Association Between Increased Platelet P-Selectin Expression and Obesity in Patients With Type 2 Diabetes: A BARI 2D (Bypass Angioplasty Revascularization Investigation 2 Diabetes) substudy Open Access

Centers participating in BARI 2D were invited to participate in this substudy, and nine clinical centers (please see acknowledgments) agreed. All patients enrolled at each participating center were eligible to participate in this substudy. Written informed consent for participation in the substudy was obtained from each participant, and the protocol was reviewed by the institutional review board of each participating clinical center. Results were reviewed by the data and safety monitoring board of the parent BARI 2D study.

Blood was obtained with the use of a two-syringe technique in which the first 3 ml blood was discarded. Blood (1 ml) to be analyzed for platelet reactivity was anticoagulated with 32 μg/ml corn trypsin inhibitor (CTI; Enzyme Research, South Bend, IN). CTI is a specific inhibitor of Factor XIIa without effect on other coagulation factors (11) and was used as the anticoagulant because we have shown that the activation of platelets is altered by conventional anticoagulants such as citrate (12).

Activation of platelets was identified with the use of flow cytometry by the binding to platelets of fluorescein isothiocyanate–labeled fibrinogen (that binds to the activated conformer of glycoprotein IIb-IIIa) and phycoerythrin-labeled anti-CD62 that binds to P-selectin on the surface of activated platelets (Becton Dickinson) as previously described (5,13,14). Platelets were identified based on size and binding of a peridinin chlorophyll protein (PerCP)-labeled anti-CD61 (Becton Dickinson), which binds to glycoprotein IIIa regardless of activation and does not interfere with the binding of fibrinogen.

To quantify nonspecific association of proteins with platelets and to define a threshold above which activation-dependent association occurs, control samples containing phycoerythrin-conjugated nonimmune mouse IgG and fluorescein isothiocyanate–conjugated albumin were assayed in samples from each subject. Assays were performed in duplicate. Activation of platelets was reported as the percentage of platelets that bound fluorescein isothiocyanate–fibrinogen or phycoerythrin–anti-CD62. This measure of platelet activation correlates with subsequent cardiac risk (13) and with platelet activation reported as the mean fluorescence intensity (14).

Blood to be analyzed for PT1.2 and FPA was added immediately to SCAT-1 tubes (Hematologic Technologies, Essex, VT). These tubes prevent protease activity and activation of coagulation factors. Aliquots of plasma were stored at −80°C until assay.

Concentrations of PT1.2 and FPA were determined with the use of commercial enzyme-linked immunosorbent assay (ELISA) kits (Dade Behring [Marburg, Germany] for PT1.2 and Vitro Chemie [Toernooiveld, the Netherlands] for FPA).

Concentrations of C-reactive protein (CRP) were determined with the use of a high-sensitivity colorimetric competitive ELISA as previously described (15). Concentrations of insulin were determined by ELISA (ALPCO, Salem, NH). Concentrations of fibrinogen were determined by nephelometry (Siemens Healthcare Diagnostics, Deerfield, IL).

Fasting lipid profile and A1C assays were performed by the Biochemistry Core Laboratory at the University of Minnesota, Minneapolis. Triglycerides were analyzed enzymatically. A1C concentrations were analyzed by high-performance liquid chromatography. For those patients for whom the core lab measures were unavailable, an estimate based on site-specific regression of the clinical site measure was used. This estimation was used for 2.5% of the triglycerides samples and 0.5% for A1C samples.

We report results from patients in which at least 80% of significant baseline data points were available for analysis (n = 201). Patients treated with either ticlopodine or clopidogrel were excluded from this analysis (n = 8) because a primary measure was ADP-induced activation of platelets. This study group (n = 193) enabled us to identify for 1 SD change in the independent variable a regression coefficient of 0.20 for 1 SD change in a dependent variable with a power of 0.77.

Because platelet reactivity results and those for markers of thrombosis were not normally distributed, Spearman's rank correlation estimates were used to evaluate relationships among the thrombosis markers and other variables of interest including BMI, A1C, lipids, and insulin. A P value of 0.05 was considered significant.

Clinical characteristics of patients are shown in Table 1.

Evidence of thrombin generation (PT1.2) and thrombin activity (FPA) was limited in these patients with stable coronary artery disease (Table 2). Similarly, evidence of platelet activation in the absence of agonist was minimal, whether assessed by the percentage of platelets that bound fibrinogen (reflecting activation of glycoprotein IIb-IIIa) or the surface expression of P-selectin (Table 2). The concentration of fibrinogen correlated positively with that of PT1.2 (r = 0.17, P = 0.034) but did not correlate with the concentration of FPA (P = 0.24).

To assess the propensity of platelets to activate in response to a stimulus (i.e., platelet reactivity), platelets were stimulated with 0.2 and 1 μmol/l ADP (Fig. 1). The activation of platelets identified by the surface expression of P-selectin in the absence of agonist and in the presence of 0.2 and 1 μmol/l ADP correlated positively (the correlation coefficient between 0.2 and 1 μmol/l ADP was 0.85, between no agonist and 0.2 μmol/l was 0.58, and between no agonist and 1 μmol/l ADP was 0.51; for each association, P < 0.0001). Correlations between the activation of glycoprotein IIb-IIIa seen with each concentration of ADP were not as strong but were still positive. The correlation coefficient between the percentage of platelets that bound fibrinogen with no agonist and in response to 0.2 μmol/l ADP was 0.31; between 0.2 and 1 μmol/l ADP, it was 0.64 (P < 0.0001 for both). The correlation coefficient between no agonist and 1 μmol/l ADP was 0.12 (P = 0.12).

Greater BMI was associated with greater platelet reactivity as assessed by the surface expression of P-selectin (Fig. 2). After adjustment for age, sex, A1C, use of insulin, and duration of diabetes in a linear model of platelet reactivity, the relationship between platelet reactivity and BMI remained significant (Table 3). A similar magnitude of increase was seen in the small group (n = 19) of subjects who were not taking aspirin at the time when blood was taken (data not shown). Identification of platelet activation based on the binding of fibrinogen (activation of glycoprotein IIb-IIIa) did not correlate with BMI. The concentration of FPA and PT1.2 did not correlate with BMI. By contrast, the concentration of fibrinogen correlated with BMI (r = 0.21, P = 0.004).

Obesity has been associated with poor glycemic control, hypertriglyceridemia, inflammation (reflected by an increased CRP), and insulin resistance (16,17). A1C correlated negatively with platelet reactivity in response to 1.0 μmol/l ADP (ρ_s = −0.16, P < 0.04). The effects of A1C and BMI were independent. The concentration in blood of triglycerides was not correlated with platelet reactivity. Similarly, the concentration of CRP did not correlate with platelet reactivity. The fasting concentration of insulin did not correlate with platelet reactivity (r_s = 0.14, P = 0.07).

In this substudy of BARI 2D, we assessed markers of thrombin generation and activity as well as platelet reactivity in blood from patients with type 2 diabetes and stable coronary artery disease. Activity of the coagulation cascade was limited, as was evidence of platelet activation in the absence of agonist. We assessed platelet reactivity by determining the propensity of platelets to activate in response to an agonist (0.2 and 1 μmol/l ADP). We found that greater BMI was associated with progressively greater platelet reactivity when platelet activation was assessed by the surface expression of P-selectin induced by 1 μmol/l ADP. A1C correlated negatively with this measure of platelet reactivity. Triglycerides did not correlate with platelet reactivity. Thus, our results suggest that obesity increases platelet P-selectin expression in patients with type 2 diabetes.

Increased platelet reactivity has been associated with a greater risk of subsequent cardiac events in apparently healthy subjects (18), in patients with previous myocardial infarction (19), and in patients undergoing percutaneous coronary intervention (13). Our results suggest that obesity may increase further the risk of subsequent cardiovascular events in patients with type 2 diabetes by increasing platelet reactivity.

Consistent with our results, previous studies have demonstrated increased activation of platelets in obese subjects. In one study, platelet activation was identified based on measurement of a thromboxane metabolite in urine and the concentration in blood of the CD40 ligand. Both were increased in obese compared with lean women, and successful weight loss led to decreased concentrations of these markers (9). A second study quantified the concentration in blood of platelet-derived microparticles that are released during the activation of platelets. The concentration of microparticles was increased in obese patients (10). Similar to the previous study, the concentration of microparticles decreased after weight reduction (10). Accordingly, obese subjects without diabetes manifest evidence of greater activation of platelets. Our results are consistent with and extend these observations to patients with type 2 diabetes.

Aspirin was used in the majority (89%) of the subjects we studied. The relationship between obesity and platelet reactivity was of similar magnitude in patients regardless of aspirin use or nonuse. These results are consistent with the limited efficacy of aspirin in patients with diabetes (20). One mechanism that may contribute to limited efficacy of antiplatelet therapy in such patients is persistently increased platelet reactivity. Our results suggest that obesity may be a cause of persistently increased platelet reactivity in patients with diabetes and thereby contribute to a lack of efficacy of aspirin.

We did not identify an association between the concentration in blood of CRP and platelet reactivity. By contrast, evidence of platelet activation has been associated with concentrations of CRP in obese women without diabetes (21). In our study, all patients had type 2 diabetes, and platelet reactivity was assessed ex vivo. The previous study compared markers of platelet activation in vivo in obese and nonobese women without diabetes. In addition, the consistent use of statins may have decreased CRP and obscured a potential interaction (22). Thus, differences in the clinical characteristics of patients and the methods used to assess platelet reactivity may account for the lack of association in our study.

Although we did not specifically measure insulin sensitivity, obesity and increased concentrations of insulin in fasting blood have been associated with insulin resistance (16). The strong positive association that we observed between obesity and platelet reactivity is consistent with our hypothesis that insulin sensitivity influences platelet reactivity when identified based on the surface expression of P-selectin and only in response to 1 μmol/l ADP. This correlation was not apparent when activation of platelets was identified based on the binding of fibrinogen. The mechanism responsible for this difference is not apparent. However, we have previously observed that platelet activation identified based on surface expression of P-selectin is altered by associated conditions or treatments (5,23). One mechanism potentially contributing is that activation of glycoprotein IIb-IIIa occurs with a low concentration of agonist (13). Thus, the low threshold for activation of glycoprotein IIb-IIIa may limit sensitivity for detection of changes. As seen in Fig. 1, surface expression of P-selectin was limited in response to 0.2 μmol/l ADP. Thus, the discrimination of inter-individual differences is reduced when the range of platelets activated is limited. Accordingly, we postulate that the lack of statistical significance between BMI and surface expression of P-selectin in response to 0.2 μmol/l ADP reflected the limited discrimination between individuals because of the limited range of activation.

Our study does not identify the mechanism by which insulin resistance increases platelet reactivity. Previous work has associated obesity with impaired synthesis and activity of cyclic nucleotides (i.e., cyclic adenosine monophosphate and cyclic guanosine monophosphate), which are key signaling molecules involved in the activation of platelets (24).

In summary, we found that obesity is associated with greater platelet reactivity in patients with type 2 diabetes and stable coronary artery disease. Our results extend previous observations made in subjects without diabetes to subjects with diabetes and suggest that insulin resistance that is associated with obesity increases platelet reactivity that may in turn increase the risk of subsequent cardiac events in patients with type 2 diabetes.

This study was supported by the National Heart, Lung, and Blood Institute (NHLBI) grant numbers R01 HL69146 (D.S., principal investigator [PI]) and R01 HL71306 [B.S., PI]). The BARI 2D trial is funded by the NHLBI, grant numbers U01 HL061746, U01 HL06171748, and U01 HL06384 and receives substantial funding from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), grant number HL061744.

BARI 2D receives significant supplemental funding from GlaxoSmithKline, Bristol-Myers Squibb Medical Imaging, Astellas Pharma US, Merck, Abbott Laboratories, and Pfizer and generous financial support from Abbott Laboratories, MediSense Products, Bayer Diagnostics, Becton Dickinson, J.R. Carlson Laboratories, Centocor, Eli Lilly, LipoScience, Merck Sante, Novartis Pharmaceuticals, and Novo Nordisk. No other potential conflicts of interest relevant to this article were reported.

The authors thank investigators who enrolled subjects in this substudy, including Charanjit Rihal (Mayo Clinic, Rochester, MN), Francisco Fuentes (University of Texas at Houston, Houston, TX), Karen Smith (University of Florida, Gainesville, FL), Frederick Feit (New York University, New York, NY), Raymond Magorien (Ohio State University, Columbus, OH), Kodangudi Ramanathan (University of Tennessee), and Alice Jacobs (Boston University, Boston, MA).