Non-stationary statistical modeling of circadian oscillations for research in cancer chronotherapy

Oscillators are found on all levels in Nature. One of the most prominent examples is the circadian rhythm which displays an endogenous, entrainable oscillation of about 24 hours. Chronotherapy is a form of personalized therapy where treatments are administered according to a schedule that corresponds to a person's rhythms in order to maximize effectiveness and minimize side effects of the therapy. There is now large evidence both in experimental models and in patients that the Circadian Timing System (CTS) of patients is affecting treatment tolerability and efficacy of anticancer drugs while the progression of the disease and treatment with anticancer drugs have disruptive effect on the CTS. Our projects focuses on the development of statistical methods that address the modeling of such time varying effects. An important aim is to develop a real-time prediction system of a patient's CTS state in order to achieve an improved personalized chronotherapy.

There is now large evidence both in experimental models and in patients that the Circadian Timing System (CTS) of the patients is affecting treatment tolerability and efficacy of anticancer drugs while, on the other hand, the development of the disease and the treatment may have a disruptive effect on the CTS. For example, Costa et al. (2013) found a significant shift in period length during anticancer treatment suggesting that the patient's daily rhythm is altered while receiving anticancer drugs. Hence a personalized chronotherapy needs to take into account the potential effect of the treatment as well as the disease on the CTS of the patient. The aim of this project is develop a statistical methodology that is able to detect and quantify (in real time) such temporal variation in the circadian oscillation and that can be used to study how these are associated with changes in covariates (tumour growth, treatment with drugs and timing of their administration, external factors such as light, temperature, food and other potential Zeitgebers). We address this through non-stationary statistical modeling with the aim of developing a system for online prediction of a patient's CTS. Using spectrum and state-space approaches we will develop non-stationary statistical methods using data from all available biomarkers to construct a robust online prediction of the current state of a patient's CTS that can be used to individualize the timing of chronotherapy. Secondly, we will enhance a mechanistic mathematical modeling approach to gene expression and provide a statistical algorithm of fitting such a model to circadian gene expression data with the aim of gaining a better understanding of how the underlying gene regulatory processes might be affected by the disease and other covariates.

This programme will develop some methodologies that will make it possible to analyze the impact of changing conditions on the circadian system in living organism. The PI has been involved for over 10 years in interdisciplinary research in systems biology and has extensive experience with the analysis of circadian data and the modeling gene expression. CI 1 Prof. Francis Levi is a world expert on chronotherapy in cancer who will direct the scientific content of our project while CI 2 will bring essential domain expertise concerning mathematical modeling in systems biology, particularly circadian systems.

We envisage an impact of this research on

(i) Statisticians working to provide appropriate statistical methods for systems biology and medicine and the statistical community working in the area of statistical signal processing.

(ii) Theoretical and experimental biologists working in the area of chronobiology.
Methods for modeling and estimating the time varying nature of circadian or other oscillations as a result of environmental or experimental conditions, with
online updating of information bases supported by appropriate inferential methodologies will continue to be developed and implemented in a wide range of applications.

(iii) There is large evidence of the Circadian Timing System (CTS) modulation of treatment tolerability and efficacy in experimental models and in cancer patients. The chronotherapy concept has thus obvious implications both for daily patient care and outcomes, and for drug development.

(iv) Our methods in assessing a patient's CTS in real time may also benefit researchers and patients of
chronotherapy in other fields, examples of this are the treatment of asthma, hypertension, and multiple types of depression.

The types of impact that we expect from this project are
- Impact on Knowledge, in particular on techniques (through beneficiaries in groups [i] and [ii]).
- Impact on Society, in particular impact on health and quality of life (through beneficiaries in group [iii]
and [iv] above).
- Impact on People, in particular on skills (through training the RA involved and facilitating the involvement of postdoctoral researchers in the workshops.