Aim, design and setting of the study
This study derived data from the CHANCE trial, a placebo-controlled, double-blind, randomized clinical trial, which was conducted in 114 sites in China from October 2009 to July 2012. Detailed design and results were published elsewhere.18-20 Sample size estimation were based on the requirement to detect the smallest expected, clinically meaningful treatment difference. In brief, eligible patients were randomly assigned into the two treatment arms: clopidogrel (initiated with a loading dose of 300mg and followed by 75mg per day) with aspirin (initiated with a range from 75 to 300mg and followed by 75mg per day in the first 21 days) arm; or aspirin arm (initiated with a range from 75 to 300mg and followed by 75mgper day for 90 days) by researchers who were blinded to the treatment regimen. Eligible patients included males or females with a ≥40 years of age, diagnosed as minor ischemic stroke (National Institutes of Health Stroke Scale, NIHSS ≤3) or moderate-to-high risk TIA (defined as the ABCD2 (the age, blood pressure, clinical features, duration of symptoms, and presence of diabetes) score ≥4) and treatment could be received within 24 hours after symptom onset.
Among 5170 patients included in the CHANCE trial, 3044 patients from 73 study sites with experience of biological sample collection process management participated
in the prespecified biomarker sub-study. Meanwhile, 1342 patients from 45 sites participated in the imaging sub-study. Patients with both biomarker data as well as
imaging data will be included in the current research.
The CHANCE trial protocol was approved by the ethics committees of all study centers and registered on www.clinicaltrials.gov (registration number: NCT00979589).
All patients or their legal proxies had given written informed consent.
Trained research coordinators collected demographic and clinical characteristics at baseline including age, gender, modified Rankin Scale (mRS) prior to the event, history of ischemic stroke, TIA, myocardial infarction, atrial fibrillation, heart failure, hypertension and other important concurrent conditions, smoking habits, baseline blood pressure, through a face-to-face interview.
Measurement and Assessment of Stress Hyperglycemia
Overnight fasting venous blood was drawn within the first 24±12 hours after enrollment and preprocessed into plasma, serum, and other forms according to requirements in the protocol. Part of the plasma samples was tested for many items, including fasting plasma glucose (FPG) and low-density lipoprotein (LDL). And the serum specimens were shipped via overnight cold-chain transportation to Beijing Tiantan Hospital, Beijing, China, and stored in the −80°C condition. Before centralized testing for glycated albumin (GA) percentage in total serum albumin using the GA kit (catalog number 4085-717, Ruiyuan Bio-Technique Co. Ltd., Ningbo, China) by a Roche Modular P800 system in the clinical laboratory of Beijing Tiantan Hospital, the freezing and thawing circle were avoided. Laboratory technicians who performed the measurements were blinded to patients’ baseline characteristics, treatment group and outcomes.21
In order to assess the stress glycemia situation by both random fasting glucose as well as background glucose level, FPG (mmol/L)/GA (%) ratio was calculated and defined as the indication of stress hyperglycemia.13 Divided by the median value of the FPG /GA ratio, patients were categorized into two groups: FPG/GA ratio < median value and FPG/GA ratio > median value.
In the prespecified imaging sub-study, patients underwent Magnetic Resonance Imaging (MRI) (3.0or1.5T) examinations following the standard protocol, including diffusion-weighted imaging (DWI), 3-dimensional time-of-flight magnetic resonance angiography (3D-TOF-MRA), T1-weighted imaging and T2-weighted imaging.17
Infarction patterns were categorized into three groups: no acute infarction (NAI), single acute infarction (SAI), and multiple acute infarctions (MAI). SAI was defined as uninterrupted lesions visible in contiguous territories. MAI indicated more than one lesion “topographically distinct” from each other (separated in space or discrete on contiguous slices).22 The study population was grouped as NAI or SAI vs. MAI in the current analysis.
Readers of the imaging data were blinded to patients’ baseline information, treatment assignment and outcomes.17
Efficacy and Safety Outcomes
The central adjudication committee, blinded to the treatment assignment, adjudicated all efficacy and safety outcomes/endpoints. All enrolled patients were interviewed by trained research coordinates at baseline and the 90-day visit18 after randomization. All the adverse events, end-point events, mRS, medication use in the period between visits,
and related medical records were collected and submitted to the central adjudication committee.
The primary efficacy outcome was recurrent stroke (ischemic or hemorrhagic) within 90 days. The secondary efficacy outcomes included combined vascular events (consist of ischemic stroke, hemorrhagic stroke, myocardial infarction, and vascular death).19 The primary safety outcome was any bleeding events according to the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) definition23.
In this study, we focused on recurrent stroke, combined vascular events, and any bleeding event.
The analysis was based on Intention-To-Treatment (ITT) principle. Baseline characteristics and outcomes were compared between groups according to FPG/GA ratio and infarction pattern. Categorical variables were presented by proportions and continuous variables as medians with interquartile ranges (IQRs). Categorical variables were compared using χ2 statistics or Fisher exact test; continuous variables, using and were compared with the Wilcoxon rank-sum test.
The multivariable Cox regression was used in analyzing the association between different groups categorized by FPG/GA ratio × infarction pattern and efficacy of dual antiplatelet treatment in reducing efficacy endpoints and other outcomes after having a minor stroke or TIA. Crude and adjusted hazard ratios (HRs) and their 95% confidence intervals (CIs) were calculated (adjusted for all potential confound factors).
A 2-sided p-value, less than 0.05, was considered of statistical significance. We used SAS software, version 9.4 (SAS Institute Inc), to perform all the statistical analyses.