BOSTON – For millions of people in the United States living with type 1 or type 2 diabetes, measuring the daily rise and fall of blood glucose (sugar) is a way of life.
Our body’s energy is importantly governed by glucose in the blood, and blood sugar itself is exquisitely controlled by a complicated set of network interactions involving cells, tissues, organs and hormones that are evolved to keep the glucose on a relatively even keel, pumping it up when it falls too low or knocking it down when it goes too high. This natural dynamical balance becomes lost when someone develops diabetes.
But diabetes is a surprisingly difficult disease to manage, even with all the modern interventions involving diet, drugs and lifestyle changes. Routine blood glucose measurements, a central part of disease management, helps patients and their caregivers plan when they should take insulin and other drugs and regulate their diet. These difficulties in managing the disease and the explosion of new cases of type 2 diabetes has motivated the search for new approaches to monitoring and therapy to further optimize clinical decisions and personalize glucose control.
Now a team led by investigators at Beth Israel Deaconess Medical Center (BIDMC), has discovered that there may be more to tiny fluctuations in blood sugar than meets the eye. In fact, these largely ignored, instantaneous dynamics may provide a wealth of information encoded in the small, seemingly inconsequential ups and downs of blood sugar.
As they described in a recent issue of the journal Chaos, extracting this information may illuminate some of the poorly understood frontiers of human physiology and possibly even suggest new ways to monitor and treat diabetes based on maintaining and restoring the complexity of the overall control system, a notion they dub using the “system as target.”
In the new paper, Madalena Costa, PhD, and Ary Goldberger, MD, of BIDMC’s Division of Interdisciplinary Medicine and Biotechnology, together with clinical colleagues, examined deidentified, retrospective data collected from 18 elderly adults with type 2 diabetes and 12 age-matched controls who did not have diabetes. All 30 people were hooked up to continuous glucose monitors – implantable devices that can measure blood sugar levels every five minutes around the clock.
Looking at several days’ worth of this data, the team saw that glucose levels undergo small fluctuations constantly – both in people with diabetes and in people without the disease. However, the fluctuations were different between the two groups, tending to be smaller and moving on a faster time scale in people who did not have diabetes.
Doctors have noticed such tiny fast frequency fluctuations in the blood sugar of people without diabetes before, but for the most part, it was always assumed that this signal was due to random noise associated with the limits of detection of the glucose meters. The fluctuations were not very large, after all. But Costa, Goldberger and their colleagues have shattered this assumption by showing that there is complex information encoded in these dynamics – information that changes when compared with subjects with type 2 diabetes.
They discovered this by applying a sophisticated mathematical technique called multiscale entropy analysis, which quantifies the complexity of data and compares the value to data sets obtained by shuffling the order of the measurements taken every five minutes in the 30 people and looking at how the variability between them changes.
“This combination of computational procedures allows us to say how unpredictable the time series is over different time scales,” said Costa.
This analysis showed that the short-term fluctuations (as well as the longer term ones) do not represent correlated randomness, but encode complex information. Moreover they found that the information encoded in these fluctuations is significantly more complex in people without diabetes – something doctors have never consistently observed before.
The apparent loss of complexity with the onset of the disease has led Costa and Goldberger to suggest a novel way of studying diabetes, which they are calling “dynamical glucometry,” an approach that would seek to uncover and make sense of the hidden information encoded in these fluctuations, rather than just relying on spot checks and average values.
In addition to Costa and Goldberger, coauthors include Teresa Henriques, Medha N. Munshi and Alissa R. Segal.
Adapted from a press release by the journal Chaos, AIP Publishing.