Trial design
This study is a multicentre, large sample, randomized controlled study with a 20 months study case recruitment, and an one-year post enrollment follow-up. Written informed consent is required for all participants. The trial is registered in the Chinese Clinical Trial Registry (ChiCTR1900026492).
The study is being conducted at 14 comprehensive first-class hospitals in Jiangxi Province. The participants are voluntary and randomly assigned to the control and intervention groups. Patients in the intervention group are tested for CYP2C19 genotypes and urine 11-dhTxB2. Finally, according to the test results, individualized medications of clopidogrel or aspirin are developed for each patient in the intervention group. We will not intervene in the clinical medication of patients in the control group. Doctors will prescribe antiplatelet medication based on the patient's condition and American Heart Association guidelines[4].
Participants and recruitment
Inpatients are screen by a researcher in the neurology department of the participating hospital. A researcher will determine eligible patients then explain the study in detail. Participants must sign the informed consent form and write down 1-2 contact numbers when signing the informed consent form to ensure that they can be dialed. Repeated contact will be given to patients who have lost to follow-up. Besides, there is a corresponding record for participants who interrupt or deviate from the intervention program. The algorithm for the study schedule/recruitment process is showed in Figure 1.
Eligibility criteria
Patient eligibility criteria include
(1) age 18 to 85 years,
(2) diagnosis of ischemic stroke (atherosclerotic cerebral infarction) or transient ischemic attack (diagnosis meets the criteria established by the Cerebrovascular Diseases Group of the Chinese Medical Association Neurology Society in 2014)[25],
(3) acute period of onset (2 weeks post-event),
(4) NIHSS score ≤ 15 points,
(5) voluntarily participation and informed consent.
Patient exclusion criteria include
(1) any cancer diagnosis,
(2) cardiac infarction, cerebral infarction with small vessel disease, other causes of cerebral infarction or unexplained cerebral infarction,
(3) stomach disease judged by doctor to be at risk of bleeding,
(4) alteplase thrombolytic usage, as well as some cases requiring anticoagulation or dual antiplatelet (PCI),
(5) hemorrhagic stroke, mixed stroke, or tumor stroke,
(6) whole blood or platelet transfusion in the past two weeks, coagulopathy or other blood diseases,
(7) renal dysfunction (24-hour urine volume<1000ml, serum creatinine>180MMOL/L, urine creatinine<5mmol/L),
(8) patients are combined with serious heart, lung and liver system diseases,
(9) poor adherence or inability to complete long-term follow-up,
(10) current involvement in any clinical trial of research drugs or medical devices .
Patients will be excluded after enrollment if
(1) they are misdiagnosed or misrepresented cases that do not meet the inclusion criteria and meet the exclusion criteria,
(2) they voluntarily withdraw from the study,
(3) the investigator determines that an effective clinical observer can not be continued for medical or other reasons,
(4) they switched to other antiplatelet drugs, or stopped taking antiplatelet drugs, or switched to anticoagulant in the course of the study,
(5) they needed blood/platelets transfusion during the 1-year follow up.
Randomization
Once eligible patients voluntarily sign the informed consent form, researchers (JJ.W, C.P, ) will register each patient in the queue, and randomize them with a 1:1 ratio to either intervention group or control group. Randomization is achieved through a electronic program of hidden random number table. In order to avoid bias, the researcher cannot obtain the random number table, only the final random results.
Participant timeline
Baseline data such as BMI, NIHSS, mRS, imaging examination, and laboratory test results are collected during hospitalization. At discharge, besides individualized drug treatment and medical expenses, the NIHSS and mRS scores for each participant are also evaluated and recorded. All patients are required to follow up at the first, third, sixth, and twelfth months after discharge, and report drug compliance and prognosis (including stroke recurrence, adverse events, mRS, etc.). Table 1 details the participants' schedule of this study.
CYP2C19 Genotype and 11-dhTxB2 tests
Three single-nucleotide polymorphisms (SNPs) for CYP2C19 (National Center for Biotechnology Information [NCBI] Genome build 37.1, GenBank NG_008384), including CYP2C19*2 (681G>A, dbSNP rs4244285), CYP2C19*3 (636G>A, dbSNP rs4986893), and CYP2C19*17 (−806C>T, dbSNP rs12248560)[26-29] will be genotyped in participantsof intervention group. Genotyping of the 3 SNPs is performed using the PCR+MassARRAY (flight mass spectrometry). Individuals with complete information for each of the 3 SNPs will be included in the current analyses. Patients are categorized by CYP2C19 metabolizer status based on *2, *3, and *17 genotypes using the common consensus star allele nomenclature.
CYP2C19*2, *3 are loss-of-function alleles, and CYP2C19*17 is gain-of-function allele[30]. Patients with two loss-of-function alleles (*2/*2, *2/*3, or *3/*3) are classified as poor metabolizers, those with one loss-of-function allele (*1/*2, *1/*3, *2/*17 or * 3/*17) are classified as intermediate metabolizers, and those without loss-of-function allele (*1/*1) are classified as extensive metabolizers[31]. Patients carrying *1 and *17 alleles (*1/*17 or *17/*17) are classified as ultra-metabolizers[32].
Aspirin is widely used in secondary prevention of stroke because it inhibits cyclooxygenase (cox) and thus reduces the production of thromboxane A (TXA2), which reflects the activation state of platelets and the function of aspirin in vivo. In this study, aspirin resistance is tested for using 11-dhTxB2 concentration in urine with 11-dhTxB2/urinary creatinine≥1500 pg/mg used to characterize the presence of aspirin resistance[33,34]. Once obtained, results of gene and urine tests will be used to develop personalized drug treatment programs.
Treatments
On days 1-8, 3-5 ml whole blood samples and 3-10 ml urine samples for the CYP2C19 genotype and 11-dhTxB2 test are taken from patients in the intervention group. The results are usually obtained on day 7-9. After the tests, doctors will adjust the antiplatelet medication according to the results. If the patients are only resistant aspirin, then clopidogrel is prescribed; if the patients are only resistant to clopidogrel, then aspirin is prescribed. If the patients are not resistance to aspirin and clopidogrel, they will be treated with aspirin[4]. If the patient has dual resistance to clopidogrel and aspirin, they will be treated with ticagrelor[4]. Patients in the control group receive neither genetic nor antiplatelet testing. The details of treatments are shown in Figure 2. This study is single blind to patients. Patients may experience discomfort while having blood drawn for blood samples, and it is an expected harm in this study. If adverse events occur during follow-up, researchers will record, report, and deal with these events. Treatment measures include sending patients to a nearby hospital as soon as possible, and/or terminating personalized medication in case of severe bleeding.
Endpoints
The primary outcome of interest is the recurrence of stroke. Secondary outcomes include bleeding events, mRS, adverse medical reaction and all-cause mortality. Adverse medical conditions include symptoms (nausea, chest pain, etc.), signs (tachycardia, tachycardia, etc.), or abnormal test results (laboratory test results, electrocardiogram, etc.).
Clinical assessment
Baseline assessment includes: (1) demographics (age, sex, nation, address, vital sign, average annual income, educational level, work, medical payment method); (2) risk factors (smoking, alcohol, hypertension, diabetes, heart diseases, hyperlipemia, dementia, hypohepatia, renal insufficiency, obesity, stroke/malignancy tumor/trauma/
surgical history, r-tPA intravenous thrombolysis); (3) family/personal medical history; (4) medications; (5) laboratory results (blood, urine, biomarkers); (6) physiological tests (BMI, blood pressure, pulse, respiratory rate, cardioechography); (7) brain imaging (MRI, CT, CTA/MRA/DSA); (8) comprehensive neurological assessment (TOAST subtypes, NIHSS, mRS, GCS); (9) symptoms and duration of hospitalization.
Sample size
In Wang et al.’s study, 58.8% of participants were carriers of CYP2C19 loss-of-function alleles. Stroke recurrence in non-carriers only: 6.7% in treatment (clopidogrel plus aspirin) vs. 12.4% control (aspirin-only). According to the above results, each group needs 417 cases, for a total of 834. We assume a loss to follow-up rate of 15% and a design effect of 1.25[34]. Finally, we calculated a sample size of 1199, which is about 1200. We aim to recruit 1,200 to 6,000 patients. If the sample size is not completed within 20 months, we will extend the study time appropriately and expand more participating hospitals.
Data collection and management
According to the requirements of the project, we designed the data collection table on the EDC platform, defined the research process, the name of the data table, the collected data items, and formed the corresponding data collection guide. Finally, eCRF is constructed based on the research of medical records. All users who access EDC need to fill in the user account application form. After confirming and approving by the sponsor, the system administrator creates a project administrator account and grants his project administrator authority. The project administrator requests to create a researcher, research assistant, auditor, data administrator account, and grant different permissions to access the EDC. For example, the researchers in each center can only see their contents and have the right to modify the data; the auditor can read the EDC data of each center, and there is no data revision authority, but can issue questions.
Once all participants have completed the trial, all medical records will be entered into the system. After the primary investigator, sponsor, statistical analyst, and data administrators confirm that the data in the database is correct, the dataadministrators will check the integrity of the data. Finally, all the data will be locked, and the data administrator will import it into the designated database and send it to the statisticians for statistical analysis. The locked data can no longer be edited, and the problems found after the data locking can be corrected in the statistical analysis program after confirmation. After the trial is over, the data administrator will put forward a request for closing EDC and cancel all account access rights after obtaining the permission of the sponsor. Finally, when the data is completely backed up, the EDC will be shut down. The EDC record system adopts dual control of role rights.
Quality control measures
This study consists of sponsor commissions and contract research organization to complete the audit work, to ensure that the rights and interests of the subjects in the clinical trials are guaranteed, the recorded data are accurate and complete, and to ensure the test follows the follows the approved program, 《Quality Management Standards for Drug Clinical Trials》[35] and related regulations. Auditor will review the records and check interventions every week, the process will not be independent from investigators and the sponsor.
If there are problems during the actual implementation of the clinical trial, it is necessary to revise the plan and submit it to the sponsor. After consultation and discussion by the multi-center coordination committee, the clinical research unit will revise the plan and then submit to the sponsor and the participating units for signature approval. Finally, it is submitted to the ethics committee for approval and implementation. If important new information related to the test drug is found, the informed consent form must be modified in writing form and sent to the Ethics Committee for approval.
All researchers must sign a confidentiality agreement with the project responsible unit to keep the relevant documents confidential and not leaked. Data collection is conducted by designated investigators, and data will not be leaked or shared during the study. After the studyis finish, the researcher needs to keep all relevant documents. Includes subjects' records, original informed consent, original documents, eCRF's PDF electronic documents, ethics committees, and research-related materials between researchers. The research unit must keep the above materials for at least five years after the end of the clinical trial.
Statistical analysis
Patients who meet the enrollment criteria will be recruited, and the hospitalization data and follow-up data of the recruited patients will be analyzed. Continuous variables that conform to the normal distribution are described using the mean (standard deviation). Those that do not conform to the normal distribution are presented as the median (inter-quartile range). Categorical variables are expressed as the number (percentage). Pearson’s χ2 test of independence will be used for categorical variables, and the analysis of variance and Mann-Whitney U test will be used for numerical variables, as appropriate. Multiple imputation will be used to impute missing values, under a missing at random assumption, so as to reduce bias and avoid excluding participants from the analysis. Comparison of endpoint incidence between intervention and control groups will be assessed by risk rate (RR). Multivariate Cox regression analysis will be used for the time-to-event data with adjustment of known confounders. In this study, the estimates of missing values for the primary variables are carried-forward to the absence of test data using the results of the closest observation. All statistical tests will be two-sided, and P <0.05 is considered statistically significant. In addition, we will use the deep learning methods that take time information into account to model the risk prediction of neurological impairment in cerebrovascular disease. Finally, the project intends to use the Recurrent Neural Network (RNN) to study the new cerebrovascular disease risk prediction model with time information in medical health electronic data. RNN can identify new high-level features that are significantly contributing and trace back to their lower-level features by examining hidden layers of well-predicted neural networks to identify new important risk factors that may be missed by conventional methods[36]. Combined with clopidogrel resistance and aspirin resistance, the Long Short Term Memory artificial neural network and Gated Recurrent Unit methods were used to establish neural network risk prediction neural network[37,38]. All of the above statistical analysis will use Statistical Analysis System (SAS) 9.3 and JetBrains PyCharm software.