Project Success Stories - In silico medicine reaches the clinic
Forget the monopoly of in vitro, the study of biology in test tubes. The real cutting edge of medical science is in silico, and European researchers have just achieved a world first in the field.
In silico is the term scientists use to describe the modelling, simulation and visualisation of biological and medical processes in computers. The emergence of in silico medicine is a result of the advance of medical computer science over the last 20 years.
'In silico refers to any application of any computer-based technologies ― algorithms, systems and data mining or analysis,' according to Professor Norbert Graf, Director of the Paediatric Oncology and Haematology Clinic at Saarland University Hospital and senior researcher with the 'Advanced clinico-genomic trials on cancer' (ACGT) project.
Since the beginning of the great race to map the human genome, computer science has begun to play a much greater role in medical science. Called bioinformatics, this combination of computer science and statistics touches on almost every area of modern medical science and molecular biology: sequencing, gene annotation, evolutionary biology, mutation analysis, high throughput image analysis, and many others.
But one of the most exciting emerging bioinformatic disciplines is modelling, simulation and visualisation. Modelling maps the elements of a biological system, simulation attempts to realistically show how that system evolves over time under given stimuli, and visualisation presents the predictions in a graphic form.
It is an almost unimaginably impressive paradigm: real biological processes simulated accurately in a virtual environment. The field is still in its infancy but already scientists have made enormous progress, and nowhere more so than the EU-funded ACGT project.
ACGT sought to give the cancer research community a state-of-the-art ICT infrastructure so that it could use applied genomics in the clinic for the treatment of cancer. Applied genomics tailors treatment to the individual genetic profile of a particular tumour and patient, and ACGT provides a range of tools to support that.
The oncosimulator: cancer in silico
And ACGT's most innovative and advanced support tool is its oncosimulator, a piece of mathematical modelling, simulation and visualisation software and an in silico experimental platform.
The In Silico Oncology Group is developing this platform in collaboration with several research centres in Europe and Japan under the lead of Research Professor Georgios Stamatakos of the Institute of Communication and Computer Systems (ICCS) at the National Technical University of Athens (NTUA).
'The oncosimulator is an integrated software system simulating in vivo tumour response to therapeutics within a clinical trial environment,' explains Prof. Graf. 'It aims to support clinical decision-making for individual patients. Cancer treatment optimisation is the main goal of the system.'
These in silico experiments can help train and inform doctors, life scientists, researchers and patients by demonstrating the likely tumour response to different therapeutic regimes. The technology is not ready for the clinic just yet, but the ACGT project took a very big step in that direction.
In the ACGT project the team focused on paediatric nephroblastoma, a childhood cancer of the kidney, and in particular on a trial run by SIOP, the International Society of Paediatric Oncology.
Thanks to that trial, ACGT researchers were able to use anonymised real data before and after chemotherapeutic treatment, and that data provided a way to adapt the software to real clinical conditions and, at the same time, validate the software using real-world results.
'By using real medical data concerning nephroblastoma for a single patient in conjunction with plausible values for the model parameters … based on available literature, a reasonable prediction of the actual tumour volume shrinkage has been made possible,' says Prof. Graf.
The work on simulation included some of the most advanced mathematical medical science, such as stochastic cellular automata, discrete event simulation, hypermatrices and discrete operators.
Prof. Graf says that using these approaches it is possible to also study genetic instability, or mutation, and mutagenesis, as well as looking at the complexity of the interactions between the immune system and the tumour.
A detailed picture
ACGT followed the well-established top-down model to develop their simulation. The top-down approach uses clinical observations and what is known about the behaviour of the cancer. This method uses physiological and biological information to build up a very detailed picture of the cancer evolution, and an iterative process constantly updates both the simulation and the model underlying it.
The range of data used by ACGT's simulator is impressive. From the literature, the system factors in the pharmacokinetics of drugs, the dynamics of interaction between drugs and specific tumour types. It is also primed with radiobiological parameters for radiotherapy and molecular data. And it includes all clinical data like age, weight, family history and so on, and imaging data from computed tomography (CT), magnetic resonance imaging (MRI) and ultrasound, or any combination.
Molecular data comes from antibody profiling, an estimated cell-type composition of the tumour and estimates of the tumour's responsiveness to candidate drugs. All this information is combined with the details of the standard treatment protocols.
So far, so state-of-the-art, but the oncosimulator hopes to go beyond that over time.
'Obviously as more and more sets of medical data are exploited, the reliability of the model's "tuning" is expected to increase,' says Prof. Graf. 'The successful performance of the initial combined ACGT oncosimulator platform, although usable up to now only as a test of principle, has been a particularly encouraging step towards the clinical translation of the system, being the first of its kind worldwide.'
The team scored a real breakthrough by demonstrating with the SIOP data that the model was able to generally produce reasonable predictions. However, more work needs to be done. The oncosimulator must undergo an exhaustive validation, adaptation and optimisation process before it can enter routine clinical practice as a decision-making tool.
Moreover, the researchers need to test and integrate molecular extraction methods of the crucial histological, or cell, constitution of the tumour. That work is underway, but ACGT's breakthrough is to show that the proof of principle is sound.