways of studying slices of tumours in the laboratory. Tumour slices provide a lot of information on the architecture and make-up of complex tumours. However, the act of creating a slice and keeping it alive can affect the cells’ behaviour. Scientists from PREDECT have found ways to get round some of these issues to create samples that give more reliable results. The models are all the result of collaboration between scientists from small companies, universities, and large pharmaceutical companies. Details of the models are published in open access papers. The new models offer a number of advantages. Most notably, their complexity means scientists will be able to study tumours in the lab in unprecedented detail, and this will reduce their reliance on animal models. ‘There is a growing wave of concern that an over-dependence on reductionist models is one factor contributing to limited success in drug discovery in all therapeutic areas,’ commented project coordinator John Hickman of Servier. ‘To go to more complex systems that model pathology usually means using more animals. This is also expensive. We are attempting some compromise solutions: complex in vitro models.’

Participants Show participants on map

EFPIA companies
  • AbbVie Inc., North Chicago, Illinois, United States
  • AstraZeneca AB, Södertälje, Sweden
  • Bayer AG, Berlin, Germany
  • Boehringer Ingelheim International GmbH, Ingelheim, Germany
  • Charles River research services Germany GmbH, Freiburg, Germany
  • F. Hoffmann-La Roche Ltd, Basel, Switzerland
  • Institut De Recherches Servier S.A.S, Suresnes, France
  • Janssen Pharmaceutica NV, Beerse, Belgium
  • Orion Corporation, Espoo, Finland
Universities, research organisations, public bodies, non-profit groups
  • Cardiff University, Cardiff, United Kingdom
  • Ecole Polytechnique Federale De Lausanne, Lausanne, Switzerland
  • Erasmus Universitair Medisch Centrum Rotterdam, Rotterdam, Netherlands
  • Robert Bosch Gesellschaft fuer medizinische Forschung mbH, Stuttgart, Germany
  • Stichting Katholieke Universiteit / Radboud university medical center, Nijmegen, Netherlands
  • Tartu Ülikool, Tartu, Estonia
  • University of Helsinki, University of Helsinki, Helsinki, Finland
  • Weizmann Institute Of Science, Rehovot, Israel
Small and medium-sized enterprises (SMEs)
  • Biomedicum Genomics Ltd , Helsinki, Finland
  • Instituto de Biologia Experimental e Tecnológica , Oeiras, Portugal


Project coordinator
Institut de Recherche Servier
+33 643928191
Managing entity
Emmy Verschuren
University of Helsinki
+358 50 415 4703