The relaTionship beTween serum ferriTin levels and serum lipids and hdl funcTion wiTh respecT To age and gender

Elevated serum ferritin (SFer) levels have been associated with chronic diseases such as coronary heart disease and diabetes mellitus type 2. the aim of this study was to examine the relationship between SFer levels and serum lipid parameters, and how this relation changes in terms of age and gender. additionally, we investigated a possible relationship between SFer levels and high-density lipoprotein (Hdl) function. SFer levels and lipid panel (total cholesterol (tc), triglyceride (tg), low-density lipoprotein-cholesterol (ldl-c) and Hdl-c) of 4205 people (3139 women, 1066 men) were examined retrospectively. Study population was classified according to age and gender. Separately, 100 subjects (52 women, 48 men) were randomly recruited to investigate the relation between SFer levels, and Hdl dependent paraoxonase-1 (PON1) and arylesterase (ARE) activities. In all age groups, women’s SFer levels were found to be significantly lower and HDL-C levels significantly higher compared to men. In the 50-70 ages range, TC and LDL-C levels of women were found to be significantly higher than those of men (P < 0.01). SFer levels tended to increase with age in women. Correlation analyses revealed a negative correlation between levels of SFer and Hdl-c, while positive correlations existed between levels of SFer, and TC, TG and LDL-C. There was no significant correlation between SFer levels and PON1 or ARE activities. The finding that increased SFer levels are accompanied by increased serum tc, tg and ldl-c levels may help us to explain the increased risk of metabolic disorders and cardiovascular disease in postmenopausal women.


The relaTionship beTween serum ferriTin levels and serum lipids and hdl funcTion wiTh respecT To age and gender
HamIt YaSar EllIdag 1 *, ESIN ErEN 1 , mEHmEt akdag 2 , OzlEm gIraY 1 , kEmal kIraz 3 , NEcat YIlmaz Elevated serum ferritin (SFer) levels have been associated with chronic diseases such as coronary heart disease and diabetes mellitus type 2. the aim of this study was to examine the relationship between SFer levels and serum lipid parameters, and how this relation changes in terms of age and gender.additionally, we investigated a possible relationship between SFer levels and high-density lipoprotein (Hdl) function.SFer levels and lipid panel (total cholesterol (tc), triglyceride (tg), low-density lipoprotein-cholesterol (ldl-c) and Hdl-c) of 4205 people (3139 women, 1066 men) were examined retrospectively.Study population was classified according to age and gender.Separately, 100 subjects (52 women, 48 men) were randomly recruited to investigate the relation between SFer levels, and Hdl dependent paraoxonase-1 (PON1) and arylesterase (ARE) activities.In all age groups, women's SFer levels were found to be significantly lower and HDL-C levels significantly higher compared to men.In the 50-70 ages range, TC and LDL-C levels of women were found to be significantly higher than those of men (P < 0.01).SFer levels tended to increase with age in women.Correlation analyses revealed a negative correlation between levels of SFer and Hdl-c, while positive correlations existed between levels of SFer, and TC, TG and LDL-C.There was no significant correlation between SFer levels and PON1 or ARE activities.The finding that increased SFer levels are accompanied by increased serum tc, tg and ldl-c levels may help us to explain the increased risk of metabolic disorders and cardiovascular disease in postmenopausal women.k e y w o r d s: ferritin, lipid metabolism, paraoxonase, arylesterase, Hdl, ldl.F ollowing the transition to aerobic respira tion, transition metal elements, through evo lutionary processes, have enabled delivery of oxygen to peripheral tissues of increasingly complex organisms.Since iron reacts with oxygen spontane ously (without energy consumption), as well as ex change electrons easily, it functions as an O 2 carrier in many organisms and is essential for these, inclu ding humans [1,2].Additionally, iron is a part of the structures of many proteins, such as haemoglobin, myoglobin and cytochromes, and functions as a co factor in various enzymes in numerous biochemical reactions [3].Nonetheless, excessive accumulation of iron in the organism leads to cellular dysfunction and death due to formation of free radicals and li pid peroxidation [4].Therefore iron metabolism re quires strict control.There is no direct physiological mecha nism in humans to excrete iron and its excess is stored in ferritin.Ferritin is an intracellular pro tein of 450 kDa molecular weight, consisting of 24 subunits.Ferritin serves to store iron in a safe form [5].
There are many publications that have found that elevated serum ferritin (SFer) levels are linked with chronic diseases, such as coronary heart disea se and diabetes mellitus [610].Sullivan et al. have proposed for the first time that in postmenopausal women, increased body iron stores may lead to an increase in cardiovascular disease (CVD) incidence [11].High SFer levels were shown to be associated with increased risk of myocardial infarction and coronary artery disease related mortality [12].Fur https://doi.org/10.15407/ubj88.06.076 thermore, links have been found between elevated SFer levels and such disorders as insulin resistance, hypertension and type2 diabetes mellitus [13,14].However, other studies have found no relationship between the level of SFer and risk of CVD [1518].
High serum levels of total cholesterol (TC), triglyceride (TG) and lowdensity lipoprotein cho lesterol (LDLC), and low levels of highdensity li poprotein cholesterol (HDLC) are among the major risk factors for CVD [19,20].There are animal and human studies indicating a link between iron and lipid metabolisms.For example, an iron-deficient diet in rats exerted broad influence on gene expres sion patterns of the lipid metabolism compared with a normal diet [21].According to two studies that examined the relationship between lipid parameters and iron deficiency anemia in women, significant variations were observed in lipid parameters before administrating iron therapy, which were partly al leviated afterwards [22,23].
Despite the cited research, the number of stud ies investigating the relation between the serum lipid panel and SFer levels in large sample sizes is limited.In particular, the categorical influence of age and gender on this association remains unknown.Fur thermore, the relationship between serum lipid and SFer levels has been probed within certain popula tions, while variations in genetic makeup and dietary habits are known to affect the lipid panel and iron stores [24].An analysis directly comparing the SFer and serum lipid levels in the Turkish population has not yet been performed.Therefore, in our study we aimed to investigate the relation between ferritin and lipid parameters, and whether and how this relation changes in terms of age and gender.

materials and methods
Study design.A total of 4205 subjects (3139 women, 1066 men, mean age: 43), who had visited various clinics of Antalya Training and Research Hospital between January 2013 and January 2015 were included in this study.The study group was se lected by examining the patient data registered in the laboratory information system (LIS) of the hospital.Accordingly, all inpatients (from the departments of cardiology, cardiovascular surgery, general surgery, neonatology, paediatrics, oncology, etc. and from the cardiac intensive care and intensive care units) were excluded from the study.Furthermore, children under 20 and people over 70 years of age, pregnant women, oncology patients, thalassemia patients, and patients with chronic liver disease or chronic kidney disease were excluded from the patients involved in the study.The lipid panel (TC, TG, LDLC and HDLC levels) and SFer levels of the study group had been determined, which was analyzed retro spectively.Throughout the said interval, the labora tory equipment and methods of measurement have remained unchanged.
An additional 100 subjects (52 women, 48 men, mean age: 46), whose SFer and HDLC parameters had been measured in our laboratory, were randomly selected for PON1 and ARE activity analysis.
The is then turned into a chromophore by peroxi dase, which is read at 660/800 nm.The increase in absorban ce is thus proportional to the triglyceride content in the sample.In cholesterol analysis, free cholesterol is formed from cholesterol esters by cholesterol estera se (EC 3.1.1.13)reaction.In the presen ce of both cholesterol oxidase (EC 1.1.3.6) and peroxidase, the resulting free choleste rol leads to a spectrophotometric absorbance change that is used for determining the cholesterol concentration in the sample.
HDLC was analyzed with Beckmann Coulter commercial reagents using an AU5800 AU Analyzer (Beckman Coulter, Inc., CA, USA).Test principle is essentially based on immunoinhibition and enzyma tic colour reaction.The antibetalipoprotein antibody binds to lipoproteins other than HDL (LDL, VLDL and chylomicrons).The enzyme activities of these antigenantibody complexes are inhibited by the addition of another reagent.HDLC concentration is determined with an enzyme chromogen system.Intra-day and inter-day coefficients of variation of TC, TG and HDLC were found to be less than 4%.LDLC concentration was calculated by the Friede wald formula [LDLC = TC -HDLC -(TG/5)] [25].
measurement of SFer levels.SFer levels were measured with the chemiluminescence method using a UniCel® DxI800 Analyzer (Beckman Coul ter, Inc., CA, USA).The ferritin assay is a twosite immunoenzymatic ("sandwich") assay.A sample is added to a reaction vessel with goat antiferritin alkaline phosphatase conjugate, and paramagnetic particles coated with goat antimouse: mouse anti ferritin complexes.Serum or plasma ferritin binds to the immobilized monoclonal antiferritin in the solid phase, while the goat antiferritin enzyme conjugate reacts with different antigenic sites on the ferritin molecules.This "sandwich" conformation serves to maximise specificity.Separation in a magnetic field and washing removes materials not bound to the solid phase.This washing step eliminates any potential crossreactants in the sample, as well as the excess antiferritinenzyme conjugate, which could otherwise increase the background signal and reduce sensitivity.A chemiluminescent substrate, LumiPhos® 530, is added to the reaction vessel and light genera ted by the reaction is measured with a luminometer.The photon production is proportional to the amount of ferritin in the sample.Intraday and inter-day coefficients of variation of ferritin were found to be less than 6%.
measurement of PON1 and arE enzyme activities in serum.PON1 and ARE enzyme activities were measured using commercially available kits (Rel Assay Diagnostics®, Gaziantep, Turkey).PON1 activity measurement method is fully automated and involves an appropriate Tris buffer containing cal cium ions, which is a cofactor of the PON1 enzyme.A linear increase in the absorbance of pnitrophenol, produced from paraoxon, is followed in a kinetic measurement mode.The nonenzymatic hydrolysis of paraoxon is subtracted from the total rate of hy drolysis.The molar absorptivity of pnitrophenol is 18.290 M −1 •cm −1 and one unit of paraoxonase activity is equal to 1 mol of paraoxon hydrolyzed per liter per minute at 37 °C [26].Phenylacetate was used as a substrate to measure ARE activity.PON1, present in the sample, hydrolyses phenylacetate to its products , phenol and acetic acid.The phenol produced is colorimetrically measured via oxidative coupling with 4aminoantipyrine and potassium ferricya nide.Nonenzymatic hydrolysis of phenyl acetate is then subtracted from the total rate of hydrolysis.The molar absorptivity of the coloured complex is 4,000 M −1 •cm −1 and one unit of arylesterase activity is equal to 1 mmol of phenylacetate hydrolyzed per litre per minute at 37 °C [27].
Statistical analysis.SPSS (version 11.0 for Windows®) statistical software was used for the evaluation of the data.The distribution of the col lected data was examined by KolmogorovSmirnov test.Comparison of data showing nonnormal dis tribution was performed via MannWhitney U test.Spearman correlation test was used for the correla tion analysis.P < 0.05 was considered as statistically significant.

results and discussion
According to the demographic data, the avera ge age was 43.34 and women accounted for 70% of the study group (Table 1, a).Once the study group was categorized by gender, the women were found to be younger.There was no significant difference be tween men and women in terms of TC and LDLC; however TG levels were significantly higher in men.As expected, compared to men, SFer levels were significantly lower and HDL-C concentrations were significantly higher in women (Table 1

, B).
The subjects in the study population were di vided into "decade subgroups" between 20 and 70 (2029, 3039, etc.), and lipid panel and SFer levels of these subgroups were analyzed for each gender.In all decade subgroups women's SFer levels were significantly lower than men's (Table 2).In the 20-29 age subgroup, women's HDLC concentrations were significantly higher and TG levels were significantly lower than men's.Women in the 3039 age subgroup had lower TC, TG and LDLC levels, and higher HDLC concentrations compared to men.In the 40 49 age subgroup, women had higher HDLC and lower TG levels compared to men.In women older than 50 years old, TC, LDLC and HDLC levels were higher compared to men (Table 2).
The correlation analysis between SFer levels and lipid parameters in the study group has revealed positive correlations between SFer and TC, TG and LDLC levels, and a negative correlation between SFer and HDLC.Once women and men were sepa rately evaluated, women had positive correlations between SFer and TC, TG and LDLC levels, and a negative correlation between SFer and HDLC.In men, there was a negative correlation between SFer and HDLC levels, and a positive correlation be tween SFer and TG (

t a b l e 1. B -distribution of parameters between genders
We divided the study group according to cut-off levels for the dyslipidemia classification recommended by the National Cholesterol Education Program and the Adult Treatment Panel III [28].Accordingly, subjects with high TC, TG and LDLC levels had higher SFer levels than those with normal lipid pa rameters.Conversely, both women and men with low HDLC levels had higher SFer levels (Table 4).
Being an important indicator of HDL func tion, PON1 and ARE enzyme activities and their relation with SFer levels were also assessed in our study.For this purpose, we applied PON1 and ARE analyses to 100 randomly selected subjects' sera that already had SFer and HDLC tests in our laboratory.Interestingly, despite there was a negative correla tion between SFer and HDLC levels (r = -0.446,P < 0.01) in these patients, in agreement with the correlation in the main study group, there was no correlation between SFer and PON1 or ARE activity levels.Likewise, there was no statistically significant relationship between the gender of these subjects and PON1 or ARE activities (Table 5).
The results of our study in which we investi gated the relation between ferritin and lipid levels can be summarized as follows: I. In all age groups, SFer was significantly lower in women compared to men (Table 1).
II.In all age groups, despite the variation in other lipid parameters, HDL-C levels were signifi cantly higher in women than those in men (Table 1).
III.In women older than 50, TC and LDLC levels were significantly higher compared to men (Table 2).
IV.When correlation analyses between SFer and lipid parameters in the study group were per formed, we found a strong correlation between SFer and HDLC levels (r = -0.247,P < 0.01).When the study group was divided by gender, this correlation was stronger in men (Table 3).
VI.No correlation was detected between SFer levels and PON1 or ARE activities, which are im portant indicators of HDL function.
The first observation that showed an association between serum ferritin levels and CVD is a large scale study (1931 Finnish men) conducted in 1992 [12].Salonen et al. have found that the risk of acute myocardial infarction was 2.2 times higher in men whose SFer levels were higher than 200 ng/ml.In the same study, high SFer levels were accompanied by high LDLC levels.A subsequent study conducted in Finland has confirmed these findings [29].In another study involving 11,471 women, postmenopausal women with higher SFer levels were associated with increased risk of ischemic stroke [30].Wolff  SFer levels could be a marker indicating early coro nary artery stenosis [32].
There is research suggesting that high SFer levels are also linked to metabolic syndrome and dia betes.Jehn et al. have found that in 6,044 Ameri can adults, there was an association between SFer levels and prevalence of metabolic syndrome in both men and women [33].Similar findings have been reported in a study of 8,441 people from different provinces of China [34].Additionally, a disorder   [24,35,36].The molecular mechanism behind the link be tween increased iron overload and metabolic syn drome, cardiovascular diseases and other diseases is not fully understood.Among proposed pathophysio logical mechanisms, the most possible one is the re lationship between iron and oxidative stress.Iron is a strong prooxidant transition element.Due to this feature, it can cause cellular damage in different tis sues by generating reactive oxygen species [24,37].Another possible mechanism could be based on Hevi et al.'s research [38].This experimental study has shown that in hepatocytes, ferritin inhibits apoli poprotein B (apoB) secretion in a posttranslational manner without affecting the secretion of apolipo protein AI, albumin or factorXIII.Interestingly, the researchers have demonstrated that in these cells, triglyceride synthesis is not reduced and cellular tri glyceride levels are in fact increased.As a result, our theory on the basis of this work is that by inhibiting apoB secretion and increasing triglyceride synthesis, ferritin increases intracellular lipid levels and leads to damage in hepatocytes, thus generating a similar pathology seen in DIOS.
Perhaps the most important finding of our study is that SFer levels in women, although remaining lower than those in men, increased with advan cing age unlike in men (Table 2; Figure).It should be noted that levels of TC, TG and LDLC also tended to increase with age, and peaked at the 5059 ages subgroup (Table 2).Many studies have reported a higher risk of CVD especially in postmenopausal women [39].This phenomenon has been attributed to the cessation of the menstrual cycle and the decline of related hormones.However, CVD risk is not di minished in women receiving hormone replacement therapy [39].Increase in iron load in postmenopausal women has been shown both in our study and others.High iron stores may be primarily responsible for the elevated CVD risk in women and ferritin could be an important marker for monitoring CVD risk.
A study conducted in patients with hereditary hemochromatosis has shown that phlebotomy re duced TG levels without altering serum glucose or cholesterol levels [40].Similarly, another study has found that phlebotomy reduced serum glucose and glycated hemoglobin levels, LDL/HDL ratio, and systolic blood pressure in patients with metabolic syndrome [41].Therefore, by keeping SFer levels low, phlebotomy can be a simple and effective pro phylactic measure for both postmenopausal women and individuals that bear risk factors for metabolic syndrome and CVD.If such positive effects of phle botomy can be proved by other studies in the future, phlebotomy not only will be used as a preventive tool for common chronic disorders such as metabolic syndrome, but also will help maintain blood supplies required for transfusion.In all age subgroups of women, HDLC levels were significantly higher compared to men in our study.On the other hand, it has been recently ac knowledged that HDL functionality is more critical than its concentration [19,42,43].PON1 and ARE, which are important indicators of HDL function, are multifunctional enzymes that have antioxidant, antiatherogenic, lactonase and homocysteinethio lactonase activities [44,45].In our smaller group of 100 individuals, although there was a negative correlation between HDLC and SFer levels, there was no correlation between SFer levels and PON1 or ARE activities.Moreover, there was no difference in PON1 and ARE activities between genders.There fore, although HDL levels in women were high in all age groups compared to men, no difference in HDL function was noted in the study.
The main limitation in our study was due to the fact that ferritin is a positive acute phase reac tant.Unfortunately, we were unable to determine the presen ce of acute or chronic inflammation in our study group in order to exclude such individuals.Similarly, even though patients that suffer acutely from CVD (inpatients) were left out of the study group, followup patients with chronic CVD could not be excluded.Also, we could not factor socioeco nomic status and dietary habits, which can influen ce serum ferritin levels and iron stores, into our catego rization of the study population.
In summary, we have examined the relation ship between SFer levels and serum lipid parame ters in each gender with respect to age in this study.In women in particular, increasing serum ferritin levels with advancing age, which is accompanied by higher serum TC, TG and LDL levels, may make postmenopausal women more prone to develop CVD distribution of SFer levels between age subgroups: a -men.B -Women Age subgroups Serum ferritin levels and meta bolic diseases.Salonen et al.'s study has es tablished a cut-off value of ferritin for men (a SFer level above 200 ng/ml increases risk of CVD by 2.2 fold in men), but a corresponding ferritin reference for women is lacking [12].The need for such a study is evident, as SFer levels in men and women exhibit physiological variance.Thus, SFer levels could be monitored in postmenopausal women and, if neces sary, the beneficial effects of phlebotomy could be utilized in order to prevent chronic diseases such as CVD and metabolic syndrome.Certainly, further largescale studies are needed to support our hy potheses.

acknowledgements
We thank Suha Sayrac for his assistance during revision of the manuscript.