High residual platelet reactivity (HRPR) on clopidogrel is a predictor of recurrent ischemic events in patients undergoing percutaneous coronary interventions (PCI). Significant intraindividual variability in platelet aggregation on repeat testing has been reported. To understand factors contributing to the variability in platelet aggregation testing, we examined clinical and laboratory elements linked to HRPR in 255 consecutive patients tested ≥12 hours after PCI using light transmission aggregometry (LTA) in response to adenosine diphosphate 5 μmol/L and VerifyNow P2Y12 assay (VNP2Y12; Accumetrics). HRPR was defined as >46% residual aggregation for LTA and >236 P2Y12 response units (PRUs) for VNP2Y12. On multivariate analysis the only variable independently associated with HRPR with both LTA and VNP2Y12 was laboratory-defined anemia. Prevalences of HRPR by LTA were 34.3% in anemic patients, 15.6% in patients with normal hemoglobin levels, and 59.8% versus 25.9% by VNP2Y12 (p <0.005 for the 2 comparisons). In a subgroup of 50 patients, testing was done before and after the clopidogrel loading dose. At baseline there were no differences in platelet aggregation with either assay; however, absolute decrease in reactivity after the clopidogrel load was significantly less in anemic patients compared to patients with normal hemoglobin (change in residual aggregation by LTA 15.8 ± 5.8% vs 28.8 ± 3.2%, p <0.05; change in PRU by VNP2Y12 56.5 ± 35.5 vs 145.0 ± 14.2 PRUs, p <0.05, respectively). In conclusion, anemia is an important contributor to apparent HRPR on clopidogrel and may explain some of the intraindividual variability of platelet aggregation testing.
The variability in the inhibitory effect of clopidogrel on platelet aggregation has been well documented. In addition to genetic variation of cytochrome P450 enzymes involved in the conversion of the drug to its active metabolite, platelet response to clopidogrel is likely affected by additional factors, in particular the overall level of platelet activation, platelet turnover, and density of the P2Y12 receptors, which cannot be predicted by genetic testing. Thus, it is likely that the clinically relevant assessment would be that of inhibition of platelet function rather than genetic testing. Large prospective clinical studies in patients with coronary artery disease using aggregation-based measurements of platelet reactivity indeed have demonstrated increased recurrent ischemic events in patients who are considered to have high residual platelet reactivity on clopidogrel (HRPR), i.e., “low responders” to clopidogrel. In addition to interindividual variability, recent data have indicated a significant intraindividual variability of response to clopidogrel. This further underscores the dynamic nature of platelet reactivity, in particular in the periprocedural setting. In the present study we sought to identify clinical and laboratory variables that would predict HRPR after percutaneous coronary interventions (PCIs) in a large cohort of consecutive patients with coronary artery disease undergoing PCI, with a longitudinal subgroup of patients tested before and after clopidogrel loading at the time of the procedure.
Methods
The patients included in this study were enrolled in a prospective ongoing PCI registry at the University of Pittsburgh Medical Center from November 2008 through March 2010 in accordance with the University of Pittsburgh institutional review board. Platelet function data were collected routinely in all patients undergoing PCI as part of their clinical risk assessment. For the analyzed cohort blood was collected the morning after the intervention (12 to 24 hours after the clopidogrel load at time of PCI; if eptifibatide was used, testing was done ≥24 hours after drug discontinuation). The nursing staff was instructed on proper blood collection techniques for platelet aggregation testing. Blood samples were couriered to the Institute for Transfusion Medicine and processed within 4 hours of collection. In a subgroup of patients, blood was also collected at the time of PCI immediately after obtaining arterial access and before administration of antiplatelet agents. The limiting factor in obtaining consecutive datasets in all the patients who underwent PCI in this interval was the limited window of availability of the reference laboratory (weekdays only from 8 a . m . to 2 p . m .). We included patients in whom eptifibatide was used if testing was done ≥24 hours after its discontinuation (and that the base P2Y12 response unit [PRU] >190 for the VerifyNow P2Y12 [VNP2Y12] assay [Accumetrics, San Diego, California] was done according to the manufacturer’s instructions). Patients in whom abciximab was used and patients with missing data from 1 of the assays were excluded.
Clinical and laboratory data were collected at the same time. Chronic kidney disease was defined as an estimated glomerular filtration rate <60 ml/min/1.73 m 2 (calculated using the Modification of Diet in Renal Disease equation). Anemia was defined according to our laboratory hemoglobin concentration cutoffs <11.7 g/dl for women and <12.9 g/dl for men. Patients’ clinical histories including a history of diabetes mellitus, hypertension, stroke (cerebrovascular accident), known coronary artery disease, peripheral vascular disease, cancer, smoking status, current medications (including proton pump inhibitor use and clopidogrel and aspirin use) were collected prospectively at the time of presentation.
Platelet-rich plasma adjusted using platelet-poor plasma to 250,000 platelets/ml in all samples was used for standard light transmission aggregometry (LTA; PAP4, Bio/Data Corp., Horsham, Pennsylvania) with adenosine diphosphate 5 μmol/L as the agonist. Peak residual platelet aggregation was recorded. Parallel samples were analyzed using the VNP2Y12 assay with clopidogrel response tested using adenosine diphosphate 20 μmol/L with prostaglandin E for selective P2Y12 activation according to the manufacturer’s instructions. The VNP2Y12 assay also generates a base value, which represents a relatively adenosine diphosphate–independent aggregation response to thrombin receptor activating peptide, and it is used as an individual reference point of maximal platelet aggregation. By normalizing the aggregation amount (measured in PRUs), the VNP2Y12 assay generates a “percent inhibition.” All samples were collected and processed within 4 hours from blood collection. Standard accepted definitions for HRPR using the 2 assays were used (>235 PRUs for VNP2Y12 and >46% for LTA using adenosine diphosphate 5 μmol/L).
Continuous variables are expressed as median (interquartile range) and categorical variables are expressed as percentage. To avoid the assumption of normality in the distribution of results of platelet testing, nonparametric statistical methods were used when applicable. We first performed univariate analysis using logistic regression for categorical variables and Spearman rank correlation for continuous variables. Univariate predictors with a p value <0.1 were included in a multivariate analysis using logistic regression in separate models with HRPR by LTA and by VNP2Y12 as outcomes. In the subgroup of patients who underwent testing before and after PCI, a nonparametric Mann–Whitney test was used to compare variables between groups. Statistical significance was defined at a p value <0.05 (2-sided). Statistical analyses were performed with STATA 10.0 (STATA Corp., College Station, Texas) and Prism 5 (GraphPad, LaJolla, California).
Results
In total 255 consecutive patients who had platelet testing results by LTA and VNP2Y12 after PCI were analyzed; of these 50 clopidogrel-naive patients had testing done before and after PCI. Baseline characteristics of patients enrolled are presented in Table 1 . In total 30.6% of patients were already on long-term clopidogrel therapy at time of presentation, and 70.5% of these patients received an additional loading dose of clopidogrel at time of PCI (median 300 mg, interquartile range 300 to 600). All clopidogrel-naive patients were treated with a loading dose of clopidogrel in the peri-PCI setting (median 600 mg, interquartile range 600 to 600).
| Characteristic | Number of Patients (%) | Median (IQR) |
|---|---|---|
| Age (years) | 65 (57–73) | |
| Men | 171 (67.1%) | |
| Caucasian | 242 (94.9%) | |
| Body mass index (kg/m 2 ) | 28.7 (25.5–33.1) | |
| Smokers | 45 (17.6%) | |
| Diabetes mellitus | 87 (34.1%) | |
| Known coronary artery disease | 122 (47.8%) | |
| Chronic kidney disease ⁎ | 69 (27.1%) | |
| Hypertension | 212 (83.5%) | |
| Hypercholesterolemia | 186 (72.9%) | |
| Cancer | 30 (11.8%) | |
| Congestive heart failure | 54 (21.2%) | |
| Stroke | 13 (5.1%) | |
| Peripheral vascular disease | 41 (16.1%) | |
| Presentation as myocardial infarction | 93 (36.5%) | |
| Eptifibatide use | 31 (12.2%) | |
| Bivaluridin use | 123 (48.8%) | |
| Platelets (10 6 /ml) | 189 (158–234) | |
| Mean platelet volume (fL) | 8.1 (7.6–8.7) | |
| White blood cell count (10 6 /ml) | 8.0 (6.4–10.0) | |
| Hemoglobin (g/dl) | 12.7 (11.5–14.1) | |
| Anemia | 95 (37.3%) | |
| Medications on presentation | ||
| Long-term clopidogrel | 78 (30.7%) | |
| Long-term aspirin | 176 (69.0%) | |
| Statin | 152 (59.6%) | |
| Proton pump inhibitor | 82 (32.3%) |
⁎ Estimated glomerular filtration rate <60 ml/min/1.73 m 2 by Modification of Diet in Renal Disease equation.
Platelet aggregation by LTA in response to adenosine diphosphate correlated well with the VNP2Y12 assay (Spearman r = 0.64, p <0.0001). The overall proportions of patients with HRPR in our entire studied population were 23.4% by LTA and 40.4% by VNP2Y12. Univariate analysis was then performed for all clinical and laboratory variables collected as listed in Table 1 and using platelet reactivity testing results as a categorical variable. Only variables associated with HRPR on univariate analysis with a p value <0.1 were then included in a multivariate logistic regression in 2 separate models for LTA and VNP2Y12 ( Figure 1 ). Goodness of fit in these 2 models was then confirmed using a Hosmer–Lemeshow test (p = 0.32 and 0.83 for VNP2Y12 and LTA, respectively, with p <0.05 indicating lack of fit). For the LTA model only anemia, gender, and administration of a clopidogrel load at time of PCI emerged as independent predictors of platelet reactivity ( Figure 1 ). For the VNP2Y12 logistic regression model the only independent variables for HRPR were anemia and diabetes mellitus ( Figure 1 ). Ultimately, the only common denominator between these assays that was independently predictive of HRPR was anemia. Using LTA proportions of patients with HRPR were 34.3% in anemic patients and 15.6% in patients with normal hemoglobin levels. For VNP2Y12 prevalences of HRPR were 59.8% in anemic patients and 25.9% in patients with normal hemoglobin (p <0.005 for the 2 comparisons).
We also investigated whether anemia influences platelet response to clopidogrel at an individual patient level using the subgroup of clopidogrel-naive patients who had testing done before and after clopidogrel therapy initiation. Of these 14 were anemic (average hemoglobin 11.3 ± 0.3 g/dl) and 36 patients had normal hemoglobin (14.2 ± 0.2 g/dl) based on postprocedure hemoglobin values. In the 2 subgroups before clopidogrel administration there was no difference in percent aggregation by LTA or PRU values between anemic and nonanemic patients. However, the change in platelet response to adenosine diphosphate was significantly decreased in anemic patients compared to patients with normal hemoglobin with either assay ( Figure 2 ). Percent maximal aggregation by LTA in response to adenosine diphosphate 5 μmol/L decreased by 15.8 ± 5.8% in anemic patients and by 28.8 ± 3.2% in patients with normal hemoglobin (p <0.05; Figure 2 ). Likewise, change in PRUs after the thienopyridine load was lower in patients with anemia (56.5 ± 35.5 PRUs) compared to patients with normal hemoglobin (145.0 ± 14.2 PRUs, p <0.05; Figure 2 ). The same was observed when comparing percent inhibition by the VNP2Y12 assay, which was significantly higher in patients with normal hemoglobin versus anemic patients (43.0 ± 4.1% vs 16.9 ± 8.8%, p <0.05). Before clopidogrel there was a very weak linear correlation between hemoglobin levels before the procedure and percent maximum aggregation by LTA (Spearman r = −0.23, p = 0.1; Figure 3 ) or PRUs by VNP2Y12 (Spearman r = −0.30, p = 0.04; Figure 3 ) before the procedure. After the clopidogrel load, however, a significant correlation of platelet aggregation response to adenosine diphosphate relative to hemoglobin levels after the procedure was present with the 2 assays (LTA, Spearman r = −0.37, p = 0.006; VNP2Y12, Spearman r = −0.46, p = 0.0007; Figure 3 ). The same robust correlation between hemoglobin and LTA or VNP2Y12 results was obtained after PCI when using the entire dataset (Spearman r = −0.25 for LTA, Spearman r = −0.45 for VNP2Y12, p <0.0001; Figure 3 ).
