TicAgrelor Versus CLOpidogrel in Stabilized Patients With Acute Myocardial Infarction: TALOS-AMI

1. Study design Pospective, multi-center, randomized, open trial
2. Study Objective To investigate the efficacy and safety of switching from ticagrelor to clopidogrel in stabilized patients with AMI with no adverse events during the first month after an index PCI
3. Study Drug Test drug: Clopidogrel Control drug: Ticagrelor
4. Study Disease Acute myocardial infarction: ST-segment elevation MI (STEMI) and non-ST-segment elevation MI (NSTEMI)
5. Study Population 2590 (loss to follow-up: 10 %) -Test group: 1295 -Control group: 1295

6. SAMPLE SIZE CALCULATIONS

   The present study is designed to show non-inferiority for the treatment group with aspirin plus clopidogrel versus the control group with aspirin plus ticagrelor. The sample size is based on the combined occurrence rate of ischaemic and bleeding events between 1 and 12 months after AMI. According to the PLATO investigators, the event rate of the primary efficacy endpoint including CV death, MI or stroke was 5.28% in the ticagrelor group and 6.60% in the clopidogrel group between 1 and 12 months after the index event1. There were no reported data on the bleeding event rate associated with ticagrelor from 1 to 12 months after AMI, especially the BARC bleeding rate at the time of the present study design. Therefore, we assumed the event rate of BARC 2, 3 or 5 bleeding from the event rates of non-coronary artery bypass graft (CABG)-related PLATO major or minor bleeding because the content of BARC 2, 3 or 5 bleeding was conceptually quite similar with non-CABG-related PLATO major or minor bleeding. With regard to the assumption of the bleeding rate from 1 to 12 months after the index PCI, we calculated the expected bleeding rate using a proportional equation under the assumption that the occurrence ratio of non-CABG-related major bleeding of the first 30 days to that of after 30 days and the ratio of non- CABG-related major or minor bleeding of the first 30 days to that of after 30 days could be equal. Thus, for the event rate of BARC 2, 3 or 5 bleeding associated with ticagrelor from 1 to 12 months after AMI, the event rate was assumed from the event rates of non-CABG-related PLATO major or minor bleeding during a year of ticagrelor therapy (8.7%) and non-CABG-related major bleeding of the first 30 days (2.47%) and after 30 days (2.17%) in the PLATO trial. For the event rate of BARC 2, 3 or 5 bleeding associated with clopidogrel from 1 to 12 months after AMI, the event rate was assumed from the event rates of non- CABG-related PLATO major or minor bleeding during a year of clopidogrel therapy (7.0%) and non-CABG-related major bleeding of the first 30 days (2.21%) and after 30 days (1.65%) in the clopidogrel group of the PLATO trial7. After applying a proportional equation, we estimated that the BARC 2, 3 or 5 bleeding would be 4.07% in the ticagrelor group and 2.99% in the clopidogrel group. Thus, the expected event rate of the primary endpoint from 1 to 12 months after the index PCI was 9.35% (ischaemic event of 5.28% + bleeding event of 4.07%) in the ticagrelor group and 9.59% (ischaemic event of 6.6% + bleeding event of 2.99%) in the clopidogrel group. We chose the non-inferiority margin in accordance with clinical judgement and other relevant studies with a non-inferiority design for the present study design. The non-inferiority margin of two contemporary trials of antiplatelet treatment after PCI that were available up to that time was equivalent to a 40% increase in the expected event rate16,17. The steering committee decided that the non-inferiority margin in our study should be less than a 40% increase compared to the expected event rate of the control group. After considering clinically acceptable relevance and the feasibility of study recruitment, we finally selected the non-inferiority margin of 3.0%, which was equivalent to a 32% increase in the expected event rate. Sample size calculations (PASS 13; NCSS, LLC, Kaysville, UT, USA) were initially performed based on a one-sided α of 0.025 and a power of 80%. To achieve these goals, a total of 2,230 patients was needed. After considering a follow-up loss rate of 10%, there should be at least 1,644 per group and a total of 3,288 patients. However, while the study was actively underway, the government policy on investigator-initiated trials (ITTs) changed: it was decided not to allow national health insurance to charge for the medical care costs of participating patients in ITTs. Soon after, the government was forced to permit the application of national health insurance to ITTs on the condition that researchers should obtain approval for their studies by the head of the Health Insurance Review and Assessment service. However, during this turmoil, researchers' willingness to register patients had been compromised, and they could not register patients as planned within the period. Thus, the steering committee held an emergency meeting with data and safety monitoring board (DSMB) members and independent statisticians and decided to recalculate the sample size for the timely completion of the trial. In recalculating the sample size, we adopted a one-sided α of 0.05 instead of 0.025. According to the CONSORT statement of non-inferiority and equivalence in trials18, a one-sided α of 0.05 was acceptable for the non-inferiority clinical trials. Moreover, in the large-scale TROPICAL-ACS (Testing Responsiveness to Platelet Inhibition Chronic Antiplatelet Treatment for ACS) trial (one of the famous CV drug trials similar to the TALOS-AMI trial), the researchers adopted a one-sided α of 0.05 for the sample size calculation4. Furthermore, of 110 CV non-inferiority trials published in JAMA, The Lancet or the New England Journal of Medicine from 1990 to 2016, a one-sided α was 0.05 in 66 trials19. Based on this external harsh environment and a review of the sample size calculation in previous large-scale randomised trials published in high-impact journals, we recalculated the sample size by using a one-sided α of 0.05, a power of 80% and a follow-up loss rate of 10%. As such, the sample size was reduced from 3,288 to 2,590 (1,295 patients in each group)

7. Study Design

1. Screening period

To conduct screening AMI patients based on the inclusion/exclusion criteria who (1) have been treated with ticagrelor+aspirin for at least 30±7 days after an index PCI, (2) received full explanation of the study details, (3) given informed consent To randomize eligible subjects within 30±7 days after AMI undergoing PCI with newer generation DES, and receiving aspirin and ticagrelor to the treatment and control groups in a 1:1 ratio.

2)Treatment period

Enrolled patients receive clopidogrel 75mg + aspirin 100 mg (treatment group) or ticagrelor 90mg bid +aspirin 100mg treatment (control group) for 11 months (post-AMI 1 month to 12 months) and evaluation safety and efficacy by conducting physical examination, checking vital sign, and collecting adverse events at post-PCI 3M, 6M, 12M visits.

Labboratory and imaging tests, which undergo according to the medical judgment of each investigator during the study period, are collected by reviewing medical records or EMR.

8. Randomization

1. Subject Assignment and Randomization

   Randomization will be performed to ensure the scientific validity of the clinical test. This will maximize the comparability of the test and control group and eliminate the subjectivity of the researchers in subject group assignment. Before PCI, a 250-325mg loading dose of aspirin is given to patients who are naïve to treatment and all patients receive a loading dose of ticagrelor 180mg. Discharge medication consists of aspirin 100mg once and ticagrelor 90mg twice per day. All patients receive treatment with aspirin plus ticagrelor for 1 month after the index PCI (screening period). At 30 ± 7days after index PCI, eligible patients were randomly assigned either to the 1) aspirin 100 mg plus clopidogrel 75mg daily (treatment group) or 2) aspirin 100 mg plus ticagrelor 90mg twice daily (control group) in a 1:1 ratio. Randomization will occur centrally. To randomize a patient, the investigative site will enter the subject into the designated electronic system and obtain the treatment assignment (clopidogrel + aspirin or ticagrelor + aspirin) in a 1:1 ratio. At 1 month visit after AMI, eligible subjects were assigned to each treatment group following an access to the interactive web-based response system (IWRS, Medical Excellence Inc., Seoul, Korea) by the investigator or designee. Randomization sequence was created by an independent statistician using SAS 9.3 (SAS Institute Inc. Cary, NC, USA) statistical software and was stratified by study center and type of AMI (STEMI or NSTEMI) and with a 1:1 allocation using hidden random block size.

2. Dosage and Method Test (Clopidogrel): 75mg oral administration, once a day Control (Ticagrelor): 1 tablet (90mg) oral administration, twice a day
3. Switching protocol (ticagrelor to clopidogrel) In the control treatment group, when switching from ticagrelor to clopidogrel, patients take a 75mg clopidogrel without loading dose at the time of the next scheduled dose after the final dose of ticagrelor (eg, ≈12 hours from last dose of ticagrelor). The steering committee decided this switching strategy of no loading dose based on the concept that our study population would be at stable status at the time point of randomization (30 days after index PCI). The data safety and monitoring board (DSMB) approved this switching strategy on the condition that initial 100 enrolled patients in the treatment group should be monitored daily during first 7days for the occurrence of adverse clinical events by telephone interviews. Thereafter, DSMB reviewed the clinical data of the initial 100 patients in the treatment group and recommended continuation of the study according to the original protocol. After randomization, patients continue the same medication for 11 months according to their group allocation (treatment period, Figure 1). Patients are evaluated at 3 (2 months after randomization), 6 (5 months after randomization), and 12 (11months after randomization) months after index PCI and monitored for the occurrence of the clinical events.

9. Statistical Analysis

1. Efficacy Test Variable Analysis

1. Primary endpoint analysis Efficacy Test The non-inferiority test between 1 and 12 months after AMI will be based on the Kaplan-Meier estimates. A 95% two-sided confidence interval will be computed for the difference event rate (clopidogrel + aspirin) - event rate (ticagrelor + aspirin). The clopidogrel group will be judged as non-inferior to the ticagrelor if the upper confidence limit is less than the predetermined non-inferiority margin of 3% (absolute risk difference).

   The hypothesis of non-inferiority test will be based on the difference of proportions. Let rT denote the true event proportion in the test arm (clopidogrel + aspirin) between 1 and 12 months, and rC denote the true event proportion in the control arm (ticagrelor + aspirin) between 1 and 12 months. The hypotheses are H0: rT - rC≥Δ HA: rT - rC\<Δ TheΔis the non-inferiority margin, and is taken to be 0.03. The test will be performed as a one-sided test at alpha=0.05.

   The null hypothesis shall be rejected at alpha=0.05 if the one-sided p-value is less than 0.05. When this occurs, the upper limit of the two-sided 95% confidence interval will be less than 3%.

   If the non-inferiority analysis passed the acceptance criterion, a superiority analysis will be performed. Statistical superiority is achieved when the upper limit of the two-sided 95% confidence interval of the risk difference is less than 0%. The type I error for this analysis is protected by the non-inferiority analysis, and no alpha adjustment would be appropriate

   Subgroup analyses will be performed by the primary endpoint categorized by type of AMI (STEMI vs NSTEMI), Subgroup analyses will be performed by the primary endpoint categorized by Type of AMI (STEMI vs NTEMI), Gender, Age (>=75 vs \<75), Diabetes, LVEF (>=40% vs \<40%), eGFR (>=60 vs \<60), type of implanted stents (Xience vs Resolute vs Synergy stents), Bleeding risk according to the ARC criteria (high vs low bleeding risk), CYP2C19 loss-of-function carrier status (carrier vs non-carrier).

   The primary analysis population for primary and secondary endpoints will be the Intention-to-Treat (ITT) population. The primary endpoint analysis will also be performed on the Per Protocol (PP) population as subsequent analysis.

   A primary endpoint analysis stratified by the institutions as a sensitivity analysis. Strata will be divided by the accrual number of institution based on quartiles.

   Implement noninferiority validation based on the tolerance limit after collecting cumulative occurrence rate of MACCE (CV death, MI, stroke) + BARC bleeding (type 2, 3, or 5) post 1M-1Y PCI and checking 95% confidence interval of [Ticagrelor occurrence rate - Clopidogrel occurrence rate]. If the upper value of the 95% confidence interval is less than 3% of the noninferiority tolerance limit, Clopidogrel is perceived noninferior to Tricagrelor. Present the cumulative limit method, Kaplan-Meier curve and conduct log-rank test to check the difference between two groups.

2. Main Secondary Endpoint Analyses

The secondary endpoints will be composed of two families. The first family consists of the composite endpoint of MACCE (CV death, MI, stroke) plus BARC bleeding (type 2, 3, or 5). The second family will consist of MACCE plus BARC bleeding (type3, or 5), MACCE, and BARC bleeding (type 2, 3, or 5). The endpoints from the second family will be tested hierarchically, thereby maintaining the study-wise alpha level. These secondary endpoints will only be tested if both the primary composite endpoint and BARC bleeding are significant at non-inferiority analysis, and superiority analysis. Composite endpoint of MACCE plus BARC bleeding (type 3, or 5) will be tested first, and only if this is significant, the composite endpoint of MACCE only will be tested afterwards. BARC bleeding (type 2, 3, or 5) will be tested only if both of the above endpoints are tested significant.

10. Analysis Population

1. The Intent to Treat (ITT) Population

   The ITT population is defined as all randomized patients at 1 month after AMI, regardless of their adherence with the entry criteria, regardless of treatment they actually received, and regardless of subsequent withdrawal from treatment or deviation from the protocol19. Only some specific reasons that might cause an exclusion of a patient from the ITT population:

   No treatment was applied at all No data are available after randomization

2. The Per Protocol (PP) Population

   The PP population is the subset of ITT population consisting of all patients who receive and retain the treatment during 12 months after PCI19. Some specific reasons that might cause an exclusion of a patient from the PP population:

   Violation of entry criteria including inclusion and exclusion criteria Withdrawal of consent Concomitant treatment of oral anticoagulant agent (vitamin-K antagonists or novel oral anticoagulants such as dabigatran, rivaroxaban, apixaban, or edoxaban) during the study period Poor compliance -Conversion from ticagrelor + aspirin to clopidogrel + aspirin during RCT procedure and vice versa

   • Discontinuation of test or control drugs for 7 days or longer
3. * In the cases of withdrawal of consent, concomitant treatment of oral anticoagulation agent and poor compliance, their data will be used for statistical analyses until such events occur.

11. Final enrollment number

The initially planned enrollment number of patients was 2,590, however, the actual number of registered patient was larger than this, which was 2,697. The randomization of the current trial was performed using interactive web-based response system and patients were competitively enrolled in 32 institutions. Despite the competitive registration among participating institutions, the random system did not have a lock on it, resulting in a registration of 107 more patients than initially planned number. When DSMB closely monitored the status of additionally registered subjects, its members agreed that there were no safety issues to them and they got an approval of IRB for using the data of additionally registered patients for the statistical analyses.