In Silico Oncology: Building and Validating Oncosimulators ... · 8/29/2019 · crucial clinical...

G.S. Stamatakos - Lecture at the Korea Institute of Science and Technology 1

Georgios S. Stamatakos

In Silico Oncology and In Silico Medicine Group,

Institute of Communication and Computer Systems,

School of Electrical and Computer Engineering,

National Technical University of Athens, Greece

& Medical School, University of Saarland, Germany

https://www.in-silico-oncology.iccs.ntua.gr/

29 Aug. 2019

In Silico Oncology:

Building and Validating Oncosimulators and Oncosimulator Based

Hypermodels as Clinical Decision Support Systems







Acknowledgements

• Prof. Norbert Graf is greatly acknowledged for the clinical drive, the clinical positioning, the provision of

crucial clinical data and the clinical overview of the work concerning nephroblastoma modelling, an

excellent Oncosimulator development paradigm.

• All my collaborators at the In Silico Oncology & In Silico Medicine Group, (ISO&ISM_G) ICCS, SECE,

NTUA are greatly acknowledged for their enthusiasm, commitment and hard and efficient work. Special

thanks are due to : Dr D. Dionysiou, Dr V. Antipas, Dr E. Kolokotroni, Dr E. Georgiadi, Dr S. Giatili, Dr E.

Ouzounoglou, Ms K. Argyri, Mr N. Christodoulou, Mr C. Antonopoulos, Mr C. Kyroudis, and Mr N.

Tousert.

• Prof. Uzunoglu is duly acknowledged for his crucial encouragement and support during the initial steps

of the endeavour.

• All partners of the 17 organizations who participated in the European Commission (EC) funded EC-US

project CHIC as well all partners involved in the Oncosimulator & Hypermodelling development and

their clinical adaptation and validation for the past 22 years are greatly acknowledged for their

important contributions.

• All partners involved in the Oncosimulator development and validation of the European Commission

funded projects ACGT, ContraCancrum, TUMOR, p-medicine, Dr Tharapat, MyHealthAvatar are duly

acknowledged.

• All external collaborators of ISO&ISM_G since 1997 are duly acknowledged.

• The European Commission, the Greek and the German States are duly acknowledged for their crucial

financial support

29 Aug. 2019 G.S. Stamatakos - Lecture at the Korea Institute of Science and

Technology 2

The CHIC Project at a

glance

29 Aug. 2019 G.S. Stamatakos - Lecture at the Korea Institute of Science and Technology

3

The CHIC Project at a glance

• The large scale EU-US integrating research project CHIC has been entitled:

“CHIC: Developing meta- and hyper-multiscale models and repositories for in silico oncology”

• Website: http://www.chic-vph.eu/ • Funded by the European Commission with a grant of

10,582,000 €.

• Seventeen academic, research and industrial partner organizations across Europe and US participated in CHIC.

• The CHIC project underwent its final review on 23 and 24 May 2017 and was assessed as "Excellent" by the Board of (five) External Reviewers appointed the European Commission.


4

http://www.chic-vph.eu/




The CHIC Project at a glance (cont.)


5

29 Aug. 2019 G.S. Stamatakos - Lecture at the Korea

Institute of Science and Technology 6



• THE CHIC PROJECT COORDINATION

SCHEME

• Overall and Scientific Coordinator: Research

Professor G. Stamatakos, ICCS-National

Technical University of Athens, Greece

• Assistant Clinical Coordinator: Professor

Norbert Graf, University Hospital of Saarland,

Germany

29 Aug. 2019

G.S. Stamatakos - Lecture at the Korea Institute of Science and Technology

7

Structure of the presentation

• A brief outline of the purpose, methods and

results

• Examples from the methods and the results

• Conclusions





A brief outline of the purpose,

methods and results

Purpose of CHIC

• to develop, clinically adapt and partly clinically

validate meta- and hyper-multiscale models and

repositories for in silico oncology

• to develop advanced technological cloud based

infrastructures supporting the process of

hypermodel development and the clinical

translation of hypermodels



Methods

• A host of clinical, experimental, mathematical, computational and software engineering strategies, methods and techniques have been devised and/or utilized in order to both develop and test multiscale hypermodels.

• A hypermodel is a complex mathematical and computational model consisting of more than one elementary component model.

• Each component model or “hypomodel” simulates a crucial biological mechanism of tumour growth and response to treatment.



Methods (cont.)

• Hypomodels are connected together in

several ways dictated by the current biological

and clinical knowledge.

• Both mechanistic and machine learning

based hypermodels have been developed

driven by clinically relevant questions

formulated by the clinical partners of the

CHIC consortium.



Methods (cont.)

• The overarching idea of the project was to exploit the accumulated quantitative experimental and clinical knowledge concerning several spatiotemporal scales of cancer biocomplexity in order to produce treatment response predictions as precise as possible based on the patient’s individual multiscale data (e.g. – imaging

– Histological

– molecular,

– Clinical

data



Methods (cont.)

• To this end several candidate treatment schemes can be simulated using detailed hypermodels fed with the actual multiscale data of the patient.

• The treatment scheme performing best in silico will serve as the optimal suggestion to the clinician to consider for their final treatment strategy decision.

• Most hypomodels or component models have been developed by different leading cancer modelling groups participating in the CHIC project scattered across EU and US.



Methods (cont.)

• A clinician friendly technological platform for hypermodel creation and execution (CRAF) has also been developed and successfully tested.

• Four paradigmatic cancer types have been considered:

– nephroblastoma,

– non small cell lung cancer

– glioblastoma (treated with immunotharepy in conjunction with radiotherapy and chemotherapy)

– prostate cancer.

29 Aug. 2019


15

Results

• Both the hypermodels and the technological platforms developed by CHIC have been documented, disseminated and demonstrated in real time and in detail to the appointed independent scientific evaluators of the European Commission.

• The overall project outcome has been finally assessed as Excellent and worth further translational development and multifaceted exploitation.





Examples from the methods

and the results

Dimensions of cancer manifestation and treatment 29 Aug. 2019


18

The Oncosimulator: a functional diagram





Basic architecture of a cancer

multimodeller hypermodel



Mathematics hidden behind each

constituent hypomodel



Nephroblastoma



Nephroblastoma

(Part of the whole table of diagrams / nephroblastoma )



Nephroblastoma

Multiscale Cancer Modelling Paradigms



The Wilms Tumour Branch of the

Oncosimulator





Chemotherapy treatment protocol. The simulated Wilms Tumour preoperative

chemotherapy treatment protocol of the SIOP/ GPOH clinical trial.

• Wilms Tumour Oncosimulator: – Tumor Free Growth - Tumor Chemotherapy



Cytokinetic Model for

Free Tumour Growth

• stem cells: cells assumed to possess unlimited proliferative potential

• limp cells: progenitor cells with limited proliferative potential

• diff cells: terminally differentiated cells

Cell Local reoxygenation Local reoxygenation

Cell Disappearance

Apoptosis

Spontaneous apoptosis

Necrosis

disappearance

G0 G1 S G2 M G

G0 M G2 S G1

Asymmetric division

DIFF STEM LIMP

Symmetric division


After n mitoses



Cytokinetic Model Treatment Response

• When cells are hit by chemo (treatment session) they enter a separate cell cycle at

which they remain till they are led to apoptotic death from a point of the cell

cycle specified by the mechanism of action of the drug (in the case of Epirubicin S

phase is considered to be that point).

chemo

G1hit Shit G2hit Mhit

Cell disappearance

G0hit

A (Apoptosis incl.

time delay)


N (Necrosis)

Cell

disappearance

G0 G1 S G2 M G0

chemo

Mhit G2hit Shit G1hit

M G2 S G1

G0hit

Asymmetric Division

DIFF

STEM LIMP



Clinical Adaptation and Validation



Successful Clinical Model Adaptation

• Case 1: [1]

Highly malignant, blastemal type of tumor

Time evolution of tumor volume and selected tumor

subpopulations. Panel A: Time evolution of tumor volume

for the four virtual scenarios of Table 1. Panels Bi and Bii,

Ci and Cii, Di and Dii, Ei and Eii: Evolution over time of

selected subpopulations of the tumors.The

chemotherapeutic scheme of Figure 2 has been

simulated. The drug administration instants are: day 3, day

10, day 17, day 24. Day 0: first MRI data set. Day 28:

second MRI data set.



Adaptation of Several Clinical Cases

No Patient Histology Risk

1 11570 Mixed Intermediate

2 11590 Mixed with focal anaplasia Intermediate


4 11628 Stromal Intermediate

5 11639 Regressive Intermediate







12 11733 Blastemal High





17 11845 Diffuse Anaplasia High

18 11862 Epithelial Intermediate





Imaging & Clinical Data



Nephroblastoma Bilateral Case

Imaging Data

1st Imaging Set 2nd Imaging Set 3rd Imaging Set

R L R R L L

DVR=90%

DVL=89%

DVR=94%

DVL=95%



Nephroblastoma Case Clinical Data

Histopathological data

• Nephroblastomatosis consists primarily of blastemal

cells which are actively cycling.

• Therefore the initial tumour is made up mainly of stem

and LIMP (progenitor) cells and fewer differentiated and

dead cells.

•Post-surgery histological data also indicated that the

remaining viable tumour was of blastemal type.



Simulation Results

Time evolution of bilateral tumour volume and selected tumour subpopulations. A: Time evolution of tumour volume for

the right and left kidney under the two scenarios of table 1. B, C, D, E: Evolution over time of the proliferating, dormant,

differentiated and dead population percentage of the bilateral tumor (respectively). Where: R: Right, L:Left, TT: Typical

tumour, CT: clinical tumour.

29 Aug. 2019 G.S. Stamatakos - Lecture at the Korea Institute of Science and Technology 38

Simulated Clinical Tumors

1st I.S.

2nd

I.S.

3rd I.S.

Right Kidney Left Kidney



29 Aug. 2019 40

AN EXAMPLE OF USING THE GBM

RADIOTHERAPY ONCOSIMULATOR TWO RTOG STUDY 83-02 BRANCHES SIMULATED

• 1) AHF-48Gy:

accelerated hyperfractionation, 48Gy total dose,

(1.6Gy twice daily to a total dose of 48 Gy)

• 2) HF-81.6Gy:

hyperfractionation, 81.6Gy total dose.(1.2Gy twice

daily to a total dose of 81.6Gy)


29 Aug. 2019 41 G.S. Stamatakos - Lecture at the Korea Institute

of Science and Technology

An MRI slice depicting a glioblastoma mutiforme. Both the clinical volume of the tumour and its central necrotic area have been delineated. The present case has been considered for the preliminary checks of the simulation model. [G.Stamatakos, D.Dionysiou, E.Zacharaki, N.Mouravliansky, K.Nikita, and N.Uzunoglu, "In Silico Radiation Oncology: Combining Novel Simulation Algorithms with Current Visualization Techniques,'' Proc. IEEE, Special Issue on "Bioinformatics: Advances and Challenges" Vol.90, No.11, November 2002, pp.1764-1777]

29 Aug. 2019 42 G.S. Stamatakos - Lecture at the Korea

Institute of Science and Technology

Spatial Discretization

29 Aug. 2019 43 G.S. Stamatakos - Lecture at the Korea Institute of Science and Technology

Mesh Initialization

NBC

Equivalence Classes

29 Aug. 2019 44 G.S. Stamatakos - Lecture at the Korea Institute

of Science and Technology

Cytokinetic Model

Cell disappearance

G1 G2 S

M

G0 N A

RI-MAD

RI-MND

SA or RI-ID

The probabilities of the alternative “death paths” due to irradiation depend

primarily on the type of tumour cell.

• In GBM the vast majority of cells undergoes a mitotic necrotic death. SA: Spontaneous Apoptosis, RI-ID: Radiation-Induced Interphase Death, RI-MAD: Radiation-

Induced Mitotic Apoptotic Death, RI-MND: Radiation-Induced Mitotic Necrotic Death

Simplified flow chart for the response of a single tumour cell to irradiation. Symbol explanation: αP and βP stand for the α and β parameters of the linear quadratic model for the tumour proliferating cells excluding those in phase S. The subscript S denotes cells in the DNA synthesis phase, whereas the subscript G0 denotes cells in the resting (dormant) phase G0.

YES NO

Irradiation (αP,βP ) (αS,βS )

Cell still cycling for a few

(e.g. 3) cell cycles

Cell lysis/apoptosis

PROLIFERATING CELL

Irradiation (αG0,βG0 )

G0- CELL

NO

Has oxygen and nutrient

supply become

adequate?

LQ cell hit

Cell disappearance Tumor shrinkage

Cell death products are diffused

LQ cell survival LQ cell hit

Cell is gradually

disintegrating

LQ cell survival

YES

Is oxygen and nutrient

supply still adequate?

[from G.Stamatakos, D.Dionysiou, E.Zacharaki, N.Mouravliansky, K.Nikita, and N.Uzunoglu, "In Silico Radiation Oncology: Combining Novel Simulation Algorithms with Current Visualization Techniques,'' Proc. IEEE, Special Issue on "Bioinformatics: Advances and Challenges" Vol.90, No.11, November 2002, pp.1764-1777]

29 Aug. 2019 G.S. Stamatakos - Lecture at the Korea Institute of Science

and Technology 45

Irradiation according to the standard fractionation scheme (2 Gy once a day, 5 days per week, 60 Gy in total). Left panel: three dimensional sections of the tumour shown in the right panel: (a) before the beginning of irradiation, (b) 1 fictitious day after the beginning of irradiation, (c) 2 fictitious days after the beginning of irradiation and (d) 3 fictitious days after the beginning of irradiation. Colour code red: proliferating cell layer, green: dormant cell layer (G0), blue: dead cell layer. The colouring criterion “99.8%” used to visualize the predictions has been defined as follows. “For a geometrical cell of the discretizing mesh, if the percentage of dead cells is lower than 99.8% then { if percentage of proliferating cells > percentage of G0 cells, then paint the geometrical cell red (proliferating cell layer), else paint the geometrical cell green (G0 cell layer) } else paint the geometrical cell blue (dead cell layer)” The values of certain parameters (e.g. cell loss) have been deliberately exaggerated in order to facilitate the demonstration of the ability of the model to simulate the shrinkage effect. [see G.Stamatakos, D.Dionysiou, E.Zacharaki, N.Mouravliansky, K.Nikita, and N.Uzunoglu, "In Silico Radiation Oncology: Combining Novel Simulation Algorithms with Current Visualization Techniques,'' Proc. IEEE, Special Issue on "Bioinformatics: Advances and Challenges" Vol.90, No.11, November 2002, pp.1764-1777]

(a)

(b)

(c)

(d)

29 Aug. 2019 G.S. Stamatakos - Lecture at the Korea Institute of Science and

Technology 46

29 Aug. 2019 47 G.S. Stamatakos - Lecture at the Korea Institute of Science and

Technology

Simulation predictions of the number of total tumour cells (mt p53 and wild type p53) for the standard fractionation scheme. An OER=3.0 has been assumed.

[see V. P Antipas, G. S Stamatakos, N. K Uzunoglu, D. D Dionysiou, R. G Dale, ” A spatio-temporal simulation model of the response of solid tumours to radiotherapy in vivo: parametric validation concerning oxygen enhancement ratio and cell cycle duration,” Phys. Med. Biol. 49 (2004) 1485–1504 [Pubmed Link: http://www.ncbi.nlm.nih.gov/entrez /query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstract&list_uids=15152687&query_hl=14] ]


TWO RTOG STUDY 83-02

BRANCHES SIMULATED

• 1) AHF-48Gy:

accelerated hyperfractionation, 48Gy total dose,

(1.6Gy twice daily to a total dose of 48 Gy)

• 2) HF-81.6Gy:

hyperfractionation, 81.6Gy total dose.(1.2Gy twice

daily to a total dose of 81.6Gy)


Number of surviving tumour cells as a function of time for a glioblastoma

tumour with mutant p53 gene. AHF-48Gy: accelerated hyperfractionation,

48Gy total dose, HF-81.6Gy: hyperfractionation, 81.6Gy total dose.

29 Aug. 2019 50

4D (3D + time)

visualization

AHF-48Gy HF-81.6Gy

GBM with mutant p53

1.0E+00

1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

1.0E+08

1.0E+09

1.0E+10

1.0E+11

0 1 2 3 4 5 6 7 8

Time (weeks)

Nu

mb

er

of

alive t

um

ou

r cells ...

AHF- 48Gy

HF- 81.6Gy


29 Aug. 2019 51

Interactive 2D sampling planes

AHF-48Gy HF-81.6Gy


CHIC SURVEY ON HYPERMODELS

• CHIC is running a survey, where patients, physicians and citizens can learn about hypermodels and can give their opinion on the usefulness of such models.

• Your feedback will help us to optimize our research results.

• The survey is available at http://www.chic-vph.eu/ Latest Highlights

or directly at http://chic-vph.eu/highlights/details/article/chic-online-survey-on-hypermodels/

• A video demonstrating the future use of hypemodels is also included in the survey

• Responsible: Prof. Norbert Graf, University Hospital of Saarland








http://chic-vph.eu/highlights/details/article/chic-online-survey-on-hypermodels/





































CHAPTER 18


65 29 Aug. 2019

Conclusions

• Based on the partial validation results and analyses that have been reported in CHIC, the highly innovative CHIC hypermodels and Oncosimulators appear to possess a great potential for serving as clinical decision support systems (CDS) and/or cores of future in silico trial platforms.

• However, additional retrospective validation work for the developed hypermodels and Oncosimulators is needed in order to more fully substantiate and support their “candidacy” for undergoing validation through prospective clinical trials.



Conclusions ( cont.)

• This is a necessary step in order to definitely assess both their clinical validity and clinical value.

• Further retrospective validation work will be carried out by specific former CHIC partners on a bilateral or small partner group basis.

• Regarding the eventual prospective clinical validation of the hypermodels, certain exploratory steps have already been taken, including focused discussions within the framework of the International Society for Pediatric Oncology (SIOP).

29 Aug. 2019


67

The BOUNCE Project

• In the context of exploitation, it is noted that

several approaches, processes, models and

tools developed in the framework of the

CHIC project have already been recruited for

the implementation needs of the EU funded

project BOUNCE under the title: “Predicting

Effective Adaptation to Breast Cancer to Help

Women to BOUNCE Back” (Grant

Agreement 777167)





Thank you

Date post:	22-Jun-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

In Silico Oncology: Building and Validating Oncosimulators ... · 8/29/2019 · crucial clinical...

Documents