Assessment of high on-treatment platelet reactivity in patients with ischemic heart disease: concordance between the Multiplate and VerifyNow assays

Authors


Correspondence: Luciano Consuegra-Sánchez, Cardiology Department, Hospital Universitario de Santa Lucia, Cartagena, Spain and Cardiovascular Research Centre, St. George’s University of London, Cranmer Terrace, London SW17 ORE, UK.Tel.: +44 658659063; fax: +44 968523865.E-mail: lconsue@gmail.com

While the clinical relevance of assessing the antiplatelet effects of available medications is established, the optimal method to quantify platelet reactivity as well as the definition (i.e. required threshold) of high on-treatment platelet reactivity (HOPR) are controversial. The VerifyNow (Accumetrics, San Diego, CA, USA) and Multiplate (Dynabite, Munich, Germany) are point-of-care analyzers which are easy to use in the clinical setting. However, only a few previous studies have compared systematically the performance of these analyzers regarding the assessment of HOPR.

From April to October 2011, we consecutively recruited patients with a diagnosis of ischemic heart disease undergoing cardiac catheterization in two Spanish University Hospitals (Hospital Universitario de Santa Lucía, Cartagena and Hospital Universitario San Juan, Alicante). Patients were considered eligible if they presented with: (i) acute coronary syndrome (ACS), defined as typical chest pain and elevation of markers of necrosis and/or abnormal T wave/ST segment shifts; or (ii) chronic stable angina, defined as ≥ 3 months history of typical exertional chest pain relieved by rest, and a positive stress test; and (iii) clopidogrel treatment ≥ 4 h from the initial loading dose (or in the absence of a clopidogrel loading dose, clopidogrel treatment for ≥ 24 h). Exclusion criteria included: heart valve disease or cardiomyopathy (n = 80), life expectancy < 1 year (n = 42), an inability to sign an informed consent form (n = 12) and treatment with IIb-IIIa inhibitors (n = 75). The present study complied with the Declaration of Helsinki and was approved by the institutional ethics committee.

Twenty millilitres of whole peripheral blood was extracted in the catheterization laboratory. The time interval between sampling and analyses was ≤ 2 h in all cases. We obtained the following results with VerifyNow: (i) ‘Base PRU’: reactivity by agonists of thrombin receptors; (ii) ‘P2Y12 Test-PRU’: reactivity by agonists of ADP; (iii) ‘Percentage of inhibition’: calculated as (PRU-basePRU)/(base PRU) ×100 and (iv) ‘Aspirin Test-ARU’: reactivity by arachidonic acid. With the Multiplate analyzer, we assessed: (i) the ‘Aspi-test’: reactivity by arachidonic acid; (ii) the ‘ADP-test’: reactivity by agonists of ADP; and (iii) the ‘ADPhs test’: reactivity by agonists of ADP with increased sensitivity.

We characterized clopidogrel resistance by applying reference values reported in the consensus report of the Working Group on High On-Treatment Platelet Reactivity to ADP [1]. Accordingly, a P2Y12 Test > 235 PRU was considered to indicate HOPR [1]. We also considered a percentage inhibition < 20% to indicate HOPR based both on the manufacturer’s specifications and reports in previous studies [2,3]. With regards to the Multiplate analyzer, the definition of a low response varies from study to study [4–6]. Here we considered both the specifications of the manufacturer (i.e. ADP Test > 53 U and ADPhs Test > 31 U) [6] and recommendations of the consensus statement (i.e. ADP test > 42 U [416 AU*min], upper quintile) [1,4]. Aspirin resistance was defined as an Aspirin Test > 550 ARU and an Aspi Test > 74 U [6–8].

To assess the degree of concordance between analyzers, we calculated the kappa coefficient (κ) and performed a graphical agreement analysis using the Bland–Altman method (calculation of the average of the differences and 95% confidence intervals [CIs]) with the macro Domenech JM. Macro! AGREE for SPSS Inc. (Chicago, IL, USA). Additionally, the percentage of observations outside the decision limit interval and Spearman’s correlation coefficient (rho) for each comparison were calculated. We used SPSS v.20.

Two hundred and fifty-one patients were recruited (mean age 68 ± 11 years; 68.5% men); 41% of the patients had type 2 diabetes mellitus and 25% were current smokers. Two hundred and thirty-one patients (92.0%) presented with ACS and 58.6% of these showed elevation of necrosis markers. The most frequent diagnosis was ACS without persistent ST segment elevation (72.1%). One hundred and sixty-eight (66.9%) patients received a loading dose of clopidogrel (151 patients received 300 mg [66.9%] and 17 patients received 600 mg [33.1%]). Among the patients who did not receive a loading dose, 40.0% were taking clopidogrel for ≥ 4 weeks prior to admission. The median time elapsed between the onset of aspirin and clopidogrel treatment and the assessment of platelet reactivity was 4 (IR 2–7) and 4 (2–8) days, respectively. One hundred and seventy-three (68.9%, 95% confidence interval [CI] 63.3–74.1) patients showed a PRU > 235, 161 (64.1%, 95% CI 58.2–69.7) percentage of inhibition < 20, 48 (19.1%, 95% CI 14.7–23.9) ADP test > 53 U, 75 (29.9%, 95% CI 25.1–35.5) ADP test > 42 U and 89 (35.5%, 95% CI 29.9–41.9) ADPhs test > 31 U. We identified 38 (15.3%, 95% CI 10.8–19.3) patients with ARU > 550 and only 6 (2.4%, 95% CI 0.8–4.4) with an Aspi Test > 74 U.

Table 1 shows the agreement coefficients between the analyzers. The greater agreement coefficients were observed between percentage inhibition < 20 vs. the ADPhs Test > 31 and percentage of inhibition < 20 vs. the ADP Test > 42 (both κ < 0.3, P < 0.001). The remaining comparisons showed κ < 0.2. Regarding aspirin, the agreement observed between the analyzers was low (κ = 0.194, P = 0.001).

Table 1. Concordance of the high on-treatment platelet reactivity by Multiplate and VerifyNow assays, according to the predefined cut points
 VerifyNowMultiplate
 ClopidogrelAspirinAspirinClopidogrel
 PRU > 235Inhibition < 20%ARU > 550 UAUC AspiTest > 74 UAUC ADPTest > 53 UAUC ADPhsTest > 31 U
  1. AUC, area under the curve; ADP test, reactivity by agonists of the ADP receptor; ADPhs Test, High Sensitivity ADP Test; ARU, reactivity by arachidonic acid; Aspi Test, reactivity by arachidonic acid; PRU, reactivity by agonists of ADP; 1 U, 10 Arbitrary Units × min. For each comparison we show the kappa coefficient and P-value. It is generally accepted that a k statistic between 0 and 0.2 translates into slight agreement, 0.2–0.4 into fair agreement, 0.4–0.6 into moderate agreement, 0.6–0.8 into substantial agreement and 0.8–1 into almost perfect agreement.

PRU > 235−0.418&#U2028;< 0.0010.115&#U2028;0.0020.182&#U2028;< 0.001
Inhibition < 20%−0.178&#U2028;0.002−0.276&#U2028;< 0.001
AUC AspiTest > 74 U0.194&#U2028;0.001
AUC ADPTest > 53 U0.446&#U2028;< 0.001
AUC ADPTest > 42 U0.184&#U2028;< 0.001−0.232&#U2028;< 0.0010.549&#U2028;< 0.001

In all cases only a few observations outside the 95% CIs were observed with the Bland–Altman method, being 10 (3.98%) with regards to the comparison of PRU vs. the ADP test, 11 (4.38%) PRU vs. the ADPhs Test, 17 [6.78%] percentage of inhibition vs. the ADP test, 14 [5.58%] percentage of inhibition vs. the ADPhs Test and 8 (3.21%) ARU vs. the Aspi-Test. The study of Spearman’s rho coefficients showed a low correlation between assays (all ρ; < 0.35, all P < 0.001).

The main finding of this study is that the concordance between VerifyNow and Multiplate measurements regarding the responsiveness to aspirin and clopidogrel is unexpectedly poor, when applying currently recommended cut points. This indicates that conclusions regarding platelet reactivity in the clinical setting are highly dependent on the analyzer used.

Studies have, in general, reported a good agreement between different methods to evaluate the degree of platelet inhibition achieved by aspirin/clopidogrel [9]. However, several previous studies have reported substantial discrepancies. Woo et al. [10], in patients with ischemic heart disease, observed κ ≤ 0.25 between light transmission aggregometry and Multiplate/VerifyNow. In a similar setting, Ko et al. [11] reported both a poor correlation and agreement between Multiplate and VerifyNow in 222 patients; these findings, in a different patient population, are consistent with our results. Importantly, in our study, a poor concordance was evident particularly when we analyzed data considering predefined cut points and was less marked when data were analyzed as a continuum using graphical methods. This suggests that the selection of cut off points represents a critical issue, as many of these have not been clinically validated. Other factors influencing the test results include: the type of patients investigated, the method for selection of cut offs and the adverse events considered when generating receiver-operating characteristic curves. We also acknowledge limitations related to the design of the study, i.e. the short interval between clopidogrel administration and sampling that might have determined the observed high proportion of HOPR.

We therefore conclude that the degree of concordance between VerifyNow and Multiplate analyzers regarding the diagnosis of HOPR with aspirin and clopidogrel is poor and largely dependent on the selected diagnostic cut points.

Acknowledgement

This study was supported by a grant from FFIS-Caja Murcia 10/006.

Disclosure of Conflict of Interest

The authors state that they have no conflict of interest.

Ancillary