If you asked me this question 10 years ago, I would have said no because knowledge of the GI and GL appeared to be sufficient to predict the insulin response. My PhD student at the time, Jiansong (Jason) Bao had proved this in an elegant study involving 121 single foods and 13 mixed meals (1). However, over the last few years, I’ve gradually come to the conclusion that more research on an insulin index of foods may shed light on the origins of obesity as well as some forms of diabetes.
I say this because I’ve learned more and more about energy metabolism and fuel use, particularly during pregnancy since then. Pregnancy is an insulin resistant state of being, meaning that the mother’s body requires the secretion of a good deal more insulin to do the job it normally does of driving glucose into tissues and cells. The largest and most insulin resistant tissue is the mother’s muscles, but interestingly, the placenta remains exquisitely insulin sensitive.
This state of affairs allows glucose (our primary fuel) to be redirected away from muscle towards the foetus which uses glucose for both energy metabolism and growth. If the mother’s insulin secretion is compromised – perhaps a smaller or dysfunctional mass of beta-cells in her pancreas – her blood glucose begins to rise, thereby exposing the placenta to larger than normal amounts of glucose.
In developed nations, pregnant women are asked to undertake a glucose tolerance test at 26-28 weeks gestation corresponding to the beginning of the third trimester. If her blood glucose level exceeds a specific threshold, she will be given a diagnosis of gestational diabetes, and she and her health care team will do their best to reduce her glucose towards the normal range. She may be told to reduce her total available carbohydrate intake, divide it more evenly between meals and consume low GI sources of carbohydrate (2). If these efforts are not sufficient on their own, she will be prescribed daily insulin injections until her baby is delivered.
The reason for making all this effort is that untreated gestational diabetes (and even high-end normal levels of glucose) results in a very large infant, defined as macrosomia when birth weight is >4 kg (8.8 pounds), or large-for-their-gestational age, when their weight is >90th percentile for the number of weeks of pregnancy. A large newborn is a concern because it increases the chances of adverse pregnancy outcomes, including an emergency C-section. Shoulder dystocia, where the shoulders are larger than the head, often predicates a planned C-section delivery.
Unfortunately, increasing rates of overweight and obesity in women (and probably men) at the time of conception have increased the odds of gestational diabetes, pre-eclampsia (hypertension in pregnancy) and macrosomia. In turn, these outcomes have driven higher risk of childhood obesity and type 2 diabetes. And of course, this means we have a viscous cycle of more obesity and more type 2 diabetes in adulthood, generation after generation.
So why would an insulin index of foods be helpful? In my mind, one reason is the undeniable fact that the foetus itself does not experience high glucose levels. It simply produces more insulin in response to high maternal glucose. This higher amount drives greater cell growth and multiplication, resulting in a large baby that has higher than normal amounts of fat or muscle, or both (3). The important point I’m making here is that insulin levels are high, despite normal levels of glycemia (blood glucose). And this is true for a sub-group of foods that have a low GI, yet high insulin index (II).
The first foods to be identified as such (low GI, high II) were dairy products, including milk and fermented products (4). Milk contains protein as well as carbohydrate (the sugar, lactose), both of which stimulate insulin secretion. Even breast milk produces this pattern of low glycemia and hyperinsulinemia (5). Since all mammal milk is designed for growth of young animals, it’s likely that this combination facilitates normal growth under normal circumstances.
However, our research has also determined that there are many other foods with this unusual combination. They happen to be foods that have proportionately large amounts of protein and/or fat relative to carbohydrate content. Many also happen to be particularly palatable foods that are easy to over-consume, such as chocolate-based products (6). Chocolate milk, for example, produces higher insulin responses that strawberry milk.
Generally speaking, the vast majority of carbohydrate-containing foods such as breads, breakfast cereals, rices, pasta and grains have a GI and II that correlate well with each other. High GI means high II and low GI means low II. Under these circumstances GI and GL will be good indicators of insulin demand.
But those low GI/high II foods may be flying under our radar. Their high insulinogenic demand may drive incremental weight gain creep, the tiny amounts of weight gain that adults tend to accumulate every year (0.5 to 1.0 kg) year in, year out, until we suddenly realise we are 10 or 20 kg heavier at 40 than we were at 20 years. And 20 to 40 kg heavier at 60 years. And those extra kilos are very hard to lose.
A better indicator of insulinogenic potential is to test foods in equivalent energetic amounts, for example 1000 kJ or 240 Calories. Our team at the University of Sydney tested ~40 foods in this way and published their ‘insulin score’ in 1997 (7) in the hope that this might be useful for people with type 1 diabetes (8). The relative differences in insulin secretion among healthy subjects to different foods could provide clues to the amounts of insulin that should be administered by those without functional beta cells in their pancreas (9).
In the future, we hope to identify more foods with a high insulin score per unit of energy, particularly those foods with a low GI/high II. In the meantime, work in animals comparing diets with the same GI but high vs low insulinogenic effects may help to identify the mechanisms by which some foods are more obesogenic or diabetogenic than others.
- Bao and colleagues. Prediction of postprandial glycemia and insulinemia in lean, young, healthy adults: glycemic load compared with carbohydrate content alone. Am J Clin Nutr 2011.
- Moses and colleagues. Effect of a low-glycemic-index diet during pregnancy on obstetric outcomes. Am J Clin Nutr 2006.
- Combs and colleagues. Relationship of fetal macrosomina to maternal postprandial glucose control during pregnancy. Diabetes Care 1992.
- Ostman and colleagues. Inconsistency between glycemic and insulinemic responses to regular and fermented milk products. Am J Clin Nutr 2001.
- Brand-Miller and colleagues. Digestion of human milk oligosaccharides by healthy infants evaluated by the lactulose hydrogen breath test. The Journal of Pediatrics 1998.
- Brand-Miller. Chocolate consumption and glucose response in people with diabetes. Edtion ed. In: Knight IOBS, 1999: 195-207., ed. Chocolate and cocoa: health and nutrition. Oxford: Blackwell Science, 1999:195-207.
- Holt S and colleagues. An insulin index of foods: the insulin demand generated by 1000-kJ portions of common foods. Am J Clin Nutr 1997.
- Bao and colleagues. A food insulin index: a physiological basis for predicting insulin demand evoked by composite meals. Am J Clin Nutr 2009.
- Bell and colleagues. Clinical Application of the Food Insulin Index for Mealtime Insulin Dosing in Adults with Type 1 Diabetes: A Randomized Controlled Trial. Diabetes Technology & Therapeutics 2016.
Professor Jennie Brand-Miller holds a Personal Chair in Human Nutrition in the Charles Perkins Centre and the School of Life and Environmental Sciences, at the University of Sydney. She is recognised around the world for her work on carbohydrates and the glycemic index (or GI) of foods, with over 300 scientific publications. Her books about the glycemic index have been bestsellers and made the GI a household word.