Toxicology Letters 156 (2005) 241–251

The effect on human eye blink frequency of exposure tolimonene oxidation products and methacrolein

Jacob Klenø Nøjgaard, Karl Bang Christensen, Peder Wolkoff∗

National Institute of Occupational Health, Lersø Parkall´e 105, DK-2100 Copenhagen Ø, Denmark

Received 25 February 2004; received in revised form 25 November 2004; accepted 26 November 2004

Abstract

Oxidation products of terpenes (e.g. limonene) contain unidentified irritants, which may be responsible for a fraction ofthe reported eye and airway complaints in indoor environments. Here we report exposure to parts per billion (ppb) levels oflimonene oxidation products (LOPs) and the terpene oxidation product methacrolein using blink frequency (BF) as a measureof trigeminal stimulation of the human eye. Ten male subjects averaging 43 (standard deviation 10.5) years were exposed for20 min to LOPs, methacrolein, and clean air, respectively. A baseline BF was measured prior to and following each exposure(8 min and 4 min, respectively). The subjects were exposed locally in the non-dominant eye and single blind at 20% relativehumidity (RH), while viewing an educational film. Blinking was video recorded and evaluated for full sessions of 36 min. MeanBF increased significantly during exposure to LOPs and methacrolein compared to the baseline of clean air, and the findingscoincided with weak eye irritation symptoms. Lowest observed effect levels were 286 ppb methacrolein and a 10-min-old LOPsm( s at 50%R ndp ngs. Thee©

K

1

ot

rected

yein-ndspec-

0

ixture of initially 92 ppb limonene and 101 ppb ozone (O3), which increased the BF comparably by 18% (p= 0.001) and 17%p= 0.003), respectively. The increase in BF was smaller, although not significantly different, during exposure to LOPH to 20% RH in mixtures prepared from ca. 350 ppb limonene and 300 ppb O3. LOPs may cause trigeminal stimulation aossibly eye irritation at O3 and limonene concentrations, which are close to high-end values measured in indoor settiffects may be exacerbated by low RH.2004 Elsevier Ireland Ltd. All rights reserved.

eywords:Blink frequency; Eye irritation; Methacrolein; Terpene oxidation products

. Introduction

Dryness, lacrimation and strained, burning, grittyr itchy eyes are common symptoms of “eye irrita-

ion” in epidemiological studies of the indoor envi-

∗ Corresponding author. Tel.: +45 39165272.E-mail address:pwo@ami.dk (P. Wolkoff).

ronment (Brightman and Moss, 2000). The causes anot well documented, although a number of suspeindoor risk factors have been explored (Wolkoff etal., 2003). The apparent difficulty of explaining eirritation including airway irritation by measureddoor concentrations of volatile organic compou(VOCs), aldehydes, and particles in offices, res

378-4274/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.toxlet.2004.11.013

242 J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251

tively (Wolkoff et al., 2003; Schneider et al., 2003) hasprompted the hypothesis that hitherto not measured re-active chemical species could be partly responsible forthe reported irritation (Wolkoff and Nielsen, 2001). Atpresent, it appears that products formed in the oxida-tion of certain biogenic VOCs, terpenes, contribute toeye and airway irritation as supported by mouse bioas-say and human exposure studies (Wolkoff and Nielsen,2001). In addition, the effect of these terpene oxidationproducts appears to decrease with increasing relativehumidity (RH) (Wilkins et al., 2003).

Terpene oxidation products are formed in the reac-tion of oxidants (e.g. ozone (O3), the hydroxyl radi-cal (OH), and the nitrate radical (NO3)) with terpenesin the presence of nitrogen oxides. Terpenes are emit-ted from vegetation and certain wood-based buildingmaterials in new housing (Hodgson et al., 2000), andadded to household products (e.g. the use of pine andcitrus oils in cleaning agents). Limonene,�-pinene andisoprene are examples of often measured terpenes in-doors (Wolkoff et al., 2000). High concentrations oflimonene can be obtained at low air exchange rates inthe absence of O3. Typical mean indoor limonene con-centrations are less than 10 parts per billion (ppb), butvalues as high as 70 ppb have been reported (Wolkoff etal., 2000; Sexton et al., 2004). The human sensory irri-tation threshold forR-(+)-limonene has been estimatedto be more than 80 parts per million (ppm) based on2 h chamber experiments (Falk-Filipsson et al., 1993).However, using the method in the present study, noa ob-s eneac lev-e dso er.W to con-t1 nw ppbO -t thes t. InO rod-u thylv OHa ,

2003). In addition to the stable irritating oxidation prod-ucts such as formaldehyde and methacrolein, terpeneozonolysis is a major source of indoor radicals, whichare suspected to cause eye- and airway irritation (Klenøand Wolkoff, 2004; Wolkoff and Nielsen, 2001). Theidentity of the radicals, which are formed in terpene ox-idation includes OH, Criegee Intermediates, and otherorganic radicals (Atkinson and Arey, 2003).

We have recently applied increases of the human eyeblink frequency (BF) as a measure of trigeminal stim-ulation of the eye (Klenø and Wolkoff, 2004). We ob-served that the BF increased significantly during 20 minexposure to limonene oxidation products (LOPs) pre-pared from ca. 200 ppb limonene and 130 ppb O3,which had reacted for 10 min at 20% RH. Thesefindings supported the hypothesis that terpene oxida-tion products can partly explain the prevalence of re-ported eye complaints in the indoor environment, al-though the applied reactant concentrations were atypi-cally high. Two major questions arose from that study,namely (1) whether LOPs prepared from lower reac-tant concentrations can produce eye irritation in hu-man subjects, and (2) how does stable terpene oxi-dation products (e.g. methacrolein) affect changes ofthe BF.

LOPs are known to cause eye and airway irritationby trigeminal stimulation in humans and mice, respec-tively (Wolkoff et al., 2000; Klenø and Wolkoff, 2004).The causative irritating products have yet not been iden-tified, which implies that a biological response cannotb s int mso eenrib c-t imet , thed .a

−

−

dverse eye-irritation or trigeminal response waserved during 20 min exposure to 116 ppb limont 20% RH (Klenø and Wolkoff, 2004). Indoor O3oncentrations are typically 20–70% of outdoorls (Weschler, 2000) varying from tenths to hundref ppb in polluted cities especially during summhile the outdoor source of O3 is the predominan

ne, electrostatic equipment, like photocopiers,ributes to the overall indoor O3 concentration (Brown,999; Leovic et al., 1996). No adverse eye-irritatioas reported during 20 min human exposure to 403 at 20% RH (Klenø and Wolkoff, 2004). The iden

ity of the terpene oxidation products depends ontructure of the reacting terpene and the oxidan3 oxidation (ozonolysis) of terpenes, abundant pcts are formaldehyde, methacrolein, OH and meinyl ketone for isoprene (a hemiterpene), andnd formaldehyde for limonene (Atkinson and Arey

e related to the concentration of a single speciehe reaction mixture. Trigeminal stimulation in terf airway irritation and thereby eye irritation has belated to the amount of chemically reacted O3, thats the reaction extent (Wilkins et al., 2003). LOPs cane generated by mixing O3 and limonene in a rea

ion flow tube, which imposes a specific reaction to the exposure mixture. As the reaction proceedsecay of reactants can be estimated by the Eqs(1)nd(2):

d(limonene)

dt= k1[limonene][O3] + k2[limonene][OH]

(1)

and

d(O3)

dt= k1[limonene][O3] + k3[O3], (2)

J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251 243

where the residual reactant concentrations are shownin brackets, andk1, k2 are second order rate constantsfor the reaction of limonene with O3 and OH, respec-tively.k3 is the first order rate constant for O3 depositiononto the flow tube surface. A limonene oxidation mix-ture absent from direct sunlight (i.e. photochemistry isnegligible) can be described by RH, reaction time, thereaction extent (the fraction of chemically convertedreactants) and the initial reactant concentrations. RHaffects the oxidation product distribution, e.g. the for-mation of hydroperoxides is favored by higher watercontent (Sauer et al., 1999). Some oxidation productsbuild up in time as the reaction proceeds and the re-actants are consumed. For example, the molar yield offormaldehyde is 19% of the reacted limonene concen-tration (Ruppert et al., 1999). Other products like theOH radical depends on the residual O3 concentration(Weschler and Shields, 1996), since they react almostinstantaneously and the driving force for OH produc-tion is ozonolysis of terpenes. Other terpene oxidationproducts show an even more complicated dependencyof time, RH and concentration of reactants and prod-ucts. In this work, the progress of the reaction will beexpressed in terms of chemically reacted O3 (as op-posed to physically consumed O3, i.e. deposition insidethe reaction tubes as wall loss). The more chemicallyreacted O3 for a particular LOPs mixture, the morelimonene consumption and higher yield of stable irri-tating products like formaldehyde. Since the concentra-tion of the OH radical depends on the O3 concentration,i mer top onso OHr xpo-s irri-t ther O

undo d asc OON thero tionw holdL

by multiplying with 0.03 (Alarie, 1973; Schaper, 1993).Furthermore, since eye- and airway irritation are bothmediated through the trigeminal nerve, eye irritation inhumans can be estimated and compared to the findingsin the present study. The estimated TLV which shouldnot alter the human BF is 10,400 ppb× 0.03∼ 300 ppb.

The purpose of this work was to evaluate BF changesand reported irritation when exposed to increasing con-centrations of: (1) methacrolein, (2) LOPs from threemixtures prepared from increasing reactant concentra-tions, and finally (3) two LOPs mixtures with high ver-sus low RH.

2. Experimental

2.1. Chemicals

Oxygen (99.999% pure) was from Hydrogas, Nor-way. O3 was generated photochemically by irradiat-ing molecular oxygen with a thermostated mercurylamp controlled by a high performance power supply(Wolkoff et al., 2000). Limonene (>99.9% pure) wasfrom Fluka and methacrolein (95% pure) was fromAldrich. A gas generator (Model 491M, KIN-TEK,TX) generated limonene and methacrolein vapors byevaporation of the liquids into a filtered air stream ofmedical grade. Clean humidified air was generated in aseparate air supply from charcoal filtered air of medicalgrade. In the supply unit, the air stream was split in a dryc cleand ingt

2

ad-j be-f nicb n,N cali-b bH& re-aR mea-s plep owt lein

ts concentration is similar in mixtures with the saesidual O3 concentration. It is therefore possiblerepare LOPs mixtures with different concentratif stable irritants, but similar concentration of theadical. Dose-response information from human eure to LOPs can thus provide information on theating product species: whether they build up witheaction extent or depend on the concentration of3.

Methacrolein was chosen as a model compof a (fairly) stable terpene oxidation product antrong airway irritant (Larsen and Nielsen, 2000). Itan be formed indoors by oxidation of isoprene by3,H and NO3 (Atkinson and Arey, 2003). Larsen andielsen (2000)studied the effect of methacrolein on

espiratory tract in a mouse bioassay. A RD50 valuef 10,400 ppb was reported, that is the concentrahich reduces the respiratory rate by 50%. Thresimit Values (TLV) can be estimated from RD50 values

hannel and one passing through a sparger witheionized water. The RH was adjusted by combin

he two channel flows.

.2. Measurements

All gas flows of oxygen, vapors and air wereusted daily by flow controllers and measuredore and after sampling with a traceable electroubble flow meter (Gillian Instrument CorporatioJ). Temperature and RH were measured by arated hygrometer (Model Testo 601, Testoterm Gm

Co., Germany). Initial concentrations of thectants were measured at sample point I inFig. 1.esidual concentrations of the reactants wereured at the end of the Teflon tube at samoint II, which connected the polyethylene (PE) fl

ube with the eyepiece. Limonene and methacro

244 J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251

Fig. 1. Experimental set-up for exposure of the human eye.

were measured daily by sampling on Tenax TA ad-sorbent tubes followed by thermal desorption–gaschromatographic–flame ionization detection. The sam-pling rate was 200± 3 ml min−1 using pump II (Ame-tek, PA). Analytical details are described elsewhere(Wolkoff, 1998). O3 was monitored continuously in thereaction mixture using a newly calibrated chemilumi-

nescence monitor (Model 265 A, API Inc., San Diego,CA). The limit of detection was 2 ppb. O3 was partlyconsumed in surface reactions on the PE reaction flowtube (wall loss). A first order rate constant was mea-sured in order to address the wall loss of O3. For thisparticular reason, the chemical loss of O3 was modeledin the four LOPs mixtures. See Section ‘2.6 Modeling

J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251 245

of the chemical reaction’ for details. The loss of O3 tothe walls of the PE reaction flow tube was measured atthe inlet (sampling point I) and outlet (sampling pointII) for the applied flow and O3 concentration.

The total number of ultrafine particles was measuredat sampling point II by a condensation particle counterin the size range 7–1000 nm and a detection limit of2 particles cm−3 (Model 3022A, TSI Inc., MN). Thedata were averaged over 10 min.

2.3. Generation of LOPs and methacrolein

LOPs were generated by mixing flows of limonenevapor (600 ml min−1), O3 (50 ml min−1), and a variableflow of air in a Teflon tube (l: 1.3 m, i.d.: 2 mm) con-nected to a wide bore PE reaction flow tube (l: 4.7 m,i.d.: 2.2 cm). The flow through the PE reaction tube(180 ml min−1) was adjusted by pump I (Model s2500,Dupont, DE) connected by a T-union to the Teflon tubeupstream the PE reaction flow tube. The flow tubewas cleaned prior to each session with 500 ppb O3(4 l min−1) for 120 min in order to remove adsorbed or-ganics. Initial reactant concentrations were measuredprior to reaction (Table 1). Following 10 min of reac-tion (i.e. the time it takes for the mixture to exit the PEreaction flow tube) the residual reactant concentrationswere measured (Table 1). The reacted amount of O3was calculated to be 32–79% from measurements of

the residual O3 concentrations and modeling of the O3wall loss (Section2.6).

Methacrolein was generated in the same systemby replacing limonene with methacrolein in the gasgenerator, and O3 turned off. All exposure concentra-tions were adjusted to 20± 3% RH and temperature21± 2◦C, except for the LOPs IV mixture which was50± 3% RH. The mixing ratio of oxygen was 0.23 inthe LOPs experiments and otherwise 0.21.

2.4. Human exposure

Ten healthy male subjects averaging 43 years (stan-dard deviation 10.5 years) participated in the study.They were non-smokers, absent from any pathologicaleye history and used no systemic medication likely topromote dry eyes (Craig, 2002; Doughty et al., 1997).Women were excluded from the study, since they areknown to be more sensitive than men with respect toreporting irritation symptoms (Wolkoff et al., 2003),which implies that the inclusion of both genders wouldrequire more subjects. The result of this study is there-fore not representative of typical office workers, butmerely serve to compare effects of methacrolein, LOPsand RH on changes of BF. Other exclusion criteriawere deviations from a healthy general condition suchas a common cold or fatigue. All subjects carried outoffice work in the same building. In the preceding 2

TE dation

R ME

L –O –

RL –O –M 2P <2

OO –T 21±R 20±

cified c n in a PEfl viation

the stamed in face (wall

l

able 1xposure conditions for methacrolein (MET) and limonene oxi

eactant concentrationsa Clean air MET I MET II

imonene (ppb) – – –

3 (ppb) – – –

esidual reactant and product concentrationsb

imonene (ppb) – – –

3 (ppb) – – –ethacrolein (ppb) – 89 (1) 189 (3)articles (cm−3) – <20 <20

ther parametersc

3 reacted (ppb/%) – – –emperature (◦C)elative humidity (%)

a LOPs are formed by mixing limonene and ozone in the speow tube. Mean concentrations are listed with one standard deb The mixtures are not fully reacted and contain reactants ofc The term “O3 reacted” refers to the amount of ozone consu

oss) according to Eq.(4).

products (LOPs)

T III LOPs I LOPs II LOPs III LOPs IV

92 (3) 204 (2) 343 (11) 350 (3)101 (1) 180 (1) 304 301

41 (1) 74 (15) 111 (12) 110 (9)39 (1) 38 (2) 40 (2) 40 (5)

86 (2)0 <60 <60 3.6 (10) 104 3.1 (10) 104

18/32 71/65 155/79 154/7923 50± 3

oncentrations (parts per billion (ppb)) following 10 min reactioin brackets.

ted residual concentrations.chemical reactions, excluding deposition onto the tubing sur

246 J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251

months before the experiments they were not exposedto methacrolein, O3 or limonene, other than to the ubiq-uitous background concentrations of these compounds,and other species encountered in typical offices. Priorwritten consent was obtained from the subjects, andthe study was approved by the local Danish ResearchEthics Committee (01-009/03). The subjects were lo-cally exposed single blind in their non-dominant eyeusing a specially designed transparent glass eyepiece,which was designed to fit the facial curve ensuring aneffective air supply at atmospheric pressure. The lin-ear velocity was calculated to be less than 4 cm s−1 ata flow of 180 ml min−1, and the subjects did not re-port any discomfort with respect to the air flow. Theair flow itself may still produce irritation causing anincrease of BF in time, irrespective of the exposure.Such an unspecific BF increase with time was observedfor clean air exposure sessions equaling 1.1% min−1

on average (Table 2). A time effect was therefore in-corporated in the model (see Section2.5). The gazedirection was slightly upward. The presence of the eye-piece may have caused some minor visual disturbances,although the dominant eye was unaffected, but the sub-jects did not pay attention to the eyepiece after a fewminutes. The subjects were neither informed about thenature of stimuli nor could they smell it, because thelow air flow exited the eye-piece above the ear follow-

ing dilution into ambient air. Blinking and commentson perceived eye irritation (intensity, location, and de-scription) were recorded using a digital video camera(Sony DCR-PC110E PAL; 25 frames per second). Thesubjects reported the intensity of the perceived irri-tation prior to the exposure on a linear scale equallydivided into none, weak, moderate and strong irrita-tion. It was stated whether irritation was located in theeye (left/right), on the skin, or at the eyelids. The sub-jects rated the intensity by marking a scale, and theywere allowed to state intermediate intensities, e.g. “be-tween weak and moderate”. Perceived irritation duringexposures was verbally assigned to this scale in shortcomments, e.g. “weak irritation, left eye ball”. An ed-ucational video film was shown to avoid drowsinessduring the entire session of 36 min.

Each of the eight sessions was made up of foursuccessive stages: an acclimatization stage, which in-cluded 3 min of clean air; an initial baseline record-ing with 8 min of clean air (stage A); 20 min of eithermethacrolein, LOPs or clean air (stage B); and finally8 min of clean air (stage C). The latter stage was dividedinto 4 min of recovery and 4 min of a final baselinerecording. The two exposure shifts in stages B and Cwere carried out by turning a tap on an exposure con-troller. When clean air was on, the LOPs mixture ormethacrolein exited a vent and vice versa. A partition

Table 2B ) and l

T M

S 4/1

P eak < k; weakR 18

C 7; 3p 0.B 1.2L 286T 300

articularticul ; “<weak”;

“ ion.ive to b the

p ual tofinal b

ease inn and

link frequency change during exposure to methacrolein (MET

est results/compounds Clean air MET I MET II

ubjects whom perceivedeye irritationa

3/10 4/10 5/10

erceived intensityb <weak <weak; weak <weak; welative change in blinkfrequency (%)c

−9 10 8

onfidence interval (%) −19; 3 −2; 23 −2; 20-value 0.14 0.09 0.13ase line trend (% min−1)d 1.1 0.8 1.5OEL (ppb)e

LVf

a The number of subject, who perceived eye irritation in the pb The average reported intensity of perceived irritation in the p

weak”; “<moderate”; “moderate”; “<strong” and “strong” irritatc The results are stated as the blink frequency change relat-value, which is the probability of the blink frequency being eqd The modeled increase in BF from the initial (stage A) to thee The Lowest Observed Effect Level (LOEL) showing an incrf The proposed occupational Threshold Limit Value (TLV) (Larse

imonene oxidation products (LOPs)

ET III LOPs I LOPs II LOPs III LOPs IV

0 4/10 4/10 5/10 4/10

weak; weak <moderate weak; <moderate weak <wea17 12 34 22

0 5; 30 9; 25 18; 52 8; 37001 0.003 0.03 <0.0001 0.002

0.1 0.4 0.6 0.718

ar session.ar session. The intensity scale is linear and ranges from “none”

aseline (see Section2.5 for details); the 95% confidence interval andbaseline.aseline (stage C) in the particular session.BF statistically different from that of clean air.

Nielsen, 2000).

J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251 247

wall separated the subjects from the exposure appara-tus. The experiments were conducted in the morningin the spring season, and there was a minimum of twodays between each session to avoid adaptation to theexposures. The temperature was 21± 2◦C and the RHwas 31± 9% in the laboratory (seeTable 1).

2.5. Modeling of blink frequency changes

A previously developed set-up for exposure of thehuman eye to irritants is used in this study (Klenøand Wolkoff, 2004). However, the original procedurewas slightly modified. First, the subjects were unawareabout the exposure shifts in stages B and C. It was rec-ognized, that a visual shift between exposures couldinfluence the subjects, although unaware of the typeof exposure. Second, the present study involved expo-sure to higher concentrations of irritants in stage B,and a possible carry-over effect from stage B to the fi-nal baseline recording of session C could increase theBF, and thereby underestimate the effect of the irri-tants. The BF changes were therefore modeled in twodifferent runs: one with a 4-min-recovery period im-posed between stage B and C, and one without. Whenthe recovery period was included, the BF changes weresystematically higher in every methacrolein and LOPsexposures (the clean air session was unaffected), whichindicated a carry-over effect and justified inclusion ofthe recovery period. Time trends in BF from stage Ato C were modeled for the clean air session, and themi llert s as-s

allv ca.5 ma-t n ac t oft rneaw ntso ndedc achs diess han2 on-o uret ern

Fig. 2. Five examples of alternating eye blink frequency curves dur-ing exposure to clean air.

(Fig. 2). Due to the alternating BF with time, blinkmeasurements shorter than ca. 4–5 min are consideredunreliable in accordance withDoughty, (2001). Thealternating pattern was conserved during exposure tomethacrolein and LOPs.

Blinking data was transformed by taking the log-arithm of the BF, since the variation increased withthe BF. A linear mixed model could then be used tomodel the BF changes. The eight exposure sessionswere modeled separately. For each of the eight expo-sure sessions, data consisted of measurements at 32time points (timet= 4, 5,. . ., 31 min andt= 36, 37, 38,39 min) on ten subjects. The first 3 min was acclima-tization, and a 4-min recovery period was imposed att= 32, 33, 34, 35 min. The logarithm of the BFtp at timet, for person p was modeled using a linear mixed model:

log(BFtp) = β + µp + tγ + I(t)δe + εtp

µp ∼ N(0, σ2B)

εtp ∼ N(0, σ2W),

using a fixed session levelβ, a random person levelµp, a linear effectγ of time, and independent nor-mally distributed error terms�tp. The variation withinpersons,σ2

w, and the variation between persons,σ2B,

was modeled. The effectδt was included only at thetime points where the exposure was present (I(t) = 1f lues

ethacrolein and LOPs sessions (Table 2). The trendn the clean air session was not significantly smahan in the other sessions, and 4-min recovery waumed to be sufficient.

The same researcher critically viewedideo recordings, and each blink (averaging00 blinks/session) excluding those in the accli

ization stage, was stored along with the time oomputer. Only blinks with a downward movemenhe upper eyelid covering more than 50% of the coere counted. Twitches were ignored. A few incidef accidentally rubbing ones eyes and forced extelosure were counted as one blink observation. Eession was counted only once, since pilot stuhowed that repetitive counting differed by less t%. The BF was calculated in sequences of nverlapping 1 min averages. Blinking during exposo clean air for 36 min exhibited an alternating patt

or t= 12–31 min and zero otherwise), estimated va

248 J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251

represented changes in the mean value of the log trans-formed BF and were transformed to relative changes.This linear mixed model is comparable to a repeatedmeasurements ANOVA analysis, but more general, be-cause it includes the time trendγ. SAS version 8.02was used to model the BF.

2.6. Modeling of the chemical reaction

A simple one-compartment model was used to es-timate the extent of the gas-phase oxidation reactionsin terms of chemically reacted O3. The involved reac-tions and rate constants are shown elsewhere (Klenøand Wolkoff, 2004). VISSIM version 3.0 (Visual Solu-tions, Incorporated, MA) was applied for this purposeusing a Backward Euler (stiff) integration method witha 0.001 minimum step size. The O3 concentrations inthe reaction flow tube were measured at sample pointsI and II in Fig. 1 in the absence of other compounds,and used to calculate the first order rate constant for thewall loss according to Eq.(3).

k3 = − ln

(O3, sample point II

O3, sample point I

)t−1 (3)

The wall loss was modeled for LOPs by implemen-tation of the 1st order rate expression using the rateconstant above. The chemically reacted O3 (excludingwall loss) in percent was calculated according to Eq.(4):

Ot II− O3, wall loss

all loss

)× 100% (4)

3

air,mC hile3 ak”,h ilea , ther f thes

cedi anw ein( ions

Fig. 3. Human eye blink frequency changes during exposure tomethacrolein and limonene oxidation products (LOPs) at 20% RH(LOPs IV is 50% RH), respectively, for 20 min. Error bars indicatethe 95% confidence interval (Table 2). The solid and dotted regres-sion lines indicate no apparent dose-response relationship.

including a regression line are shown inFig. 3. Nodose-response relationship is apparent at concentra-tions below the proposed TLV. The lowest observedeffect level (LOEL) was 286 ppb methacrolein, whichincreased the BF by 18% (p= 0.001).

Forty to fifty percent of the subjects reported eye ir-ritation rated as “weak” to “less than moderate” duringexposure to LOPs I-III, and between “less than weak”

to “weak” during exposure to LOPs IV. Eye irritationwas experienced as a feeling of gravel in the eyes or astinging sensation. All LOPs exposures increased theBF significantly more than clean air (Table 2). No dose-response relationship was apparent for LOPs as shownin Fig. 3. LOPs III, which was prepared from the high-est reactant concentrations and had reacted the most,increased the BF significantly more than LOPs I andII (34% versus 17% and 12%). It is of note, that themean BF increases of LOPs I and II did not overlapthe confidence interval of LOPs III. This shows thatthe effects were separated by more than two standarddeviations. Similarly, an increase of the RH from 20%to 50% showed a reduction of BF (34% versus 22%),

3, reacted=(

O3, sample point I− O3, sample poin

O3, sample point I− O3, w

. Results

The changes in BF from exposure to cleanethacrolein and LOPs are shown inTable 2andFig. 3.lean air decreased the BF by 9% on average, w0% of the subjects rated irritation as “less than weowever typically transient within the session. Whperception generally involved an increase of BF

everse situation was not always true, and 40% oubjects did not report irritation in any session.

Forty to fifty percent of the subjects experientching, stinging or gritty eyes rated as “less theak” to “weak” during exposure to methacrol

Table 2). The methacrolein exposure concentrat

J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251 249

but not significantly. The LOEL of LOPs was 18 ppb ofchemically reacted O3 in a mixture of initially 92 ppblimonene and 101 ppb O3.

4. Discussion and conclusion

The purpose of this study was to measure weakeye irritation, during exposure to oxidation products.An exposure set-up developed for eye irritation assess-ments using increases of BF as a measure of trigemi-nal stimulation was applied. Eye irritation effects havepreviously been measured using this set-up for LOPs,isoprene oxidation products, and nitrate radicals (Klenøand Wolkoff, 2004). We now report changes in BF fromexposure to low concentrations of methacrolein at 20%RH corresponding to ca. 1/3, 2/3 and 3/3 of a recentlyproposed TLV of 300 ppb, which had been estimatedfrom a mouse bioassay (Larsen and Nielsen, 2000).Changes in BF by low concentrations of methacroleinwere subject to large variation, and a dose-responserelationship was not apparent at concentrations belowthe proposed TLV (Table 2andFig. 3). Concentrationsabove TLV are irrelevant for indoor environments andtherefore not examined. Forty to fifty percent of the sub-jects perceived irritation during the methacrolein expo-sures, and the subjective ratings were comparable in allthree exposure settings (Table 2). LOEL was 286 ppbmethacrolein, which increased the BF by 18% equal toan average increase from 14 to 16.5 min−1. Indeed, theL thep ion.H nty,s be-t r,1

20%R 50%R theB oretb to1 y3 arent( cor-r Fs vent ore

than LOPs I and II, the subjective ratings failed to showsuch a relationship. Fifty percent of the subjects re-ported irritation from LOPs III versus 40% in the otherLOPs sessions, but the rating was somewhat lower(Table 2).

The subjective irritation ratings of LOPs I andII were higher than all methacrolein exposures eventhough the BF increases were of the same order. Itis noteworthy, that the BF increased during expo-sure to both LOPs and methacrolein as the concen-tration/reaction extent increased, however not in a lin-ear manner. An increase of the reaction extent neitherinfluenced the subjective ratings of irritation, nor thenumber of subjects whom perceived irritation. An ex-planation could be that blinking increases prior to per-ceived irritation. In LOPs I, II, III and IV 32%, 65%,79% and 79% of the initial O3 was consumed in chem-ical reactions, respectively. The highest concentrationsof stable irritants could thus be found in LOPs III andIV, and vice versa. The amount of reacted O3 is a mea-sure of the reaction extent and thereby the concentra-tion of stable products, which build up during the re-action. The OH radical behaves differently; it is tooreactive to build up in the reaction mixture and de-pends on the concentration of residual O3 (Weschlerand Shields, 1996). The reactant concentrations werecarefully chosen in order to maintain similar OH con-centrations in LOPs mixtures I–IV. Similar residual O3concentrations ensured OH concentrations of 1× 106

molecules cm−3 estimated by the model. For this rea-s rvede l re-s estt ione

ceoP ix-t rkedaT blyb owne -t daryo tains ucts( alr er

OEL was expected to be somewhat higher thanroposed TLV, which should cause no eye irritatowever, the proposed TLV is subject to uncertaiince it is estimated from an empirical relationshipween RD50 and TLV values (Alarie, 1973; Schape993).

LOPs were prepared in three concentrations atH, and the highest concentration was prepared atH as well. Exposure to LOPs I and II increasedF by 17% and 12%, which was significantly m

han the clean air exposure (Table 2andFig. 3). Dou-ling the reaction extent from 71 ppb (LOPs II)55 ppb reacted O3 (LOPs III) increased the BF b4%. A dose-response relationship was not appFig. 3), and the regression analysis was poorlyelated (R2 = 0.7). LOPs II did not increase the Bignificantly more than LOPs I, and vice versa. Ehough LOPs III increased the BF significantly m

on OH is not considered responsible for the obseffect, because the significantly greater trigeminaponse of LOPs III relative to LOPs I and II sugghat irritants in the mixture build up with the reactxtent.

LOPs III differed from LOPs I and II by the presenf particles, i.e. secondary organic aerosols (Table 1).articles were expected to be formed in all LOPs m

ures, but the PE reaction flow tube apparently wos a reverse denuder for low particle counts (Table 1).he higher BF increase of LOPs III can presumae ascribed to gas-phase products, which are knye irritants (Klenø and Wolkoff, 2004), but an addiional effect of particles cannot be excluded. Seconrganic aerosols from ozonolysis of limonene coneveral higher molecular weight oxygenated prodGlasius et al., 2000), which may exert a trigeminesponse. A higher RH (LOPs IV) resulted in low

250 J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251

rating of subjective irritation and a smaller trigem-inal response. This was anticipated, because a lowRH increases water evaporation from the precornealtear film (PTF), which may exacerbate the susceptibil-ity towards irritants (Wolkoff et al., 2005). Althoughthe trigeminal responses at 20% and 50% RH werenot significantly different, a similar negative biolog-ical effect of low RH on irritation from LOPs hasbeen observed in a mouse bioassay (Wilkins et al.,2003).

Subjects were exposed to a LOPs mixture in a pre-vious study, which was comparable to LOPs II (Klenøand Wolkoff, 2004). The observed BF increase washigher (42% versus 12%) than in this study, and thedifference is likely to be explained by the modificationof the set-up, the inclusion of other human subjects,and day-to-day variation in the subjects.

The LOEL for LOPs corresponded to 18 ppb con-sumed O3 in a mixture prepared from ca. 92 ppblimonene and 101 ppb O3. This increased the BFby 17% during 20 min exposure comparable to thetrigeminal effect of 286 ppb methacrolein. The appliedlimonene- and O3 concentrations in LOPs I were higherthan typical indoor values, but close to high-end val-ues measured in various indoor settings (Wolkoff etal., 2000). Since the reaction time in the LOPs mix-ture was 10 min, the products can be formed at air ex-change rates as high as 6 h−1. The result indicates thatincidents of high air exchange rate and high limonene-and O3 concentrations may cause trigeminal stimula-t icalc ypi-c n inL ec ork-i heh ornei aya rl n ino tures ork-i

A

dy.

References

Alarie, Y., 1973. Sensory irritation by airborne chemicals. Crit. Rev.Toxicol. 2, 299–363.

Atkinson, R., Arey, J., 2003. Gas-phase troposheric chemistry ofbiogenic volatile organic compounds: a review. Atmos. Environ.37, S197–S219.

Brightman, H.S., Moss, N., 2000. Sick building syndrome stud-ies and the compilation of normative and comparative val-ues. In: Spengler, J.D., Samet, J.M., McCarthy, J.F. (Eds.), In-door Air Quality Handbook. McGraw-Hill, New York, pp. 3.1–3.32.

Brown, S.K., 1999. Assessment of pollutant emissions from dry-process photocopiers. Indoor Air 9, 259–267.

Craig, J.P., 2002. Structure and function of the preocular tear film.In: The Tear Film: Structure Function and Clinical Examination.Butterworth-Heinemann, Oxford, pp. 18–50.

Doughty, M.J., 2001. Consideration of three types of spontaneouseyeblink activity in normal humans: during reading and videodisplay terminal use, in primary gaze, and while in conversation.Optom. Vis. Sci. 78, 712–725.

Doughty, M.J., Fonn, D., Richter, D., Simpson, T., Caffery, B., Gor-don, K., 1997. A patient questionnaire approach to estimatingthe prevalence of dry eye symptoms in patients presenting tooptometric practices across Canada. Optom. Vis. Sci. 74, 424–431.

Falk-Filipsson, A., Lof, A., Hagberg, M., Hjelm, E.W., Wang, Z.,1993.d-Limonene exposure to humans by inhalation: uptake,distribution, elimination, and effects on the pulmonary function.J. Toxicol. Environ. Health 38, 77–88.

Glasius, M., Lahaniati, M., Calogirou, A., Di Bella, D., Jensen, N.R.,Hjorth, J., Kotzias, D., Larsen, B.R., 2000. Carboxylic acids insecondary aerosols from oxidation of cyclic monoterpenes byozone. Environ. Sci. Technol. 34, 1001–1010.

Hodgson, A.T., Rudd, A.F., Beal, D., Chandra, S., 2000. Volatile or-man-

K as aoxi-icals.

L n the

L gni,ionsage.

R p-lab-

rell,opo-WIT

S 99.reneiron.

ion and possibly eye irritation, indoors under atyponditions. Average limonene concentrations in tal indoor environments are 10–20 times lower thaOPs I (Wolkoff et al., 2000), this, however, may bounteracted by longer exposure time during a wng day (typically 8 h versus 0.33 h in this study). Tuman eye may become further susceptible to airb

rritants under typical working conditions, which mlter the PTF (Wolkoff et al., 2003, 2005). Whethe

ower limonene concentrations lead to eye irritatioccupational environments should be pursued in futudies of prolonged exposure under simulated wng conditions, including both genders.

cknowledgement

We thank all subjects for participating in this stu

ganic compound concentrations and emission rates in newufactured and site-built houses. Indoor Air 10, 178–192.

lenø, J., Wolkoff, P., 2004. Changes in eye blink frequencymeasure of trigeminal stimulation by exposure to limonenedation products, isoprene oxidation products, and nitrate radInt. Arch. Occup. Environ. Health 77, 235–243.

arsen, S.T., Nielsen, G.D., 2000. The effects of methacrolein orespiratory tract in mouse. Toxicol. Lett. 114, 197–202.

eovic, K.W., Sheldon, L.S., Whitaker, D.A., Hetes, R.G., CalcaJ.A., Baskir, J.N., 1996. Measurement of indoor air emissfrom dry-process photocopy machines. J. Air Waste ManAssoc. 46, 821–829.

uppert, L., Becker, K.H., Noziere, B., Spittler, M., 1999. Develoment of monoterpene oxidation mechanisms: results fromoratory and smog chamber studies. In: Borrell, P.M., BorP. (Eds.), Transport and Chemical Transformation in the Trsphere - Proceedings of the EUROTRAC-2 Symposium ’98.Press, Southhampton, UK, pp. 63–68.

auer, F., Schafer, C., Neeb, P., Horie, O., Moortgat, G.K., 19Formation of hydrogen peroxide in the ozonolysis of isopand simple alkenes under humid conditions. Atmos. Env33, 229–241.

J.K. Nøjgaard et al. / Toxicology Letters 156 (2005) 241–251 251

Schaper, M., 1993. Development of a database for sensory irritantsand its use in establishing occupational exposure limits. Am. Ind.Hyg. Assoc. J. 54, 488–544.

Schneider, T., Sundell, J., Bischof, W., Bohgard, M., Cherrie, J.W.,Clausen, P.A., Dreborg, S., Kildesø, J., Kjærgaard, S.K., Løvik,M., Pasanen, P., Skyberg, K., 2003. EUROPART. Airborne par-ticles in the indoor environment. A European interdisplinary re-view of scientific evidence on associations between exposure toparticles in buildings and health effects. Indoor Air 13, 38–48.

Sexton, K., Adgate, J.L., Ramachandran, G., Pratt, G.C., Mongin,S.J., Stock, T.H., Morandi, M.T., 2004. Comparison of personal,indoor, and outdoor exposures to hazardous air pollutants in threeurban communities. Environ. Sci. Technol. 38, 423–430.

Weschler, C.J., 2000. Ozone in indoor environments: concentrationsand chemistry. Indoor Air 10, 269–288.

Weschler, C.J., Shields, H.C., 1996. Production of hydroxyl radicalin indoor air. Environ. Sci. Technol. 30, 3250–3258.

Wilkins, C.K., Wolkoff, P., Clausen, P.A., Hammer, M., Nielsen,G.D., 2003. Upper airway irritation of terpene/ozone oxidation

products (TOPS). Dependence on reaction time, relative humid-ity and initial ozone concentration. Toxicol. Lett. 143, 109–114.

Wolkoff, P., 1998. Impact of air velocity, temperature, humidity, andair on long-term VOC emissions from building products. Atmos.Environ. 32, 2659–2668.

Wolkoff, P., Clausen, P.A., Wilkins, C.K., Nielsen, G.D., 2000. For-mation of strong airway irritants in terpene/ozone mixtures. In-door Air 10, 82–91.

Wolkoff, P., Klenø, J., Troiano, P., Piccoli, B., 2005. Eye complaintsin the office environment: precorneal tear film integrity influ-enced by eye blinking efficiency. Occup. Environ. Med. 62, 4–12.

Wolkoff, P., Nielsen, G.D., 2001. Organic compounds in indoor air –their relevance for perceived indoor air quality. Atmos. Environ.35, 4407–4417.

Wolkoff, P., Skov, P., Franck, C., Pedersen, L.N., 2003. Eye irritationand environmental factors in the office environment. Hypothe-ses, causes, and a physiological model. Scand. J. Work Environ.Health 29, 411–430.