Article Text

Download PDFPDF

Original research
Platelet reactivity and hemorrhage risk in neurointerventional procedures under dual antiplatelet therapy
  1. Hidehisa Nishi1,
  2. Ichiro Nakahara2,
  3. Shoji Matsumoto1,3,
  4. Tetsuya Hashimoto4,
  5. Tsuyoshi Ohta5,
  6. Nobutake Sadamasa1,
  7. Ryota Ishibashi1,
  8. Masanori Gomi1,
  9. Makoto Saka1,
  10. Haruka Miyata1,
  11. Sadayoshi Watanabe2,
  12. Takuya Okata1,
  13. Kazutaka Sonoda1,
  14. Junpei Kouge1,
  15. Akira Ishii1,
  16. Izumi Nagata1,
  17. Jun-ichi Kira3
  1. 1Department of Neurosurgery, Kokura Memorial Hospital, Fukuoka, Japan
  2. 2Department of Neurosurgery, Fujita Health University, Aichi, Japan
  3. 3Department of Neurology, Neurological Institute, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan
  4. 4Department of Cerebrovascular Medicine, National Cerebral and Cardiovascular Center, Osaka, Japan
  5. 5Department of Neurosurgery, Kochi Health Science Center, Kochi, Japan
  1. Correspondence to Dr Hidehisa Nishi, Department of Neurosurgery, Stroke Center, Kokura Memorial Hospital, 3-2-1 Asano, Kokurakita-ku, Kitakyushu-shi, Fukuoka 802-8555, Japan; venturahighway83{at}


Background and purpose Hemorrhagic complications during neurointerventional procedures have various etiologies and can result in severe morbidity and mortality. This study investigated the possible association between low platelet reactivity measured by the VerifyNow assay and increased hemorrhagic complications during elective neurointervention under dual antiplatelet therapy.

Methods From May 2010 to April 2013 we recorded baseline characteristics, P2Y12 reaction units (PRU), and aspirin reaction units using VerifyNow. The primary endpoint was post-procedural hemorrhagic complications.

Results A total of 279 patients were enrolled and 31 major hemorrhagic complications (11.1%) were identified. From receiver-operating characteristic curve analysis, PRU values could discriminate between patients with and without major hemorrhagic complications (area under the curve 0.63). Aspirin reaction unit values had no association with the primary outcome. The optimal cut-off for the primary outcome (PRU ≤175) was used to identify the low platelet reactivity group. The incidence of hemorrhagic complications was 20.0% in this group and 8.9% in the non-low platelet reactivity group. Multivariate analysis identified low platelet reactivity as an independent predictor for hemorrhagic complications.

Conclusions The risk of hemorrhagic complications during elective neurointervention including cerebral aneurysm coil embolization and carotid artery stenting under dual antiplatelet therapy is associated with the response to clopidogrel but not to aspirin. A PRU value of ≤175 discriminates between patients with and without hemorrhagic complications. Future prospective studies are required to validate whether a specific PRU value around 170–180 is predictive of hemorrhagic complications.

  • Intervention
  • Drug
  • Pharmacology
  • Hemorrhage

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Similar to percutaneous coronary intervention (PCI), dual antiplatelet therapy with aspirin and an antagonist of the P2Y12 receptor (a chemokine receptor for ADP) such as clopidogrel is widely used as a standard intervention during elective neurointerventional procedures to prevent thrombotic complications.1

To assess platelet reactivity, the VerifyNow P2Y12 assay (Accumetrics, San Diego, California, USA) is widely used as a point-of-care platelet function test that can measure the degree of P2Y12 receptor inhibition within minutes. This assay strongly correlates with light transmission aggregometry, the gold standard for quantification of platelet reactivity in patients treated with clopidogrel, prasugrel, or ticagrelor.2–5 The VerifyNow aspirin assay can measure the degree of aspirin inhibition, which indicates the amount of thromboxane A2-mediated activation of the glycoprotein IIb/IIIa receptors that contribute to platelet aggregation.

Several studies have shown that low reactivity to clopidogrel is associated an increased risk of early major bleeding or entry site complications after PCI.6–9 On the other hand, the relationship between P2Y12 reaction unit (PRU) values and complications resulting from neurointervention have rarely been reported.10–12 Unlike PCI, neurointervention requires a larger sheath size, occasional multiple access sites, and catheterization of smaller vessels. Therefore, the risk of hemorrhagic complications during neurointervention may be independent of and greater than those observed during PCI.

The present study investigated possible associations between platelet reactivity and the occurrence of hemorrhagic complications during elective neurointervention while undergoing dual antiplatelet therapy.


Patient population and study design

From May 2010 to April 2013 we retrospectively reviewed consecutive patients pretreated with aspirin and clopidogrel prior to scheduled neurointervention procedures including carotid artery stenting, vertebral artery stenting, and coil embolization of unruptured cerebral aneurysm at our institution. Patients received 100 mg/day aspirin and 75 mg/day clopidogrel for at least 5 days before the procedure, or 200 mg aspirin and 300 mg clopidogrel as a loading dose. Procedural anticoagulation consisted of administration of unfractionated heparin administration to achieve an activated clotting time (ACT) of 250–300 s. Patients who used concomitant antiplatelet drugs or anticoagulants other than aspirin and clopidogrel were excluded.

Blood sampling and platelet function analysis

A blood sample was drawn from the subject's medial cubital vein with a 21-gauge needle and platelet reactivity was measured in the catheterization laboratory using VerifyNow PRU and aspirin reaction unit (ARU) assays immediately before treatment.

Endpoints and definitions

The primary endpoint was the presence of hemorrhagic complications within 1 week following neurointervention, using the International Society on Thrombosis and Hemostasis (ISTH) criteria.13 We divided the study population based on their platelet aggregation values using a cut-off value derived from receiver-operating characteristic (ROC) curve analysis to discriminate between patients with or without ISTH bleeding events. Patients with PRU values below the optimal cut-off for hemorrhagic complications were classified as having low platelet reactivity (LPR) and the other patients were classified as members of the non-LPR group. This strategy of classification using ROC curve analysis was based on a previous report in the cardiac literature.6 ,7 The incidence of hemorrhagic complications was also compared between the LPR and non-LPR groups.

Statistical methods

Variables are presented as mean± SD, counts, percentages, or median with IQR. Categorical variables were compared using the χ2 test, likelihood ratio test, or Pearson product-moment correlation coefficient. ROC curve analysis was calculated to determine the optimal cut-off value for discrimination based on the primary outcome. The study population was stratified according to the results of the ROC curve analysis. The LPR group consisted of patients with PRU values below the optimal cut-off for hemorrhagic complications and all other patients were placed in the non-LPR group.

Demographic and procedural variables were first evaluated in a univariate model. Variables with p values <0.15 were entered into a multiple logistic regression model to test for an independent association with the primary endpoint. ORs and the corresponding 95% CIs were calculated for each variable included in the multivariate model. A p value of <0.05 was considered statistically significant. The p value is expressed using two significant figures; if the value is <0.01, it is expressed as p<0.01. All analyses were performed using the JMP V.10.0.2 software (SAS Institute, Cary, North Carolina, USA).


Study population

Of 324 patients recruited into the study (204 cases of coil embolization for unruptured cerebral aneurysm, 3 cases of vertebral artery stenting, and 117 cases of carotid artery stenting for stenosis), 5 patients were excluded because of a lack of historical clinical data and 37 patients were excluded because of the presence of concomitant antiplatelet drugs or anticoagulants.

Following the exclusion of ineligible subjects, data obtained from 279 patients who underwent neurointerventional procedures were analyzed (174 cases of coil embolization and 105 cases of carotid artery stenting). The clinical characteristics and procedural data of these patients are shown in table 1. The mean patient age was 64.1±12.5 years, 136 (48.7%) were women, and 160 (57.3%) had a history of hyperlipidemia, 158 (56.6%) of whom had received statin therapy. All procedures were elective, and 217 (79.4%) patients had received dual antiplatelet drugs for >5 days before treatment. During the study procedure the sheath size varied from 5 Fr to 9 Fr and 38 patients (13.6%) required multiple catheter access sites. A percutaneous closure device was used to treat 241 patients (86.3%). The mean±SD maximum ACT value during the procedure was 341±71.6 s (median 329 (IQR 296–367)).

Table 1

Characteristics of study patients

Response to various antiplatelet medications

Figure 1 shows the ARU and PRU distribution values. The ARU value range was 269–597 (mean 429; median 407; IQR 389–465) and the PRU value range was 7–444 (mean 227; median 229; IQR 162–291). The PRU SD was greater than the ARU SD.

Figure 1

Distribution of aspirin reaction unit (ARU) and P2Y12 reaction unit (PRU) values among the study population.

Primary endpoint

Within the study groups, a total of 31 major hemorrhagic events occurred (11.1%). Of the bleeding events, 10 (3.5%) were intracranial hemorrhages, 9 were subarachnoid hemorrhages, and 1 was a cerebral hemorrhage. There were no cases of hemorrhagic infarction among these intracranial hemorrhages. Seventeen events (6.0%) were puncture site hemorrhages, 3 (1.0%) were retroperitoneal hemorrhages, and 1 (0.3%) was hemoptysis. The PRU values were significantly associated with hemorrhagic complications (p<0.01), but ARU values were not (p=0.70). As shown in figure 2, ROC curve analysis showed that PRU values could discriminate moderately between patients likely to experience hemorrhagic complications (area under the curve (AUC) 0.63). A PRU value of ≤175 was found to be the optimal cut-off point for predicting hemorrhagic complications, with a sensitivity of 54%, specificity of 72%, negative predictive value (NPV) of 81%, and positive predictive value of 19%. Patients with PRU ≤175 were assigned to the LPR group and patients with PRU >175 were assigned to the non-LPR group.

Figure 2

Receiver-operating characteristic curves for International Society on Thrombosis and Hemostasis (ISTH) bleeding events. Area under the curve 0.63; optimal cut-off 175; sensitivity 54%; specificity 72%; negative predictive value 81.0%; positive predictive value 19.0%.

The LPR group accounted for 30% of all patients in the present study. Patients in the LPR group were younger than those in the non-LPR group (p=0.01). The LPR group also had a lower hemoglobin count (p<0.01), a lower hematocrit level (p<0.01), and a higher platelet count (p<0.01) than those in the non-LPR group. For procedure-related factors, more patients in the LPR group required multiple access sites than patients in the non-LPR group. No differences were found in maximum sheath size, use of a percutaneous closure device, or clopidogrel regimen (loading dose or maintenance dose).

Group characteristics are shown in table 1 and the incidence of hemorrhagic complications by group are reported in table 2. Patients in the LPR group experienced more bleeding events than those in the non-LPR group (ISTH bleeding events 19.0% vs 7.69%, p<0.01). In the LPR group, three subarachnoid hemorrhages occurred (3.57%), two of which were caused by intraprocedural perforation. In the non-LPR group, six subarachnoid hemorrhages (3.07%) and one intracerebral hemorrhage (0.51%) occurred, with three of the seven intracranial hemorrhages being caused by intraprocedural perforation and the other three intracranial hemorrhages caused by hyperperfusion syndrome after carotid artery stenting. In both groups the puncture site was the most common location of hemorrhagic complication (LPR group 13.0% vs non-LPR group 3.07%). Among other locations, two retroperitoneal hematomas occurred in the LPR group and one retroperitoneal hematoma and one hemoptysis occurred in the non-LPR group. In subcategories, the LPR group reported more hemorrhagic complications only at the puncture site and the retroperitoneum.

Table 2

Bleeding outcomes

Bleeding events categorized by the endovascular procedure are reported in table 3. Patients in the LPR group had more bleeding events than patients in the non-LPR group for both cerebral aneurysm coil embolization and carotid artery stenting (coil embolization: ISTH bleeding events 17.5% vs 7.69%, p=0.05; carotid artery stenting: ISTH bleeding events 22.2% vs 7.69%, p=0.04). With cerebral aneurysm coil embolization, three intracranial hemorrhages were observed in both the LPR and non-LPR groups (5.26% vs 2.56%). All of these intracranial hemorrhages resulted from intraprocedural perforation, except one case in the LPR group who had a subarachnoid hemorrhage on postoperative day 1. With carotid artery stenting, no intracranial hemorrhages were observed in the LPR group and four intracranial hemorrhages were observed in the non-LPR group (0.00% vs 5.12%). Three of the four intracranial hemorrhages in the non-LPR group were caused by hyperperfusion syndrome.

Table 3

Bleeding outcomes according to endovascular procedure

Multivariate analysis

In a univariate analysis of the primary endpoint, the factors with a p value <0.15 were hypertension, body mass index, and LPR. All factors were entered into the multivariate analysis and the results are shown in table 4. Multivariate analysis identified a PRU value ≤175 as an independent predictor of an increased risk for hemorrhagic complications (OR 2.61; 95% CI 1.19 to 5.72).

Table 4

Univariate and multivariate predictors of major hemorrhagic complications


In the present study we found that the clopidogrel response is associated with the risk of developing hemorrhagic complications from neurointervention under dual antiplatelet therapy, and a threshold PRU value of ≤175 is suitable to predict hemorrhagic complications during elective neurointervention under dual antiplatelet therapy. Those patients with a PRU ≤175 accounted for 30% of the study population and were shown to have a nearly threefold greater risk of hemorrhagic complications than patients with a PRU >175.

The relationship between PRU and hemorrhagic complications during neurointervention has been reported previously. Delgado Almandoz et al11 ,12 ,14 reported that a PRU <60 was an independent predictor of major perioperative hemorrhagic complications for all types of elective cerebral aneurysm treatments, including simple or balloon-assisted aneurysm coiling, stent-assisted aneurysm coiling, and flow diverter stenting. However, in that report the PRU cut-off line was set arbitrarily. Goh et al10 investigated 47 patients and found that PRU inhibition ≥72% was an optimal cut-off point for predicting hemorrhagic complications stemming from a neurointeventional procedure. However, since August 2012, VerifyNow no longer reports ‘percent inhibition’ and only the PRU value is provided. Moreover, previous reports did not assess aspirin reactivity, which may also influence the incidence of hemorrhagic complications.

In the cardiology literature, various studies have reported an optimal cut-off value for PCI under dual antiplatelet therapy (eg, PRU ≤1786 or 1898). Factors such as a larger sheath often require multiple access sites and smaller vessel diameters, which may discriminate the hemorrhagic risk of patients undergoing neurointerventional procedures from those undergoing PCI. However, our cut-off value for LPR (PRU ≤175) was similar to that mentioned in these reports on coronary intervention. Since most of the hemorrhages occurred at the puncture site, our results suggest that the hemorrhagic risk based on the variability of platelet reactivity may be similar under different conditions, such as both PCI and neurointervention.

We also evaluated patient response to aspirin using the VerifyNow assay and found that the ARU value was distributed across a narrower range than the PRU value, but with no apparent association with hemorrhagic events. No reports have suggested a correlation between ARU and hemorrhagic complications in the presence of either PCI or neurointervention. Aspirin may lack a sufficient range of antiplatelet response types to cause patients in that category to be at a higher risk of bleeding events.

In both PRU groups, most of the bleeding complications identified were hematomas at the puncture site. The major known risk factors for puncture site vascular complications are advanced age, arteriosclerosis obliterans, female gender, emergency procedure, and low body surface area.15–20 A percutaneous vascular closure device has also been reported to cause a risk of hematoma and pseudoaneurysm.21 ,22 We evaluated all of these factors, except for arteriosclerosis obliterans, and found no apparent association with the observed major bleeding events. Practitioners at our institution always inject a contrast agent into the entry site artery immediately after introduction of the sheath to assess the presence and level of stenosis and calcification. If arteriosclerosis is observed, we avoid using a percutaneous closure device in order to prevent an entry site complication, particularly ischemic complications.

In our analysis, procedure-related factors such as maximum sheath size, use of a percutaneous closure device, maximum ACT during the procedure, or clopidogrel regimen (loading dose or maintenance dose) were not associated with hemorrhagic complications at the puncture site. Although the mean maximum ACT of 341 s in our study was relatively higher than previously reported, it was not significantly associated with the primary outcome in univariate estimates (p=0.20). Because it cannot be excluded with the type 1 error, high ACT may have influenced the relatively high rate of hemorrhagic complications in this case series.

With regard to the 10 intracranial hemorrhages, five (50%) were due to intraprocedural perforation and three (30%) were due to hyperperfusion syndrome. The remaining two (20%) were subarachnoid hemorrhages detected on postoperative day 1, with no known cause for the bleeding. In general, post-procedural intracranial hemorrhages after neurointerventions are caused by intraprocedural wire perforation, catheter perforation, traction injury of the arteries, aneurysm rupture by coils, hyperperfusion syndrome, or hemorrhagic infarction. In our analysis, intracranial hemorrhage was less correlated with the PRU value than puncture site hemorrhages. This result may suggest that many post-procedural intracranial hemorrhages within 1 week of surgery are the result of these procedures or anatomical conditions rather than platelet reactivity.

Based on the results of the present study, it is plausible that patients with P2Y12 receptor overinhibition may be at a high risk of bleeding events, particularly those with a larger body mass index. Because many of the hemorrhagic complications were at the puncture site, cautious observation of the puncture site will be required to detect potential complications in these patients. Future prospective research is required to determine whether P2Y12 receptor antagonist management of at-risk patients and maintaining a PRU >175 will lower the risk of hemorrhagic complications from neurointervention.

Study limitations

There are several limitations to the present study. First, this study only assessed the incidence of hemorrhagic complications over 1 week. Second, this was an observational study that was subject to limitations inherent to such an analysis. Third, we did not assess CYP2C19 allele genetic polymorphisms, which have been implicated as contributing to both hyper- and hypo-response to clopidogrel.23


Patient response to clopidogrel, as indicated by the results of the VerifyNow assay, was independently associated with hemorrhagic complications during elective neurointerventions including cerebral aneurysm coil embolization and carotid artery stenting under dual antiplatelet therapy. We determined that a PRU value of ≤175 is the optimal cut-off value to discriminate between patients with and without hemorrhagic complications during neurointervention. Future prospective studies are required to validate whether a specific PRU value of 170–180 is predictive of hemorrhagic complications.



  • Contributors SM, TH, JK conceived the study. SM and TH helped with implementation. IN, TO and RI provided statistical expertise in clinical trial design. All authors contributed to data acquisition, refinement of the study protocol and approved the final manuscript.

  • Funding This work was supported by Grants-in-Aid for Scientific Research (C) KAKENHI (No.22591591).

  • Competing interests JK is an advisory board member for Merck Serono and a consultant for BiogenIdec Japan. He has received payments for lectures from Bayer Schering Pharma, Cosmic Cooperation, and Biogen Idec Japan. These sponsors had no control over the interpretation, writing, or publication of this work.

  • Ethics approval Ethical approval was obtained from the institutional ethics review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Any additional data regarding this submission can be requested from the corresponding author via email.