Weather forecasting models could predict brain tumor growth

Meteorologists predict the weather through complex mathematical models, monitoring changes in state in a given time and space.  Today, an innovative study published in Biology Direct shows that state estimation schemes, applied operationally for weather forecasting, can be utilised in principle to predict the growth and spread of malignant brain tumors.

In an effort to demonstrate the potential utility of spatiotemporal models of biological processes, Kostelich et al. focused on the possibility of creating clinically useful forecasts of glioblastoma multiforme, the most common and aggressive type of human brain cancer.

Glioblastoma has a poor prognosis, with an average survival time of approximately 14 months from diagnosis, and it is largely resistant to conventional therapies such as chemo- and radiotherapy. The location and density of the glioblastoma tumor cell population influences patient symptoms and treatment planning. This, combined with the complex geometry of the tumor, made it an important cancer against which to test this mathematical methodology.

A modern state estimation algorithm previously used for numerical weather prediction, referred to as the Local Ensemble Transform Kalman Filter (LETK), was applied to two different mathematical models of glioblastoma. Data assimilation techniques were then employed for updating the state vector, that is the initial condition of the glioblastoma growth model, via a combination of new observations with one or more prior forecasts.

The feasibility of this approach for making short-term (60 day) forecasts of glioblastoma spread and growth was successfully demonstrated, in individual patient cases, using the synthetic magnetic resonance images of a hypothetical tumor.

Although preliminary, this intelligent forecasting method for glioblastoma progression could prove useful in a clinical setting, with potential to aid treatment planning and patient counseling. On a wider scale, it also shows much promise in being applicable to the modeling efforts of other cancers and diseases in the future.

View the latest posts on the On Biology homepage