Oncosimulator Models as Components of a Personal Health Record Platform can Enable and Enhance the Provision of Personalized Medical Treatment

Full text

!
!
Translating*VPH*to*the*Clinic*
!
!
!26-28!September!2016! Amsterdam,!the!Netherlands! www.vph-conference.org!
!
BOOK$OF$ABSTRACTS$
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
!
info@vph-conference.org!
www.vph-conference.org!

!
!
!
!
Title:!!! ! VPH2016,!book!of!abstracts!
!
Editor:!! ! Alfons!G.!Hoekstra!
!
Publisher:!!University!of!Amsterdam!
!
Year:! ! 2016!
!
ISBN:!! ! 978-90-826254-0-0!
!

Oncosimulator Models as Components of a Personal Health Record Platform
can Enable and Enhance the Provision of Personalized Medical Treatment
Eleni Ch. Georgiadi 1, Nikolaos A. Christodoulou 1, Christos Kyroudis 1, Feng Dong 2, Norbert Graf 3
and Georgios S. Stamatakos 1*
1. In Silico Oncology and In Silico Medicine Group, Institute of Communication and Computer Systems,
National Technical University of Athens, Greece
2. Department of Computer Science and Technology, University of Bedfordshire, UK
3. Pediatric Oncology and Hematology Clinic, University of Saarland, Germany
* Correspondence: gestam@mail.ntua.gr, 9 Iroon Polytechniou, GR 157 80, Zografos, Greece
1. Introduction
The risk, the development and the treatment of several major diseases such as cancer are affected by a
large number of factors. In this context maintaining clinical records of individual citizens that involve
cross-border activities in a consistent manner is a difficult task due to the diversity and the complexity of
the national healthcare systems. In recent years, personalized computational models simulating the
complex natural phenomenon of cancer have provided extensive quantitative insight into the mechanisms
involved in cancer. Therefore, they might contribute to the patient-specific therapy optimization on a
global level. They are to be viewed as potential powerful tools for personalized medicine. Nevertheless,
data collection, access, sharing and analysis constitute key issues regarding the application of models to a
wide number of real clinical data categories and sets. A solution to those problems would be of benefit to
the researchers regarding the effective validation of computer simulation models, to the clinicians in view
of the clinical translation of validated models in the form of clinical decision support systems and
obviously to the individual patients who could directly benefit from such a clinical translation.
To address these needs, the European Commission funded MyHealthavatar (MHA) project (FP7-ICT-
2011–9–600929) has designed and implemented a platform which acts as a citizen’s “bag” of personal
data that can be utilized for the aforementioned needs. A personalized model of nephroblastoma cancer
growth and response to chemotherapy referred to as the Wilms (Tumour) (WT) or Nephroblastoma
Oncosimulator has been integrated into the MHA platform through an external application.
MyHealthAvatar offers access, collection and sharing of long term and consistent personal health status
data as a proof of concept for the digital representation of the patient’s health status by putting special
emphasis on engaging the wider public. The Wilms Oncosimulator prediction model serves as a proof of
principle for a clinical decision support system aiming at a better modulating cancer treatment through the
utilization of the collected personal health data via in-silico models. The latter produces an in-silico
profiling of patients which further enriches their digital personal health record. Furthermore, the
integration of an external prototype web application called IAPETUS into the MHA platform serves as a
learning environment which helps to disseminate the importance of in-silico models in medicine to the
wider public, medical stakeholders, industry and funding agencies.
2. Materials and Methods
2.I The Nephroblastoma Wilms Tumour Oncosimulator
Nephroblastoma is the most common malignant renal tumour in children. Today treatments are based on
several multicenter trials and studies conducted by the International Society of Paediatric Oncology
(SIOP) society in Europe and the COG group in North America. The main objectives of these trials and
studies are to treat patients according to well defined risk groups in order to achieve the highest possible
cure rates, to decrease the frequency and the intensity of acute and late toxicities and to minimize the cost
of therapy. In that way the SIOP trials and studies largely focus on the issue of preoperative therapy. SIOP
Georgiadi et all. VPH2016 book of abstracts
297 Personalised Medicine II

has enrolled children with Wilms tumour into 6 studies up to now (SIOP 1, SIOP 2, SIOP 5, SIOP 6,
SIOP 9, SIOP 93-01). Graf et al [1] have conducted a review of these studies. The hallmark of the SIOP
RTSG approach is the preoperative chemotherapy with Vincristine and Actinomycin-D without preceding
mandatory histological assessment. Although the overall and the event-free survival of most renal
tumours is excellent, further improvements are needed to find better risk stratifications and corresponding
treatments since certain patients still have a poor clinical outcome despite intensive treatment. Therefore,
a better subtyping of WT and other renal tumours is mandatory. Such developments can only be achieved
through the design of biology driven approaches.
The Wilms Oncosimulator is a clinically oriented simulation model that has been developed by the In
Silico Oncology and In Silico Medicine Group (ICCS- National Technical University of Athens) [2,3,4,5].
Its design and implementation is based on SIOP trial protocols in order to provide personalized
modelling of the growth of nephroblastoma tumours and their response to pre-operative combined
chemotherapeutic schemes of Actinomycin-D and Vincristine. In order to produce more accurate
predictions a variety of multiscale data are required to create a multiscale tumour profile utilizing
personalized patient data such as imaging, molecular, histological and clinical data. The model integrates
several cellular biological features and phenomena such as the varying mitotic potential of tumour cells,
the categorization of tumour cells into stem cells, limited mitotic potential (progenitor) cells,
differentiated cells and dead cells; the cycling of cells into the the cell cycle phases; the mitotic division
of cells; the symmetric and asymmetric modes of stem cell division; the mechanisms of cell death through
spontaneous apoptosis and necrosis; the effect of chemotherapy drugs; the neutralization of proliferating
cells and their transition to the dormant phase due to eventually inadequate supply of oxygen and
nutrients and the eventual differentiation of tumour cells. A combination of discrete entity-discrete event
with continuous mathematics has been adopted for the mathematical and computational modelling needs.
2.II Nephroblastoma Use Case ( NEPH – UC) Scenarios through the MyHealthAvatar Platform
The Wilms Oncosimulator has been integrated into the MHA platform and two use case scenarios have
been formulated: the educational scenario and the clinical one. The outcome of the integration of the
Wilms Oncosimulator into the MHA platform is to provide a tool that produces the ‘in-silico profiling’ of
a nephroblastoma patient and performs ‘in-silico’ predictions of a given therapeutic scheme outcome. The
nephroblastoma use case in the framework of MyHealthAvatar targets patients and parents of patients,
paediatric oncologists, researchers and the general public. The “Nephroblastoma clinical scenario” has
been developed and demonstrated through IAPETUS implemented by the In Silico Oncology and In
Silico Medicine Group. It features a modular designing apporach [6] and contains a tool/model repository
and interfaces for interacting with stored models. The particular use case has been implemented in order
to facilitate the patient/patient’s parent – doctor interaction. The latter is a right way through which
sensitive patient data and the model execution outputs can be viewed and implemented while at the same
time results interpretation by the clinical expert, guidance and advice are provided. For the needs of this
scenario the Oncosimulator operates on synthetic data acquired from an external repository of the
European Commission funded project CHIC (FP7-ICT-2011–9-600841) and produces a simple set of
results which are stored back to the MHA platform. Since MHA is also targeted at citizens, it can also
inform the latter about what VPH multiscale models are capable of doing and how they can contribute to
the battle against severe diseases such as cancer. Therefore, a “Nephroblastoma educational scenario” has
been formulated and integrated within the MHA platform.
3. Results
As a proof of concept the two scenarios of the nephroblastoma use case have been applied to synthetic
Wilms tumour patients treated with combined pre-operative chemotherapy of actinomycin-D and
vincristine. Patients’ anonymized data have been collected and provided to the modelling team in the
framework of the CHIC project by the Institute of Pathology, University Hospital of Saarland, and the
Saarland Tumour Centre, Germany in order to form a multiscale tumour profile of the synthetic patient.
Georgiadi et all. VPH2016 book of abstracts
298 Personalised Medicine II

3.1 Educational scenario: As an indicative example of the added value of the nephroblastoma use case
educational scenario implemented at the MHA platform, a hypothetical scenario of postponing the
beginning of chemotherapy for one week and presenting the effect on the disease progress is addressed
and presented. Two treatment schemes have been simulated on the same synthetic patient. Both treatment
schemes are in accordance with the SIOP 2001/GPOH clinical trial protocol of preoperative
chemotherapy with a combination of actinomycin-D and vincristine for stage I-III nephroblastoma
tumours. The first simulated scheme begins one day whereas the second one eight days after the diagnosis
of the tumour. The results of the two simulations presented in Fig.1 depict the great effect of the delay at
the start of the pre-chemotherapy on the final chemo-induced reduction of the tumour.
Fig. 1 (↓) NEPH- UC Education Scenario example: A. Simulated effect of the SIOP 2001/GPOH clinical trial
protocol of preoperative chemotherapy administered 1 day after diagnosis. B. Simulated effect of the SIOP
2001/GPOH clinical
trial protocol of
preoperative
chemotherapy
administered 8 days
after diagnosis. Grey
colour: initial tumour,
blue colour: final
tumour.
3.2 Clinical Scenario:
As an indicative
example of the
nephroblastoma use
case clinical scenario
implemented at the
MHA platform, a
hypothetical scenario
of simulating a certain
delay in one
administration time
point and shifting or
not shifting the delay
to the rest of the
administration time
points is presented. The added value for the MHA platform as a tool for the physician when treating a
patient with nephroblastoma is depicted. This refers to both predicting how this specific nephroblastoma
tumour will respond to preoperative treatment with vincristine and actinomycin-D and performing
multiple simulations of hypothetical treatment schemes. Three treatment schemes have been simulated on
the same patient. The standard scheme, a scheme with two days delay at the third administration time
point and a shift of the delay to the fourth administration time point and a scheme with two days delay at
the third administration time point and no shift of the delay to the forth administration. The results of the
three simulations presented in Fig 2 show that an overall shift of the rest of the administrations due to a
potential delay in one administration time point will not result to better chemo-induced tumour reduction.
A detailed result interpretation by the clinician as well as their guidance and advice are very important.
4. Discussion
Integrating the Wilms Oncosimulator into the MyHealthAvatar ( MHA) platform offers the opportunity
to demonstrate the platform capabilities for assisting a third party application by providing their own
collected data and/or access to external data sources. The latter may aim at training, adapting and
validating simulation models under development or feeding already clinically utilized models and
Georgiadi et all. VPH2016 book of abstracts
299 Personalised Medicine II

providing actual predictions for patients whose data may reside outside the MyHealthAvatar platform.
Storing the execution results back to the platform indicates that the latter can further add to its collected
data by incorporating outputs from simulation model executions which in turn can support the function of
other modules within the platform or other models inside or outside the platform. Education-wise,
partially validated models simulating a hypothetical patient’s disease progress and running through a
platform such as the MyHealthAvatar one are suitable for general public education. The nephroblastoma
educational scenario can be used to inform the patient and/or their parents of the new methods that are
used to individualize the treatment scheme. By extending this notion, the general public can be informed
that simulation models can mimick the basic biological mechanisms, disease progression, and response to
treatments (health literacy). The Nephroblastoma Clinical Educational scenario can be used to
demonstrate how in-silico models provide the ability to simulate the individualized nephroblastoma
patient response to treatment and following the completion of a strict prospective clinical validation how
an in silico model might serve as a prototype clinical decision support system able to enhance
personalized medicine. It also provides clinical researchers and modellers alike a powerful tool to define
an in silico patient profile and further exploit it in other modelling approaches and VPH related
endeavours.
Figure 2 () : NEPH-UC A
clinical scenario example
implemented on the MHA
platform : A. Simulated effect
of the SIOP 2001/GPOH
clinical trial protocol of
preoperative chemotherapy on
a specific patient B.
Simulated effect of the SIOP
2001/GPOH clinical trial
protocol of preoperative
chemotherapy with a 2 day
delay on the third and the
fourth administration time
points. C. Simulated effect of
the SIOP 2001/GPOH clinical
trial protocol of preoperative
chemotherapy with a 2 day
delay on the third
administration time point.
4.1 Acknowledgements
This work has been supported in
part by the European
Commission under the projects
Computational Horizons in
Cancer (CHIC): Developing
meta- and Hyper-multiscale Models and Repositories for In Silico Oncology (FP7-ICT-2011–9-600841) and MyHealthAvatar: a
Demonstration of 4D Digital Avatar Infrastructure for Access of Complete Patient Information (FP7-ICT-2011–9–600929). The
support of Dr D. Dionysiou, ICCS-National Technical University of Athens is duly acknowledged.
5. References
[1] Graf N, et al,, Urologic Clinics of North America,2000; 27:443-454
[2] Stamatakos G et al., IEEE Journal of Biomedical and Health Informatics. 2014;18(3):840-854
[3] Georgiadi E et al., Computers in Biology and Medicine. 2012;42(11):1064-1078
[4] Stamatakos G et al., PLoS ONE. 2011;6(3):e17594
[5] N. Graf et al.,2009; Klin. Paediatr. 221, 141-149
[6] Christodoulou N and G. Stamatakos G., Proceedings of the 2014 6th International Advanced Research Workshop on In Silico
Oncology and Cancer Investigation - The CHIC Project Workshop (IARWISOCI). 2014. doi:10.1109/iarwisoci.2014.7034640,
pp.52-55
Georgiadi et all. VPH2016 book of abstracts
300 Personalised Medicine II