Risk of infections in patients treated with ticagrelor vs. clopidogrel: a systematic review and meta-analysis

Abstract

Introduction

Cardiovascular diseases (CVDs) are the leading causes of mortality and morbidity worldwide.^1,2 P2Y12 inhibitors, such as clopidogrel and ticagrelor, are effective in preventing myocardial infarction, stroke, and cardiovascular death.³ Both European and US guidelines recommend the use of ticagrelor, together with aspirin as dual antiplatelet therapy, in patients with acute coronary syndrome after percutaneous coronary intervention (ST-segment elevation myocardial infarction or non-ST-segment elevation acute coronary syndrome) to improve cardiovascular outcomes, although at a cost of higher risk of bleeding compared to clopidogrel; while clopidogrel is preferred for patients with stable coronary artery diseases undergoing percutaneous coronary intervention.^4–7 So far, these recommendations were made mainly based on the evidence about their effects on cardiovascular outcomes and bleeding, while other outcomes, such as infections, which are not rare and could be life-threatening, have been much less addressed.^8,9

Oral P2Y12 inhibitors are believed to increase the risk of infections.³ Previous cohort studies showed that clopidogrel increased the risk of infections by 48%⁸ to 51%⁹ compared to placebo. This is supported by findings from both in vitro and in vivo studies.¹⁰ Interestingly, ticagrelor, but not clopidogrel, has been suggested to have protective effects against infections. The Platelet Inhibition and Patient Outcomes (PLATO) trial showed that patients treated with ticagrelor had a lower risk of pneumonia and death due to pneumonia and sepsis than clopidogrel.^11,12 In a randomized trial of patients with pneumonia, ticagrelor reduced thrombo-inflammatory biomarkers, improved lung function, and reduced the need for supplemental oxygen.¹³ Moreover, ticagrelor had a protective effect on renal function in sepsis-induced acute kidney injury mice models.¹⁴ Recent in vivo and in vitro studies have revealed its antibacterial and other anti-infection effects.^15,16

Nevertheless, the comparison between ticagrelor and clopidogrel in terms of infection outcomes yielded inconsistent findings. A post hoc analysis of the PLATO trial¹¹ showed that ticagrelor significantly reduced the risk of infections, but this effect was not statistically significant in the Examining Use of Ticagrelor in Peripheral Artery Disease (EUCLID) trial.¹⁷ Furthermore, some trials observed reverse association in that the patients on ticagrelor had a higher risk of pneumonia, pharyngitis, upper respiratory tract infection (URTI), and urinary tract infection (UTI).^18,19 However, their sample sizes were small. It remains uncertain whether the inconsistencies were due to small sample size, patient characteristics, specific infection types, or chance. Therefore, the objective of this systematic review and meta-analysis was to compare the effect of ticagrelor vs. clopidogrel on common infection outcomes (pneumonia, URTI, UTI, and sepsis) in patients with CVDs at baseline.

Methods

This systematic review and meta-analysis were conducted and reported in accordance with the recommendations of the Cochrane Handbook (Version 5.1.0)²⁰ and the PRISMA statement.²¹ This study was registered on PROSPERO (CRD42020154506).

Literature search

We conducted a literature search in Embase, MEDLINE, Cochrane Library, and ClinicalTrials.gov for eligible studies through 15 October 2019. The following keywords and their synonyms were used: (i) ticagrelor (Brilinta, Brilique, Possia), (ii) clopidogrel (Plavix), and (iii) randomized controlled trial. The search strategy is shown in Supplementary material online, Appendix S1. The reference lists of eligible studies and relevant reviews were also manually searched for additional studies.

Study selection and data extraction

We included randomized controlled trials that compared ticagrelor with clopidogrel in adult patients (aged ≥ 18 years) with CVDs and reported outcome data for infections. The baseline CVDs included but not limited to acute coronary syndrome, stroke, angina, myocardial infarction, and peripheral artery diseases. The primary outcome was pneumonia, while secondary outcomes included URTI, UTI, and sepsis. There were no restrictions on treatment duration, follow-up time, sample size, and whether using ticagrelor or clopidogrel alone or being used in dual antiplatelet therapy with aspirin. We excluded studies that included immunocompromised (e.g. HIV) and immunosuppressed patients. When multiple studies were found to be based on the same trial, we included the one with the biggest eligible sample size or longest follow-up period; sample size was prioritized over follow-up length when the study with the biggest sample size and the study with the longest follow-up period were different. All the inclusion and exclusion criteria were specified prior to literature search and screening. The titles and abstracts of all retrieved citations were first screened to assess their potential eligibility, and final eligibility was determined after examining their full texts.

Data extraction and quality assessment

We extracted the following information using a pre-designed data extraction form: bibliographic information (author and year of publication), study information (trial name, trial registration number, country, and sample size), patient characteristics (age, proportion of male patients, baseline CVDs, comorbidities, and surgical procedures), treatment information (regimen, dose, and treatment duration), and outcome data (event number for each outcome and follow-up time). Since we only included randomized trials and all outcomes of interest were binary, we extracted the original 2 × 2 table for each outcome. When a trial included multiple arms of the same drug but at different doses, we combined them into one single drug arm. For example, we combined two arms of ticagrelor at 45 mg and 90 mg twice daily²² into one-single ticagrelor arm and comparing it with the clopidogrel arm. This method is recommended in the Cochrane Handbook²⁰ and adopted in previous systematic reviews.²³ Many studies did not report data on infection outcomes of interest in their journal-published reports, in which case we examined their ClinicalTrials.gov webpages for more information or contacted the corresponding authors by e-mail.

We extracted other information necessary for the assessment of methodological quality. The Cochrane Collaboration’s tool for assessing risk of bias was used to assess methodological quality.²⁴ This tool evaluated potential bias from seven domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other biases. The risk of bias was judged as high, low, or unclear in each domain, and the overall risk of bias was judged based on the risks in all seven domains, as low when all domains were judged as low, as high when any domains were judged as high, or as unclear otherwise.

Study selection, data extraction, and quality assessment were conducted by three independent authors (Q.F., H.L.L., and M.F.T.). Any disagreement was resolved by discussion until consensus was reached or by consulting a fourth author.

Data analysis

The clopidogrel arm was used as the reference in all the analyses. Intention-to-treat analysis was employed. Relative risks (RRs) and their 95% confidence intervals (CIs) were estimated from the extracted 2 × 2 table for each outcome and then pooled across studies to obtain an overall effect estimate by using a random-effects model with inverse variance weighting. Relative risk >1 would favour clopidogrel, otherwise ticagrelor. Subgroup analysis was prespecified for pneumonia based on baseline CVD type, therapy type (monotherapy vs. dual therapy), treatment duration (<12  vs. ≥12 months). Subgroup analyses were also performed based on the proportion of diabetic patients in each trial (≥30% vs. <30%) because diabetic patients are generally at higher risk of infections,²⁵ and 30% was the median proportion of diabetic patients in all included studies. Sensitivity analysis was performed by excluding studies with a high risk of bias and by using odds ratio (OR) as the effect measure. Statistical heterogeneity across studies was measured by the I² statistic and Cochrane’s Q test. An I² >50% or a P-value <0.10 was suggestive of substantial heterogeneity, in which case meta-regression would be used to investigate potential sources of heterogeneity. Publication bias was investigated with funnel plot, but Egger’s test for asymmetry in funnel plot would be performed only when the number of eligible studies was 10 or more because it would generate misleading results otherwise.^20,26 Asymmetries in the funnel plot by visual assessment or a P-value for Egger’s test <0.10 would suggest potential publication bias, and the trim-and-fill method would be employed to adjust for potential bias.²⁷ The statistical significance level was decided at 0.05 unless specified otherwise. All data analyses were performed with the ‘meta’ package in R software (Version 3.4.3).

Results

Among the 5231 citations identified by literature search, 10 trials^{17–19,22,28–33} with altogether 37 514 patients (18 790 on ticagrelor and 18 724 on clopidogrel) were considered eligible (Figure 1). The mean age was 62.7 (range 58.7–67.0) years old, and 76.8% (range 70.0–89.2%) patients were males. The median sample size was 105.5 (range 36–18 624). The median proportion of diabetes was 29.8% (range 16.8–52.5%). The median proportion of current or previous smokers was 52.5% (range 35.8–77.9%) (Table 1). Eight trials enrolled patients from multiple centres.^{17,18,22,28–31,33} Patients had acute coronary syndrome in four trials,^18,19,28,29 stable coronary disease in three trials,^22,30,33 peripheral artery disease in one trial,¹⁷ myocardial infarction in one trial,³¹ and coma with cardiac arrest and percutaneous coronary intervention in one trial.³² Ticagrelor and clopidogrel were often used in dual antiplatelet therapy with aspirin, but used alone in three trials.^17,19,33 The trials encompassed a wide range of treatment duration, from 2 days to 30 months with a median of 42 days. Four trials had a follow-up period of 12 months or longer.^17,28,29,31 Four trials included more than 30% of the study patients with baseline diabetes.^17,18,28,33 More baseline characteristics are shown in Supplementary material online, Appendix S2. According to the Cochrane Collaboration’s tool for assessing risk of bias, 5 out of the 10 trials had an overall high risk of bias, due to their open-label design.^{18,19,31–33} The other five trials had low risk of bias (Supplementary material online, Appendix S3).^{17,22,28–30}

Figure 1

Flowchart of study selection.

Open in new tabDownload slide

Seven studies with altogether 37 228 patients reported data on pneumonia. The RRs ranged from 0.32 to 3.20. Overall, ticagrelor was associated with a 20% risk reduction in pneumonia compared to clopidogrel (RR 0.80, 95% CI 0.67–0.95) (Figure 2A). Prespecified subgroup analyses demonstrated that ticagrelor significantly reduced the risk of pneumonia when it was used for ≥12 months (RR 0.80, 95% CI 0.67–0.96), in dual therapy (RR 0.73, 95% CI 0.55–0.96), and in patients with acute coronary syndrome (RR 0.72, 95% CI 0.53–0.98), although the difference between subgroups did not reach the prespecified significance level of 0.05, mainly due to the wide CIs of the other subgroup(s) (Table 2). Since the PLATO and the EUCLID trials contributed more than 90% of total weight and might drive the overall effect, we conducted a sensitivity analysis removing these two trials, which yielded an overall RR 0.79 (95% CI 0.42–1.48). The point estimate was very close to the primary analysis, although the CI became wider due to the smaller sample size.

Figure 2

Forest plots of comparing ticagrelor with clopidogrel regarding infection outcomes (A) pneumonia, (B) URTI, (C) UTI, and (D) sepsis. URTI, upper respiratory tract infection; UTI, urinary tract infection.

Open in new tabDownload slide

Seven trials with 33 579 patients, four trials with 33 333 patients and two trials with 32 466 patients reported the outcome of URTI, UTI, and sepsis, respectively. By pooling results across studies, we did not find that ticagrelor reduced risk of URTI (RR 0.71, 95% CI 0.34–1.48), UTI (RR 1.09, 95% CI 0.73–1.64), or sepsis (RR 0.79, 95% CI 0.50–1.26) (Figure 2B–D).

The heterogeneity in all these meta-analyses ranged from no to low. Sensitivity analyses by excluding trials with a high risk of bias yielded similar results to the primary analyses (Supplementary material online, Appendix S4). Substituting RR with OR produced similar results: OR 0.79 (95% CI 0.66–0.95) for pneumonia, 0.70 (95% CI 0.33–1.49) for URTI, 1.09 (95% CI 0.73–1.65) for UTI, and 0.79 (95% CI 0.50–1.26) for sepsis. Subgroup analyses based on diabetic status showed no subgroup difference in the effect of ticagrelor on pneumonia [RR 0.83 (95% CI 0.66–1.04) vs. 0.74 (95% CI 0.55–1.01) for diabetic and non-diabetic patients, respectively, P for subgroup difference = 0.55], URTI [RR 0.47 (95% CI 0.13–1.65) vs. 0.79 (95% CI 0.31–2.01), P = 0.51], UTI [RR 1.09 (95% CI 0.56–2.15) vs. 1.31 (95% CI 0.72–2.38), P = 0.70], and sepsis [RR 1.01 (95% CI 0.55–1.84) vs. 0.63 (95% CI 0.34–1.15), P = 0.28]. Symmetry was observed in the funnel plots for pneumonia, URTI, and sepsis, but not for UTI (Supplementary material online, Appendix S5). Since the numbers of eligible studies for all outcomes were below 10, we did not test for publication bias formally as explained in Methods section. Trim-and-fill method generated an overall RR of 0.98 (95% CI 0.66–1.46) for UTI.

Discussion

In this systematic review and meta-analysis of 10 trials with 37 514 CVD patients, we found that patients treated with ticagrelor had a lower risk of pneumonia, but not of URTI, UTI, or sepsis, compared with clopidogrel. To our knowledge, this is the first systematic review and meta-analysis that addresses this research question.

The PLATO trial²⁹ was the first study revealing the protective effect of ticagrelor on infection outcomes, in which patients treated with ticagrelor were at a lower risk of pulmonary adverse effects and related death but not of sepsis, URTI, or UTI. These findings were consistent with our results. However, the PLATO trial was the only one that showed a protective effect of ticagrelor against pneumonia,²⁹ while the other trials failed to show a significant effect, which may be due to reasons, such as small size, short follow-up, or insufficient events. Two trials even suggested an association in the opposite direction.^19,31 Our study is of great value in resolving the inconsistencies across studies and provides more reliable evidence. The post hoc analysis of the PLATO trial¹¹ revealed that ticagrelor reduced the risk of pulmonary adverse effects by 17% (RR 0.83, 95% CI 0.71–0.97; computed from extracted data). However, dyspnoea, which counted as a pulmonary adverse effect, is more common in patients treated with ticagrelor compared to clopidogrel.³⁴ Ticagrelor’s effect on reducing pneumonia might therefore be more than 17%. In PLATO, the reduction in pneumonia was 29% (RR 0.71, 95% CI 0.53–0.98; as shown in Figure 2A). In our meta-analysis, the overall risk reduction in pneumonia was around 20%.

The primary use of ticagrelor and clopidogrel, as P2Y12 inhibitors, is protection against cardiovascular events through platelet inhibition.³⁵ Inhibition of platelets has downstream effects on inflammation and immunity, such as reducing platelet release of pro-inflammatory alpha-granule contents and the formation of pro-inflammatory platelet-leucocyte aggregates.¹⁰ These features lead to a potentially higher risk of infection, as demonstrated by previous cohort studies of clopidogrel.^8,9 As a more potent platelet-inhibitor than clopidogrel, the protective effect of ticagrelor on infection outcomes was unlikely to be due to its platelet-inhibition function. This hypothesis is also supported by the observation that clopidogrel and prasugrel had similar effects on infection outcomes despite their difference in platelet-inhibition efficacy.³⁶

Unlike clopidogrel, ticagrelor can additionally inhibit cellular uptake of adenosine via inhibiting the equilibrative nucleoside transporter 1 (ENT1) receptor,³⁷ which increases intracellular levels of adenosine. Adenosine has been shown to activate neutrophils to release cytokines, chemokines, and arachidonic acid-derived lipid mediators, to promote neutrophil chemotaxis, phagocytosis, and degranulation via low-affinity G protein-coupled receptors A_2A and A_2B at high concentration and play a role in the resolution of lung injury.^16,38–41 Lancellotti et al.¹⁵ showed that ticagrelor also has a direct anti-microbial effect on multiple Gram-positive bacteria strains in vitro, which is not seen in other P2Y12 inhibitors. Ticagrelor inhibited biofilm growth and dissemination to surrounding tissues in Staphylococcus aureus-infected mice at conventional dosage.¹⁵ Ticagrelor has also been believed to have a promising role in preventing multi-organ failure among patients with sepsis due to resistant Gram-positive cocci.⁴² These are the two potential mechanisms by which ticagrelor reduces pulmonary infections, but more research is needed.

Previous studies have focused on ticagrelor’s effect on preventing pneumonia and improving lung function,^11,13 but few on its effect on other infections. Currently, there is no evidence of a reduction in infections in the upper respiratory tract or urinary tract. Pneumonia is usually due to Gram-positive pathogens, whereas URTI is usually caused by virus, and UTI and septicaemia are usually caused by Gram-negative bacteria. Therefore, our findings are actually in line with Lancellotti et al.’s finding¹⁵ that ticagrelor has anti-microbial effects on Gram-positive bacteria, rather than other bacteria strains.

We found that ticagrelor showed a significant protective effect when used in dual therapy, for a long duration, and in patients with acute coronary syndrome, which is consistent with the recommended usage in current guidelines.^5,6 Aspirin in dual therapy is not known to be associated with the risk of infection, and so it can be combined with ticagrelor as dual therapy.^43,44 Ticagrelor is preferred over clopidogrel in the management of ST-segment elevation myocardial infarction and non-ST-segment elevation acute coronary syndrome, especially for those undergoing percutaneous coronary intervention.^5,6,45 Although current recommendations on the usage of dual antiplatelet therapy were made mainly based on ticagrelor’s efficacy regarding cardiovascular outcomes and safety regarding bleeding,⁴⁵ they are now further supported by our findings that ticagrelor could provide patients with the additional benefit of reducing pneumonia risk.

Three randomized trials compared ticagrelor with placebo in patients with baseline CVDs: stable coronary disease and diabetes in the THEMIS trial,⁴⁶ acute stroke or transient ischaemic attack in the SOCRATES trial,⁴⁷ and previous heart attack in the PEGASUS trial.⁴⁸ All three trials consistently showed that ticagrelor tended to reduce the risk of pneumonia, although the results did not reach statistical significance. Combining them using meta-analysis yielded similar results [RR 0.91 (95% CI 0.77–1.07), Supplementary material online, Appendix S6]; the lack of significance could be attributed to the low number of pneumonia cases in these studies. This finding seems also consistent with ticagrelor’s antibacterial effect in vitro and in vivo¹⁵; however, future studies with larger sample size or longer follow-up are required for confirmation. Therefore, physicians may consider ticagrelor as the preferred therapy for eligible acute coronary syndrome patients undergoing percutaneous coronary intervention and at high risk of pneumonia, such as the elderly, the bedbound, smokers, and chronic obstructive pulmonary disease patients. Moreover, polymorphisms in CYP2C19 can cause variability in a patient’s response to clopidogrel, which is a prodrug that requires activation.⁴⁹ Ticagrelor does not have this issue.

Since pneumonia and CVDs are both leading causes of morbidity and mortality in the elderly, dual antiplatelet therapy with ticagrelor should be further recommended to eligible patients because it is effective in reducing the risk of both conditions. In people over the age of 65 years in the general population, the incidence of pneumonia ranges from 25 to 44 per 1000 person-year,⁵⁰ higher than that in the PLATO trial (10.2 per 1000 person-year) and the EUCLID trials (9.5 per 1000 person-year). This is because the trials tend to exclude patients that are very old. Translating the relative effect (RR = 0.80) into absolute effect using a baseline incidence of 10 per 1000 person-year, the 10-year risk difference is 2.0% and the number needed to treat is 50, meaning that treating 50 eligible patients with ticagrelor would prevent one case of pneumonia in 10 years. However, when using the baseline incidence of 44 (or 25) per 1000 person-year, the 10-year risk difference and the number needed to treat are 8.8% (or 5.0%) and 12.5 (or 20), respectively.

Some limitations have to be acknowledged in this study. First, we only included studies published in English, which might cause selection bias. Second, we did not perform Egger’s test for examining asymmetry in funnel plots. Egger’s test would give misleading results in the case of a small number of eligible studies (<10) and is therefore not recommended. The funnel plots are shown in Supplementary material online, Appendix S5. Third, we did not include the outcome of infection-related mortality, which is clinically more important but rarely reported in randomized trials. The PLATO trial showed that ticagrelor and clopidogrel had a similar risk of infection-related death,¹² but more data are required for conducting a meta-analysis. Although the effect on pneumonia-related mortality remains unknown, pneumonia per se can cause substantial disease burdens and decreased quality of life. Fourth, the results of prespecified subgroup analyses did not reach statistical significance. Although we recommend long-term use of ticagrelor in dual therapy in acute coronary syndrome, we could not conclude that ticagrelor did not have an effect in treatment duration <12 months, in patients with other conditions than acute coronary syndrome, or in monotherapy. More studies are required to explore its potential clinical applications. Fifth, the PLATO and the EUCLID trials contributed disproportionately large weights in the meta-analysis of pneumonia, but the sensitivity analysis by removing them generated similar point estimates although wider CIs. Sixth, the two big trials included, PLATO and EUCLID, had long follow-up periods (12 and 30 months, respectively), which made it difficult to identify whether other factors that may have occurred during follow-up that would affect the final results. Seventh, this study was conducted with summary data instead of individual patient data, which prevented us from performing more flexible subgroup analysis or regression. Eighth, the studies included in this meta-analysis were randomized controlled trials that were not specifically designed with infection as the primary outcome, and so were underpowered to allow firm conclusions on infection risk. Future randomized controlled trials investigating infections as the prespecified outcome are warranted. Ninth, more epidemiological evidence is required to judge whether ticagrelor per se reduces the risk of pneumonia compared to placebo.

Conclusion

This meta-analysis found that compared to clopidogrel, ticagrelor reduces the risk of pneumonia, but not URTI, UTI, or sepsis. This further supports the guideline-recommended use of ticagrelor with aspirin in dual antiplatelet therapy in the management of patients with acute coronary syndrome undergoing percutaneous coronary intervention, especially in those with a high risk of pneumonia. However, further research is needed to confirm the causality and investigate the mechanism of this protective effect.

Supplementary material

Supplementary material is available at European Heart Journal – Cardiovascular Pharmacotherapy online.

Conflict of interest: none declared.

References

Author notes