A pharmacist's perspective on health and metabolic disease
The Journal of Insulin Resistance (edited by Jason Fung) has just published our article from my PhD looking at the repeatability characteristics of HOMA 2 measures.
When I first started my research, I thought that fasting insulin or HOMA measures would be useful way to measure insulin resistance and hopefully hyperinsulinaemia – because they were used everywhere.
But I was also feeling very bitter and disillusioned as I was learning that everything I thought I knew about cholesterol and low fat diets was based on poor science. So in a fairly naive and hopeful way I was determined to do “good science” – which meant questioning everything. This meant I didn’t like all I read about HOMA.
HOMA is supposed to be an easy way to assess insulin resistance based on your fasting insulin and fasting glucose levels. Insulin resistance is how easy or hard it is for glucose to get into your muscle cells, the harder it is, the higher your degree of insulin resistance.
But it wasn’t clear if measuring insulin resistance would translate into assessing hyperinsulinaemia.
What I also did not know was by how much HOMA needed to change to be able to tell whether we were still within normal biological variation, or had we reached clinical change. It is normal for biological markers, such as blood glucose levels, to vary a little as body’s a constantly in a state of flux and change. The question was how much is too much to be considered “normal”, which then becomes clinical change. I could not get any straight answers from the literature, so had to devise a study to look at this. This was just published and can be accessed here:
Xiaomiao Lan-Pidhainy and Thomas Wolever (University of Toronto) very kindly gave me the raw data from one of their studies and I looked at how much variation you can get when you repeatedly assess the same measures in the same subjects under the same conditions.
Below are the raw results from 19 people with normal glucose tolerance and fairly normal insulin levels (couldn’t do Kraft patterns) and 10 people with type 2 diabetes. Their fasting glucose and fasting insulin levels were measured eight times. Each diamond shape represents a test result.
When you look at the fasting glucose results, all the results for everyone are fairly close together. This is a good example of a very repeatable result. We know that if we get change of > 0.5mmol/L or 10mg/dL we would be concerned that the person’s clinical state has changed.
However, fasting insulin levels, and the HOMA 2%B have a much wider variation . A normal level could range from anywhere from 2mU/L to 20mU/L.
Not useful in my opinion on the individual level. (Possibly good for a large scale intervention study).
But when we look at the results based on insulin response patterns produced after an oral glucose tolerance test (in this case, OGIS), the results are very different
Here people had their baseline blood taken then were given 50g glucose to drink. Further blood samples were taken for insulin and glucose. The results are more tightly clustered together.
Sure, there are a number of limitations within this study including only having access to three OGIS results, not eight as per HOMA. But to me, it suggests that there are a number of challenges with HOMA, and we need to look further at its usefulness as a measure for the individual.
Please download the article for a full discussion, including the full statistical analysis of the repeatbility coefficients. I am very grateful to Mark Wheldon for devising an analysis plan that I could actually understand.