J

E

B

DEa

b

c

d

e

a

ARA

KMICHF

I

bicsp

DmrS

R

h0

ARTICLE IN PRESSG ModelTEMB-25514; No. of Pages 7

Journal of Trace Elements in Medicine and Biology xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Journal of Trace Elements in Medicine and Biology

jou rn al homepage: www.elsev ier .de / j temb

pidemiology

iomarkers of iron status and trace elements in welders�

ag G. Ellingsena,∗, Maxim Chashchinb,1, Balazs Berlingera, Tobias Konzc,vgenij Zibarevb, Jan Aasethd, Valery Chashchinb,e, Yngvar Thomassena

National Institute of Occupational Health, P.O. Box 8149 Dep, N-0033 Oslo, NorwayNorthwest Public Health Research Centre, 2-Sovetskaya 4, St. Petersburg 191036, RussiaDepartment of Physical and Analytical Chemistry of the University of Oviedo, ES-33006, SpainDepartment of Medicine, Innlandet Hospital Trust, N-2226 Kongsvinger, NorwayNorth-Western State Medical University, St. Petersburg 191015, Russia

r t i c l e i n f o

rticle history:eceived 20 December 2013ccepted 3 March 2014

eywords:anganese

ronarbohydrate deficient transferrinepcidinerritin

a b s t r a c t

Iron status was studied in 137 welders exposed to a geometric mean (GM) air concentration of 214 �g/m3

(range 1–3230) of manganese (Mn), in 137 referents and in 34 former welders. The GM concentrations ofS-ferritin were 119 (3–1498), 112 (9–1277) and 98 (12–989) �g/L (p = 0.24) in the three groups, respec-tively. Also the GM concentrations of S-hepcidin were not significantly different between the groups(8.4 �g/L (2.8–117); 6.6 �g/L (1.8–100); 6.5 �g/L (1.2–22)) (p = 0.22). Multiple linear regression analysisincluding all welders and referents showed an increase in the concentration of S-ferritin associated withhaving serum carbohydrate deficient transferrin (S-CDT) above the upper reference limit of ≥1.7%, indi-cating high alcohol consumption. Serum C-reactive protein was not associated with exposure as welders,but an association with S-ferritin was shown. The GM S-ferritin concentrations among all welders andreferents with S-CDT ≥ 1.7% were 157 �g/L (95% CI 113–218) as compared to 104 �g/L (95% CI 94–116)(p = 0.02) in those with S-CDT < 1.7%. The GM concentrations of Mn in biological fluids were higher in the

welders as compared to the referents, while S-Fe, S-Co and B-Co were statistically significantly lower. Thiscould suggest a competitive inhibition from Mn on the uptake of Fe and Co. Increasing concentrationsof S-CDT was associated with higher S-Mn, S-Fe and B-Co in the multiple linear regression analysis. Theassociation between S-CDT and S-Fe remained when all subjects with high S-CDT (≥1.7%) were excluded,suggesting increased uptake of Fe even at lower alcohol consumption.

© 2014 Published by Elsevier GmbH.

ntroduction

Potentially high manganese (Mn) content in welding fumes maye a risk factor for neurological disturbances in welders [1]. Weld-

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of i(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

ng fume consists of primary particles that are agglomerated intohainlike structures, containing chemically complex compoundsuch as KMnF3, Fe3O4, MnFe2O4 or K2MnO4 [2]. The predominantarticle sizes of welding aerosols are below 1 �m in aerodynamic

� The study was supported by Grant Number W81XWH-05-1-0239 from the U.S.epartment of Defense, United States Army Medical Research and Material Com-and. Its contents are solely the responsibility of the authors and do not necessarily

epresent the official views of the United States Department of Defense, Unitedtates Army Medical Research and Material Command.∗ Corresponding author. Tel.: +47 23195377; fax: +47 23195377.

E-mail address: dag.ellingsen@stami.no (D.G. Ellingsen).1 Current address: City Centre of Occupational Health, St. Petersburg 191014,ussia.

ttp://dx.doi.org/10.1016/j.jtemb.2014.03.004946-672X/© 2014 Published by Elsevier GmbH.

diameter, thus the particles can easily penetrate into the deeperparts of the lung [3,4]. The welding fume content of iron (Fe) is sub-stantially higher than Mn in both manual and gas metal arc welding[5].

Mn and Fe share several transporters in animals and humans[6]. Non-heme Fe(II) and Mn(II) are absorbed in duodenal entero-cytes through the divalent metal transporter 1 (DMT1). The cellularexport of Fe(II) through ferroportin has also been proposed forMn(II) [7,8]. Fe(III) and Mn(III) bind to transferrin and are takenup by cells through high affinity transferrin receptor (TfR) medi-ated endocytosis. After acidification, both metals are transportedout of the endosome through DMT1 [9]. Hepcidin is important forthe Fe metabolism by regulating ferroportin, perhaps through Fesensors in the hepatocytes [10].

ron status and trace elements in welders. J Trace Elem Med Biol

Thus, there is no surprise that high exposure to Mn may alter Femetabolism. Some studies have addressed this issue. Oral admin-istration of high amounts of Mn resulted in lower serum Fe andhemoglobin concentrations in lambs and lower gastrointestinal

ING ModelJ

2 nts in

ahdFebaotrrtt

ekaDwrc

rsosofik

imiaesfe

ief[eeimea

M

ewIrmhrifr

ARTICLETEMB-25514; No. of Pages 7

D.G. Ellingsen et al. / Journal of Trace Eleme

bsorption of Fe in humans [11,12]. Brain Mn concentrations wereigher in Mn injected rats both at increased and at decreasedietary Fe as compared to the administration of normal dietarye [6]. Mn metabolism may also be altered by Fe deficiency, asvidenced by significantly higher concentrations of Mn in wholelood of Fe deficient women [13]. It also appears that Mn can inter-ct with basic regulatory systems of Fe metabolism, e.g. inhibitionf mitochondrial aconitase, probably by insertion of Mn instead ofhe labile Fe into its 4Fe-4S cluster [14], or by activation of the Feesponsive protein (IRP), perhaps by depleting Fe with concomitanteduction of intracellular ferritin [15]. A recent study suggestedhat Mn can competitively inhibit gastrointestinal Fe absorptionhrough DMT1 [16].

Welders are exposed to Mn and Fe by inhalation. Pulmonarypithelial cells and alveolar macrophages have on their surfaces alley transporters of Fe such as TfR, DMT1 and ferroportin, as wells intracellular ferritin [17]. However, the pattern of pulmonaryMT1 expression is not altered in Fe-deficient rats in contrast tohat is observed in enterocytes [18]. In contrast, Fe-loaded rats had

educed transport of Mn across the air–blood barrier after intratra-heal instillation [19].

It has long been recognized that high alcohol consumption canesult in liver Fe accumulation [20,21]. Studies of mice and rats havehown that high alcohol consumption results in down-regulationf hepcidin expression with a concomitant increase in the expres-ion of DMT1 and ferroportin in duodenum [22,23]. Hepatocyticxidative stress induced by alcohol has been proposed as a causeor reduced hepcidin expression [24]. Whether this may have anmpact on the transport of other divalent metal ions, is to ournowledge not known.

The potential for systemic Fe overload has been little studiedn welders. Case reports have suggested that body Fe overload

ay occur in welders’ siderosis [25]. However, welders had sim-lar serum ferritin concentrations when compared to referents,lthough the concentrations were associated with duration ofxposure [26]. That study also reported reduced concentrations ofoluble TfR in serum. A recent study did not show increased serumerritin concentrations during a one week follow up of long termxposed welders [5].

This study is part of a larger study of nervous system impairmentn welders. We have previously reported that these welders arexposed to high amounts of Fe, but mostly in a non-bioaccessibleorm, while the bioaccessibility was substantially higher for Mn27]. The aim of this study is to assess whether Fe status biomark-rs are altered in welders, to assess the impact of welding fumexposure on selected trace element concentrations and to study thempact of biomarkers of alcohol consumption and systemic inflam-

ation on the above biomarkers. Some of the participants were alsoxamined six years previously and results from that follow-up arelso presented.

aterials and methods

One hundred and forty-nine welders and 178 potential refer-nts were invited to participate in this cross-sectional study, ofhom 12 welders and 41 potential referents refused participation.

n all, 137 welders and 137 referents were included (participationates 91.9% and 77.0%, respectively). The study was restricted toen with at least one year of employment at one facility producing

eavy machinery or at two shipyards in St. Petersburg (Russia). The

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of i(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

eferents were turners/fitters. The predominant welding methodsn use were manual metal arc and gas metal arc welding. Criteriaor exclusion have been published [27]. Sixty-three welders and 65eferents had been examined around six years earlier [28].

PRESS Medicine and Biology xxx (2014) xxx–xxx

Also 34 former welders (four females) that had been diagnosedwith manganism caused by welding were examined, of whom 17had been examined around six years earlier [28]. The latter sub-jects belonged to a group of 27 patients examined at that time,of whom two had died, seven had moved out of the region andwere thus not invited to participate and one declined the requestto participate. Additional 20 patients were identified from thepatient records kept at the clinic of the Northwest Public HealthResearch Centre (NWPHRC). They were invited to participate, butthree subjects declined. Thus, the participation rate among the eli-gible patients was 89.5%. Background data of all participants arepresented (Table 1).

The Norwegian Regional Ethical Committee for MedicalResearch (REK2) approved the study that was also approved by theEthics Committee of the NWPHRC (St. Petersburg, Russia) and theOffice of Research Protection, US Army Medical Research and Mate-rial Command (Fort Detrick, MD, USA). Participation was strictlyvoluntary and informed written consent was obtained from all par-ticipants.

Collection of biological samples

Welders and referents went through a structured interview andcollection of biological samples at the occupational health clinicsof the respective plants, while the patients were examined at theNWPHRC.

First voided morning urine samples were collected in 10 mLSarstedt® polypropylene (PP) tubes (Sarstedt AG, Nümbrecht,Germany). Blood samples were collected from the cubital veinbetween 8.30 and 9.30 the same morning after cleaning of theskin. Whole blood for the determination of trace elements wascollected in 4 mL Lithium-Heparine Vacuette® vacutainers. NinemL Vacuette® vacutainers without additives (Greiner Labortech-nik GmbH, Austria) were used for the harvest of serum. After await of 30 min the tubes were centrifuged for 10 min at 1500 × g.Serum was pipetted into 1.5 mL Sarstedt® cryotubes (Sarstedt AG,Nümbrecht, Germany) for storage. The samples were kept frozenat −20 ◦C until analysis.

Ten referents refused to give a urine sample, while 14 refer-ents declined blood sampling. Three former and one current welderrefused to give urine samples. Due to lack of serum, trace elementsin serum was determined in 124 welders, 33 former welders and83 referents. For the same reason, serum hepcidin was measuredin 45 welders, 19 former welders and 35 referents.

Determination of biomarkers

Serum ferritin (S-ferritin) was measured by an immunoturbi-metric method using Advia 2400 CapillarysTM (Siemens HealthcareDiagnostics Inc. NY, US) at Fürst Medical Laboratory (Oslo, Norway).The method’s reproducibility was 2.3%.

Serum hepcidin (S-hepcidin) was determined by a quantita-tive ELISA sandwich immunoassay (BlueGene Biotech, Shanghai)at the University of Oviedo (Spain). The assay used a microtiterplate pre-coated with a monoclonal antibody specific for hepcidin.Standards and samples were added to the plate wells and finally, asecond polyclonal antibody conjugated to horseradish peroxidasewas added. The color change in the plate was measured spec-trophotometrically at 450 nm. The immunoassay was previouslycross-validated. The method’s DL was 0.1 ng mL−1 and the intra-assay CV was <3%. A more detailed protocol has been presented[29].

ron status and trace elements in welders. J Trace Elem Med Biol

Serum carbohydrate deficient transferrin (S-CDT) was measuredby capillary electrophoresis using CapillarysTM (Sebia Inc., Georgia,USA) at Fürst Medical Laboratory (Oslo, Norway). The method’slimit of detection (DL) was 0.4%. Details have been presented [30].

ARTICLE IN PRESSG ModelJTEMB-25514; No. of Pages 7

D.G. Ellingsen et al. / Journal of Trace Elements in Medicine and Biology xxx (2014) xxx–xxx 3

Table 1Background characteristics and airborne exposure data among the study participants.

Welders (N = 137) Former welders (N = 34) Referents (N = 137)

AMa Range AM Range AM Range

Age (yrs.)** ,*** 39.9 19–70 58.8 47–77 40.1 19–70Education (yrs.)* ,** ,*** 11.8 7–17 10.9 4–16 12.8 8–19Current smokers (%) 52.6 - 38.2 – 56.9 –Alcohol cons. (L/year)*,*** 5.1 0–47 6.0 0–56 3.1 0–35Years welding** 16.6 1–45 23.7 7–33 –Cessation of welding (yrs.) – – 10.2 4–17 – –HatchsolMn (�g/m3)b,d 22 <DL-425 – – – –Hatchnon-solMn (�g/m3)b,c 185 1–2860 – – – –HatchsolFe (�g/m3)b 13 <DL-183 – – – –Hatchnon-solFe (�g/m3)b 1505 11–10960 – – – –HatchsolCo (�g/m3)b 0.02 <DL-0.46 – – – –Hatchnon-solCo (�g/m3)b 0.29 <DL-3.5 – – – –

a Arithmetic mean.b Geometric mean.c No air samples in seven subjects.d Hatchsol: Soluble in Hatch solution, Hatchnon-sol: non-soluble in Hatch solution.

M(d

octamMs

M

NiSB1s(itmf(ssTtwS

arnnff10

* p < 0.05 between welders and referents.** p < 0.05 between welders and former welders.

*** p < 0.05 between former welders and referents

High sensitivity CRP in serum (S-CRP) was measured at Fürstedical Laboratory (Oslo, Norway) by CardioPhase TM, Advia 2400

Siemens Healthcare Diagnostics Inc., NY, US). The method’s repro-ucibility was 2.2% and the DL was 0.1 mg/L.

The methods for the determination of serum concentrationsf soluble transferrin receptor (S-TfR) and S-ferritin in samplesollected for the study carried out six years previously werehe following; S-TfR was measured by immunoenzymometricssay (IEMA) as previously described [31]. S-ferritin was deter-ined using an immunoturbidimetric assay method (Cobas, Roche,annheim). The reproducibility as determined in reference human

erum material for this method was 3%.

easurements of elements in blood and urine

The biological samples were analyzed for elements at theational Institute of Occupational Health, Oslo, Norway (NIOH), by

nductively coupled plasma sector-field mass spectrometry (ICP-F-MS) using an Element 2 mass spectrometer (Thermo Electron,remen, Germany). Two mL of ultrapure nitric acid was added to

mL of whole blood or serum in polypropylene tubes for the mea-urements of Mn in whole blood (B-Mn) and Mn and Fe in serumS-Mn and S-Fe). 100 �L of an internal standard solution contain-ng 1 mg/mL of gallium was added to the sample that was dilutedo 14 mL with deionized (DI) water, after heating to 90 ◦C for 90

inutes and cooling to room temperature. Urine was diluted 5-old with DI water for the measurements of Fe and Mn in urineU-Fe and U-Mn) with the same amount of the internal standardolution added. The instrument was calibrated with whole blood,erum and urine matrix matched standard solutions. SeronormTM

race Elements human whole blood, serum and urine quality con-rol materials were used for quality assurance. The method’s DLsere 0.083, 0.12, 0.035, 30 and 0.82 �g/L for B-Mn, S-Mn, U-Mn,

-Fe and U-Fe, respectively.The results obtained for Mn and Fe in SeronormTM Quality Assur-

nce materials were within the producer’s recommended referenceange for Whole Blood Lot OK0337 (B-Mn: found 9.3 ± 0.5 �g/L,

= 33, acceptable range: 8.3–10.7; B-Co: found 0.13 ± 0.01 �g/L, = 33, acceptable range: 0.12–0.18), Serum Lot N00371 (S-Mn:

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of i(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

ound 16.7 ± 0.4 �g/L, n = 29, acceptable range: 13–21; S-Co:ound 2.9 ± 0.1 �g/L, n = 29, acceptable range: 2.9–3.4; S-Fe: found.93 ± 0.05 mg/L, n = 29, acceptable range: 1.71–2.18) and Urine Lot511545 (U-Mn: found 10.7 ± 0.4 �g/L, n = 12, acceptable range:

10.9–13.7; U-Fe: found 15.5 ± 0.8 �g/L, n = 12; no recommendedvalue by the producer; U-Co: found 10.0 ± 0.4 �g/L, n = 12, accept-able range 9.4–10.6). Creatinine in urine was measured by Jafféreaction using a SFA-200 flow injection analyzer (Burkard Scien-tific Ltd., Uxbridge, UK). Details for themeasurements of Mn and Fehave been published [27].

Co was also measured by ICP-SF-MS using the 59Co+ ion, mediumresolution and 69Ga+ ion as internal standard. Otherwise, the ana-lytical details and use of quality control materials was similar tothose used for Mn and Fe. The DLs of Co in whole blood (B-Co),serum (S-Co) and urine (U-Co) were 0.014, 0.021 and 0.0048 �g/L,respectively.

Welding aerosol exposure assessment and measurements

Personal full-shift sampling of the welding aerosol particulatematter, ideally on the two days before collection of biologicalsamples, was carried out. All welders used welding helmets for pro-tection, while two welders reported the use of respirators duringwork. Millipore (25 mm) aerosol plastic cassettes equipped with5.0 �m pore-size polyvinyl chloride membrane filters (SKC Ltd.,Dorset, UK) were mounted in the breathing zone underneath thewelding helmet. SKC Sidekick pumps (SKC Ltd., Dorset, UK) oper-ated at an initial flow rate of 2.0 L/min were used for sampling.Further details have been published [27].

Solubility of the welding fume components was assessed byleaching in the lung lining fluid simulant Hatch solution, followedby a two-step acid digestion procedure to dissolve the non-Hatchsoluble material [32]. The amount of Hatch-soluble (Hatchsol) andHatch insoluble (Hatchnon-sol) Fe, Mn and Co on the collectedaerosol filters were measured. The filters were leached in 50 mLVectaSpin 20TM PP centrifuge tubes with 25 mL filter cup insertequipped with 0.45 �m pore size nylon membranes (WhatmanInternational Ltd., Maidstone, England, UK), with 10 mL of the artifi-cial lung lining fluid simulant. The Hatch non-soluble acid digestedsamples were analyzed using a Perkin-Elmer Optima 7300 induc-tively coupled plasma optical emission spectrometer (ICP-OES)(Perkin-Elmer, Waltham, Massachusetts, USA). The Hatch leachedsolutions were analyzed using ICP-SF-MS. Further details have been

ron status and trace elements in welders. J Trace Elem Med Biol

published except for Co [27]. The same analytical methods wereused for measurements of Co in the soluble and non-soluble Hatchfractions. The relatively high amounts of Co in the protein part ofthe Hatch reagent solution resulted in a considerable higher DL for

ARTICLE IN PRESSG ModelJTEMB-25514; No. of Pages 7

4 D.G. Ellingsen et al. / Journal of Trace Elements in Medicine and Biology xxx (2014) xxx–xxx

Fau

C0

S

tfsuwsocraasw

R

psdcsrtcinc

eciosa(taCaC

0

0.5

1

1.5

2

2.5

3

W FW R W FW R W FW R

Rat

io (9

5%C

I)

S-Ferritin B-Mn U-Mn

Fig. 2. The difference in the concentrations of S-ferritin, B-Mn and U-Mn between

ig. 1. The GM (and 95% CI) concentration of S-ferritin in welders and referentsccording to alcohol consumption as assessed by S-CDT higher or lower than thepper reference limit of the laboratory of ≥1.7% (adjusted for S-CRP and age).

o in the Hatchsol than in the Hatchnon-soluble fraction; 0.048 and.014/filter of Co, respectively.

tatistics

Variables were log-transformed when the skewness of a dis-ribution exceeded 2.0. The geometric mean (GM) is presentedor these variables, otherwise the arithmetic mean (AM) is pre-ented. Group differences of continuous variables were assessedsing analysis of variance (ANOVA) and the least square differenceas calculated when more than two groups were compared. Least

quare regression analysis was used for the statistical assessmentf univariate associations, yielding Pearson’s correlation coeffi-ient (Pearson’s r) as the measure of association. Multiple linearegression analysis (backwards procedure) was applied to studyssociations between one dependent and several independent vari-bles. Two-tailed p-value <0.05 was considered as the level oftatistical significance. The statistical package SPSS®, version 18.0,as used.

esults

The welders’ current airborne exposure to Fe, Mn and Co wasredominantly non-soluble in the lung lining fluid simulant Hatcholution (Table 1). The concentrations of S-ferritin and S-hepcidinid not differ significantly between the groups (Table 2). The meanoncentrations of S-CRP were low in all groups, but statisticallyignificantly higher in the former welders as compared to the cur-ent welders and referents. Alcohol consumption, as assessed byhe S-CDT concentrations, was comparable in the groups. The meanoncentrations of Mn in biological fluids were as expected highern the welders, whereas S-Fe, S-Co and B-Co were statistically sig-ificantly lower. B-Mn was also higher in the former welders asompared to the referents.

Multiple linear regression analysis including all welders and ref-rents showed that S-ferritin was not associated with S-CDT as aontinuous variable. However, when S-CDT was stratified accord-ng to concentrations higher or lower than the upper reference limitf ≥1.7%, S-ferritin was associated with S-CDT and S-CRP. Fig. 1hows that the GM (and 95% CI) S-ferritin concentrations wereround 50% higher in both welders and referents with S-CDT ≥ 1.7%p = 0.02), while being exposed as a welder did not contribute tohe differences (p = 0.67). The GM S-ferritin concentrations among

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of i(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

ll welders and referents with S-CDT ≥ 1.7% were 157 �g/L (95%I 113–218) as compared to 104 �g/L (95% CI 94–116) (p = 0.02)mong all welders and referents with s-CDT < 1.7% adjusted for S-RP and age (not tabulated). The prevalence of having S-ferritin

this study and study carried out six years previously in 63 welders (W), 65 referents(R) and 17 former welders (FW). The differences are expressed as ratios because thevariables were log-transformed.

concentrations above the upper reference limit of the laboratory(300 �g/L) was 20% among subjects with S-CDT ≥ 1.7% as comparedto 6.6% among those with S-CDT < 1.7% (p = 0.03). The concentra-tions of S-hepcidin were not associated with any of the aboveindependent variables.

The concentrations of S-TfR and S-ferritin were determined ina study carried out six years previously (Table 3). S-ferritin wassignificantly higher in the welders as compared to the referents inthat study, while S-TfR was highest in the former welders. How-ever, when excluding subjects with low Fe stores as defined byS-ferritin <15 �g/L, the S-TfR concentrations were comparable inthe three groups. The S-ferritin concentrations determined in thepresent study were compared with those determined six years ear-lier among those 63 welders and 65 referents that were examined atboth occasions. Fig. 2 shows the mean ratios (and 95% CI) betweenthe concentrations of S-ferritin, U-Mn and B-Mn determined in thepresent study and six years previously. A ratio >1 indicates that theconcentration determined in the present study is higher than sixyears previously. These follow-up data show that six years of addi-tional welding did not increase S-ferritin beyond the increase in thereferents or the former welders during the same time period. B-Mnin the welders was around 50% higher in the present study, but onlyslightly higher in the patients and the referents. U-Mn was around60% higher in the welders, but not significantly higher in the formerwelders and in the referents at follow-up.

The results from the multiple linear regression analysisassessing the concentrations of elements in biological fluids areshown in Table 4. Being exposed as a welder was associated withhigher B-Mn, U-Mn and S-Mn and lower S-Fe, S-Co and B-Co.Alcohol consumption as assessed by S-CDT was associated withhigher S-Mn, S-Fe and B-Co. When excluding all subjects with S-CDT ≥ 1.7%, the S-CDT remained as an independent variable in theregression model for S-Fe, but not for B-Co and S-Mn. However,S-CDT was included in the model describing the concentrationsof B-Mn. S-Ferritin was statistically significantly included in allmodels except for B-Mn, with negative associations for U-Co, S-Coand B-Co and positive associations for the remaining metal con-centrations measured. The associations with S-ferritin were onlyobserved in the welders and not in the referents when the subjectswere stratified according to exposure status (results not shown).

We have previously reported the associations between theconcentrations of Mn in biological fluids and in the welding

ron status and trace elements in welders. J Trace Elem Med Biol

aerosol [27]. When adding S-ferritin and S-CDT to those previouslyreported models, it appears that S-ferritin modifies the concentra-tions of S-Mn and U-Mn (Table 5). The data also suggest that alcoholconsumption as assessed by S-CDT is positively associated with

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of iron status and trace elements in welders. J Trace Elem Med Biol(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

ARTICLE IN PRESSG ModelJTEMB-25514; No. of Pages 7

D.G. Ellingsen et al. / Journal of Trace Elements in Medicine and Biology xxx (2014) xxx–xxx 5

Table 2Iron status parameters and biological concentrations of elements in whole blood (B), Serum (S) and urine (U).

Welders (N = 137) Former welders (N = 34) Referents (N = 137) PANOVA

GMa Range GM Range GM Range

S-Hepcidin (�g/L) 8.4 2.8–117 6.6 1.8–100 6.5 1.2–22 0.22S-Ferritin (�g/L) 119 3–1498 112 9–1277 98 12–989 0.24S-CRP (mg/L)** ,*** 1.2 0.1–28 2.1 0.1–34 1.2 0.1–29 0.04S-CDT (%) 0.71 <DL-9.1 0.65 <DL-9.3 0.65 <DL-8.9 0.59U-Co (�g/g cr.)* ,*** 0.22 0.07–4.4 0.15 0.07–1.0 0.18 0.06–3.72 0.002U-Mn (�g/g cr.)* ,*** 0.36 0.03–12.9 0.07 0.01–0.83 0.07 0.01–3.1 <0.001U-Fe (�g/g cr.) 9.6 1.2–158 8.5 2.5–59 8.0 2.4–269 0.18B-Mn (�g/L)*,** ,*** 12.8 5.9–40.3 9.3 4.2–22.4 8.0 4.1–13.9 <0.001B-Co (�g/L)*,** 0.16 0.10–0.69 0.15 0.09–0.44 0.20 0.09–0.80 <0.001S-Co (�g/L)*,** 0.15 0.02–2.3 0.13 0.06–0.45 0.17 0.07–0.92 0.004S-Fe (mg/L)*,** 1.3 0.4–5.0 1.3 0.5–3.2 1.6 0.5–8.9 <0.001S-Mn (�g/L)*,*** 1.04 0.2–5.4 0.79 0.5–1.8 0.77 0.44–5.0 <0.001

a Geometric mean.* p < 0.05 between welders and referents.

** p < 0.05 between former welders and referents.*** p < 0.05 between welders and former welders.

Table 3Iron status parameters collected in Study 1.

Welders (N = 96) Former welders (N = 27) Referents (N = 96) PANOVA

GMa Range GM Range GM Range

S-TfR (mg/L)** ,*** 2.0 1.0–4.4 2.4 1.4–6.8 2.1 1.3–3.6 0.02S-Ferritin (�g/L)* 112 13–833 102 6–1831 73c 8–1347 0.003S-TfR/lg S-Ferritin*,*** 1.01 0.52–2.26 1.24 0.64–6.61 1.16c 0.61–2.95 0.008U-Mn (�g/g cr.)** ,*** 0.17c 0.03–5.5 0.07c 0.03–0.17 0.12 0.02–10.2 0.001B-Mn (�g/L)b,* ,** 8.6c 3.7–23.5 8.7 5.2–19.1 6.9 2.5–14.3 <0.001

a Geometric mean.b Arithmetic mean.c One subject missing.* p < 0.05 between welders and referents.

** p < 0.05 between former welders and referents.*** p < 0.05 between welders and former welders.

Table 4Multiple linear regression analysis among 137 currently exposed welders and 137 referents (Model: Exposure, S-CDT(lg), S-CRP (lg), S-Ferritin (lg), Smoking (1/0), Age).

� Mult. r

Exposure S-CDT S-Ferritin S-CRP Age Smoke

S-Mna (�g/L) +0.12**** +0.11** +0.12*** – – – 0.42****

S-Fe (mg/L) −0.12**** +0.11**** +0.11*** −0.06** −0.002* – 0.40****

S-Co (�g/L) −0.06** – −0.10** – – – 0.23***

U-Mn (�g/g cr.) +0.73**** – +0.20** – – +0.12* 0.59****

U-Fe (�g/g cr.) – – +0.26**** – +0.003* – 0.30****

U-Co (�g/g cr.) +0.11*** – −0.20**** – +0.003** – 0.32****

B-Mn (�g/L) +0.20**** – +0.05* – – – 0.58****

B-Co (�g/L) −0.07**** +0.07** −0.08**** − −0.002** – 0.35****

a All dependent variables are log-transformed.* p < 0.10.

** p < 0.05.*** p < 0.01.

**** p < 0.001.

Table 5Multiple linear regression analysis among 92 currently exposed welders (Model: Air- Mn, S-CDT(lg), S-CRP (lg), S-Ferritin (lg), Smoking (1/0), Age).

Mult. r

S-Mn −0.04ns +0.08 lgAir-Mnsol2a,*** +0.12lgFerri** +0.18lgCDT** 0.53****

U-Mn −0.81*** +0.24lgAirMnmeanb,**** −0.45lgFerri*** +0.32 lgCDT* 0.56****

B-Mn 1.22**** +0.08lgAir-Mnsol2**** +0.003yrs.welding** 0.51****

a Mn concentrations in air samples collected two days before biological samples.b Mean Mn concentration in air samples collected the day before and two days before biological samples.* p < 0.10.

** p < 0.05.*** p < 0.01.

**** p < 0.001.

ING ModelJ

6 nts in

SCH

D

iericsccieAao

tiToio[taShppoe

iSTetorsdfihlc7ttsl

cpuitToew

ARTICLETEMB-25514; No. of Pages 7

D.G. Ellingsen et al. / Journal of Trace Eleme

-Mn and possibly also U-Mn. No significant association betweeno in biological samples and the welding aerosol concentrations ofatchsol or Hatchnon-sol Co was found.

iscussion

The results from this study show that the concentrations ofmportant biomarkers of Fe status are not different in weldersxposed to welding aerosols containing high amounts of Fe andeferents. The concentrations of Mn in all examined biological flu-ds were higher in the welders than in the referents, while theoncentrations of S-Co, B-Co and S-Fe were lower. High alcohol con-umption, as assessed by S-CDT, had an impact by increasing theoncentrations of S-Mn, S-Fe and B-Co as well as the S-ferritin con-entrations. S-Ferritin was highly statistically significantly includedn the models describing most of the element concentrations, how-ver after stratification, only in the welders and not in the referents.lso when taking into account the air exposure measurements, highlcohol consumption appears to increase the Mn uptake into serumf the welders.

Although the airborne exposure to Fe on average was higherhan 1.5 mg/m3 in the welders, there were no significant differencesn S-ferritin and S-hepcidin between the welders and the referents.his may partly be explained by the low pulmonary bioaccessibilityf Fe [27]. However, also the low deposition efficiency of particlesn the size range present in welding fumes, suggesting that muchf the inhaled fume will be exhaled, may be a contributing factor33]. The fraction of the welding fume particles available for gas-rointestinal absorption is unknown, but in this group of welderspparently not large enough to have an impact on the S-ferritin or-hepcidin levels. The welders that were followed during six yearsad the same increase in S-ferritin as the referents and non-exposedatients followed up for the same time period. This extends therevious observation of unaltered S-ferritin concentrations duringne week of occupational exposure to welding fume in long-termxposed welders [5].

Alcohol consumption as assessed by S-CDT has a substantialmpact on S-ferritin which is around 50% higher in subjects with-CDT above the upper reference limit of the laboratory (≥1.7%).his corresponds to a daily consumption of at least 60–80 g ofthanol [34]. The impact of high alcohol consumption on Fe sta-us markers is well known, although sparsely studied by the usef S-CDT [35,36]. Mechanistically it has been shown that ethanoleduces the liver hepcidin expression, perhaps due to oxidativetress, in exposed rats and mice, accompanied by elevated duo-enal expression of DMT1 and ferroportin [22–24]. We did notnd any association between S-CDT and S-hepcidin. However, S-epcidin was determined in a limited number of samples due to

ack of serum and in few subjects only with S-CDT ≥ 1.7%. The meanoncentration in the five samples with S-CDT ≥ 1.7% was 5.4 vs..7 �g/L in the remaining subjects (p = 0.25). The S-ferritin concen-rations measured six years previously was significantly higher inhe welders, which could be related to differences in alcohol con-umption. However, we had no information on S-CDT or S-CRPevels in that study.

The S-Fe concentration correlated positively with the S-CDToncentration. S-CDT remained in the regression model as inde-endent variable after excluding all subjects with S-CDT above thepper reference limit of ≥1.7% (results not shown), suggesting an

mpact of alcohol consumption on S-Fe at lower alcohol consump-ion. This is compatible with the study by Whitfield et al. [35].

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of i(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

he concentrations of S-Mn were also related to S-CDT, which tour knowledge has not been reported. The reduced liver hepcidinxpression related to increased ethanol consumption is compatibleith reduced internalization of ferroportin in enterocytes. This is

PRESS Medicine and Biology xxx (2014) xxx–xxx

compatible with increased uptake of Mn as evidenced by higher S-Mn in the welders, as it has been shown that ferroportin is a cellularexporter of Mn [7,8]. If this interpretation is correct, alcohol con-sumption might be a risk factor for increased Mn uptake. Whetherthis also could be related to increased risk of acquiring Mn relatednervous system disturbances remains to be elucidated. Also B-Co isincreased related to increasing S-CDT, suggesting Co to use ferro-portin as an exporter out of the enterocytes as well, although littleis known about the Co absorption.

The GM concentration of S-Fe was lower in the welders althoughthe welding fume content of Fe was high. Also the concentrationsof S-Co and B-Co were lower in the welders, although they hada slight exposure to Co. Both Co and Fe are transported throughDMT1 together with Mn, and it has been shown that Mn compet-itively can inhibit the uptake of Fe [16]. This is compatible withstudies showing lower serum Fe and hemoglobin levels in lambsadministered high amounts of Mn and to findings of decreased gas-trointestinal absorption of Fe in Mn exposed humans [11,12]. It istherefore tempting to speculate that even low amounts of Mn caninhibit the gastrointestinal uptake of metals transported throughDMT1, although we are not aware of any such studies.

There is a consistent pattern in the present study of S-ferritinbeing included in the statistical models explaining the elementalconcentrations in the welders, but not in the referents. Horses-pleen ferritin, traditionally considered as a mammalian ferritinmodel, consists of a protein shell with a hollow core capable ofcontaining several thousand Fe atoms as trivalent ferrihydrite [37].The 24 polypeptide shells are classified as heavy (H) or light (L)subunits, where the H-unit is associated with rapid detoxificationof Fe(II) because of its content of ferroxidase, while the L-unit isassociated with mineralization and long term storage in the cav-ity. It is assumed that Fe(II) moves through hydrophilic channelsuntil oxidized by a ferroxidase center in the H-subunit. It has beenshown that other elements than Fe can interact with ferritin, thusdisturbing its function as a Fe detoxifier. Cadmium is competingwith Fe for the same sites within the cavity but also on the exter-nal shell surface, in particular around the entry of the hydrophilicchannels. This may hinder the entry of Fe. Zinc can also bind atthe channel entrances, and its binding to ferroxidase site mayreduce its oxidation capability. Copper is also known to interactwith the ferritin molecule. We have not been able to find any dataon Mn, but given all metabolic similarities between the transitionelements Mn and Fe, this possibility should, in our opinion, bestudied.

The concentrations of U-Co, S-Co and B-Co were all negativelyassociated with the concentration of S-ferritin. The results are sup-ported by a previous study showing that an inverse associationexists between markers of Fe status and biological Co concentra-tions [38]. The biological mechanisms for this association have toour knowledge not been elucidated.

We have previously reported that the concentration of S-Mn isassociated with the welding fume concentration of Mn soluble inthe artificial lung fluid stimulant Hatch solution [27]. When com-bining the results from that study of biological monitoring withthe results from the present study, a statistically significant con-tribution from alcohol consumption as assessed by S-CDT on theS-Mn concentrations in the welders was observed. This associationhas, to our knowledge, previously not been shown, but may pointto higher uptake of Mn in exposed subjects that consume higheramounts of alcohol. This association should be assessed in futurestudies.

This study shows that important biomarkers of Fe status are not

ron status and trace elements in welders. J Trace Elem Med Biol

altered in welders with high exposure to Fe and Mn. In contrast,alcohol consumption, as assessed by the concentration of S-CDT, isan important modifier of the measured concentrations of biologicalmarkers of several divalent metals as well as S-ferritin. The results

ING ModelJ

nts in

mo

C

R

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

ARTICLETEMB-25514; No. of Pages 7

D.G. Ellingsen et al. / Journal of Trace Eleme

ay also be interpreted that exposure to Mn may reduce the uptakef Fe and Co.

onflict of interest statement

The authors declare no conflict of interest.

eferences

[1] Flynn MR, Susi P. Neurological risks associated with manganese exposurefrom welding operations – a literature review. Int J Hyg Environ Health2009;212:459–69.

[2] Voitkevich V. Welding fumes – formation, properties and biological effects.Cambridge, England: Abington Publ; 1995. p. 18–71.

[3] Antonini JM, Roberts JR, Stone S, Chen BT, Schwegler-Berry D, Frazer DG. Short-term inhalation exposure to mild steel welding fume had no effect on lunginflammation and injury but did alter defense responses to bacteria in rats.Inhal Toxicol 2009;21:182–92.

[4] Berlinger B, Benker N, Weinbruch S, L‘vov B, Ebert M, Koch W, et al. Physi-cochemical. characterisation of different welding aerosols. Anal Bioanal Chem2011;399:1773–80.

[5] Pesch B, Weiss T, Kendzia B, Henry J, Lehnert M, Lotz A, et al. Levels and predic-tors of airborne and internal exposure to manganese and iron among welders.J Expo Sci Environ Epidemiol 2012;22:291–8.

[6] Fitsanakis VA, Zhang N, Garcia S, Aschner M. Manganese (Mn) and iron (Fe):interdependency of transport and regulation. Neurotox Res 2010;18:124–31.

[7] Yin Z, Jiang H, Lee ESY, Ni M, Erikson KM, Milatovic D, et al. Ferroportin isa manganese-responsive protein that decreases manganese cytotoxicity andaccumulation. J Neurochem 2010;112:1190–8.

[8] Madejczyk MS, Ballatori N. The iron transporter ferroportin can also functionas a manganese exporter. Biochim Biophys Acta 2012;1818:651–7.

[9] Chua ACG, Graham RM, Trinder D, Olynyk JK. The regulation of cellular ironmetabolism. Crit Rev Clin Lab Sci 2007;44:413–59.

10] Ganz T. Hepcidin and iron regulation, 10 years later. Blood 2011;117:4425–33.11] Hartman RH, Matrone G, Wise GH. Effect of high dietary manganese on

hemoglobin formation. J Nutr 1955;57:429–39.12] Rossander-Hultéen L, Brune M, Sandström B, Lönnerdal B, Hallberg L. Compet-

itive inhibition of iron absorption by manganese and zinc in humans. Am J ClinNutr 1991;54:152–6.

13] Meltzer HM, Brantsaeter AL, Borch-Iohnsen B, Ellingsen DG, Alexander J,Thomassen Y, et al. Low iron stores are related to higher blood concentrationsof manganese, cobalt and cadmium in non-smoking, Norwegian women in theHUNT 2 Study. Environ Res 2010;110:497–504.

14] Zheng W, Ren S, Graziano JH. Manganese inhibits mitochondrial aconitase: amechanism of manganese neurotoxicity. Brain Res 1998;799:334–42.

15] Li Q, Chen H, Huang X, Costa M. Effects of 12 metal ions on iron regulatory pro-tein 1 (IRP-1) and hypoxia-inducible factor-1 alpha (HIF-1) and HIF-regulatedgenes. Toxicol Appl Pharmacol 2006;213:245–55.

16] Illing AC, Shawki A, Cunningham CL, Mackenzie B. Substrate profile andmetal-ion selectivity of human divalent metal-ion transporter-1. J Biol Chem

Please cite this article in press as: Ellingsen DG, et al. Biomarkers of i(2014), http://dx.doi.org/10.1016/j.jtemb.2014.03.004

2012;287:30485–96.17] Ghio AJ, Turi JL, Yang F, Garrick LM, Garrick M. Iron homeostasis in the lung.

Biol Res 2006;39:67–77.18] Heilig E, Molina R, Donaghey T, Brain JD, Wessling-Resnick M. Pharmacokinetics

of pulmonary manganese absorption: evidence for increased susceptibility to

[

[

PRESS Medicine and Biology xxx (2014) xxx–xxx 7

manganese loading in iron-deficient rats. Am J Physiol Lung Cell Mol Physiol2005;288:887–93.

19] Thompson K, Molina R, Donaghey T, Brain JD, Wessling-Resnick M. Theinfluence of high iron diet on rat lung manganese absorption. Toxicol ApplPharmacol 2006;210:17–23.

20] Chapman RW, Morgan MY, Laulicht M, Hoffbrand AV, Sherlock S. Hepatic ironstores and markers of iron overload in alcoholics and patients with idiopathichemochromatosis. Dig Dis Sci 1982;27:909–16.

21] Irving MG, Halliday JW, Powell LW. Association between alcoholism andincreased hepatic iron stores. Alcohol Clin Exp Res 1988;12:7–13.

22] Harrison-Findik DD, Schafer D, Klein E, Timchenko NA, Kulaksiz H, Clemens D,et al. Alcohol metabolism-mediated oxidative stress down-regulates hepcidintranscription and leads to increased duodenal iron transporter expression. JBiol Chem 2006;281:22974–82.

23] Harrison-Findik DD. Role of alcohol in the regulation of iron metabolism. WorldJ Gastroenterol 2007;1337:4925–30.

24] Harrison-Findik DD, Klein E, Crist C, Evans J, Timchenko NA, Gollan J. Iron-mediated regulation of liver hepcidin expression in rats and mice is abolishedby alcohol. Hepatology 2007;46:1979–85.

25] Doherty MJ, Healy M, Richardson SG, Fisher NC. Total body iron overload inwelder’s siderosis. Occup Environ Med 2004;61:82–5.

26] Lu L, Zhang L, Li GJ, Guo W, Liang W, Zheng W. Alterations of serum con-centrations of manganese, iron, ferritin, and transferring receptor followingexposure to welding fumes among career welders. NeuroToxicology 2005;26:257–65.

27] Ellingsen DG, Zibarev E, Kusraeva Z, Berlinger B, Chashchin M, Bast-PettersenR, et al. The bioavailability of manganese in welders in relation to its solubilityin welding fumes. Environ Sci: Processes Impacts 2013;15:357–65.

28] Ellingsen DG, Konstantinov R, Bast-Pettersen R, Merkurjeva L, Chashchin M,Thomassen Y, et al. A neurobehavioral study of current and former weldersexposed to manganese. NeuroToxicology 2008;29:48–59.

29] Konz T, Alonso-Garcia J, Montes-Bayón M, Sanz-Medel A. Comparison of copperlabeling followed by liquid chromatography-inductively coupled plasma massspectrometry and immunochemical assays for serum hepcidin-25 determina-tion. Anal Chim Acta 2013;79:1–7.

30] Ellingsen DG, Kusraeva Z, Bast-Pettersen R, Zibarev E, Chashchin M, ThomassenY, et al. The interaction between manganese exposure and alcohol on neuro-behavioral outcomes in welders. Neurotoxicol Teratol 2014:8–15.

31] Ellingsen DG, Haug E, Ulvik R, Thomassen Y. Iron status in manganese alloyproduction workers. J Appl Toxicol 2003;23:239–47.

32] Berlinger B, Ellingsen DG, Naray M, Zaray G, Thomassen Y. A study of the bio-accessibility of welding fumes. J Environ Monit 2008;10:1448–53.

33] Sriram K, Lin GX, Jefferson AM, Roberts JR, Chapman RS, Chen BT, et al. Dopami-nergic neurotoxicity following pulmonary exposure to manganese-containingwelding fumes. Arch Toxicol 2010;84:521–40.

34] Bortolotti F, De Paoli G, Tagliaro F. Carbohydrate-deficient transferrin (CDT)as a marker of alcohol abuse: a critical review of the literature 2001–2005. JChromatogr B 2006;841:96–109.

35] Whitfield JB, Zhu G, Heath AC, Powell LW, Martin NG. Effects of alcohol con-sumption on indices of iron stores and of iron stores on alcohol intake markers.Alcohol Clin Exp Res 2001;25:1037–45.

36] Lieb M, Palm U, Hock B, Schwarz M, Domke I, Soyka M. Effects of alcoholconsumption on iron metabolism. Am J Drug Alcohol Abuse 2011;37:68–73.

ron status and trace elements in welders. J Trace Elem Med Biol

37] Carmona F, Palacios O, Gálvez N, Cuesta R, Atrian S, Capdevila M, et al. Fer-ritin iron uptake and release in the presence of metals and metalloproteins:chemical implications in the brain. Coord Chem Rev 2013;257:2752–64.

38] Sorbie J, Valberg LS, Corbett WEN, Ludwig J. Serum ferritin, cobalt excretionand body iron status. Can Med Assoc J 1975;112:1173–8.