Arsenic speciation in human hair: a new perspective for

epidemiological assessment in chronic arsenicism

Jorge Yanez,*a Vladimir Fierro,a Hector Mansilla,b Leonardo Figueroa,c Lorena Cornejoc

and Ramon M. Barnesd

aDepartment of Analytical & Inorganic Chemistry, Faculty of Chemical Sciences, University ofConcepcion, Concepcion, P.O. Box 160-C, Chile

bDepartment of Organic Chemistry, Faculty of Chemical Sciences, University of Concepcion,Concepcion, P.O. Box 160-C, Chile

cDepartment of Chemistry, Faculty of Sciences, University of Tarapaca, Arica, ChiledUniversity Research Institute for Analytical Chemistry, Amherst, Massachusetts, USA

Received 5th May 2005, Accepted 6th October 2005First published as an Advance Article on the web 20th October 2005

The analysis for arsenic in hair is commonly used in epidemiological studies to assess exposure to this toxicelement. However, poor correlation between total arsenic concentration in hair and water sources have beenfound in previous studies. Exclusive determination of endogenous arsenic in the hair, excluding externalcontamination has become an analytical challenge. Arsenic speciation in hair appears as a new possibility foranalytical assessing in As-exposure studies. This study applied a relative simple method for arsenic speciationin human hair based on water extraction and HPLC-HG-ICP-MS. The concentration of arsenic species inhuman hair was assessed in chronically As(V)-exposed populations from two villages (Esquina and Illapata)of the Atacama Desert, Chile. The arsenic concentrations in drinking water are 0.075 and 1.25 mg L�1,respectively, where As(V) represented between 92 and 99.5% of the total arsenic of the consumed waters. Onaverage, the total arsenic concentrations in hair from individuals of Esquina and Illapata were 0.7 and 6.1 mgg�1, respectively. Four arsenic species, As(III), DMA(V), MMA(V) and As(V), were detected and quantified inthe hair extracts. Assuming the found species in extracts represent the species in hair, more than 98% of thetotal arsenic in hair corresponded to inorganic As. On average, As(III) concentrations in hair were 0.25 and3.75 mg g�1 in Esquina and Illapata, respectively; while, the As(V) average concentrations were 0.15 and 0.45mg g�1 in Esquina and Illapata, respectively. Methylated species represent less than 2% of the extracted As(DMA(V) þ MMA(V)) in both populations. As(III) in hair shows the best correlation with chronic exposureto As(V) in comparison to other species and total arsenic. In fact, concentrations of As(total), As(III) andAs(V) in hair samples are correlated with the age of the exposed individuals from Illapata (R ¼ 0.65, 0.69,0.57, respectively) and with the time of residence in this village (R ¼ 0.54, 0.71 and 0.58, respectively).

Introduction

Arsenic is a toxic element for humans and is commonlyassociated with serious health disruptions. The principal man-ifestations of arsenicism affecting health are melanosis, kera-tosis and different forms of cancer (skin, bladder, lung, liverand prostate among others).1 The most common form ofmassive and chronic exposure is by consumption of contami-nated drinking water. Bangladesh, India, Mongolia, China,Taiwan, Mexico, Argentina and Chile are countries wherearsenic poisoning appears as a public health problem resultingmainly from consumption of As-contaminated water.2

The northern zone of Chile, and especially the AtacamaDesert, has been described as an arsenic-rich environment.Minerals of metallic sulfides containing arsenic are dissolved inthe Andes Mountains, affecting superficial and ground watersthat cross the Atacama Desert and are used as drinking watersources. Since 1970, drinking water is specially treated toremove arsenic in all the large cities of the Atacama Region,such as Antofagasta.2 However, the populations of severalsmall rural villages remain exposed to arsenic in drinkingwater. The problem of chronic arsenicism affects around50 000 people, mainly in rural populations of the AtacamaDesert in northern Chile. The affected populations drink waterfrom small waterfalls and rivers with arsenic contents greater

than 1 mg L�1.3 This situation greatly surpasses World HealthOrganization and the U.S. Environmental Protection Agencyrecommendations for As concentrations up to 10 mg L�1,respectively.4–6

It is well known that the toxicity of arsenic is highlydependent on its chemical form. In fact, As(III) is more toxicthan As(V) and methylated compounds that contain trivalentarsenic are more cytotoxic and genotoxic than arsenite.1,7–9

Other organic compounds of arsenic, such as arsenobetaine,arsenocoline and arsenosugars, can be ingested by seafood andseaweed consumption, although their toxicity is lower thanfound for inorganic species.1,10

Exposed individuals transform, accumulate, and eliminatethe ingested arsenic. Inorganic arsenic can be transformed intoorganic arsenic, mainly to methylated species such as dimethyl-arsinate (DMA(V)) and monomethylarsenate (MMA(V)).Around 60–75% of the inorganic arsenic ingested by a normalindividual is excreted in urine in a few days, principally asDMA(V) (60–80%) and MMA(V) (10–15%).11 Also arseno-sugars are metabolized by humans into DMA(V), and theneliminated through the urine. This fact restricts the use ofDMA(V) as a bioindicator of inorganic arsenic exposure,especially in seafood- or seaweed-consuming individuals.10

On the other hand, reduced methylated arsenical species(MMA(III) and DMA(III)) have been measured in urine. Since

P A P E R

ww

w.rsc.o

rg/jem

DO

I:1

0.1

03

9/b

50

63

13

b

J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1 1 3 3 5T h i s j o u r n a l i s & T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 5

these species present low stability even at low temperatures(�20 1C),12,13 changes in arsenic species composition duringthe transportation and storage before analysis need to beconsidered in data interpretation. This is an important limita-tion of using As-speciation in urine for epidemiological studiesespecially when sampling takes place in remote areas.14,15 As aresult, sample preservation requires special consideration inorder to assure the reliability of speciation analysis in urine.

Part of the ingested arsenic can be found in keratin-richtissues, such as hair and nails. Metal and non-metal elements,such as arsenic, are transported in the blood and included inthe fiber. Alpha-keratine of human hair contains about10–14% of cysteine,16 offering abundant thiol groups forreaction with As compounds. Due to the high affinity of arsenicfor keratin, arsenic concentrates in hair much higher than inother tissues or biological fluids.17 A normal concentration ofarsenic in hair ranges from 0.08 to 0.25 mg g�1 in an unexposedpopulation. In contrast, in chronically exposed populations,concentrations ranging from 1 to greater than 9 mg g�1 havebeen reported.18 Owing to the hair’s capacity to accumulatearsenic and its slow growth (0.44 mm day�1 or 13 mmmonth�1), the total As concentration in hair has often beenused for epidemiological studies in chronically exposed popu-lations. However, poor correlation between total arsenic con-centration in hair and water sources have been found indifferent studies.17,18 Possible explanations for this observationare the existence of exogenous contamination and the poorefficiency for removing exogenous arsenic from the hair priorto analysis.

In the last few years, arsenic speciation in hair has beenreported.19–22 One important advantage of arsenic speciationin hair is related to the good stability of arsenic species whencompared to other biological samples. Other advantages aresample size that can be obtained (approximately 1 g) and therelative facility for sampling, transport and storage. Addition-ally, there is the possibility of differentiating endogenous andexogenous arsenic species when they are characterized.

However, there is a lack of information about arsenic speciesconcentrations in hair for exposed populations. Shraim et al.reported low extraction recoveries and the possibility of specieschanges during extraction steps.20 Recently, Raab and Feld-mann described a new method for arsenic speciation in hair byextraction with boiling water.22 They determined the speciesstability during extraction and quantified four species (As(III),As(V), DMA(V) and MMA(V)) in exposed populations fromWest Bengal and Central India, where As(V) was the mainspecies found in hair (55 and 63%, respectively).

The most widely used method for arsenic speciation is highperformance liquid chromatography (HPLC) coupled to spe-cific element detectors, usually elemental spectroscopy includ-ing atomic absorption spectrometry (AAS), inductivelycoupled plasma optical emission spectrometry (ICP-OES) orinductively coupled plasma mass spectrometry (ICP-MS). Re-verse phase ion-pairing HLPC presents the advantages that itcan separate with high resolution inorganic and methylatedspecies simultaneously in relatively short times (e.g.,o10 min).Additionally, IP-HPLC has a good robustness for biologicalmatrices23,24 and it have been coupled directly with ICP-MSfor arsenic speciation in hair.25 Hydride generation improvessensitivity and detection limits, and it provides supplementaryselectivity for the most toxic species (As(III), As(V), MMA(V),MMA(III), DMA(V) and DMA(V)), excluding non-toxic speciesthat do not form hydrides (arsenocoline (AsC) and arseno-betaine (AsB)). Although, the determination of AsC and AsBby using on-line photo oxidation HG been reported,26 con-sidering the detection limits, the most powerful technique forarsenic speciation is HPLC-HG-ICP-MS.

To the best of our knowledge, there is a lack of availableinformation on arsenic speciation in hair from As(V)-exposedpopulations. The objective of this study is to characterize

arsenic species in human hair collected from As(V)-exposedpopulations living in the Atacama Desert, Chile and to explorepossible correlation between arsenic species concentration inhair with the exposure time (age and time of residence in theselected villages).

Material and methods

Reagents

All reagents were of analytical grade. Milli-Q water (Millipore,Bedford, MA) was used for all experiments. Standard stocksolutions of arsenic species, containing 1000 mg L�1 of As, wereprepared by dissolving an appropriate amount of the followingsalts: NaAsO2 and Na2HAsO4 � 7H2O from Merck, Ger-many, C2H6AsO2Na (DMA(V)) from SIGMA, USA andCH5AsO2Na (MMA(V)) from Chem Service, USA. NaBH4,HCl, NaOH, NaH2PO4, Na2HPO4 and tetrabutylammoniumhydrogen sulfate (TBAHSO4) were purchased from Merck,Germany. The stock solutions were stored in the dark at4 1C, except for NaBH4, which was prepared daily in NaOH(0.05%) owing to its low stability in neutral water.Human hair certified reference material (GBW 09101 No 18)

from the National Research Center for Certificated Materials,Beijing, China was kindly provided by Dr Chitra J. Amarasir-iwardena. This standard contains a certified total As concen-tration of 0.59 � 0.07 mg g�1. GBW 09101 No 18 and was usedto determine the accuracy, precision, and extraction recoveriesfor total arsenic and species. The water standard referencematerial SRM 1643c, trace elements in water was obtainedfrom the National Institute of Standards and Technology(NIST, USA). The total arsenic concentration is 82.1 � 1.2mg L�1. No individual As species is certified.

Populations

Two small rural villages, Esquina and Illapata, were selectedfor this study. They are located in the middle of AtacamaDessert, Valley of Camarones River, Region of Atacama,Chile, 250 km southeast of the city Arica. The population ofEsquina consumes drinking water principally from smallwaterfalls present in the valley, while the population of Illapataconsumes drinking water mainly from the river and waterfalls.In both villages, the drinking water was not treated prior toconsumption. Esquina and Illapata were selected because ofprevious information reporting the total arsenic concentrationspresent in Camarones Valley waters as published by Figueroaet al.3 According to this study, the arsenic concentration indrinking water of Esquina was 39 mg L�1. In spite of thisinformation, the population of Illapata consumed As-contami-nated drinking water containing 55 and 1090 mg As L�1 fromthe waterfall and river, respectively.3

The populations in the Atacama Dessert and in both studiedvillages live in similar geological environments, consume thesame foods, and are exposed to the same sunlight intensity.However, the magnitude of arsenic exposure by drinkingwaters is very different.3 Illapata has about 60 inhabitants, 21of whom were sampled for the present work; 13 adults(5 females and 8 males) and 8 children. Esquina has a stablepopulation of about 50 people, 22 of whom were sampledincluding 11 adults (6 females and 5 males) and 11 children.Children were considered as individuals from 2 up to 12 yearold.All sampled individuals were interviewed about their general

health condition, food consumption, drug use, drinking watersources, the time of residence in the village, sun exposure andtype of activity. As criteria for exclusion, sampled individualshad to be older than 3 years old, have no incapacitating illness,not consume drugs and have lived at least one month in thevillage. Clinical examinations were not performed in this study.All individuals selected for sampling were previously informed

1 3 3 6 J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1

about the purposes of this work and voluntarily consented toparticipate in the study.

Sampling and sample pretreatment

Water. Water samples were collected in 1-L polyethylenecontainers from drinking water sources (waterfalls and Camar-ones River). To assess the total arsenic and speciation analysis,two different waterfalls were sampled in Esquina. In Illapata,samples from one waterfall and the river were collected. Thesewater sources are available to the population by installation ofpublic water faucets. Samples were stabilized with 0.1% HCland kept in an ice-cooler at 0 1C during the transport to thelaboratory. In the laboratory, samples were stored at �20 1Cuntil analysis. All samples were filtered with 0.5 mm membranedisks prior to analysis.

Hair. 0.5–1 g hair was collected from different parts of thescalp using a stainless-still scissors, cutting at a distance of ca.1 cm from scalp. Hair was placed in polyethylene bags fortransport and storage. Since no well-established procedureexists to differentiate between endogenous and exogenousarsenic, in this work, samples were washed following theInternational Atomic Energy Agency (IAEA) protocol of1978 for removing exogenous As contamination.27 This simplemethod does not remove the endogenous arsenic in compar-ison to other described methods.28 Briefly, hair samples werefirst washed using a sufficient volume of acetone to cover thehair sample in a 50-mL polyethylene centrifuge tube and theacetone was separated by centrifugation. The hair (pellet) waswashed using de-ionized water at room temperature withsuccessive shaking in a high-speed lab shaker for 1 min forproper homogenization followed by manual shaking for 10min. After the water-cleaning step, the same washing proce-dure was repeated twice using acetone as described previously.Samples were dried overnight in an oven at 50 1C. The driedhair was cut into small pieces (o1 mm) using stainless-steelscissors, and the pieces were stored in polyethylene tubes atroom temperature until analysis. Samples were weighed im-mediately before the digestion or extraction procedure.

Total arsenic in hair

The digestion method was adapted from Flores et al.29 Briefly,0.1 g of sample was digested by adding 3 mL of concentratedHNO3 and 1 mL H2O2 (30%) in a Teflon PFA vessel using amicrowave oven (CEM MDS-2000). The vessels were closedand heated following the MW-program: 100 W (5 min), 250 W(3 min), 400 W (5 min), 450 W (3 min), 630 W (1 min). Toavoid overpressure, each heating step was followed by 3 minwithout power. After cooling, the solutions were transferredand diluted in 25-mL polyethylene tubes. Determination oftotal arsenic was performed using hydride generation (HG)and inductively coupled plasma-mass spectrometry (ICP-MS)(see below).

Speciation in hair

The arsenic leaching procedure was performed after washingthe samples. Water has been described as a simple and effectivesolubilization reactive for arsenic species in hair.19,20 In gen-eral, water incubation softens keratine-rich tissues, increasingleaching agent accessibility and facilitating the dissolution ofbonded compounds and ions. The effectiveness of water as aleaching agent depends directly on the temperature. Previouswork of Mandal et al. found that at room temperature, arsenicextraction from hair and fingernails was less that 1% of thetotal arsenic.19 Increasing the temperature close to boilingimproves the leaching of the total arsenic from hair.

First, 25–100 mg of washed and dried hair was placed inpolyethylene tubes. Then, 10 mL de-ionized water (o18 MO)was added to the tube with hair. Samples were leached at 90 1Cfor 3 h in an oven, with manual shaking every 30 min. Afterleaching, the samples were centrifuged at 3000 rpm for 10 min.The leaching solution (supernatant) was carefully separatedfrom the hair (pellet) and stored at �20 1C until speciationanalysis. When leaching was performed at lower temperatures(room temperature and 50 1C), lower recoveries were obtained.By boiling water, non-reproducible recoveries of As occurredpresumably by volatile arsenic compound losses. Accordingly,the leaching temperature for samples and certified referencematerial (GBW 09101 No 18) was 90 1C.

Instrumentation for arsenic speciation

Arsenic speciation was performed using ion-pair chromatogra-phy (IP-HPLC) combined with hydride generation (HG) andinductively coupled plasma-mass spectrometry (ICP-MS) as aspecific arsenic detector. The HPLC system consisted in aHPLC from Merck-Hitachi, Germany (model L-7100 La-Chrom), a six-port HPLC valve from Rheodyne, USA (model7725i) with a 20 mL sample loop and a monolithic HPLCcolumn RP-C18 (100� 4.6 mm) fromMerck, Germany (modelChromlith). The separation was performed at room tempera-ture and a flow rate of 1 mL min�1. The HPLC mobile phasecontained 0.35 mM of tetrabutylammonium hydrogen sulfate(ion pair reagent), and a pH value of 5.75 was regulated usingphosphate buffer at 0.5 mM. The column was connecteddirectly to the hydride generation system, and consisted in ahomemade flow injection device with two T-joints for contin-uous flow of HCl (15%) and NaBH4 (0.6%, in NaOH 0.05%).They were pumped at a flow 1 mL min�1 using a peristalticpump into the HPLC effluent. Chemical reaction by volatilehydride generation took place in a 1-mL loop made of PTFEtubing. Separation of gaseous hydrides from the liquid wasperformed in a glass gas–liquid separator (GLS). The GLSdesign was previously described for HG-AFS (atomic fluores-cence spectrometry),30 and it is used here for the first timeapplied to HG-ICP-MS. Hydrides were carried from the GLSto the ICP-MS using an argon flow rate of 0.75 L min�1. Asecond makeup gas (argon) was necessary for obtaining opti-mal sensitivity. For this purpose, 0.5 mL min�1 Ar wasintroduced after the GLS and before the plasma torch. TheICP-MS (Agilent Technologies Model 7500a, Wilmington,Delaware) was operated under optimal conditions. The Ar-senic ion signal was monitored atm/z 75. Them/z 77, 82 and 83signals were also monitored for interference correction. RFpower, sample depth, plasma and auxiliary gas flow rate were1200 W, 6 mm, 16 and 1 L min�1, respectively. Chromato-graphy software of the instrument was used for quantification.Fig. 1 shows a typical chromatogram of four arsenic in water;As(III), As(V), MMA(V) and DMA(V), 50 mg L�1 each, underthe optimized chromatographic conditions. Acceptable separa-tion (resolution) for the four species was achieved to quantifytarget species. As(III) and MMA(V) have higher sensitivity incomparison with other species, which can be explained due tothe differences of hydride species generation efficiency.

Results and discussion

Arsenic concentration in water sources

The population of Illapata principally consumes water fromthe Camarones River. The total arsenic concentration in thisdrinking water source was 1252 mg L�1, more than 100 timesthe accepted international levels (10 mg L�1) and 25 timesgreater than the Chilean standard (50 mg L�1). Waterfall watersampled in Illapata contains 48.7 mg L�1 of total arsenic.However, waterfall water is not the preferred drinking waterbecause of its taste. In Esquina, the population principally

J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1 1 3 3 7

consumes waterfall water from two sources with arsenic con-centrations of 74 and 12.2 mg L�1, respectively, slightly over themaximum permitted levels by Chilean Standards. The waterwas consumed without any prior treatment in both villagesbefore sampling. The total arsenic concentrations found inwater sources agree with values reported by Figueroa.3 Smalldifferences can be explained by seasonal changes of arsenicconcentration in water. Concentrations of total As and Asspecies in drinking water sources of Esquina and Illapata arepresented in Table 1. The speciation analysis shows that theprincipal species is As(V), representing between 92 and 99.5%of the total arsenic of the consumed waters. No organic arsenicspecies were found in any of the analyzed water samples. Thisresult permits the conclusion that the Esquina and Illapatapopulations are exposed almost exclusively to As(V) in drinkingwater, although intake of other arsenic species containedthrough vegetable consumption is also possible.

Total arsenic in human hair

The total arsenic concentrations in hair samples from Esquinaand Illapata were determined using HG-ICP-MS after nitricacid and hydrogen peroxide digestion in MW oven. For qualitycontrol, the total arsenic was determined in the hair CRM(GBW 09101 No 18), finding the value 0.62 � 0.06 mg g�1 (N ¼3), which agrees with CRM that contains 0.59 � 0.07 mg g�1.The CRM was not washed prior to the total arsenic analysis.

In Esquina and Illapata, the average total arsenic valueswere 0.7 and 5.8 mg g�1, respectively. The median values fortotal arsenic are 0.4 and 4.7 mg g�1, respectively. The values inIllapata were around ten times greater than the acceptednormal values for non-exposed individuals (o0.5 mg g�1). Avalue greater than 1 mg g�1 is considered an indication ofchronic exposure and toxicity.18 In Esquina, 8 of the 23 (35%)individuals studied presented As concentrations over the re-ferenced concentration (0.5 mg g�1) with 4 individuals (17%)

presenting concentrations over the toxicity level (41 mg g�1).The maximum concentration found in Esquina was 3.3 mg g�1

for a 9-year-old child. In Illapata, 97% (29 from 30) of theindividuals present values that exceed the normal and toxicitylevels (41 mg g�1), with 30% of these presenting values morethan 10 times the normal concentration. These values demon-strate the chronic exposure of the Illapata population toarsenic in contrast to the concentrations found in the Esquinapopulation, evidently less chronically exposed than in Illapata.Correlation between the total arsenic in hair and the resi-

dence time was found for both populations. In general, thelonger the stay in town, the higher the arsenic concentrationfound in hair. In the Illapata population, the total As concen-tration in hair is highly correlated with the individual’s age.For individuals of Illapata, the correlation (R) between totalAs in hair and age was 0.64, while the correlation of total Asconcentration in hair and the residence time in the village was0.53. Fig. 2 presents the plotted correlation of total As in hairbetween age and time of residence for the Illapata individuals.

Stability of As species during leaching from hair samples

The stability of the arsenic species was studied during theleaching procedure. Clean hair was separately incubated at90 1C with standard solutions of 100 mg L�1 As(III), As(V) andDMA(V). No significant changes between As(III) and As(V)were observed up to 3 h of extraction at 90 1C. In samplescollected after 4 h of incubation, a partial oxidation of As(III)to As(V) (15%) and no reduction of As(V) to As(III) was found.After 6 h of incubation, 16% oxidation of As(III) to As(V) and12% reduction of As(V) to As(III) was found. DMA(V) resultsstable up to 10 h incubation. Based on these results, hairsample leaching was performed at 90 1C for 3 h, excludingthe possibility of species changes during extraction. Furtherstudies about the stability of arsenic species during the extrac-tion procedure are presently being performed. The recovery ofarsenic with 3 h leaching at 90 1C in the human hair CRMGBW 09101 No 18 (0.59 � 0.07 mg g�1) was 66% (0.39 � 0.09mg g�1, N ¼ 3). For all samples (N ¼ 43), the mean recoveryusing the same conditions was 66% � 21 mg g�1. Furtherimprovements in extraction efficiency could be achieved bycontrolling the size of the hair pieces.

Arsenic species in human hair

The concentration of the arsenic species in water extracts ofhuman hair samples from Esquina and Illapata was assessed.As(III), As(V), MMA and DMA concentrations were measured

Fig. 1 Typical chromatogram of four arsenic species in water: As(III), DMA(V), MMA(V) and As(V) 50 mg L�1 each. Separation conditions aredescribed in text.

Table 1 Arsenic species in Esquina and Illapata water sources

Village Water source

As in water/mg L�1

As(III) DMA MMA As(V) Sum

Esquina Waterfall 1 1.1 N.D. N.D. 72.9 74.0

Waterfall 3 1.0 N.D. N.D. 11.2 12.2

Illapata Camarones river 5.0 N.D. N.D. 1247 1252

Waterfall 1 1.4 N.D. N.D. 47.3 48.7

N.D. ¼ not detected.

1 3 3 8 J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1

using HPLC-HG-ICP-MS. The results indicate greater abun-dance of inorganic species (As(III) and As(V)), which repre-sented close to 98% of the total arsenic extracted from the hairsamples in both populations. In contrast, the organic species ofarsenic have lower concentrations, representing less than 2% ofthe total arsenic in hair. As(III) was detected in hair extracts of95% of the individuals from the Esquina who are exposed to alower arsenic concentration. In all the calculations of arsenicspecies concentration in hair, we have assumed that the speciesfound in extracts represent the original species in the hair.Nevertheless changes of the original species can occur whenarsenic is extracted. This issue is still not clear and furtherstudies are required. Considering that the concentrations inextracts represent the original species in the hair, the As(III)average concentration was 0.25 mg g�1 (median value was 0.14mg g�1. As(V) was detected in 70% of the individuals with anaverage concentration of 0.15 mg g�1 (median value 0.11 mgg�1). Considering the methylated species, only MMA(V) wasdetected in 40% of the samples with an average and mediavalue of 0.02 mg g�1. DMA(V) was not detected in any extractfrom exposed individuals from Esquina.

In the highly exposed population from Illapata, both inor-ganic arsenic species, As(III) and As(V), were detected in allextracts, presenting an average concentration 3.75 and 0.45 mg

g�1, respectively. Mean values of As(III) and As(V) were 2.64and 0.27 mg g�1, respectively. MMA(V) and DMA(V) weredetected in 71 and 24% of the samples from highly exposedindividuals with an average concentration of 0.16 and 0.04 mgg�1, respectively. An interesting result was that four of the fivesamples where DMA(V) was detected, corresponded to chil-dren. No conclusion can be made due to the small number ofsamples. Table 2 presents the mean values of the four Asspecies for the two villages studied.Fig. 3 presents a typical chromatogram of the extracted

species from hair of an exposed individual (male, 75 yearsold) containing 15 mg g�1 of total arsenic, distributed as 13.7As(III); oD.L. (0.07) DMA(V); 0.15 MMA(V) and 1.14 As(V)(all in mg g�1). An unidentified As species was detected in hair,eluting between MMA and As(V).The arsenic species were also analyzed in human hair CRM

(GBW 09101 No 18), which was used for recovery andaccuracy studies. CRM was not washed prior to the speciationanalysis. Human hair CRM contains a certified total Asconcentration of 0.59 � 0.07 mg g�1. Nevertheless the concen-tration of each arsenic species is not certified in CRM. Nowa-days there is no available CRM of human hair containingcertified concentration of arsenic species. The concentrationsfound in CRM were 0.14, 0.04 and 0.21 mg g�1 for As(III),MMA and As(V), respectively. The sum (0.39 mg g�1) repre-sents only 66% extraction. This result confirms that the mainproportion of As in hair is due to inorganic species, and thatorganic species represent a minimum fraction. Recently, Raaband Feldmann have reported the predominance of inorganicarsenic.22 They also found low arsenic recoveries when usingboiling water leaching (68.9%).Moreover, comparing the species distribution obtained in

the exposed population and CRM, some differences have beenfound. As(III) represents only 36% in CRM, contrasting with68 and 88% found in Esquina and Illapata, respectively. Theconcentration of As(V) in the hair CRM represents 53% of theextracted arsenic in contrast to the 30 and 11% found inEsquina and Illapata, respectively. The higher proportion ofAs(V) could occur due to exogenous arsenic in CRM (noprevious washings were performed). No DMA was detectedin this CRM, coinciding with the low concentration found insamples, where DMA was detected only in 5 of 43 samples.

Correlation of arsenic species in human hair with exposure time

The correlation between inorganic species concentration andthe exposure time of the highly exposed population of Illapatawas assessed. Even though the number of sampled individuals(N ¼ 43) is limited, the number of studied individuals repre-sents an important proportion of the total population living inthe studied villages (45 and 35% in Esquina and Illapata,respectively) validating interpretation regarding with thosepopulations.

Fig. 2 Plotted regression between total arsenic concentration in hair(mg g�1) and (a) age of the sampled individual (years); (b) residencetime (years) in Illapata.

Table 2 Arsenic species (As(III), As(V), MMA(V) and DMA(V)) in hair of the Esquina and Illapata populations

Village Population

As in hair, mg/g average (min. � max.)

As(III) DMA MMA As(V)

Esquina (n ¼ 22) Children (n ¼ 11) 0.40 (N.D.–1.53) N.D. 0.01 (N.D.–0.02) 0.15 (N.D.–0.37)

Adults (n ¼ 11) 0.13 (0.02–0.71) N.D. 0.02 (N.D.–0.03) 0.17 (N.D.–0.41)

SD 0.37 0.01 0.11

Average Esquina 0.26 N.D. 0.02 0.15

Illapata (n ¼ 21) Children (n ¼ 08) 2.15 (0.56–3.24) 0.18 (N.D.–0.35) 0.02 (N.D.–0.03) 0.18 (0.1–0.47)

Adults (n ¼ 13) 4.74 (0.23–13.72) 0.07 (N.D.–0.07) 0.05 (N.D.–0.15) 0.614 (0.21–1.52)

SD 3.87 0.11 0.04 0.38

Average Illapata 3.75 0.16 0.04 0.45

For calculation of averages, only values over the detection limits were considered.

J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1 1 3 3 9

People in both villages have been drinking contaminatedwater as long as they have lived there. The degree of exposurecan be considered to be proportional to the individual’s ageand residence time in village (exposure time). For this popula-tion, the calculated As(III) and As(V) concentrations in haircorrelates with the age of the sampled individuals, presenting acorrelation coefficient of 0.6923 and 0.5678, respectively. Thecorrelations of As(III) and As(V) with the time of residence inIllapata are 0.7109 and 0.5799, respectively. Fig. 4 shows theplotted regression of As(III) and As(V) in relation with the ageand time of residence for the Illapata population. For thestudied population, As(III) provided a better correlation thanthe total arsenic and As(V). The better correlation betweenAs(III) and time of residence indicates that As(III) concentrationin hair should be more closely related with the degree ofexposure in the studied individuals. The explanation for thisphenomenon is still not clear, although it is likely that theexplanation depends on understanding the arsenic inclusionprocesses in hair. It is not known, if during inclusion in hair,

reduction of arsenic occurs, as occurs in the As-methylationpathways into the liver. Additionally, external contaminationdue to As(V) increases the possibility of an interpretation errorsince almost 99% of the external arsenic (from the CamaronesRiver) is in the form of As(V). Further experiments are requiredto explain this phenomenon. Additionally, another studysuggested the limitations of hair analysis in exposure assess-ment.31 According to our results, As speciation in human hairoffers more complete analytical information that should permitbetter assessment in As-exposure studies.

Conclusions

In both villages, the arsenic concentration in drinking watersurpasses the recommended WHO value (10 mg L�1). Thepopulations of Esquina and Illapata are exposed to As-con-taminated drinking water exceeding between 7.5 and 125 timesthe international referenced values. The result of arsenic

Fig. 3 Typical chromatogram of water-extracted arsenic species from a highly exposed individual from Illapata (male, 75 years old and 60 yearsresidence time in Illapata). Insert shows reduced scale of the same chromatogram.

Fig. 4 Correlations of As species in hair in highly exposed population of Illapata. Plotted regression between: (A) As(III) in hair (mg g�1) and age ofthe sampled individual (years); (B) As(III) and residence time (years) in Illapata; (C) As(V) and age of the sampled individual (years); (D) As(V) andresidence time (years).

1 3 4 0 J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1

speciation in water sources shows that theses populations areprincipally exposed to inorganic arsenic, mainly As(V).

Arsenic species found in water extracts of human hair aremainly inorganic arsenic (close to 98%), where As(III) is theprincipal species. This high proportion of inorganic arsenic wasfound in both studied populations. Considering that the con-centrations in extracts represent the original species in the hair(no changes during extraction procedure), mean concentra-tions of As(III) were 0.26 mg g�1 (68%) and 3.75 mg g�1 (88%)in Esquina and Illapata, respectively. As(V) was found as wellin hair extracts, although at lower concentration than As(III) inboth studied populations with values of 0.15 mg g�1 (30%) and0.45 mg g�1 (10.5%) in Esquina and Illapata, respectively.Further studies are necessary to have a better understandingof possible changes during the extraction of original arsenicspecies from the hair.

The presence of inorganic arsenic as the principal arsenicspecies in hair agrees with the arsenic distribution found in ahair CRM, although in CRM the major part of arsenic isAs(V). Standard reference material of human hair with certifiedarsenic species is strongly needed for better analytical assess-ment.

In the studied populations, total arsenic, As(III) and As(V)concentration in hair correlated with the degree of exposure,but the As(III) concentration exhibits better correlation, withfactors such as individual’s age and time of residence in thevillage. The correlations between total arsenic, As(III), As(V)and the age of the exposed individuals were 0.65, 0.69, 0.56,respectively. The correlations with the time of residence were0.54, 0.71 and 0.58, respectively. These results indicate that, atleast in the studied populations, As(III) was the most accurateindicator for chronic As(V) exposure. Because of the bettercorrelation found for As(III) in this study, arsenic speciationappears as a promissory tool for more complete analyticalassessment in epidemiological studies on arsenicism. Furtherstudies are required to confirm our results using larger numberof samples of exposed populations with well-characterizedarsenic intake. Organic arsenic species are unlikely indicatorssince they are present in very low and variable concentrationsin comparison to inorganic arsenic and they were not detectedin all the samples.

Acknowledgements

The authors acknowledge the Fulbright Alumni InitiativeAward Program (AIA), Grant AIA-FY2001, for its financialsupport, and the cooperation of the University ResearchInstitute for Analytical Chemistry (URIAC), Amherst, MA,USA.

References

1 Element Speciation in Bioinorganic Chemistry, ed. S. Caroli, JohnWiley & Sons, Inc., New York, 1196, pp. 445–463.

2 B. K. Mandal and K. T. Suzuki, Arsenic round the world: areview, Talanta, 2002, 58, 201–235.

3 Arica inserta en una region arsenical: El Arsenico en el Ambienteque la Afecta y 45 Siglos de Arsenicismo Cronico, ed. L. Figueroa,Editorial Universidad de Tarapaca, Arica, Chile, 1st edn, 2001.

4 A. H. Smith, P. Lopipero, M. N. Bates and C. M. Steinmaus,Arsenic epidemiology and drinking water standards, Science,2002, 296, 2145–2146.

5 World Health Organization (WHO), Guidelines for DrinkingWater Quality, 1993, Vol. 1, Recommendations.

6 Environmental Protection Agency, 40 CFR Parts 9, 141 & 142.National Primary Drinking water Regulations; arsenic and Clari-fications to Compliance and New Source Contaminants Monitor-ing: Final rule, 2001.

7 M. Styblo, Z. Drobna, I. Jaspers, S. Lin and D. J. Thomas,Environ. Health Perspect., 2002, 110(Suppl. 5), 767–771.

8 D. J. Thomas, M. Styblo and S. Lin, Toxicol. Appl. Pharmacol.,2001, 176, 127–144.

9 National Research Council (NRC), Subcommittee on Arsenic inDrinking Water, Arsenic in Drinking Water, National AcademyPress, Washington, DC, 1999, vol. 2, pp. 16–25.

10 M. Ma and X. Le Chris, Effects of arsenosugar ingestion onurinary arsenic speciation, Clin. Chem., 1998, 44(3), 539–550.

11 G. Samantaa, U. K. Chowdhurya, B. K. Mandala, D. Chakra-bortia, N. C. Sekaranb, H. Tokunagab and M. Ando, Highperformance liquid chromatography inductively coupled plasmamass spectrometry for speciation of arsenic compounds in urine,Microchem. J., 2000, 65, 113–127.

12 X. Le Chris, X. Lu, M. Ma, W. R. Cullen, H. V. Aposhian and B.Zheng, Speciation of key Arsenic methabolic intermediates inhuman urine, Anal. Chem., 2000, 72, 5172–5177.

13 L. M. Del Razo, M. Styblo, W. R. Cullen and D. J. Thomas,Determination of trivalent methylated arsenicals in biologicalmatrices, Toxicol. Appl. Pharmacol., 2001, 174, 282–293.

14 Y. C. Chen, C. J. Amarasiriwardena, Y. M. Hsueh and D. C.Christiani, Stability of arsenic species and insoluble arsenicin human urine, Cancer Epidemiol., Biomarkers Prev., 2002, 11,1427–1433.

15 Z. Gong, X. Lu, W. R. Cullen and X. Le Chris, Unstable trivalentarsenic metabolites, monomethylarsonous acid and dimethylarsi-nous acid, J. Anal. At. Spectrom., 2001, 16, 1409–1413.

16 J. Yu, D. W. Yu, D. M. Checkla, I. M. Freeberg and A. P.Bertolino, Human hair keratins, J. Invest. Dermatol., 1993,101(Suppl 1), 56S–59S.

17 J. T. Hindmarsh, Arsenic, its clinic and environmental signifi-cance, J. Trace Elem. Exp. Med., 2000, 13, 165–172.

18 J. T. Hindmarsh, Caveats in hair analysis in chronic hair poison-ing, Clin. Biochem., 2002, 35, 1–11.

19 B. K. Mandal, Y. Ogra and K. T. Suzuki, Speciation of arsenic inhuman nail and hair from arsenic-affected area by HPLC-induc-tively coupled argon plasma mass spectrometry, Toxicol. Appl.Pharmacol., 2003, 189, 73–83.

20 A. Shraim, S. Hirano and H. Yamauchi, Extraction and specia-tion of arsenic in hair using HPLC-ICPMS, Anal. Sci., 2001, 17,i1729–i1732.

21 A. Raab, H. Hansen, L. Zhuang and J. Feldmann, Arsenicaccumulation and speciation analysis in wool from sheep exposedto arsenosugars, Talanta, 2002, 58, 67–76.

22 A. Raab and J. Feldmann, Arsenic speciation in hair extracts,Anal. Bioanal. Chem., 2005, 381, 332–338.

23 E. Hakala and L. Pyy, Selective determination of toxicologicallyimportant arsenic species in urine by high performanceliquidchromatography hydride generation atomic absorption spectro-metry, J. Anal. At. Spectrom., 1992, 7, 191–196.

24 X. Le Chris, X. Lu and X. G. Li, Arsenic speciation, Anal. Chem.,2004, 1, 27A–33A.

25 A. Shraim, S. Hirano and H. Yamauchi, Extraction and specia-tion of arsenic in hair using HPLC-ICPMS, Anal. Sci., 2001, 17,i1729–i1732.

26 X.-Y. Wei, C. A. Brockhoff-Schwegel and J. T. Creed, Applicationof sample pre-oxidation of arsenite in human urine prior tospeciation via on-line photo-oxidation with membrane hydridegeneration and ICP-MS detection, Analyst, 2000, 125, 1215–1220.

27 International Atomic Energy Agency (AIEA), Activation analysisof Hair as an Indicator of Contamination on Man by EnvironmentalTrace Element Pollutants, AIEA/RL/50, Vienna, 1978.

28 J. Morton, V. A. Carolan and P. H. E. Gardiner, Removal ofexogenously bound elements from human hair by various washingprocedures and determination by inductively coupled plasma,Anal. Chim. Acta, 2002, 21759, 1–12.

29 E. M. Flores, A. P. F. Saidelles, J. S. Barin, S. R. Mortari and A.Figueiredo Martins, Hair sample decomposition using polypro-pylene vials for determination of arsenic by hydride generationatomic absorption spectrometry, J. Anal. At. Spectrom., 2001, 16,1419–1423.

30 Y. Bohari, A. Astruc, M. Astruc and J. Cloud, Improvements ofhydride generation for the speciation of arsenic in natural fresh-water samples by HPLC-HG-AFS, J. Anal. At. Spectrom., 2001,16, 774–778.

31 H. Yamauchi, K. Takahashi, M. Mashiko and Y. Yamamura,Biological monitoring of arsenic exposure of gallium arsenide- andinorganic arsenic-exposed workers by determination of inorganicarsenic and its metabolites in urine and hair, Am. Ind. Hyg. Assoc.J., 1989, 50, 606–612.

J . E n v i r o n . M o n i t . , 2 0 0 5 , 7 , 1 3 3 5 – 1 3 4 1 1 3 4 1