Novel Food Algorithm to Improve Mealtime Insulin Dosing

CHICAGO — A novel algorithm based on the normal pancreatic response to food could represent an alternative to carbohydrate counting for premeal insulin bolusing in patients with type 1 diabetes.

Information about the limitations of carbohydrate counting and the potential of the investigational algorithm, dubbed the "food insulin index" (FII), was presented here at the American Diabetes Association (ADA) 2013 Scientific Sessions by Kristine Bell, a dietician, diabetes educator, and PhD candidate at the University of Sydney, Australia.

Currently, patients with type 1 (and some with type 2) diabetes are taught to base their premeal insulin "bolus" doses solely on the number of grams of carbohydrates in the food they are about to eat, without regard to the protein or fat content of the food or to the variable effect of different carbohydrates, as expressed by the glycemic index.

However, Ms. Bell told Medscape Medical News: "There are some fundamental limitations to carbohydrate counting, both in theory and in clinical practice… Carbohydrate is the main macronutrient affecting the blood glucose response. However, stimulation of normal insulin release is multifactorial. In addition, the efficacy of the intervention is also limited by how accurately and precisely people can count carbs and then…use this information to adjust their insulin dose."

The food insulin index does not attempt to account for all the individual factors — which have been shown to include fat and protein — that influence blood glucose response to food. Rather, it is a relative measure of the overall normal insulin demand of a food, derived from testing the insulin response generated by 1000-kJ (240-kcal) portions of single foods in healthy adults without diabetes (J Clin Nutr. 2009;90: 986-992).

"The food insulin index measures the postprandial increase in insulin secretion of a whole food in normal healthy subjects and thus provides a more holistic approach to predicting the prandial insulin requirements in type 1 diabetes," she told Medscape Medical News.

Howard A. Wolpert, MD, from the Joslin Diabetes Center, Boston, Massachusetts, is somewhat ambivalent about this new advance, however. While the FII could be particularly advantageous for rapidly absorbed high-glycemic-index meals — breakfasts in particular — it could underestimate the amount of insulin needed for meals containing more fat, since they are absorbed more slowly, he told Medscape Medical News. Also, the FII is similar to carbohydrate counting in that it adds complexity to the already-difficult medical regimens for people with type 1 diabetes, he observed. That said, the FII has "some important implications," he conceded.

The Limits of Carb-Counting

Ms. Bell summarized the findings of her group's systematic review and meta-analysis demonstrating a lack of evidence to recommend carbohydrate counting as the standard dietary therapy in type 1 diabetes.

Of 294 potentially relevant studies, only 6 met the criteria of randomized controlled trials of more than 3 months' duration comparing carbohydrate counting with other dietary advice in adults and children with type 1 diabetes. The total study population comprised 563 adults and 104 children.

"This is a very limited number of studies to be forming the evidence base for such an integral element of type 1 diabetes management and highlights the need for further research to support and improve clinical practice," Ms. Bell told Medscape Medical News.

Study quality scores averaged just 7.7 out of 13. Overall, carbohydrate counting reduced HbA_1c nonsignificantly, by 0.3 percentage points, compared with controls or other dietary advice (P = .185). The adults had a 0.4-percentage-point improvement, a marginally significant difference (P = .048). Nonsignificant decreases in hypoglycemia were reported with carb counting in 5 of the studies.

"There is limited evidence to recommend carbohydrate counting as the standard dietary therapy in type 1 diabetes. Additional studies are required to support this strategy in routine clinical care," she concluded.

The Food Insulin Index: There Might Be an App for That

Ms. Bell described the FII in a separate talk during a symposium at the meeting. Expressed on a scale of 0-100, the FII is derived by the following formula (Am J Clin Nutr. 1997;66:1264-1276):

(120 min area under the curve [insulin] for 1000 kJ of test food/100) x 120 min AUC (insulin) for 1000 kJ of reference food.

Published FIIs, of which there are now about 130, are based on insulin responses from 10 to 13 healthy subjects each.

For example, for a 1000-kJ portion of whole-grain bread with 40 g of carbohydrates, the FII would be 41. For an equivalent serving of white bread, with a similar 44 g of carbohydrates, the FII would be vastly greater, at 73, because white bread contains less fiber than whole-grain bread and is metabolized more rapidly.

A serving of low-fat yogurt with 38 g of carbs has an FII of 84, whereas a 58-g apple's FII is only 43. That's due to the dairy protein in the yogurt, Ms. Bell explained.

And beef and chicken, which have zero carbs, actually have FIIs of 37 and 19, respectively. "The FII relates not just to carbohydrate foods but to foods you would normally ignore with carbohydrate counting," she noted.

An initial study from the Sydney group showed that the FII is a more accurate predictor of observed insulin response among healthy subjects than is carbohydrate content (Am J Clin Nutr. 2009;90:986-992). Subsequently, they showed that the FII algorithm significantly decreased postmeal glucose spikes following 2 different breakfast meals among 28 insulin-pump–using patients with type 1 diabetes (Diabetes Care. 2011;34:2146-2151).

And in a new study, presented separately at the ADA meeting by one of Ms. Bell's colleagues, 11 adults with type 1 diabetes on insulin-pump therapy were given 1 of 6 single protein-containing foods (steak, battered fish, poached eggs, low-fat yogurt, baked beans, and salted peanuts) and given insulin doses based on either the FII or carb counting.

Compared with carb counting, the FII algorithm significantly reduced mean blood glucose levels (P = .003) and produced a smaller mean change in blood glucose level (P = .001) and a smaller peak postmeal glucose excursion (P = .835), all without significantly increasing the risk for hypoglycemia (P = .155).

Ms. Bell and her colleagues are now conducting a real-world 3-month randomized controlled trial in 36 adults with type 1 diabetes who are educated in either carb counting or use of FII. The patients are provided with written resources and pictorial guides, just as is now done to teach carb counting. Patients typically learn FII in a couple of hours, "the same as when someone first starts carb counting," Ms. Bell told Medscape Medical News.

"It takes practice to get familiar with the foods you eat frequently. People really have a small selection of foods they eat 90% of time," she noted.

HbA_1c and continuous glucose monitoring data are being collected. The study began in May and is expected to be completed this November. If all goes well, she ultimately envisions FII training on Web sites and the FII formula being worked into apps "to make it as accessible as possible."

Carbohydrate Counting Here for Foreseeable Future

Dr. Wolpert, who recently published a study on the effect of dietary fat on postmeal glucose levels and insulin requirements (Diabetes Care. 2013;36:810-816), told Medscape Medical News that fat sensitivity varies from person to person, with some needing much more additional insulin for high-fat meals than others. And it is not clear as yet whether there is a dose-response with fat as there is with carbohydrates, he noted, adding that, for these reasons, it would seem the FII is a better tool for carbohydrate counting than for fat-containing foods.

Moreover, Dr. Wolpert said, "You have to be very wary of adding to the complexity and challenges people [with diabetes] face.

"On the other hand," he said, the FFI is "novel. The way I envision [it] is that one could use FII as a starting point for people to get a better sense of how much insulin they need for specific meals or specific foods."

Ms. Bell emphasized that carb counting is not going away any time soon. "At this stage, carbohydrate counting is still the best-known method for adjusting mealtime insulin dose. Clinicians do, however, need to be aware of the potential limitations of the intervention, to inform their own clinical practice and better support their patients."

Ms. Bell has no reported no relevant financial relationships. Dr. Wolpert is a consultant to Abbott Diabetes Care, Becton Dickinson, and Novo Nordisk. He receives research support from Abbott Diabetes Care.

American Diabetes Association 2013 Scientific Sessions. Abstracts 164-OR and 165-OR, presented June 23 and 24, 2013.

Medscape Medical News © 2013  WebMD, LLC

Send comments and news tips to news@medscape.net.

Cite this: Novel Food Algorithm to Improve Mealtime Insulin Dosing - Medscape - Jun 28, 2013.