High on-treatment platelet reactivity assessed by various platelet function tests: is the consensus-defined cut-off of VASP-P platelet reactivity index too low?

Authors


Paul A. Gurbel, Sinai Center for Thrombosis Research, Cardiac Catheterization Laboratory, 2401 W. Belvedere Ave., Baltimore, MD 21215, USA.
Tel.: +1 410 601 9600; fax: +1 410 601 9601.
E-mail: pgurbel@lifebridgehealth.org

Recent consensus of the Working Group has suggested that high on-treatment platelet reactivity (HPR) to ADP is a major cardiovascular risk factor [1]. This document also has provided the potential cut-offs of HPR to ADP for the most commonly used platelet function tests (PFTs), including light transmittance aggregometry (LTA), the VerifyNow P2Y12 assay (VerifyNow), Multiplate and the vasodilator-stimulated phosphoprotein-phosphorylation (VASP-P) assay. Although the concept of HPR is a major step forward for a personalized antiplatelet therapy strategy in high-risk patients undergoing percutaneous coronary intervention (PCI), several issues have been raised and delayed the clinical adoption of PFT [2,3]. The different correlations and degree of agreement between HPR determined by various PFTs are important concerns. Although the various PFTs identify different aspects of platelet biology, several studies have suggested matched cut-offs of HPR measured by LTA, VerifyNow and Multiplate [4,5], strengthening the reliability of the hypothesis of HPR being an important indicator of risk.

The higher prevalence of HPR determined by the VASP-P assay in clopidogrel- or prasugrel-treated patients (> 50% and ∼25%, respectively) compared with other assays during clopidogrel therapy (20∼40%) [6–10], questions that the consensus-defined cut-off value of > 50% platelet reactivity index (PRI) may be so low that the true proportion of high-risk patients for ischemic events is overestimated [2,3]. If we keep in mind the relatively low rate of post-PCI ischemic events, an estimation of HPR exceeding 50% during clopidogrel therapy may cause unnecessary application of potent P2Y12 receptor inhibitors with an attendant increased risk of bleeding. We therefore performed the current pharmacodynamic analysis to determine the agreement for the definition and the prevalence of HPR among various PFTs in patients with stable coronary artery disease treated with dual antiplatelet therapy [6].

Platelet function measurements from two prospective randomized studies [6] were used in this analysis; 37 and 39 patients were enrolled from our institution and treated with ticagrelor (180-mg loading-dose [LD] and 90-mg twice a day maintenance-dose [MD]) and clopidogrel (600-mg LD and 75-mg d−1 MD), respectively. Serial LTA, VerifyNow and VASP-P assays were performed, and a total of 1002 matched pairs (ticagrelor = 498 pairs; clopidogrel = 504 pairs) were available for this analysis.

The overall inter-assay correlation and receiver-operating characteristics (ROC) curve analysis determined by the level of platelet inhibition were consistent irrespective of clopidogrel or ticagrelor administration. The LTA values correlated most with VerifyNow (Fig. 1A, = 0.821) and less well with the VASP-P assay (Fig. 1B, r = 0.688). PRI values in low/medium responsive patients were more scattered compared with LTA measurements, a finding that is in line with previous observations [11,12].

Figure 1.

 Inter-assay agreement and ROC curve analysis for the cut-off of HPR. (A) Agreement between 5 μm ADP-PA > 46% (black arrow) and PRU > 235 (black arrow). (B) Agreement between 5 μm ADP-PA > 46% (black arrow) and VASP-P PRI > 50% (black arrow) vs. > 60% (green arrow). (C) The cut-off of P2Y12 reaction units corresponding to 5 μm ADP-PA > 46%. (D) The cut-off of VASP-P PRI corresponding to 5 μm ADP-PA > 46%. ROC, receiver-operating characteristics; HPR, high on-treatment platelet reactivity; PA, platelet aggregation; VASP-P, vasodilator-stimulated phosphoprotein-phosphorylation; PRI, platelet reactivity index; AUC, area under curve; CI, confidence interval.

The frequency of HPR (based on consensus-defined cut-offs) was the greatest by VASP-P PRI (47.5%) compared with 5 μm ADP-induced aggregation > 46% (37.3%) and P2Y12 reaction units (PRU) > 235 (40.0%). A PRU cut-off > 235 showed substantial agreement with > 46% LTA cut-off (κ = 0.737; concordance, 87.5%; balanced distribution of discordance) (Fig. 1A). The VASP-P PRI cut-off showed moderate agreement with the LTA cut-off (κ = 0.585; concordance, 79.6%; relatively unbalanced distribution of discordance) (Fig. 1B). ROC curve analysis was used to assess the corresponding points to 5 μm ADP-induced aggregation > 46%, PRU > 234 and VASP-P PRI > 60% (Fig. 1C, D). When the new HPR criteria of a VASP PRI > 60% were applied, the agreement with the LTA cut-off improved (κ = 0.634; concordance, 82.6%; narrowed discordance) (Fig. 1B: green arrow and numbers). We also performed ROC curve analysis to assess the corresponding cut-offs to VASP-P PRI > 50%. The matched point with 5 μm ADP-induced aggregation was > 37% (AUC [area under curve], 0.887; 95% confidence interval [CI], 0.843–0.890; P < 0.001; sensitivity, 79.5%; specificity, 82.7%). PRU > 167 was identified as the matched point to > 50% VASP-P PRI (AUC, 0.897; 95% CI, 0.875–0.918; P < 0.001; sensitivity, 87.1%; and specificity, 80.3%).

To the best of our knowledge, this is the first comparative analysis describing the agreement between the definitions of the HPR cut-off and the frequency of HPR by LTA, VerifyNow and VASP-P assays in patients treated with clopidogrel or ticagrelor. The present study demonstrates that the consensus-defined cut-offs of HPR correlate well between VerifyNow and LTA, whereas the cut-off of > 50% VASP-P PRI appears to overestimate the frequency of HPR. This observation was consistent irrespective of clopidogrel or ticagrelor treatment. Based on these data, a VASP-P PRI > 60% is suggested as the criterion for HPR to be studied in future trials. In addition, our findings support that the cut-offs of HPR may be used interchangeably in investigations of the relation of platelet reactivity to post-PCI ischemic event occurrence. [1].

However, the clinical meaning of < 50% VASP-P PRI should not be overlooked. Patients with VASP-P PRI < 50% seldom showed ischemic events occurrence [1], similar to < 170 PRU as the suggested ‘immunity’ cut-off from the results of the GRAVITAS (Gauging Responsiveness with A VerifyNow assay – Impact on Thrombosis And Safety) trial [13]. As the optimal cut-off of HPR suggested by ROC curve analysis is the point showing the greatest sum of sensitivity and specificity in a certain assay, it does not mean that patients below this cut-off will not suffer from an ischemic event. The above findings from clinical trials suggest that an ‘immunity’ cut-off to nearly eliminate the risk of post-PCI thrombosis can be achievable in the era of potent P2Y12 inhibitors and further enhance our understanding of a ‘therapeutic window’ for P2Y12 inhibitors. Along with this concept, an optimal cut-off value for bleeding needs to be clarified with a delicate bleeding scale in future adequately powered clinical trials.

It is noteworthy to remember that clinical studies have demonstrated a wide range of VASP-P PRI cut-offs (48∼57%) depending on the characteristics of the patients enrolled [1,14]. Although El Ghannudi and Freynhofer et al. recently reported that ∼60% cut-off of VASP-P PRI optimally predicted the risk of ischemic events in PCI-treated patients [15,16], their studies used median fluorescence intensity, contrary to the previous studies where mean fluorescence intensity was used [2,6–9,14]. Therefore, clinical outcome data to demonstrate increased clinical efficacy and safety are needed before the consensus cut-off values can be modified and to implement the proposed VASP-P PRI cut-off value in future studies.

Acknowledgements

The authors acknowledge T. Gesheff and M.J. Antonino for technical assistance.

Disclosures of conflict of interests

The ONSET/OFFSET (NCT00528411) and RESPOND (NCT00642811) studies were funded by AstraZeneca LP (Wilmington, Delaware, USA). Y.-H. Jeong received honoraria for lectures from Sanofi-Aventis, Daiichi Sankyo Inc. and Otsuka. P. A. Gurbel received research grants, honoraria and consultant fees from Haemoscope, AstraZeneca, Schering-Plough/Merck, Medtronic, Lilly/Daiichi Sankyo Inc., Sanofi Aventis/Bristol Myers, Portola, Boston-Scientific, Bayer, Norvatis, Accumetrics, Boehringer Ingelheim and Johnson and Johnson.

Ancillary