4. Monte Carlo Workshop

Program

Time		Talk

9.00-10.00		Annette Kopp-Schneider (DKFZ, Heidelberg)
		Statistics in Toxicology: Using biologically-based stochastic models to describe the carcinogenic process and using statistic models for dose-response analyses
10.00-11.00		Meinhard Kieser (U Heidelberg)
		Cancled due to illness

11.00-11.30		Coffee Break

11.30-12.00		Geraldine Rauch (U Heidelberg)
		Application of Monte-Carlo Simulations when Planning Clinical Trials with Composite Endpoints

12.00-13.00		Lunch Break

13.00-14.00		Harald Paganetti (MGH, Boston)
		The use of Monte Carlo methods in proton radiation therapy of cancer

14.00-15.00		Martin Soukup (Elekta CMS Software)
		Elekta: Monte Carlo in radiotherapy treatment planning

15.00-15.30		Coffee Break

15:30-16.00		Katia Parodi (U Heidelberg)
		Monte Carlo modeling to support high precision ion beam therapy

Abstracts

Annette Kopp-Schneider (DKFZ, Heidelberg)
Statistics in Toxicology: Using biologically-based stochastic models to describe the carcinogenic process and using statistic models for dose-response analyses
The talk will be divided into two parts. The connection between the two parts is the use of models in the field of toxicology. In the first part, stochastic models will be presented which are used to describe the process of formation and growth of premalignant and malignant lesions. There are several reasons for formulating models of carcinogenesis. One is to elucidate the biological process of carcinogenesis and another is to provide a rational basis for risk assessment, e.g. the extrapolation of dose-response curves of environmental agents down to a range of interest for regulating agencies. In addition, carcinogenesis models can be used to analyse the mode of action of carcinogens. Data from classical carcinogenesis experiments report rates of tumor bearing animals in the experimental groups or time to tumor for individual animals. More recent studies have focussed on preneoplastic endpoints as these occur earlier in life and their occurrence is less harmful for the animals. Stochastic models of carcinogenesis describe both the carcinoma endpoint and the number and sizes of preneoplastic lesions. Monte-Carlo-methods are used to verify theoretical derivations and to investigate model properties too complex for analytic solution. The second part of the talk addresses the use of statistical models, specifically the log-logistic model, for dose-response analyses of in vitro experiments. Log-logistic dose-response modelling was used in the statistical analysis of the ACuteTox project, an EU-FP6 project aimed at developing a simple and robust in vitro testing strategy for prediction of human acute systemic toxicity, which could replace the animal acute toxicity tests used nowadays for regulatory purposes. In the ACuteTox project, 57 chemicals were tested in 71 in vitro assays. Dose-response experiments were performed and a characteristic value such as the EC50 were derived. Using the characteristic values for all assay x chemical combinations, a subset of assays was identified which is able to classify the given set of chemicals into five toxicity categories.
Geraldine Rauch (U Heidelberg)
Application of Monte-Carlo Simulations when Planning Clinical Trials with Composite Endpoints
Composite endpoints combine several events of interest within a single variable in order to increase the number of expected events. They are of particular interest in the field of cardiologic trials. To demonstrate the significance of an overall clinical benefit, it is thus sufficient to assess the test problem formulated for the composite. However, even if a statistically significant and clinically relevant superiority is shown for the composite endpoint, there is the need to evaluate the treatment effects for the constituting components. For example, the Points to Consider on Multiplicity (2002) require that “… if all cause mortality is a component, a separate analysis of all cause mortality should be provided to ensure that there is no adverse effect on this endpoint.” In the ICH E9 Guideline, it is stated that the use of composite endpoints addresses the multiplicity problem without requiring adjustment to the type I error (ICH, 1999). However, if the effects of single endpoints are tested additionally, this results in a multiple test problem.
The single endpoints constituting a composite are usually correlated. This correlation structure can be used to construct a multiple test procedure that controls the overall type I error and avoids at the same time a too strict adjustment of the local significance levels. However, even if fixed correlations between the single endpoints are assumed, the correlation structure between the test statistics is not obvious. We use Mont-Carlo simulations to address this problem in the planning stage. Application is illustrated by a clinical trial example.
Harald Paganetti (MGH, Boston)
The use of Monte Carlo methods in proton radiation therapy of cancer
Monte Carlo simulations are a powerful tool in radiation oncology. They are being used in designing treatment heads or radiation detectors. Further, various research projects are relying on Monte Carlo simulations to characterize the radiation field. This presentation does focus on two key applications of Monte Carlo in proton radiation therapy: Precise dose calculation in patients for radiation therapy treatment planning and the simulation of the health risk of secondary radiation to the patient outside of the treatment area.
Katia Parodi (U Heidelberg)
Monte Carlo modeling to support high precision ion beam therapy
The application of ion beams (from protons up to carbon ions) to external beam radiotherapy is rapidly increasing worldwide. The main rationale is the favorable ionization energy-loss of swift charged ions in matter, resulting in the characteristic dose maximum at the end of range known as Bragg-peak. Proper superimposition of several Bragg-peaks enables optimal conformation of the delivered dose to the tumor, with better sparing of surrounding healthy tissue in comparison to conventional photon and electron radiation. However, full clinical exploitation of the physical advantages requires millimeter accuracy in the localization of the beam stopping point and lateral field position in human tissue. This demands precision range measurements in representative tissue-equivalent materials as well as accurate calculation tools for realistic description of the electromagnetic and nuclear interactions of ion beams in matter. Besides, non-invasive imaging techniques for in-vivo verification of the actual beam delivery and, in particular, of the beam range in the patient would be highly beneficial. This talk will address two emerging areas of application of Monte Carlo to promote improved accuracy of ion beam therapy in clinical practice. It will first illustrate the value of Monte Carlo modeling to support treatment planning, showing examples of applications to proton and carbon ion therapy at Massachusetts General Hospital (MGH), USA, and at the Heidelberg Ion Beam Therapy Center (HIT), Germany. These include the generation of basic data required by the treatment planning system (TPS) and verification of the TPS analytical dose computation in critical cases, e.g., in the presence of tissue inhomogeneities and metallic implants. Moreover, it will review the application of Monte Carlo to support the first clinical study on the usage of post-treatment PET/CT (Positron-Emission-Tomography / Computed- Tomography) imaging for in-vivo verification of proton therapy at MGH. Due to the different electromagnetic and nuclear processes underlying dose deposition and +-activation, verification of the delivered treatment is achieved by comparing the measured irradiation- induced activity distribution with a detailed Monte Carlo calculation. Finally, an overview of ongoing Monte Carlo research at the HIT center will be given.