OUP user menu

Characterization of dyspnoea in PLATO study patients treated with ticagrelor or clopidogrel and its association with clinical outcomes

Robert F. Storey, Richard C. Becker, Robert A. Harrington, Steen Husted, Stefan K. James, Frank Cools, Philippe Gabriel Steg, Nardev S. Khurmi, Håkan Emanuelsson, Anna Cooper, Richard Cairns, Christopher P. Cannon, Lars Wallentin
DOI: http://dx.doi.org/10.1093/eurheartj/ehr231 2945-2953 First published online: 30 July 2011

Abstract

Aims To describe the incidence of dyspnoea and its associations with demographic characteristics and clinical outcomes in patients with acute coronary syndromes (ACS) treated with ticagrelor or clopidogrel in the PLATelet inhibition and patient Outcomes (PLATO) study.

Methods and results In the PLATO study, 18 624 patients were randomized to receive either clopidogrel [300–600 mg loading dose (LD), 75 mg daily] or ticagrelor (180 mg LD, 90 mg b.i.d.). The occurrence of reported dyspnoea adverse events (AEs) was analysed in the 18 421 patients who received at least one dose of study medication in relation to demographic characteristics, clinical outcomes and other associations of patients with and without dyspnoea. A total of 1339 ticagrelor-treated patients (14.5%) and 798 clopidogrel-treated patients (8.7%) had a dyspnoea AE following randomization, with respectively 39 (0.4%) and 24 (0.3%) classified as severe in intensity. Excluding dyspnoea AEs occurring after the secondary endpoint of myocardial infarction (MI), the yearly rates of the efficacy endpoints in dyspnoea AE patients in the ticagrelor and clopidogrel groups were: for the primary composite of CV death, MI, and stroke, 8.8 and 10.4% (unadjusted P = 0.25; adjusted P = 0.54); for CV death, 3.1 and 4.8% (unadjusted P = 0.024; adjusted P = 0.18); and for total death 3.7 and 6.2% (unadjusted P = 0.004; adjusted P = 0.06), respectively.

Conclusions Ticagrelor-related dyspnoea is usually mild or moderate in intensity and does not appear to be associated with differences concerning any efficacy or safety outcomes with ticagrelor compared with clopidogrel therapy in ACS patients.

  • Platelet
  • Platelet inhibitor
  • Clopidogrel
  • Ticagrelor
  • P2Y12 receptor
  • Coronary artery disease

Introduction

Ticagrelor, an oral, reversibly binding platelet P2Y12 receptor inhibitor, yields greater inhibition of platelet aggregation than clopidogrel1,2 and was shown to reduce the risk of ischaemic events and death compared with clopidogrel in the PLATO study.3,4 Ticagrelor therapy is, however, associated with an increased incidence of episodes of dyspnoea.3,57 The episodes of dyspnoea tend to occur early in the course of ticagrelor treatment, and are usually transient and of mild or moderate severity.7 Despite the more common occurrence of episodes of dyspnoea with ticagrelor when compared with clopidogrel in a prospective study in patients with stable coronary artery disease (CAD), there were no changes of cardiac or pulmonary function measurements.7 Therefore, the mechanisms for the ticagrelor-related dyspnoea remain to be further elucidated. The aim of the current analysis of the PLATO study was to further characterize the nature of this dyspnoea in acute coronary syndromes (ACS) patients and its relations to major clinical outcome events.

Methods

Patient population

The design of the PLATO study has been described previously.3,8 In brief, 18 624 patients with moderate-to-high-risk non-ST elevation ACS [unstable angina or non-ST elevation myocardial infarction (MI)] or ST elevation MI planned for primary percutaneous coronary intervention were randomized to receive either clopidogrel (300–600 mg loading dose (LD), 75 mg daily maintenance dose) or ticagrelor (180 mg LD, 90 mg b.i.d. maintenance dose) for up to 12 months. Patients with asthma, chronic obstructive pulmonary disease and other respiratory diseases were included in the study. A total of 18 421 (98.9%) patients (9235 ticagrelor patients and 9186 clopidogrel patients) received at least one dose of study drug and were considered in this analysis. A total of 6762 (73.2%) patients in the ticagrelor group and 6915 (75.3%) of patients in the clopidogrel group were exposed for >180 days to study treatment, and 3138 (34.0%) and 3184 (34.7%) had >360 days of exposure to the respective study treatments.

Adverse event and serious adverse event recording

The Medical Dictionary for Regulatory Activities (MedDRA) version 11.1 was used for coding adverse events (AEs). A serious adverse event (SAE) was defined as an AE occurring during any study phase that fulfiled one or more of the following criteria: results in death; is immediately life-threatening; requires in-patient hospitalization or prolongation of existing hospitalization; results in persistent or significant disability or incapacity; is a congenital abnormality or birth defect; is an important medical event that may jeopardise the patient or may require medical intervention to prevent one of the outcomes listed above. Adverse events were recorded from enrolment until completion of the last scheduled study visit. Any AEs remaining unresolved at the follow-up visit were recorded as ongoing. The date when the AE started and stopped, maximum intensity, seriousness (‘yes’ or ‘no’), action taken with regard to the study medication and the outcome of the event were recorded for each AE. The intensity was rated according to the following definitions: (i) mild (awareness of sign or symptom, but easily tolerated); (ii) moderate (discomfort sufficient to cause interference with normal activities); (iii) severe (incapacitating, with inability to perform normal activities). The relationship of AEs to study treatment (‘causality’) was assessed by the investigators, who were required to answer ‘yes’ or ‘no’ to the question ‘Do you consider that there is a reasonable possibility that the event may have been caused by any of the following: study medication; other medication?’.

Investigators were informed that dyspnoea had been reported previously with ticagrelor administration and were requested to record occurrences of dyspnoea as AE/SAEs as well as record at enrolment whether there was a history of dyspnoea and/or current dyspnoea. Dyspnoea AEs were identified using one of five terms, namely ‘dyspnoea’, ‘dyspnoea at rest’, ‘dyspnoea exertional’, ‘dyspnoea paroxysmal nocturnal’, and ‘nocturnal dyspnoea’, and all five terms were considered together in this analysis since ‘dyspnoea’ and ‘dyspnoea exertional’ represented over 95% of dyspnoea AEs. Investigators were required to evaluate patients experiencing dyspnoea with regard to underlying cause and classify the suspected aetiology according to the following categories: (i) pulmonary oedema (cardiac, non-cardiac); (ii) other cardiac aetiology (e.g. chronic heart failure); (iii) asthma; (iv) chronic obstructive pulmonary disease; (v) pulmonary vascular disease (pulmonary hypertension, pulmonary embolism); (vi) parenchymal lung disease; (vii) infection (e.g. pneumonia or bronchitis); (viii) metabolic disorder; (ix) anxiety disorder; (x) other known cause; (xi) unexplained. Dyspnoea AEs that were identified by one of the five terms but not recorded in the case report form as a dyspnoea event with suspected aetiology are included as dyspnoea AEs with ‘missing’ suspected aetiology.

Evaluation of clinical outcomes

The definitions used for recording clinical efficacy and safety endpoints in the PLATO study have been previously described and were used in this analysis.3,8 In brief, the primary endpoint was a composite of MI (not including silent MI), stroke, and death from vascular causes. Death from vascular causes, stroke, and total MI (including silent MI) were assessed as secondary endpoints. Bleeding was classified as major or minor according to the PLATO study definitions and frequency of blood transfusions was assessed separately.

Statistical analysis

Statistical methods used for the sample size calculation and endpoint analysis for this trial have been reported previously.3 The data presented here are from the safety population, consisting of subjects that received at least one dose of the study medication. Dyspnoea events occurring both during treatment and following cessation of study medication are included up to the completion of subjects' involvement in the study. Dyspnoea events that were not recorded as resolved at the final study visit were recorded as ‘ongoing’. Data are presented as mean and standard deviation (SD) for normally distributed data and median and inter-quartile range (IQR) for non-normal data. Two-sample comparisons, including those of subject baseline characteristics, were performed using a standard t-test (for normally distributed continuous data), Wilcoxon test (for other continuous data) or χ2 test (for categorical data), as appropriate. Time to event analyses were performed by fitting a Cox proportional hazards survival model, including factors for study treatment and incidence of first dyspnoea event as a time-dependent covariate, with adjustment where indicated to account for differences in relevant clinical variables between treatment groups. Analyses were performed to either include or exclude dyspnoea AEs occurring after the secondary endpoint of MI. Efficacy endpoint rates are quoted as the Kaplan–Meier percentage at 12 months.

Since analysis of the prognostic significance of dyspnoea may be confounded by the fact that there may be overlap between factors contributing to the aetiologies of dyspnoea and clinical events and since most cases of ticagrelor-related dyspnoea occur early after commencing treatment,7 a landmark analysis was performed of clinical outcomes from 31 days onwards in those patients who reported dyspnoea in the first 30 days following randomization, excluding patients who suffered the endpoint of interest in the first 30 days.

Results

Incidence of dyspnoea and associated patient characteristics

At enrolment, ∼20% patients in both the ticagrelor and clopidogrel arms had a history of dyspnoea, with many continuing to experience this symptom, heart failure being the most common underlying suspected aetiology and chronic obstructive pulmonary disease or asthma being the suspected aetiology in approximately one-third of cases (Table 1). Following study drug administration, 1339 ticagrelor-treated patients (14.5%) and 798 clopidogrel-treated patients (8.7%) reported dyspnoea as an AE at any time during follow-up (on and off study drug), with only 39 (0.4%) and 24 (0.3%) cases being judged as severe by the investigator. A greater proportion of the dyspnoea events in the ticagrelor group compared with the clopidogrel group were judged by the investigator to be of unexplained/unknown aetiology (27.3 vs. 20.1%, respectively), while a lesser proportion of these events were judged to be due to heart failure of cardiac aetiology (23.7 vs. 30.8%) (Table 2). The proportions of events attributed to other suspected aetiologies were similar in the two groups with an even distribution of numerical excess in the ticagrelor-treated patients across most of these aetiologies. Patients in both groups reporting dyspnoea were older than those who did not report dyspnoea and were more likely to have chronic renal disease or diabetes mellitus (Table 3). Mean weight was slightly greater in ticagrelor-treated patients with dyspnoea compared with those without, which was not the case in the clopidogrel group although mean waist circumference was higher in both groups for dyspnoea patients. In both treatment groups, patients with dyspnoea during follow-up were more likely to have been experiencing dyspnoea prior to enrolment, often associated with a history of heart failure, chronic obstructive pulmonary disease, and/or asthma, and this association appeared stronger for clopidogrel-treated patients (Table 3). The excess of dyspnoea AEs in the ticagrelor group compared with the clopidogrel group was similar in those reporting dyspnoea prior to enrolment [160/794 (20.2%) vs. 108/733 (14.7%), respectively] compared with those without prior dyspnoea [959/7345 (13.1%) vs. 518/7330 (7.1%), respectively]. Rates of dyspnoea in each group were similar between races or ethnic groups (data not shown).

View this table:
Table 1

History of dyspnoea and suspected aetiology at enrolment according to treatment group

n (%)Ticagrelor, n = 9235Clopidogrel, n = 9186
Dyspnoea prior to enrolment1890 (20.5)1856 (20.2)
 Currently experiencing dyspnoea1295 (14.0)1246 (13.6)
Suspected aetiology
 Heart failure (cardiac aetiology)790 (8.6)803 (8.7)
 Heart failure (non-cardiac aetiology)10 (0.1)19 (0.2)
 Other cardiac aetiology294 (3.2)287 (3.1)
 Asthma125 (1.4)132 (1.4)
 Chronic obstructive pulmonary disease330 (3.6)291 (3.2)
 Pulmonary vascular disease6 (0.1)11 (0.1)
 Parenchymal lung disease7 (0.1)10 (0.1)
 Infection17 (0.2)20 (0.2)
 Metabolic disorder5 (0.1)6 (0.1)
 Anxiety disorder29 (0.3)25 (0.3)
 Other known cause127 (1.4)94 (1.0)
 Unexplained/unknown151 (1.6)158 (1.7)
View this table:
Table 2

Suspected aetiology of dyspnoea reported after randomization according to treatment group

n (%)Ticagrelor, n = 1339Clopidogrel, n = 798
Suspected aetiology (%)
 Unexplained/unknown366 (27.3)160 (20.1)
 Heart failure (cardiac aetiology)318 (23.7)246 (30.8)
 Missing284 (21.2)175 (21.9)
 Other cardiac aetiology132 (9.9)70 (8.8)
 Other known cause115 (8.6)70 (8.8)
 Chronic obstructive pulmonary disease81 (6.0)42 (5.3)
 Anxiety disorder66 (4.9)42 (5.3)
 Infection18 (1.3)11 (1.4)
 Asthma13 (1.0)6 (0.8)
 Heart failure (non-cardiac aetiology)10 (0.7)4 (0.5)
 Parenchymal lung disease5 (0.4)3 (0.4)
 Pulmonary vascular disease1 (0.1)3 (0.4)
 Metabolic disorder3 (0.2)0 (0)
View this table:
Table 3

Characteristics of patients reporting dyspnoea post randomization compared with those not reporting dyspnoea, according to treatment group

Ticagrelor, n = 9235Clopidogrel, n = 9186
Dyspnoea, n = 1339No dyspnoea, n = 7896P-valueDyspnoea, n = 798No dyspnoea, n = 8388P-value
Age, mean years (SD)64.3 (10.8)61.7 (11.2)<0.000164.6 (11.1)62.0 (11.2)<0.0001
Sex, female, n (%)410 (30.6)2224 (28.2)0.066241 (30.2)2362 (28.2)0.221
Race, Caucasian, n (%)1244 (92.9)7235 (91.6)0.115733 (91.9)7687 (91.6)0.836
Weight, kg, mean (SD)82.6 (16.1)80.2 (15.9)<0.000180.5 (16.3)80.3 (15.9)0.739
Waist circumference, cm, mean (SD)101.3 (13.1)98.1 (14.4)<0.000199.8 (13.3)98.5 (14.4)0.0016
History of smoking, n (%)894 (66.7)4779 (60.5)<0.0001510 (63.9)5054 (60.3)0.033
Dyspnoea prior to enrolment, n (%)380 (28.4)1510 (19.1)<0.0001280 (35.1)1576 (18.8)<0.0001
 Current dyspnoea at enrolment252 (18.8)1043 (13.2)<0.0001186 (23.3)1060 (12.6)<0.0001
History of congestive heart failure, n (%)164 (12.2)845 (10.7)0.094129 (16.2)923 (11.0)<0.0001
History of asthma, n (%)63 (4.7)206 (2.6)<0.000133 (4.1)233 (2.8)0.029
History of chronic obstructive pulmonary disease, n (%)134 (10.0)442 (5.6)<0.000187 (10.9)462 (5.5)<0.0001
History of diabetes mellitus, n (%)368 (27.5)1937 (24.5)0.021211 (26.4)2105 (25.1)0.403
History of chronic renal disease, n (%)68 (5.1)305 (3.9)0.03746 (5.8)356 (4.2)0.045

Patients with dyspnoea in the ticagrelor group were more likely to have onset of this symptom within 7 days compared with dyspnoea patients in the clopidogrel group and the median time to onset of dyspnoea was 23 vs. 43 days, respectively (P < 0.0001, Figure 1). Fifteen per cent of the dyspnoea cases in the ticagrelor group were categorized by the investigator in the case report form as being causally related to study medication compared with only 6.9% of cases in the clopidogrel group (P < 0.0001); 26.2% of the ticagrelor patients with dyspnoea prematurely discontinued study medication for any reason compared with 22.3% of clopidogrel patients with dyspnoea (P = 0.043), the excess in the ticagrelor group being attributable to discontinuation due to the dyspnoea: 79/1339 (5.9%) and 13/798 (1.6%), respectively, discontinued study medication due to dyspnoea (P < 0.0001) representing 0.9 and 0.1% of treated patients, respectively. At the end of the study, ∼30 days following discontinuation of study medication, 463 ticagrelor patients (5.0%) and 284 clopidogrel patients (3.1%) were recorded as having ongoing dyspnoea (P < 0.0001). Despite discontinuation of study medication in some patients with dyspnoea, the exposures to study medication and durations of follow-up were similar in those with and without dyspnoea AEs both within each treatment group and between these groups (see Supplementary material online).

Figure 1

Landmark analyses of the incidence of new dyspnoea AE post randomization (investigational product, IP) showing (A) onset of any dyspnoea AE in the first 30 days; (B) onset of any dyspnoea AE from 31 days onwards; (C) onset of a dyspnoea AE judged to be unexplained or unknown aetiology in the first 30 days; and (D) onset of a dyspnoea AE judged to be unexplained or unknown aetiology from 31 days onwards.

Clinical outcomes in patients reporting dyspnoea

The incidence of the primary composite endpoint in patients reporting dyspnoea was greater compared with those not reporting dyspnoea in both ticagrelor and clopidogrel groups, primarily due to a higher incidence of MI (Table 4). This is explained in part by the occurrence of dyspnoea following the secondary endpoint of MI (see Supplementary material online). Exclusion of patients with dyspnoea AEs occurring after MI showed favourable trends for ticagrelor compared with clopidogrel, particularly for mortality, that were consistent with the overall trial results (Table 5). The incidence of bleeding was numerically higher in patients reporting dyspnoea compared with those without dyspnoea in both groups but the differences in major bleeding were not significant after adjusting for relevant clinical variables (Table 4).

View this table:
Table 4

Clinical outcomes of patients reporting dyspnoea post randomization compared with those not reporting dyspnoea, according to treatment group

n (K–M %)Ticagrelor, n = 9235Clopidogrel, n = 9186
Dyspnoea, n= 1339No dyspnoea, n = 7896P-valueDyspnoea, n = 798No dyspnoea, n = 8388P-value
Primary composite endpoint (%)151 (11.9)701 (9.4)<0.001117 (15.7)882 (11.2)0.008
Myocardial infarction (%)112 (8.7)393 (5.4)0.00883 (11.3)515 (6.6)0.173
Stroke (%)21 (1.7)102 (1.4)0.4239 (1.3)95 (1.2)0.278
CV death (%)39 (3.3)306 (4.1)<0.00137 (4.8)391 (5.0)0.036
Total mortality (%)47 (3.9)342 (4.6)<0.00148 (6.4)443 (5.7)0.007
Major bleed (%)164 (13.7)797 (11.2)0.59196 (13.5)833 (11.0)0.436
Major or minor bleed (%)256 (21.4)1083 (13.7)0.117136 (18.8)1079 (14.2)0.032
  • P-values are from a Cox proportional hazards model with explanatory variables for treatment group, occurrence of first dyspnoea event (as a time-dependent covariate) and treatment-dyspnoea interaction, and adjusted for age, weight, diabetes mellitus, history of congestive heart failure, smoking status, history of chronic renal disease, and prior dyspnoea; all hazard ratios for Dyspnoea vs. No dyspnoea are >1.

  • K–M, Kaplan–Meier.

View this table:
Table 5

Clinical outcomes of patients reporting dyspnoea post randomization according to treatment group, excluding patients with dyspnoea AE following the secondary endpoint of non-fatal myocardial infarction

Ticagrelor, n = 1293Clopidogrel, n = 753Unadjusted HR (95% CI)P-valueAdjusted HR (95% CI)P-value
Patients with events (K–M %)Patients with events (K–M %)
Primary composite endpoint 106 (8.8)72 (10.4)0.84 (0.62, 1.13)0.2540.91 (0.67, 1.23)0.542
Myocardial infarction66 (5.4)38 (5.7)0.99 (0.66, 1.47)0.9561.01 (0.68, 1.52)0.945
Stroke19 (1.6)7 (1.0)1.55 (0.65, 3.69)0.3201.62 (0.68, 3.87)0.279
CV death36 (3.1)35 (4.8)0.59 (0.37, 0.93)0.0240.72 (0.45, 1.16)0.179
Total mortality43 (3.7)45 (6.2)0.54 (0.36, 0.83)0.0040.67 (0.44, 1.02)0.060
Major bleed151 (13.1)87 (12.6)1.01 (0.78, 1.32)0.9151.03 (0.79, 1.35)0.802
Major or minor bleed239 (20.7)126 (18.4)1.12 (0.90, 1.39)0.2931.14 (0.91, 1.41)0.252
  • K–M, Kaplan–Meier; HR, hazard ratio. HRs and P-values are from a Cox proportional hazards model with an explanatory variable for treatment group, and, where indicated, adjusted for age, weight, diabetes mellitus, history of CHF, smoking status, history of chronic renal disease and prior dyspnoea.

Landmark analyses of clinical outcomes from 31 days onwards for those who reported a dyspnoea AE in the first 30 days were performed to determine whether ticagrelor-related dyspnoea appears to be associated with reduced efficacy of ticagrelor compared with clopidogrel (Table 6, Figure 2 and Supplementary material online). With the caveat that the causes of dyspnoea are different between the treatment groups, we found that rates of MI after 30 days were similar in the ticagrelor- and clopidogrel-treated patients with these early dyspnoea events, while CV and total mortality rates were substantially lower in the ticagrelor-treated patients. While a dyspnoea AE in the first 30 days was associated with greater subsequent CV and total mortality in the clopidogrel group, this was not the case for the ticagrelor group. History of congestive heart failure was more frequent in the clopidogrel group compared with the ticagrelor group (16.9 vs. 11.8%; P = 0.022) but the difference in mortality remained significant after adjusting for important baseline clinical variables (Table 6). In addition, the estimated yearly rate of the primary composite endpoint from 31 days was numerically lower (6.2%) in ticagrelor-treated patients who had dyspnoea of unknown or unexplained aetiology in the first 30 days compared with the whole cohort of ticagrelor-treated patients with early dyspnoea events (8.4%).

View this table:
Table 6

Clinical outcomes from 31 to 360 days for patients reporting dyspnoea in the first 30 days post randomization according to treatment group

TicagrelorClopidogrelAdjusted HR (95% CI)P-value
nPatient with events, n (K–M %)nPatient with events, n (K–M %)
Primary composite endpoint 31–360 days69753 (8.4)31434 (11.7)0.72 (0.47, 1.11)0.1385
Myocardial infarction 31–360 days70137 (5.9)31415 (5.2)1.04 (0.57, 1.91)0.8926
Stroke 31–360 days7227 (1.1)3315 (1.7)0.67 (0.21, 2.14)0.4981
CV death 31–360 days72616 (2.6)33121 (7.0)0.43 (0.22, 0.83)0.0117
Total mortality 31–360 days72619 (3.0)33126 (8.5)0.40 (0.22, 0.73)0.0027
Major bleed 31–360 days61432 (6.4)27212 (5.0)1.19 (0.62, 2.32)0.599
 Non-CABG-related major bleed63921 (4.0)2887 (2.8)1.37 (0.58, 3.22)0.471
 CABG-related major bleed62214 (2.9)2786 (2.4)1.05 (0.40, 2.72)0.926
  • K–M, Kaplan–Meier; HR, hazard ratio. HRs and P-values are from a Cox proportional hazards model with an explanatory variable for treatment group, and adjusted for age, weight, diabetes mellitus, history of CHF, smoking status, history of chronic renal disease and prior dyspnoea.

Figure 2

K–M curves for events between 31 and 360 days in (A and C) ticagrelor-treated patients or (B and D) clopidogrel-treated patients with or without onset of dyspnoea up to 30 days post randomization showing (A, B) primary composite endpoint; and (C, D) total death.

Other analyses of associations with dyspnoea events

In patients reporting dyspnoea after randomization, 1062 ticagrelor-treated and 633 clopidogrel-treated patients had serum NT-pro-BNP levels available at randomization and these levels were numerically lower in the ticagrelor patients (224 ± 480 vs. 310 ± 591 pmol/L), with similar trends seen in patients with dyspnoea attributed to unexplained/unknown aetiology (156 ± 312 vs. 222 ± 391 pmol/L) or heart failure of cardiac aetiology (327 ± 673 vs. 487 ± 836 pmol/L). These trends towards lower serum NT-pro-BNP levels in the ticagrelor patients reporting dyspnoea were maintained at subsequent follow-up visits (data not shown). Heart failure was reported as an AE in 470 (5.1%) and 483 (5.3%) patients in the ticagrelor and clopidogrel groups, respectively.

Of patients reporting dyspnoea after randomization, 483/1339 (36.1%) ticagrelor-treated patients and 300/798 (37.6%) clopidogrel-treated patients were initiated on a loop diuretic compared with 1683/7896 (21.3%) ticagrelor-treated and 1750/8388 (20.9%) clopidogrel-treated patients without dyspnoea after randomization, indicating a small numerical excess of diuretic prescription (a total of 2166 vs. 2050 patients) associated with the excess dyspnoea in the ticagrelor group.

In patients enrolled in the Holter monitoring substudy, ventricular pauses more than 3 s were not more common in ticagrelor-treated patients with dyspnoea compared with those without dyspnoea: 89 out of 1472 ticagrelor-treated patients had pauses >3 s of whom 11/89 (12.4%) reported a dyspnoea AE compared with 267/1383 (19.3%) patients who reported a dyspnoea AE but did not have pauses >3 s. 62 out of 1436 clopidogrel-treated patients had pauses >3 s of whom 6/62 (9.7%) reported a dyspnoea AE compared with 144/1374 (10.5%) who reported a dyspnoea AE but did not have such pauses.

Plasma ticagrelor levels at Day 4 or hospital discharge was available for 965 ticagrelor-treated patients with dyspnoea and 5261 ticagrelor-treated patients without dyspnoea post randomization. Median (IQR) plasma levels were no different in those with dyspnoea [423 (256–664) ng/mL] vs. those without dyspnoea [417 (245–642) ng/mL; P = 0.35], with no trends evident for any of the various attributed aetiologies. However, plasma ticagrelor levels were higher in those patients developing dyspnoea in the first 7 days of treatment [455 (272–771) ng/mL] compared with those who did not have any dyspnoea AE at any time (P = 0.0156) and were also higher than in those who had a first dyspnoea AE more than 7 days after randomization [399 (250–632) ng/mL; P = 0.026].

Discussion

In this analysis of the PLATO study, we have confirmed that there is an excess of dyspnoea associated with ticagrelor compared with clopidogrel, with much of this excess occurring early in the first month of treatment but a small excess also arising later in the course of treatment. This dyspnoea is mostly mild or moderate in intensity and may sometimes be recognized as being related to the medication. While dyspnoea may occasionally lead to discontinuation of therapy, the discontinuation rate due to dyspnoea in all ticagrelor-treated patients is low (0.9%). Patients with a prior history of congestive heart failure, chronic obstructive pulmonary disease, or other causes of dyspnoea do not appear to be more likely to develop ticagrelor-related dyspnoea compared with those without a history of these conditions.

The substantially lower mortality rates in ticagrelor-treated patients with dyspnoea compared with clopidogrel-treated patients with dyspnoea are consistent with two effects: first, the treatment benefit of ticagrelor compared with clopidogrel seen in the overall trial appears to be preserved in ticagrelor-treated patients with dyspnoea compared with those without dyspnoea; and, second, there is a favourable mortality prognosis in patients with ticagrelor-related dyspnoea compared with other aetiologies of dyspnoea. This conclusion is reinforced by the analysis of outcomes after 1 month in patients developing dyspnoea in the first 30 days after initiating study medication, for whom there was a 66% lower risk of CV death with ticagrelor compared with clopidogrel. This difference in CV death is much >21% CV mortality risk reduction seen with ticagrelor in the trial overall and suggests that patients with dyspnoea related only to ticagrelor do not have the increased mortality risk associated with the various pathological causes of dyspnoea occurring in both the clopidogrel- and ticagrelor-treated patients.3 Indeed our landmark analyses show how a dyspnoea AE in the first 30 days is associated with a much greater risk of subsequent death in clopidogrel-treated patients compared with those who do not report a dyspnoea AE. The fact that there was no evidence that the mortality benefit associated with ticagrelor in the whole trial was attenuated in the subgroup of patients with a dyspnoea AE suggests that patients with tolerable dyspnoea should be encouraged to continue ticagrelor, while patients who cannot tolerate dyspnoea that is believed to be an adverse effect of ticagrelor may be switched to either clopidogrel or prasugrel.

The early onset of ticagrelor-related dyspnoea has been previously characterized in patients with stable CAD, with the majority of cases occurring within 1 week of commencing ticagrelor, and this may help to differentiate it from dyspnoea of other aetiologies that occurs in both ticagrelor-treated and clopidogrel-treated patients. Inevitably in the PLATO ACS population compared with stable CAD populations, there are more cases of dyspnoea arising early in the course of treatment due to onset of heart failure following MI, which complicates the analysis performed here and necessitates the making of some assumptions in determining what influence ticagrelor-related dyspnoea has on clinical outcomes. However, the data presented here at least provide reassurance that there does not appear to be any loss of treatment effect in those ticagrelor-treated patients who develop dyspnoea as a side effect. It is particularly encouraging that the ticagrelor-treated patients with dyspnoea had similar mortality rates to those without dyspnoea despite having overall higher risk characteristics including greater mean age and higher incidence of diabetes mellitus and chronic renal disease. These higher risk features may partly explain the higher rates of bleeding seen in patients with dyspnoea in addition to the fact that bleeding itself may contribute to the development of dyspnoea through various mechanisms including ischaemia, anaemia, and development of heart failure.

While the proportion of dyspnoea cases attributed to heart failure in the ticagrelor group was lower than the proportion in the clopidogrel group, there was a numerical excess of dyspnoea cases attributed to heart failure in the ticagrelor group. Previously, it has been demonstrated that ticagrelor-related dyspnoea is not associated with any increase in serum NT-pro-BNP level or change in left ventricular systolic function.7 The nominally lower NT-pro-BNP levels in ticagrelor-treated patients with dyspnoea compared with clopidogrel-treated patients with dyspnoea are consistent with this prior observation and suggest that measurement of serum NT-pro-BNP may help to differentiate ticagrelor-related dyspnoea from heart failure and avoid unnecessary prescription of loop diuretics. Chest X-ray, echocardiography and pulmonary function tests may also help in differentiating ticagrelor-related dyspnoea from pathological causes. While ticagrelor-related dyspnoea is generally not associated with bradycardia or any other change in heart rhythm,7 it is recognized that another side-effect of ticagrelor is increased propensity to ventricular pauses, particularly nocturnal sinoatrial pauses. We found no increased incidence of ventricular pauses in ticagrelor-treated Holter substudy patients with dyspnoea compared with those without dyspnoea but this increased propensity to nocturnal bradycardia may still warrant consideration in infrequent cases of nocturnal waking with dyspnoea.

In the DISPERSE-2 study in ACS patients, there was evidence that the incidence of ticagrelor-related dyspnoea is dose-dependent, suggesting that plasma ticagrelor levels influence whether or not patients develop this adverse effect.6 This was supported by our observation that ticagrelor-treated patients developing dyspnoea during the first 7 days of treatment had modestly higher plasma levels at Day 4 or hospital discharge than those who either did not develop dyspnoea or developed dyspnoea at a later time point. However, the modest nature of this association and the lack of any strong association between dyspnoea and clinical characteristics in the present analysis suggest that factors, apart from ticagrelor plasma levels, that influence the probability of ticagrelor-related dyspnoea remain to be determined.

While the majority of dyspnoea AEs had resolved by the end of the study in both treatment groups, the reasons why more ticagrelor-treated patients compared with clopidogrel-treated patients had an unresolved dyspnoea AE at the end of the study despite discontinuation of study medication are unclear. Previously it has been shown that ticagrelor-treated dyspnoea, if sustained during treatment, tends to resolve after the medication is discontinued and there has been no evidence of any compromise to pulmonary or cardiac function that would lead to sustained dyspnoea after drug discontinuation.7

Currently the mechanism for ticagrelor-related dyspnoea is unknown. It is however recognized that adenosine infusion can cause dyspnoea, in the absence of bronchospasm,9 and that ticagrelor inhibits adenosine re-uptake and can affect adenosine-mediated increases in blood flow in preclinical models.10 Thus, one hypothesis is that ticagrelor-related dyspnoea is due to increased levels of extracellular adenosine levels. Further work is required to explore this hypothesis and search for other potential mechanisms.

Limitations

This was a post hoc analysis of subgroups of patients defined by AE reporting of dyspnoea with the ticagrelor-treated patients having an additional cause for dyspnoea, namely a drug-related AE, compared with clopidogrel-treated patients and so the clinical outcomes data should be viewed with caution. The excess of dyspnoea AEs in the ticagrelor group inevitably led to an imbalance in the numbers of dyspnoea AE patients in the two treatment groups for the landmark and other analyses. Despite adjustment for differences in baseline clinical characteristics, we cannot exclude confounding variables leading to differences in event rates in dyspnoea AE patients in the two treatment groups and our analyses should therefore be considered observational. Investigations for establishing the aetiology of dyspnoea were not protocol-mandated and were at the investigators' discretion.

In conclusion, ticagrelor-related dyspnoea is usually mild or moderate in intensity, resolves spontaneously or upon discontinuation of medication in the majority of patients and does not appear to be associated with any differences in any efficacy or other safety outcomes compared with clopidogrel therapy in ACS patients.

Funding

This work was supported by AstraZeneca.

Conflict of interest: R.F.S.: research/educational grants from AstraZeneca, Dynabyte, Eli Lilly/Daiichi Sankyo alliance, Schering-Plough/Merck, and Accumetrics; honoraria from AstraZeneca, Eli Lilly/Daiichi Sankyo alliance, Medscape, Novartis, GlaxoSmithKline, and Schering-Plough/Merck; consultant fees from AstraZeneca, Eli Lilly/Daiichi Sankyo alliance, Merck, Novartis, Sanofi-Aventis/Bristol-Myers Squibb, Eisai. The Medicines Company and Accumetrics; travel support from AstraZeneca, Eli Lilly/Daiichi Sankyo alliance, and Schering-Plough/Merck. R.C.B.: research grants from AstraZeneca, Regado Biosciences, Johnson and Johnson, Bayer, and Bristol-Myers Squibb; consultant fees from Daiichi Sankyo, Boehringer Ingelheim, and AstraZeneca. R.A.H.: consulting fees from AstraZeneca, Bristol-Myers Squibb, Schering Plough/Merck, Novartis, Portola, Sanofi-Aventis, and The Medicines Company; honoraria/lecture fees from Eli Lilly, AstraZeneca; grant support from AstraZeneca, Bristol-Myers Squibb, Portola, Schering-Plough/Merck. S.H.: research grants from AstraZeneca, Bristol-Myers Squibb, Pfizer, and Bayer; consultant fees from Sanofi-Aventis, Pfizer, and AstraZeneca. S.K.J.: research grants from AstraZeneca, Bristol-Myers Squibb, and Eli Lilly; honoraria and advisory board compensation from AstraZeneca, Merck, Sanofi-Aventis and Eli Lilly. P.G.S.: research grants from Servier; consultant fees/advisory board membership; Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi/Sankyo/Eli Lilly alliance, GlaxoSmithKline, Medtronic, Merck, Otsuka Pharmaceutical, Roche, Sanofi-Aventis, Servier, and The Medicines Company; and having equity ownership in Aterovax. N.S.K.and H.E.: employees of AstraZeneca and having equity ownerships in AstraZeneca. F.C. and A.C.: no conflicts of interest declared. C.P.C.: research grants from Accumetrics, AstraZeneca, GlaxoSmithKline, Intekrin Therapeutics, Merck, and Takeda; advisory board participation with Alnylam, Novartis and Bristol-Myers Squibb/Sanofi (but funds donated to charity); being a clinical advisor and having equity ownership interest in Automedics. L.W.: research grants from AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, and Schering-Plough; honoraria from AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, Schering-Plough, and Eli Lilly; consultant fees from Regado Biotechnologies, Athera Biotechnologies, Boehringer Ingelheim, AstraZeneca, GlaxoSmithKline, and Eli Lilly; lecture fees from AstraZeneca, Boehringer Ingelheim, and Eli Lilly.

Acknowledgements

We are grateful to Emily de Looze of Gardiner Caldwell Communications for her support in preparing the figures, funded by AstraZeneca.

References

View Abstract