I am dedicated to the founding, shaping and advancement of the emergent clinical trial driven scientific and technological discipline of in silico oncology and more broadly in silico medicine. The proposed notion and system of "Oncosimulator," aiming at individualized treatment optimization via mathematics and multiscale modelling, is the focus of my group's research. Many different cancer types and treatment modalities are addressed. Research is done on an outstanding EU-US clinical network.
Skills and Expertise
Research Professor, ICCS-NTUA and Visiting Professor, School of Electrical and Computer Engineering, NTUA
National Technical University of Athens · Institute of Communication and Computer Systems (ICCS)
CHIC: Computational Horizons in Cancer - Developing Meta- and Hyper-Multiscale Models and Repositories for In Silico Oncology
Institute of Communications and Computer Systems - National Technical University of Athens
Zográfos, Attica, Greece
Georgios Stamatakos is the Scientific and the Overall Coordinator of of the large scale EU-US integrating research project CHIC on in silico oncology (http://www.chic-vph.eu/ ) which is mainly funded by the European Commission with 10,582,000 €.
Mar 2013 - Feb 2016
MyHealthAvatar - "A Demonstration of 4D Digital Avatar Infrastructure for Access of Complete Patient Information "
Institute of Communications and Computer Systems - National Technical University of Athens
Zográfos, Attica, Greece
Georgios Stamatakos is the Leader of WP5 (Models & Repositories) and WP 10 (Dissemination and Exploitation) of the EC funded project (Grant Agreement: 600929). http://www.myhealthavatar.eu/
Sep 1991 - Feb 1997
National Technical University of Athens
Physics (Theoretical Biophysics)
Oct 1987 - Sep 1988
University of Strathclyde
Oct 1981 - Mar 1987
National Technical University of Athens
Electrical (and Computer) Engineering
The EU-US large scale integrating project CHIC ( http://www.chic-vph.eu/ ) proposes the development of clinical trial driven tools, services and secure infrastructure that will support the creation of multiscale cancer hyper-models (integrative models). The latter are defined as choreographies of component models, each one describing a biological process at a characteristic spatiotemporal scale, and of relation models/metamodels defining the relations across scales. Integrative models can become component models for other integrative models. The development of a secure hypermodelling infrastructure consisting primarily of a hypermodelling editor and a hypermodelling execution environment is a central generic VPH geared objective of CHIC. In order to render models developed by different modellers semantically interoperable, an infrastructure for semantic metadata management along with tools and services for ontology-based annotations will be developed. Existing approaches such as the one developed by the EC funded RICORDO project will be exploited and extended. Facilitated operations will range from automated dataset matching to model merging and managing complex simulation workflows. In this way standardization of cancer model and data annotation allowing multiscale hypermodelling will be fostered. The following entities will also be developed: a hypermodel repository, a hypermodel-driven clinical data repository, a distributed metadata repository and an in silico trial repository for the storage of executed simulation scenarios, an image processing toolkit, a visualization toolkit and cloud and virtualization services. In order to ensure that the entire project will be clinically driven and clinically oriented, three concrete clinical trials/studies will be adopted and addressed. They concern nephroblastoma treated by combined chemotherapy, glioblastoma treated by immunotherapy in combination with chemotherapy and radiotherapy and non-small cell lung cancer treated by a combination of chemotherapy and radiotherapy. The multiscale data generated by these trials/studies will be exploited so as to both drive the development of a number of integrative multiscale cancer models (hypermodels) and hypermodel oncosimulators and clinically adapt and partly validate them.
Research Items (164)
Efficient use of Virtual Physiological Human (VPH)-type models for personalized treatment response prediction purposes requires a precise model parameterization. In the case where the available personalized data are not sufficient to fully determine the parameter values, an appropriate prediction task may be followed. This study, a hybrid combination of computational optimization and machine learning methods with an already developed mechanistic model called the acute lymphoblastic leukaemia (ALL) Oncosimulator which simulates ALL progression and treatment response is presented. These methods are used in order for the parameters of the model to be estimated for retrospective cases and to be predicted for prospective ones. The parameter value prediction is based on a regression model trained on retrospective cases. The proposed Hybrid ALL Oncosimulator system has been evaluated when predicting the pre-phase treatment outcome in ALL. This has been correctly achieved for a significant percentage of patient cases tested (approx. 70% of patients). Moreover, the system is capable of denying the classification of cases for which the results are not trustworthy enough. In that case, potentially misleading predictions for a number of patients are avoided, while the classification accuracy for the remaining patient cases further increases. The results obtained are particularly encouraging regarding the soundness of the proposed methodologies and their relevance to the process of achieving clinical applicability of the proposed Hybrid ALL Oncosimulator system and VPH models in general.
DC-vaccination integrated in multimodal temozolomid-based treatment for patients with GBM prolong OS in a fraction of patients. Biomarkers predicting outcome after active specific immunotherapy, however, fail. Blood samples from 134 adults, randomized for DC-vaccination during or after TMZ maintenance (TMZm), were taken before and after radiochemotherapy. Data from circulating lymphocytes were determined by FACS in 101 patients. To quantify immune profiles more systematically, nonlinear features of these data were introduced, corresponding to all meaningful ratios between immune cell subpopulations. To quantify immune profile changes during radiochemotherapy for each such quantity, the ratio of its values after/before radiochemotherapy was also calculated. Canonical correlation analysis was performed exhaustively between all possible combinations of these quantities taken 1, 2 or 3 at a time versus OS. Median OS of the total group of patients was 18 months, 2-year OS was 31.3% (+4). There was no difference in OS for patients treated with immunotherapy during versus after TMZm. Patients with postoperative residual tumor volume had a significantly worse median OS (16 versus 20 months). For the entire population, no strong correlations were detected between the immune profiles and OS. However, stratifying the patients into subgroups defined by extent of resection (residual tumor volume RTV = 0 or >0) and vaccination schedule (during or after TMZm), and repeating this analysis for each of the 4 subgroups, revealed for all subgroups strong correlations between FACS data immune profiles derived from the pre- and post-radiochemotherapy blood samples and ultimate OS of the patients. These data suggest strong influences from the immune status at start of treatment on the OS outcome of the patients. The data depict the first in silico oncology model using immune profiles to potentially predict outcome of clinical risk profile-stratified patients treated with standard treatment and immunotherapy (www.chic-vph.eu).
Glioblastoma (GB) implies a devastating prognosis with an average survival of 14–16 months using the current standard of care treatment . GB is the most frequent malignant tumour originating from the brain parenchyma, and it is characterised by a marked intratumoural heterogeneity, proneness to infiltrate throughout the brain parenchyma, robust angiogenesis and necrosis as well as intense resistance to apoptosis and genomic instability.
Glioblastoma remains a clinical challenge in spite of years of extensive research. Novel approaches are needed in order to integrate the existing knowledge. This is the potential role of mathematical oncology. This paper reviews mathematical models on glioblastoma from the clinical doctor’s point of view, with focus on 3D modeling approaches of radiation response of in vivo glioblastomas based on contemporary imaging techniques. As these models aim to provide a clinically useful tool in the era of personalized medicine, the integration of the latest advances in molecular and imaging science and in clinical practice by the in silico models is crucial for their clinical relevance. Our aim is to indicate areas of GBM research that have not yet been addressed by in silico models and to point out evidence that has come up from in silico experiments, which may be worth considering in the clinic. This review examines how close these models have come in predicting the outcome of treatment protocols and in shaping the future of radiotherapy treatments.
A novel explicit triscale reaction-diffusion numerical model of glioblastoma multiforme tumor growth is presented. The model incorporates the handling of Neumann boundary conditions imposed by the cranium and takes into account both the inhomogeneous nature of human brain and the complexity of the skull geometry. The finite-difference time-domain method is adopted. To demonstrate the workflow of a possible clinical validation procedure, a clinical case/scenario is addressed. A good agreement of the in silico calculated value of the doubling time (ie, the time for tumor volume to double) with the value of the same quantity based on tomographic imaging data has been observed. A theoretical exploration suggests that a rough but still quite informative value of the doubling time may be calculated based on a homogeneous brain model. The model could serve as the main component of a continuous mathematics-based glioblastoma oncosimulator aiming at supporting the clinician in the optimal patient-individualized design of treatment using the patient’s multiscale data and experimenting in silico (ie, on the computer).
Background Antiangiogenic agents have been recently added to the oncological armamentarium with bevacizumab probably being the most popular representative in current clinical practice. The elucidation of the mode of action of these agents is a prerequisite for personalized prediction of antiangiogenic treatment response and selection of patients who may benefit from this kind of therapy. To this end, having used as a basis a preexisting continuous vascular tumour growth model which addresses the targeted nature of antiangiogenic treatment, we present a paper characterized by the following three features. First, the integration of a two-compartmental bevacizumab specific pharmacokinetic module into the core of the aforementioned preexisting model. Second, its mathematical modification in order to reproduce the asymptotic behaviour of tumour volume in the theoretical case of a total destruction of tumour neovasculature. Third, the exploitation of a range of published animal datasets pertaining to antitumour efficacy of bevacizumab on various tumour types (breast, lung, head and neck, colon). Results Results for both the unperturbed growth and the treatment module reveal qualitative similarities with experimental observations establishing the biologically acceptable behaviour of the model. The dynamics of the untreated tumour has been studied via a parameter analysis, revealing the role of each relevant input parameter to tumour evolution. The combined effect of endogenous proangiogenic and antiangiogenic factors on the angiogenic potential of a tumour is also studied, in order to capture the dynamics of molecular competition between the two key-players of tumoural angiogenesis. The adopted methodology also allows accounting for the newly recognized direct antitumour effect of the specific agent. Conclusions Interesting observations have been made, suggesting a potential size-dependent tumour response to different treatment modalities and determining the relative timing of cytotoxic versus antiangiogenic agents administration. Insight into the comparative effectiveness of different antiangiogenic treatment strategies is revealed. The results of a series of in vivo experiments in mice bearing diverse types of tumours (breast, lung, head and neck, colon) and treated with bevacizumab are successfully reproduced, supporting thus the validity of the underlying model. Reviewers This article was reviewed by L. Hanin, T. Radivoyevitch and L. Edler.
The plethora of available disease prediction models and the ongoing process of their application into clinical practice – following their clinical validation – have created new needs regarding their efficient handling and exploitation. Consolidation of software implementations, descriptive information, and supportive tools in a single place, offering persistent storage as well as proper management of execution results, is a priority, especially with respect to the needs of large healthcare providers. At the same time, modelers should be able to access these storage facilities under special rights, in order to upgrade and maintain their work. In addition, the end users should be provided with all the necessary interfaces for model execution and effortless result retrieval. We therefore propose a software infrastructure, based on a tool, model and data repository that handles the storage of models and pertinent execution-related data, along with functionalities for execution management, communication with third-party applications, user-friendly interfaces to access and use the infrastructure with minimal effort and basic security features.
During the last decades, medical observations and multiscale data concerning tumor growth are mounting. At the same time, contemporary imaging techniques well established in clinical practice, provide a variety of information on real-time, in-vivo tumor growth. Mathematical and in-silico modeling has been widely recruited to provide means for further understanding of pertinent biological phenomena. However, despite the vast amounts of new evidence compiled by medical doctors, there are still many aspects of tumor growth that remain largely unknown. There is still a large variety of mechanisms to be better understood and therefore, many hypotheses to be tested. To approach this problem, starting from mathematical elaborations, we have developed a model of the early phases of tumor growth consisting of several algorithmic modules, each one corresponding to a particular biological mechanism. The modularity of the model allows keeping track of the assumptions made in each step and facilitates re-adjustment, in case new hypotheses need to be considered. Simulations showed good qualitative agreement with biological observations, and revealed a non-trivial interplay between between oxygen requirements of cancer cells and their maximum mitosis rates. The proposed model has, at least in principle, the potential to exploit data from contemporary imaging techniques and is eligible for utilizing multicore computation.
The large scale integrating Euro-American research project CHIC (“Computational Horizons in Cancer: Developing Meta- and Hyper-Multiscale Models and Respositories for In Silico Oncology”)  which is funded by the European Commission aims at developing an efficient and robust scientific and technological framework able to support the development of complex cancer hypermodels. The latter are driven by and oriented towards real clinical needs [2-6]. Hypermodels may be developed by linking and integrating component models (hypomoderls), eventually created by different modelling teams (multimodeller hypermodels). The reason behind this approach is that integrating the expertise and models of different cancer modelling research teams across the globe could greatly accelerate the process of exploiting the accumulated knowledge and the ever increasing number of multiscale data sets pertaining to cancer for the benefit of the patient. The hypercomplexity of cancer itself naturally calls for such an approach. The long term goal of the endeavour is to provide clinicians with the possibility to experiment in silico (i.e. on the computer) in order to select the optimal treatment for a given patient based on the patient’s own multiscale data such as clinical, imaging, histological and molecular. A sine qua non condition for the clinical translation of this process is a strict clinical adaptation and validation - both retrospective and prospective - of the hypermodels. In this context excellent progress has been achieved so far regarding all aspects of the CHIC project including the hypermodel development, its clinical relevance and the initial steps of clinical adaptation and validation of hypermodels. The aim of this paper is to briefly outline selected indicative facets of the project through specific paradigms.
The 5-year survival of non-small cell lung cancer patients can be as low as 1% in advanced stages. For patients with resectable disease, the successful choice of preoperative chemotherapy is critical to eliminate micrometastasis and improve operability. In silico experimentations can suggest the optimal treatment protocol for each patient based on their own multiscale data. A determinant for reliable predictions is the a priori estimation of the drugs' cytotoxic efficacy on cancer cells for a given treatment. In the present work a mechanistic model of cancer response to treatment is applied for the estimation of a plausible value range of the cell killing efficacy of various cisplatin-based doublet regimens. Among others, the model incorporates the cancer related mechanism of uncontrolled proliferation, population heterogeneity, hypoxia and treatment resistance. The methodology is based on the provision of tumor volumetric data at two time points, before and after or during treatment. It takes into account the effect of tumor microenvironment and cell repopulation on treatment outcome. A thorough sensitivity analysis based on one-factor-at-a-time and latin hypercube sampling/partial rank correlation coefficient approaches has established the volume growth rate and the growth fraction at diagnosis as key features for more accurate estimates. The methodology is applied on the retrospective data of thirteen patients with non-small cell lung cancer who received cisplatin in combination with gemcitabine, vinorelbine or docetaxel in the neoadjuvant context. The selection of model input values has been guided by a comprehensive literature survey on cancer-specific proliferation kinetics. The latin hypercube sampling has been recruited to compensate for patient-specific uncertainties. Concluding, the present work provides a quantitative framework for the estimation of the in-vivo cell-killing ability of various chemotherapies. Correlation studies of such estimates with the molecular profile of patients could serve as a basis for reliable personalized predictions.
Background: The adoption in oncology of Clinical Decision Support (CDS) may help clinical users to efficiently deal with the high complexity of the domain, lead to improved patient outcomes, and reduce the current knowledge gap between clinical research and practice. While significant effort has been invested in the implementation of CDS, the uptake in the clinic has been limited. The barriers to adoption have been extensively discussed in the literature. In oncology, current CDS solutions are not able to support the complex decisions required for stratification and personalized treatment of patients and to keep up with the high rate of change in therapeutic options and knowledge. Results: To address these challenges, we propose a framework enabling efficient implementation of meaningful CDS that incorporates a large variety of clinical knowledge models to bring to the clinic comprehensive solutions leveraging the latest domain knowledge. We use both literature-based models and models built within the p-medicine project using the rich datasets from clinical trials and care provided by the clinical partners. The framework is open to the biomedical community, enabling reuse of deployed models by third-party CDS implementations and supporting collaboration among modelers, CDS implementers, biomedical researchers and clinicians. To increase adoption and cope with the complexity of patient management in oncology, we also support and leverage the clinical processes adhered to by healthcare organizations. We design an architecture that extends the CDS framework with workflow functionality. The clinical models are embedded in the workflow models and executed at the right time, when and where the recommendations are needed in the clinical process. Conclusions: In this paper we present our CDS framework developed in p-medicine and the CDS implementation leveraging the framework. To support complex decisions, the framework relies on clinical models that encapsulate relevant clinical knowledge. Next to assisting the decisions, this solution supports by default (through modeling and implementation of workflows) the decision processes as well and exploits the knowledge embedded in those processes.
As in many cancer types, the G1/S restriction point (RP) is deregulated in Acute Lymphoblastic Leukemia (ALL). Hyper-phosphorylated retinoblastoma protein (hyper-pRb) is found in high levels in ALL cells. Nevertheless, the ALL lymphocyte proliferation rate for the average patient is surprisingly low compared to its normal counterpart of the same maturation level. Additionally, as stated in literature, ALL cells possibly reside at or beyond the RP which is located in the late-G1 phase. This state may favor their differentiation resistant phenotype. A major phenomenon contributing to this fact is thought to be the observed limited redundancy in the phosphorylation of retinoblastoma protein (pRb) by the various Cyclin Dependent Kinases (Cdks). The latter may result in partial loss of pRb functions despite hyper-phosphorylation. To test this hypothesis, an in silico model aiming at simulating the biochemical regulation of the RP in ALL is introduced. By exploiting experimental findings derived from leukemic cells and following a semi-quantitative calibration procedure, the model has been shown to satisfactorily reproduce such a behavior for the RP pathway. At the same time, the calibrated model has been proved to be in agreement with the observed variation in the ALL cell cycle duration. The proposed model aims to contribute to a better understanding of the complex phenomena governing the leukemic cell cycle. At the same time it constitutes a significant first step in the creation of a personalized proliferation rate predictor that can be used in the context of multiscale cancer modeling. Such an approach is expected to play an important role in the refinement and the advancement of mechanistic modeling of ALL in the context of the emergent and promising scientific domains of In Silico Oncology and more generally In Silico Medicine.
Intensive glioma tumor infitration into the surrounding normal brain tissues is one of the most critical causes of glioma treatment failure.To quantitatively understand and mathematically simulate this phenomenon, several diffsion-based mathematical models have appeared in the literature.The majority of them ignore the anisotropic character of diffsion of glioma cells since availability of pertinent truly exploitable tomographic imaging data is limited. Aiming at enriching the anisotropy-enhanced glioma model weaponry so as to increase the potential of exploiting available tomographic imaging data, we propose a Brownian motion-based mathematical analysis that could serve as the basis for a simulation model estimating the infitration of glioblastoma cells into the surrounding brain tissue. Th analysis is based on clinical observations and exploits diffsion tensor imaging (DTI) data. Numerical simulations and suggestions for further elaboration are provided.
Over the previous years, semantic metadata have largely contributed to the management, exchange and querying of health-related data, including mathematical and computational disease simulation model descriptions, implementations and output results. In this paper, we present a proposal for an abstract semantic metadata infrastructure layout, indicating its modularity, and thus its capability to operate with different combinations of software tools. Its potential contribution for the purposes of the CHIC project is also reported.
The goal of this article is to present basic scientific principles and core algorithms of the simulation module of the CERvical cancer ONCOsimulator (CERONCO) developed within the context of the DrTherapat project (FP7-ICT-600852). CERONCO simulates the response of cervical tumours to radiotherapy treatment (external beam radiotherapy followed by brachytherapy) with concomitant weekly cisplatin, in the patient-individualized context. Results from a preliminary clinical adaptation study based on the MR imaging data of a clinical case are presented as well.
A couple of multiscale spatiotemporal simulation models of glioblastoma multiforme (GBM) growth and invasion into the surrounding normal brain tissue is presented. Both models are based on a continuous and subsequently finite mathematical approach centered around the non-linear partial differential equation of diffusion-reaction referring to glioma tumour cells. A novel explicit, strict and thorough numerical treatment of the three dimensional adiabatic Neumann boundary conditions imposed by the skull is also included in both models. The first model assumes a homogeneous representation of normal brain tissue whereas the second one, assuming an inhomogeneous representation of normal brain tissue, distinguishes between white matter, grey matter and cerebrospinal fluid. The predictions of the tumour doubling time by both models are compared for specific data sets. Clinical observational data regarding the range of the GBM doubling time values are utilized in order to ensure the realism of both models and their predictions. We assume that the inhomogeneous normal brain tissue representation is a virtual rendering of reality more credible than its homogeneous counterpart. The simulation results for the cases considered show that using the homogeneous normal brain based model may lead to an error of up to 10% for the first 25 simulated days in relation to the predictions of the inhomogeneous model. However, the error drops to less than 7% afterwards. This observation suggests that even by using a homogeneous brain based model and a realistic weighted average value of its diffusion coefficient, a rough but still informative estimate of the expected tumour doubling time can be achieved. Additional in silico experimentation aiming at statistically testing and eventually further supporting the validity of this hypothesis is in progress. It is noted that the values of the diffusion coefficients and the cell birth and death rates of the model are amenable to refinement and personalization by exploiting the histological and molecular profile of the patient. Work on this aspect is in progress.
The aim of this paper is to present the development of a multi-scale and multiphysics approach to tumor growth. An existing biomodel used for clinical tumor growth and response to treatment has been coupled with a biomechanical model. The macroscopic mechanical model is used to provide directions of least pressure in the tissue, which drives the geometrical evolution of the tumor predicted at the cellular level. The combined model has been applied to the case of brain and lung tumors. Results indicated that the coupled approach provides additional morphological information on the realistic tumor shape when the tumor is located in regions of tissue inhomogeneity. The approach might be used in oncosimulators for tumor types where the morphometry information plays a major role in the treatment and surgical planning.
Significant Virtual Physiological Human (VPH) efforts and projects have been concerned with cancer modeling, especially in the European Commission Seventh Framework research programme, with the ambitious goal to approach personalized cancer simulation based on patient-specific data and thereby optimize therapy decisions in the clinical setting. However, building realistic in silico predictive models targeting the clinical practice requires interactive, synergetic approaches to integrate the currently fragmented efforts emanating from the systems biology and computational oncology communities all around the globe. To further this goal we propose an intelligent graphical workflow planning system that exploits the multiscale and modular nature of cancer and allows building complex cancer models by intuitively linking/interchanging highly specialized models. The system adopts and extends current standardization efforts, key tools and infrastructure in view of building a pool of reliable and reproducible models capable of improving current therapies and demonstrating the potential for clinical translation of these technologies.
This paper outlines the major components and function of the Technologically Integrated Oncosimulator developed primarily within the ACGT (Advancing Clinico Genomic Trials on Cancer) project. The Oncosimulator is defined as an information technology system simulating in vivo tumor response to therapeutic modalities within the clinical trial context. Chemotherapy in the neoadjuvant setting, according to two real clinical trials concerning nephroblastoma and breast cancer, has been considered. The spatiotemporal simulation module embedded in the Oncosimulator is based on the multiscale, predominantly top-down, discrete entity - discrete event cancer simulation technique developed by the In Silico Oncology Group, National Technical University of Athens. The technology modules include multiscale data handling, image processing, invocation of code execution via a spreadsheet-inspired environment portal, execution of the code on the grid and visualization of the predictions. A refining scenario for the eventual coupling of the Oncosimulator with immunological models is also presented. Parameter values have been adapted to multiscale clinical trial data in a consistent way, thus supporting the predictive potential of the Oncosimulator. Indicative results demonstrating various aspects of the clinical adaptation and validation process are presented. Completion of these processes is expected to pave the way for the clinical translation of the system.
This paper presents a brief outline of the notion and the system of oncosimulator in conjunction with a high level description of the basics of its core multiscale model simulating clinical tumor response to treatment. The exemplary case of lung cancer preoperatively treated with a combination of chemotherapeutic agents is considered. The core oncosimulator model is based on a primarily top-down, discrete entity - discrete event multiscale simulation approach. The critical process of clinical adaptation of the model by exploiting sets of multiscale data originating from clinical studies/trials is also outlined. Concrete clinical adaptation results are presented. The adaptation process also conveys important aspects of the planned clinical validation procedure since the same type of multiscale data - although not the same data itself- is to be used for clinical validation. By having exploited actual clinical data in conjunction with plausible literature-based values of certain model parameters, a realistic tumor dynamics behavior has been demonstrated. The latter supports the potential of the specific oncosimulator to serve as a personalized treatment optimizer following an eventually successful completion of the clinical adaptation and validation process.
This paper discusses the need for interconnecting computational cancer models from different sources and scales within clinically relevant scenarios to increase the accuracy of the models and speed up their clinical adaptation, validation, and eventual translation. We briefly review current interoperability efforts drawing upon our experiences with the development of in silico models for predictive oncology within a number of European Commission Virtual Physiological Human initiative projects on cancer. A clinically relevant scenario, addressing brain tumor modeling that illustrates the need for coupling models from different sources and levels of complexity, is described. General approaches to enabling interoperability using XML-based markup languages for biological modeling are reviewed, concluding with a discussion on efforts towards developing cancer-specific XML markup to couple multiple component models for predictive in silico oncology.
This short VIDEO demonstrates a four dimensional, discrete mathematics based, top-down simulation of tumor response to two radiation therapy scenarios according to the predictions of an in silico model described mainly in the following two publications: 1.G S Stamatakos, V P Antipas, N K Uzunoglu, R G Dale, “A four-dimensional computer simulation model of the in vivo response to radiotherapy of glioblastoma multiforme: studies on the effect of clonogenic cell density.” The British Journal of Radiology. 06/2006; 79(941):389-400. and 2. Georgios S Stamatakos, Dimitra D Dionysiou, “Introduction of hypermatrix and operator notation into a discrete mathematics simulation model of malignant tumour response to therapeutic schemes in vivo. Some operator properties.” Cancer Informatics. 01/2009; 7:239-51. It is noted that the version of the model used in the particular oncosimulator is an early one which, however,is still sufficient for the needs of a gross visual demonstration of the fundamental macroscopic behavior of the model. Crucial extensions taking inter alia into account the different levels of the mitotic potential of progenitor tumour cells have been developed more recently . The use of pseudocolors is explained within the video. In order to clearly demonstrate the capability of the model to simulate shrinkage and repopulation (regrowth) during the intervals between treatment sessions and especially during weekends, we had to proceed to the selection of exaggerated values of certain model parameters and “squeeze” the simulated period of weeks into seconds of visualization. However, models should first demonstrate their capability to behave qualitatively or semi-quantitatively as expected and at a subsequent stage they may readily be tuned to reality by carefully selecting certain parameter values during the clinical adaptation process. The video was created by G. Stamatakos (Institute of Communication and Computer Systems - National Technical University of Athens ( ICCS-NTUA), D.Dionysiou (ICCS-NTUA) and R. Belleman (University of Amsterdam) in 2010 within the framework of the large scale integrated research project ACGT (FP6-2005-IST-026996) funded jointly by the European Commission and the Japanese State. It was uploaded on ResearchGate on 20 Jan. 2013.
Improving the initial diagnosis and the assessment of response to treatment in malignant gliomas, while avoiding invasive methods as much as justifiable, is one major aspect actual research is focusing on. Imaging studies are used to calculate tumor volume and define vital, necrotic and cystic areas within a tumor. Though the visual interpretation of magnetic resonance (MR) images is based on qualitative observation of variation in signal intensity, a correlation of signal intensities with histological features of a tumor is not possible. Better methods are needed for a reliable interpretation of follow-up studies in single patients. Histograms of signal intensities might serve as a method adding quantitative data to the description of a tumor. Using DoctorEye software, tumors can be easily rendered and histograms of the signal intensities within a tumor as well as mean and median signal intensities are possible to calculate. Our results in glioblastoma suggest that these histograms are an innovative method of gaining new tumor-specific information without performing additional investigations in a patient. It can be an additional diagnostic tool in differentiating various intracranial lesions from each other, as well as in assessing response to treatment or progression of malignant glioma.
I. INTRODUCTION Cancer is a natural phenomenon and consequently is amenable to mathematical and computational description. Clinically driven complex multi-scale cancer models are capable of producing realistic spatio-temporal and patient-specific simulations of commonly-used clinical interventions such as radio-chemotherapy. Clinical data-processing procedures and computer technologies play an important role in this context. Following clinical adaptation and validation within the framework of clinico-genomic trials, models are expected to advance the prospect of individualized treatment optimization, this being the long term goal of the emergent scientific, technological and medical discipline of in silico oncology. Treatment optimization is to be achieved through experimentation in silico i.e. on the computer. Moreover, provision of improved insight into tumor dynamics and optimization of clinical trial design and interpretation constitute short- and mid-term goals of this new domain. The IEEE-EMBS technically co-sponsored 5th International Advanced Research Workshop on In Silico Oncology and Cancer Investigation (5th IARWISOCI) (www.5th-iarwisoci.iccs.ntua.gr), being also the transatlantic TUMOR project workshop (www.tumor-project.eu), proved an excellent opportunity for contributing to the shaping of the discipline. The presented papers deal with modeling of tumor dynamics and response to treatment from the biochemical to the macroscopic level and from basic science to clinics via information technology. They have been contributed (some by invitation) by top international researchers and research groups. The workshop took place in Athens, Greece on 23-24 October 2012. II. IN SILICO ONCOLOGY In silico oncology could be formally defined as being "…a complex and multiscale combination of sciences, technologies and clinical medicine intending to simulate malignant tumor growth and tumor and normal tissue response to therapeutic modalities at all biomedically meaningful spatio-temporal scales". Its long term goal is to quantitatively understand cancer and related phenomena and optimize therapeutic interventions by performing in silico experiments using clinical, imaging, histopathological, molecular and pharmacogenomic data from individual patients. In order to achieve such an ambitious goal translation of cancer models and oncosimulators into the clinical trials arena is a sine qua non condition. III. TOWARDS IN SILICO MEDICINE In silico oncology serves as an excellent paradigm of the emergent generic domain of in silico medicine, which has been designated as one of the key research areas to be supported by the European Commission’s upcoming research funding framework, "HORIZON 2020" (http://ec.europa.eu/research/horizon2020/index_en.cfm ). Horizon 2020 is the financial instrument implementing the Innovation Union, a Europe 2020 flagship initiative aimed at securing Europe's global competitiveness. Running from 2014 to 2020 with an €80 billion budget, the EU’s new program for research and innovation is part of the drive to create new growth and jobs in Europe. A branch of the program is entitled: “Using in-silico medicine for improving disease management and prediction.” The current and future importance of both in silico oncology and in silico medicine is therefore unquestionable.
In this paper, a previous continuum approach describing vascular tumor growth under angiogenic signaling is developed and extended via the inclusion of bevacizumab pharmacokinetics. The modeling approach to the problem addressed includes inter alia the building of the model (selection of equations, related assumptions, coupling with a pharmacokinetic model tailored to the bevacizumab paradigm, implementation and numerical solution) as well as a study of the vascular tumor growth model with results for free growth and an intermittent bevacizumab mono-therapy schedule.
The Continuous Mathematics Based Glioblastoma Oncosimulator is a platform for simulating, investigating, better understanding, and exploring the natural phenomenon of glioma tumor growth. Modelling of the diffusive-invasive behaviour of glioma tumour growth may have considerable therapeutic implications. A crucial component of the corresponding computational problem is the numerical treatment of the adiabatic Neumann boundary conditions imposed by the skull on the diffusive growth of gliomas and in particular glioblastoma multiforme (GBM). In order to become clinically acceptable such a numerical handling should ensure that no potentially life-threatening glioma cells disappear artificially due to oversimplifying assumptions applied to the simulated region boundaries. However, no explicit numerical treatment of the 3D boundary conditions under consideration has appeared in the literature to the best of the authors’ knowledge. Therefore, this paper aims at providing an outline of a novel, explicit and thorough numerical solution to this problem. Additionally, a brief exposition of the numerical solution process for a homogeneous approximation of glioma diffusion-invasion using the Crank – Nicolson technique in conjunction with the Conjugate Gradient system solver is outlined. The entire mathematical and numerical treatment is also in principle applicable to mathematically similar physical, chemical and biological diffusion based spatiotemporal phenomena which take place in other domains for example embryonic growth and general tissue growth and tissue differentiation. A comparison of the numerical solution for the special case of pure diffusion in the absence of boundary conditions with its analytical counterpart has been made. In silico experimentation with various adiabatic boundary geometries and non zero net tumour growth rate support the validity of the corresponding mathematical treatment. Through numerical experimentation on a set of real brain imaging data, a simulated tumour has shown to satisfy the expected macroscopic behaviour of glioblastoma multiforme, on concrete published clinical imaging data, including the adiabatic behaviour of the skull. The paper concludes with a number of remarks pertaining to the potential and the limitations of the diffusion-reaction approach to modelling multiscale malignant tumour dynamics.
Two blastemal nephroblastoma cases have been successfully clinically adapted under two simulation scenarios of a clinically-oriented multiscale computational model, providing insight into the tumor characteristics.
Nephroblastoma is the most common malignant renal tumor in children. Today about 90 % of the patients can be cured by chemotherapy and surgery. Most of the patients are enrolled in prospective clinical trials. In the SIOP (International Society of Pediatric Oncology) approach all children do receive preoperative chemotherapy to shrink the tumor before surgery. Patients with an excellent response to preoperative chemotherapy have a better outcome than those with a poor response. Response is not only defined by tumor volume shrinkage but also by the vanishing of blastemal tumor cells. Imaging studies are used to calculate tumor volume and to define vital, necrotic and cystic areas within a tumor. In case of nephroblastoma it would be of most importance to further separate the vital tumor in blastemal and non-blastemal components. This would allow changing treatment at an early phase in case of remaining blastema. Using the DoctorEye software, tumors can be easily rendered and histograms of the signal intensities within a tumor are possible to calculate. Our preliminary results show that these histograms give further insight in the tumors in single patients by correlating them with the histological findings.
In the present study, methods aiming at supporting the personalization of an Acute Lymphoblastic Leukemia (ALL) Model (ALL Oncosimulator), already in development by the In Silico Oncology Group, National Technical University of Athens, are provided. Specifically, a population pharmacokinetic model for orally administered prednisone in children with ALL is developed, and the ability of classification algorithms to predict the prednisone response using gene expression data is studied.
In the past decades a great progress in cancer research has been made although medical treatment is still widely based on empirically established protocols which have many limitations. Computational models address such limitations by providing insight into the complex biological mechanisms of tumor progression. A set of clinically-oriented, multiscale models of solid tumor dynamics has been developed by the In Silico Oncology Group (ISOG), Institute of Communication and Computer Systems (ICCS)-National Technical University of Athens (NTUA) to study cancer growth and response to treatment. Within this context using certain representative parameter values, tumor growth and response have been modeled under a cancer preoperative chemotherapy protocol in the framework of the SIOP 2001/GPOH clinical trial. A thorough cross-method sensitivity analysis of the model has been performed. Based on the sensitivity analysis results, a reasonable adaptation of the values of the model parameters to a real clinical case of bilateral nephroblastomatosis has been achieved. The analysis presented supports the potential of the model for the study and eventually the future design of personalized treatment schemes and/or schedules using the data obtained from in vitro experiments and clinical studies.
The aim of this chapter is to provide a brief introduction into the basics of a top-down multilevel tumor dynamics modeling method primarily based on discrete entity consideration and manipulation. The method is clinically oriented, one of its major goals being to support patient individualized treatment optimization through experimentation in silico (= on the computer). Therefore, modeling of the treatment response of clinical tumors lies at the epicenter of the approach. Macroscopic data, including i.a. anatomic and metabolic tomographic images of the tumor, provide the framework for the integration of data and mechanisms pertaining to lower and lower biocomplexity levels such as clinically approved cellular and molecular biomarkers. The method also provides a powerful framework for the investigation of multilevel (multiscale) tumor biology in the generic investigational context. The Oncosimulator, a multiscale physics and biomedical engineering concept and construct tightly associated with the method and currently undergoing clinical adaptation, optimization and validation, is also sketched. A brief outline of the approach is provided in natural language. Two specific models of tumor response to chemotherapeutic and radiotherapeutic schemes are briefly outlined and indicative results are presented in order to exemplify the application potential of the method. The chapter concludes with a discussion of several important aspects of the method including i.a. numerical analysis aspects, technological issues, model extensions and validation within the framework of actual running clinico-genomic trials. Future perspectives and chal-lenges are also addressed.
Modelling of the diffusive-invasive behaviour of glioma tumour growth is an active field of Virtual Physiological Human (VPH) research with considerable therapeutic implications. A crucial component of the corresponding computational problem is the numerical handling of the adiabatic Neumann boundary conditions imposed by the skull on the diffusive growth of gliomas and in particular glioblastoma multiforme (GBM). In order to become clinically acceptable such a numerical handling should ensure that no potentially life-threatening glioma cells disappear artificially due to oversimplifying assumptions applied to the simulated region boundaries. However, to the best of the authors’ knowledge no explicit numerical treatment of the 3D boundary conditions under consideration has appeared in the literature. Therefore, this paper aims at providing an outline of a novel, explicit and thorough numerical solution to this problem. Additionally, a brief exposition of the numerical solution process for a homogeneous approximation of glioma diffusion-invasion using the Crank – Nicolson technique in conjunction with the Conjugate Gradient system solver is outlined. The entire mathematical and numerical treatment is also in principle applicable to mathematically similar physical, chemical and biological phenomena. A comparison of the numerical solution for the special case of pure diffusion in the absence of boundary conditions with its analytical counterpart has been made. In silico experimentation with various adiabatic boundary geometries and non zero net tumour growth rate support the validity of the corresponding mathematical treatment. Through numerical experimentation on a set of real brain imaging data, a simulated tumour has shown to satisfy the expected macroscopic behaviour of glioblastoma multiforme including the adiabatic behaviour of the skull. The expected GBM macroscopic behaviour has been based on concrete published clinical imaging data. The paper concludes with a number of remarks pertaining to the potential and the limitations of the diffusion-reaction approach to modelling multiscale malignant tumour dynamics.
The TUMOR project aims at developing a European clinically oriented semantic-layered cancer digital model repository from existing EU projects that will be interoperable with the US grid-enabled semantic-layered digital model repository platform at CViT.org (Center for the Development of a Virtual Tumor, Massachusetts General Hospital (MGH), Boston, USA) which is NIH/NCI-caGRID compatible. In this paper we describe the modular and federated architecture of TUMOR that effectively addresses model integration, interoperability, and security related issues.
The study of the diffusive behavior of glioma tumor growth is an active field of biomedical research with considerable therapeutic implications. An important aspect of the corresponding computational problem is the mathematical handling of boundary conditions. This paper aims at providing an explicit and thorough numerical formulation of the adiabatic Neumann boundary conditions imposed by the skull on the diffusive growth of gliomas and in particular on glioblastoma multiforme (GBM). Additionally, a detailed exposition of the numerical solution process for a homogeneous approximation of glioma invasion using the Crank–Nicolson technique in conjunction with the Conjugate Gradient system solver is provided. The entire mathematical and numerical treatment is also in principle applicable to mathematically similar physical, chemical and biological phenomena. A comparison of the numerical solution for the special case of pure diffusion in the absence of boundary conditions or equivalently in the presence of adiabatic boundaries placed in infinity with its analytical counterpart is presented. Numerical simulations for various adiabatic boundary geometries and non zero net tumor growth rate support the validity of the corresponding mathematical treatment. Through numerical experimentation on a set of real brain imaging data, a simulated tumor has shown to satisfy the expected macroscopic behavior of glioblastoma multiforme including the adiabatic behavior of the skull. The paper concludes with a number of remarks pertaining to both the biological problem addressed and the more generic diffusion–reaction context.
Image-based modeling of tumor growth combines methods from cancer simulation and medical imaging. In this context, we present a novel approach to adapt a healthy brain atlas to MR images of tumor patients. In order to establish correspondence between a healthy atlas and a pathologic patient image, tumor growth modeling in combination with registration algorithms is employed. In a first step, the tumor is grown in the atlas based on a new multiscale, multiphysics model including growth simulation from the cellular level up to the biomechanical level, accounting for cell proliferation and tissue deformations. Large-scale deformations are handled with an Eulerian approach for finite element computations, which can operate directly on the image voxel mesh. Subsequently, dense correspondence between the modified atlas and patient image is established using nonrigid registration. The method offers opportunities in atlas-based segmentation of tumor-bearing brain images as well as for improved patient-specific simulation and prognosis of tumor progression.
The book is available at http://www.5th-iarwisoci.iccs.ntua.gr/
In the present study, methods aiming at supporting the personalization of an Acute Lymphoblastic Leukemia (ALL) Model (ALL Oncosimulator), already in development by the In Silico Oncology Group, National Technical University of Athens, are provided. Specifically, a population pharmacokinetic model for orally administered prednisone in children with ALL is developed, and the ability of classification algorithms to predict the prednisone response using gene expression data is studied.
Prior to an eventual clinical adaptation and validation of any clinically oriented model, a thorough study of its dynamic behavior is a sine qua non. Such a study can also elucidate aspects of the interplay of the involved biological mechanisms. Toward this goal, the paper focuses on an in-depth investigation of the free growth behavior of a macroscopically homogeneous malignant tumor system, using a discrete model of tumor growth. We demonstrate that when a clinical tumor grows exponentially, the following preconditions must be fulfilled: (a) time- and space-independent tumor dynamics, in terms of the transition rates among the considered cell categories and the duration of the cell cycle phases, and (b) a tumor system in a state of population equilibrium. Moreover, constant tumor dynamics during the simulation are assumed. In order to create a growing tumor, a condition that the model parameters must fulfill has been derived based on an analytical treatment of the model’s assumptions. A detailed parametric analysis of the model has been performed, in order to determine the impact and the interdependences of its parameters with focus on the free growth rate and the composition of cell population. Constraining tumor cell kinetics, toward limiting the number of possible solutions (i.e., sets of parameters) to the problem of adaptation to the real macroscopic features of a tumor, is also discussed. After completing all parametric studies and after adapting and validating the model on clinical data, it is envisaged to end up with a reliable tool for supporting clinicians in selecting the most appropriate pattern, extracted from several candidate therapeutic schemes, by exploiting tumor- and patient-specific imaging, molecular and histological data.
Simulators of tumour growth can estimate the evolution of tumour volume and the quantity of various categories of cells as functions of time. However, the execution time of each simulation often takes several dozens of minutes (depending upon the dataset resolution), which clearly prevents easy interaction. The modern graphics processing unit (GPU) is not only a powerful graphics engine but also a highly parallel programmable processor featuring peak arithmetic performance and memory bandwidth that substantially outpaces its CPU counterpart. However, despite this, the GPU is little used in the context of the Virtual Physiological Human (VPH). This paper provides a case study to demonstrate the performance advantages that can be gained by using the GPU appropriately in the context of a VPH project in which the study of tumour growth is a central activity. We also analyse the algorithm performance on different modern parallel processing architectures, including multicore CPU and many-core GPU.
Modeling of tumor growth has been performed according to various approaches addressing different biocomplexity levels and spatiotemporal scales. Mathematical treatments range from partial differential equation based diffusion models to rule-based cellular level simulators, aiming at both improving our quantitative understanding of the underlying biological processes and, in the mid- and long term, constructing reliable multi-scale predictive platforms to support patient-individualized treatment planning and optimization. The aim of this paper is to establish a multi-scale and multi-physics approach to tumor modeling taking into account both the cellular and the macroscopic mechanical level. Therefore, an already developed biomodel of clinical tumor growth and response to treatment is self-consistently coupled with a biomechanical model. Results are presented for the free growth case of the imageable component of an initially point-like glioblastoma multiforme tumor. The composite model leads to significant tumor shape corrections that are achieved through the utilization of environmental pressure information and the application of biomechanical principles. Using the ratio of smallest to largest moment of inertia of the tumor material to quantify the effect of our coupled approach, we have found a tumor shape correction of 20% by coupling biomechanics to the cellular simulator as compared to a cellular simulation without preferred growth directions. We conclude that the integration of the two models provides additional morphological insight into realistic tumor growth behavior. Therefore, it might be used for the development of an advanced oncosimulator focusing on tumor types for which morphology plays an important role in surgical and/or radio-therapeutic treatment planning.
The challenge of modelling cancer presents a major opportunity to improve our ability to reduce mortality from malignant neoplasms, improve treatments and meet the demands associated with the individualization of care needs. This is the central motivation behind the ContraCancrum project. By developing integrated multi-scale cancer models, ContraCancrum is expected to contribute to the advancement of in silico oncology through the optimization of cancer treatment in the patient-individualized context by simulating the response to various therapeutic regimens. The aim of the present paper is to describe a novel paradigm for designing clinically driven multi-scale cancer modelling by bringing together basic science and information technology modules. In addition, the integration of the multi-scale tumour modelling components has led to novel concepts of personalized clinical decision support in the context of predictive oncology, as is also discussed in the paper. Since clinical adaptation is an inelastic prerequisite, a long-term clinical adaptation procedure of the models has been initiated for two tumour types, namely non-small cell lung cancer and glioblastoma multiforme; its current status is briefly summarized.
Time evolution of various tumor subpopulations. Alternative presentation of various tumor subpopulations for the four virtual tumor scenarios implemented (T1: Tumor1, T2: Tumor2, T3: Tumor3, T4: Tumor4, defined by the parameter values indicated in Table 1). Time evolution of A) proliferating, B) dead, C) terminally differentiated, D) stem, and E) LIMP (committed progenitor) cells. (TIF)
The development of computational models for simulating tumor growth and response to treatment has gained significant momentum during the last few decades. At the dawn of the era of personalized medicine, providing insight into complex mechanisms involved in cancer and contributing to patient-specific therapy optimization constitute particularly inspiring pursuits. The in silico oncology community is facing the great challenge of effectively translating simulation models into clinical practice, which presupposes a thorough sensitivity analysis, adaptation and validation process based on real clinical data. In this paper, the behavior of a clinically-oriented, multiscale model of solid tumor response to chemotherapy is investigated, using the paradigm of nephroblastoma response to preoperative chemotherapy in the context of the SIOP/GPOH clinical trial. A sorting of the model's parameters according to the magnitude of their effect on the output has unveiled the relative importance of the corresponding biological mechanisms; major impact on the result of therapy is credited to the oxygenation and nutrient availability status of the tumor and the balance between the symmetric and asymmetric modes of stem cell division. The effect of a number of parameter combinations on the extent of chemotherapy-induced tumor shrinkage and on the tumor's growth rate are discussed. A real clinical case of nephroblastoma has served as a proof of principle study case, demonstrating the basics of an ongoing clinical adaptation and validation process. By using clinical data in conjunction with plausible values of model parameters, an excellent fit of the model to the available medical data of the selected nephroblastoma case has been achieved, in terms of both volume reduction and histological constitution of the tumor. In this context, the exploitation of multiscale clinical data drastically narrows the window of possible solutions to the clinical adaptation problem.
In silico (on the computer) oncology is a complex and multiscale combination of sciences and technologies that focuses on the study and modelling of biological mechanisms related to the phenomenon of cancer at all levels of biocomplexity. In silico oncology simulation models may be used for evaluating and comparing different therapeutic schemes, while at the same time considering different values of critical parameters which present substantial inter-patient variability. As the number of the involved parameters characterizing both the complex tumour biosystem and possible treatment schemes increases, the resulting exponential increase in computational requirements makes the use of a grid environment for the execution of the simulations a particularly attractive solution. In this paper, a grid-enabled simulation environment for the execution of in silico oncology radiotherapy simulations on grid infrastructures is presented and implementation details are discussed. The environment provides a web portal as the end-user interface and contains advanced features that facilitate the execution of in silico oncology simulations in grid environments. Special consideration has been given during the development of the environment in order to simplify the maintenance and extension of the application, while additional services for Quality of Service provisioning have been applied. The simulation environment has been employed in order to perform several scenarios of glioblastoma multiforme radiotherapy simulations on the Enabling Grids for E-sciencE (EGEE) grid infrastructure. Indicative simulation results, as well as statistics regarding execution times on the grid, are presented.
A brief introduction into the basics of a top-down multiscale tumour dynamics modelling method primarily based on the consideration and manipulation of discrete biological entities and events is presented. The method is clearly clinically oriented. One of its major goals is to support patient individualized treatment optimization through experimentation in silico (=on the computer). Therefore, modelling of the treatment response of clinical tumours lies at the epicenter of the approach. Macroscopic data, including i.a. anatomic and metabolic tomographic images of the tumour, provide the Virtual Physiological Human (VPH) framework for the integration of data and mechanisms pertaining to lower and lower biocomplexity levels such as clinically approved cellular and molecular biomarkers. The method also provides a powerful framework for the investigation of multiscale tumour biology in the generic investigational context. The Oncosimulator, a multiscale basic science and biomedical engineering concept and construct tightly associated with the method and currently undergoing clinical adaptation, optimization and validation, is also sketched. Indicative results are presented in order to exemplify the application potential of the method. The paper concludes with a brief discussion of several aspects of the approach and future perspectives.
The aim of this chapter is to provide a brief introduction into the basics of a top-down multilevel tumor dynamics modeling method primarily based on discrete entity consideration and manipulation. The method is clinically oriented, one of its major goals being to support patient individualized treatment optimization through experimentation in silico (=on the computer). Therefore, modeling of the treatment response of clinical tumors lies at the epicenter of the approach. Macroscopic data, including i.a. anatomic and metabolic tomographic images of the tumor, provide the framework for the integration of data and mechanisms pertaining to lower and lower biocomplexity levels such as clinically approved cellular and molecular biomarkers. The method also provides a powerful framework for the investigation of multilevel (multiscale) tumor biology in the generic investigational context. The Oncosimulator, a multiscale physics and biomedical engineering concept and construct tightly associated with the method and currently undergoing clinical adaptation, optimization and validation, is also sketched. A brief outline of the approach is provided in natural language. Two specific models of tumor response to chemotherapeutic and radiotherapeutic schemes are briefly outlined and indicative results are presented in order to exemplify the application potential of the method. The chapter concludes with a discussion of several important aspects of the method including i.a. numerical analysis aspects, technological issues, model extensions and validation within the framework of actual running clinico-genomic trials. Future perspectives and challenges are also addressed. Comment: Dedicated to 75 anniversary of Prof. Duechting, Siegen. Keywords: top down model, discrete event based cancer simulation technique, DEBCaST, cancer modeling, cancer multiscale modeling, discrete event simulation, clinically oriented cancer modeling, oncosimulator, in silico oncology, glioblastoma multiforme, radiation therapy, temozolomide, chemotherapy, cancer biomathematics, cancer bioinformatics, cancer integrative biology, clinical trials, in silico experiment; virtual physiological human, VPH
A brief introduction to the emergent discipline of in silico oncology along with its predominantly Platonic character is provided. The conceptual origins of the oncosimulator, one of the key notions and constructs of the discipline are traced back to antiquity.
In silico modeling of tumor growth and treatment response within the ContraCancrum oncosimulator context requires a specific handling of the molecular variation of patient tumors which correspond to variable drug response. In this context we have looked to correlate the in vitro expression profiles with drug sensitivity in order to achieve a broad indication of sensitivity that can be used as a modifier of cell kill within the simulator. Using the exemplar case of the drug temozolomide in glioblastoma, we consider how a gene signature capable of predicting drug sensitivity can be utilized within a simulation of tumor growth and treatment response.
A hypermatrix-operator formulation of a multiscale top-down tumor dynamics modeling method primarily based on the consideration and manipulation of discrete biological entities and events is outlined. The method is clearly clinically oriented. One of its major goals is to support patient individualized treatment optimization through experimentation in silico (=on the computer). Therefore, modeling of the treatment response of clinical tumors lies at the epicenter of the approach. Macroscopic data, including i.a. anatomic and metabolic tomographic images of the tumor, provide the Virtual Physiological Human (VPH) framework for the integration of data and mechanisms pertaining to lower and lower biocomplexity levels such as clinically approved cellular and molecular biomarkers. The method also provides a powerful framework for the investigation of multiscale tumor biology in the generic investigational context. The Oncosimulator, a multiscale basic science and biomedical engineering concept and construct tightly associated with the method is also sketched. The current status of the simulation executions, clinical adaptation and validation of the corresponding models is briefly outlined. The paper concludes with a brief discussion of several aspects of the approach and future perspectives.
This short paper provides a brief outline of the main components and the developmental and translational process of the ACGT Oncosimulator. The Oncosimulator is an integrated software system simulating in vivo tumor response to therapeutic modalities within the clinical trial environment. It aims at supporting patient individualized optimization of cancer treatment. The four dimensional simulation module embedded in the Oncosimulator is based on the multiscale, top-down, discrete entity – discrete event cancer simulation approach and has been specified for the cases of nephroblastoma and breast cancer. Chemotherapeutic treatment in the neoadjuvant setting according to protocols included in the SIOP 2001/GPOH and the TOP clinical trial respectively is considered. The technology modules of the Oncosimulator include i.a. anonymized / pseudonymized multiscale data handling (including imaging, histopathological, molecular, clinical and treatment data), image processing and molecular/histopathological data preprocessing, invocation of code execution via an intelligent portal (“RecipeSheet”), execution of the code on either a cluster or grid, collection and visualization of the predictions and numerical analysis (including sensitivity) of the simulation model. Exploratory analyses have revealed the importance of critical model parameters such as the symmetric division probability, the dormancy probability of a newborn tumor cell and the cell kill probability for the response of a solid tumor to chemotherapeutic treatment. Furthermore, plausible parameter values have been adapted to real multiscale clinical trial data in a consistent way, thus supporting the predictive potential of the Oncosimulator. Clinical adaptation and validation are in progress. Completion of these time demanding phases is expected to lead to the clinical translation of the system. In parallel, the Oncosimulator simulation component is currently expanded in order to allow studying the immune system response to cancer as well as related phenomena and treatment techniques.
In the present paper, the dynamic behavior of a clinically-oriented simulation model of breast tumor response to chemotherapy is investigated. The model incorporates various biological processes such as cycling of proliferating cells, quiescence, differentiation and cell death. Indicative results drawn from an extensive parametric analysis of the model are presented.
In this paper indicative sensitivity studies performed with an already developed simulation model of nephroblastoma response to chemotherapy are presented. The sensitivity analyses reported involve model parameters with a critical effect on the following aspects of the model: tumor initialization, tumor growth and tumor response to chemotherapy.
This short communication briefly outlines the major components and the integration steps of the Oncosimulator that is being developed within the framework of the European Commission funded ContraCancrum project. The Oncosimulator is a technologically advanced multiscale tumor growth and treatment response system aiming at supporting patient individualized treatment decisions. An indicative example of the adopted mathematical approaches as well as a simple example of numerical code validation are provided. The document concludes with a short discussion on the characteristics of the major modeling approaches that refer to the cellular and higher biocomplexity levels since the latter constitute the basis for the entire Oncosimulator integration.
Glioblastoma is the most aggressive type of glioma. During the last decades, several models have been proposed for simulating the growth procedure of glioma. Diffusive models have been used for simulating the spatiotemporal change of glioma cell concentration. The most recent ones take into account tissue heterogeneity and the anisotropic migration of glial cells along white fibers. The main purpose of this paper is to present a novel method for computing the coefficients for the diffusion, taking into account the proportion of white and gray matter in atlases extracted by medical images. Some initial experiments are presented
A careful boundary condition handling is a sine qua non prerequisite for a reliable diffusion based solution to the problem of clinical tumor growth and in particular glioma progression. However, to the best of our knowledge no explicit treatment of the numerical application of boundary conditions in this context has appeared in the literature as yet. Therefore, the aim of this paper is to outline a detailed numerical handling of the boundary conditions imposed by the presence of the skull in the case of glioblastoma multiforme (GBM), a highly diffusive and invasive brain tumor.
Simulators of tumor growth can estimate the evolution of tumor volume and the quantity of cells as functions of time. However, the execution time of each simulation often takes several dozens of minutes (depending on the dataset resolution), which clearly prevents easy interaction. The modern graphics processing unit (GPU) is not only a powerful graphics engine but also a highly parallel programmable processor featuring peak arithmetic and memory bandwidth that substantially outpaces its CPU counterpart. In this work, by programming an NVIDIA GPU device with CUDA, we have designed algorithms to parallelise the time-consuming process of tumor simulation on the GPU, which allows the local application of basic biological rules and subsequently leads to the spatiotemporal simulation of the tumor system
The ACGT Oncosimulator is an integrated Grid-based system, under development within a 25-partner European-Japanese project, for patient-specific simulation of the response of a tumor and its surrounding tissue to various forms of therapy. The validation of the simulation code is an activity requiring extensive human-driven visual investigation of the influence of each of the dozens of parameters to the code, initially by comparing results from simulations carried out with different parameter values. This activity requires that users be supported in specifying simulation runs based on chosen parameter-value combinations, submitting the runs for execution on the Grid, then obtaining result visualisations that help in making the necessary comparisons. We report on our development and early use of the OncoRecipeSheet, an environment designed to meet these requirements.
In silico oncology is anticipated to gain a more individualized treatment for patients with cancer. To run in silico oncology models data from individual patients are essential. The more and the more accurate these data are the more precise the results of the in silico oncology models will be. Imaging studies are used to calculate tumor volume and define vital, necrotic and cystic areas within a tumor. Though the visual interpretation of magnetic resonance (MR) images is based on qualitative observation of variation in signal intensity a correlation of signal intensities with histological features of a tumor is not possible. Quantitative methods are needed for reliable follow-up or inter-individual studies. Using DoctorEye tumors can be easily rendered and histograms of the signal intensities within a tumor as well as mean and median signal intensities are calculated. In gliomas the histogram of signal intensities of cerebrospinal fluid is used as a reference for standardization of signal intensities. Our results in gliomas suggest that these histograms add value for a better description of tumors for the use in insilico oncology models.
In silico oncology is a demanding research field that entails the development of complex simulation models for the prediction of malignant tumor growth and the response of normal tissue to therapeutic modalities. A system that would enable researchers exchange, share and validate their models could therefore aid the in silico domain to the ultimate benefit of cancer patients. This paper presents the requirements of such a system, the solutions that can be implemented to address them, as well as a number of guidelines for its successful implementation.
ContraCancrum means ‘Against Cancer’ in Latin. This is because the ContraCancrum project aims to pave the way for translating clinically validated multilevel cancer models into clinical practice. The models will assist the clinician to define the optimal therapy for the individual patient taking into consideration all the available clinical data from different scales (molecular to tissue level), modalities (e.g. multimodal cancerimaging) and examinations (e.g. before and after therapy).
In this paper an advanced, clinically oriented multiscale cancer model of breast tumor response to chemotherapy is presented. The paradigm of early breast cancer treated by epirubicin according to a branch of an actual clinical trial (the Trial of Principle, TOP trial) has been addressed. The model, stemming from previous work of the In Silico Oncology Group, National Technical University of Athens, is characterized by several crucial new features, such as the explicit distinction of proliferating cells into stem cells of infinite mitotic potential and cells of limited proliferative capacity, an advanced generic cytokinetic model and an improved tumor constitution initialization technique. A sensitivity analysis regarding critical parameters of the model has revealed their effect on the behavior of the biological system. The favorable outcome of an initial step towards the clinical adaptation and validation of the simulation model, based on the use of anonymized data from the TOP clinical trial, is presented and discussed. Two real clinical cases from the TOP trial with variable molecular profile have been simulated. A realistic time course of the tumor diameter and a reduction in tumor size in agreement with the clinical data has been achieved for both cases by selection of reasonable model parameter values, thus demonstrating a possible adaptation process of the model to real clinical trial data. Available imaging, histological, molecular and treatment data are exploited by the model in order to strengthen patient individualization modeling. The expected use of the model following thorough clinical adaptation, optimization and validation is to simulate either several candidate treatment schemes for a particular patient and support the selection of the optimal one or to simulate the expected extent of tumor shrinkage for a given time instant and decide on the adequacy or not of the simulated scheme.
Glioma is the most aggressive type of brain tumor. Several mathematical models have been developed during the last two decades, towards simulating the mechanisms that govern the development of glioma. The most common models use the diffusion-reaction equation (DRE) for simulating the spatiotemporal variation of tumor cell concentration. The proposed diffusive models have mainly used finite differences (FDs) or finite elements (FEs) for the approximation of the solution of the partial differential DRE. This paper presents experimental results on the comparison of the FEs and FDs, especially focused on the glioma model case. It is studied how the different meshes of brain can affect computational consistency, simulation time and efficiency of the model. The experiments have been studied on a test case, for which there is a known algebraic expression of the solution. Thus, it is possible to calculate the error that the different models yield.
The ACGT Oncosimulator is an integrated Grid-based system, under development within a 25-partner European-Japanese project, for patient-specific simulation of the response of a tumour and surrounding tissue to various forms of therapy. The validation of the simulation code is an activity requiring extensive human-driven visual investigation of the influence of each of the dozens of parameters to the code, and comparison of the simulation results against the known outcomes of past patient treatments. This activity therefore calls for a visualisation environment that supports users in working with an extremely large potential result space, and in rapidly setting up visualisations that highlight the differences between chosen subsets of available results. We describe the innovative features of the OncoRecipeSheet, an environment designed to meet these requirements.
The tremendous rate of accumulation of experimentally and clinically extracted knowledge concerning cancer at all levels of biocomplexity dictates the development of integrative in silico models of tumour dynamics in order to better understand and treat the disease. Since the eventual translation of biomodels into clinical practice presupposes successful clinical validation we have developed a number of multiscale cancer simulation models oriented towards patient individualized treatment optimization. A top-down modelling approach based primarily on discrete event/state simulation has been proposed, developed and implemented. The emerging simulators have been serving as the core simulation modules (oncosimulators) of both the EC funded research projects Contra- Cancrum and ACGT. In this paper a brief outline of the basics of the approach along with paradigmal results demonstrating the effect of symmetric division of cancer stem cells on the cellular constitution of a tumour are presented. The potential and extensibility of the models are discussed.
The ContraCancrum project aims at developing a composite multilevel platform for simulating malignant tumor development and tumor and normal tissue response to therapeutic modalities and treatment schedules. The project aims at having an impact primarily in (a) the better under-standing of the natural phenomenon of cancer at different levels of biocomplexity and most importantly (b) the disease treatment optimization procedure in the patient’s individualized context by simulating the response to various therapeutic regimens. Fundamental biological mechanisms involved in tumor development and tumor and normal tissue treatment response such as metabolism, cell cycle, tissue mechanics, cell survival following treatment etc. are modeled also addressing stem cells in the context of both tumor and normal tissue behavior. The simulators exploit several discrete and continuous mathematics methods such as cellular automata, the generic Monte Carlo technique, finite elements, differential equations, novel dedicated algorithms etc. The predictions of the simulators rely on the imaging, histopathological, molecular and clinical data of the patient. ContraCancrum deploys two important clinical studies for validating the models, one on lung cancer and one on gliomas. The crucial validation work is based on comparing the multi-level therapy simulation predictions with multi-level patient data, acquired before and after therapy. ContraCancrum aims to pave the way for translating clinically validated multilevel cancer models into clinical practice.
The e-book of the Proceedings is available on www.4th-iarwisoci.iccs.ntua.gr
The tremendous rate of accumulation of experimental and clinical knowledge pertaining to cancer dictates the development of a theoretical framework for the meaningful integration of such knowledge at all levels of biocomplexity. In this context our research group has developed and partly validated a number of spatiotemporal simulation models of in vivo tumour growth and in particular tumour response to several therapeutic schemes. Most of the modeling modules have been based on discrete mathematics and therefore have been formulated in terms of rather complex algorithms (e.g. in pseudocode and actual computer code). However, such lengthy algorithmic descriptions, although sufficient from the mathematical point of view, may render it difficult for an interested reader to readily identify the sequence of the very basic simulation operations that lie at the heart of the entire model. In order to both alleviate this problem and at the same time provide a bridge to symbolic mathematics, we propose the introduction of the notion of hypermatrix in conjunction with that of a discrete operator into the already developed models. Using a radiotherapy response simulation example we demonstrate how the entire model can be considered as the sequential application of a number of discrete operators to a hypermatrix corresponding to the dynamics of the anatomic area of interest. Subsequently, we investigate the operators' commutativity and outline the "summarize and jump" strategy aiming at efficiently and realistically address multilevel biological problems such as cancer. In order to clarify the actual effect of the composite discrete operator we present further simulation results which are in agreement with the outcome of the clinical study RTOG 83-02, thus strengthening the reliability of the model developed.
The present paper outlines the initial version of the ACGT (Advancing Clinico-Genomic Trials) - an Integrated Project, partly funded by the EC (FP6-2005-IST-026996)I-Oncosimulator as an integrated software system simulating in vivo tumour response to therapeutic modalities within the clinical trials environment aiming to support clinical decision making in individual patients. Cancer treatment optimization is the main goal of the system. The document refers to the technology of the system and the clinical requirements and the types of medical data needed for exploitation in the case of nephroblastoma. The outcome of an initial step towards the clinical adaptation and validation of the system is presented and discussed. Use of anonymized real data before and after chemotherapeutic treatment for the case of the SIOP 2001/GPOH nephroblastoma clinical trial constitutes the basis of the clinical adaptation and validation process. 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. Obviously as more and more sets of medical data are exploited the reliability of the model “tuning” is expected to increase. 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 tremendous rate of accumulation of experimentally and clinically extracted knowledge concerning cancer at all levels of biocomplexity dictates the development of integrative in silico models of tumour dynamics in order to better understand and treat the disease. Since the eventual translation of biomodels into clinical practice presupposes successful clinical validation we have developed a number of multiscale cancer simulation models oriented towards patient individualized treatment optimization. A top-down modelling approach based primarily on discrete event/state simulation has been proposed, developed and implemented. The emerging simulators have been serving as the core simulation modules (oncosimulators) of both the EC funded research projects Contra-Cancrum and ACGT. In this paper a brief outline of the basics of the approach along with paradigmal results demonstrating the effect of symmetric division of cancer stem cells on the cellular constitution of a tumour are presented. The potential and extensibility of the models are discussed.
A constantly increasing number of applications from various scientific sectors are finding their way towards adopting Grid technologies in order to take advantage of their capabilities: the advent of Grid environments made feasible the solution of computational intensive problems in a reliable and cost-effective way. The aim of this paper is to demonstrate how multilevel tumour growth and response to therapeutic treatment models can be used in order to simulate clinical trials, with the long-term intention of better designing clinical studies and understanding their outcome based on basic biological science. For this purpose, a computer simulation model of glioblastoma multiforme response to radiotherapy has been applied to perform the aforementioned simulation in a real Grid environment by also taking into account historical data. The proposed approach yields very good results for the conducted virtual trial since these are in agreement with the outcome of the real clinical study, while the use of Grid technologies demonstrate and highlight their added-value.
The aim of this paper is to demonstrate how multilevel tumour growth and response to therapeutic treatment models can be used in order to simulate clinical trials, with the long-term intention of both better designing clinical studies and understanding their outcome based on basic biological science. For this purpose, an already developed computer simulation model of glioblastoma multiforme response to radiotherapy has been used and a clinical study concerning glioblastoma multiforme response to radiotherapy has been simulated. In order to facilitate the simulation of such virtual trials, a toolkit enabling the user-friendly execution of the simulations on grid infrastructures has been designed and developed. The results of the conducted virtual trial are in agreement with the outcome of the real clinical study.
PHILOSOPHIAE NATURALIS PRINCIPIA MATHEMATICA - PARS SECUNDA - DE MATERIA VIVENTI - DE PHENOMENIS IN MULTIS PLANIS -DE CANCRO ----- [A book to be jointly written by thousands of interdisciplinary researchers worldwide]----- Dr. Georgios Stamatakos discusses the fundamentals of multiscale cancer modelling------ The VIDEO is available at http://ecancer.org/tv/pubdate/105------ ( ecancertv is the video section of the clinical journal ecancermedicalscience ( ecancer.org/ecms )----- The Workshop Website is http://ec.europa.eu/information_society/events/ict_bio/2008/ta-cancer-wkshp/index_en.htm -------------------------------------------------------------------- Newton’s classic book “PHILOSOPHIAE NATURALIS PRINCIPIA MATHEMATICA” (Mathematical Principles of Natural Philosophy) has always been a source of inspiration. In this context the collective global efforts to mathematically model life and disease could be viewed as an obvious extension of the Principia approach. The fundamentals of this domain - still to be determined - could be entitled “PHILOSOPHIAE NATURALIS PRINCIPIA MATHEMATICA – PARS SECUNDA – DE MATERIA VIVENTI” ( Mathematical Principles of Natural Philisophy – Second Part – of Living Matter) . In the specific case of cancer the latter could be complemented by the subtitle “DE PHENOMENIS IN MULTIS PLANIS – DE CANCRO” ( of Multilevel Phenomena – of Cancer). Due to the "monstrous" complexity of life and disease, contemporary “science servants” need inspiration, strength and courage. I do believe that reflecting on Newton’s Principia can provide all three abundantly.
Mathematical and computational tumor dynamics models can provide considerable insight into the relative importance and interdependence of related biological mechanisms. They may also suggest the existence of optimal treatment windows in the generic setting. Nevertheless, they cannot be translated into clinical practice unless they undergo a strict and thorough clinical validation and adaptation. In this context one of the major actions of the EC funded project ldquoAdvancing Clinico-Genomic Trials on Cancerrdquo (ACGT) is dedicated to the development of a patient specific four dimensional multiscale tumor model mimicking the nephroblastoma tumor response to chemotherapeutic agents according to the SIOP 2001/GPOH clinical trial. Combined administration of vincristine and dactinomycin is considered. The patient#x2019;s pseudoanonymized imaging, histopathological, molecular and clinical data are carefully exploited. The paper briefly outlines the basics of the model developed by the In Silico Oncology Group and particularly stresses the effect of stem/clonogenic, progenitor and differentiated tumor cells on the overall tumor dynamics. The need for matching the cell category transition rates to the cell category relative populations of free tumor growth for an already large solid tumor at the start of simulation has been clarified. A technique has been suggested and succesfully applied in order to ensure satisfaction of this condition. The concept of a nomogram matching the cell category transition rates to the cell category relative populations at the treatment baseline is introduced. Convergence issues are addressed and indicative numerical results are presented. Qualitative agreement of the modelpsilas behavior with the corresponding clinical trial experience supports its potential to constitute the basis for an optimization system within the clinical environment following completion of its clinical validation and optimization. In silico treatment experimentation in - - the patient individualized context is expected to constitute the primary application of the model.
The potential of cancer multilevel modeling has been particularly emphasized over the past years. Integration of multiscale experimental and clinical information pertaining to cancer via advanced computer models seems to considerably accelerate optimization of cancer treatment in the patient individualized context. However, a sine qua non prerequisite for such models to reach clinical practice is to be thoroughly tested through clinical trials for validation and optimization purposes. This is one of the major goals of the European Commission funded ldquoadvancing clinico-genomic trials on cancerrdquo (ACGT) project. This paper presents a discrete state based, four dimensional, multiscale tumor dynamics model that has been specially developed by the in silico oncology group in order to mimick the trial of principle (TOP) clinical trial concerning breast cancer treated with epirubicin. The TOP trial constitutes one of the ACGT clinical trials. A substantial part of the model can address other tumor types as well. The actual pseudoanonymized imaging, histopathological, molecular and clinical data of the patient are exploited. Special emphasis is put on the effect of cancer stem/clonogenic, progenitor, differentiated and dead cells, the cell category transition rates and the cell category relative populations within the tumor from the treatment baseline onwards. The importance of adaptation of the cell category relative populations to the cell category transition rates for free tumor growth is revealed and the concept of a pertinent nomogram is introduced. A method which ensures adaptation of these two sets of entities at the beginning of the simulation execution is proposed and subsequently successfully applied. Convergence and code checking issues are addressed. Indicative parametric/sensitivity studies are presented along with specific numerical findings. The modelpsilas behavior substantiates its potential to serve as the basis of a treatment optimization system fol- - lowing an eventually succesful completion of the clinical validation and optimization process.