Association between dietary iron intake and the prevalence of nonalcoholic fatty liver disease: A cross-sectional study.

The aim was to test the association between dietary iron intake and the prevalence of nonalcoholic fatty liver disease (NAFLD) in a large sample of middle-aged and elderly Chinese population.The data included in this analysis were collected from a population-based cross-sectional study, that is, the Xiangya Hospital Health Management Center Study. Dietary iron intake was assessed using a validated semiquantitative food frequency questionnaire. The relationship between dietary iron intake and the prevalence of NAFLD was examined using logistic and spline regressions.A cross-sectional study including 5445 subjects was conducted. The prevalence of NAFLD was 36.9%. Compared with the lowest quintile, the energy-adjusted odds ratios (ORs) of NAFLD were 1.33 (95% confidence interval [CI]: 1.07-1.64), 1.80 (95% CI: 1.41-2.29) and 2.11 (95% CI: 1.60-2.80) in the 3rd, 4th, and 5th quintile of iron intake, respectively (P-value for trend <.001). In addition, dietary iron intake was positively associated with the OR of NAFLD in a dose-response relationship manner (test for trend P < .001). However, after stratifying the data by gender, such association only remained in the male, but not in the female population. With adjustment of additional potential confounders, the results did not change materially.Subjects with higher dietary iron intake were subject to a higher prevalence of NAFLD in a dose-response relationship manner. However, such association probably only exists in males, but not in females.

Introduction

Nonalcoholic fatty liver disease (NAFLD) is nowadays recognized as one of the major causes of chronic liver disease worldwide,[ which might develop to cryptogenic cirrhosis at the end stage.[ A meta-analysis involving a total of 8515, 431 participants from 22 countries estimated that 25.2% of the population aged ≥18 years in the world have NAFLD,[ and its prevalence and impact are expected to increase continuously with the growing prevalence of overweight and obesity in some developing countries.[ However, to date, the etiology and pathogenesis of NAFLD has not been fully elucidated. Although obesity has been proved to be strongly associated with NAFLD, there are still a large proportion of nonobese NAFLD patients without known risk factors.[

Iron is an essential microelement that plays an important role in red cell function, oxygen transport, as well as the synthesis of DNA, protein, and hormone.[ Previous in vivo studies have reported that high iron exposure could increase the risk of NAFLD,[ probably by inducing hepatic oxidative stress, inflammation, hepatocellular ballooning injury, lipid accumulation, or insulin resistance.[ However, there is a lack of data on the association between iron intake and the risk of NAFLD for human beings. The relevant findings would not only have important implications in understanding the etiology of NAFLD, but also contribute to the development of preventive and treatment measures.

To fill the gap of knowledge in this field, a cross-sectional study was conducted to test the association between dietary iron intake and the prevalence of NAFLD in a large sample of middle-aged and elderly population.

Methods

Study population

This cross-sectional study targeted the general Chinese participants who were undergoing routine health checkup from October 2013 to November 2014 at Xiangya Hospital, Central South University in Changsha, Hunan Province of China. Details of the study design and population have been published previously.[ Briefly, the routine health checkup included anthropometric and basic clinical examinations, as well as biochemical and imaging tests. Ethical approval for this study was provided by the Ethics Committee of Xiangya Hospital, Central South University (reference number: 201312459). All subjects gave their written informed consent before participating in the study.

The subjects meeting the following criteria were qualified for the present study: aged 40 years or order; underwent abdominal ultrasound examination; availability of basic characteristics (e.g., age, gender, body mass index [BMI], alcohol drinking status, smoking status and physical activity); completion of the semiquantitative food frequency questionnaire (SFFQ); and underwent blood biochemistry test.

Of the 14,715 participants who underwent routine health examinations during the targeted period, 6240 met the inclusion criteria. Then 795 subjects were excluded due to the following reasons: with alcoholic fatty liver disease (n = 786); with severe elevation of serum alanine aminotransferase (n = 9; normal range: 0–40 U/L, ≥200 U/L was defined as severe elevation). Eventually, 5445 subjects were qualified for the final analysis.

Assessment of NAFLD

All subjects in this study had underwent abdominal ultrasound examinations. According to the definition of NAFLD, subjects who met the following requirements were considered as NAFLD patients[: imaging evidence of hepatic steatosis; no excessive alcohol consumption (<140 g/wk for men and <70 g/wk for women); no specific diseases (e.g., parenteral nutrition, viral hepatitis, drug hepatitis, Wilson disease, etc) that can lead to NAFLD. The diagnosis of fatty liver was assessed by experienced and trained radiologists who were blinded to the aim of the present study and the subjects’ clinical diagnosis. Ultrasonographic diagnosis of hepatic steatosis was determined by the presence of at least 2 of 3 abnormal findings on abdominal ultrasonography: diffusely increased echogenicity (“bright”) liver with liver echogenicity greater than kidney or spleen, vascular blurring, and deep attenuation of ultrasound signal.[

Dietary assessment

Dietary iron intake was evaluated by using a validated SFFQ, which has been used in our previous studies,[ performed on the same day with anthropometric measurements, biochemical tests, and abdominal ultrasound for each participant. Additionally, the SFFQ was carried out before the participants were informed of the results of the examination. The reproducibility of the evaluating of dietary iron intake by the SFFQ, which was evaluated by comparing two same SFFQs administered at least 1 week apart, was 0.67. The validity of the SFFQ was examined by comparing its results with that of the 24-hour dietary recall method for the same population, and the correlation between the SFFQ and the 24-hour recall method on the measurement of iron intake was 0.40. Briefly, this SFFQ consisted of 63 commonly consumed local food items. Participants were requested to answer their average consumption frequency (none, once a month, 2–3 times/mo, 1–3 times/wk, 4–5 times/wk, once a day, 2 times/d, or 3 times or above/d) and the average consumption amount of each food item at one time (<100 g, 100–200 g, 201–300 g, 301–400 g, 401–500 g, and >500 g). Pictures of food samples at standard weight were provided to the participants to help them make choices quickly and accurately. For each food item, the reported frequency and average consumption amount were converted to a daily consumption. Iron intake was calculated by multiplying the daily consumption of edible parts for corresponding food item (g/d) by the assigned mean iron value (mg/100 g) according to the Chinese Food Composition Table.[ Total iron intake (mg/d) was determined by summing up the iron contents of all food items.

Assessment of nondietary exposures

The BMI was calculated based on each subject's weight and height measurements. All blood samples were drawn after a 12-hour overnight fast and were kept at 4°C till analysis. The glucose oxidase enzyme method was adopted to detect the blood fasting glucose, and an electronic sphygmomanometer was used to detect the blood pressure. A subject would be diagnosed as a diabetes patient if his/her concentration of fasting blood glucose ≥7.0 mmol/L or if he/she was currently taking any antidiabetic drugs. Hypertension was defined as the systolic blood pressure ≥140 mm Hg or the diastolic blood pressure ≥90 mm Hg or who were currently undergoing antihypertensive treatments.

Demographic characteristics and lifestyle habits, including age, gender, education level, cigarette smoking status, alcohol drinking status, and physical activity were collected through a standard questionnaire. Besides, subjects were asked whether they were taking any oral nutritional supplementation or not. Nutritional supplementation was simply classified into 4 groups: calcium, vitamins, minerals, and others.

Statistical analysis

The continuous data were expressed as mean ± standard deviation, while the category data were expressed in proportion (percentage). The dietary iron intake was classified into 5 categories according to the quintile distribution: ≤16.68 mg/d, 16.69 to 22.84 mg/d, 22.85 to 29.60 mg/d, 29.61 to 40.78 mg/d, and ≥40.79 mg/d. Statistical differences in continuous data were determined by using the 1-way analysis of variance (normally distributed data) or the Kruskal–Wallis H test (nonnormally distributed data), and differences in category data were determined by the Chi-squared test. The logistic regression models were used to examine the association between dietary iron intake and the prevalence of NAFLD. The odds ratio (OR) and the related 95% confidence interval (CI) were calculated for each category of dietary iron intake, and the lowest quintile was considered as the reference. Covariates were selected based on similar studies published previously.[ Three models were established for multivariate analysis: model 1 was adjusted for dietary energy intake (quintiles); model 2 was adjusted for age (continuous data), gender (male, female), BMI (≥28 kg/m2, <28 kg/m2) and dietary energy intake (quintiles); model 3 was adjusted for additional factors including education level (with or above high school background or not), smoking status (yes or no), hypertension (yes/no), diabetes (yes/no), activity level (quintiles), dietary fat intake (quintiles), dietary fiber intake (quintiles), and nutritional supplementation (yes or no) on the basis of model 2. Tests for linear trends were conducted using logistic regression with a median value of dietary iron intake level in each category. Subgroup analysis was conducted by repeating the multivariate analysis in the male and female population, respectively; then, the dose–response relationship between the dietary iron intake and the prevalence of NAFLD was evaluated using the restricted cubic spline regression with 4 knots defined by the quintile distribution of dietary iron intake.[ In addition, sensitivity analysis was conducted by excluding patients with hypertension or diabetes as suggested by previous studies.[ All data analyses were performed using Statistical Package for the Social Sciences (SPSS) version 21.0 (SPSS Inc, Chicago, IL) and STATA 11.0 (StataCorp LP, College Station, TX). P < .05 was regarded as statistically significant.

Results

A total of 5445 subjects (2590 males, 2855 females) aged from 40 to 84 years old (average 53.19 ± 7.50 years) were included in the current analysis. The overall prevalence of NAFLD was 36.9% (44.2% in males and 30.2% in females). Referring to the basic characteristics related to the NAFLD status, as shown in Table 1, subjects with (n = 2007) or without NAFLD (n = 3438) were different in terms of gender ratio, BMI, education level, smoking ratio, hypertension, diabetes, activity level, dietary fat intake, dietary fiber intake, and nutritional supplementation. The information according to the quintiles of dietary iron intake of the included subjects are listed in Table 2, and significant differences were observed in gender ratio, BMI, education level, smoking ratio, diabetes, activity level, dietary energy intake, dietary fat intake, dietary fiber intake, and nutritional supplementation.

Table 1

Basic characteristics among 5445 participants according to nonalcoholic fatty liver disease status.

Table 2

Basic characteristics among 5445 participants according to quintiles of dietary iron intake.

The association between dietary iron intake and the prevalence of NAFLD is shown in Table 3. Compared with the lowest quintile, the ORs adjusted by dietary energy intake (model 1) were 1.33 (95% CI: 1.07–1.64), 1.80 (95% CI: 1.41–2.29), 2.11 (95% CI: 1.60–2.80) in the 3rd, 4th, and 5th quintile of iron intake respectively (P-value for trend <.001). With further adjustment of age, gender, and BMI on the basis of model 1 (model 2), the OR for prevalence of NAFLD increased by 1.66 times in the 4th quintile (OR = 1.66, 95% CI: 1.28–2.15) and 1.90 times in the highest quintile (OR = 1.90, 95% CI: 1.41–2.57) compared with the reference (P-value for trend <.001). The multivariable adjusted ORs in model 3 still suggested a positive association between dietary iron intake and the prevalence of NAFLD (OR = 1.77, 95% CI: 1.20–2.61 in the 4th quintile; OR = 2.10, 95% CI: 1.30–3.41 in the highest quintile; P-value for trend .001). Furthermore, dietary iron intake, which was similar to the daily intake level of iron in other studies conducted in China,[ was positively associated with the OR for prevalence of NAFLD even below the tolerable upper intake level of the Chinese population (42 mg/d)[ in a dose–response relationship manner (Fig. 1, test for trend P < .001).

Table 3

Relationship between dietary iron intake and prevalence of nonalcoholic fatty liver disease.

Figure 1

Dose–response relationship between dietary iron intake level and the odds ratio for nonalcoholic fatty liver disease. CI = confidence interval, NAFLD = nonalcoholic fatty liver disease, OR = odds ratio, UL = tolerable upper intake level.

The results for subgroup analysis of gender are presented in Table 3. It was found that the positive association between dietary iron intake and the prevalence of NAFLD remained significant in the male population (model 1: OR = 4.07, 95% CI: 2.66–6.23 in the 5th quintile, P for trend <.001; model 2: OR = 3.64, 95% CI: 2.31–5.73 in the 5th quintile, P for trend <.001; model 3: OR = 3.76, 95% CI: 1.85–7.64 in the 5th quintile, P for trend .001), but not in the female subgroup.

Sensitivity analysis by excluding participants with hypertension or diabetes (n = 3, 471) did not change the results materially (see Table S1 and Fig. S1, Supplemental Content 1 and Content 2, which illustrate the sensitivity analysis by excluding participants with hypertension or diabetes).

Discussion

Main findings

The overall prevalence of NAFLD was 36.9%, which was similar to the prevalence presented in some previous studies targeted the middle-aged and elderly Chinese population.[ A positive association between dietary iron intake and NAFLD was observed in this study in a dose–response relationship manner. Our findings were independent of the effects of major confounders, including dietary energy intake, age, gender, BMI, educational level, smoking status, hypertension, diabetes, activity level, dietary fat intake, dietary fiber intake, and nutritional supplementation. However, such association only existed in the male population, but not in the female population.

Comparisons with previous studies

While previous in vivo studies reported that excessive intake of dietary iron-induced liver inflammation and fibrogenesis in rat models,[ there is lack of data on the association between iron intake and NAFLD for human beings. A small study (27 patients) conducted by Yamamoto and colleagues showed that dietary restriction on the combination of calories, fat and iron could improve NAFLD.[ However, it remained unclear whether such effect was contributed solely by the restriction on iron. In another small-sample case–control study (215 NAFLD patients and 215 healthy control population) conducted by Zheng and colleagues, the relationship between dietary iron intake and NAFLD was tested without adjusting for activity level and nutrition supplementation, and the results were consistent with ours.[ Furthermore, a randomized controlled trial conducted by Lachili and colleagues showed that high iron supplementation, especially the combination of iron and vitamin C supplementation, might lead to increased lipid peroxidation in 2 groups of 27 pregnant women.[

Potential explanations

It has been generally recognized that insulin resistance and oxidative stress are the key factors in the development and progression of NAFLD.[ Excessive accumulation of iron may exert effects on the synthesis and secretion of insulin and interfere with insulin receptor, thus leading to insulin resistance.[ Consequently, the glucose production is inhibited whereas the fatty acid synthesis is retained, which may eventually lead to NAFLD.[ Another important mechanism linking iron with NAFLD might be oxidative stress.[ Excessive iron in liver can generate reactive oxygen species through Fenton reaction and activate Kupffer cells; thus, it may place hepatocytes under oxidative stress and release profibrogenic mediators, then trigger damage to the liver.[ The augmented generation of reactive oxygen species can also lead to lipid peroxidation by activating stellate cells, and in turn result in hepatic inflammation. In addition, reactive oxygen species tend to inhibit the secretion of very low-density lipoprotein in liver and contribute to the accumulation of liver fat.[ Furthermore, hepatic copper concentrations was found to vary inversely according to iron status,[ and impaired bioavailability of copper has been proved to play an important role in the pathogenesis of NAFLD.[ On the one hand, low copper bioavailability can attribute to lipid metabolism and that it may therefore be related to the development of NAFLD.[ On the other hand, systemic copper deficiency could also lead to changes in mitochondrial morphology owing to its effects on respiratory chain physiology and function, which has been implicated in the pathogenesis of NAFLD.[

The gender-based difference observed in the present study might be explained by estrogen-related hormones, especially the estradiol.[ Previous studies have revealed that estrogen could modulate iron homeostasis by governing the expression of hepatic hepcidin.[ Under the circumstance of estrogen deficiency, the hepcidin level will increase, leading to tissue iron retention and eventually progressing to iron overload.[ In addition, it is deemed that estradiol and its derivatives have strong antioxidant capacity that can reduce the lipid peroxide levels in the liver and serum.[ Furthermore, estradiol has the ability to suppress the generation of iron-induced reactive oxygen species and the lipid peroxidation, and thus it may prevent hepatocytes from oxidative damage, inflammation, and cell death.[ However, further mechanistic clarification is still warranted.

Strengths and limitations

Several strengths of this study are noteworthy. Firstly, this is the 1st study that evaluated the association between dietary iron intake and the prevalence of NAFLD in a large sample (5445 subjects). Secondly, the present results were independent of the effect of the major confounders, which improved the reliability of the findings. Thirdly, the dietary iron intake was measured by a validated SFFQ, which is an effective method to study dietary intake over the past 1 year. Thus, it might better represent the average intake of iron compared to the measurement of instant intake.

Nevertheless, several limitations should also be acknowledged. First, liver biopsy, which remains the gold standard for the diagnosis of NAFLD, was not available in the present study. However, a requirement of liver biopsy to define NAFLD is often impractical for several reasons (e.g., cost, access, and acceptability).[ Instead, determine fatty liver using ultrasound has been widely used in both clinical and population settings owing to its noninvasive nature and acceptable degree of diagnostic accuracy for steatosis.[ Second, due to the cross-sectional nature, it is unable to infer the causality between dietary iron intake and the incidence of NAFLD. In this regard, further prospective studies should be carried out to confirm the causal relationship. Third, since subjects may take iron supplementation of different types, or in different durations or doses, it is difficult to quantify the specific iron intake from nutritional supplements, which may influence the results.

Conclusion

Subjects with higher dietary iron intake, even below the tolerable upper intake level, were subject to a higher prevalence of NAFLD in a dose–response relationship manner. However, such association may only exist in males, but not in females, which still warrants further studies to explore.

Acknowledgment

The authors appreciate the support of Orthopedics Research Institute of Xiangya Hospital.

Author contributions

Conceptualization: Zidan Yang, Tubao Yang.

Data curation: Xiaoxiao Li, Dongxing Xie, Yilun Wang.

Formal analysis: Jing Wu.

Methodology: Zidan Yang, Jing Wu, Tubao Yang.

Writing – original draft: Zidan Yang.

Writing – review & editing: Tubao Yang.

Tubao Yang orcid: 0000-0001-8762-1229.