Using weather forecasting to predict brain tumor growth

The Arizona State approach to tumor forecasting in action.

Arizona State University researchers believe their research in improving weather forecasting could be applied to brain cancer. Their proof-of-concept study, published by Biology Direct, shows they might be correct.

Brain cancer is chaotic. The most common form, glioblastoma multiforme, is also the most aggressive, said surgeon research team member Mark Preul, the director of neurosurgery research at Phoenix’s Barrow Neurological Institute. Glioblastoma yields a life expectancy only around a year and a half. But it is also chaotic in the sense of chaos theory. Like the weather, the factors involved in the spread of brain cancer are extraordinarily complex and produce increasingly inaccurate as predictions are made for further and further into the future. A forecast of tomorrow’s weather, or tomorrow's spread of a brain tumor, is likely to be nearly accurate. A forecast for six months from now likely is not.

“The amazing thing is since first identified in the early 1900s, survival rates haven’t shown many improvements,” Preul said.

One of the problems with weather prediction is the exponential increases in error as the forecast bases new predictions on old predictions that were created using imperfect data. Mathematician Eric Kostelich, who headed this study, was on a team that developed a formula for combining a prior forecast and new measurements to get better initial data. They called it the Local Ensemble Transform Kalman Filter.

But the filter was not specific to weather. Mathew Hoffman, a graduate students of Kostelich, used it to determine oceanic conditions.

“I thought about applying the filter to cancer because I had a family member who was suffering from cancer,” Kostelich said. Using the existing models and the filter’s more accurate initial measurements, Kostelich was able to better predict the spread of Glioblastoma through the brain.

There is no way to fully remove the cancer, but Preul and Kostelich note that better data can lead to more accurate and more proactive treatments that can improve a patient’s quality of life “The main value would be for the radiation oncologist while they plan the treatment volume,” said Rush University Medical Center neurosurgeon Richard Byrne.

The ASU study was preliminary, based only on a handful of patients. The next step in testing the filter’s applications to brain cancer will be to test on mice, he said. Kostelich is confident their methods could be applied to other biological phenomena including other forms of cancer.