1008 Diabetes Care Volume 38, June 2015

Kirstine J. Bell,1,2 Carmel E. Smart,3,4 Impact of Fat, Protein, and Garry M. Steil,5,6 Jennie C. Brand-Miller,1 Bruce King,3,4 and Howard A. Wolpert2,6

TYPE 1 DIABETES AT A CROSSROADS on Postprandial Control in Type 1 Diabetes: Implications for Intensive Diabetes Management in the Continuous Glucose Monitoring Era Diabetes Care 2015;38:1008–1015 | DOI: 10.2337/dc15-0100

BACKGROUND Continuous glucose monitoring highlights the complexity of postprandial glucose patterns present in type 1 diabetes and points to the limitations of current approaches to mealtime insulin dosing based primarily on counting.

METHODS A systematic review of all relevant biomedical databases, including MEDLINE, 1Charles Perkins Centre and the School of Molecular Embase, CINAHL, and the Cochrane Central Register of Controlled Trials, was Bioscience, The University of Sydney, Sydney, conducted to identify research on the effects of dietary fat, protein, and glycemic Australia 2Joslin Diabetes Center, Boston, MA index (GI) on acute postprandial glucose control in type 1 diabetes and prandial 3Department of Paediatric Endocrinology and insulin dosing strategies for these dietary factors. Diabetes, John Hunter Children’s Hospital, Newcastle, Australia RESULTS 4Hunter Medical Research Institute, School of All studies examining the effect of fat (n = 7), protein (n =7),andGI(n =7)indicated Medicine and Public Health, University of Newcastle, Rankin Park, Australia that these dietary factors modify postprandial glycemia. Late postprandial hyper- 5Children’s Hospital, Boston, MA glycemia was the predominant effect of dietary fat; however, in some studies, 6Harvard Medical School, Boston, MA glucose concentrations were reduced in the first 2–3 h, possibly due to delayed Corresponding author: Howard A. Wolpert, gastric emptying. Ten studies examining insulin bolus dose and delivery patterns [email protected]. required for high-fat and/or high-protein meals were identified. Because of meth- Received 15 January 2015 and accepted 26 odological differences and limitations in experimental design, study findings were February 2015. inconsistent regarding optimal bolus delivery pattern; however, the studies in- This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/ dicated that high-fat/protein meals require more insulin than lower-fat/protein suppl/doi:10.2337/dc15-0100/-/DC1. meals with identical carbohydrate content. © 2015 by the American Diabetes Association. Readers may use this article as long as the work CONCLUSIONS is properly cited, the use is educational and not These studies have important implications for clinical practice and patient edu- for profit, and the work is not altered. cation and point to the need for research focused on the development of new See accompanying articles, pp. 968, insulin dosing algorithms based on meal composition rather than on carbohydrate 971, 979, 989, 997, 1016, 1030, and content alone. 1036. care.diabetesjournals.org Bell and Associates 1009

Currently, are consid- the standard of care in the management QUESTION 1: WHAT EFFECTS DO ered the predominant macronutrient of type 1 diabetes (3), the challenges of GI, PROTEIN, AND FAT HAVE ON affecting postprandial glucose control keeping postprandial glucose concentra- ACUTE POSTPRANDIAL GLUCOSE and the primary determinant for calcu- tions in range will become an inescapable CONCENTRATIONS IN TYPE 1 lating mealtime insulin doses in type 1 and increasing focus in the daily lives of DIABETES? diabetes (1). Emerging evidence from people with diabetes. As summarized in Table 1, 16 studies recent research and the use of continu- examining the effects of nutritional fac- ous glucose monitoring have shown SYSTEMATIC REVIEW PROCEDURE tors on postprandial glycemia were that other nutritional properties of For questions 1 and 2, a search of all identified (4–19) (see Supplementary food, including fat, protein, and glyce- relevant biomedical databases was con- Table 1 for more details). mic index (GI), can significantly affect ducted, including MEDLINE, Embase, postprandial glucose excursions. These CINAHL (Cumulative Index to Nursing and Fat findings highlight the need for alterna- Allied Health Literature), Thomson Reu- Seven studies examining the effect of tive mealtime insulin dosing algorithms ters Web of Science (formerly ISI Web of dietary fat on postprandial glycemia in and have important implications for nu- Knowledge), and the Cochrane Central 103 subjects with type 1 diabetes were trition education and counseling in pa- Register of Controlled Trials. The follow- identified (5,7,8,11,15,18,19). All stud- tients with diabetes. ing key words were used to identify ies added fat to the test meal, which The American Diabetes Association potentially relevant articles: “type 1 concurrently increased the energy con- (ADA) recommends that people with di- diabetes” AND “blood glucose” OR “in- tent of the test meal. The amount of di- abetes who have mastered carbohy- sulin” AND “dietary carbohydrate” OR etary fat added to the control meals drate counting receive education on “dietary protein” OR “dietary fat” OR ranged from 6.6 to 52 g. the glycemic impacts of protein and “glycemic index.” The search strategy All studies reported that dietary fat fat (1). To our knowledge, neither a was deliberately broad to identify all rel- modified postprandial glycemia. One comprehensive review of the literature evant articles to answer the two review study did not find an increase in glucose examining the relative effects of pro- questions. concentration (5) possibly because the tein, fat, and GI on postprandial glyce- Original controlled studies were in- postprandial monitoring period was mia nor guidelines for clinicians on how cluded if they were published in English only 3 h. Two studies reported that the insulindosesshouldbeadjustedintype between March 1995 and November addition of dietary fat reduced the area 1 diabetes for various meal composi- 2014, conducted in adults and/or chil- under the curve in the first 2–3 h (7,8). tions exist. dren with type 1 diabetes, used rapid- This may be due to fat delaying gastric This article reviews the evidence ad- acting insulin analogs (lispro, aspart, emptying, as indicated by one of the dressing the following questions: or glulisine), and 1) compared post- studies demonstrating that the gastric prandial glycemia following different emptying rate was significantly reduced 1. What effects do GI, protein, and fat foods/meals with the same insulin dos- in the first 2 h after consuming a high-fat have on acute postprandial glucose ing strategy or 2) different prandial dos- meal (7). This in turn delayed the rise in concentrations in type 1 diabetes? ing strategies for the same foods/meals. postprandial glycemia, with four studies 2. What prandial insulin dosing strate- A start date of March 1995 was inten- reporting a lag in time to peak blood gies work best for GI, protein, and fat tionally chosen because this was when glucose concentrations between 6 and in type 1 diabetes? the first rapid-acting insulin analog (lis- 90 min (mean 29 min) (5,7,15,18). pro) was available in the U.S. Studies of Evidence suggests that meals contain- In light of the evidence reviewed in participants of any age, sex, or race ing carbohydrates and that are high in these questions and our clinical insights, were included. Studies where both the dietary fat cause sustained late post- this article also discusses the following meal and the insulin dose were concur- prandial hyperglycemia. One study questions: rently altered or the meal composition showed the addition of 35 g dietary fat was not controlled and studies in preg- significantly increased postprandial glu- 3. What are the implications for clinical nancy or involving exercise were cose concentrations by 2.3 mmol/L at 5 practice? excluded. h (15). Wolpert et al. (19) demonstrated 4. What are the knowledge gaps, and The literature search identified 143 that the addition of 50 g fat caused sig- how can technology be leveraged to studies, and 13 more studies were iden- nificant hyperglycemia over 5 h, even improve postprandial glucose control? tified through hand-searching refer- when additional insulin was adminis- ence lists (Supplementary Fig. 1). Of tered using a closed-loop glucose con- Current clinical approaches to inten- these studies, 127 were excluded pri- trol system. Free fatty acids (FFAs) sive diabetes management tend to be marily due to duplication (n =23)and directly induce insulin resistance, and insulin-centric; however, a focus on di- not meeting the inclusion criteria for one study postulated that the mecha- etary quality and mealtime routine, this review, including 21 in type 2 dia- nism for the delayed hyperglycemic ef- with referral to a registered dietitian betes, 6 comparing various types of in- fect of dietary fat is FFA-induced insulin for medical nutrition therapy, may be sulin or regimens, and 4 varying insulin resistance with increased hepatic glu- just as important for optimizing glyce- dosing and meals concurrently. The re- cose output (20). Consistent with the mic control (1,2). In the coming years, as maining 29 studies were included in this observed time course of hyperglycemia continuous glucose monitoring becomes review. following higher-fat meals in type 1 1010 Postprandial Glycemia and Insulin Dosing in T1D Diabetes Care Volume 38, June 2015

Table 1—Summary of systematic review: effects of fat, protein, and GI on acute postprandial glycemia in type 1 diabetes Nutritional factor Summary of findings Clinical implications Fat c Seven studies (total 103 subjects) (5,7,8,11,15,18,19). c Increase in dose required for coverage of higher-fat c All studies reported significant differences in meals needs to be individualized. glycemia with addition of fat. c Delicate balance in calculation and timing of insulin c Fat reduces early glucose response (first 2–3 h) (7,8) and action: needs more insulin to prevent late postprandial delays peak blood glucose (5,7,15,18) due hyperglycemia; however, if too much insulin upfront, to delayed gastric emptying. there is a risk for early postprandial hypoglycemia. c Fat leads to late postprandial (.3 h) hyperglycemia (18,19). c Addition of 35 g fat can increase blood glucose by 2.3 mmol/L (15), and in some individuals, 50 g of fat can increase insulin requirements by twofold (19). c Marked interindividual differences in the glycemic effect of fat. Protein c Seven studies (total 125 subjects) (5,8,11,13,15–17). c Protein-only meals (e.g., $230 g lean steak with salad) c All studies reported significant differences in may require a different insulin dosing strategy than for glycemia with addition of protein. protein and carbohydrate meals. c Effect of protein is delayed (effects seen ;100 min postmeal) (11,13,15,17). c Protein has different effects when consumed with and without carbohydrates [e.g., 30 g protein with carbohydrates will affect blood glucose (15,16), whereas at least 75 g protein is needed to see an effect when consumed in isolation (13)]. GI c Seven studies (total 98 subjects) (4,6,8–10,12,14). c Mismatch between insulin action and carbohydrate c All studies reported significant differences in absorption following high-GI foods can be problematic, glycemia with differing GI (same carbohydrate). leading to a rapid glucose spike. c High-GI foods have rapid glucose spike (9,14). c Total carbohydrate content still important: a large c Low-GI foods lower overall glucose response carbohydrate serving of low-GI food will still cause (8–10,12,14), reduce glucose peak (4,9,14), and large glycemic response. increase risk of hypoglycemia (when usual CIR is c Low-GI foods with high fructose and/or sucrose content used) (6,9,10). (e.g., juice) will still produce a rapid glucose spike. See Supplementary Table 1 for details.

diabetes, Gormsen et al. (21) demon- The effect of fat and protein was addi- concentrations were noted after 3–4h strated in healthy subjects that insulin tive, with blood glucose concentrations (11,15,17). resistance develops at least 120 min af- increasing by 5.4 mmol/L at 5 h, the sum ter circulating FFAs increase. of the individual incremental increases Glycemic Index for protein and fat. Paterson et al. (13) Seven studies examining the effects of Protein was the only study examining the effect GI on postprandial glycemia in 98 sub- Seven studies examining the effect of of protein only (in the absence of carbo- jects with type 1 diabetes (range 8–20 dietary protein on postprandial glyce- hydrates and fat) and found that the subjects) were identified (4,6,8– mia in 125 subjects with type 1 diabetes addition of 12.5–50 g protein did not 10,12,14). All studies compared higher- (range 8–33 subjects) were identified significantly affect glycemia, although versus lower-GI foods or meals. Two (5,8,11,13,15–17). All studies kept car- the addition of 75 and 100 g significantly studies concurrently varied the energy bohydrate content consistent, with six increased glucose concentrations, and macronutrient compositions of the adding protein to the test meal, thereby reaching a peak at the conclusion of test meals, confounding the interpreta- concurrently increasing the energy con- the 5-h study and causing an increase tion (4,8). tent. Winiger et al. (17) was the only in glucose concentrations similar to All seven studies reported significant study to keep energy levels constant; that of 20 g carbohydrates given without differences in blood glucose concentra- however, to keep energy and carbohy- insulin. These results suggest that pro- tions, with low-GI foods and meals pro- drates constant, both protein and fat tein has differential effects when con- ducing lower glycemic responses. Three were simultaneously varied. sumed with and without carbohydrates. studies suggested that the risk of mild All included studies reported that All studies agreed that protein affects hypoglycemia is greater with low-GI postprandial glycemia was modified by blood glucose concentrations in the late than with high-GI foods (6,9,10); how- the addition of protein, with all but one postprandial period. When protein was ever, the timing of the episodes was (5) reporting significant differences. the only macronutrient consumed, glu- not reported. One study suggested Smart et al. (15) reported that the addi- cose concentrations began to rise after that low-GI foods are more likely to tion of 35 g protein to 30 g carbohy- 100 min for protein loads of $75 g cause early hypoglycemia, reporting a drates significantly increased blood (13). For meals where carbohydrates correlation between GI and time to hy- glucose levels by 2.6 mmol/L at 5 h. were also consumed, increased glucose poglycemia, with each unit increase in GI care.diabetesjournals.org Bell and Associates 1011

delaying hypoglycemia by 1 min (9). This calculate fat-to-insulin and protein-to- (23) and mixed meals (22). The FII is is consistent with the reduced overall insulin ratios, with the additional insulin based on the physiological insulin de- glycemic response with low-GI foods/ delivered as an extended bolus. Simi- mand evoked by 1,000-kJ food portions meals. larly, another two studies (28,33) calcu- as measured in healthy subjects. Be- lated additional insulin for fat-protein cause food energy is the constant, the QUESTION 2: WHAT PRANDIAL units (FPUs) (defined as 100 kcal or method accounts for all nutritional and INSULIN DOSING STRATEGIES 420 kJ fat and/or protein) (35), with metabolic factors that influence insulin WORK BEST FOR GI, PROTEIN, AND the duration of the extended bolus ad- demand, not just the macronutrient FAT IN TYPE 1 DIABETES? justed from 3 to 6 h depending on the content. With this method, bolus insulin Fifteen studies examining the effects of number of FPUs consumed. A major lim- is calculated for foods not traditionally varying prandial dosing strategies, in- itation of this equation was the high rate covered by insulin, including eggs and cluding various bolus types, timing of of clinically significant hypoglycemia steak. Both studies reported signifi- the meal bolus, and methods to calcu- compared with carbohydrate counting cantly improved glycemia in the 3-h late the bolus dose, on postprandial gly- (28,33). Another study using closed- postprandial period. However, the rela- cemia were identified (14,19,22–34) loop glucose control measured the tively short postprandial monitoring pe- (Supplementary Table 2). additional insulin required to maintain riod may not have detected the delayed glucose control after a high-fat meal glycemic impact of fat and protein. The Fat and Protein (19). This study revealed that the addi- rates of hypoglycemia were high in both Ten studies examining insulin bolus tion of 50 g of fat markedly increased the FII and the carbohydrate-counting dose and delivery patterns and timing insulin requirements over the 5-h post- groups in one study (23), suggesting required to cover meals high in fat prandial period, with one subject that the initial CIR and basal rates were and/or protein were identified (19,22– needing a more than twofold increase. not adequately optimized at study com- 24,26–30,33). This study identified marked interindi- mencement. A pilot study using the FII in Insulin Dose vidual differences in fat sensitivity, high- practice did not find an increased risk of As summarized in Table 2, six studies lighting the need for individualized hypoglycemic events (36). investigated the additional insulin re- incremental dosing for fat and pointing quirements for food or meals high in to the limitations of equations that cal- Bolus Timing and Type fat or protein (19,22,23,28,29,33). The culate individual insulin dosing for fat Whereas all the studies in the literature amount of additional insulin varied and protein as a proportion of the indi- examining the optimal timing of the across studies depending on the vidual’s CIR (28,29,33). prandial bolus (25,26,34) uniformly method used to estimate the bolus in- Two studies investigated another demonstrated that delivering a bolus sulin requirements. Lee et al. (29) used a novel insulin dosing algorithm, the 15–20 min before eating rather than predetermined value of 50% of the Food Insulin Index (FII), to calculate immediately before or after the meal carbohydrate-to-insulin ratio (CIR) to the insulin demand for individual foods significantly improves postprandial glycemia, studies examining the optimal bolus type required to cover higher-fat Table 2—Summary of systematic review: studies evaluating prandial insulin meals showed discrepant results. Four dosing equations and requirements for high-fat meals with or without protein studies reported that the combo wave Methoda Limitation bolus was the most effective method Two studies (28,33) calculated additional High rate of clinically significant (24,27,29,33), whereas one suggested insulin for fat and protein using a complex postprandial hypoglycemia that the standard bolus with insulin de- dosing equation with FPUs (35) and the compared with carbohydrate livered 15 min before the meal was su- subject’s CIR. counting. perior (26) and another reported no One study (29) used a predetermined value of Suboptimal postprandial glucose differences in glucose excursions among 50% of the CIR to calculate fat-to-insulin control. normal, combo wave, and extended bo- and protein-to-insulin ratios and added the luses after meals (30). additional insulin to extended bolus. Methodological differences and is- Two studies (22,23) used the FII to calculate Because of the short postprandial sues confound the evaluation of the additional insulin for high-fat and/or monitoring period, may have -protein meals. Both studies reported failed to detect delayed impact published literature. The duration and significantly improved glycemia in the of fat and protein on glucose split of the bolus types varied across 3-h postprandial period. concentrations. studies. Furthermore, failure to opti- One study (19) used closed-loop control to Findings need to be translated into mize basal rates as part of the study pro- determine additional insulin required to a dosing equation for use in tocol could have affected the findings of maintain postprandial glycemic control clinical practice. some studies. In addition, differences in after a high-fat meal; definitively the GI and macronutrient content of the established that high-fat meals need more test meals and duration and split of the insulin coverage than lower-fat meals with identical carbohydrate content. bolus types in the studies may have con- tributed to the differences in the results. a See Supplementary Table 2 for details. Method used to calculate the additional insulin and the Of note, some studies had a relatively amount given varied across studies. short postprandial monitoring period 1012 Postprandial Glycemia and Insulin Dosing in T1D Diabetes Care Volume 38, June 2015

and, thus, may not have detected the increase in the total insulin dose is re- risk for early postprandial hypoglycemia delayed hyperglycemia from dietary fat. quired for meals containing 40 g satu- from delayed gastric emptying induced rated fat (38). As outlined in Fig. 1, an by the dietary fat. Based on physiologi- Glycemic Index Two studies examining insulin dosing empirical approach is necessary to iden- cal considerations, it seems likely that ’ strategies to improve postprandial gly- tify an individual s nutrient sensitivities the percentage of the total insulin cemia for low-GI meals were identified and optimize insulin dosing for complex dose delivered during the initial phase (14,32). One study in patients using a meals. As a starting point, initial insulin of a combo wave bolus for higher-fat pump demonstrated that a 50:50% doses are calculated based on the car- meals will also need to vary depending combo wave bolus over 2 h resulted bohydrate content of the meal and the on the GI of the meal (more insulin up- ’ in a 47% decrease in the postprandial individual s CIR, with subsequent incre- front for higher-GI carbohydrate loads). glucose area under the curve compared mental dose increases guided by retro- However, as outlined in the previous ’ fi with a standard bolus for a low-GI meal spective review of the patient s previous section, limited scienti c data are avail- (32). Another study examined bolus postprandial glucose responses. This re- able regarding the optimal split and du- strategies for low-GI meals in children view of records can also identify alter- ration of the advanced pump boluses for using multiple daily injections and found native favorite foods with less glycemic various meal types. that insulin administration 15 min before effect, and these insights can be help- In individuals on multiple dose insulin the meal resulted in better postprandial ful in patient education and nutrition therapy, strategies frequently sug- glycemic control than insulin administra- counseling. gested in clinical practice for high-fat/ tion 15 min after the meal (14). Of note, in covering higher-fat meals, low-GI meals (e.g., pizza) are to inject No studies investigated changes in in- the amount of insulin given upfront an additional insulin bolus 1 h after the sulin dose for meals of a higher GI. The needs to be adjusted to minimize the meal to match the delayed absorption mismatch between the action of analog insulins and the rapid glucose spike caused by high-GI meals remains a clini- cal challenge, and practical insulin dosing strategies are needed to address this. QUESTION 3: WHAT ARE THE IMPLICATIONS FOR CLINICAL PRACTICE? From the findings of this review, GI, pro- tein, and fat substantially modulate postprandial glucose concentrations in individuals with type 1 diabetes. The impact on 3-h postprandial glucose concentrations of the addition of 35 g fat and 40 g protein to a meal (37) is equivalent to that resulting from the consumption of 20 g carbohydrates without insulin (13). The addition of 50 g fattoamealcanincreaseinsulinre- quirements by more than twofold (19). This demonstrates that to optimize post- prandial glucose control, some mealtime insulin doses need to be adjusted based on the complete meal composition rather than solely on the carbohydrate content. Whereas mealshigh in fat andprotein require more insulin to control late postprandial hyperglycemia than lower-fat and -protein meals with the same carbohydrate content, as outlined in the previous section, the actual amount of additional insulin and deliv- ery pattern required for high-fat or high- protein meals is not clear. Preliminary studies have indicated that in some individuals, a combo wave bolus extend- Figure 1—Clinical application of insights about the effects of fat, protein, and GI on postprandial ing over 2–2.5 h with as much as a 125% glucose control. care.diabetesjournals.org Bell and Associates 1013

or, alternatively, to cover the meal Finally, clinical education and recom- meal patterns to identify whether mac- with a preprandial injection of regular mendations should reinforce healthy ronutrients are contributing to glycemic with or without analog insulin. How- eating principles and not simply focus fluctuations and to provide individual- ever, these approaches have not been on insulin adjustments. In clinical prac- ized dosing recommendations to pa- investigated in controlled studies. tice,itmaybemorepertinenttoaddress tients for their common meals, thereby Interindividual differences in the im- issues in dietary quality and meal- eliminating the need for patients to rou- pact of macronutrients on postprandial time routines to improve postprandial tinely count carbohydrates and other glucose control are another factor com- control rather than to just focus on macronutrients. pounding the development of stan- fine-tuning mealtime insulin therapy. A An opportunity exists to capitalize on dardized insulin dosing algorithms for registered dietitian can provide a thor- the control algorithms used in artificial complex meals (15,19). To date, studies ough assessment and medical nutrition pancreas (AP) technology to optimize have not been able to identify pheno- therapy (1,2). current open-loop mealtime insulin dos- type markers that correlate with interin- ing. Although AP technology ultimately dividual differences in the glycemic QUESTION 4: WHAT ARE THE has the potential to remove any need to effect of dietary fat. Furthermore, these KNOWLEDGE GAPS, AND HOW calculate insulin dosing, considerable differences do not appear to correlate CAN TECHNOLOGY BE LEVERAGED regulatory and technical hurdles must with the CIR, and there is no scientific TO IMPROVE POSTPRANDIAL be overcome before an AP system will foundation for the use of the CIR as the GLUCOSE CONTROL? become a routine tool for the manage- basis for calculating incremental insulin Further research in this area is crucial to ment of type 1 diabetes. Going back doses to cover fat [as has been sug- provide a stronger evidence base for almost a decade, methods based on pro- gested by others (33)]. clinical decision-making. Table 3 out- portional integral derivative control The mismatch between the rapid glu- lines some of the issues of practical im- strategies have been used to effect cose absorption following higher-GI portance to address in future research. changes in basal insulin delivery with meals and the relatively delayed action This increased knowledge about the im- improved fasting glucose control of subcutaneously administered insulin pact of macronutrients on postprandial (19,40,41). Metabolic models for pre- leads to an initial postprandial glycemic glucose control and improved insulin dicting future glucose excursions have spike that in practice can be difficult to dosing algorithms will be of practical rel- been developed for use with AP systems overcome. As indicated in the ADA evance only if a pathway for translating (36), and the same approach can be used guidelines, substituting low–glycemic the research findings into clinical care is to analyze postprandial glucose patterns load foods for higher–glycemic load established. and provide adaptive open-loop bolus foods may modestly improve glycemic Digital health tools and cloud com- dose recommendations (38). control (2). Increasing the insulin dose puting will open opportunities to de- The introduction of continuous glu- to reduce the spike can lead to overin- velop sophisticated analytic systems to cose monitoring into diabetes care has sulinization in the late postprandial pe- automatically evaluate postprandial glu- revealed the complex picture of post- riod with an associated increased risk cose data and provide dosing recom- prandial glucose patterns in type 1 dia- for hypoglycemia. Use of a superbolus mendations. Carbohydrate counting betes, including rapid glucose spikes (an increased insulin bolus upfront fol- is a challenging aspect to diabetes self- from higher-GI carbohydrates and late lowed by basal rate reduction) has been management, and requiring that fat and postprandial hyperglycemia from dietary proposed to provide a better match be- protein intake also be quantitated and fat and protein, and highlights the limi- tween insulin action and glucose ab- incorporated in insulin dosing decisions tations of the traditional carbohydrate- sorption following higher-GI meals. will create an additional burden that few based approaches for mealtime insulin Technosphere insulin (Afrezza), the in- patients will be able to accomplish. The dosing. New insights about the effect haled prandial insulin recently released need for both practical simplicity and of dietary macronutrients on post- in the U.S., has an action profile that widespread use of advanced dosing al- prandial glucose control has been in- may allow for better coverage of gorithms will ultimately be resolved cluded in the recent ADA standards higher-GI carbohydrates than insulin with the development of data analysis of medical care, which mention that analogs delivered subcutaneously (39). and decision-support tools that evaluate “for selected individuals...the impact The timing of the insulin dose in re- lation to the meal is an important factor in optimizing postprandial control. Early Table 3—Unanswered questions about the effect of dietary fat and protein on delivery of a preprandial insulin bolus postprandial glucose control in type 1 diabetes How much fat does there need to be in a meal before a clinically significant glycemic effect for higher-GI carbohydrate loads is an- becomes apparent? other strategy that can be at least par- Is there a threshold and/or dose response (i.e., more fat requires more insulin)? tially effective in blunting the initial Do all types of fat and protein have similar effects? postprandial spike. In this context, it Are there phenotypic characteristics that can be used as markers to identify individuals should be mentioned that unless the in- with diabetes who are more nutrient sensitive and will require more insulin to cover dividual has gastroparesis, preprandial in- higher-fat/protein meals? sulin is preferable to insulin administered What are the optimal insulin dose adjustments needed for common meals with varying fat during or after the meal for all meal types, and protein content? including low-GI and high-fat meals. 1014 Postprandial Glycemia and Insulin Dosing in T1D Diabetes Care Volume 38, June 2015

of protein and fat on glycemic excursions Practice Guideline. J Clin Endocrinol Metab 17. Winiger G, Keller U, Laager R, Girard J, can be incorporated into diabetes man- 2011;96:2968–2979 Berger W. Protein content of the evening agement” (1). At a practical level, in re- 4. Elleri D, Allen JM, Harris J, et al. Absorption meal and nocturnal plasma glucose regulation patterns of meals containing complex carbohy- in type-I diabetic subjects. Horm Res 1995;44: viewing patient glucose records, the drates in type 1 diabetes. Diabetologia 2013;56: 101–104 possible role of dietary fat and protein 1108–1117 18. Wolever TMS, Mullan YM. Sugars and fat as contributors to glycemic fluctuations 5. Garc´ıa-Lopez´ JM, Gonzalez-Rodriguez´ M, have different effects on postprandial glucose needs to be considered. Although this Pazos-Couselo M, Gude F, Prieto-Tenreiro A, responses in normal and type 1 diabetic sub- review shows that high-fat meals require Casanueva F. Should the amounts of fat and jects. Nutr Metab Cardiovasc Dis 2011;21: protein be taken into consideration to calculate 719–725 more insulin coverage than lower-fat the lunch prandial insulin bolus? Results from a 19. Wolpert HA, Atakov-Castillo A, Smith SA, meals with identical carbohydrate con- randomized crossover trial. Diabetes Technol Steil GM. Dietary fat acutely increases glucose tent and provides some indication of Ther 2013;15:166–171 concentrations and insulin requirements in the additional insulin dosages de- 6. Lafrance L, Rabasa-Lhoret R, Poisson D, patients with type 1 diabetes: implications for carbohydrate-based bolus dose calculation and manded for higher-fat foods (19), we Ducros F, Chiasson JL. Effects of different gly- caemic index foods and dietary fibre intake on intensive diabetes management. Diabetes Care do not yet have simple and easy-to-use glycaemic control in type 1 diabetic patients on 2013;36:810–816 insulin dosing algorithms for dietary fat. intensive insulin therapy. Diabet Med 1998;15: 20.RodenM,PriceTB,PerseghinG,etal. Ongoing research in conjunction with the 972–978 Mechanism of free fatty acid-induced insulin development of digital tools to as- 7. Lodefalk M, Aman J, Bang P. Effects of fat resistance in humans. J Clin Invest 1996;97: 2859–2865 sist patients in dosage decision-making supplementation on glycaemic response and gastric emptying in adolescents with type 1 di- 21. Gormsen LC, Nielsen C, Jessen N, Jørgensen promise solutions to address this need abetes. Diabet Med 2008;25:1030–1035 JO, Møller N. Time-course effects of physio- for a better approach to optimizing post- 8. MacDonald K, Lowe JM, Barker D, Mensch M, logical free fatty acid surges on insulin sensi- prandial insulin coverage that could be Attia J. Effect of popular takeaway foods on tivity in humans. Acta Physiol (Oxf) 2011;201: readily and widely applied in the man- blood glucose levels in type 1 diabetes mellitus 349–356 agement of type 1 diabetes. patients on intensive insulin therapy. Int J Clin 22. Bao J, Gilbertson HR, Gray R, et al. Improv- Pract 2009;63:189–194 ing the estimation of mealtime insulin dose in 9. Mohammed NH, Wolever TMS. Effect of car- adults with type 1 diabetes: the Normal Insulin bohydrate source on post-prandial blood glu- Demand for Dose Adjustment (NIDDA) study. Duality of Interest. No potential conflicts of cose in subjects with type 1 diabetes treated Diabetes Care 2011;34:2146–2151 interest relevant to this article were reported. with insulin lispro. Diabetes Res Clin Pract 23. Bell KJ, Gray R, Munns D, et al. Estimating Author Contributions. K.J.B. contributed to 2004;65:29–35 insulin demand for protein-containing foods us- the conception and design of the study, col- 10. Nansel TR, Gellar L, McGill A. Effect of vary- ing the food insulin index. Eur J Clin Nutr 2014; lected and interpreted the data, contributed ing glycemic index meals on blood glucose 68:1055–1059 to the writing of the first draft of the manuscript control assessed with continuous glucose mon- 24. Chase HP, Saib SZ, MacKenzie T, Hansen and subsequent revisions, critically reviewed itoring in youth with type 1 diabetes on basal- MM, Garg SK. Post-prandial glucose excursions the drafts of the manuscript, and contributed bolus insulin regimens. Diabetes Care 2008;31: following four methods of bolus insulin admin- to intellectual content. C.E.S. contributed to 695–697 istration in subjects with type 1 diabetes. Diabet the design of the study, collected and inter- 11. Neu A, Behret F, Braun R, et al. Higher glu- Med 2002;19:317–321 preted the data, contributed to the writing of cose concentrations following protein- and fat-rich 25. Cobry E, McFann K, Messer L, et al. Timing the first draft of the manuscript, critically meals - the Tuebingen Grill Study: a pilot study of meal insulin boluses to achieve optimal post- reviewed the drafts of the manuscript, and in adolescents with type 1 diabetes. Pediatr Di- prandial glycemic control in patients with type 1 contributed to intellectual content. G.M.S. abetes 20 October 2014 [Epub ahead of print] diabetes. Diabetes Technol Ther 2010;12:173– contributedtotheconceptionanddesignof 12. Parillo M, Annuzzi G, Rivellese AA, et al. Ef- 177 the study, contributed to the writing of the fects of meals with different glycaemic index on 26. De Palma A, Giani E, Iafusco D, et al. Low- manuscript, critically reviewed the drafts of postprandial blood glucose response in patients ering postprandial glycemia in children with the manuscript, and contributed to intellectual with type 1 diabetes treated with continuous type 1 diabetes after Italian pizza “margherita” content. J.C.B.-M. and B.K. critically reviewed subcutaneous insulin infusion. Diabet Med (TyBoDi2 Study). Diabetes Technol Ther 2011; the drafts of the manuscript and contributed 2011;28:227–229 13:483–487 to intellectual content. H.A.W. conceived and 13.PatersonMA,SmartCE,McElduffP,etal. 27. Jones SM, Quarry JL, Caldwell-McMillan M, designed the study, contributed to the data in- Influence of pure protein on postprandial Mauger DT, Gabbay RA. Optimal insulin pump terpretation, contributed to the writing of the blood glucose levels in individuals with type dosing and postprandial glycemia following a first draft of the manuscript and subsequent 1 diabetes mellitus (Abstract). Diabetes 2014; pizza meal using the continuous glucose moni- revisions, critically reviewed the drafts of the 63:A15 toring system. Diabetes Technol Ther 2005;7: manuscript, and contributed to intellectual 14. Ryan RL, King BR, Anderson DG, Attia JR, 233–240 content. Collins CE, Smart CE. Influence of and optimal 28. Kordonouri O, Hartmann R, Remus K, Blasig¨ insulin therapy for a low-glycemic index meal in S, Sadeghian E, Danne T. Benefit of supplemen- References children with type 1 diabetes receiving intensive tary fat plus protein counting as compared with 1. American Diabetes Association. Approaches insulin therapy. Diabetes Care 2008;31:1485– conventional carbohydrate counting for insulin to glycemic treatment. Sec. 7. In Standards of 1490 bolus calculation in children with pump therapy. Medical Care in Diabetesd2015. Diabetes Care 15. Smart CEM, Evans M, O’Connell SM, et al. Pediatr Diabetes 2012;13:540–544 2015;38(Suppl. 1):S41–S48 Both dietary protein and fat increase postpran- 29. Lee SW, Cao M, Sajid S, et al. The dual-wave 2. American Diabetes Association. Foundations dial glucose excursions in children with type 1 bolus feature in continuous subcutaneous insu- of care: education, nutrition, physical activity, diabetes, and the effect is additive. Diabetes lin infusion pumps controls prolonged post- smoking cessation, psychosocial care, and immu- Care 2013;36:3897–3902 prandial hyperglycaemia better than standard nization.Sec.4.InStandards of Medical Care in 16. Uthoff H, Lehmann R, Sprenger M, Wiesli bolus in type 1 diabetes. Diabetes Nutr Metab Diabetesd2015. Diabetes Care 2015;38(Suppl. P. Skipping meals or carbohydrate-free meals 2004;17:211–216 1):S20–S30 in order to determine basal insulin require- 30. Lindholm-Olinder A, Runefors J, Smide B, 3. Klonoff DC, Buckingham B, Christiansen JS, ments in subjects with type 1 diabetes melli- Kernell A. Post-prandial glucose levels following et al.; Endocrine Society. Continuous glucose tus? Exp Clin Endocrinol Diabetes 2010;118: three methods of insulin bolusing: a study in monitoring: an Endocrine Society Clinical 325–327 adolescent girls and in compliance with girls care.diabetesjournals.org Bell and Associates 1015

without diabetes. Pract Diabetes Int 2009;26: 34. Scaramuzza AE, Iafusco D, Santoro L, et al. their caregivers estimate the carbohydrate content 110–115 Timing of bolus in children with type 1 diabetes of meals and ? Diabet Med 2010;27:348–353 31. Lopez P, Smart C, Morbey C, McElduff P, using continuous subcutaneous insulin infusion 38. Bell K, Toschi E, Wolpert H, Steil GM. Use of Paterson M, King BR. Extended insulin boluses (TiBoDi Study). Diabetes Technol Ther 2010;12: model predicted bolus estimation to optimize cannot control postprandial glycemia as well 149–152 insulin dosing strategies covering high fat meals as a standard bolus in children and adults using 35. Pankowska´ E, Szypowska A, Lipka M, (Abstract). Diabetes Technol Ther 2015;17:A-22 insulin pump therapy. BMJ Open Diab Res Care Szpotanska´ M, Błazik M, Groele L. Application 39. Nuffer W, Trujillo JM, Ellis SL. Technosphere 2014;2:e0000050 of novel dual wave meal bolus and its impact on insulin (Afrezza): a new, inhaled prandial insulin. 32. O’Connell MA, Gilbertson HR, Donath SM, glycated hemoglobin A1c level in children with Ann Pharmacother 2015;49:99–106 Cameron FJ. Optimizing postprandial glycemia in type 1 diabetes. Pediatr Diabetes 2009;10:298– 40. Dauber A, Corcia L, Safer J, Agus MS, Einis S, pediatric patients with type 1 diabetes using insulin 303 Steil GM. Closed-loop insulin therapy improves pump therapy: impact of glycemic index and pran- 36. Bell K, Gray R, Munns D, et al. Food Insulin glycemic control in children aged ,7years: dial bolus type. Diabetes Care 2008;31:1491–1495 Index (FII) vs. traditional carbohydrate counting a randomized controlled trial. Diabetes Care 33. Pankowska´ E, Błazik M, Groele L. Does the for glycemic control in adults with type 1 diabe- 2013;36:222–227 fat-protein meal increase postprandial glucose tes: a 3-month pilot study (Abstract). Diabetes 41. Steil GM, Agus MS. Critical illness hyper- level in type 1 diabetes patients on insulin 2014;63(Suppl.):A189 glycemia: is failure of the beta-cell to meet pump: the conclusion of a randomized study. 37. Smart CE, Ross K, Edge JA, King BR, McElduff P, extreme insulin demand indicative of dysfunc- Diabetes Technol Ther 2012;14:16–22 Collins CE. Can children with Type 1 diabetes and tion? Crit Care 2009;13:129