Association between erythrocyte parameters and metabolic syndrome in urban Han Chinese: a longitudinal cohort study

Background Although various cross-sectional studies have shown that erythrocyte parameters, including red blood cell (RBC), hemoglobin (Hb) and hematocrit (HCT), were linked with metabolic syndrome (MetS), few longitudinal studies have been used to confirm their relationship. The study, therefore, constructed a large-scale longitudinal cohort in urban Chinese population to highlight and confirm the association between erythrocyte parameters and MetS/its components. Methods A longitudinal cohort with 6,453 participants was established based on the routine health check-up systems to follow up MetS, and Generalized Estimating Equation (GEE) model was used to detect the association between erythrocyte parameters and MetS/its components (obesity, hyperglycemia, dyslipidemia, and hypertension). Results 287 MetS occurred over the four-year follow-up, leading to a total incidence density of 14.19 per 1,000 person-years (287/20218 person-years). Both RBC and Hb were strongly associated with MetS (RR/95% CI, P value; 3.016/1.525-5.967, 0.002 for RBC; 3.008/1.481-6.109, 0.002 for Hb), with their dose–response trends detected. All three erythrocyte parameters (RBC, Hb and HCT) were found to be associated with obesity, hypertension and dyslipidemia with similar dose–response trends respectively, while only Hb showed a significant association with hyperglycemia. Conclusions Elevated erythrocyte parameters were confirmed to be associated with MetS/its components in urban Chinese population, suggesting that erythrocyte parameters might be served as a potential predictor for risk of MetS.

The study, we established a longitudinal cohort with 6,453 participants based on the routine health check-up systems in urban Chinese population to follow up MetS, and each individual in this cohort was undergone at least three repeated health checks in the five years (January 2005 to January 2010). Furthermore, Generalized Estimating Equation (GEE) model, which could handle the repeat measurement data with high autocorrelation in the framework of logistic regression model [11][12][13][14], was used to detect the association between erythrocyte parameters and MetS/its components (obesity, hyperglycemia, dyslipidemia, and hypertension). All individuals in the longitudinal cohort underwent a general health questionnaire, anthropometric measurements, and laboratory tests. The general health questionnaire covered the current status of smoking, alcohol intake, diet, sleeping quality and physical activity. Anthropometric measurements involved height, weight, and blood pressure. Both height and weight were measured with light clothing without shoes. Body mass index (BMI) was calculated as weight/height 2 (kg/m 2 ) as an evidence of obesity. Blood pressure was measured using Omron HEM-907 by the cuff-oscillometric method on the right arm in sitting position after a 5-min rest, and the mean systolic and diastolic blood pressure values of two measurements were recorded respectively. While the participant was fasting, a venous blood sample was taken for laboratory test. Laboratory tests included RBC, Hb, HCT, white blood count (WBC), platelet distribution width (PDW), mean platelet volume (MPV), thrombocytocrit (PCT), glucose, total cholesterol (CHOL), low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides (TG), gamma-glutamyl transpeptidase (GGT), serum albumin (ALB), serum globulins (GLO), blood urea nitrogen (BUN), and serum creatinine (SCr), etc. This study was approved by the Ethics Committee of School of Public Health, Shandong University, and all participants were given informed written consent.

Definition of the metabolic syndrome
Considering the target population was Chinese with their specific physiological characteristics, Diabetes Branch of the Chinese Medical Association (CDS) [15] was used as the MetS diagnostic criteria, which is very popular for the Chinese population in clinical practice. MetS was defined as presence of three or more of the following four medical conditions: 1) overweight or obesity, i.e. BMI ≥25.0 Kg/m 2 ; 2) hypertension, i.e. systolic blood pressure (SBC) ≥140 mmHg, or diastolic blood pressure (DBP) ≥90 mmHg, or previously diagnosed; 3) dyslipidemia, i.e., fasting TG ≥1.7 mmol/L, or fasting HDL <0.9 mmol/L; 4) hyperglycemia, i.e. fasting bloodglucose (FPG) ≥6.1 mmol/L, or 2 h Post-meal Glucose (PG) ≥7.8 mmol/L, or previously diagnosed.

Statistical analysis
To account for missing values, multiple imputation was performed. Since imputation method was depended on   the patterns of the missing data and the types of the imputed variables, without loss of generality, the Markov chain Monte Carlo (MCMC) method was chosen according to MI Procedure of SAS 9.1.3 [16]. Most variables had less than 2% missing observations before imputation except diet, drinking, smoking, quality of sleep and physical activity having less than 10% missing values. The original continuous erythrocyte parameters were categorized into 4 levels (Q1-Q4) using the 3 quartiles of P25, P50 and P75 as critical values, with ≤ P25 for Q1, >P25 and ≤ P50 for Q2, >P50 and ≤ P75 for Q3, and > P75 for Q4 respectively. Summary statistics were used to illustrate the distribution characteristics for variables of interest at each repeated surveys, and student's t test for continuous variables and chi-square test for categorical variables were used to detect the statistical significances compared with the first survey (baseline). As GEE model could handle the repeat measurement data with high autocorrelation in the framework of logistic regression model [11][12][13][14], it was used to detect the association between erythrocyte parameters and MetS/ its components. Simple GEE model was firstly used to select variables associated with MetS/its components, then variables which were significant at the level of 0.05 in the simple GEE analysis entered the multiple GEE model to adjust the potential confounding. The 'Logit' link function was chosen in GEE analysis, with significance level 0.05. All the statistical analyses were performed on SAS 9.1.3. Table 1 summarized the characteristics of erythrocyte parameters levels and other potential confounding factors of the participants at each repeated survey, which showed that most factors were generally higher than that in the first survey (baseline). A total of 294 cases of MetS occurred over the four-year follow-up, leading to a total incidence density of 14.19 per 1,000 person-years (287/20218 person-years). During the follow up, 3 participants were diagnosed as cerebral infarction (personyears), 141 participants were diagnosed as cardiovascular disease (person-years) and no-one underwent coronary artery bypass surgery (see Additional file 1: Table S20). Table 2 showed the selected variables associated with MetS at α = 0.05 level. It indicated that each of the 3 erythrocyte parameters (RBC, Hb, and HTC) with 9 potential confounding factors, including gender, age, GGT, GLO, BUN, WBC, diet, drinking and smoking might be linked with MetS. Also, each of the 3 erythrocyte parameters might be linked with the four components of MetS with their specific potential confounding factors respectively (see Additional file 2: Table S1, Additional file 3: Table S2, Additional file 4: Table S3 and Additional file 5: Table S4 for details).

Discussions
This study mainly attempted to confirm the association between erythrocyte parameters and MetS/its components using the longitudinal cohort. Although the longitudinal cohort study was based on routine health check-up in urban Han Chinese population from middle to upper socioeconomic strata, the positive associations between erythrocyte parameters and MetS/its components were observed, which were also detected in other two cohort study [9].
RBC had been reported to be associated with MetS in various populations by cross-sectional studies [2][3][4][5][6] and in Ethiopia by a cohort studies [9]. At present study, we not only confirmed that RBC was associated with MetS and its 3 single components (dyslipidemia, obesity and hypertension), but also observed the dose-response trends (seeing Table 3 or Additional file 7: Table S6, Additional  file 8: Tables S7 and Additional file 9: Table S8). These results highlighted the positive association between RBC and MetS. In pathogenesis, this linkage might be explained by the insulin resistance(IR) mechanisms in the development of MetS, because insulin and insulin growth factors I and II supporting erythropoiesis in both vitro [17][18][19][20][21] and vivo [22][23][24] had been detected in laboratory studies.
As another important erythrocyte parameter, Hb also had been reported to be associated with MetS in Thailand [7] and Japan [6] by Cross-sectional studies, as well as in Ethiopia [9] by a cohort study. This positive association was also detected in our longitudinal cohort study with a potential dose-response trend between them (seeing Table 3 or Additional file 11: Table S10, Additional file 12: Table S11, Additional file 13: Table S12, Additional file 14: Table S13 and Additional file 15: Table S14). The possible mechanism might be supported by the following pathogenesis. Hb is a well recognized carrier and buffer of nitric oxide (NO), and various compounds of Hb with NO can affect Hb-oxygen affinity of the whole blood [25]. Disturbed NO synthesis may exert an adverse effect on endothelial dysfunction through the L-arginine-  NO pathway [26]. Furthermore, endothelial dysfunction was reported to be associated with MetS [27,28]. All these evidences expect the association between Hb and MetS/its components in population level. Elevated HCT could increase blood viscosity and peripheral resistance to blood flow, and further contribute to IR [29][30][31]. Therefore, the association between HCT and MetS/its components should be observed in population level. In this paper, HCT associating with obesity, hypertension and dyslipdemia were all detected in urban Han Chinese population (seeing Table 3 or Additional file 17: Table S16, Additional file 18: Table S17,  Additional file 19: Table S18), while no statistical significant association between HCT and MetS/ hyperglycemia was found (seeing Table 3, Additional file 16: Table S15 or Additional file 20: Table S19). Similar results were also observed in Ethiopia [9] and Japan [9] by cohort studies, as well as in Thailand [7] and Japan [9,11] by Cross-sectional studies.
Several limitations of this study must be considered. a) Selection bias might exist due to the samples just from routine health check-up population for middle-toupper class urban Han Chinese. b) Owing to the absence of waist circumference measurement, the diagnostic criteria of MetS was just based on China Diabetes Federation, rather than international standard criteria.  c) The medication history and menstrual history of participants might be significant confounding factors, but they were absent in our database. d) Hematological parameter categories were based on a single assessment of blood, which may cause a misclassification bias. It is, therefore, desired to conduct a perfect longitudinal cohort study in general population for further highlighting the association between erythrocyte parameters and MetS.