Subjects
The population of this study including 1324 subjects lived in the Pingguoyuan area of Beijing, China. Baseline data obtained initially from 1680 subjects between September 2007 and January 2009 after a routine health check-up. We followed this population prospectively from February 1 to September 30, 2013 for the first time, and obtained follow-up data from 1499 subjects (181 subjects were lost, follow-up rate was 89.2%). The median follow-up interval for the original 1499 subjects was 4.8 years. Of the 1499 subjects, 175 were excluded from analyses because of because of a history of cardiovascular disease, remaining 1324 subjects for the analysis. No differences other that baseline risk factors were noted in those who completed baseline and follow-up assessments.
The study was approved by the ethics committee of the People’s Liberation Army General Hospital, and each subject provided informed written consent.
Clinical data collection
All subjects received a face-to-face questionnaire survey to ascertain new CVD events during these visits. Prevalent diseases, medical histories, lifestyle factors and family history were collected. Urine and fasting blood samples were also obtained. Blood pressure and anthropometric measurements were obtained by trained physicians.
Biomarker variable determination
We collected blood samples from all 1324 subjects in centrifuge tubes after overnight fast and centrifuged them for 15 min at 1200×g. Serum aliquots were frozen at −80℃ until assays were performed. Concentrations of PCSK9 were measured after a second thaw using a commercially available quantitative sandwich ELISA assay by following the manufacturer's instructions (CY-8079; CycLex Co., Nagano, Japan) at the laboratory of MBL (Medical & Biological Laboratories Co., Ltd., Nagoya, Japan) [17]. Concentrations of fasting blood glucose (FBG), lipids (TC, TG, LDL-C, HDL-C) were measured on a Roche autoanalyzer (Roche Diagnostics, Indianapolis, IN, USA) by the Roche enzymatic assays (Roche Diagnostics GmbH, Mannheim, Germany). Concentrations of serum creatinine were measured on a Hitachi 7600 autoanalyser (Hitachi, Tokyo, Japan) by enzymatic assay (Roche Diagnostics GmbH). All testing was performed in the Department of Biochemistry of Chinese PLA General Hospital by well trained personnel following the criteria of the World Health Organization Lipid Reference Laboratories.
Definition of variables
CVD events were defined as fatal and non-fatal myocardial infarctions, unstable angina, stable angina, deaths from coronary heart disease. Body mass index (BMI) = weight (kg) / height2 (m2). The estimated glomerular filtration rate (eGFR) was calculated using the following Chronic Kidney Disease Epidemiology Collaboration equation: eGFR = 141 × min (Scr/κ,1)α× max (Scr/κ, 1)-1.209 × 0.993Age × 1.018 [if female] × 1.159 [if black], where Scr is plasma creatinine (mg/dL), κ is 0.7 for females and 0.9 for males, α is -0.329 for females and -0.411 for males, min indicates the minimum of Scr/κ or 1, and max indicates the maximum of Scr/κ or 1. Hypertension was defined as a mean SBP ≥140 mmHg, mean DBP ≥90 mmHg, both, or the use of antihypertensive medication. Diabetes mellitus (DM) was defined as a fasting glucose ≥7.0 mmol/L, glucose ≥11.1 mmol/L at two hours after an oral 75 g glucose challenge, the use of antihyperglycaemic medication, or both.
Statistical analyses
Baseline characteristics were expressed as the median (interquartile range) or mean ± standard deviation (SD) for continuous variables and percentages for dichotomous variables. Subjects were categorized into quartiles according to baseline distribution of PCSK9 levels. Q1 was considered low level group whereas Q2-Q4 was considered high level group.
Differences in the baseline levels of risk factors and clinical characteristics between subjects based follow-up CVD or non-CVD groups were analyzed using a t-test for continuous variables and a chi-square test for categorical variables. Cox proportional hazard regression model was used to evaluated the association between baseline PCSK9 levels and future CVD events. Regression models were adjusted for age and gender as well as hypertension, SBP, DBP, Diabetes (DM), smoking, BMI, levels of glucose, TC, HDL-C, TG, and eGFR (model 1); Model 2 was adjusted for model 1 plus levels of LDL-C; Model 3 was adjusted for model 1 plus levels of sd-LDL-C.
All analyses were conducted using Stata software (version14.0; Stata Corporation, College Station, TX). P-values <0.05 were considered statistically significant.
Systematic Review and meta-analysis
We performed a systematic review and meta-analysis that incorporated results from the current study into findings from previous studies on the association between plasma PCSK9 and risk of CVD. We searched Pubmed and EMBASE from inception until December 1, 2016, with no language restrictions applied. The following full search strategy was used: (proprotein convertase subtilisin/kexin type 9 OR PCSK9) AND (cardiovascular disease OR cardiovascular events OR cardiovascular risk OR coronary heart disease OR myocardial infarction OR stroke OR mortality OR all cause death). Articles describing levels of plasma PCSK9 and risk of CVD retrieved through this search were screened by two investigators (YB and XW) and any disagreement was resolved by consensus. Reference lists of the included articles were hand-searched for potentially relevant articles.
Studies included were prospective population-based cohort which measured plasma PCSK9 levels at baseline and then reported CVD risk during follow-up. Each included cohort study had to report either risk estimates (relative risks, odds ratios, or hazard ratios) with 95% confidence intervals (CI), or provide sufficient data to estimate these.
The following information was obtained from eligible studies: study design, population, risk estimates, and their 95% CIs. In studies not reporting these data, we calculated risk estimates from the survival curves. Data extraction was performed by two authors separately (YB and XW) to ensure accuracy and disagreements were discussed in a consensus conference. Random effects meta-analysis was used due to the presence of high heterogeneity (I2 >50%). Test for interaction between subgroups was performed using Cochran’s Q test. Sensitivity analyses were conducted to evaluate the robustness of our results. We removed each study individually to evaluate that study’s effect on the summary estimates. Small study bias was assessed with funnel plots and Egger regression test. Statistical analyses were performed with Stata software (version14.0; Stata Corporation, College Station, TX).