The Role of Dietary Protein and Fat in Glycaemic Control in Type 1 Diabetes: Implications for Intensive Diabetes Management

Unlike proteins or CHO, fat is rarely consumed in isolation as a single macronutrient, therefore the impact of fat is usually considered in terms of its ability to modify the CHO-induced blood glucose response.

In healthy people and people with type 1 diabetes, adding fat to a CHO meal reduces the glycaemic response in the early postprandial period (first 1–3 h) but extends the glycaemic response over hours. Using continuous glucose monitoring (CGM) profiles in 24 healthy non-diabetic adults, Freckmann et al. demonstrated that meals with a higher fat content induced a smaller increase and a slower decrease of postprandial glucose concentrations than a high-carbohydrate low-fat meal [28].

As with healthy individuals, the addition of fat tends to decrease the glycaemic response in the early postprandial period in people with type 1 diabetes, and thus increases the risk of hypoglycaemia. However, in the late postprandial period, high-fat meals often cause significant hyperglycaemia lasting for several hours following the meal. This hyperglycaemia is problematic for good glycaemic control, particularly since high-fat meals are often consumed for the evening meal when prolonged hyperglycaemia may not be detected until the following morning. A recent study of the dietary factors associated with nocturnal hypoglycaemia in adults has shown a significant association with post-dinner fat intake and higher rates of nocturnal hyperglycaemia occurrence [29].

Several studies looking at the effects of dietary fat have been published in recent years. We have shown that the addition of 35 g of dietary fat to 30 g of CHO initially reduces the glycaemic excursion for up to 90 min following the meal but then significantly increases the postprandial glycaemic response from 3 h onwards, with the BGL increased by 2.3 mmol/L at 5 h [18••]. In an insulin clamp study, Wolpert et al. also demonstrated that a high-fat meal (50 g of fat) causes significant hyperglycaemia over 5 h, even with additional insulin infused [19••].

Unlike proteins or CHO, fat is rarely consumed on its own as a single macronutrient. Triacylglycerols (TAGs) constitute more than 90 % of lipids in the human diet, and this review will focus on their role in the management of BGLs.

TAGs can influence the glycaemic response in type 1 diabetes by (1) gluconeogenesis of glycerol, (2) a direct effect of free fatty acids (FFA), (3) effects on other hormones and (4) effects on gastric emptying.

The impact of dietary fat on gastric emptying of the subsequent meal regardless of its content also needs consideration, as dietary fat from breakfast has been shown to delay the rate of gastric emptying of the subsequent lunch meal [41]. Furthermore, intake of high-fat diet for 1 week desensitises gastrointestinal transit and leads to a more rapid gastric emptying [42].

Because of fats impact on gastric emptying, modification of dietary fat content is used in the management of diabetes-related gastroparesis.

In individuals without diabetes, dietary protein does not alter postprandial glycaemia [43], however, it stimulates a significant postprandial insulin response, which is required for amino acid uptake [44]. In order to maintain euglycaemia, protein concurrently stimulates glucagon secretion, thereby promoting hepatic glucose release and regulation of BGL [45‐47].

In contrast, when people with type 1 diabetes consume protein, there is an increase in postprandial BGLs and insulin requirements. Studies comparing the postprandial glycaemic responses to standard meals with meals containing between 28 and 57 g of additional protein have demonstrated significantly higher postprandial glycaemic excursions and insulin requirements in the 2–5 h postprandial period in type 1 diabetes [17, 18••, 45, 48]. More recently, our group looked at the glycaemic impact of ingested protein only (independent of CHO and fat) in children and young adults with type 1 diabetes [49]. Consumption of ≥75 g of protein resulted in a significantly delayed, sustained, elevated, postprandial glycaemic excursion from 180 to 300 min compared with a control drink (water). The peak BGL from ≥75 g of protein was similar to that of 20 g of glucose given without insulin; however, the shape of the glycaemic response was significantly different. Following the ingestion of protein, the BGL only began to rise after 100 min and reached the peak BGL after 5 h.

These findings demonstrate that for people with type 1 diabetes the addition of ≥28 g of protein to a mixed meal or consuming ≥75 g of protein alone is likely to result in significant and sustained postprandial hyperglycaemia commencing in the late postprandial period (2–3 h) and continuing beyond 5 h. Clinically, this is important when determining whether additional insulin is required and, importantly, how to safely calculate and distribute these increased doses.

The sustained postprandial hyperglycaemic effect of protein could potentially be utilised to decrease nocturnal hypoglycaemia, however, research has shown that high-fat meals (which also have sustained postprandial hyperglycaemia) did not prevent nocturnal hypoglycaemia [50]. Hence, further research is required to determine if high-protein meals would decrease nocturnal hypoglycaemia.

Two mechanisms have been proposed by which dietary protein may cause delayed and sustained postprandial glycaemic excursions in people with type 1 diabetes: (1) the alteration of hormones which affects glucose homeostasis and (2) the conversion of amino acids to glucose by gluconeogenic pathways.

A study by Khan et al. showed that 50 g of ingested protein resulted in an additional 9.7 +/− 5.7 g of glucose in the circulation over 8 h in healthy adult males [43]. This demonstrates that by weight, 19.4 % of the total amount of protein ingested can be converted to glucose. However, considering only the amount of glucose that enters the circulation ignores the protein-induced hormonal response (as discussed above) and that some additional insulin may be needed to counter the resulting decrease in insulin sensitivity.

Alternatively, Pańkowska et al. have proposed that 100 kcal of fat/protein requires the equivalent amount of insulin as 10 g of glucose [16, 59]. Using this ratio (protein contains 4.1 kcal/g), 100 kcal of protein = 24.4 g of protein = 10 g of glucose. This is approximately twice the amount of glucose that Khan and colleagues found was produced from protein in healthy volunteers [43]. This apparent discrepancy may be explained because Khan et al. measured the amount of glucose produced, whereas Pańkowska et al. used clinical response (which would account for glucose production and hormonal responses). Kordonouri et al. used the Pańkowska equation in children with type 1 diabetes using IPT and found that although glycaemic control was improved, the risk of postprandial hypoglycaemia was increased threefold [20••]. This suggests that further research is required to determine how much additional insulin is required to cover the protein component of a meal.

Managing the early postprandial period is the first step in adjusting mealtime insulin doses, as changes implemented here will have flow-on effects for the late postprandial period.

For high CHO meals, controlling the immediate postprandial glycaemic rise is important as an increased initial glucose spike can increase the entire postprandial excursion [61]. For example, an increase in the early postprandial response by 2 mmol/L, with a consequential increase in BG at all time points, could result in a 2–3× fold greater glycaemic excursion. This is especially true for meals also containing significant amounts of fat and protein, which will further increase the late postprandial period.

Changing the timing of the insulin bolus to minimise the immediate postprandial response has been investigated, however, not in the context of high-fat and/or high-protein meals. Although the immediate postprandial glucose rise is blunted after a high-fat meal, there is still a need for insulin to cover the CHO component of the meal [18••]. Therefore, in the author’s opinion, delaying the insulin injection for high-fat meals would cause unacceptable postprandial hyperglycaemia. Research looking at the timing of the insulin bolus has shown that giving the insulin bolus before starting to eat gives superior glycaemic control compared to delaying the insulin bolus [61‐63]. Conversely, for high-CHO meals also containing fat and protein, two studies have shown that dosing insulin 15 min prior to the meal was able to improve control over the initial postprandial spike, significantly reducing the peak BGL [62, 64]. However, in practice, dosing 15 min prior to a meal may prove too onerous for some patients and may result in insulin doses being forgotten. As with all diabetes management recommendations, advice should be tailored to the individual.

For patients using IPT, the insulin delivery pattern can be adjusted to better match insulin requirements. The bolus for a meal can be given as a standard bolus (where the entire dose is given within 5 min), an extended bolus (where the dose is infused at a constant rate over a chosen number of hours) or a combination bolus (where a proportion of the dose is delivered immediately and the remainder is infused continuously as an extended bolus). However, not all boluses are able to control the immediate postprandial rise [65, 66]. Chase et al. found that extended boluses did not control the immediate postprandial glucose rise with a high-fat meal [65]. Lopez et al. attempted to optimise the amplitude (height of the wave) of the extended bolus but was not able to control the immediate postprandial glucose rise compared to the standard bolus [66].

Hypoglycaemia in the early postprandial period is also a concern for low-CHO meals containing significant amounts of protein and/or fat, with patients often reporting hypoglycaemia soon after the meal when insulin is dosed upfront. This is likely due to mismatch between the rapid-acting insulin and the delayed gastric emptying following fat and the delayed glycaemic responses for both protein and fat. For these meals, the initial insulin dose needs to be reduced upfront but potentially increased in the latter phase, with the total dose delivered over an extended period to match the food absorption.

Changing the type of insulin delivered is an option for reducing the risk of early hypoglycaemia with MDI. To our knowledge, there are no studies comparing regular insulin with rapid-acting insulin (such as aspart or lispro insulin) for high-fat and/or high-protein meals. However, in theory, regular insulin with its slower onset of action and longer duration of action may offer an advantage in meals high in fat where gastric emptying is delayed and the initial postprandial glucose rise is blunted [18••].

Fat and protein cause increased glycaemia continuing to 6–12 h after meal ingestion [16, 18••, 20••, 50]. To cover this extended glycaemic effect of fat and/or protein, the duration of the insulin effect needs to be extended and total dose may need to be increased.

There is much debate over how additional insulin doses should be calculated and how much protein or fat needs to be consumed before insulin doses need to be adjusted. Two different algorithms have been proposed to calculate insulin for meal fat and protein content. Pankowska et al. have proposed that the insulin dose for 10 g of CHO should be equivalent to the insulin needed for every 100 kcal of fat/protein. Alternatively, Brand-Miller et al. have proposed a food insulin index, which is a measure of the postprandial insulin response to foods in healthy subjects, as a basis for determining mealtime insulin doses in type 1 diabetes [12, 67, 68]. However, both of these methods have resulted in significant postprandial hypoglycaemia and require further research.

Closed-loop studies have suggested that for high-fat and/or high-protein meals, the insulin dose needs to be increased by 42 % to as much as 125 %. Wolpert et al. revealed a mean insulin dose increase of 42 % was needed for a high-fat meal (60 g fat) compared to the low-fat meal (10 g fat), however, given patients still experienced significantly hyperglycaemia following the high-fat meal, this dose is likely to be an underestimation [19••]. Importantly, there was a wide variation in the dose adjustment needed to cover the high-fat meal (range −17 to +108 %), highlighting the importance of individualised advice. A subsequent study by this group used predictive modelling to optimise the insulin dose in patients using insulin pump therapy for the addition of 35 g of fat and 32 g of protein to a meal, finding a mean insulin dose increase of 78 %, using a combination-wave with 30/70 % split over 2.4 h [69]. Similarly, there was a wide variation in the total dose increase (+38 to +125 %) and the optimal split and duration (range 20/80 % to 50/50 % and 2–3 h).

Since rapid-acting insulin given as a standard bolus will not last long enough to cover the long tail caused by protein and fat, altering the insulin delivery pattern (i.e., bolus type) may assist with controlling late postprandial hyperglycaemia in patients using IPT. Extended boluses are also ineffective for high-protein/fat meals, as they are unable to control the immediate postprandial BGL rise [65, 66, 70], even though the rate of the rise is blunted [18••]. Currently, the most effective form of insulin pump bolus for meals containing protein and/or fat is the combination-wave bolus [65, 70, 71]. Jones et al. demonstrated that compared with a standard bolus, a combination-wave bolus significantly reduced the mean BGL and improved the amount of time spent within the normal BGL range for a high-fat pizza meal [71]. Similarly, Chase et al. also found that the combination-wave gave the best postprandial glycaemic control following a high-fat pizza and tiramisu meal [65]. To date, there are no studies looking at the optimal split of insulin between the immediate and extended bolus components for combination-wave boluses. However, studies have used 50 % of insulin given immediately and 50 % given as the extended bolus, and this can be used as an appropriate starting point in practice and adjusted as needed. The duration has been found to have an improved glycaemic profile when the duration of the bolus is increased from between 4 and 8 h [5].

Extending the insulin action time is more problematic for patients using MDI. Theoretically, the insulin injection could be given as a split bolus with two separate injections: one injection prior to the meal and the other given 1 h following the meal. However, this method needs to be investigated to determine its efficacy in practice.

For small meals and snacks, large adjustments of insulin doses may not be required. Wilson et al. compared a low-fat (30 g CHO, 2.5 g protein, 1.3 g fat; 138 kcal) and a high-fat (30 g CHO, 2 g protein, 20 g fat; 320 kcal) snack at bedtime and showed that during an 8-h overnight period, the average BGL following the low-fat snack was 8.7 and 9.5 mmol/L following the high-fat snack [50]. In this scenario with a high-fat snack (rather than a large meal), a standard bolus for the CHO component and a temporary 10 % increase in basal overnight would appear appropriate.