Ticagrelor with or without aspirin in high-risk patients with anaemia undergoing percutaneous coronary intervention: a subgroup analysis of the TWILIGHT trial

Abstract

Introduction

Dual antiplatelet therapy (DAPT), which refers to the combination of aspirin and a P2Y12 inhibitor, is the standard of care for the prevention of ischaemic events in patients undergoing percutaneous coronary intervention (PCI).^1,2 However, this benefit occurs at the expense of higher rate of bleeding complications, which gradually increases with prolonged or more potent DAPT.^3,4 Given the enhanced morbidity and mortality associated with bleeding events, novel antiplatelet treatment regimens have been examined to reduce the risk of bleeding while maintaining protection from ischaemic events after PCI.⁵ Among those, a short period of DAPT followed by P2Y12 inhibitor monotherapy has emerged as a promising bleeding reduction strategy in several studies.^6–8 The TWILIGHT (Ticagrelor with Aspirin or Alone in High-Risk Patients after Coronary Intervention) trial showed that among high-risk patients undergoing PCI, after 3 months of DAPT, P2Y12 inhibitor monotherapy with ticagrelor reduced bleeding without increasing ischaemic harm compared with a standard ticagrelor-based DAPT.⁹

Anaemia is a common condition in patients undergoing PCI, with prevalence >30% in all-comer cohorts.¹⁰ Anaemia is a marker of frailty and comorbidity and has been associated with increased mortality, bleeding, and ischaemic complications after PCI.^11,12 Several bleeding scores and the criteria proposed by the Academic Research Consortium (ARC) to identify patients at high bleeding risk (HBR) include anaemia.^13–15 Several studies have confirmed that shorter DAPT regimens after PCI are effective in reducing bleeding risk without increasing ischaemic events in HBR patients.^7,8,16,17 However, specific evidence for patients with anaemia is lacking.

Accordingly, we aimed to assess the effect of ticagrelor monotherapy vs. ticagrelor plus aspirin among patients with anaemia included in the TWILIGHT trial.

Methods

TWILIGHT (NCT02270242) was a randomized, placebo-controlled trial conducted at 187 sites in 11 countries. The rationale, design, and principal results of the trial have been reported previously.¹⁸ TWILIGHT was an investigator-initiated trial designed, coordinated, and sponsored by the Icahn School of Medicine at Mount Sinai. AstraZeneca provided an investigator-initiated grant and supplied ticagrelor for the trial but had no role in the design, collection, analysis, or interpretation of the data. The executive and steering committees were responsible for trial conduct, integrity of data analysis, and reporting of results. The trial protocol was approved by national regulatory agencies and institutional review boards or ethics committees of participating centres. The safety of trial participants was ensured by an external and independent data and safety monitoring board.

Study population

Patients were eligible to participate if they underwent a successful PCI with at least one commercially available drug-eluting stent and were prescribed at discharge a regimen of ticagrelor plus aspirin by the treating clinician. The presence of at least one clinical and one angiographic feature associated with a high risk for ischaemic or bleeding events was required for trial inclusion.^9,18 Clinical criteria were as follows: age ≥65 years, female sex, troponin-positive acute coronary syndrome (ACS), atherosclerotic vascular disease (prior myocardial infarction [MI], coronary revascularization, or peripheral arterial disease), diabetes mellitus requiring medication, and chronic kidney disease (CKD) (estimated glomerular filtration rate <60 mL/min/1.73 m² or creatinine clearance <60 mL/min). Angiographic criteria included multivessel coronary artery disease, total stent length >30 mm, thrombotic target lesion, bifurcation lesion requiring two stents, obstructive left main or proximal left anterior descending coronary artery lesion, and calcified target lesion requiring debulking devices. Key exclusion criteria included presentation with an ST-segment elevation MI (STEMI), cardiogenic shock, prior stroke, need for oral anticoagulation, or contraindication to aspirin or ticagrelor.

All enrolled patients received open-label ticagrelor (90 mg twice daily) and enteric-coated aspirin (81–100 mg/day) after index PCI. At 3 months, patients were randomized 1:1 in a double-blind fashion to aspirin or matching placebo for 12 months in addition to open-label ticagrelor.^9,18 Patients who were non-adherent to ticagrelor or aspirin or who experienced Bleeding Academic Research Consortium (BARC) type 3b or higher bleeding events or ischaemic events (stroke, MI, or coronary revascularization) between the index PCI and 3 months were not eligible for randomization (Figure1). Follow-up occurred 1 month after randomization via telephone and in person at 6 and 12 months after randomization. At the end of protocol-mandated therapy, patients were switched to a standard-of-care antiplatelet regimen at the discretion of their treating physicians, followed by final telephone follow-up 3 months later.

Figure 1

Study design. The Ticagrelor with Aspirin or Alone in High-Risk Patients after Coronary Intervention trial enrolled patients undergoing percutaneous coronary intervention and at high risk of ischaemic or bleeding complications. After 3 months of ticagrelor-based dual antiplatelet therapy, patients free from ischaemic or bleeding events and adherent to dual antiplatelet therapy were randomized to aspirin or matching placebo for 12 months in addition to open-label ticagrelor. Outcomes were assessed at 12 months after randomization. BARC, Bleeding Academic Research Consortium; MI, myocardial infarction.

Open in new tabDownload slide

Clinical endpoints

The primary endpoint was the composite of BARC type 2, 3, or 5 bleeding through 1 year after randomization.^9,18 The key secondary endpoint was the composite of all-cause death, MI, or stroke. Other secondary bleeding endpoints included BARC type 3 or 5 bleeding; TIMI (Thrombolysis in Myocardial Infarction) major or minor bleeding; GUSTO (Global Use of Strategies to Open Occluded Arteries) moderate, severe, or life-threatening bleeding; and major bleeding as defined by the International Society on Thrombosis and Haemostasis (ISTH).^9,18

Additional secondary endpoints were cardiovascular death, non-fatal MI, ischaemic stroke, and definite or probable stent thrombosis. Myocardial infarction was defined according to the third universal definition, and stent thrombosis was classified according to the ARC definition.^9,18 All clinical events were adjudicated by an independent committee, blinded to treatment assignment.

Statistical analysis

In this post-hoc analysis, patients were stratified into two groups according to the presence or absence of anaemia at the time of index PCI, defined as haemoglobin (Hb) <13 g/dL for men and <12 g/dL for women. For additional analyses, anaemia was further classified into mild (Hb between 11 and 12.9 g/dL for men and between 11 and 11.9 g/dL for women) and moderate–severe (Hb < 11 g/dL for both sexes). These definitions were in line with those proposed by the ARC-HBR consensus.¹⁵ Patients with any Hb value were included; only subjects with missing Hb values were excluded.

Clinical and procedural characteristics were summarized by anaemia status and randomized group as mean and standard deviation for continuous variables and frequencies and percentages for categorical variables. The cumulative incidence of both primary and secondary endpoints was estimated using the Kaplan–Meier method. Patients without primary endpoints within 12 months after randomization were censored at the time of death, last known contact, or 365 days, whichever came first. Hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using unadjusted Cox proportional hazards model.

Analyses for the bleeding outcomes were performed in the intention-to-treat cohort, ischaemic outcomes in the per-protocol cohort (i.e. randomized participants who completed all study-related contacts without any major protocol deviations).^9,18 Treatment effects were estimated according to anaemia status with formal interaction testing to assess for effect modification. Finally, the association between Hb (as a continuous variable) and the rates of the primary and key secondary endpoints was also evaluated by fitting a smoothing spline curve with four knots placed at equally spaced percentiles across the range of Hb. All analyses were performed using Stata version 16.0 (StataCorp, College Station, Texas).

Results

Population characteristics

Among 6828 patients randomized in the TWILIGHT trial with available Hb values at baseline, 1329 (19.5%) had anaemia (Figure1 and Supplementary material online, Figure 1). Of these, 50.8% were randomized to ticagrelor plus placebo and 49.2% to ticagrelor plus aspirin. Among non-randomized patients, not included in this analysis, frequency of anaemia was slightly higher (24.9%).

Compared to patients without anaemia, anaemic patients were on average older, more likely to be women, from North America or Asia, of non-white race, and to have comorbidities, such as diabetes, hypertension, prior coronary artery bypass graft surgery, peripheral artery disease, CKD, and a history of major bleeding (Supplementary material online, Table 1). Mean Hb was 11.6 ± 1.1 g/dL in anaemic patients and 14.5 ± 1.2 g/dL in non-anaemic patients and was similar in the two randomized arms (Table1). Among patients with anaemia, femoral artery access, multivessel coronary artery disease, and calcific lesions were more common (Supplementary material online, Table 2).

Within patients with or without anaemia, demographic, clinical, and procedural characteristics were well balanced between treatment arms, except for the rates of current smoking and total stent length, which were higher in the ticagrelor plus placebo arm than in the ticagrelor plus aspirin arm (Table1 and Supplementary material online, Table 3).

At 1 year after randomization, patients with anaemia were more likely to have permanently discontinued ticagrelor (17.1% vs. 12.4%, P < 0.001) or the study drug (21.9% vs. 16.3%, P < 0.001) than patients without anaemia.

Bleeding events

Bleeding event rates were inversely associated with Hb values (Figure2A). Patients with anaemia compared to those without anaemia had significantly higher rates of BARC 2, 3, or 5 (8.3% vs. 4.9%) and BARC 3 or 5 bleeding (3.1% vs. 1.1%) (Supplementary material online, Table 4).

Figure 2

One-year rates of bleeding (A) and ischaemic (B) events according to baseline haemoglobin values. A smoothing spline curve was used to plot the rates of Bleeding Academic Research Consortium (BARC) type 2, 3, or 5 bleeding (A) and all-cause death, myocardial infarction (MI), or stroke (B) at 1 year after randomization. The lines indicate Kaplan–Meier estimated event rates in the overall population (black), in the ticagrelor plus placebo arm (light blue), and in the ticagrelor plus aspirin arm (dark blue). The mean haemoglobin value was 13.9 ± 1.6 g/dL in the overall population.

Open in new tabDownload slide

Among patients with anaemia, the primary outcome of BARC type 2, 3, or 5 bleeding occurred in 42 patients (6.4%) randomized to ticagrelor plus placebo and 67 patients (10.4%) randomized to ticagrelor plus aspirin (HR: 0.60; 95% CI: 0.41–0.88; P = 0.009) (Figure3A ). In anaemic patients, rates of BARC type 3 or 5 bleeding were 1.7% in the ticagrelor plus placebo group and 4.5% in the ticagrelor plus aspirin group (HR: 0.37; 95% CI: 0.18–0.73; P = 0.005) (Figure4). This treatment effect was consistent across different bleeding scales, including TIMI, GUSTO, and ISTH (Table2).

Figure 3

Kaplan–Meier curves for BARC 2, 3, or 5 bleeding (A) and death, MI, or stroke (B) in patients with or without anaemia stratified by randomized treatment. BARC, Bleeding Academic Research Consortium; MI, myocardial infarction.

Open in new tabDownload slide

Figure 4

Effect of randomized treatment on events at 1 year after randomization in patients with or without anaemia. BARC, Bleeding Academic Research Consortium; MI, myocardial infarction.

Open in new tabDownload slide

There was no significant interaction between anaemia status and treatment group with respect to any of the bleeding endpoints.

Ischaemic events

The key secondary endpoint of all-cause death, MI, or stroke increased with decreasing values of Hb (Figure2B ) and was significantly higher in patients with anaemia than without anaemia (5.0% vs. 3.7%); this difference was driven by an excess of all-cause death in anaemic patients (Supplementary material online, Table 4). Among patients with anaemia, all-cause death, MI, or stroke occurred in 34 patients (5.2%) randomized to ticagrelor plus placebo vs. 31 patients (4.8%) randomized to ticagrelor plus aspirin (HR: 1.07; 95% CI: 0.66–1.74; P = 0.779) (Figure3B).

Rates of all-cause death (1.5% vs. 2.2%), MI (3.5% vs. 3.5%), ischaemic stroke (0.5% vs. 0.3%), and definite or probable stent thrombosis (0.5% vs. 0.8%) were similar between treatment groups (P > 0.20 for all) (Figure4). There was no significant interaction between anaemia status and treatment group with respect to the ischaemic endpoints.

Outcomes by anaemia severity

The risk of clinically relevant or major bleeding increased progressively in patients with mild (6.7%) and moderate–severe (14.7%) anaemia, as compared to non-anaemic patients (4.9%). Moreover, moderate–severe anaemia was associated with a significantly higher risk of all-cause death and MI as compared to no anaemia (5.7% vs. 3.7%) (Supplementary material online, Table 5).

The reduction in BARC type 2, 3, or 5 bleeding associated with ticagrelor monotherapy was consistent across levels of anaemia severity (P interaction = 0.769) (Figure5). In contrast to the findings for our primary endpoint, the relative risk reduction in major bleeding tended to be greater in patients with moderate–severe anaemia than in patients with no anaemia irrespective of the bleeding scale applied (Figure5 and Supplementary material online, Table 6). There were no differences between the two treatment arms with respect to the key secondary endpoint of death, MI, or stroke between the three groups (P interaction = 0.962) (Figure5 and Supplementary material online, Table 6).

Figure 5

Effect of randomized treatment on events at 1 year after randomization according to degree of anaemia severity. Mild anaemia was defined as haemoglobin between 11 and 12.9 g/dL for men and between 11 and 11.9 g/dL for women; moderate–severe as haemoglobin <11 g/dL in both sexes. BARC, Bleeding Academic Research Consortium; MI, myocardial infarction.

Open in new tabDownload slide

Discussion

In this post-hoc analysis of the TWILIGHT trial, we assessed the effect on clinical outcomes of ticagrelor with or without aspirin after 3 months of ticagrelor-based DAPT in 1329 patients with anaemia undergoing PCI. The key findings from our analysis can be summarized as follows:

1. Patients with anaemia were older, had more comorbidities, were more likely to have multivessel disease, and experienced higher rates of bleeding complications, and—to a lesser extent—of death, MI, or stroke than patients without anaemia.

2. In anaemic patients, ticagrelor monotherapy reduced the 1-year risk of clinically relevant BARC 2, 3, or 5 bleeding by 40%, of BARC 3 or 5 bleeding by 63%, and was not associated with an increased risk of all-cause death, MI, or stroke as compared to ticagrelor plus aspirin.

3. The treatment effects of ticagrelor monotherapy with respect to ischaemic and bleeding outcomes were consistent irrespective of the presence of anaemia or its severity; the benefit of major bleeding reduction tended to be larger in patients with moderate–severe anaemia.

In real-world PCI cohorts, anaemia can be found in >30% of patients and is also one of the most frequently fulfiled ARC-HBR criteria among HBR patients.^10,11,19 Anaemic patients are more likely to discontinue antiplatelet therapy than non-anaemic patients²⁰ and to receive suboptimal therapy for secondary prevention after PCI, including antiplatelet therapy.^21,22 Several studies confirmed that in HBR patients shorter DAPT regimens after PCI are effective in reducing bleeding risk without ischaemic harm. However, analyses focusing on patients with anaemia are limited.

In this study, anaemia, which was defined based on the ARC-HBR definition as Hb <13 g/dL in men and <12 g/dL in women, was found in 19.5% of included patients. The lower frequency of anaemia compared to all-comer observational studies is probably due to the exclusion from the trial of patients on dialysis, with cardiogenic shock, STEMI, or liver cirrhosis, who frequently present with low Hb values, and to the selection of event-free patients for randomization. Indeed, the prevalence of anaemia was slightly higher (24.9%) among non-randomized subjects. In agreement with the main findings of the TWILIGHT trial, we found that the withdrawal of aspirin after a brief period of DAPT significantly reduced bleeding and did not incur any increase in ischaemic complications irrespective of anaemia status. Interestingly, the reduction of major bleeding associated with aspirin withdrawal seemed to be amplified in patients with anaemia irrespective of the bleeding scale used (BARC, TIMI, GUSTO, or ISTH), especially in patients with moderate or severe anaemia.

A number of trials have investigated the safety and efficacy of P2Y12 inhibitor monotherapy after a minimal duration (1–3 months) of DAPT following PCI.^6–8 However, TWILIGHT was the only trial to be placebo controlled and to enroll patients with both clinical and angiographic features associated with increased risk for ischaemic or bleeding complications post-PCI. In some of the abovementioned trials, the effect of P2Y12 inhibitor monotherapy was assessed in HBR patients, whereas analyses focused on anaemic patients are lacking. The benefit of ticagrelor monotherapy in reducing bleeding complications in patients with anaemia (40% for BARC 2, 3, or 5 and 63% for BARC 3 or 5) was similar to that observed in the HBR cohort of the TWILIGHT trial (47% for BARC 2, 3, or 5 and 69% for BARC 3 or 5).¹⁷

In keeping with previous studies, our study confirms that anaemic patients have a high burden of comorbidities, an increased risk of bleeding events, and—to a lesser extent—a higher risk of ischaemic complications after PCI.^{11,12,20,21,23} The increased rate of complications in anaemic patients undergoing PCI has been attributed to their old age and the associated comorbidities^21,24,25 but also to direct pathophysiological changes due to anaemia. Indeed, anaemia leads to impaired oxygen delivery to the myocardium and induces myocardial cell ischaemia²⁶; it also increases myocardial oxygen demand through increments in heart rate, cardiac index, and stroke volume.²⁷ In addition, patients with anaemia have high platelet reactivity on clopidogrel,^28,29 probably related to the inflammatory status generated by the same chronic diseases that promote anaemia.

Several interventions were tested to improve outcomes of anaemic patients undergoing PCI. Routine blood transfusion or erythropoietin administration has been shown to be harmful in the context of PCI.^30,31 Blood transfusion remains an effective measure in presence of low Hb values (≤8 g/dL) and symptoms.³² Other bleeding avoidance strategies, such as use of transradial access and use of bivalirudin or enoxaparin rather than unfractionated heparin in ACS, might reduce the risk of bleeding, even though dedicated studies in patients with anaemia are lacking.⁵

Our analysis demonstrated that aspirin withdrawal after a short period of ticagrelor-based DAPT represents an effective intervention to prevent bleeding events in patients with anaemia.

Limitations

These results should be interpreted in light of several limitations. First, this analysis was not prespecified; therefore, the findings should be considered exploratory. Additional caution is needed in interpreting the results in the context of an underpowered subgroup analysis. In particular, patients with moderate–severe anaemia represented a small proportion of the study population. Moreover, the diagnosis of anaemia was based on the Hb levels at the time of index PCI, which reflects solely the status of the patients before PCI. The cause of anaemia and additional data such as mean corpuscular volume, mean corpuscular Hb, and iron studies were not available. In this study, since pharmacodynamics or platelet function data were not collected, mechanistic insights on the interaction between antiplatelet agents and platelet function in anaemic patients could not be provided and should be explored in dedicated studies. Our findings may not generalize to patients treated with other oral P2Y12 inhibitors, particularly clopidogrel, or to patients presenting with STEMI, who were excluded from participation in TWILIGHT. Finally, limitations of the main trial also apply to the current analysis, including the lack of power to detect differences in the risk of important yet rare clinical events (e.g. stent thrombosis and stroke) and applicability of the findings restricted to patients who were able to adhere to 3 months of DAPT without experiencing any major bleeding or ischaemic event.

Conclusions

Among high-risk patients undergoing PCI, ticagrelor monotherapy following 3 months of DAPT significantly reduced clinically relevant bleeding compared with ticagrelor plus aspirin without evident ischaemic harm, irrespective of the presence of baseline anaemia. The reduction in major bleeding tended to be greater in patients with moderate–severe anaemia. These findings support that ticagrelor monotherapy can be implemented without any signals for harm in patients with anaemia and concomitant increased risk for ischaemic complications.

Acknowledgement

We thank all patients who participated in the TWILIGHT trial.

Funding

The TWILIGHT trial was an investigator-initiated trial designed, coordinated, and sponsored by the Icahn School of Medicine at Mount Sinai. AstraZeneca provided an investigator-initiated grant and supplied ticagrelor for the trial but had no role in the design, collection, analysis, or interpretation of the data.

Conflict of interest: A.S. received a research grant from the Swiss National Science Foundation (SNSF).

R.M. reports institutional research payments from Abbott, Abiomed, Alleviant Medical, Amgen, AM-Pharma, Arena, AstraZeneca, Atricure, Bayer, Biosensors, Biotronik, Boston Scientific, Bristol Myers Squibb, CardiaWave, CeloNova, Chiesi, Concept Medical, CSL Behring, Cytosorbents, Daiichi Sankyo, Element Science, Faraday, Humacyte, Idorsia Pharmaceuticals, Janssen, Medtronic, Novartis, OrbusNeich, PhaseBio, Philips, Pi-Cardia, PLx Pharma, RenalPro, RM Global, Shockwave, Vivasure, and Zoll; personal fees from Cine-Med Research, Novartis, and WebMD; equity <1% in Applied Therapeutics, Elixir Medical, Stel, and ControlRad (spouse); Scientific Advisory Board for AMA, ACC (BOT Member), and SCAI (Women in Innovations Committee Member); JAMA Associate Editor; and faculty CRF (no fee). A.K. is an inventor in a patent application related to drug-eluting stent technology; he also serves in the Data and Safety Monitoring Board of the TARGET IV trial sponsored by the Cardiovascular Research Foundation in New York, USA.

D.J.A. declares that he has received consulting fees or honoraria from Abbott, Amgen, AstraZeneca, Bayer, Biosensors, Boehringer Ingelheim, Bristol Myers Squibb, Chiesi, Daiichi Sankyo, Eli Lilly, Haemonetics, Janssen, Merck, Novartis, PhaseBio, PLx Pharma, Pfizer, and Sanofi; D.J.A. also declares that his institution has received research grants from Amgen, AstraZeneca, Bayer, Biosensors, CeloNova, CSL Behring, Daiichi Sankyo, Eisai, Eli Lilly, Gilead, Idorsia, Janssen, Matsutani Chemical Industry Co., Merck, Novartis, and the Scott R. MacKenzie Foundation.

U.B. has received honoraria from AstraZeneca and Boston Scientific.

D.J.C. has received grant support, paid to his institution, and consulting fees from AstraZeneca, Medtronic, and Abbott Vascular; and has received grant support, paid to his institution, from Boston Scientific.

G.D. has received consulting fees and advisory board fees from AstraZeneca; has received consulting fees from Biosensors; and previously held stock in Medtronic.

J.E. has received consulting and lecture fees from Abbott, Philips, Boston Scientific, and Medtronic; and has received lecture fees from Abiomed, Terumo, and Biosensors.

C.M.G. has received grant support and consulting fees from Angel Medical, Bayer, CSL Behring, Janssen Pharmaceuticals, Johnson & Johnson, and Portola Pharmaceuticals; has received consulting fees from The Medicines Company, Eli Lilly, Gilead Sciences, Novo Nordisk, WebMD, UpToDate Cardiovascular Medicine, Amarin Pharma, Amgen, Boehringer Ingelheim, Chiesi, Merck, PharmaMar, Sanofi, Somahlution, Verreseon, Boston Scientific, Impact Bio, MedImmune, Medtelligence, MicroPort, the PERT Consortium, and GE Healthcare; holds equity in Inference; serves as a chief executive officer of the Baim Institute; and has received grant support, paid to the Baim Institute, from Bristol Myers Squibb and AstraZeneca.

K.H. has received lecture fees from AstraZeneca and Bayer.

S.R.M. has received grant support from and has served on an executive committee and as a site investigator for AstraZeneca.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References

Author notes