Introduction

In recent years, the Western lifestyle has gradually pervaded the Chinese people’s daily life which characterizes with an increased consumption of red meat in the total amount of calories. The excess intake of red meat contributed to the rapid increase of the prevalence of type 2 diabetes mellitus (T2DM) in China [1]. It has been confirmed that the high content of iron in the red meat serves as a catalyst in the formation of reactive oxygen species (ROS) such as hydroxyl radical, which induced the oxidative stress of the liver and muscle, as well as the apoptosis of pancreatic β cells [2]. Serum ferritin is widely used as a clinical biomarker of body iron stores which is determined by the effect of genetic and dietary factors [3]. The pathologically increased ferritin level in genetic diseases, such as hereditary hemochromatosis, contributes to the occurrence of T2DM [4]. However, the association between moderately increased ferritin level caused by the over-intake of red meat and the risk of abnormal glucose metabolism (AGM) is still unclear.

The study investigates the association between serum ferritin levels with abnormal glucose metabolism risk, especially in the prospective cohort or nested case-control studies which have provided strong evidence about the interrelation of them, and most of these studies are done in western countries [5–8]; however, the relevant studies among the Asian countries are quite rare [9, 10] and no one reports them in Southwestern China to date. In this study, we designed a prospective cohort to explore the association between serum ferritin levels and AGM risk in Southwestern Chinese population.

Materials and Methods

Study Design and Subjects

The Chinese Diabetes Society conducted a cross-sectional study about diabetes and relevant metabolic risk factors by using multistage, stratified sampling method between 2007 and 2008 in China [11]. A total of 689 adults (age of 20 years old) who came from two urban communities and two rural villages in Southwestern China were involved in this study. Based on the previous cross-sectional study, these participants have been followed up for 4 years. The present study was reviewed and approved by The Ethics Committee of China-Japan Friendship Hospital (location: 2 Cherry Pank Street, Chaoyang District, Beijing 100029, China). Written informed consent was obtained from every subject before the study. Our study was registered in the Chinese clinical trial registry (#TR-CCH-ChiCTR-OCS-09000361).

Exclusion criteria included the following: the participants with AGM or who were taking medications for diabetes in the cross-sectional study at baseline, the subjects with chronic diseases such as hepatic or renal diseases, the subjects with thyroid or parathyroid diseases, the subjects taking osteoporosis medications or other drugs that might influence glucose metabolism such as glucocorticoid, pregnant women, and the subjects with incomplete data. According to the exclusion criteria, the final sample size for the present analysis was 383 participants (including 133 men and 250 women).

Subjects Classification

Our study got the blood samples of participants who did not have AGM at baseline between 2007 and 2008, and the serum ferritin levels were measured. The study was designed as a prospective cohort study, which used the baseline serum ferritin levels as the exposure. The subjects whether developed into AGM including type 2 diabetes (T2DM), impaired fasting glucose (IFG), and impaired glucose tolerance (IGT), according to the diagnosis criteria of World Health Organization (WHO) in 1999, during the follow-up, or maintained normal glucose tolerance (NGT) were considered as the result. We make use of the quartile methods in classifying the subjects into four groups according to the baseline serum ferritin levels in females and males, respectively, and compared the differences in developing AGM under different baseline serum ferritin levels and analyzed the association between baseline serum ferritin levels and AGM. During the follow-up, we got the blood samples of the 383 participants and measured their serum ferritin levels.

Demographic Characteristics

The clinical characteristics and lifestyle of participants were recorded by trained staff using a standard questionnaire [12]. Cigarette smoking was defined as having smoked at least 100 cigarettes in one’s lifetime. Alcohol consumption was defined as consumption of ≥30 g of alcohol per week for 1 year or more. Leisure-time physical activity was defined as participation in ≥30 min of moderate or vigorous activity per day at least 3 days per week. A physical examination was performed on all the participants and included an assessment of body height, weight, blood pressure, pulses, and waist circumference. Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters.

After an overnight fast ≥8 h, the Oral Glucose Tolerance Test (OGTT) using a 75 g glucose load was performed to measure fasting plasma glucose (FPG), 2 h-plasma glucose (2 h-PG), fasting serum insulin (FINS) and 2 h-serum insulin (2 h- INS).

Biochemical Parameter Measured

Plasma glucose levels were measured by using a hexokinase enzymatic method. Insulin was measured by a radioimmunoassay with human insulin as a standard (Linco, St. Charles, MO). Triglyceride (TG), total cholesterol (TC), high-density lipoprotein-cholesterol (HDL-C), low-density lipoprotein-cholesterol (LDL-C), aspartate aminotransferase (AST), alanine transaminase (ALT), and blood creatinine (Cr) values were measured for all the patients by using the automated standardized equipment from the Clinical Laboratory of West China Hospital, Sichuan University. Serum ferritin level was measured with a radioimmunoassay kit (Beijing North Institute of Biological Technology). Both at baseline and during the follow-up, the questionnaire, physical examination, OGTT, biochemical indexes, and serum ferritin level were recorded and measured. At baseline, homeostasis model assessment formula insulin resistance (HOMA-IR) was used to measure the insulin resistance as the equation,\( \mathrm{HOMA}-IR\kern0.50em = \left[\mathrm{FINS}\ \left(\frac{\mathrm{mIU}}{\mathrm{L}}\right)\times FPG\left(\frac{\mathrm{mmol}}{\mathrm{L}}\right)\right]/22.5 \), and homeostasis model assessment formula beta-cell function (HOMA-β) was used to measure the function of pancreatic β-cells as the equation \( \mathrm{HOMA}-\upbeta \kern0.5em = \left[20\times \mathrm{FINS}\ \left(\frac{\mathrm{mIU}}{\mathrm{L}}\right)\ \right]/\ \left[FPG\ \left(\frac{\mathrm{mmol}}{\mathrm{L}}\right)-3.5\right] \).

Statistical Analysis

Normally distributed data are expressed as means ± standard deviation (SD) for continuous variables and percentage for categorical variables. Skewed variables were logarithmically transformed before analysis and were expressed as median (interquartile range). The quantitative variables were compared using Student’s t test, while the qualitative variables were compared using the chi-square (χ 2) test. The association of serum ferritin levels at baseline with AGM risk was analyzed using a logistic regression model to calculate the relative risk (RR) and 95 % confidence interval (CI). The variables included in the logistic regression model were those that were statistically significant in univariate analyses or were biologically relevant. Values with p ≤ 0.05 were considered statistically significant. All the statistical analysis was performed using SPSS 16.0 (SPSS, Inc., Chicago, IL).

Results

Baseline and Follow-Up Characteristics

During 4-years follow-up, 243 non-AGM participants did not develop into AGM, and 140 non-AGM participants (91 women and 49 men) developed into AGM. The 140 subjects developed into AGM included 28 subjects with T2DM,15 subjects with IFG, and 107 subjects with IGT; there were 10 subjects developed into IFG and IGT at the same time.

The characteristics of participants at baseline and after the follow-up were shown in Tables 1 and 2, separately. At baseline, compared with the participants who did not develop into AGM, participants who developed into AGM were older and the waist, BMI, systolic blood pressure (SBP), TG, FPG, 2 h-PG, FINS, 2 h-INS, and HOMA-IR were higher significantly both in females and males. The baseline serum ferritin levels were higher in men than the ferritin concentration in women. Moreover, the concentration of serum baseline ferritins were markedly higher in subjects who developed into AGT compared to those who remained NGT, 110.9 (124.8) vs. 63.9 (108.6) in women and 199.5 (169.2) vs. 150.8 (159.8) in men, separately. During the follow-up, the age, waist, BMI, SBP, and DBP were significantly higher and the levels of TG, FPG, and 2 h-PG increased significantly both in women and men who developed AGM compared with those who did not.

Table 1 Baseline characteristics of participants in 2007–2008
Full size table
Table 2 Follow-up characteristics of participants in 2011–2012
Full size table

The Relationship Between Baseline Serum Ferritin Levels and AGM Risk in Females and Males

Table 3 showed the logistic regression analysis of the relationship between baseline serum ferritin levels and AGM risk in females and males, respectively. After adjusting the confounders of age, the family history of diabetes, current smoking, and alcohol, compared with the lower quartile of baseline serum ferritin levels, the females in the upper quartile of baseline serum ferritin levels had an increased trend of developing AGM [RR 2.86 (1.25–6.56), p = 0.013]. After adjusting the confounders of age, the family history of diabetes, current smoking, alcohol, and components of metabolic factors (waist circumference, SBP, TG, and HOMA-IR), compared with females in the lower quartile of baseline serum ferritin levels, the females in the upper quartile of baseline serum ferritin levels had no significant difference in the risk of developing AGM [RR 2.23 (0.94–5.30), p = 0.068].

Table 3 Relative risk for incident AGM according to serum ferritin level in female and male, 2007–2008
Full size table

For the males, the study results were consistent with those in females. In multivariable-adjusted models, the RRs for developing AGM compared subjects in the highest with those in the lowest quintile serum ferritin levels were 3.50 (p = 0.029) after adjusting age, the family history of diabetes, current smoking, and alcohol and 2.74 (p = 0.103) after adjusting age, the family history of diabetes, current smoking, alcohol, and components of metabolic factors (waist circumference, SBP, TG, and HOMA-IR).

In order to find the most effective factor of AGT, confounding factors, including ferritin, age, the family history of diabetes, current smoking, alcohol, waist circumference, SBP, TG, and HOMA-IR, were analyzed using logistic regression model (forward, conditional). Waist and SBP were included in the final equation of logistic regression in females. Unlike the result of logistic regression in females, serum ferritin, age, and waist appeared in the final equation of logistic regression in males.

Discussion

In this prospective cohort study, we found that incident AGM patients had higher baseline and follow-up serum ferritin levels than those in non-AGM subjects. Subjects with the upper quartile of baseline serum ferritin levels had the highest risk of developing AGM. These results indicated that elevated serum ferritin levels might be associated with the increased risk of AGM in Southwestern Chinese. However, after adjusting the age and components of metabolic factors (waist circumference, SBP, TG, and HOMA-IR), the subjects with the highest quartile of baseline serum ferritin levels did not have higher risk of developing AGM, which suggested that high serum ferritin levels might not be an independent factor for the occurrence of AGM.

It is widely recognized that the pathologically high level of serum ferritin is the major cause of AGM in hereditary hemochromatosis [13]. The excess iron in patients with hereditary hemochromatosis could reduce the insulin secretion by destroying pancreatic β cells and induce insulin resistance based on which deposited in liver and muscle tissues [13]. However, it is still unclear whether moderately increased ferritin level directly affects the AGM. From the aspect of cause and effect, although the extremely high level of ferritin contributes directly to the pathogenesis of AGM, it is highly possible that moderately elevated levels of serum ferritin only play a role in metabolic intermediate or a biomarker of metabolic abnormalities.

The mechanism of excess iron in hereditary hemochromatosis leading to AGM is summarized as above (elevated serum ferritin levels → AGM). However, many literatures have demonstrated that the abnormalities in metabolic factors, such as hyperlipidemia, obesity, and hypertension, may increase the level of serum ferritin (metabolic factors → elevated serum ferritin levels) [14, 15, 16]. The hyperinsulinemia in metabolic syndromes could interfere with iron metabolism by stimulating iron uptake and ferritin synthesis [14]. Oxidative stress also exists in metabolic syndromes, which induced insulin resistance of peripheral tissues and increases the ferritin mRNA expression [15]. Qin Tang found that serum ferritin level is correlated with higher waist circumference, SBP, HOMA-IR, and concentration of TG in Chinese males [16]. Based on these, the moderately elevated serum ferritin might be the result of metabolic syndromes and AGM or the metabolic intermediate for the occurrence of AGM.

Salonen et al. [17], who designed a case-control study, reported that men with the lowest quarter of the ratio of transferrin receptors to ferritin had higher risk of developing AGM [17]. Some case-control or cross-sectional studies found that AGM subjects had higher ferritin levels than those in non-AGM subjects [9, 18]. Jiang et al. [6] published the first follow-up study (a nested case-control study), explored the association between serum ferritin levels and T2DM risk [6], and confirmed the higher serum levels increased the T2DM risk [RR 2.68 (95 % CI, 1.75–4.11)].

A systematic review and meta-analysis summarized the published literatures about the association between body iron stores and T2DM risk [19]. Most of the literatures supported that the elevated body iron stores were significantly associated with increased risk of AGM. Some studies showed that gender difference might significantly influence on these results. Dekker et al. [20] and Shi et al. [21] found that only females who had higher ferritin levels had increased risk of developing T2DM while Le et al. [8] came out with the opposite conclusion. Jehn et al. [22] demonstrated when the components of metabolic factors and inflammatory indexes were not adjusted, a higher plasma ferritin level was significantly associated with increased T2DM risk [RR 1.74 (95 % CI, 1.14–2.65)]; however, the result was not significant when these factors were adjusted, which implicated that metabolic factors play an important role in association of serum ferritins levels and T2DM risk. They concluded that elevated ferritin might not be a causal factor for T2DM, but a biomarker of metabolic abnormalities related to the pathogenesis of T2DM [22], which is consistent with our findings.

Previously, many studies among the Chinese population were designed as cross-sectional or case-control studies, and their results indicated that the high ferritin levels could merely reflect the systemic inflammation rather than the cause of AGM [9, 22–28] and no one studies them in the community of Southwestern China. Our study investigated the association between serum ferritin and AGM in Southwestern Chinese population. In addition, the limitation of the present study, during the follow-up, many participants only received anthropometric measurements, but did not want to provide blood sample; thus, the development of AGM between visits was unknown. Due to the small study sample, the power to study the relationship between AGM and serum ferritin in males was low; prospective cohort studies with larger samples were needed to confirm the relationship between serum ferritin levels and AGM.

Conclusion

In a word, elevated serum ferritin levels are associated with AGM, but not an independent risk factor in Southwestern Chinese population. The higher levels of serum ferritin might play a metabolic intermediate part in the occurrence of AGM.