|Home | About | Journals | Submit | Contact Us | Français|
In the present study, we tested the hypothesis that calories consumed at a prior meal (lunch) may impair glycemic control after a subsequent meal (supper) even if the pre-supper glucose did not differ regardless of the size of the lunch meal.
Nine subjects with Type 1 diabetes using continuous subcutaneous (s.c.) insulin infusion (CSII) therapy were studied on two separate days. Lunch (1200 h) was randomly assigned as 25% or 50% of the usual daily intake on alternate study days. The CSII was stopped at 1000 h on the day of the study and glucose was controlled until supper by adjusting the rate of intravenous (i.v.) insulin based on glucose measurements every 15 min. The CSII was restarted 1 h before supper and i.v. insulin discontinued 15 min before the first bite of supper. An identical supper meal and pre-supper s.c. bolus of short-acting insulin were administered on both visits.
Pre-supper glycemia was nearly identical on each of the two study days. However, the post-supper glucose area under the curve was 27.5% greater on the day of the antecedent large lunch compared with the small lunch (P = 0.0039).
For optimal postprandial glucose control, people with Type 1 diabetes may need to consider not only anticipated meal calories, but also prior food intake, a practice not commonly recommended based on currently used insulin dosing algorithms.
To assist in controlling glycemia, various algorithms have been developed to guide insulin therapy.1–3 Typically these involve subcutaneous (s.c.) basal and bolus insulin delivery through an insulin infusion pump or by multiple daily injections. Patients with Type 1 diabetes are commonly taught to adjust scheduled premeal insulin dosing according to anticipated food intake, often with correction based on pre-meal blood glucose levels.
One factor generally not considered in existing algorithms is the caloric amount consumed in recent hours prior to the meal in question. However, there is theoretical reason to believe that this may be important. Both hepatic and muscle glycogen content are increased after a mixed meal and can persist for several hours.4,5 Moreover, fat intake raises circulating fatty acid concentrations and increases lipid deposition in liver and skeletal muscle, which can impair insulin-sensitive glucose utilization.6 The translational implications of this are not clear, but suggest that prior food intake, in larger compared with smaller amounts, may impair the blood glucose response to insulin at a subsequent feeding, with important consequences for the control of postprandial glycemia.
Based on these considerations, we hypothesized that greater caloric intake at an antecedent meal will result in greater glucose excursion following a subsequent meal. To test this hypothesis, we administered lunch feedings of different caloric content on each of two study days to subjects with Type 1 diabetes treated using continuous s.c. insulin infusion (CSII) pumps. Post-lunch glycemic control was maintained using intravenous (i.v.) insulin targeting equivalent pre-supper blood sugar on each of the two study days. Supper feedings, consisting of the same nutrient composition on each of the two study days, were then administered, along with equivalent doses of pre-supper bolus insulin. The major outcome variable was post-supper glucose excursion.
Procedures followed were in accordance with the ethical standards of the Univerisity of Iowa. The study was approved by the University of Iowa Institutional Review Board and was performed in the Clinical Research Unit (CRU) of the Institute for Clinical and Translational Science. Nine subjects, between 18 and 60 years of age, with Type 1 diabetes were recruited to the study. Patient demographics are listed in Table 1.
Inclusion criteria included Type 1 diabetes for at least 5 years (diagnosed after assessment by an endocrinologist), current use of continuous CSII therapy, glycosylated hemoglobin (HbA1c) <8% within the past 3 months, and body mass index <30 kg/m2. Exclusion criteria included pregnancy or lactation, renal dysfunction, hypoglycemic unawareness, gastroparesis, or any active medical or surgical disorder likely to interfere with safe participation in or the validity of the results of the study.
After initial contact by a nurse coordinator by telephone or in person, potential participants were asked to sign informed consent and were interviewed and examined by a study investigator. Eligible participants took part in 2 day-long visits to our CRU. Subjects were asked to strictly follow their usual insulin and diet regimen for 72 h prior to the visit and were to eat the same breakfast on each of the two study days. If the pre-breakfast blood sugar level was >200 mg/dL, the subject was asked to reschedule; if the blood sugar level was <200 mg/dL, the patients was asked to come to the CRU, arriving at 0900 h.
An i.v. line was placed in each arm, one for insulin infusion and the other for venous blood sampling at 15 min intervals. Blood glucose levels were determined in blood samples with a Beckman Coulter (Fullerton, CA, USA) automated glucose analyzer. The CSII was suspended at 1000 h. Human regular insulin was administered i.v. and the rate was adjusted targeting a pre-lunch glucose level of 90–130 mg/dL and safe return of the post-lunch glucose towards a target of 90–130 mg/dL. The i.v. insulin infusion was initiated at a rate equivalent to the subject’s usual CSII basal rate at 1000 h. Thereafter, adjustments were made based on investigator judgment. Decisions were guided, but not absolutely determined, by an empirical algorithm calling for raising or lowering of the insulin infusion rate by 25% if glucose levels were >110 or <90 mg/dL, respectively, and for raising the insulin infusion rate by 33% if the glucose level was >125 mg/dL.
At 1200 h, on alternate, randomly assigned study visits, subjects received a lunch meal of either 50% or 25% of the usual total daily caloric requirement as assessed by a General Clinical Research Center nutritionist at screening. The lunch foods and the percentage of carbohydrate, fat, and protein were identical on each visit, differing only in portion size. The supper meal on both study days was identical and provided 25% of the usual total daily caloric intake.
In addition to the continuous i.v. infusion described above, post-lunch i.v. insulin was administered as three bolus doses at 15, 30, and 45 min after the onset of the lunch feeding. Each i.v. bolus dose consisted of 16.7% of the usual s.c. lunch dose so that the total bolus amount was 50% of the usual s.c. lunch dose. The usual s.c. dose (used to calculate the i.v. bolus doses) was greater for the larger lunch because the subjects routinely adjusted pre-lunch insulin according to intended carbohydrate intake.
The CSII was restarted 1 h before supper and the i.v. insulin was discontinued 15 min before the first bite of the supper meal. The usual pre-supper s.c. insulin dose (identical on each study day) was given through the insulin pump with the first bite of food. Glucose monitoring continued every 15 min for 2 h after the supper meal. The major outcome variable was the post-supper glucose excursion, defined as the area under the curve (AUC) for glucose 0–2 after the start of supper feeding (AUC0–2). Plasma free fatty acids were determined in a venous sample obtained prior to the supper meal and were measured using a colorimetric kit purchased from Roche Diagnostics (Mannheim, Germany).
Data were compared by t-test using the Wilcoxon signed-rank method, as described in the figure legends. Comparisons were within individual subjects following the larger or smaller lunch feeding, so that each individual served as his or her own matched control.
As indicated in Table 2, subjects consumed almost all calories provided. Figure 1(a,b) shows blood glucose levels and insulin infusion rates on each study day according to clock time up to 1600 h. Supper feedings were initiated from 4 to 5.5 h after the onset of lunch, dependent on achieving a target glucose level of 90–130 mg/dL (or at least <180 mg/dL) by 5.5 h. The actual interval between lunch and supper was 4.28 ± 0.13 h after the larger lunch and 4.36 ± 0.16 h after the smaller lunch (P = 0.88). Figure 2(a,b) shows insulin infusion rates and glucose values plotted against time relative to the onset of supper. Because not all supper feedings were administered at the same time, the values in Fig. 2(a,b) do not exactly correspond to those in Fig. 1(a,b). The AUC0–2 is shown in Fig. 2(c) and demonstrates a significant increase in glucose excursion after supper following the larger lunch compared with the smaller lunch.
Our protocol called for oral carbohydrate for any glucose <60 mg/dL. No subject required oral carbohydrate from 45 min before lunch until the end of the study. On the study day of the lower calorie lunch, four subjects required glucose in the pre-lunch period before 1115 h and received a total of 69 g carbohydrate. On the study day of the higher calorie lunch, three subjects required glucose in the period before 1115 h and received a total of 54 g carbohydrate.
Pre-supper free fatty acid concentrations following the larger and smaller lunches were 59.2 ± 18.9 and 95.3 ± 46.7 μmol/L, respectively. These values varied considerably and did not differ significantly.
The results of the present study show greater post-supper glucose excursion following an antecedent lunch of high caloric content compared with a lunch of lower content. We could find no reports of similar past studies. Older studies examined the dietary effects on insulin needs using an “artificial pancreas”, termed a Biostator, to regulate glycemia by infusing i.v. insulin based on continuous glucose sensing.7,8 These studies showed that meal macronutrient composition and the time of feeding affect insulin needs. In addition, it is well established that equivalent carbohydrate administration may result in differing meal glucose excursion based on the specific nature (glycemic index) of the food provided.9 Moreover, it is clear that chronic dietary perturbations affect insulin sensitivity.10 However, we could find no study that examined the effect of directly antecedent meal calories on subsequent meal glucose excursion at equivalent pre-meal glycemia.
In the present study, pre-supper glucose, supper food intake, and subcutaneous pre-supper insulin dosing were equivalent after each of the antecedent lunch meals. Moreover, relative nutrient composition at the lunch meal did not differ; the only difference was portion size. Hence, the difference in supper glucose excursion resulted from more calories at lunch. Although post-lunch glycemia was not controlled in the non-diabetic range, these readings do appear representative of those expected in the daily lives of people with Type 1 diabetes.
One limitation of the present study is that our data do not discriminate between the effects of one particular macronutrient type versus another. Another limitation is that there were some asymptomatic low glucose readings in the early portion of the study days for both the high- and low-caloric lunches. However, these all occurred before the lunch meal and required few extra calories, which did not differ according to study day.
There appeared to be a greater immediate pre-supper downward momentum of blood glucose following the larger compared with the smaller lunch (Fig. 2b). However, this would not alter data interpretation because the post-supper glycemic excursion following the larger lunch, if anything, would be less if downward glucose momentum carried into the supper meal.
Our intent was to perform this study in a clinically meaningful manner. However, we do acknowledge the limitation that we used i.v. insulin rather than s.c. to maintain post-lunch glycemia. We believe our objective of matching pre-supper glycemia would have been much more difficult if we were dependent on adjustments of s.c. insulin administration, even by CSII. Thus, we examined the effect of differing antecedent food intake with attention to providing enough insulin so that glucose concentrations by the subsequent meal were essentially equivalent. This, in fact, is what patients attempt to achieve on a daily basis by taking meal-time insulin according to carbohydrate and calories consumed.
We considered that higher circulating free fatty acids after the larger lunch may explain the greater glycemic excursion following supper by decreasing muscle glucose utilization.6,11 However, plasma concentrations of free fatty acids did not significantly differ and, if anything, were lower following the larger lunch. This may have resulted from the greater insulin infusion rates required after the larger of the antecedent lunch meals. So, although we can only speculate, it remains possible that even without an increase in pre-supper circulating free fatty acids there could have been an increase in intramyocellular or intrahepatic fat, which would decrease insulin sensitivity at the time of the subsequent meal. In this respect, resistance to insulin is associated with increased triglyceride in the liver and skeletal muscle.12,13 Moreover, postprandial triglyceride accumulation in the liver and skeletal muscle has been documented using 13C magnetic resonance spectroscopy, at least in subjects with Type 2 diabetes, and postprandial accumulation of liver triglyceride has been documented in non-diabetic subjects.14
In conclusion, caloric intake at an antecedent meal impacts glucose control following a subsequent meal. For optimal postprandial glucose control, patients with Type 1 diabetes may need to consider not only anticipated meal calories, but also prior food intake, a practice not commonly recommended based on currently used insulin algorithms.
This work was supported by grant M01-RR00059 from the National Institutes of Health, National Center for Research Resources, and General Clinical Research Centers Program, by funds donated by the Iowa Affiliate of the Fraternal Order of the Eagles, and by VA Medical Affairs medical research.