Study Design and Subject Recruitment
The randomized crossover intervention design was adopted in this study. 24 patients with SCAD were recruited to carry out the intervention trial of air purifier in their residences (air purifier with HEPA was “real-intervention”, and without HEPA was “sham-intervention”). The experiment was completed in 4 batches with 6 subjects in each batch. Every intervention period was 3 days and the washout period was more than 14 days. The arrangement of the intervention is shown in Fig. 1.
The subjects were recruited from 3 community-level (or above) general hospitals in Beijing, who were clinically diagnosed with SCAD according to the diagnostic requirements of 2013 ESC Guidelines on the Management of Stable Coronary Artery Disease (Montalescot et al. 2013) published by the European College of Cardiology and 2010 Chinese Consensus on the Management of Chronic Stable Coronary Artery Disease (Hu 2010) issued by China. In addition, the inclusion criteria include: (1) the age was between 55 and 80 years old; and (2) they lived near the hospital for a long time and the house was not decorated within one year. The exclusion criteria are as follows: (1) someone in their house smoked or themselves had just quitted smoking for less than one year; (2) patients were diagnosed with acute myocardial infarction within 3 months, or receive percutaneous coronary intervention and bypass transplantation within 6 months; (3) patients wore a pacemaker; and (4) patients suffered from liver injury, cancer and other serious diseases.
During the intervention, the individual exposure to PM2.5 and the concentration of PM2.5 indoor and outdoor the residence was monitored continuously. The age and body mass index (BMI) of the subjects was investigated, and their travel behaviors, window opening time and medication were recorded. The subjects stayed at home for more than 20 h/d, and the window opened for no more than 2 h/d during the intervention. Drug use records focused on conventional medicine for CAD treatment. Blood samples were collected at the end of each intervention, and the following indicators were detected: (1) inflammation indicators such as IL-6, TNF-α, CRP and FIB; (2) coagulation indicators such as CD62P, CD40L, intercellular adhesion molecule-1 (ICAM-1), nitric oxide (NO) and ET-1; (3) plaque stability indicators such as MMP-2, MMP-9 and TIMP-1; (4) blood lipid indicators such as CHO, TG, HDL-C and LDL-C.
Environmental Monitoring
Particulate matter sampling pumps MicroPEMTM (RTI Company, USA) based on the principle of light scattering were used to monitor PM2.5 concentration with sampling flow of 500 mL/min. It can monitor the environmental temperature and relative humidity (RH) simultaneously. Daily average concentration of PM2.5 was calculated from 8 a.m. to 8 a.m. Lag and moving average levels of PM2.5 exposure are represented by "lag".
Sampling pumps for monitoring the indoor PM2.5 concentration were placed in the open area of the house to avoid air conditioner vent, kitchen, doors and windows, about 1.5m above the ground, more than 1m far from the wall, and at least 2 m far from the air purifier. Sampling pumps for monitoring the outdoor PM2.5 concentration were placed outside the residential window and avoid air conditioner unit and kitchen and its air outlet. Sampling pumps for monitoring individual exposure to PM2.5 were taken along by the subjects.
Blood Collection and Indicator Measurements
Fasting blood samples (5 mL anticoagulant and 5 mL non-anticoagulant) were collected from 8:00 to 9: 00 on the day at the end of each intervention. IL-6, TNF-α, CRP, FIB, CD62P, CD40L, ICAM-1, NO, ET-1, MMP-2, MMP-9, and TIMP-1 levels were tested by a professional testing institution. CHO, TG, HDL-C and LDL-C levels were tested by the laboratory of the hospital. IL-6, TNF-α, TIMP-1, CD62P, CD40L and ICAM-1 was detected by a bead-based multiple flow cytometry on Luminex 200 system (Luminex Corporation, Austin, TX, USA) with chips (IL-6, TNF-α and TIMP-1: Merck Millipore, Germany; CD62P, CD40L and ICAM-1: R&D, USA). CRP, NO, MMP-2, MMP-9, FIB and ET-1 was separately detected by ELISA kit (CRP, NO, MMP-2, MMP-9: R&D, USA; FIB: NOVUS, USA; ET-1: Elabscience Biotechnology Co. Ltd, Wuhan, China).
In addition, a compound index-atherogenic index of plasma (AIP) was calculated by the results of TG and HDL-C levels and the formula: AIP=log (TG/HDL-C).
Statistical Analyses
Descriptive statistics were conducted for general characteristics of the environmental exposure and effects. Data with normal distribution were described by mean and standard deviation (mean±SD), and data with skewed distribution were described by median and interquartile range as M (P25, P75). The paired Student’s t-test was employed to compare the PM2.5 concentrations between two intervention periods, or the indoor and outdoor environment. The Linear mixed effect models were conducted to estimate the correlation between PM2.5 exposure and cardiovascular indicators. The results were reported by the percentage change of indicators associated with an interquartile range (IQR) decreases of PM2.5 exposure. The interpretation contribution rate of independent variables in the model were calculated by the determination coefficient R2 of independent variables/ of the model.
In the linear mixed effect models, the "effective drug" variable is defined as one or more kinds of the effective drug had been taken by the subject (dichotomous variable). The kinds of drug need to be controlled were selected according to Guidelines for Rational Drug Use of Coronary Artery Disease (2nd Edition) (Expert Committee on Rational Drug Use of National Health and Family Planning Commission and Chinese Pharmacists Association 2018) and the actual medication of the subjects (taken by more than 20% of the subjects). Whether the drug is effective was determined by searching the literatures on "Pubmed", "Web of Science", "China National Knowledge Infrastructure", "Wanfang" and "VIP" (before December 30, 2020). It needs to meet the following requirements: (1) there were two or more case-control trials or pre-and post-treatment control trials of cardiovascular disease patients treated with the drug, and the results of the trials were statistically significant; or (2) there was a case-control trial or pre-and post-treatment control trial of cardiovascular disease patients treated with the drug, and the results were statistically significant, and there was one or more animal experiment of the drug, and the results were statistically significant.
Influencing factors of PM2.5 exposure and cardiovascular indicators need to be considered as independent variables in the linear mixed effects models. It includes the "effective drug", age, sex and BMI, as well as temperature and relative humidity. The stepwise elimination method based on the principle of minimization of Akaike’s Information Criterion (AIC) was adopted for the validity of models. We screened three additional independent variables on the basis of preserving key variables (PM2.5 exposure and the "effective drug"). The model was established for each indicator respectively, given the influencing factors differences.
The parameters of the model were tested by bilateral t-test, with statistical significance of 0.05. The "lme4" package and "r2glmm" package in the R (Version 4.0.3) was separately used for linear mixed effect model and R2 calculation.
Quality Control
The monitoring instruments were cleaned and calibrated before sampling. Instrument condition review was done at the end of the first and last day of each intervention. The blood samples were collected by the nurses. Blood lipid indicators were detected on the same day, and samples were stored at-80℃ before further detection. The detection process was controlled strictly in accordance with the testing procedures and the requirements of the chips/kits.
The subjects of the study were trained by professional personnel before recording the questionnaire themselves. Questionnaires were re-checked at the end of the first and last day of each intervention for ensuring recording accuracy. Questionnaire and data input were done by two personnel, and then reviewed and analyzed by special personnel.