Laboratory assessments of therapeutic platelet inhibition in endovascular neurosurgery: complication prediction using the VerifyNow P2Y12 assay and thromboelastography with platelet mapping

OBJECTIVE: Inhibition of platelet aggregation is universally used to prevent thromboembolic complications related to stent placement in endovascular neurosurgery, but excessive inhibition potentiates hemorrhagic complications. Previously, the authors demonstrated that two different commercially available measures of adenosine diphosphate (ADP)-dependent platelet inhibition-the VerifyNow P2Y12 clopidogrel assay (measured in platelet reactivity units PRU) and maximal amplitude (MA) attributable to ADP activity (MA-ADP) derived from thromboelastography (TEG) with platelet mapping (PM)-yielded wildly different results. This study sought to analyze observed complications to quantify the ideal therapeutic windows for both tests. METHODS: Ninety-one patients with simultaneous or near-simultaneous PRU and TEG-PM results who underwent craniocervical endovascular stenting at the authors' institution between September 2015 and November 2017 were identified and retrospectively enrolled. From November 2017 until June 2019, 109 additional patients were prospectively enrolled. For this study, in-hospital thrombotic and hemorrhagic complications (both CNS and non-CNS) were tabulated, and receiver operating characteristic (ROC) curve analysis was used to identify threshold values of the PRU and MA-ADP for predicting each type of complication. RESULTS: Of the 200 patients enrolled, 7 were excluded because of anemia or thrombocytopenia outside of the test manufacturer's specified ranges and 1 was excluded because they did not have a TEG-PM result. Including complications of all severities, there were a total of 15 CNS thrombotic complications, 1 access-site thrombotic complication, 3 CNS hemorrhages, 8 access-site hemorrhagic complications, and 3 hemorrhagic complications not affecting either the CNS or the access site. ROC curve analysis yielded therapeutic threshold values of 118-144 PRU. The results demonstrated PRU has a significant dose-dependent effect on the rates of thrombosis and hemorrhage. Logistic regression models did not demonstrate statistically significant relationships between the MA-ADP and either thrombosis or hemorrhage. ROC analysis based on these models is of little value and did not identify significant threshold values for MA-ADP. CONCLUSIONS: There continues to be poor correlation between the results of TEG-PM and PRU. PRU accurately predicted complications, with a relatively narrow ideal value range of 118-144. The MA-ADP alone does not appear able to accurately predict either hemorrhagic or thrombotic complications in this group.