Introduction

Aspirin (acetylsalicylic acid) is a well-known baseline anti-platelet agent for the treatment and prevention of coronary artery disease (CAD). It irreversibly acetylates a serine residue at position 529 in platelet prostaglandin synthase, and inhibits cyclooxygenase (COX) channel associated with platelet aggregation. However, up to 24% patients were reported to be resistant to aspirin1. This mechanism of resistance and its clinical impact are under investigation.

PlA1/A2 polymorphism, a single nucleotide substitution (T → C) at position 1565 in exon 2 of the GP IIIa (a component of the final platelet aggregation pathway GPIIb/IIIa) gene has been reported to be associated with the laboratory detected aspirin resistance2,3,4,5 and adverse clinical outcomes6,7,8,9. However, the results are inconsistent among different studies10,11,12. In addition, study results regarding whether aspirin resistance is associated with adverse cardiovascular events are also inconsistent13,14,15,16,17.

This meta-analysis aimed to include the latest studies and update the concept regarding whether the PlA1/A2 gene polymorphism predicts laboratory aspirin resistance and/or cardiovascular outcomes, and whether laboratory detected aspirin resistance predicts cardiovascular outcomes in CAD patients who are on aspirin treatment. The results of this study will provide evidence for the individualized anti-platelet treatment.

Materials and Methods

Eligibility and search strategy

We performed a comprehensive literature search to identify all the studies investigating the association among PlA1/A2 gene polymorphism, aspirin resistance and cardiovascular outcomes in patients with coronary artery disease treated with aspirin. The literature was scanned by computerized searches of Pubmed, Embase, Cochrane library and Chinese Medical Journal Network databases from establishment to September 2018. The search strategy included a combination of medical subject headings and text words as follows: A = (aspirin OR acetylsalicylic acid), B = (platelet aggregation OR platelet activity OR aspirin resistance OR aspirin non-responder OR aspirin low response), C = (gene OR polymorphism OR mutation OR genotype OR allele OR genetic), D = (death OR stroke OR myocardial infarction OR revascularization). The Medical Subject Headings terms and text words A, B, C were used for the search to investigate the association between PlA1/A2 gene polymorphism and aspirin resistance; The Medical Subject Headings terms and text words A, C, D were used for the search to assess the association between PlA1/A2 gene polymorphism and cardiovascular outcomes; The Medical Subject Headings terms and text words A, B, D were used for the search to investigate the association between aspirin resistance and cardiovascular outcomes. Reference literatures of the appropriate trials were hand searched. The search results were limited to human. No language restriction was enforced.

Study selection

The inclusion criteria of this study include: (1) studies that include patients with confirmed CAD; (2) studies that include patients who were treated with aspirin for secondary prevention of cardiovascular events; (3) studies that contain a clear description of the method used to establish the effects of aspirin on platelet reactivity; (4) studies that contain a clear description of the PlA1/A2 polymorphism; (5) studies that report the incidence of either death, myocardial infarction (MI), ischemic stroke, or target vessel revascularization (TVR). Studies that meet the criteria (1), (2), (3), (4) were adopted to analyze the relationship between PlA1/A2 gene polymorphism and aspirin resistance. Studies that meet criteria (1), (2), (4), (5) were adopted to analyze the relationship between PlA1/A2 gene polymorphism and cardiovascular outcomes. Studies that meet criteria (1), (2), (3), (5) were adopted to analyze the relationship between aspirin resistance and cardiovascular outcomes (Fig. 1). We assessed the included observational studies according to the Newcastle-Ottawa Scale.

Figure 1
figure 1

Flowchart of the study selection.

Full size image

Clinical outcomes

The adverse clinical outcomes involve all-cause death, non-fatal MI, ischemic stroke and TVR. The definition of each event in the original articles was accepted.

Statistical analysis

Statistical analysis was performed using Review Manager 5.3.5 (The Cochrane Collaboration, Oxford, England). The odds ratio (OR) or relative risk (RR), and 95% confidence interval (CI) for categorical variables were calculated using a fixed-effect model with the Mantel-Haenszel method. The DerSimonian and Laird random effect model was applied to the calculated OR and RR in case of significant heterogeneity across studies. Statistical heterogeneity was evaluated using the Q statistic with P < 0.1. Points were evaluated at the longest follow-up available. Statistical significance was considered as P < 0.05.

Results

PlA1/A2 gene polymorphism and aspirin resistance

Sixteen studies18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33 including 3077 CAD patients were selected to assess the association between PlA1/A2 gene polymorphism and aspirin resistance. Details of included studies are summarized in Table 1. Heterogeneity testing showed considerable bias between different studies (P = 0.004), so a random effect model was adopted. An OR of 0.94 (95% CI 0.63 to 1.40, P = 0.74) was observed for aspirin resistance in patients carrying the PlA2 allele (PlA1/A2 + PlA2/A2) (Fig. 2).

Table 1 Characteristics of eligible studies referring to the association between PlA1/A2 gene polymorphism and aspirin resistance.
Full size table
Figure 2
figure 2

Association between PlA1/A2 polymorphism and laboratory aspirin resistance. The position of the blue squares corresponds to the odds ratio (OR) per study and the horizontal black line to the 95% confidence intervals (CI). The size of the square is proportional to the relative weight of that study w(%) to compute the overall OR (black diamond). The width of the diamond represents the 95% CI of the overall OR. If a 95% CI spans one (indicated by the black vertical solid line), this study has found no significant difference in the incidence of aspirin resistance (AR) between patients carrying the PlA1 and PlA2 alleles. This meta-analysis shows no significant change in the incidence of AR of patients carrying the PlA2 allele over those carrying PlA1 allele. The overall OR is 0.94 (P = 0.74). Note that the p-value mentioned here is the p-value for Z test.

Full size image

A subgroup analysis was conducted considering that the methodology used to assess aspirin resistance might influence the association. The data available allowed the comparison of two methods: light transmission aggregometry (LTA) and point-of-care assay PFA-100 (Supplemental Figs 1 and 2).

The LTA subgroup analysis showed significant homogeneity between studies (P = 0.22), but did not reveal a significant association between carriage of the PLA2 allele and aspirin resistance (OR 1.35, 95% CI 0.96, 1.91; P = 0.09). The PFA-100 subgroup analysis revealed a significant association between carriage of PLA2 allele and aspirin sensitivity (OR 0.7, 95% CI 0.52, 0.94; P = 0.02), but showed significant heterogeneity between studies (P = 0.02). This significance was lost when using the random effects model (OR 0.79, 95% CI 0.45, 1.38; P = 0.4).

PlA1/A2 gene polymorphism and adverse cardiovascular outcomes

Eight studies6,7,8,9,10,11,12,34 including 4091 CAD patients were selected to investigate the association between PlA1/A2 gene polymorphism and adverse cardiovascular events. Details of the included studies are summarized in Table 2. Among the recruited studies, data on death was available in eight studies6,7,8,9,10,11,12,34; data on non-fatal MI was available in seven studies6,8,9,10,11,12,34; and data on TVR was available in five studies6,8,10,11,12.

Table 2 Characteristics of eligible studies referring to the association between PlA1/A2 gene polymorphism and adverse cardiovascular events.
Full size table

Meta-analysis showed that PlA2 gene carriers had similar risk of death compared with PlA1/A130. The incidences of MI and TVR in PlA2 gene carriers (PlA1/A2 or PlA2/A2) were also similar to those with the wild genotype (PlA1/A1) with ORs of 1.13 (95% CI 0.79 to 1.61, P = 0.51) and 0.89(95% CI 0.46 to 1.61, P = 0.71) respectively (Fig. 3).

Figure 3
figure 3

Association between PlA1/A2 polymorphism and clinical outcomes. Symbols and conventions are the same as in Fig. 2. This meta-analysis shows no significant change in the incidences of death, MI or TVR of patients carrying the PlA2 allele over those carrying PlA1 allele. The overall ORs of death, MI and TVR are 0.51(P = 0.28), 1.13 (P = 0.51) and 0.89 (P = 0.72) respectively. Note that the p-values mentioned here are the p-values for Z test. CI, confidence interval; MI, myocardial infarction; TVR, target vessel revascularization.

Full size image

Laboratory aspirin resistance and adverse cardiovascular outcomes

Eleven studies13,14,15,16,17,35,36,37,38,39,40 with 11857 aspirin-treated CAD patients were included to analyze the association between laboratory aspirin resistance and adverse cardiovascular outcomes. Details of included studies are summarized in Table 3, where mortality was reported in 10 studies13,14,15,16,17,36,37,38,39,40, ischemic stroke was reported in 5 studies16,17,36,37,38, and revascularization was reported in 4 studies35,37,38,39.

Table 3 Characteristics of eligible studies referring to the association between laboratory aspirin resistance and adverse cardiovascular outcomes.
Full size table

Meta-analysis showed that the laboratory aspirin resistance significantly increased the risk of all-cause death (OR = 2.42, 95% CI 1.86 to 3.15, I2 = 0%, P < 0.00001) and TVR (OR = 2.20, 95% CI 1.19 to 4.08, I2 = 13%, P = 0.01) (Fig. 4).

Figure 4
figure 4

Association between laboratory aspirin resistance and clinical outcomes. Symbols and conventions are the same as in Fig. 2. This meta-analysis shows significant overall increase in the incidences of death and TVR of patients with AR over those without AR, while no significant change in the incidences of MI or stroke. The overall ORs in death, MI, stroke and TVR are 2.42 (P < 0.00001), 0.95 (P = 0.79), 1.64 (P = 0.46) and 2.2 (P = 0.01) respectively. Note that the p-values mentioned here are the p-values for Z test. CI, confidence interval; ASA, aspirin; MI, myocardial infarction; TVR, target vessel revascularization.

Full size image

Discussion

In this systematic review and meta-analysis we have found that: (1) there is no significant association between the PlA1/A2 polymorphism and aspirin resistance, or the PlA1/A2 polymorphism and worse clinical outcomes. (2) Laboratory aspirin resistance predicts all-cause death and TVR.

The PlA1/A2 gene polymorphism generated great interest since Weiss et al.41 first reported that PlA2 gene carriers presented a two-fold increase in risk of acute coronary syndromes. It has been clarified that PlA1/A2 encodes the platelet membrane glycoprotein IIIa which is integrated with glycoprotein IIb to form a complex. People with wild-type PlA1 have leucine at position 33 of mature glycoprotein IIIa, while those with point mutation PlA2 have proline at this position as a consequence of the substitution of cytosine for thymidine at position 1565 in exon 2 of the glycoprotein IIIa gene42. However, whether this mutation affects the anti-platelet effect of aspirin remains controversial.

By this meta-analysis, we found that PlA1/A2 polymorphism does not predict laboratory aspirin resistance. In fact, aspirin reduces the activation of platelets by irreversibly acetylating serine at position 529 of cyclooxygenase-1 (COX-1), and thereby reduces thromboxane A2 (TXA2) formation from the platelets43. So when aspirin effectively blocks the COX-1 channel, platelet aggregation would be effectively inhibited as a consequence of less TXA2 formation even in PlA2 carriers who have an increased activity of the GpIIb/IIIa receptor44. We suggest that PlA1/A2 polymorphism does not affect the anti-platelet effect of aspirin as aspirin has an upstream inhibitory effect on platelet aggregation, which may also account for its inability to predict the clinical outcomes.

Goodman et al.45 reported their meta-analysis which showed a genetic association between the PlA1/A2 molecular variant and aspirin resistance in healthy subjects who took aspirin alone. However, they failed to find significant association between carriage of the PlA2 allele and aspirin resistance in subjects with cardiovascular disease, which was consistent with our findings. In fact, in our study 51.1% of the CAD patients are on dual anti-platelet treatment with aspirin and a P2Y12 receptor antagonist, and it has been demonstrated that P2Y12 antagonists potentiate the inhibitory actions of PGI2, which would be converted to TXA2 46. So it is possible that P2Y12 antagonists cause certain degree of platelet inhibition through blockage of TXA2 formation, and thereby obscuring the presence of aspirin resistance, as well as bias our study results. However, we believe that the efficacy of PlA1/A2 on aspirin resistance as well as the clinical outcomes would be too weak to be significant, should it exist.

Our study found that laboratory aspirin resistance predicted all-cause death and TVR, though the following studies reported negative results. Kim et al.37 included 220 patients who were planned to receive off-pump coronary artery bypass surgery (OPCAB), but it only recorded in-hospital clinical outcomes during the same admission period with relatively small sample size and very short follow-up intervals. This could lead to a study bias. The study conducted by Gulmez et al.16 was similar in this aspect.

Floyd et al. investigated the association between PlA1/A2 polymorphism of glycoprotein IIIa and the efficacy of aspirin by meta-analysis in 201447, in which they included 14 papers18,19,21,22,24,25,26,29,30,31,32,33,48,49 published before 1 April 2013, and 1463 subjects who were homozygous for the PlA1 allele and 622 who carried the PlA2 allele were enrolled in their meta-analysis. In our study, however, we only included papers studying on patients with coronary artery disease, so we excluded 2 papers48,49 included in Floyd’s study, but recruited another 4 papers20,23,27,28 in the analysis. As a result, sixteen studies18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33 including 3077 CAD patients were enrolled to assess the association between PlA1/A2 gene polymorphism and aspirin resistance. To the best of our knowledge, this meta-analysis is the first to simultaneously investigate the relationship among PlA1/A2, aspirin resistance and clinical outcomes, which included the latest studies and recruited the largest study population, and would come out with the most updated, accurate and realistic results on this topic.

Although we adopted a random effect model in the process of this meta-analysis, the following limitations could not be avoided: (1) The recruited studies differed in genders and the duration of follow-up time, which might cause heterogeneities. If more data were available in the future, it would be preferable to perform a refined stratification analysis with data being adjusted for these factors. (2) The recruited studies differed in geographical areas, of which 4 were performed in Asia, 8 in America, and 12 in Europe. It was reported that the prevalence of the PlA2 allele is dependent on ethnicity, with a frequency of approximately 15 per 100 in Caucasian populations falling to 1 per 100 in Oriental populations50,51. If we have a detailed data on the ethnicity of each included patient, a stratification analysis on different ethnicity would be valuable to elucidate whether the results would be differ by ethnicity. (3) As pointed out by Floyd et al., mortality bias may attenuate or entirely obscure any true association52. Almost a third of individuals with a first major coronary event die out-of-hospital, and are not accounted for in the predominantly retrospective data presented in this meta- analysis52.

In conclusion, in aspirin-treated CAD patients, the laboratory aspirin resistance predicts all-cause death and TVR. However, the PlA1/A2 gene polymorphism predicts neither the laboratory aspirin response nor the clinical outcomes. Given this result, individualized anti-platelet treatment with the guidance of PlA1/A2 genetic testing may not be meaningful. This is in accordance with the 2018 ESC/EACTS guidelines on myocardial revascularization which recommended that genetic testing can not be recommended on a routine basis for tailoring and escalating dual anti-platelet treatment after stent implantation in all percutaneous coronary intervention (PCI)-treated patients53. However, as the results of laboratory platelet function test on individual aspirin response predict clinical outcomes, patients with aspirin resistance would benefit from intensified anti-platelet treatment.