Several prospective studies have shown that high on-clopidogrel platelet reactivity (HPR) in patients undergoing percutaneous coronary intervention (PCI) is a risk factor for ischemic events. All studies were insufficiently powered to detect differences in stroke between patients with HPR and those without. Therefore, we performed a systematic review and meta-analysis of available publications aimed at determining whether patients undergoing PCI with HPR are also at increased risk of stroke. We searched for prospective studies enrolling patients undergoing PCI and treated with aspirin and clopidogrel that reported on clinical relevance of HPR to adenosine diphosphate. Study end point was the rate of stroke. We also investigated whether there was an interaction on the relative risk of stroke between HPR, clinical presentation, duration of follow-up, or laboratory methods. Fourteen studies including 11,959 patients were deemed eligible. On pooled analysis, the risk of stroke was higher in patients with HPR compared with patients with no HPR (1.2% vs 0.7%, relative risk on fixed effect 1.84, 95% confidence interval 1.21 to 2.80). There was no heterogeneity among the studies (I 2 = 0%, p = 0.5). Clinical presentation (p = 0.39 for interaction), duration of follow-up (p = 0.87 for interaction), and laboratory method for detection of HPR (p = 0.99 for interaction) did not affect the relative increase in the risk of stroke in patients with HPR compared with patients with no HPR. In conclusion, in patients with coronary artery disease undergoing PCI, the presence of HPR to adenosine diphosphate is a risk factor for stroke.
Dual antiplatelet therapy with aspirin and the adenosine diphosphate (ADP) receptor inhibitor clopidogrel is the most widely used treatment to inhibit platelet reactivity and prevent ischemic events after percutaneous coronary intervention (PCI). However, a significant individual variability in platelet reactivity after clopidogrel administration exists related to nongenetic and genetic factors. A large amount of evidence derived from longitudinal prospective studies and meta-analyses have shown that patients undergoing PCI and disclosing a high on-clopidogrel platelet reactivity (HPR) are at increased risk of ischemic events, including cardiovascular death and stent thrombosis. However, the relative risk (RR) of stroke of patients with HPR and those without has not been investigated in depth, because all studies were insufficiently powered for this purpose. Therefore, we performed a systematic review and meta-analysis of available publications aimed at determining whether patients undergoing PCI with HPR are also at increased risk of stroke.
Methods
We carried out a systematic review of the available publications according to the current Meta-analysis Of Observational Studies in Epidemiology guidelines to perform meta-analyses of observational studies. We searched for relevant reports published from January 1, 2002 to September 30, 2013 in the MEDLINE database and the Cochrane Library using the following keywords that were variously combined: “clopidogrel,” “clopidogrel resistance,” “platelet reactivity,” “high on-treatment platelet reactivity,” “outcome,” and “stroke.” No language restriction was used. We also checked the reference lists of reviews and relevant reports.
Inclusion criteria were as follows: (1) prospective longitudinal design, (2) studies that enrolled patients receiving aspirin and clopidogrel, (3) patients divided into patients with HPR or not on the basis of absolute measure of on-treatment platelet reactivity by current accepted ADP-specific platelet function assays (ADP-stimulated light transmission aggregometry [LTA ADP ], VerifyNow P2Y12, [Accumetrics, San Diego, California], flow cytometric assessment of vasodilator-stimulated phosphoprotein phosphorylation [BioCytex, Marseille, France], and multiple electrode aggregometry with ADP stimuli [Dynabyte, Munich, Germany]), (4) availability of numbers of cases in both HPR and no HPR groups, and (5) data on stroke prospectively collected. Two investigators (NT and TP) independently reviewed the titles, abstracts, and studies to determine whether they met the inclusion criteria. Conflicts between reviewers were resolved by consensus.
RRs and 95% confidence intervals were used as the summary statistic. The pooled RR was calculated using both fixed-effects (inverse variance weighted) and random-effects (DerSimonian and Laird) models. Between-study heterogeneity of effects was analyzed using the chi-square and inconsistency across study results quantified by I 2 statistics, with I 2 <25%, 25% ≤ I 2 ≤ 50%, and I 2 >50%, respectively, representing mild, moderate, and severe inconsistency.
Sensitivity analysis was performed by evaluating the influence of removing individual studies on the pooled RR. Sensitivity analysis was also performed with reference to the following study characteristics: clinical presentation, duration of follow-up, and laboratory methods used to measure platelet reactivity. The possibility of publication bias was assessed by Harbord’s test.
Statistical analyses were performed using Stata/SE 11.2 (StataCorp LP, College Station, Texas).
Results
Figure 1 shows the flow chart for the study analysis. Of 1,322 potentially relevant reports initially screened, 14 met the inclusion criteria and were included in the meta-analysis, with a total of 11,959 patients. Table 1 lists the main characteristics of the studies included. Of these, 4 included patients with acute coronary syndrome (ACS), 6 all comers, and 4 patients with stable coronary artery disease or low risk ACS. Mean age of patients was 66 ± 3 years ranging from 61 to 71. All studies enrolled predominantly male patients (prevalence ranging from 59% to 83%). The prevalence of smokers and of patients with diabetes ranged from 11% to 56% and from 19% to 44%, respectively. HPR was measured by LTA ADP in 7 studies, by VerifyNow in 5, and by multiple electrode aggregometry with ADP stimuli in 1 study. In 1 study, HPR was measured by both LTA ADP and VerifyNow. The mean prevalence of HPR was 30 ± 15% ranging from 6% to 67% and was greater in studies using the VerifyNow compared with those using the LTA ADP (42 ± 13% vs 22 ± 10%, p = 0.006).
| Author | Year of Publication | Design | Patient Profile | Patients | Laboratory Method | Selected Cutoff for HPR | HPR (%) | Clopidogrel (LD/MD, mg) | Aspirin (LD/MD, mg) | Platelet Function Test Timing | Duration of DAPT (Mo) | Follow-Up (Mo) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Geisler et al | 2006 | Single center | Symptomatic CAD | 363 | LTA, ADP 20 μmol/L | Max AGG >70% (a priori) | 6 | 600/75; LD before PCI | 0/100; MD daily before enrollment | >6 h after LD | 3 | 3 |
| Bliden et al | 2007 | Single center | Nonemergent PCI | 100 | LTA, ADP 5 μmol/L | AGG >50% (a priori) | 22 | 0/75; MD ≥1 month | 0/≥81; MD ≥7 days before PCI | Before PCI in the cath lab | 6 | 12 |
| Gurbel et al | 2008 | Single center | Nonemergent PCI | 297 | LTA, ADP 5 μmol/L | >46% (ROC curve defined) | 30 | 600/75, 300/75, 0/75; LD on day of PCI | 0/81–325; MD ≥7 days before PCI plus 325 on the day of PCI | At discharge | Not specified | 24 |
| LTA, ADP 20 μmol/L | >59% (ROC curve defined) | 34 | ||||||||||
| Migliorini et al | 2009 | Single center | Stable CAD/ACS LM PCI, DES | 215 | LTA, ADP 10 μmol/L | AGG ≥70% (a priori) | 19 | 600/75, LD at least 12 h before PCI | 325/325; LD ≥12 h before PCI | 12–18 h after LD | 12 | 6 |
| Sibbing et al | 2009 | 2 Participating centers | Stable CAD/ACS DES | 1,608 | MEA | Highest quintile >416 AU | 20 | 600/75, LD before PCI | 500 IV/100 twice daily; LD after angiography and before PCI | >2 h after LD encouraged | Not specified | 1 |
| Geisler et al | 2010 | Single center | Stable CAD/ACS | 1,019 | LTA, ADP 20 μmol/L | Late AGG >42.5 (border of upper tertile) | 32 | 600/75, LD before PCI | 500 IV/100; LD before PCI | >6 h after 600-mg LD, >24 h for those on long-term MD + plus 300-mg LD | 3 | 3 |
| Breet et al | 2010 | Single center | Stable CAD | 1,052 | VerifyNow | ≥236 PRU (ROC curve defined) | 39 | 300/75 (LD ≥24 h), 600/75 (LD ≥4 h), 0/75 (MD for at least 5 days) before PCI | 0/80–100; MD ≥10 days before PCI | Before PCI in the cath lab | 12 | 12 |
| 1,049 | LTA, ADP 5 μmol/L | Max AGG ≥42.9% (ROC curve defined) | 42 | |||||||||
| 606 | Plateletworks | AGG ≥80.5% (ROC curve defined) | 43 | |||||||||
| Campo et al | 2010 | Multicenter RCT substudy | Stable CAD/UA | 468 | VerifyNow | Platelet inhibition <40% (a priori) | 39 | 300/75 (LD ≥6 h), 600/75 (LD ≥2 h), 0/75 (MD for at least 7 days) before PCI | 0/≥80; MD ≥5 days before PCI | >6 h after 300-mg LD; >2 h after 600-mg LD, >7 days of MD | 12 | 12 |
| Parodi et al | 2011 | Single center | ACS | 1,772 | LTA, ADP 10 μmol/L | Max AGG >70% (a priori) | 14 | 600/75 or 600/150–300 in patients with HPR, LD before PCI | 325/325; LD before PCI | 12–18 h after LD | 6 | 24 |
| Park et al | 2011 | Single center | Stable CAD/NSTE-ACS DES, Asian population | 2,849 | VerifyNow | PRU >235 (a priori) | 67 | 600/75, 300/75 (LD ≥12 h), 0/75 (MD >5 days), before PCI | 200/100, 200/200; LD before PCI | 24–48 h after PCI | 12 | 24 |
| Yu et al | 2012 | Single center | Stable CAD/NSTE-ACS STEMI >12 h | 186 | VerifyNow | PRU ≥240 (a priori) | 41 | 300/75; LD >6 h | 300/100; LD >6 h | 12–24 h after PCI | 12 | 12 |
| Saia et al | 2013 | Multicenter | NSTE-ACS | 833 | VerifyNow | PRU ≥208 (a priori) | 39 | 600/75 (LD >2 h), 300/75 (LD >6 h) before PCI | Not reported | 30 Days after the procedure | 12 | 12 |
| Liang et al | 2013 | Single center | ACS, DES, Chinese patients | 1,016 | LTA, ADP 20 μmol/L | Max AGG >60.7% (ROC curve defined) | 28 | 600/75, LD before PCI | 300/100; LD before PCI | 12 h after PCI | 12 | 12 |
| Jin et al | 2013 | 3 Centers | STEMI, DES | 181 | VerifyNow | PRU ≥282 (ROC curve defined) | 30 | 600/75, LD at admission | 300/100, LD after hospital admission | At discharge | 12 | 12 |
In the 11,959 patients included, the rate of stroke was 0.9%. On pooled analysis, the risk of stroke was higher in patients with HPR compared with patients with no HPR (1.2% vs 0.7%, RR on fixed effect 1.84, 95% confidence interval 1.21 to 2.80). There was no heterogeneity among the studies (I 2 = 0%, p = 0.5; Figure 2 ). One study including 100 patients in which no strokes occurred was excluded from the main analyses.
By reestimating the summary RR in the absence of each study, we found that no individual study significantly influenced the summary effect estimate; indeed, even after removing the study with the greatest effect estimate, patients with HPR were still at increased risk of stroke (RR 1.66, 95% confidence interval 1.07 to 2.58).
On further sensitivity analyses, clinical profile (p for interaction = 0.39), duration of follow-up (p for interaction = 0.87), and laboratory method for detection of HPR (p for interaction = 0.99) did not affect the relative increase in the risk of stroke in patients with HPR compared with patients with no HPR ( Figure 3 ). There was no small-study effect according to Harbord’s test (p = 0.49).
