Dr Ann-Charlotte Orre-Pettersson, Department of Medicine and Care, Internal Medicine, The University Hospital, S-581 85 Linköping, Sweden (fax: +46–13222804; e-mail: firstname.lastname@example.org).
Objectives. To evaluate how a snack influences the blood glucose profile during treatment with preprandial regular human insulin.
Design. In a randomized study a mid-morning snack either was or was not served. Insulin was given 30 min before the usual breakfast of the patients. Plasma free insulin and blood glucose were repeatedly determined for 5 h.
Setting. Outpatient clinic at a university hospital.
Subjects. Twenty patients with type 1 diabetes treated with multiple injections of regular insulin (Actrapid®) and eight non-diabetic subjects.
Interventions. A mid-morning snack either was or was not served 2 h after the usual morning insulin injection.
Main outcome measures. A difference in the blood glucose profile after a mid-morning snack.
Results. With a snack there was no difference in blood glucose fasting and at 12.30 h, whilst without a snack there was a decrease of almost 4 mmol L−1, several patients experienced low blood glucose and three had hypoglycaemia. An extended peak of free insulin was reached 30 min after the insulin injection with a slow decrease to the fasting level after 5 h. After the insulin injection a significant decrease in blood glucose occurred within 30–45 min
Conclusions. A snack 2 h after the insulin injection results in a smoother blood glucose profile and reduces the risk of hypoglycaemia in patients with type 1 diabetes treated with preprandial regular human insulin. Furthermore, the recommended interval of 30 min between insulin injection and a meal may be too long.
Insulin treatment of patients with type 1 diabetes may be regarded as a substitution therapy aiming at providing insulin according to the physiological requirements. When regular human insulin is injected subcutaneously, the insulin peak obtained is blunted and extended compared with the meal-related insulin peaks in healthy subjects, resulting in an insulin profile with prandial hypoinsulinaemia and postprandial hyperinsulinaemia [ 1, 2]. The slow absorption of subcutaneously injected regular human insulin is mainly due to the association of the insulin molecules in hexamers at high concentrations (10−4–10−3 mol L−1) [ 3]. Due to the slow absorption of subcutaneously injected regular human insulin, it is commonly recommended to inject the insulin 30 min before the meals [ 4]. To take a snack after the main meal is commonly recommended in dietary advice for patients with diabetes [ 5]. In a study by Primavesi et al. [ 6], intake of a snack 2–3 h after breakfast and lunch was compared with no snack during treatment with a mixture of short- and intermediate-acting lente insulin given before breakfast and dinner in children with type 1 diabetes. No difference in the blood glucose profile was found. In other studies the use of different kinds of snacks has been compared, but giving no snack at all was not included in these protocols [ 7, 8]. There is little scientific information about if and when a snack should be taken in multiple-injection insulin regimens with preprandial regular insulin. We have investigated how such a snack influences the blood glucose profile.
Materials and methods
Nine men and 11 women with type 1 diabetes, aged 35 ± 2.0 (mean ± SEM) years and with a duration of 19 ± 2 years, took part. HbA1c was 6.4 ± 0.2% (normal range 3.5–5.4%) and body mass index (BMI) was 23.1 ± 0.5 kg m−2. They were treated with a four to five dose insulin regimen with preprandial regular insulin and intermediate-acting insulin at bedtime. The total insulin dose was 40.2 ± 2.5 U day−1 (range 28–53 U), with 38% given as intermediate-acting insulin. The patients were matched with eight non-diabetic subjects (four men and four women) aged 35 ± 2 years, with body weight 69.1 ± 2.9 kg and BMI 23.1 ± 0.2 kg m−2.
The patients and non-diabetic subjects arrived fasting at 07.00 h, and a cannula (Venflon, Viggo, Helsingborg, Sweden) was inserted into a vein. After the initial blood sampling the patients took their usual morning insulin dose, on average 9.1 ± 0.6 U (Actrapid® 100 U mL−1; Novo-Nordisk AS, Denmark) at their usual injection site (the abdominal wall), and 30 min later their usual breakfast was served. The energy content of the breakfast was 1753 ± 100 kJ and the nutrient content was 57.7 ± 3.8 g carbohydrate (55.8 ± 1.2 E%, percentage of total energy intake in the breakfast), 11.7 ± 0.7 g fat (25.0 ± 1.1 E%) and 19.9 ± 1.2 g protein (19.3 ± 0.4 E%). The patients were randomized either to receive or not to receive a snack 2 h after the insulin injection. The snack consisted of bread, margarine, cheese and coffee or tea. The energy content was 628 kJ and 16 g carbohydrates. The study was repeated within 3 weeks the other way around. Blood samples for glucose and free insulin were collected and the profiles are illustrated in Fig. 1. If hypoglycaemic symptoms occurred and/or if hypoglycaemic blood glucose values < 2.0 mmol L−1 were registered, i.v. glucose was given. The non-diabetic subjects were served breakfast with a similar composition and energy content as the mean of the breakfast served to the patients. No snack was served. The study was approved by the local ethics committee and the patients gave their informed consent.
Blood glucose was measured by a hexokinase glucose-6-phosphate dehydrogenase method. HbA1c was analysed by reversed phase partition chromatography on a cation exchanger using high-pressure liquid chromatography (HPLC; Auto A1c HA 8110, Boehringer Mannheim, Germany). Free insulin was determined using the Phadeseph Insulin RIA kit (Pharmacia Diagnostics AB, Uppsala, Sweden) after polyethylene glycol precipitation of antibody-bound insulin using the method described by Kuzuya et al. [ 9], as modified by Arnqvist et al. [ 10]. The intra-assay CV for free insulin is 4.2–5.5% at insulin concentrations of 54–600 pmol L−1 and the interassay CV is 7.8–10.7%. The detection limit is 2 pmol L−1 and analytical recovery is 99.3%.
The results are means ± SEM. Data were tested with two-tailed Student's t-test for paired data and with Wilcoxon's signed rank test where appropriate.
The insulin profile obtained in the patients differed considerably from that in the non-diabetic subjects ( Fig. 1a). The rise in free insulin was similar to begin with and half of the maximum of the insulin peak was already reached 15 min after insulin injection, but the maximal rise was significantly smaller than in the healthy controls. In fact, there was little further increase in free insulin concentration 30 min after the insulin injection and a plateau without a clear peak was obtained. The free insulin concentration thereafter slowly decreased to the fasting concentration at 12.30 h. There was no significant difference in the insulin profiles with or without snack ( Fig. 1a). In the patients, the blood glucose concentration decreased during the 45-min period from the insulin injection to 15 min after the start of breakfast from 8.9 ± 0.7 mmol L−1 to a nadir of 7.6 ± 0.8 mmol L−1 (P= 0.06). The corresponding values on the day when the snack was not served were 9.6 ± 0.7 and 8.4 ± 0.8 mmol L−1 (P < 0.01), respectively. Before the intake of the snack there was no difference in the blood glucose concentration, but thereafter a clear-cut difference in the blood glucose profiles was found. There was no significant difference in blood glucose concentration between fasting and 12.30 h with the snack, whilst without the snack there was a decrease of almost 4 mmol L−1 ( Fig. 1b). This resulted in a significant difference in blood glucose area under the curve (P= 0.02) during the period after the snack until the end of the study. None of the patients were given i.v. glucose owing to hypoglycaemia when the snack had been served, while three patients received 20 mL of 300 mg mL−1 solution at 10.45, 11.45 and 12.00 h when no snack was served according to the study design. Still, six patients had blood glucose concentration < 4 mmol L−1 at 12.30 h, but with the snack this occurred in only three patients. In the non-diabetic subjects, fasting blood glucose concentration was 3.7 ± 0.1 mmol L−1 and the maximum value after breakfast 4.7 ± 0.2 mmol L−1.
We found that a snack 2 h after the insulin injection has a pronounced effect on the blood glucose profile and reduces the risk of hypoglycaemia in patients with type 1 diabetes treated with preprandial regular human insulin. Administration of a snack also results in a smoother blood glucose profile. This provides the basis for introducing a snack in the treatment with preprandial regular insulin.
Many patients leave out the snack when measuring a high blood glucose concentration such as that found 2 h after the insulin injection. As the insulin concentration beyond this time is unphysiologically high, it is important to take the snack to avoid later hypoglycaemia. The mean blood glucose concentration that was achieved at the end of the study without a snack might be considered normal, but as there is a high variation in the uptake of insulin, the treatment must take into account these fluctuations to avoid hypoglycaemia on certain days, and several of our patients who did not take a snack experienced low blood glucose values. Moreover, the difference would have been even greater if i.v. glucose had not been administered when hypoglycaemia occurred. During the profile days the patients had somewhat high fasting blood glucose concentrations in spite of the fact that they were in good metabolic control with HbA1c values similar to those of the DCCT study [ 11, 12].
Possible approaches to take other than administering a snack might be to lower the morning insulin dose, to increase the size of the breakfast or to use a shorter-acting insulin analogue. Lowering of the morning insulin dose might result in a problem of controlling the postprandial glucose excursion after breakfast. In the present study we could not prevent such a rise, and lowering of the morning insulin dose would be disadvantageous, as it is well known that reduction of the insulin dose will result in lower glucose utilization and higher blood glucose concentration after a meal [ 1]. Furthermore, the insulin concentration at the end of the morning had returned to the baseline fasting level which indicates that the insulin doses used were adequate. In certain instances, hypoinsulinaemia may occur during the late morning if the dose is lowered, as this will also affect the duration of the dose. Increasing the size of the breakfast without changing the insulin dose would exaggerate the breakfast glucose excursion, which is already difficult to minimize, and it is also uncertain how much it would affect the glucose concentration during the late morning. The third alternative is to use a shorter-acting insulin analogue instead of usual regular insulin [ 13]. This might theoretically diminish the need for a snack, but it is uncertain whether this would hold true in practice. Owing to the shorter duration of these preparations, combination with intermediate-acting insulin might be needed [ 14] and this will probably increase the need for a snack. The time for the snack was selected according to the action profile of regular insulin. The greatest glucose utilization after a subcutaneous bolus injection of regular insulin in the abdominal wall is observed after 2–3 h [ 13, 15]. Intake of a snack is therefore recommended 2 h after injection. A later intake coincides with a decreasing insulin action [ 13] and might therefore cause late hyperglycaemia.
The increase of plasma concentrations of free insulin is rapid when using regular insulin, and after 15 min half the peak concentration was already noted. This means that a clearly higher insulin concentration is achieved during the interval between injection and food intake in patients with type 1 diabetes than in healthy subjects. Also, the onset of action can be observed soon after injection as was demonstrated by the lowering of blood glucose 30–45 min after injection, in agreement with Pampanelli et al. [ 16] and Dimitriadis and Gerich [ 4], but not Kinmonth and Baum [ 17] when a mixture of lente and regular insulin was given. It ought to be mentioned that mixing regular insulin with lente insulin may change the proporties of regular insulin [ 18]. The practical implication of these results is that hypoglycaemia might evolve if regular insulin is injected 30 min before the meal if the blood glucose concentration at the time of injection is not elevated. In that case, the time interval between injection and the meal might need to be reduced [ 19]. It should also be noted that there is little further increase in free insulin concentration 30 min after the insulin injection.
In summary, a snack 2 h after insulin injection has a pronounced effect on the blood glucose profile and reduces the risk of hypoglycaemia in patients with type 1 diabetes treated with preprandial regular human insulin. Furthermore, the recommended interval of 30 min between insulin injection and a meal may be too long.
Financial support for this study was obtained from the County Council of Östergötland, the Swedish Medical Research Council (04952), the Swedish Diabetes Association and Novo-Nordisk Sverige AB.