Chemotherapy and Cancer Modelling

Chemotherapy and Cancer Modelling

Organised By

Carl Panetta & David Cameron

 

Accelerated chemotherapy - do we know why it works?

David Cameron
Western General Hospital, UK
dcameron@srv0.med.ed.ac.uk

 

Following theoretical modelling using the Gompertz function by Larry Norton and others, there has been increasing interest in the Clinical Oncology community for giving dose-dense chemotherapy. One of the main obstacles to curing cancer is the limitation in dose imposed by the tolerance of the normal tissues, such as the bone marrow. Attempts had been made to overcome this with high dose chemotherapy and bone marrow or stem cell transplantation. This approach has not proved very succesful in solid cancers. Further work has suggested that an alternative strategy is to give the chemotherapy more frequently with GCSF to assist in a ore rapid recovery of the bone marrow. Theoretical modelling has predicted that it would be better in the adjuvant therapy of breast cancer, and recently early data from the CALGB co-operative group in the USA has confirmed this. However, in my hands, modelling based on Gompertz does not necessarily show an advantage, particularly when one considers that the benefit seen with adjuvant therapy for many patients is not due to cure, but delayed recurrence. This will be discussed, and a challenge issued to modellers to get involved in this field!


Integrating Knowledge Engineering with Cancer Modelling

Roger Day
University of Pittsburgh, USA
day@upci.pitt.edu

Coauthor(s): William E. Shirey, Michele Morris

In cancer research, clinical testing is a permanent bottleneck. Due to sample size limitations, few of the promising ideas can be tested clinically, so clinical questions must be chosen carefully. A principal investigator designing a new cancer clinical study for the chosen question must exercise ``due diligence''. This comes in two parts: gathering relevant information from prior studies, and using that information judiciously to make study design choices. The bridge from the first part to the second is often the formation of mental model of the disease from the literature review. This is not so easy; only a small subset of the ballooning treasury of human knowledge about cancer can be grasped by one person, resulting in the neglect of relevant facets, while mental modelling may suffer from poor intuition about the behavior of complex systems.

To augment the usual informal thought experiments, the OncoTCap cancer software workbench is designed to support, integrate and document the information-gathering and modelling tasks. An application to lung cancer is demonstrated.

 


Simple mathematical models reliably yet indiscriminately
reproduce the outcome of breast cancer chemotherapy trials

Eliezer Shochat
Weizmann Institute of Science, Israel
shochate@wisdom.weizmann.ac.il

Coauthor(s): David Cameron and Sallo Stemmer

In clinical practice the derivation of a new chemotherapy protocol is a very tedious, time consuming and often-futile process. It is thus often questioned whether mathematical simulations coupled with experimentally derived parameters of the physical phenomena may assist to planning and predicting the course of cancer treatment. We apply mathematical modeling to simulate the course and outcome of some of the recent breast cancer chemotherapy trials. Several simple phenomenological modes of cancer growth were simulated in the study. In addition, a detailed model that explicitly incorporates the biology of the vasculature and growth factors into the tumour dynamics was derived. Each model represents an unperturbed tumour growth, superimposed by periods of treatment applications. To estimate the treatment effect we have simulated the survival data of cohorts of patients using plausible distributions of model parameters (a Monte-Carlo procedure). The simulated survival (Kaplan-Mayer) curves derived by ALL of the models are similar to what may be clinically observed. These results caution against a naive application of any particular model in clinical trials and reiterate the need of specific parameter estimation in individual patients. Some basic notions of mathematical modelling in medicine will be discussed.