OUP user menu

★ Editor's choice ★

Accelerated platelet inhibition by switching from atorvastatin to a non-CYP3A4-metabolized statin in patients with high platelet reactivity (ACCEL-STATIN) study

Yongwhi Park, Young-Hoon Jeong, Udaya S. Tantry, Jong Hwa Ahn, Tae Jung Kwon, Jeong Rang Park, Seok-Jae Hwang, Eun-Ha Gho, Kevin P. Bliden, Choong Hwan Kwak, Jin-Yong Hwang, Sunjoo Kim, Paul A. Gurbel
DOI: http://dx.doi.org/10.1093/eurheartj/ehs083 2151-2162 First published online: 16 April 2012


Aims CYP3A4-metabolized statins can influence the pharmacodynamic effect of clopidogrel. We sought to assess the impact of switching to a non-CYP3A4-metabolized statin on platelet function among patients receiving clopidogrel and atorvastatin with high on-treatment platelet reactivity (HPR).

Methods and results Percutaneous coronary intervention (PCI)-treated patients (n= 50) with HPR [20 μM adenosine diphosphate (ADP)-induced maximal platelet aggregation (MPA) >50%] were enrolled during chronic administration of atorvastatin (10 mg/day) and clopidogrel (75 mg/day) (≥6 months). They were randomly assigned to a 15-day therapy with either rosuvastatin 10 mg/day (n= 25) or pravastatin 20 mg/day (n= 25). Platelet function was assessed before and after switching by conventional aggregometry and the VerifyNow P2Y12 assay. Genotyping was performed for CYP2C19*2/*3, CYP3A5*3, and ABCB1 C3435T alleles. The primary endpoint was the absolute change in 20 μM ADP-induced MPA. After switching, MPAs after stimuli with 20 and 5 μM ADP were decreased by 6.6% (95% confidence interval: 3.2–10.1%; P < 0.001), and 6.3% (95% confidence interval: 2.5–10.2%; P = 0.002), respectively. Fifty-two P2Y12 reaction units fell (95% confidence interval: 35–70; P < 0.001) and the prevalence of HPR decreased (24%; P < 0.001). Pharmacodynamic effects were similar after rosuvastatin and pravastatin therapy. In addition to smoking status, the combination of calcium channel blocker usage and ABCB1 C3435T genotype significantly affected the change of 20 μM ADP-induced MPA.

Conclusions Among PCI-treated patients with HPR during co-administration of clopidogrel and atorvastatin, switching to a non-CYP3A4-metabolized statin can significantly decrease platelet reactivity and the prevalence of HPR. This switching effect appears similar irrespective of the type of non-CYP3A4-metabolized statin.

  • Platelet
  • Clopidogrel
  • Atorvastatin
  • Non-CYP3A4-metabolized statin
  • High platelet reactivity
See page 2121 for the editorial comment on this article (doi:10.1093/eurheartj/ehs126)


Clopidogrel added to aspirin therapy is efficacious in preventing ischaemic events in patients with high-risk coronary artery disease (CAD) and is recommended in the guidelines.1,2 However, clopidogrel hyporesponsiveness and high on-treatment platelet reactivity (HPR) to adenosine diphosphate (ADP) measured by multiple platelet function assays have been linked to an increased risk of the occurrence of ischaemic event.3 High on-treatment platelet reactivity is now considered a major risk factor in patients undergoing percutaneous coronary intervention (PCI). Variability in the response to clopidogrel is directly related to the level of active metabolite generated.4 In the two-step hepatic oxidation of clopidogrel, cytochrome P450 (CYP) 2C19 is a major contributor to active metabolite generation, followed by CYP3A4/5.5 Single nucleotide polymorphisms (SNPs) of genes encoding proteins related to absorption and hepatic metabolism have been associated with the pharmacokinetic and pharmacodynamic properties of clopidogrel therapy, and clinical outcomes.3,6 In addition, drug–drug interactions influencing the function of CYP isoenzymes have been shown to affect the response to clopidogrel.7

Most of the lipophilic statins, such as atorvastatin, simvastatin, and lovastatin, are predominantly metabolized by CYP3A4.8,9 In addition, these statins can also affect the bioavailability of clopidogrel by inhibiting the P-glycoprotein efflux transporter.10,11 Lau et al.12 first reported a significant influence of atorvastatin on the pharmacodynamics of clopidogrel. However, many subsequent studies could not confirm this interaction.13 These conflicting data may be explained by the early assessment of the interaction after drug initiation, use of different doses or classes of statin, and patient selection. Lau et al.14 recently reported that St John's wort, an inducer of the CYP3A4 gene, enhances the pharmacodynamic response in clopidogrel hyporesponders. Furthermore, a novel in vitro pharmacodynamic assay demonstrated that CYP3A4/5 is involved in both steps the hepatic oxidation of clopidogrel.15 Based on these observations, use of non-CYP3A4-metabolized statins may be an alternative strategy to enhance responsiveness to clopidogrel in patients with HPR during treatment with a CYP3A4-metabolized statin such as atorvastatin.

We hypothesized that the antiplatelet effect of clopidogrel is enhanced by switching to a non-CYP3A4-metabolized statin among patients with HPR during the co-administration of clopidogrel and atorvastatin. We enrolled patients with HPR during chronic treatment with clopidogrel and atorvastatin (≥6 months) after PCI, and then evaluated the influence of switching to a non-CYP3A4-metabolized statin.


Patient selection

Between April 2009 and December 2011, patients were prospectively recruited at the Department of Cardiology of the Gyeongsang National University Hospital (Jinju, Korea) at the time of follow-up angiography. Patients were eligible for enrolment if they were ≥18 years of age, had no angina symptoms, and had been treated with clopidogrel (75 mg/day), atorvastatin (10 mg/day), and aspirin (100 mg/day) for at least 6 months (Figure 1A). Major exclusion criteria were post-angiography repeated PCI, previous coronary stenting for a high-risk complex lesion, active bleeding and bleeding diatheses, warfarin therapy, left ventricular ejection fraction <30%, leucocyte count <3000/mm3, platelet count <100 000/mm3, aspartate aminotransferase or alanine aminotransferase level ≥3 times upper normal, stroke within 3 months, non-cardiac disease with a life expectancy <1 year, and inability to follow the protocol. This study complied with the Declaration of Helsinki; the Institutional Ethics Committee approved the study protocol, and the patients provided written informed consent.

Figure 1

Flow diagram: (A) random group and (B) sham group. PCI, percutaneous coronary intervention; ADP, adenosine diphosphate; PA, platelet aggregation.

Study design

This ACCEL-STATIN (ACCELerated platelet Inhibition by switching from atorvastatin to a non-CYP3A4-metabolized STATIN in patients with high platelet reactivity) study was a prospective, randomized, parallel-group pharmacodynamic study (Figure 1A). Blood samples for platelet function assays were collected just before angiography and 2–6 h after the ingestion of the last dose. We enrolled 50 patients with HPR [20 μM ADP-induced maximal platelet aggregation (PA) >50%].16 Patients were randomly allocated to rosuvastatin 10 mg daily (ROSU group) or pravastatin 20 mg daily (PRA group) based on a computer-generated randomization sequence. Any change in other medications was not permitted during the study period. Aspirin (100 mg/day), clopidogrel (75 mg/day), and the randomized non-CYP3A4-metabolized statin were co-administered in the morning. At 15 ± 3 day follow-up, compliance and adverse events were assessed by the attending physician based on interview, pill counting, a questionnaire, and laboratory evaluation. If patients showed 100% compliance, blood samples for follow-up platelet function measurements were performed 2–6 h after the last ingested dose. Patients who exhibited incomplete compliance to medication were excluded from the data analysis. Fasting blood samples were obtained in the morning before the procedure and at follow-up to determine the lipid profile and high-sensitivity C-reactive protein. Because there may be variability in the responsiveness to clopidogrel irrespective of the allocated treatment, a third reference group without change of atorvastatin (sham group: n = 25) was separately enrolled (Figure 1B), and platelet measurements were performed at baseline and 15 ± 3 day follow-up in this group.

Platelet function assays

Light transmittance aggregometry (LTA) and the VerifyNow P2Y12 assay (Accumetrics, San Diego, CA, USA) were performed within 2 h of blood sampling. The protocol and validation data were described elsewhere.17 Blood samples were drawn into Vacutainer tubes containing 0.5 mL of sodium citrate 3.2% (Becton-Dickinson, San Jose, CA, USA). Platelet-rich plasma (PRP) was obtained after centrifuging blood samples at 120 g for 10 min. The remaining blood was further centrifuged at 1200 g for 10 min to recover platelet-poor plasma (PPP). Platelet-rich plasma was adjusted to a platelet count of 250 000/mm3 by adding PPP if needed. Platelet reactivity was evaluated for 10 min at 37°C using an AggRAM aggregometer (Helena Laboratories Corp., Beaumont, TX, USA) after the addition of 5 and 20 μM ADP. Maximal and final (at 5 min) PAs were determined by laboratory personnel blinded to the study protocol. The absolute change in PA (ΔPA) was calculated as follows: ΔPA = (PA during atorvastatin therapy − PA during a non-CYP3A4-metabolized statin therapy).

The VerifyNow P2Y12 assay is a whole blood, point-of-care turbidimetric assay that measures the responsiveness to P2Y12 antagonists.17 Blood samples were collected in 3.2% citrate Vacuette tubes (Greiner Bio-One Vacuette® North America, Inc., Monroe, NC, USA). The cartridge consists of two channels; one channel contains fibrinogen-coated polystyrene beads, 20 μM ADP, and 22 nM prostaglandin E1; the optical signal of this channel is reported as P2Y12 reaction units (PRU). The second channel contains fibrinogen-coated polystyrene beads, 3.4 μM iso-thrombin receptor-activating peptide [protease-activated receptor (PAR)-1 agonist], and PAR-4-activating peptide. This channel was incorporated to estimate the maximal platelet function independent of P2Y12 receptor blockade (BASE). The instrument reports ‘percent inhibition (PI)’ to indicate the extent of P2Y12 blockade by P2Y12 inhibitors: PI = [(BASE-PRU)/BASE] × 100 (%). The absolute changes in PRU (ΔPRU) and PI (ΔPI) were defined as ΔPRU = (PRU during atorvastatin therapy − PRU during a non-CYP3A4-metabolized statin therapy) and ΔPI = (PI during atorvastatin therapy – PI during a non-CYP3A4-metabolized statin therapy).


Genotyping was performed for CYP2C19*2 (rs4244285, c.681G>A) and *3 (rs4986893, c.636G>A), CYP3A5*3 (rs776746, g.6986A>G), and ABCB1 C3435T alleles using a commercially available kit (QIAamp DNA Blood Mini Kit, Qiagen, Hilden, Germany) after extracting genomic DNA from whole blood leucocytes.18 Genotypes were identified using the TaqMan method and a commercially available detection system (ABI PRISM 7900HT Sequence Detection System, Applied Biosystems). The polymerase chain reaction (PCR) amplification protocol for the TaqMan assays included denaturation at 95°C for 10 min, followed by 40 cycles at 92°C for 15 s, 60°C for 1 min, and 72°C for 45 s, followed by elongation at 72°C for 5 min. The TaqMan assay products were then read on a 7900HT Fast Real-Time PCR system. Patients were classified based on the presence of SNP's of the CYP2C19 and ABCB1 C3435T genes, whereas for the CYP3A5 gene, patients were categorized as expressers (*1/*1 or 1/*3) or non-expressers (*3/*3).18


The primary endpoint was Δ maximal PA (20 μM ADP). Secondary endpoints were Δ maximal PA (5 μM ADP), Δ final PA (5 and 20 μM ADP), ΔPRU, ΔPI, the decrease in the prevalence of HPR, and determinants of platelet reactivity during the co-administration of atorvastatin and clopidogrel. We compared these endpoints based on rosuvastatin and pravastatin therapies. We also evaluated whether the observed pharmacodynamic effects were influenced by genotypes and demographic variables.

Sample size calculation and statistical analysis

To observe a Δ maximal PA (20 μM ADP) by switching to a non-CYP3A4-metabolized statin of 10%, at least 20 patients per group would be required to provide 80% power to detect a statistical difference with a two-sided α-value of 0.05 and standard deviation of 15%. Considering an ∼25% dropout rate, 25 patients were needed per group.

Continuous variables were expressed as means ± SD or median and inter-quartile (IQR) range, and categorical variables were expressed as percentages. Student's unpaired t or the Mann–Whitney U test, and the ANOVA test were used to compare continuous variables, and comparisons between categorical variables were performed using χ2 statistics or Fisher's exact test. If a variable showed a significant difference between groups, adjustment was performed. Comparisons of platelet function before and after switching were performed with Student's paired t-test or Wilcoxon signed rank test, as appropriate. A multivariate linear regression model was employed to test for an independent linkage of demographic or genetic variables to baseline platelet reactivity or Δ maximal PA (20 μM ADP). Independent variables were chosen from genotypes and known variables with reported significant influence.19,20 A P-value <0.05 was considered statistically significant, and statistical analyses were performed using SPSS 18.0 software (SPSS, Inc., Chicago, IL, USA).


Baseline characteristics and laboratory measurements

All patients were 100% compliant with the study protocol. The average age was 61 years, and most of the patients were men (Table 1). About 70% of the patients previously presented with myocardial infarction, and the majority of patients were treated with drug-eluting stents; standard cardiovascular medications were used frequently and no patient was treated with proton pump inhibitors.

View this table:
Table 1

Baseline characteristics according to the study group

VariablesTotal (n = 50)Rosuvastatin (n = 25)Pravastatin (n = 25)
Duration of antiplatelet therapy, months10 ± 110 ± 110 ± 1
Age, years61 ± 1060 ± 1062 ± 10
Male, n (%)34 (68)15 (60)19 (76)
Body mass index, kg/m225.6 ± 3.725.6 ± 3.025.6 ± 4.3
Medical history, n (%)
 Previous myocardial infarction34 (68)17 (68)17 (68)
 Previous CVA0 (0)0 (0)0 (0)
 Diabetes mellitus9 (18)4 (16)5 (20)
 Hypertension24 (48)15 (60)9 (36)
 Hypercholesterolaemia50 (100)25 (100)25 (100)
 Current smoking16 (32)8 (32)8 (32)
 Chronic kidney disease3 (6)2 (8)1 (4)
Concomitant medications, n (%)
 Atorvastatin50 (100)25 (100)25 (100)
 Calcium channel blocker9 (18)5 (20)4 (16)
 Proton pump inhibitor0 (0)0 (0)0 (0)
 Beta-blocker42 (84)21 (84)21 (84)
 Angiotensin antagonist48 (96)24 (96)24 (96)
 Nitrate22 (44)11 (44)11 (44)
Laboratory data
 LV ejection fraction, %61 ± 762 ± 661 ± 8
 Haemoglobin, g/dL13.2 ± 1.413.0 ± 1.413.3 ± 1.5
 Platelet count, ×103/mm3242 ± 55243 ± 57242 ± 54
 HbA1C, %6.2 ± 0.86.1 ± 0.76.3 ± 0.8
 Creatinine, mg/dL0.9 ± 0.30.9 ± 0.30.9 ± 0.3
 Total cholesterol, mg/dL148 ± 33157 ± 32141 ± 33
 Triglyceride, mg/dL153 ± 70149 ± 62155 ± 71
 HDL cholesterol, mg/dL40 ± 1242 ± 1139 ± 14
 LDL cholesterol, mg/dL89 ± 2795 ± 2783 ± 27
 High sensitivity C-reactive protein, mg/L0.9 ± 0.81.0 ± 1.20.8 ± 0.5
 Drug-eluting stent, n (%)46 (92)23 (92)23 (92)
 Bare-metal stent, n (%)1 (2)1 (4)0 (0)
 Ballooning only, n (%)3 (6)1 (4)2 (8)
 Stents per patient1.9 ± 1.32.2 ± 1.51.6 ± 1.0
  • Values are expressed as mean ± SD unless otherwise indicated.

  • CVA, cerebrovascular accident; LV, left ventricular; HbA1C, haemoglobin A1C; HDL, high density lipoprotein; LDL, low density lipoprotein.

The lipid profile and C-reactive protein levels did not change significantly after switching (data not shown). In the randomized group, the median baseline values of 20 and 5 µM ADP-induced maximal PA were 67.0% (IQR: 60.9–73.3%) and 52.7% (IQR: 44.5–58.9%), respectively; and those of 20 and 5 µM ADP-induced final PA were 60.0% (IQR: 50.0–70.0%) and 42.5% (IQR: 29.8–50.3%), respectively (Figure 2A and B, left panel). The median PRU was 297 (IQR: 243–329), and the median PI was 18.0% (IQR: 9.8–25.3%) (Figure 2C and D).

Figure 2

(A and B) Maximal and final platelet aggregations and (C and D) P2Y12 reaction units and percent inhibition during atorvastatin therapy vs. after switching to a non-CYP3A4-metabolized statin. The sham group represents patients without change of atorvastatin. The central box represents the values between the lower and upper quartiles, and the middle line is the median. The vertical line extends from the minimum to the maximum value, excluding outside values, which are displayed as separate points. ADP, adenosine diphosphate; CYP3A4-MET, cytochrome P450 3A4-metabolized.

Platelet function measurements in the sham group

Platelet aggregation was the same at baseline and follow-up (Figure 2A and B, right panel). Δ maximal PA after stimulation with 20 and 5 μM ADP was −1.5% [61.9 ± 6.8% vs. 63.2 ± 8.3%; 95% confidence interval (CI): −4.2 to 1.8%; P = 0.412] and −0.2% (49.1 ± 10.0% vs. 49.3 ± 9.4%; 95% CI: −4.4 to 4.1%; P = 0.943), respectively. Δ final PA also showed the same trend. The results of the VerifyNow P2Y12 assay did not differ in terms of PRU (259 ± 36 vs. 246 ± 53; P = 0.146) or PI (11.6 ± 9.3% vs. 14.1 ± 10.7%; P = 0.226).

Pharmacodynamic effect of statin switching

Primary endpoint

Δ maximal PA (20 μM ADP) was 6.6% (95% CI: 3.2–10.1%; P < 0.001); 20 μM ADP-induced maximal PAs before and after switching were 66.9 ± 8.4 and 60.3 ± 15.0%, respectively (Figure 2A, left panel).

Secondary endpoints

Maximal PA (5 μM ADP) was reduced by switching (52.7 ± 10.7 to 46.4 ± 16.7%; ΔPA: 6.3%; 95% CI: 2.5–10.2%; P = 0.002) (Figure 2A, left panel). The switching to a non-CYP3A4-metabolized statin significantly decreased 20 and 5 μM ADP-induced final PA (ΔPA: 8.2%; 95% CI: 3.6–12.7%; P = 0.001 and ΔPA: 6.7%; 95% CI: 2.5–10.9%; P = 0.002, respectively) (Figure 2B, left panel). A significant decrease in PRU (291 ± 64 to 239 ± 76; ΔPRU: 52; 95% CI: 35–70; P < 0.001) and an increase in PI (18.3 ± 14.8% to 28.5 ± 18.3%; ΔPI: −10.1%; 95% CI: −15.6 to −4.6%; P = 0.001) were also observed after switching (Figure 2C and D). The prevalence of HPR was reduced by 24% (P < 0.001).

Switching effect according to non-CYP3A4-metabolized statin type

Baseline characteristics did not differ between the ROSU and PRA groups (Table 1). Platelet function and the prevalence of HPR decreased significantly and to a similar extent during rosuvastatin or pravastatin treatment (Table 2).

View this table:
Table 2

Platelet reactivity and prevalence of high on-treatment platelet reactivity during atorvastatin therapy vs. after switching to a non-CYP3A4-metabolized statin

VariablesRosuvastatin (n = 25)Pravastatin (n = 25)P-value
Maximal PA, %
 20 μM ADP
  Atorvastatin68.8 ± 8.265.0 ± 8.30.102
  Post-switching61.5 ± 15.159.1 ± 15.10.568
  Δ mean (95% CI)7.3 (2.7–11.9)5.9 (0.4–11.4)0.687
 5 μM ADP
  Atorvastatin55.2 ± 9.050.2 ± 11.90.102
  Post-switching49.9 ± 17.342.8 ± 15.60.131
  Δ mean (95% CI)5.3 (0.1–10.6)7.4 (1.8–13.1)0.574
Final PA, %
 20 μM ADP
  Atorvastatin61.8 ± 12.856.1 ± 13.60.133
  Post-switching52.4 ± 20.649.2 ± 20.60.581
  Δ mean (95% CI)9.4 (3.2–15.6)6.9 (−0.2 to 14.0)0.590
 5 μM ADP
  Atorvastatin45.5 ± 13.138.2 ± 16.00.085
  Post-switching39.4 ± 20.430.9 ± 18.20.126
  Δ mean (95% CI)6.1 (0.4–11.8)7.3 (0.9–13.8)0.767
VerifyNow P2Y12 assay
 P2Y12 reaction units
  Atorvastatin305 ± 62277 ± 640.131
  Post-switching242 ± 64236 ± 880.765
  Δ mean (95% CI)63 (39–86)42 (15–69)0.232
 Percent inhibition
  Atorvastatin14.2 ± 9.922.5 ± 17.60.046
  Post-switching26.7 ± 17.230.2 ± 19.50.496
  Δ mean (95% CI)−12.5 (−18.4 to −6.6)−7.8 (−17.5 to 1.9)0.394
HPR, n (%)
 20 μM ADP-maximal PA >50%
 Atorvastatin25 (100)25 (100)1.000
 Post-switching20 (80)18 (72)0.508
  • Values are expressed as mean ± SD unless otherwise indicated.

  • HPR, high platelet reactivity; CYP, cytochrome P450; PA, platelet aggregation; ADP, adenosine diphosphate; CI, confidence interval.

Predictors of platelet reactivity before switching

Genotype was determined in 45 patients. The results followed Hardy–Weinberg equilibrium (Appendix 1). The prevalence of the CYP2C19 loss-of-function (LoF) allele carriage was high (66.7%). Baseline characteristics did not differ between the genotype groups, except for the lower prevalence of current smoking in the CYP2C19 wild types, higher creatinine levels in the CYP3A5 non-expressers, and lower platelet count in the ABCB1 C3435T high expression (CC type) group (Appendix 2). However, these differences did not influence the results of platelet function testing after adjustment (data not shown). While the CYP3A5 genotype was not associated with platelet reactivity (Appendix 3), carriers with the CYP2C19 LoF allele and ABCB1 C3435T CC type tended to have greater platelet reactivity.

In a multivariate linear regression model (Appendix 4), only gender (male) was independently associated with the pre-switching level of 20 µM ADP-induced maximal PA (β coefficient: −7.1; SE: 2.5; P = 0.008). The CYP2C19 LoF allele (β coefficient: 5.3; SE: 2.6; P = 0.055) and calcium channel blocker (CCB) therapy (β coefficient: 6.3; SE: 3.3; P = 0.064) tended to have higher 20 µM ADP-induced maximal PA. There was no pharmacodynamic difference according to the CCB type (amlodipine, n = 4; lacidipine, n = 1; diltiazem, n = 4) (data not shown).

Predictors of platelet reactivity change after switching

In a multivariate analysis, current smoking was independently associated with Δ maximal PA (20 μM ADP) (β coefficient: 10.4; SE: 4.9; P = 0.043) (Table 3). The carriers of the ABCB1 C3435T T allele who were not treated with a CCB exhibited a significant switching effect by Δ maximal PA (20 μM ADP) (β coefficient: 9.1; SE: 3.8; P = 0.021). One patient with the CYP2C19*1/*1 genotype who smoked showed a marked switching effect in this model (β coefficient: 45.5; SE: 15.1; P = 0.005).

View this table:
Table 3

Determinants of the change of 20 µM adenosine diphosphate-induced maximal platelet aggregation after switching to a non-CYP3A4-metabolized statin (n = 45)

Variablesβ CoefficientP-value
Type of statin: rosuvastatin vs. pravastatin2.14.30.631
Age ≥65 years−
Body mass index (increment per 1 kg/m2)−
CYP2C19 LoF (*2 or *3) allele carriage−
CYP3A5*3/*3 genotype−
ABCB1 C3435T T allele carriage7.44.60.115
History of myocardial infarction1.84.80.713
Current smoking10.44.90.043
Diabetes mellitus−
Chronic kidney disease6.49.10.486
Calcium channel blocker−
Platelet count (increment per 1000/mm3)−
  • CYP, cytochrome P450; SE, standard error; LoF, loss of function; ABCB1, P-glycoprotein gene.


The important findings of this prospective investigation are as follows: (i) in clopidogrel-treated patients with HPR during chronic co-administration of atorvastatin, switching to a non-CYP3A4-metabolized statin resulted in a significant decrease in platelet reactivity and the prevalence of HPR; (ii) the switching effect was not influenced by the type of non-CYP3A4-metabolized statin. The magnitude of the switching effect may be influenced by genetic polymorphisms, smoking, and drug–drug interactions. However, a larger number of patients must be studied to confirm this observation.

The absolute or relative change in platelet reactivity has been used to evaluate the pharmacodynamic effect of an antiplatelet agent.3 Although the relation of non-responsiveness to clopidogrel to the occurrence of a post-PCI ischaemic event was initially reported, multiple translational studies have utilized HPR as a predictor of the occurrence of an ischaemic event instead of non-responsiveness to clopidogrel.

To the best of our knowledge, the ACCEL-STATIN study is the first to support the beneficial effect of replacing atorvastatin therapy with a non-CYP3A-metabolized statin with regard to reducing platelet reactivity in clopidogrel-treated patients with HPR.13 We tried to reduce potential confounders present in previous clopidogrel–atorvastatin interaction studies by enrolling stable patients treated with a commonly prescribed regimen of clopidogrel (75 mg/day) and atorvastatin (10 mg/day) for ≥6 months. In the current study, enhanced platelet inhibition was more likely related to CYP3A4-related metabolism, since a similar improvement in clopidogrel-induced platelet inhibition was observed irrespective of the type of non-CYP3A4-metabolized statin. Moreover, an enhanced pharmcodynamic effect was demonstrated by both LTA and the VerifyNow P2Y12 assay.

Since CYP2C19 is a major enzyme involved in clopidogrel metabolism,21 the US Food and Drug Administration noted that alternative antiplatelet regimens or clopidogrel dosing strategies are considered in genetically predicted poor metabolizers.22 Several alternative strategies have been proposed to achieve enhanced platelet inhibition in carriers of the CYP2C19 LoF allele.21 However, uniform implementation of new P2Y12 receptor blockers may be associated with higher bleeding risk and increased cost.23 In the genetic substudy of the GRAVITAS (Gauging Responsiveness with A VerifyNow assay-Impact on Thrombosis And Safety) trial, it was demonstrated that 150 mg/day clopidogrel was not a highly effective strategy to overcome HPR in patients carrying the CYP2C19 LoF allele.19 The addition of CYP3A4 inducers such as rifampin and St John's wort was associated with enhanced responsiveness to clopidogrel, particularly in clopidogrel hyporesponders.14,24 Likewise, the present study suggested that switching from a CYP3A4-metabolized statin to a non-CYP3A4-metabolized statin enhanced the antiplatelet effect of clopidogrel and overcame HPR observed during atorvastatin therapy in some patients. Furthermore, switching to a non-CYP3A4-metabolized statin reduced platelet reactivity (∼50 PRU) to the same extent as reported by switching to double-dose clopidogrel (∼40 PRU).19

Recent studies have indicated that in addition to CYP2C19, CYP3A4 has an important role in biotransformation of clopidogrel.57,13,15 In addition, CYP3A4 is the most abundant hepatic CYP (∼30%) and is also present in the intestine. CYP 3A4 metabolizes >50% of the clinically used drugs.25 Because the CYP3A4 gene does not have common polymorphisms that alter its function, competition between the drugs that are metabolized by CYP3A4 or also inhibition of CYP3A4 activity by its substrates may be responsible for reduced CYP3A4 activity and biotransformation of clopidogrel. In addition, SNP's in the gene encoding the P-glycoprotein multi-drug resistant-1 efflux transporter, ABCB1, that is responsible for regulating clopidogrel absorption may affect the antiplatelet effect of clopidogrel.2628 The contribution of SNP's or drug–drug interactions on clopidogrel metabolism may be influenced by the clinical presentation, patient characteristics, or the time when the response to clopidogrel was measured. The present study suggested that demographic variables, SNP's in the ABCB1 C3435T and CYP2C19 genes, and drug–drug interactions together influenced the switching effect.

The bioavailability of lipophilic statins is largely dependent on CYP3A-mediated first pass metabolism in the intestine and liver.10,11,29 Lipophilic statins, such as atorvastatin, are inhibitors of CYP3A4 and ABCB1 in the intestine and liver. Hydrophilic statins such as rosuvastatin and pravastatin have little interactions with CYP3A4 and drug transporters.10,29 Interestingly, in the present study, the effect of switching to a non-CYP3A4-metabolized statin was prominent in carriers of the ABCB1 C3435T T allele who were not treated with CCB, a CYP3A4-metabolized drug. However, no significant influence on the response to clopidogrel was observed when these factors were evaluated separately in the statistical analysis. Atorvastatin may influence functional variability at the ABCB1 and CYP3A4 isoenzymes together. The current finding may indicate that the switching effect is the cumulative result following the elimination of the influence of atorvastatin on intestinal absorption and hepatic metabolism.3 Therefore, future studies should consider combining the influence of various factors that are associated with clopidogrel metabolism. As mentioned in the consensus document,3 while relating HPR to the occurrence of clinical events in patients receiving clopidogrel, demographic variables, SNP's, and drug–drug interactions should be considered as a combined variable influencing the development of HPR.

The reduced clopidogrel response among PCI-treated patients co-administered with atorvastatin in the current study has not been replicated in clinical trials. In the CREDO (Clopidogrel for the Reduction of Events During Observation) trial,30 although the 1-year composite of ischaemic events in the clopidogrel-treated patients was numerically lower in patients co-administered with statin compared with no statin (7.4 vs. 10.1%), a non-significant interaction was observed between patients treated with a CYP3A4- vs. a non-CYP3A4-metabolized statin (7.6 vs. 5.4%). Similar findings were observed in a post hoc analysis of the CHARISMA (Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance) trial.31 The observed clinical composite at 28 months was 8.7% in the clopidogrel-treated group without statin therapy, 5.7% in the atorvastatin group, and 5.1% in the pravastatin group. Although statins have numerous beneficial pleiotropic effects including lipid lowering,32,33 clopidogrel and CYP3A4-metabolized statin co-administration may decrease the antiplatelet effect of clopidogrel, with important clinical implications in selected patients with borderline platelet inhibition.13 However, a definitive clinical statin-clopidogrel interaction has yet to be demonstrated in a large-scale randomized clinical trial.

In the present study, platelet reactivity decreased proportionally with the number of ABCB1 C3435T T alleles (gene-dose effect) (20 µM ADP-induced maximal PA: CC type, 69.7 ± 8.9% vs. CT type, 66.6 ± 7.6% vs. TT type, 60.0 ± 6.3%; P = 0.026), although it remained non-significant in the multivariate linear regression model. This finding was in line with a pharmacodynamic study of Rideg et al.,27 and the observations from the PLATO (PLATelet inhibition and patient Outcomes) genetic substudy.34 In the latter study, carriers of the ABCB1 3435 CC type had a numerically higher rate of ischaemic events compared with those of CT or TT type. However, contrasting data also existed in several other pharmacodynamic and clinical studies26,27 that reported higher platelet reactivity and occurrence of ischaemic events in TT carriers. Although high expression of the ABCB1 C3435T (CC type) has been considered to increase drug efflux from enterocytes into the intestinal lumen, P-glycoprotein activity is influenced by multiple factors including environmental and genetic factors.35 The influence of atorvastatin on ABCB1 activity and the enrolment of a pure East Asian population also might have affected the results of the present study.

Results from this study should be interpreted with consideration of the potential limitations. Firstly, this was not a crossover study, but the data of the sham group reinforced the pharmacodynamic effect achieved by switching. Secondly, the enrolment guidelines in the present study differed from those of the consensus-defined HPR.3 When the study started, no definite criteria for HPR were available. However, if we adapted the criteria of consensus-defined HPR, the impact of switching on HPR prevalence was consistent. Finally, in addition to genetic polymorphisms and drug–drug interactions, the functional variability in CYP3A4 activity has been associated with age, race, disease state, organ function, and dietary factors.25 In addition, this study enrolled a prespecified population; results may differ in a general population.


Among PCI-treated patients with high platelet reactivity during co-administration of clopidogrel and atorvastatin, switching to a non-CYP3A4-metabolized statin enhanced the response to clopidogrel and reduced the prevalence of HPR. This switching effect was similar irrespective of the type of the non-CYP3A4-metabolized statin administered.


This study was partly supported by grants from Institute of the Health Sciences, Gyeongsang National University and the Sinai Hospital of Baltimore.

Conflict of interest: Y.H.J. has received honoraria for lectures from Sanofi-Aventis, Daiichi Sankyo, Inc., and Otsuka. P.A.G. has received research grants, honoraria, and consultant fees from AstraZeneca, Merck, Medtronic, Lilly/Daiichi Sankyo, Inc., Sanofi Aventis/Bristol Myers, Portola/Novartis, Boston-Scientific, Bayer, Accumetrics, Boehringer Ingelheim, and Johnson and Johnson. The other authors report no conflicts of interest.

Appendix 1

View this table:

Distributions of genotype (n = 45)

GeneGenotypesPredicted functionDistribution, n (%)
CYP2C19*1/*1Normal15 (33.3)
*1/*2Decreased14 (31.1)
*1/*3Decreased6 (13.4)
*2/*2Decreased or absent4 (8.9)
*2/*3Decreased or absent6 (13.3)
*3/*3Decreased or absent0 (0)
CYP3A5*1/*1Normal7 (15.6)
*1/*3Normal11 (24.4)
*3/*3Decreased or absent27 (60.0)
ABCB1 C3435TCCHigh expression16 (35.5)
CTIntermediate expression21 (46.7)
TTLow expression8 (17.8)
  • CYP, cytochrome P450; ABCB1, P-glycoprotein gene.

Appendix 2

View this table:

Baseline characteristics according to genotype (n = 45)

VariablesCYP2C19CYP3A5ABCB1 C3435T
*1/*1 (n = 15)LoF carriage (n = 30)*1/*1 + *1/*3 (n = 18)*3/*3 (n = 27)CC (n = 16)CT + TT (n = 29)
Duration of antiplatelet therapy, months10 ± 110 ± 110 ± 110 ± 110 ± 110 ± 1
Age, years60 ± 1362 ± 961 ± 1061 ± 1161 ± 861 ± 11
Male, n (%)10 (66.7)20 (66.7)11 (64.7)19 (67.9)9 (56.3)21 (72.4)
Body mass index, kg/m225.6 ± 4.425.8 ± 3.526.3 ± 3.925.4 ± 3.725.3 ± 4.026.0 ± 3.7
Medical history, n (%)
 Previous myocardial infarction12 (80.0)19 (63.3)12 (70.6)19 (67.9)10 (62.5)21 (72.4)
 Previous CVA0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
 Diabetes mellitus2 (13.3)4 (13.3)2 (11.8)4 (14.3)2 (12.5)4 (13.8)
 Hypertension6 (40.0)15 (50.0)8 (47.1)13 (46.4)7 (43.8)14 (48.3)
 Hypercholesterolaemia15 (100)30 (100)18 (100)27 (100)18 (100)27 (100)
 Current smoking1 (6.7)*13 (43.3)5 (29.4)9 (32.1)3 (18.8)11 (37.9)
 Chronic kidney disease1 (6.7)2 (6.7)0 (0)3 (10.7)1 (6.3)2 (6.9)
Concomitant medications, n (%)
 Atorvastatin15 (100)30 (100)18 (100)27 (100)16 (100)29 (100)
 Calcium channel blocker1 (6.4)6 (20.0)2 (11.8)5 (17.9)2 (12.5)5 (17.2)
 Proton pump inhibitor0 (0)0 (0)0 (0)0 (0)0 (0)0 (0)
 Beta-blocker14 (93.3)24 (80.0)16 (94.1)22 (78.6)14 (87.5)24 (82.8)
 Angiotensin antagonist15 (100)29 (96.7)16 (94.1)28 (100)15 (93.8)29 (100)
 Nitrate6 (40.0)13 (43.4)6 (35.3)13 (46.4)6 (37.5)13 (44.8)
Laboratory data
 LV ejection fraction, %61 ± 761 ± 761 ± 662 ± 761 ± 661 ± 7
 Haemoglobin, g/dL12.9 ± 1.613.3 ± 1.412.9 ± 1.713.3 ± 1.312.8 ± 1.613.4 ± 1.4
 Platelet count, ×103/mm3236 ± 65240 ± 52244 ± 71235 ± 45216 ± 30*251 ± 63
 HbA1C, %6.0 ± 0.76.2 ± 0.86.1 ± 0.46.1 ± 0.95.8 ± 0.36.3 ± 0.8
 Creatinine, mg/dL0.9 ± 0.40.9 ± 0.30.8 ± 0.2*1.0 ± 0.30.9 ± 0.40.9 ± 0.3
 Total cholesterol, mg/dL145 ± 34151 ± 33154 ± 34146 ± 33143 ± 33152 ± 33
 Triglyceride, mg/dL144 ± 65139 ± 66128 ± 46148 ± 73141 ± 57141 ± 70
 HDL cholesterol, mg/dL46 ± 1439 ± 1042 ± 1341 ± 1140 ± 1241 ± 12
 LDL cholesterol, mg/dL83 ± 2890 ± 2795 ± 3183 ± 2586 ± 3089 ± 27
 High sensitivity C-reactive protein, mg/L0.6 ± 1.10.9 ± 0.81.0 ± 1.20.5 ± 0.50.7 ± 1.10.8 ± 0.9
 Drug-eluting stent, n (%)15 (100)27 (90.0)17 (100)25 (89.3)15 (93.8)27 (93.1)
 Bare-metal stent, n (%)0 (0)1 (3.3)0 (0)1 (3.6)0 (0)1 (3.4)
 Ballooning only, n (%)0 (0)2 (6.7)0 (0)2 (7.1)1 (6.3)1 (3.4)
 Stents per patient1.5 ± 0.72.2 ± 1.52.2 ± 1.31.9 ± 1.41.8 ± 1.42.1 ± 1.4
  • Values are expressed as mean ± SD unless otherwise indicated.

  • CYP, cytochrome P450; ABCB1, P-glycoprotein gene; LoF, loss-of-function; CVA, cerebrovascular accident; LV, left ventricular; HbA1C, haemoglobin A1C; HDL, high density lipoprotein; LDL, low density lipoprotein.

  • *P < 0.05 between the metabolic phenotype groups.

Appendix 3

View this table:

Influence of switching on platelet reactivity and prevalence of high on-treatment platelet reactivity according to genotype (n = 45)

VariablesCYP2C19CYP3A5ABCB1 C3435T
*1/*1 (n = 15)LoF carriage (n = 30)P-value*1/*1 + *1/*3 (n = 18)*3/*3 (n = 27)P-valueCC (n = 16)CT + TT (n = 29)P-value
Maximal PA, %
 20 μM ADP
  Atorvastatin63.2 ± 7.068.2 ± 8.70.05766.0 ± 9.166.9 ± 8.10.71769.7 ± 8.964.8 ± 7.80.060
  Post-switching54.4 ± 14.562.3 ± 15.50.10557.5 ± 18.461.1 ± 13.40.44367.0 ± 13.855.6 ± 15.00.016
  Δ mean (95% CI)8.8 (0.8–16.8)5.9 (1.5–10.4)0.4818.5 (1.0–16.0)5.8 (1.4–10.2)0.4902.7 (−2.2–7.6)9.2 (3.9–14.5)0.102
 5 μM ADP
  Atorvastatin48.6 ± 8.454.5 ± 11.00.07553.0 ± 11.252.3 ± 10.30.80955.0 ± 11.351.2 ± 10.10.255
  Post-switching41.0 ± 16.148.8 ± 17.90.16244.8 ± 19.447.1 ± 16.50.67154.8 ± 17.041.4 ± 16.20.013
  Δ mean (95% CI)7.6 (−0.2–15.4)5.7 (0.5–11.0)0.6768.2 (−0.1–16.4)5.1 (0.4–9.9)0.4750.2 (−5.0–5.4)9.8 (4.1–15.5)0.026
Final PA, %
 20 μM ADP
  Atorvastatin52.1 ± 13.761.6 ± 12.80.02757.7 ± 14.159.0 ± 13.60.75963.4 ± 13.055.7 ± 13.50.068
  Post-switching42.1 ± 19.653.6 ± 20.90.08348.1 ± 22.950.9 ± 20.00.66660.8 ± 17.643.6 ± 20.40.007
  Δ mean (95% CI)10.1 (0.0–20.2)8.0 (2.0–14.1)0.7069.6 (0.0–19.2)8.1 (2.1–14.1)0.7722.6 (−4.0–9.2)12.1 (5.2–18.9)0.070
 5 μM ADP
  Atorvastatin35.6 ± 11.344.8 ± 15.20.04442.4 ± 14.841.3 ± 14.70.81545.9 ± 14.939.4 ± 14.20.156
  Post-switching28.3 ± 17.638.4 ± 21.20.12034.2 ± 22.035.6 ± 19.70.82446.7 ± 19.228.7 ± 18.50.003
  Δ mean (95% CI)7.3 (−1.4–15.9)6.4 (0.8–12.0)0.8588.2 (−0.4–16.7)5.7 (0.3–11.1)0.596−0.8 (−6.9–5.4)10.8 (5.0–16.6)0.012
VerifyNow P2Y12 assay
 P2Y12 reaction units
  Atorvastatin284 ± 58296 ± 680.561284 ± 58296 ± 680.930318 ± 61277 ± 580.144
  Post-switching217 ± 69243 ± 780.346226 ± 73238 ± 770.546282 ± 73209 ± 680.001
  Δ mean (95% CI)67 (34–100)53 (31–76)0.46067 (36–98)52 (29–74)0.39136 (0–73)69 (49–90)0.138
 Percent inhibition
  Atorvastatin19.6 ± 11.217.9 ± 16.60.72519.6 ± 11.217.9 ± 16.60.64515.5 ± 19.520.1 ± 11.80.328
  Post-switching34.0 ± 16.727.4 ± 18.30.24330.2 ± 17.329.2 ± 18.60.85620.7 ± 16.834.5 ± 16.80.012
  Δ mean (95% CI)−14.5 (−22.4 to −6.5)−9.5 (−17.7 to −1.3)0.431−10.5 (−23.0 to 2.0)−11.6 (−17.8 to −5.4)0.854−5.2 (−19.4 to 9.0)−14.4 (−19.8 to −9.0)0.137
HPR, n (%)
 20 μM ADP-maximal PA >50%
  Atorvastatin15 (100)30 (100)1.00018 (100)27 (100)1.00016 (100)29 (100)1.000
  Post-switching9 (60.0)24 (80.0)0.17411 (61.1)22 (81.5)0.17514 (87.5)19 (65.5)0.164
  • Values are expressed as mean ± SD unless otherwise indicated.

  • HPR, high platelet reactivity; CYP, cytochrome P450; ABCB1, P-glycoprotein gene; LoF, loss-of-function; PA, platelet aggregation; ADP, adenosine diphosphate; CI, confidence interval.

Appendix 4

View this table:

Determinants of 20 µM adenosine diphosphate-induced maximal platelet aggregation during co-administration of clopidogrel and atorvastatin (n = 45)

Variablesβ CoefficientP-value
Age ≥65 years0.32.70.900
Body mass index (increment per 1 kg/m2)
CYP2C19 LoF (*2 or *3) allele carriage5.32.60.055
CYP3A5*3/*3 genotype−
ABCB1 C3435T T allele carriage−
History of myocardial infarction4.74.80.713
Current smoking0.42.80.101
Diabetes mellitus1.64.20.708
Chronic kidney disease0.19.10.486
Calcium channel blocker6.33.30.064
Platelet count (increment per 1000/mm3)−
  • SE, standard error; CYP, cytochrome P450; LoF, loss-of-function; ABCB1, P-glycoprotein gene.


View Abstract