Incremental role of glycaemic variability over HbA1c in identifying type 2 diabetic patients with high platelet reactivity undergoing percutaneous coronary intervention

A schematic diagram of the study protocol was reported in Fig. 1. We performed a blood sample in all patients immediately before the percutaneous procedure in the catheterization laboratory, right after the insertion of the radial/femoral sheath, for the assessment of platelet reactivity using the VerifyNow P2Y12 assay (Accumetrics Inc., San Diego, California). Results were expressed as P2Y12 reaction units (PRU) and HPR was defined as a PRU value > 240 [18, 19]. Fasting blood glucose levels and HbA1c were also obtained before PCI in all patients. Furthermore, blood samples were drawn before and at 8 and 24 h after the procedure for creatine kinase-MB (CK-MB) and troponin I (TnI) levels detection; further values were obtained if symptoms suggested myocardial ischemia. These measurements were performed by standard enzymatic procedures (LOCI™ immunochemiluminometric assay—SIEMENS); normal limits were considered 3.6 ng/ml for CK-MB and 0.06 ng/ml for TnI. PCI-related myocardial damage was defined by an elevation of TnI > 5 the 99th percentile URL in patients with normal baseline values and by an increase of TnI > 20% if the baseline values were elevated (> 0.06 ng/ml).

The glucose monitoring methods were previously reported [20]. Briefly, all enrolled patients were equipped with the iPro™ continuous glucose recorder (Medtronic, Northridge, CA) and a CGM sensor (Enlite^® Sensor) was inserted into the subcutaneous abdominal fat tissue about 12 h before the revascularization procedure and platelet reactivity assessment. CGM measured subcutaneous tissue interstitial glucose levels every 5 min, within a range 40–400 mg/dl. The analysis was performed on data obtained in the entire peri-procedural period, from 12 h before to 12 h after PCI. During iPro™ CGM, patients also checked their capillary blood glucose levels with a self-monitoring at least 4 times per day. The FreeStyleLite (Abbott Laboratories, Abbott Park, IL) BG-monitoring system was used to calibrate the iPro™ continuous glucose recorder. After 24-h monitoring, the recorded data were downloaded for the analysis of the glucose excursion parameters using the CareLinkiPro System. Intraday GV was expressed by the GV indexes: standard deviation (SD); coefficient of variation (CV); mean amplitude of glycaemic excursions (MAGE), representing an average size of glycaemic fluctuations; MAGE-up, an average size of glycaemic fluctuations from nadirs to peaks, thus indicating hyperglycaemic excursions; MAGE-down, indicating glycaemic variations from peaks to nadirs, thus hypoglycaemic excursions; continuous overall net glycemic action (CONGA) (1, 2 and 4) assessing magnitude and timing of blood glucose fluctuations captured over various time periods.

Data are presented as percentages and frequencies for categorical variables; whereas, continuous data are presented as mean ± standard deviation (SD). To identify deviations from the normal distribution the Kolmogorov–Smirnov test was used. Comparisons between groups were evaluated using the Mann–Whitney U test, the Chi-square or the Fisher exact test according to distribution. Bivariate correlations between variables were performed and partial correlations were used to adjust for covariates (e.g., BMI and age). HPR and PCI-related myocardial damage were used as dependent variables and only variables which showed a p-value of < 0.1 in the two-way analysis were tested in the hierarchical multivariate model. Results of logistic regression are expressed as odds ratios (ORs) with their 95% confidence intervals (95% CI). Two-tailed p-value < 0.05 was considered statistically significant. The statistical analyses were performed using SPSS version 20.

Guidelines recommend that patients with type 2 DM should be treated to achieve a HbA1c level between 7 and 8%, although an even less stringent target has been proposed in those with advanced macrovascular complications, such as established cardiovascular disease [21]. In our population, more than 70% of patients had an HbA1c value ≤ 7%, indicating a well-controlled glycaemic status; however, the incidence of HPR was still 40%. With respect to external validity of the present analysis, these findings are consistent with previous literature reporting reduced responsiveness to antiplatelet drugs in 36–60% of diabetic patients [22‐24], with the highest values in those receiving insulin therapy [25]. The high prevalence of HPR among diabetic patients should be properly taken into account on the clinical ground because HPR represents a recognized predictor of post-PCI long-term cardiovascular events, including death, MI and stent thrombosis in patients on antiplatelet therapy [26].

In our study, HbA1c was observed to be an independent predictor of HPR, with each 1% increase in HbA1c levels translating into a seven-fold rise in the risk of PRU > 240. Previous studies investigating the impact of this glycaemic parameter on platelet reactivity have shown contradictory results [6‐9]. Gaborit et al. [6] reported no correlation between ADP-induced platelet aggregation and glycaemic variables such as fasting blood glucose and glycated haemoglobin in a large cohort of diabetic patients. On the other hand, a specifically designed analysis of the prospective, multicenter Assessment of Dual AntiPlatelet Therapy With Drug-Eluting Stent (ADAPT-DES) Registry showed HPR prevalence increasing progressively with HbA1c levels [9]. Unfortunately, no previous study took into account the effects of GV on platelet reactivity.

While HbA1c represents a measure of chronic hyperglycaemia [10, 27], it does not reflect acute glucose fluctuations (hyper/hypoglycaemic spikes). Acute glycaemic variations are closely associated with cardiovascular complications and may be more adequately assessed by using GV [11‐17]. Long-term GV has been demonstrated to represent a better glycaemic parameter than mean HbA1c for assessing the risk of micro and macrovascular complications in diabetic patients [11]. Moreover, HbA1c is characterized by several other shortcomings including worse diagnostic performance in different populations such as elderly, patients with iron deficiency anaemia or increased red blood cell turnover, those with end-stage renal disease or alcohol consumption [10].

Consequently, several studies described the clinical impact of GV in patients with coronary artery disease undergoing PCI [14‐16]. A high MAGE value has been demonstrated an independent predictor of long-term poor prognosis in a large cohort of patients with acute coronary syndromes undergoing coronary stenting [15].

According to our results, GV showed to be relevant in determining platelet reactivity despite an adequate HbA1c level: indeed, at the multivariate analysis, MAGE revealed to be an independent predictor of HPR even after the adjustment for HbA1c. Importantly, GV also exhibited an incremental value over HbA1c in detecting patients at higher thrombotic risk: adding MAGE over HbA1c to the hierarchical multivariate model significantly increases the chance to identify a non-responder to clopidogrel (76% compared with 50% gained with HbA1c measurement only).

To the best of our knowledge, this is the first study investigating the correlation between GV assessed by CGM and platelet reactivity, particularly in the setting of percutaneous coronary revascularization. Previous studies only focused on hyper or hypoglycaemia at one time, failing to demonstrate the biological significance of the combined, which can be only expressed by GV. For instance, induced hypoglycaemia (50 mg/dl) in patients with type 2 DM has been demonstrated to enhance platelet hyperactivity by impairing sensitivity to prostacyclin [28]. On the other side, acute, short-term hyperglycaemia has been proposed to produce transient increased platelet activation after exposure to high shear stress both in vitro and in vivo in diabetic patients, possibly involving enhanced levels of Von Willebrand factor [29]. Furthermore, persistent in vivo platelet activation has been observed in response to postprandial hyperglycaemic spikes in patients with early diagnosed type 2 DM and with HbA1c levels < 7% [30]. Nevertheless, GV is the only measure including positive and negative acute glycaemic fluctuations and, therefore, it might play a more relevant role in promoting a prothrombotic state.

In our population higher GV values were observed in patients with PCI-related myocardial damage, confirming our previous evidence that diabetic patients with poor glycaemic control may have a significantly higher risk of suffering from this complication [17]. This is particularly relevant considering PCI-related myocardial injury associated with unfavourable long-term outcome [31]. Interestingly, high levels of MAGE were associated with increased myocardial damage, assessed by myocardial blush grade and ST-segment resolution, also in non-diabetic patients undergoing primary PCI [32].

MAGE-up confirmed to be an independent predictor of this ischemic complication as well as high white blood cells levels. These findings are consistent with experimental studies reporting acute blood glucose fluctuations significantly associated with increased oxidative stress, inflammation and endothelial dysfunction, all factors involved in the pathophysiology of myocardial damage during PCI [33]. Moreover, the occurrence of PCI-associated myocardial damage may reflect a more extensive and unstable atherosclerotic burden, as observed in patients with high GV. An intravascular ultrasound study reported increased MAGE levels in patients with larger plaque volumes and greater lipid plaque components [34].

Since HPR is a well-described predictor of adverse thrombotic events [4, 5, 26], our findings underscore the clinical importance of maintaining an adequate glycaemic control, assessed by GV and HbA1c, in diabetic patients undergoing PCI.

For instance, our study generates the hypothesis that periprocedural GV and HbA1c assessment may help to identify those diabetic patients that should be moved towards more aggressive antiplatelet management, such as ticagrelor, also in the setting of an elective procedure [35, 36].

Similarly, given the impact of glycaemic control on HPR, efforts to reduce GV in diabetic patients undergoing PCI may have the same clinical relevance than an intensified platelet inhibition. Indeed, improvement in glucose metabolism assessed by fasting glucose levels, HbA1c and GV has been demonstrated to be associated with reduced oxidative stress, this latter playing an important role in the development of cardiovascular complications in diabetic patients [37]. Moreover, in the last decade, several novel oral glucose lowering drug classes (dipeptidyl peptidase 4 inhibitors, glucagon-like peptide-1 agonists, and sodium-glucose cotransporter 2 inhibitors) have been demonstrated to improve GV indexes [38, 39] and to reduce adverse cardiovascular outcomes in large clinical trials [40]. Few studies have also investigated possible effects of these agents on platelet function with favourable results [41, 42]. Nevertheless, whether the use of these new glucose-lowering drugs may also provide antithrombotic effects improving GV and HbA1c in diabetic patients undergoing PCI should be investigated in further studies.

The study population was limited, but this is the first study indicating a significant association between GV and HPR. There is not yet a consensus regards the appropriate methods and indexes for assessing GV; consequently our results showing that patients with higher MAGE-up levels had an increased incidence of PCI-related myocardial damage need further confirmation. Nevertheless, in our study we used CV and MAGE for the correlation with HPR, the most accepted metrics for GV according to the previous literature [11, 20]. Furthermore, CGM accuracy may be reduced during the first hours after sensor insertion due to local trauma; however, in our cohort, the mean absolute relative difference (MARD) value, an accepted metric of CGM precision, was < 10% indicating an adequate monitoring performance [43]. CGM was maintained for 24 h, within the 12 h before and the 12 h after PCI; even though a longer period of monitoring might translate in a more precise assessment of GV, nevertheless it would be less applicable to current clinical practice where patients with stable coronary disease are admitted few hours before the procedure and discharged shortly after; we believe that been in keeping with current recommended clinical practice may represent a further strength of the present study. Additionally, GV is likely affected by several pre-, intra- and post-procedural factors in patients undergoing PCI; thus, a 24-h GV monitoring peri-procedural interval may be an adequate time frame taking into account all these aspects and potentially useful, on the clinical ground, for the tailored antiplatelet management of these patients.

Finally, our sample size did not allow to investigate the effects of different glucose lowering agents on GV and, consequently, on platelet reactivity. In addition, we did not explore the effects of glycaemic state on prasugrel or ticagrelor antiplatelet action. However, even if both these two agents are considered the first choice in the acute coronary setting according to guidelines, clopidogrel still remains the more frequently used antiplatelet drug in elective procedures, and we observed as glycaemic state significantly impact patients’ responsiveness to this drug.