Younger age, higher body mass index and lower adiponectin concentration predict higher serum thromboxane B₂ level in aspirin-treated patients with type 2 diabetes: an observational study

Treatment with acetylsalicylic acid (ASA) has proven effective in the reduction of cardiovascular morbidity and mortality. Type 2 diabetic (DM2) patients are known to be at a very high cardiovascular risk, therefore it could be anticipated that they should potentially benefit the most from ASA treatment [1]. However, results of the Antithrombotic Trialists' Collaboration's meta-analysis of 195 clinical trials with over 135,000 patients, including almost 5000 diabetic patients, suggest that ASA therapy in diabetic patients may be less effective in cardiovascular prevention than in normoglycemic individuals [2]. Concomitant DM2 also increases the risk of high on-aspirin platelet reactivity (HAPR), defined as inadequate inhibition of platelet function, assessed in vitro with laboratory tests [3]–[9]. According to the position paper of the Working Group on Antiplatelet Drugs Resistance, endorsed by the Working Group on Thrombosis of the European Society of Cardiology, laboratory response to ASA treatment should be assessed with aspirin-specific tests evaluating the degree of cyclooxygenase 1 (COX-1) inhibition, either directly, by measuring serum concentration of thromboxane B₂ (TXB₂) - a stable metabolite of thromboxane A₂ (COX-1 product) or indirectly, by assessing platelet aggregation induced by arachidonic acid (COX-1 substrate) [10].

Patients with DM2 exhibit abnormalities of both platelet and plasma hemostasis, and differ from other patients in terms of thrombus structure and kinetics [11]. The pathomechanism of HAPR in patients with DM2 is complex and, so far, not entirely understood [12],[13]. It is hypothesized that under hyperglycemic conditions nonenzymatic glycation of COX-1 may competitively inhibit its acetylation by ASA [14]. In addition, nonenzymatic glycation of platelet membrane proteins may result in reduced membrane fluidity, increasing the propensity of platelets to activate [15]. Other possible mechanisms leading to enhanced platelet reactivity under hyperglycemia include osmotic effect of glucose, increased activation of protein kinase C and decreased activation of nitric oxide (NO) - cyclic guanosine monophosphate (cGMP) - cGMP-dependent protein kinase pathway [15]–[17]. These assumptions are supported by clinical studies demonstrating relationship between HAPR and inadequate glycemic control, as well as in vitro studies showing enhanced platelet activation, assessed using flow cytometry, with increasing glucose concentrations despite incubation with ASA [6],[7],[18]–[23]. Alongside with poor glycemic control, HAPR in DM2 patients has been linked to high triglyceride, total cholesterol and low-density lipoprotein (LDL) concentrations, as well as low high-density lipoprotein (HDL) concentration [18],[20],[24]. Hyperglycemia and hyperlipidemia induce oxidative stress, which, in turn, leads to fatty acids peroxidation and, consequently, formation of products that may change physicochemical properties of platelet plasma membrane or - as described for isoprostanoids - act as ligands for platelet membrane receptors, triggering platelet activation and aggregation via alternative pathways [25]. Furthermore, oxidative stress may also augment platelet reactivity indirectly, as it results in endothelial dysfunction, diminished endothelial synthesis of NO and prostacyclin, and attenuation of biological effects of NO [15].

Much as the relationship of HAPR with glycemic and lipid control of DM2 has been well documented, only few small studies regarding the association of platelet reactivity with insulin resistance and adipokine concentrations have been conducted so far [5],[26]–[28]. Among 60 healthy women, a lower insulin sensitivity index was predictive of higher urine 11-dehydro-TXB₂ concentration and improvement in insulin sensitivity resulted in decreasing urinary 11-dehydro-TXB₂ excretion [26]. In a group of 55 patients with DM2, higher Homeostatic Model Assessment-Insulin Resistance (HOMA-IR) index was an independent predictor of poor responsiveness to ASA assessed using urine 11-dehydro-TXB₂ concentration measurement [27]. Increased rates of HAPR observed in obese patients also imply its potential association with insulin resistance [6],[21],[26]. On the other hand, similar prevalence of poor responsiveness to ASA in patients with type 1 and type 2 diabetes seems to contradict such relationship [29]. Moreover, a negative correlation of serum TXB₂ concentration with body mass index (BMI) and leptin concentration (which was shown to positively correlate with HOMA-IR) reported by Graziani et al. also question the previously suggested pathogenetic role of insulin resistance in the development of HAPR [28]. Thus, consistent data on the association of HAPR with insulin resistance and adipose tissue metabolic activity are lacking.

In our previous work, we have evaluated the relationship between HAPR assessed with various laboratory methods and genetic factors, clinical variables and routine laboratory parameters in DM2 [30],[31]. The aim of this study was to assess the relationship between laboratory response to ASA therapy and insulin resistance, adipose tissue metabolic activity and the achievement of metabolic control in DM2 patients.

In the studied group of DM2 patients, age proved to be the most important predictive factor of laboratory response to ASA therapy. Some published data suggest an association between HAPR and older age, which could be explained by an increasing number of comorbidities and other coexisting conditions, including risk factors for HAPR [38]. However, our analysis revealed younger age as an independent predictor for HAPR. Similar results have been previously reported by Maree et al. [34]. In a group of 131 stable cardiovascular patients taking 75 mg of enteric-coated ASA daily, HAPR, defined as serum TXB₂ concentrations above 2200 pg/mL, was found in 44% of patients. In a multivariate analysis, younger age and higher body weight were the only predictive factors of HAPR. In all cases of HAPR in vitro incubation of participants’ blood samples with ASA solution resulted in a complete inhibition of platelet aggregation, which implied a pharmacokinetic mechanism of HAPR (i.e. related to diminished bioavailability of ASA) [34].

Lower bioavailability of ASA in younger patients may be attributable to a higher activity of plasma, hepatic and intestinal esterases, which hydrolyze ASA to salicylic acid - a product deprived of permanent antiplatelet activity. Review of published literature confirms that plasma esterase activity, including aspirin esterase, decreases in older people burdened with additional risk factors, especially a coexisting inflammatory state [39]–[41]. Furthermore, compared with healthy volunteers, higher aspirin esterase activity has been reported in DM2 patients, leading to faster ASA hydrolysis and HAPR [42]. These data suggest that the inverse relationship between serum TXB₂ concentration and age observed in our study could arise from age-related differences in ASA bioavailability in DM2 patients. It can be hypothesized that in younger DM2 patients a total daily dose of ASA higher than 75 mg may be required to exhibit full antiplatelet effect. Low doses of ASA are a known risk factor for HAPR and increasing ASA doses have been reported to overcome HAPR [6],[19],[25].

The etiology of HAPR in younger DM2 patients might involve the above described pharmacokinetic mechanism, as well as pharmacodynamic abnormalities. A higher rate of platelet turn-over resulting in a faster recovery of COX-1 activity after ASA administration has been proposed as a potential pharmacodynamic mechanism of HAPR in DM2 after several studies demonstrated increased platelet aggregation at the end of the 24-hour dosing interval, compared to platelet aggregation measured early after ASA ingestion [43]–[45]. In 100 DM2 patients receiving enteric-coated ASA 100 mg once daily, serum TXB₂ concentration recovery rate between 12 and 24 hours after ASA intake was measured by Rocca et al. [45]. Younger age, higher BMI and higher mean platelet volume (MPV) were independent predictors of the most rapid increase in TXB₂ concentration. Positive correlations between TXB₂ recovery rate and MPV as well as between TXB₂ recovery rate and the percentage of newly-released, messenger ribonucleic acid (mRNA)-positive platelets were observed, which suggests that the recovery of COX-1 activity during the 24-hour dosing interval largely depends on accelerated platelet turn-over. Most importantly, in a subgroup of 33 patients with the most rapid TXB₂ recovery on ASA 100 mg o.d., changing ASA regimen to 100 mg b.i.d. resulted in a persistent inhibition of COX-1 activity, while changing to 200 mg o.d. displayed an intermediate pattern of TXB₂ recovery [45]. Several other platelet function studies conducted in DM2 patients, confirmed the beneficial effect of a twice-daily ASA regimen compared to an equivalent ASA dose given once daily [46]–[49].

Thus, the accumulated evidence indicates that in DM2 a twice-daily ASA regimen may be favorable compared to merely increasing ASA given as a single daily dose. This approach might be of value especially in younger DM2 patients, as it would counteract both the pharmacokinetic (esterase-activity-dependent) and the pharmacodynamic (related to the increased platelet turn-over) mechanism of HAPR. However, no threshold allowing identification of such "younger" patients can be warranted based on the results of our study, as the observed difference in age between the patients with TXB₂ concentrations above the upper quartile and the patients with TXB₂ concentrations below the lower quartile was only 4.5 years. Therefore, clinical significance of this finding should be verified by further studies.

The second independent predictor of serum TXB₂ concentration above the upper quartile in the studied population was a higher BMI. In a univariate analysis, higher BMI was also significantly predictive of serum TXB₂ concentration above the median, however, this effect was attenuated in a multivariate analysis. Indices of obesity, including higher BMI and body weight, as well as markers of visceral obesity, such as waist circumference and waist-hip ratio (WHR) have been previously described as risk factors for HAPR [6],[21],[26],[27],[34],[38]. Platelet hyperaggregability in obese patients may result from enhanced lipid peroxidation due to increased generation of intraplatelet reactive oxygen species and low-grade inflammation, which may be related to coexisting metabolic (e.g. lipid) abnormalities [50]–[53]. Fatty acid peroxidation may, in turn, induce changes in the physicochemical properties of platelet plasma membrane and formation of products which activate platelets by binding to their surface receptors, thus, triggering HAPR in a pharmacodynamic manner [25]. Nevertheless, the observed association between BMI and HAPR in our study, might have also resulted from reduced bioavailability of an enteric-coated ASA preparation, representing a pharmacokinetic mechanism of HAPR. Recently, in a group of 400 healthy volunteers, Grosser et al. [54] have assessed platelet aggregation and serum TXB₂ concentration after ingestion of a single oral dose of enteric-coated versus immediate-release ASA. Ingestion of an immediate-release preparation led to a substantial inhibition of platelet aggregation and TXB₂ generation in all of the patients, while a single dose of enteric-coated preparation was associated with high rates of HAPR - higher if assessed 4 hours after ASA intake compared to the measurement performed 8 hours after ASA ingestion. Ex vivo addition of ASA to blood samples of non-responders resulted in amelioration of HAPR, suggesting that in those patients HAPR was related to reduced and delayed absorption of enteric-coated ASA tablets [54]. In our study, we used enteric-coated ASA preparations and collected blood samples relatively early (i.e. 2-3 hours) after ASA ingestion. The extent to which this approach might have influenced the results of our analysis remains uncertain, since in the study by Grosser et al. [54], prolongation of enteric-coated ASA treatment to one week led to overcoming of HAPR in all but one of the 27 patients who were unresponsive to a single dose of the enteric-coated preparation. Still, other studies have demonstrated that compared to immediate-release preparations, enteric-coated ASA tablets are less effective at inhibiting TXB₂ generation, even if taken chronically [55],[56]. Interestingly, in these studies, higher body weight was found to be predictive of HAPR - analogously as in the above cited study by Maree et al. [34],[55],[56]. It has been hypothesized that due to reduced bioavailability, low doses of enteric-coated ASA may be insufficient in heavier patients with a larger volume of distribution [34]. Thus, a conversion from 75 mg daily of enteric-coated ASA to 75 mg daily of immediate-release ASA has been suggested by Peace et al. [56] for patients over 90 kg, with a further dose increase to 150 mg daily in patients over 120 kg.

Despite significant differences in HOMA-IR across increasing quartiles of serum TXB₂ concentration, logistic regression analysis did not reveal HOMA-IR as a predictive factor of HAPR in the studied population. This might suggest that the observed relationship of higher TXB₂ levels with lower HMW adiponectin concentrations is not associated with its insulin-sensitizing properties. Moreover, HMW adiponectin remained borderline significant even after correction for BMI. No association of serum TXB₂ concentration with neither leptin nor resistin levels would also imply that the relationship of low HMW adiponectin concentrations with increased platelet reactivity is independent from its insulin-sensitizing course of action.

To our knowledge, this is the first study to demonstrate the association of lower HMW adiponectin concentration with higher serum TXB₂ level in aspirin-treated patients. Under physiological conditions, adiponectin is the hormone most abundantly secreted from adipocytes, with serum concentrations ranging between 3 and 30 μg/mL in healthy individuals [57]. In the blood, it circulates as three different oligomeric complexes, including low-molecular weight (a trimer), medium-molecular weight (a hexamer) and HMW adiponectin - the latter is considered the most biologically active form and constitutes approximately half of adiponectin's plasma amount [57]. In contrast to leptin and resistin, adiponectin levels have been shown to correlate inversely with BMI, the amount of visceral adipose tissue and insulin resistance [58]–[60]. Low adiponectin concentration is an independent predictive factor of the future development of DM2 [61]. Hypoadiponectinemia has been also associated with an increased risk of metabolic syndrome, hyperlipidemia, nonalcoholic steatohepatitis and polycystic ovary syndrome, as well as hypertension, coronary artery disease and acute coronary syndrome [59],[60]. Adiponectin acts via two distinct types of receptors, identified in skeletal muscles (type 1 receptor) and liver (type 2 receptor), as well as on the surface of pancreatic β-cells, endothelial cells, smooth muscle cells, cardiomyocytes, monocytes, macrophages and platelets. It improves insulin sensitivity by increasing fatty acid β-oxidation and glucose utilization, as well as decreasing lipogenesis and gluconeogenesis in the liver. Unlike leptin and resistin, adiponectin has also been shown to exert anti-inflammatory and antiatherogenic effects. Its protective influence on the endothelium results largely from the enhancement of NO generation and attenuation of oxidative stress [57],[59],[60]. Recently, Okada-Iwabu et al. [62] have discovered an orally active, synthetic adiponectin receptor agonist (AdipoRon), which was found to ameliorate insulin resistance, glucose intolerance and diabetes in experimental rodent models, offering a new promising therapeutic approach to DM2 prevention and treatment.

Although antiplatelet properties of adiponectin have been already described before, the existing evidence stems mostly from in vitro studies and the exact mechanism of action remains unknown. Data from the literature imply a possibility of both, a direct and an indirect effect of adiponectin on platelet function. It has been demonstrated that human and animal platelets contain mRNA for adiponectin receptors and that in healthy volunteers, as well as in patients with the metabolic syndrome, platelets express both types of adiponectin receptors on their surface [63],[64]. In comparison to wild-type mice, adiponectin knockout mice showed enhanced platelet aggregation induced by adenosine diphosphate (ADP) and collagen, and accelerated thrombus formation on carotid arterial injury [63]. In vitro incubation of platelets with adiponectin decreased spontaneous, epinephrine- and ADP-induced platelet aggregation in several, although not in all studies [64]–[66]. Furthermore, adiponectin could influence platelet reactivity also in an indirect manner, by counteracting the deleterious effects of perioxide generation and by increasing the activity of the endothelial NO synthase (eNOS), NO generation, endothelial COX-2 expression and prostacyclin production [67]–[69]. Considerably less evidence is available from the clinical studies. In a group of 30 healthy volunteers and 30 patients with the metabolic syndrome, adiponectin concentration correlated negatively with plasma soluble P-selectin and soluble CD40 ligand (CD40L) levels, as well as with epinephrine- and ADP-induced platelet aggregation [64]. In 277 patients with DM2 low concentration of adiponectin was an independent predictive factor for the formation of leukocyte-platelet aggregates [70]. So far, only one study has investigated the relationship between adiponectin level and laboratory response to ASA. Following the incubation of blood samples with ASA solution, Takahashi et al. assessed platelet reactivity with Platelet Function Analyzer-100® (PFA-100®) in 168 men without known cardiovascular disease [5]. Poor responsiveness to ASA was defined as collagen-epinephrine closure time (CEPI-CT) below 250 seconds. The subgroup of 40 poor responders was characterized by a higher prevalence of diabetes and, surprisingly, by a higher adiponectin concentration (8.8 ± 4.1 μg/mL versus 7.3 ± 2.9 μg/mL, p = 0.010). These results, however, were not verified with a multivariate analysis [5].

In the present study, we observed associations of serum TXB₂ concentration with fasting glycemia, HbA_1c proportion and triglyceride concentration. Nevertheless, the results of logistic regression analysis did not support the existence of a relationship between achieving treatment targets [i.e. HbA_1c below 7% and triglyceride level below 150 mg/dL] and serum TXB₂ levels. There was no association of serum TXB₂ concentration with total, LDL or HDL cholesterol. This is not in concordance with the data from the previous studies, which have reported a significant relationship between HAPR and the indices of metabolic control of diabetes [6],[7],[18]–[22],[24]. However, a more thorough analysis of the available literature reveals that studies which showed such relationship mostly included patients with poor or suboptimal diabetes control and, in some cases, patients with diabetes and healthy controls. Such characteristics of a studied population would result in a broader spectrum of values of the analyzed parameters, which could facilitate a demonstration of significance between the observed differences. Furthermore, the higher glucose, triglyceride and cholesterol concentrations, the higher the probability of development of platelet function disorders severe enough to be reflected by laboratory tests. On the other hand, patients included in our analysis comprised a more homogeneous group and were characterized by a relatively good metabolic control, with median HbA_1c of 6.5% [in comparison to 7.3% in a study by Cohen et al., and 9.4% in a study by Watala et al.], mean total cholesterol of 166.9 mg/dL (in comparison to 207 mg/dL in a study by Watala et al.), mean LDL cholesterol of 89.1 mg/dL (in comparison to 135.4 mg/dL in a study by Hovens et al.), mean HDL cholesterol of 48.9 mg/dL (in comparison to 43 mg/dL in a study by Watala et al.), and mean triglyceride concentration of 139.6 mg/dL (in comparison to 159.4 mg/dL in a study by Hovens et al.) [18],[20],[21]. These differences may at least partly account for the lack of relationship between platelet reactivity and metabolic control in our study. Another explanation for the lack of significant association between serum TXB₂ concentration and glycemic control may derive from the findings of Russo et al., who proved that incubation of platelets with hyperosmolar glucose solutions prevented ASA-induced activation of the NO - cGMP - cGMP-dependent protein kinase pathway and, consequently, reduced the ability of ASA to inhibit platelet responses to agonists but did not modify ASA-dependent inhibition of TXB₂ synthesis [17].

In the studied group of DM2 patients, the most important independent predictor of HAPR was younger age. Younger patients with DM2 may therefore require total daily ASA doses higher than 75 mg, preferably as a twice-daily regimen, to achieve full therapeutic effect. Higher BMI and lower HMW adiponectin concentration were also associated with less potent ASA effect. This is the first study to demonstrate an association of lower adiponectin concentration with higher serum TXB₂ level in patients treated with ASA.