Postprandial hyperglycemia is associated with platelet activation. We thus investigated if meal-induced platelet activation could be attenuated by meal insulin. A randomized, double-blind, cross-over study was performed to compare postprandial platelet activation after premeal injections of placebo or insulin aspart (0.1 and 0.2 units/kg) in 18 patients with type 2 diabetes mellitus (T2DM). Platelet activation was assessed by flow cytometry, without and with stimulation by the thromboxane analog U46619 or ADP. Measurements were before and after premeal blood glucose standardization (to 6–7 mmol/L by insulin infusion, if needed) and at 90 min after the meal. Premeal insulin reduced postprandial hyperglycemia by 2–3 mmol/L compared with placebo. Postmeal insulin levels were doubled with placebo and further elevated with insulin injections. The standardized meal enhanced U46619-induced platelet P-selectin expression by 23% after placebo; this response was more than doubled after premeal insulin. U46619-induced fibrinogen binding was unchanged after meal intake with placebo but was markedly enhanced (by ∼50–60%) after premeal insulin. Postprandial platelet activation correlated positively to postprandial insulin levels and inversely to glucose levels. Premeal insulin infusion was also associated with platelet activation. Our results suggest that postprandial insulin rather than glucose accounts for postprandial platelet activation in T2DM patients.
Postprandial hyperglycemia (PPH) is thought to be an independent risk factor for cardiovascular complications in diabetes mellitus (DM) (1). PPH is associated with increased oxidative stress (2), endothelial dysfunction (3), and platelet activation (4–6). In addition, control of PPH is essential to achieve the recommended target HbA1c levels (7).
Several new drugs targeting PPH have been developed (1), including the rapidly acting insulins. Insulin aspart reduces PPH and improves endothelial dysfunction (3). We have found that a carbohydrate-rich meal increases platelet activation and platelet–leukocyte conjugation stimulated by ADP (4) and, in particular, the thromboxane A2 analog U46619 (5) in type 2 DM (T2DM). Because healthy control subjects developed neither PPH nor increased platelet reactivity after this meal (5), we hypothesized that the postprandial platelet hyperactivity was related to the hyperglycemia. Treatment with repaglinide or glibenclamide reduced PPH only mildly and did not counteract the postprandial platelet activation (4,5).
The current study was designed to evaluate if platelet activation after a carbohydrate-rich meal is related to the PPH and thus can be attenuated by premeal insulin treatment in patients with T2DM.
RESEARCH DESIGN AND METHODS
The study was a randomized, double-blind, cross-over study comparing postprandial platelet activation after three premeal treatments: placebo (saline) or either of two doses of subcutaneous insulin aspart (Novorapid, Novo Nordisk, Bagsvaerd, Denmark), 0.1 or 0.2 units/kg body weight. The study was performed according to good clinical practice rules and was approved by the regional ethical review board and the Swedish Medical Products Agency (EudraCT No. 2006-007031-27). All patients gave written informed consent.
Patients.
T2DM patients were eligible if younger than age 70 years, without known cardiovascular disease, and treated with insulin and/or metformin. HbA1c levels were between 52 and 83 mmol/mol (6.9–9.7% by the Diabetes Control and Complications Trial method or 6–9% by the Mono-S method) Exclusion criteria were contraindication to insulin, acute or chronic kidney or liver diseases, or treatment with sulfonylurea, glitazones, antiplatelets, or anticoagulants.
Study design.
Each patient participated in the study on three occasions separated by at least 2 weeks. Patients were instructed to abstain from food, tobacco, or nicotine on days of the experiment. No blood glucose–lowering medication was given in the mornings before experiments. The basal insulin dose was halved the evening before to minimize the risk of hypoglycemia. Patients arrived in the morning and rested in a semireclining position for 30 min, after which baseline blood samples (baseline 1 [B1]) were drawn for platelet function, glucose, and insulin levels. Premeal blood glucose levels were “standardized” to 6–7 mmol/L on every occasion by intravenous infusion of insulin aspart, if needed. A second set of blood tests was taken (baseline 2 [B2]) 15 min after the target premeal glucose level was achieved. Patients who already had glucose below 7 mmol/L at B1 rested 45 min before B2 measurements. After B2 sampling, one of the three treatments—insulin aspart at 0.1 or 0.2 units/kg body weight or placebo (in a randomized order)—was injected subcutaneously immediately before the standardized meal (54% carbohydrates, 30% fat, and 16% protein, 620 kcal) (4,5) was ingested during a 15-min period. Patients rested another 90 min before the postprandial (PP) sampling.
Blood tests.
At B1, after premeal glucose titration (B2), and 90 min after the meal (PP), a complete blood cell count, triglyceride levels, and platelet activation measurements were taken. Plasma glucose (glucose oxidase technique; Hemocue Glucose 201+, Hemocue AB, Ängelholm, Sweden), insulin (radioimmunoassay [6]), and C-peptide (Human C-Peptide RIA, Millipore, Billerica, MA) were measured before and repeatedly after the meal.
Flow cytometry.
Platelet P-selectin expression and fibrinogen binding were measured by whole blood flow cytometry (8,9) and are reported as percentages of positive cells in the platelet population. Platelet agonists were U46619 (thromboxane A2 analog, 0.3 μmol/L), ADP (1 μmol/L), and a collagen-related peptide (CRP-18/i; 0.3 µg/mL), which activates platelet glycoprotein VI (GPVI) (10).
Statistics.
Continuous data were compared by repeated-measures ANOVA, with Fisher post hoc testing to control for multiplicity. Differences between two treatments were analyzed by the Student paired t test after validation for normal distribution (Shapiro-Wilk test). Wilcoxon matched pairs test was used for skewed variables. The Pearson correlation coefficient was used to test independence between variables. Trend analyses were performed by means of regression analysis. Mean values and SDs are given unless otherwise specified. Analyses were done using STATISTICA 7 software (StatSoft, Tulsa, OK), and probability (P) values are given.
We estimated that 15 subjects were required to detect a 20% effect of interventions at the 5% level of significance using ADP-induced platelet P-selectin expression. When stimulated by 0.3 μmol/L U46619, the percentage of P-selectin–positive platelets increased by 9.66% (Δ value) after the meal in ref. 5. Assuming an SD of 6.63 for a 50% reduction of this Δ value, we would need 15 patients to achieve the desired statistical power (α = 0.05; 1−β = 0.8).
RESULTS
Patients.
Of 27 screened patients, 18 were enrolled. Reasons for exclusion were failure to meet HbA1c criteria, anticoagulant treatment, thrombocytopenia, or technically suboptimal venipunctures. One patient withdrew consent after the first visit. Patient characteristics are presented in Table 1.
Glucose, insulin, and C-peptide levels.
Basal fasting levels (B1) of glucose, insulin, and C-peptide were similar on the three occasions (Fig. 1, Table 2). Insulin infusion was needed in 16 patients on 43 occasions to reduce premeal glucose to 6–7 mmol/L; 2 patients needed no insulin. After the glucose standardization procedure (B2), insulin levels increased, and glucose and C-peptide levels decreased compared with B1. Premeal glucose, insulin, and C-peptide levels at B2 were very similar on the three experimental occasions (Fig. 1, Table 2).
Premeal insulin reduced postprandial plasma glucose significantly (by 2–3 mmol/L at PP) compared with placebo (Fig. 1), but the difference between doses was not significant (P = 0.16). Postprandial plasma insulin increased with placebo (representing endogenous insulin secretion) and was further increased after premeal injections of insulin aspart. Maximal levels of insulin were achieved 30–60 min after the meal (Fig. 1); 90 min after the meal (PP), plasma insulin was higher after 0.2 compared with 0.1 units/kg insulin aspart (P = 0.004). Postprandial C-peptide levels increased after the meal, with the greatest increment after placebo (Fig. 1, Table 2).
Platelet activation after insulin infusion.
Six patients did not require insulin infusion on 11 different occasions (B1 glucose was below 7 mmol/L); these occasions were not included in this analysis. Significant platelet activation occurred after intravenous insulin infusion on 43 occasions. P-selectin–positive platelets were mildly but significantly enhanced by 10% in unstimulated blood (from 2.3 ± 0.6 to 2.6 ± 0.7; P = 0.005) and by 7.5% in ADP-stimulated samples (from 56.3 ± 8.0 to 60.6 ± 8.5; P < 0.001), whereas marked enhancement was seen with U46619 (67%, from 19.1 ± 13.7 to 31.7 ± 22.1; P < 0.001).
Platelet activation after the meal.
Platelet activation was enhanced 90 min after the meal (PP), most markedly in response to U46619 stimulation (Fig. 2). Modest but significant meal effects on platelet P-selectin expression were also found in unstimulated and ADP-stimulated samples (P = 0.006 and P < 0.001, respectively, in the overall ANOVAs). Thus, ADP increased P-selectin expression from 2.3 ± 0.7% to 56.2 ± 8.0% at baseline and from 2.5 ± 0.6% to 62.4 ± 8.1% after the meal on the placebo occasion. GPVI stimulation increased P-selectin expression to ∼30% and fibrinogen binding to ∼20% on all occasions (i.e., without effects of meal intake or insulin treatment, results not shown).
The following net meal effects on platelet activation (PP vs. B2) could be seen after placebo injection: U46619–stimulated P-selectin expression increased by 23% (P = 0.02, Fig. 2A) and ADP-induced P-selectin expression increased by 5% (P = 0.005, data not shown). Other measures, such as unstimulated platelet P-selectin expression and fibrinogen binding, did not increase after placebo injection (Fig. 2).
Postprandial platelet activation was much more consistent and pronounced after premeal insulin compared with placebo, with no significant difference between 0.1 and 0.2 units/kg insulin. There was no significant interaction term for treatment effects on platelet P-selectin expression in unstimulated or ADP-stimulated samples, but U46619-stimulated samples showed larger responses after premeal insulin (P = 0.013 for interaction term in the ANOVA; Fig. 2A). Postprandial P-selectin expression increased by 10% (P = 0.01) and 15% (P = 0.02) in unstimulated samples after premeal injections of 0.1 and 0.2 units/kg insulin, respectively. Increases were 57% and 54% in samples stimulated with U46619 (P < 0.001 for both; Fig. 2A) and 5% (P = 0.006) and 9% (P < 0.001) with ADP (data not shown).
Meal effects on platelet fibrinogen-binding were more complex: unchanged in unstimulated samples, slightly but significantly decreased with ADP-stimulation, and increased by 46–49%, on average, in U-46619-stimulated samples after premeal insulin (P = 0.007 for meal effects in the ANOVA; P = 0.045 for treatment effect; Fig. 2B).
The postprandial enhancement of U46619-induced platelet P-selectin expression was inversely correlated with changes in blood glucose (Fig. 2C) and positively correlated with changes in plasma insulin from B2 to PP (Fig. 2D). Similarly, meal-induced changes in platelet fibrinogen binding tended to correlate inversely with plasma glucose (P = 0.06) and positively with plasma insulin (P = 0.04; data not shown).
DISCUSSION
This study was designed to evaluate whether platelet activation after a carbohydrate-rich meal is related to the PPH. However, attenuation of PPH by premeal insulin injections did not result in reduced platelet activation. On the contrary, platelet activation evoked by thromboxane receptor stimulation (U46619) was markedly enhanced after premeal insulin, and a positive correlation was observed between postprandial platelet activation and insulin levels. In support of direct or indirect platelet activating effects of insulin in vivo, we found that insulin infusion to standardize premeal glucose levels was also associated with increased platelet reactivity. These results suggest that postprandial hyperinsulinemia, rather than PPH, causes platelet hyperactivity mediated via pathways stimulated by thromboxane and ADP but not collagen after food intake in patients with T2DM.
Several studies, including work from our laboratory, have shown that high glucose can increase platelet reactivity via an elevation of osmolality (11,12). Our findings of postprandial platelet hyperreactivity in T2DM patients but not in healthy control subjects led us to propose that PPH causes platelet dysfunction (5). In line with this notion, acarbose treatment, which attenuates PPH, reduced the urinary thromboxane metabolite excretion in T2DM (6). Our present finding challenges the concept that PPH enhances platelet activation (6) and indicates that other mechanisms may underlie postprandial platelet hyperactivity in T2DM patients.
Platelets express insulin and IGF-1 receptors, which may be activated by insulin (13–15), but the literature on insulin effects on platelet activation is inconsistent. We have found mild platelet-activating effects of insulin at physiologic concentrations (10–100 µU/mL [60–600 pmol/L]) in vitro (9,16). This is in line with the present in vivo results. However, several laboratories have found platelet-inhibiting effects of higher insulin concentrations (≥1 nmol/L) (14,17,19), and platelet-activating effects of very high insulin concentrations (100 nmol/L) (15,17,19). Others found no direct effects of insulin on platelets in vitro but found inhibitory effects during coincubation with endothelial cells (20). In healthy individuals, insulin appears to suppress platelet activation through inhibition of the P2Y12 pathway (18). However, patients with T2DM and even obese insulin-resistant subjects seem to have “platelet insulin resistance” and lose their responsiveness to insulin (18,21,22). In insulin-treated obese T2DM patients, there are increases in collagen- and ADP-induced platelet adhesion and aggregation, and increased phosphatidylserine exposure (18), which is consistent with our observations.
In the second Diabetes Mellitus Insulin Glucose Infusion in Acute Myocardial Infarction (DIGAMI 2) study, insulin treatment of T2DM patients after a myocardial infarction was associated with an increased risk of myocardial infarction or stroke compared with sulfonylurea treatment, whereas metformin treatment, which increases insulin sensitivity, was associated with a decreased risk compared with sulfonylurea treatment (23). Our findings of platelet activation by insulin in vivo may be relevant in this context. Also of interest are findings of higher platelet activity among T2DM patients receiving dual-antiplatelet treatment who were also treated with insulin compared with those who were not (24). The importance of insulin resistance for these findings in T2DM patients and the mechanisms involved in the presently observed meal effects, such as increased oxidative stress (2), deserve further investigation.
In conclusion, our results suggest that postprandial insulin rather than glucose levels may account for meal-induced platelet hyperactivity in patients with T2DM.
Clinical trial reg. no. NCT00771693, clinicaltrials.gov.
ACKNOWLEDGMENTS
This work was supported by grants from the Swedish Research Council/Medicine, the Swedish Heart-Lung Foundation, Karolinska Institutet, and the Stockholm County Council.
Novo Nordisk AB kindly provided a research grant for partial support of the study, which was academically initiated and conducted. No other potential conflicts of interest relevant to this article were reported.
G.S. designed the research, researched and analyzed the data, and wrote the manuscript. C.-G.Ö. and N.L. designed the research and revised the manuscript. P.H. designed the research, analyzed the data, and wrote the manuscript. P.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
This work was presented as a poster at the XXIII Congress of the International Society on Thrombosis and Haemostasis (ISTH), Kyoto, Japan, 23–28 July 2011. The abstract is published in J Thromb Haemost 2011;9(Suppl. 2):344.
The authors thank Maud Daleskog, Maj-Christina Johansson, Kathinka Leine, and Ragnhild Stålesen in the Clinical Pharmacology Unit, and Kajsa Sundqvist in the Endocrinology & Diabetology Unit, Karolinska University Hospital-Solna, for expert technical assistance, and Per Näsman, PhD, Royal Institute of Technology, Stockholm, for valuable statistical advice. The authors are grateful for the opportunity to use the collagen-related peptide CRP-18/I, which was kindly provided by Dr. Richard Farndale, Biochemistry, Cambridge University, U.K.