Document MMY9VNoZOr3kQQwNjjQk2XN1x

Gnter for Enwonmend Toucolog & Technology Dcpamnent of Enwonmental Health,Colorado State Umvemty, Fort Cob, CO 80523 Varhb@kyinBiologicalExposure IndicesUdngPhysblogicallyBased PhatmacokheticModelingand Monte Carlo Simulation / By using physiologically based pharmacokinetic (PBPK) modeling coupled with Monte Carlo simulation,the interindividualvariability in the concentrationsof chemicals in a worker's exhaled breath and urine were estimated and compared with existing biological exposure indices (BEls).The PBPK model simulated an exposure regimen similar to a typical workday, while exposure concetltrations were set to equal the ambient threshold limit values (TLV"s) of six industrial solvents (benzene, chloroform,carbon tetrachloride, methylene chloride, methyl chloroform, and trichloroethylene).Based on model predictions incorporating interindividual Tvariability, the perc tage of population protected was derived usingTLVs as the basis for worker protectionb. esults showed that current BEls may not protect the majority or all of the --workers ir; an oc@pational setting.For instance, current end-expired air indices for benzene and methyl chloroform protect 95% and less than 10% of the worker population, respectively. Urinary metabolite concentrationsfor benzene, methyl chloroform, and trichloroethylene were also estimated.The current BE1 recommendationfor phenol melabolite concentrationat the end-of-shiftsampling interval was estimated to protect 6896 of the worker population,while trichloroacetic acid (TCAA) and trichloroethanol(TCOH) concentrationsfo: methyl chloroform exposure were estimated to protect 54% and 97%, respectively.The recommended concentration ofTCAA in urine as a determinant of trichloroethyleneexposure protects an estimated 84% of the workers. Although many of the existing BEls consideredappear to protect a majority of the worker population,an inconsistent proportion of the population is protected. The informationpresented in this study may provide a new approach for administrative decisions establishing BEls and allow uniform application of biological monitoring among different chemicals. re assessment of human exposure to mdustrial solvents is a difficult and complea prohlem fw mdusmal hygemsts who seek to malw e and conuol factors &tmg to worker health Tdtlonahy, mdustnal hygenm have re)led on au morutonng techtuques to estlmate hu man exposures, while the relative sd+ of the =-worker is evaluated through cornpansons with ref- erence standards However, the momtomg of borne concenuanons suffers fiom several shortc o m g s that have heen o u b e d and dscused premousl! 'J ~nmaril!, avbome concentrations of the w g h t 1996, American Industrial Hygiene Association H y p n i m (ACGIK) to cnamage an mtepted a.These standards define sue groups consisting of the liver, lung, slowly perhsrd, rapidly perfused, and fat, while each compartment is represented by mass hazardtohealthor balance differential equations that incorporate blood flows, parti- tion coefficients, m d tissue volumes. The concentration of the chemical of interest in the blood leaving the lung is assumed to be are collected fbm a in equilibrium with the concentration in alveolar air as determined worker 6 has k e n exposed to an airborne chemical at the thresh- by the b1ood:air partition coefficient. The chemical is distributed old limit value ( T L V T and, :opcdv. provide the same margrn of to all tissues and eliminated by metabolism in the liver, as well as satetv Js n v s . by exhalation. Excretion of metabolites is assumed to occur For the purpose of cstabhhmg BEIs, phmiolo@ly based phar- through a tirst-order process (KJ, and the h&onal metabolite macokmenc (PBPK) models have been emploved to relate &me formation (F,J is included for phenol, trichloroacetic acid, and concentrauons of chemcals to the levels of the parent compound or mchloroethanol. The mass balance equation b r the liver can de- ther metabolites ui a vanety of bodv duds or assues.*5' In phyio- scribe metabolism in terms of a single saturableprocess, a h t - o r - lqcally based pharmacoluneuc models, the body IS subdivlded lnto der process, or a combination of the two. Spedficayl, the saturable m a t o m d compartments represennng m d i w d d organs or tissue process has a maximum metabolic rate, V,, (mg/hr), and a groups The movement of chemcalsbetween the compartmentsIS de- Michaelis constant, &, (mg/Liter). The h - o r d e r constant, r<, has scribed bv mass balance ilifferenual equauons that mcorporate blood units of hr-'. -4description of the mm-balance differential equa- flows, pamuon coefficients, and m u e volumes. .%tier mcorporatmg tions used in the model is presented in de& elsewhere.tSi metabolic and potenual pharmacodvnarmcprocesses, the lite and dis- Physiological values used in the present model are scaled as a p i n o n ofthe c h e m d and metabolites can be predicted and e q - b c t i o n of body weight or cardiac o u t p ~ t . ( ' ~IJt s~h)ould be noted olated to a vanety ofexposure scenarios. Although physloloslcalmod- that many pharmacokinetic studies reb on interspecies (i.e., cross- els have been successtUv applied to the descnpuon of lndustd species) scaling of alveolar vendation and cardiac output as a h c - hermcah wtthm the human body the pnmary weaknessUI m g BEIs tion of body surface area (e.g., body weight raised to the two- denved tiom phvsiologdy based pharmacoluneuc modeling ts the thirds power). However, intraspecies (i.e., witbin species) scaling vanabdityw d m the human populauon;'6)speufically, human vanabd- haon are used in this model and calculated as a direct function of ity dirrrrmshes the relauonshp between absorbed dose and biological body weight:('*' mdicaton. In an attempt to address h s problem, Droz et al.78' created a "populanon-based" physiologd model in whch selected physlolog- rcal and exposureparameterswere assigned realisuc stamcal distnbu- nons. The model predimons of expued dv and unnarv metabolite concentawns were compared wtth the results of field studiestoevd- uate the reliabdity of bioiogcal mdicaton. However, the model uu- lized by Droz et 51. 'n' did not mcorporate manv ofrhe physolo%lcal md pharmacokmetlc pnnaples currently emplovrd ln PBPK models. Current PBPK models differ tiom dasslcal or convennonal pharma- cohencs 111 that f 1) they unlize a large body of phmologd and physiochemd data that x e not chemcal-speafic; 12) they &rd, wtth more confidence, '~tnspc~emsd interroute extrapdaaon; and ( 3) they may be used to predict a prim, tiom h t e d data, the phar- , m a c h e u c behawor of c e m chemcals. 9' %s study attempts to expand on the work of Droz et al."J' by crombmng current PBPK models wth stamucd simulauon techques ci.e., Monte Carlo sunutauon ). The linkage of PBPK models wxh Monte Carlo mulamn allows the assessment of lntenndiwdd vanabdity by conducnngpharmacokmetic studies computauody on a very large number of humans tin this case n = lo00 humans) wth "rymg p h y d o g d and I Flt Llllll I QF cm metabdic paramcrers. A detailed descnpnon of coup@ PBPKmod- eis wth Monte Carlo mulation is descnbed elsewhere."O' The obpaivcs a f b study were to (1)ases the vadnhyin bi- obgical exposure indices tbc bcnzcne, carbon temachh. k,* h,Inefhyid O i - o k J m l , mthyknc chhide, and tnchkrocthylm - V - b 8th I using PBPKmoQlingand hCad0jimuhtion;and (2)eoapprr the current B E I s W in the ACGIH handboohcli)whbtftc moM damddues. MATERIALS AND MRHoos .r hlqpbasic pfykologid modd struchur is ck5cliw by and illustrated in Figure 1. Thc body is subdivided into fk6- & = Q p c * BW (1) & . a .. ddu a l variability, or vui.lbility witbin thc same worker fiom * m & t w a s ~ d h ~ m <K w e*Ilatn-vm- lmiaityqrs,ttwrr- bR,nq&ixcd Bor s i q l i c q Unlscdmxwk- I per- ~ c r p r c s s c d i n ~ M ; u a i ? l s u r d M a h o d s a c c t i o n u e c o c f -' ficjliBtufafvuiorian. IFrfrbsEtlmHodelR#- Lbcnicdinput to the modd (C,) was set equal to the TLV summary ofestimates for the inmindhidual variability in physi'cal5mpar;lmctcrs is oudimd io Table I. Ahrtolv ventila- plltntionof thc PffocLtOd ch~mical.(~AJs~s)tated prewoudy, tion was mcasurcd as afunaionahlgr and activityin sublcco;din- reprcscnt the concentration of the chcmical or metabolite i d l y h e of pulmonary and cardiovascular While no sig- SeDd &om a worker who has been exposed toan airborne con- nificant differences for interindividual variabilitywere observed with Therefore, the duration of exposure was agc, variability during exercise was consistendy hgher (25 to 29%) an 8-hour workday per day for five days than that found at rest (10 to 19%).This suggests that the individ- was assumed to perform light-duty exer- ual variability normally associated with phpcal conditioning may cisc (50 W) during the 8-hr workday, followed bv 16 hours of rest not be manifested at basal condmonsto the extent found during ex- prior to the next-shift. The concentration 111 h e d expired air wcise. As a result, the variabihty associated with alveolar ventilation CC,) was calculated by assuming that t w o - h d s of the exhaled air in the physiological model during resting and light work con&tions a-asalveolar (C,j.'5i was assigned to be 20 and 30X, respectively. The basic model structure was moddied for Monte Carlo sim- Variabiliq of cardiac output in resting subjectswas similarto that dation by resampling model parameters within assigned distribu- reported for alveolar ventilation.'1G19V)ariability ranged from 15% tions. Consequently, central tendencies and associated dispersions in 15 males of unreported age("' to 22%in 11 females at 24 vean wae assigned to each model parameter. The central tendeecies of age.(16M) easurements during physical activity were reported by were obtained from L ~ u n g 'w~h'ile the dispersions were acquired Holmgren et al.[l9' for subjects r i b g a bicycle ergometer at 300 a variety of published studies. It was assumed that the in- kpm/min, but the interindividual variability in thosemeasurements was not sigdicantly different -Summary of Estimatesfor InterindividualYrMbilityinPhysiologicalFlow from that observed at rest. In contrast, studies by Mitchell et hmnetersUsedin MonteCarloSimulation m ?maleta No. dsubjccb kx I mdn 4J@ W) CocffiaiCntofV~ (%I ltcsting Exercise Reference no. al.(17)andChapman and F r a s e P revealed variability changes in cardlac output-al'iiox subjects during exercise. specifically, Mitchell et al.[l7'observed a 24% Flows variability in 15 males w b g on Alveolar A M 255 10 25' 15 a treadmill, while subjects .at rest 1 M 42.7 19 28' 15 possessed a variabilit\. of only A M 59.6 14 2 9 15 15%.Chapman and FraseP re- Cardiac output 23 M 25.2 15 25' 16 ported variabilities for male and 11 F 24.2 22 42' 16 female subjects during exercise of 15 MC 15 24' 17 25%and 42%,respectivel!-. Con- 12 M 303 18 - 18 sequently, variabfity in cardiac Liver 18 M, F 21.3 17 17' 10 M 24.8 28 -14 A I 28 19 output was assumed to increase 20 with exercise and estimated to be 21 equal to that of alveolar ventila- Rapidly perfused 9 9 M 50.1 22' - MA 17" - 22 tion. 23 The variability associated Slowly perfused 15 M 25.1 - 56,40,29 24 with liver blood flow was consid- 10 M, F 32.8 2 u - 25 erably greater than that reported 10 M, F 32.8 SOk - 25 for cardiac output during resting Fat 18 M 43.5 26 - 26 con&tions. Using the clearance 14 F 45.3 19 - 26 of radoactive xenon as an in&- ^Notstated in article 'Exercise consistedof stepping upon platform (20 cm high)and down again at a rate of 30 cycles/minute 'Exercise consistedof walking on treadmill at 3 mph on 5% grade DAgerange reponedfrom 20-50 yrs 'Exercise consistedof walking on a treadmili at maximumoxygen uptake 'Exercise consistedof ridinga bicycle ergomete: at 300 kpm/minute 'Cerebral blood flow *Kidney bloodflow cator, investigations by Lundbergh and Strandell,zo and Sherriff et al.'z' produced varabhty estimates of nearly 30%. However, the vanabihn-of liver blood 0ow is compounded by individual variability in carckac 'Quadriceps'bloodflow during bicycleergometer workout at 1628,40-51, and 6246% of maxima!oxygen uptake, respectively 'Cutaneous blood flow at ambienttemperature of 19-22 OC 'Subcutaneous bloodflow at ambienttemperature of 19-22 OC output. Therefwe, the uncer- tainty reported by L u n w and Strand~,izoa~s well as by S h e d et is a composite AlHA JOURNAL (57) January 1996 25 t of liver blood flow and cardiac output variabilities. Consequently, an estimate of the independent vanability of liver blood flow was derived using a standard method in the propagation of uncertainty represented by the tbllowing equation:tZ7' CV,,& = J C V L f CV-, (2) Summaiyof Estimatesfor Interindividual Variabilii inPhysiological Volume Parameters Usedin Monte Carlo Simulation PBPK Parameter No. of subjects sex Mean cocffidcnt Age ofVariation Reference m) 6) NO. where CV-*dc is the coefficient of variation for the tissue and cardiac output combined, wMe CY6 and CV- are the coefficients for cardiac output and tissue alone. In the current model, the total variability dssociated with liver blood flow and cardiac output was estimated to be 30%based on the xenon clearance studies, while the variability associated with cardiac output was estimated to be 20%.Based on Equation 2, a coeffiaent of variation of 22%was calculated for liver blood flow alone. The variabdity associated with slowlyp e k d and tit blood flows Weights: MY Liver Rapidly perfused 21,752 38 54 58 39 M,F M M M M 3 M,F AM 39 25 35 45 55 52 25 13 21 20 19 15 13' 21' 29 30 30 30 30 31 32 are compounded by the same problems associated with the liver blood flow. Estimates of variability associated with tit blood flow in men and women j h c h indude cardiac output variability) have been reported as 26%and 19%,respectlvdy.'26'Apptylngthe same methods of uncertainty propagation and asguming a c d a e n t of variation of 25%for the tit blood flow plus cardiac variability, the uncertainty associated with fit blood Bow alone was calculated to be 15%. For variability in slowly perfused blood flow, studies involving cutaneous and muscle blood flow measurements were used. A Fat 12 12 12 12 A Not stated in article 'Variability for lungweights CVariabiliifor heart weights 'Sedentaly menor women F M M F F 25 23.6 21.1 21.9 27.2 23' 31' 32' 16' 22' 32 33 33 33 33 Whereas variability in cutaneous and subcutaneous blood flow has '"Muscular' menor women enaaaed in comwtitive athletics been reported to be 20 and 50%,respe~tively,!~G~r)imby et al.Q4) found that variability in blood flow to the muscles that comprise the quadriceps decreased as a h a i o n of exercise intensity. VariabiriinModelMetabolic parameters ASpecifically,variabdity in blood flow to the quadriceps during bicy- cle e m m s workouts of progressive intensitiesdropped &om56 summary of the variability in metabolic and excretion parame- t t y is presented in Table 111.The variability in V- and K,,, was to 29%;") but the consistent decrease in individual variability can detdmmed primarily through in ~ Z T Ostudies by Peter et al.(x)and be explained through physiological blood flow autoregulation. As Re& et al.,(35b) ut the tbcus of the information on variability was the intensity of the exercise increases, skeletal muscles exhibit au- G t e d to the cytochrome P-4502E1 isozyme due to its role in corcgulatory characteristics in which blood flow is regulated meta- metabolizing the solvents in this study.(m1)Variability III V,, and bolically at the local level and, therefbre, removed &omchanges in K,, was reported to be 44% and 18%,respectively, using chlorzox- cardiac output.(28'Although the methods of uncertainty propaga- azone as a probe for cytochrome P-4502E1.(MI)n contrast, Reitz tion no longer apply during increasing activity, the individual vari- et al.'35r'eported considerably larger variations for V, (73%)and ability associated with cardiac output must be accounted for dur- ,K,,, (41%)using methylene chloride as a substrate. Reitz et al.'35)at- ing resting conditions. Therefore, assuming a variability of 50%in mbuted the large vanation in metabolic activity to a smgle liver the combined slowly perfused and cardiac output, the uncertainty sample with abnormally high activity. When the value for the spu- for slowly perfused blood flow alone was calculated to be 45%. rious liver sample was removed from the analysis, variabdity in V,, In contrast to blood Bow to the fat and skeletal muscles,blood and K, was 57%and la%,respectively. As expected, the interindi- flow to the majority of organs assigned to the richly perfused com- vidual vanability in V, is considerably greater than the variability partment is autoregulated even at basal levels.'28)Therefore, the associated with K,,,. T ~ IisSamibuted to the individual variability in variabdity estimates reported b r cerebral and kidney blood flows chemical intake, namely ethanol, which is known to induce cy- of 22 and 17%,'22J3) respectively, were not adjusted for the vari- tochrome P-4502E1.(41i)nduction of this isozyme would greatly ability in cardiac output. Variability in richly p e k d blood flow aftitct the maximum metabolicrate of the enzyme (V-) while hav- within the mode1 was estimated to be 20%. ing little effect on enzyme affinity (K)T.he in pitro variability in- V~isWvokrmcRraRcters f m a t i o n was supported by immunohistochemical studies by Wrighton and Stemns(u)WOI tbund a variability of 44% in im- As u m n r ~ yofestim;acstbr tk interk~dividdvariability in php- munodetmabk cytodrromc P-4502E1 among 14 human liver 'ologidvdume pvrmetas isummarized in Tabk LI. With thc samples. As a result, V- and & were assigncd model variabilities exception of body weight ami itd u n e , the inmindhidud vui- of5016 and 20%,mqxaively. atniity in tissuevohumswu Bma;lyr cooJistent at 20%.Aa a re- Thcvvhbilityinfirst-ordamtabolicictmify(~~estimated SUI& livcr and rapidiyperfipscdtissue vdumcs were estimated to &ansmdk by Hunter et 117'37) Dattaa*@@) &fW vary 2oKin d u d . Haarna,dtbc solvcnp&*mahyknt ben, The mriabikty io 6t tissue was reported m a study by Qmomatrdtopcrrscss a sigrtifir?nt &sc-wdrramju@onwtiv- WbmmlCJ! et alj- a d m t i qrid "rmwulu"M p md WmtKn bchacndke ages dB6 a n d 4 9 T~Be:caeS5aad- f%umCndmmcnwaea p p m h d y 3 0 ; i n d 2 0 5 ( , ~ . Thcreforc, model variabilityin fit tissue waghts was c o w gluta~stranstirasc(~)rtivity~mthykocchloridcIs arnuwd~bc30%. ssabatnst.H-,thebigb~af-btbcRcibt 36 A l U A I ~ I I I J N A I f % 7 ) Tin.,..,.. 1 0 0 ~ ; me excretion mer a 24-hour period ~ i 1 6c a l d a r d to bc 1694 rklicadLraretion while creatinine variability was reported at 32%.(%Al)though UD- SxrGIinty in Urhypanunaers lruynrywithdcgrocofhydntlm tad kvddmivity, m o d d v p r h r b i l i t g i n ~emmuion pad urinary vohrrne was cstinutrd to bc 3096 and 2oK, rcqamdy. Vuiability in d~ ht-wde Of urinary m b d i t ~(KJ~ was?IsolssuaKdmbc30K. h d e x a d o n as an indicatordbcnzcnc exposure was ad- justedfor background levels of p h d baaedon a study of40 urim 35 sampics by van Haaften and Sic.(@) Tbe variability in unoary phc- 34 d concentrations was reported to be 64% wth a mean of ap 35 proximately 6.6 mg/L of urine Solubility parameta werc ob- 35 tamed &omGargas et al.J45/and the standard deviations werc d - ,L,'-' 10 B B 15 21 IB 26 36 culated based on the standard errors and the number of samples 37 (Table IV)reported by Gargas.(%I 20 6B 25-45 7 ParameterOistriiutionrandConstraints 7 M 49 32 T38 he statistlcal distributions for model parameters were assumed to be either lognormal or normal"' and are summarized in 9 M 49 16' 38 Tables V and VI. With the exception of body weight and slowly _ _ _ _ _A\LrLa-Lr-w#:-uiiicy in IiietaDL -oI iic- parame*te_r_s_o_erIe_rrn~i n_e_d b_y.in vitro analysis .U^. IWL >.-+t-a*-At- :" ..*.A^ 111 at LKIC 'The highcoefficient of variation was skewed due to an outlying liver samde When the sample was removed,the coefficientsof variationfgr perfused blood flow, the &stributions of model parameters were truncated at 2 3 standard deviations to e h n a t e any outhers that would not remain within the bounds of physiologd consrrmts. It should be noted that fractlonal flow, metabohc, and volume were 57 and 18%, respectively ~ of variation was skewed due to a liver sample parameters were truncated at 2 3 standard deviauons or constraned between 0 and 1 to maintain mass balance Body weight was not truncated, and the blood flow to slowly perfused tlssues was truncated at 22 standard dekiations due to its high variabil- ity (45%)A. lthough metabolic pa meters, such as \'-, h pos- m.When the valuc for that sample was omitted. a ctkfficient don of only 8%was calcula;ed., Investigation another enzyme system, UDP-glucuronylmferase (UDP-GT), sessed relatively high variabilin; t"ey were truncated at 2 3 stan- dard deviations due to the mherent ability of the human p- op- ula- tion to induce e n z p,atic fUnCltion. Th xandau-d- chosen after Ionside:ring; the ranges srted in pons was literaitwe ed b\. Hunter et in which the conccntration of D- listed above. [# & m cacid wasmeasured fiom the inhibitory effect of gjucarolac- For model stability, as well as for practlcal reasons, alveolar ven- tam on ~-glucuromdaseA. vanabhty of 26%was observed in tilatlon was set equal to car&ac output dunng resung conditions, UDP-GT activin m h a group of 21 control and the sum of the lndimdual organ masse5 was c o n s m e d to be h d e n n g the wde range of uncertamty estimates, a conservatrve less than the total body mass. This was acheved by semng the vol- &e of 30%was used for K, model vanabhw. ume of slowl\ perfused m u e equal to the ddference between bodv #riabili in Model Exuetionand SolubilityParameters mass and r e m k g tissues.A conservation of blood t l o ~during resting condtions was accomplished by setting the flow to rapidly hriabihty estimates for creatinine excretion and urinary volume perf&ed tissues equal to the difference between cardac output and v w e r e obtained from a study of nine male lead workers whose re- flows to remaining compartments. During light work activity, the d function and serum creatinine concentrationswere considered flow to slowiy perfused tissues was set to equal the difference be- ~rarmaJ'~(T*)able 111). Be.twcen-subject variability in urinary vol- tween cardiac output and remaining flows. A total of 1000simula- tions wa.~performed for each solvent using the SimuSolP ..I t:. (Dow Chemical, Co.. Mid- !Ib id * UOdAr IhUAir FW w PcrfuscdAk slorrtr Perh&*Air land, Mich.) sofnvare package; the distribution algorNitahymlosr ewt earle.'4o8btained from r: Qentene 8.79(0.17) 17.0 (2.6) 499(24) 17.0 (2.6) 10.3 (1.8) Carbon .,._...-.r.,n.f.,nI r,m.I dene 2.73 (0.46) 6.85 (1.02) ,".&",3 G? in >&\ L.dd a 7n 2.1" iIn".n",7/i 14.2 (2.0) 359 (211 21.1 (2.96) 203 (9) "."" ,J-,Q M(,&3.al",Q\3'"6"a2 lldl >1 A I7.L I\L>.-Ar\, 1 7n I 1 7 \ 14.2 (2.Oi 21.1 (2.96) vR.vAvn ,1&3 .38) 14.2 (2.4) 27.2 (10.2) 4.57 (1.18) 13.9 (4.7) 3.1 (o.81) 7.92 (3.54) 10.1 (6.6) Data Treatment Statistical analysis of the biological concentrations was performed using standard methods based on normal theor). via the Minitab software program (Minitab, Inc., State College, Pa.). Normality of . . each data set was staristically assessed using Filliben-5 proce- AlHA JOURNAL (57) January 1996 27 1 .. dux.,*' A one-sided normal percenule was calculated according to urborne ihemcal at the TLV.'l'In turn, the following equation: weighted average concentrationk a a 10-hour workweek, to which a wwka BC,,,, = p - Z * o (3) posed, day d e r day,without ahrnseAQ tors ntervene (e.%.,dermal a w x where BC is the biological concentration at the lower normal per- exposure, or personal habits of the WO&US); centile (Le., for the lower 5th percentile, 95% of all values would margm of d k t v as TLVs. Consequently.TL.Vs were employed as be higher), p is the estimated mean, Q is the estimated standard de- the "gold-standard" and used as the mput concentration for the viation, and 2 is the standard normal value corresponding to the PBPK model (see Matenals and Methods). The theoretlcal basis, cvmulative probability and desired percende (e.g., 1.645 for the therefore, is based on the protecnon fiorded by the TLV. lower 5th percende). in the event that a transformation was needed to achieve normality,the BC values were computed using SummaryStatistics Tthe manstormed variable, and Equation 3 was followed by the ap- propriate back-transformation. he results of the model predictions hr expired air concentrations are described in Tables VI1 and VIII. of the solvents studied, all six had sigNficant concentrationsin the expired air at the end-of- shift sampling, which suggests that expired air concentrations can RESULTS TheomthlBasisFor 0eriVing"PmtectedPopulatiolu" be a suitable biologcal index if the typr of expired air (end or mixed) is specified. Comparisons with current expired air recommendations were BEIs represent the levels of biological determinants likely to be possible for benzene and methyi chloroform at the prior-to-next observed in specimens collected fiom a worker exposed to an shift and prior-to-last shift sampling times. The current BE1 for benzene in end-exhaled air is listed as 0.12 ppm(") and was ModelParametersUsed in PBPwMonteCarloSimulationto Determine Vahbilhy in BiologicalExposure Indices anmated by the model to pro- tect 95% of the worker popula- tion (Table MI).In contrast, PWK Parameter Weights Body (kg) L - 70.0 13 the current BE1 of methvl chloroform in end-exhaled ax (40p p m P ' was much lower than the corresponding model predictions and protects, in Eact, less than 10% of the PercentagedW& wo&r population (TableVII). Liver Rapidly perfused Slowly perfused Fat Flows Alveolar (Vhr) Cardiac output (Uhr) N N - N L L B 0.03 20 B 0.06 1c B C- 6 0.23 30 S 4.97 I 20 S D- Model variability in urinary metabolite concentrations IS summanzed in Table IX. The current BE1 recommendanon tbr phenol metabolite concentraaon with benzene exposure at the end-of-sM samphg time was 50 mg/g creati- PercentageofCardiiOutput nine. ") Based on model predictions, the cvrrent BE1 rec- Liver Rapidly perfused N C 0.25 22 N C E- ommendation for phenol in urine protects an estlmated slowly perfused N C 0.19 45 68% ofthe worker population. Fat N C 0.05 15 Methyl chloroform and Metabolic aichloroahyknedctcrminants V- (rnghr) L were estimated &om the 16 (mg/L) K, (hr-') L L mchloroacetic aad (TCAA) and- d (TCOH) F, L bcretion K. (hr-') -L I 30 Creatinine (mgR) N - 135 30 Urinaryvd. (VShrs) t - 0 5 2 20 coilcentl;Ltioils. Modei prcdktiwsofTCMtndTCQHbr 28 AlHA JOURNAL (57) January 1996 111 Mad?lPu;lanQwStt#dinph~l while TCOH in ltriDt b a bet- 195.7 -- - -035 0.2 70 f& cnd-exh;rled & ir 4- 0.25 &uaEdtiopm&eciartsra16y --0.45 - - of the worker population, 5.75 Ox126,O.W 75,25' while mcabolitc rrc~mmar- 0.58 053c - - dations fbr urinary TCAA and -0.25 0.026,0.0069 45,32e TCOH uc to pro- t m t 54% md 9?%, XSPCC- anal and tively. Signifiwt dispuity, thercforc, exists among the bi- chloridehas beendemonstratedto possess significantfim-order ological indicators relative to the proportion of population protected. Furchmnorc, re- %w (Table IX) Finally, the recommended concentration of TCAA in sults from model cstimatcs of exhaled breath and TCAA suggest that workers are not being ad- ~ .f 2. urine as a determinant oftrichiorocthylene exposure is 100 mg/g quatcly protected with respect to c m n t indicators of inte- . 3 , creatinine,"" whch protects an estimated 84% of the workers grated exposure. Biolotzical monitoring of trichloroethvlene exposure is also performLd using the presence of TCAA.and TCOH metabo- lites in urine, However, current ACGIH recommendations in- DISCUSSION clude a combined TCAA and TCOH concentration of 300 lthough mathematical models with biologically interpretablepa- Variabilii in PBPKModel Prediicted End-ExhaledAir Concentrationsas Compared with Current BiologicalExposure Indices" mjoxity of the worker population despite the proteaion"aforded through the TLV. According to ACGIH, phenol concentrations in the urine are awnid only to be applied to a group of workers, vther than to an in&- Benzene vidual worker, due to the large variabiiity in background urine con- End-of-shift centration~.A(s~~a result, a direct companson of model-derived Prior-to-next-shift results and the recommended levels can be deceiving. Instead,the Prior-to-last-shift it recommended concentration of 50 mg/g creatinine should, theo- Carbon tetrachloride retically, compare with the model estimate protecting 50% of the End-of-shift worker population. According to model estimates the recom; mended phenol concentration protects an estimated 68%of the Prior-tonext-shift Prior-to-last-shift population and, when applied to a group ofindividuals according Chloroform to ACGIH r e c e t ~ o n s s,hould provide adequate protec- End-of-shift tion. Prior-to-next-shift As a supplement to urinary phenol, the monitoring of exhaled Prior-to-last-shift air is recommended for workers exposed to benzene.'50'The cur- Methyl chloroform r a t BE1 recommendationsof 0.08 and 0.12 ppm for mixed- and End-of-shift end-exhakdair collected prior to next shift,(11)respectively, protect Prior-to-next-shift an estimated 9596 of the worker population. However, the in- Prior-to-last-shift tuindmdual variability in expired air concentrations is com- Methylenechloride pounded by differences in physical acnvity during and after expo- Endof-shift sure. Therefore, reducing the potennal for varianons in physical ac- Prior-to-next-shift h t y would result in a decrease in the total uncertainq Smce Prior-to-last-shift concentranons of benzene IIY exhaled air rise sharply at the beLgm- Trichloroethylene nux of the exmsure and reach steady state within 1to 2 hour~,'~O End-of-shift cd k- at the end-of-shift Prior-to-next-shift - 5.7 6 3 7.0 0.12 0.16 022 0.30 0.38 0.49 4.0 4 3 4.4 0.02 0.03 0.05 0.06 0.10 0.14 4.9 5.6 6.4 0.11 0.16 0.2 1 0.18 0.24 0.31 31 1 2.4 7.3 318 33 9.4 326 4.7 12.2 27 0.37 0.36 31 0.76 0.84 34 1.2 1.3 29 0.56 32 0.78 35 1.1 Prior-to-last-shift h c i x i d to momtor 1.5 1.9 2.5 chloroform, TCAA, ed air priorred as indicators A End-of-shift= shortly beforeend of an 8-hr exposure. Prior-to-next-shift = beforethe second shift of fiveday workweek after 8-hr exposureand 16-hr rest. Prior-to-last-shift= beforefifth shift of fiveday workweekafter four consecutive8-hr exposures and 16-hr rests ingday and prewous days AlHA JOURNAL (57) January 1996 29 Variabiiii in PBPKModelPredicted Mixed-ExhaledAir Concentrationsas Compared with Current BiologicalExposure Indices* PredictedMixed-Exhakd Air Concentrations(ppm) Benzene End-of-shift Prior-to-next-shift Prior-to-last-shift Carbon tetrachloride End-of-shift Prior-to-next-shift Prior-to-last-shift Chloroform End-of-shift Prior-to-next-shift Prior-to-last-shift Methyl chloroform End-of-shift Prior-to-next-shift Prior-to-last-shift Methylene chloride End-of-shift Prior-to-next-shift Prior-to-last-shift Trichloroethylene End-of-shift P-rior-to_--n-ext-shift Prior-to-last-shift 7.1 7.5 8.0 0.08 0.11 0.15 0.20 0.25 0.33 4.4 0.02 0.04 4.5 0.02 0.07 4.6 0.03 0.09 6.6 7.1 7.6 0.08 0.10 0.14 0.12 0.16 0.21 324 1.6 4.9 329 2.2 6.3 334 3.2 8.1 35 0.25 0.24 37 0.51 0.56 40 0.78 0.89 36 0.37 1 .o 38 0.52 1.3 40 0.73 1.7 *End-of-shift= shortly before end of an 8-hr~<posureP.rior-to-next-shift = before the second shiftof five-day work-#eela. fter 8-hrexposureand l d h r r e s t Prior-to-last-shift= before fifth A f . of five-day workweek after four consecutive8-hr exposuresand 16-hr rests. mg/g creatinine, as weil as TCAA Concentration alone (100 mg/g creatinine).'"' .Uthough the current TCAA recommendation is estimated to protect 84% of the worker population, the combined TCAA and TCOH recommendation would protect greater than 95% of the worker population, since the TCOH concentration alone is greater than 400 mg/g creati- nine at 9596 (Table \TI). By combining the tools currently affiirded the toxicology and pharmacokinetic tields (Le., PBPK modeling and Monte Carlo simulation), the establishment of standards designed for worker protection can be carried out with less uncertainty. Spedcally, the addition of realistic variability information to established PBPK models may provide a consistent basis for administrative deasions concerning BEIs. While many existing BEIs addresscd in thisstudy appear to protect a majority of the worker population, an inconsistent proportion ofthe population is protected. Therefore, reevalu- aaon of cvrrent indicators according to "population-based" BEIs m y be necessary to allow unitbnn ami reiiabte appbcatirm dbio- logical monitoring among difErmt chemicals. Ofthe solvents presented in this study, only carbon tetrachlo- d e is bdicvcdto have tbc pacrrci?r b r ~ drm idkited by thc skin notrrtioaio &e ACGM hulclbooLfll) Although no attempt was made to consider dermal expowre in this model, previous efforts by McDougal et al.'519h)ave incOrPorated "&&-ginnt model that needs to be addressed. Recent a&ces in molkular md biochemical toxicology have permitted an understandmg of the toxic mechanisms underlying many of the common industnd chemicals. From a predictive standpoint, mechanistic information can be incorporated into what are presently called pharmacodynamic models. In contrast to pharmacohetic models that describe what the body does to the chemical, pharmacodynamic models characterize what the chemical does to the body. The real potential Lies in the ability to link the two types of models in order to quantitate the target tissue dose ofthe reactive chemical or its metabolite and describe the associated biologlcal effects. Together with Monte Carlo simulation, physioioIpcally based pharmacohetic and pharmacodynamic modeling (PBPK/PD) can provide a mechanistic basis &om whch to establish not only BEIs, but dso TLVs. A PBPK/PD model tbr the hepatotoldcity of carbon tetrachloride, alone and following kepone pretreatment, has been developed and evaluated by El-Masri et al."O' Further research associated with PBPKJPD modeling may also address issues associated with mixed chemical exposures. According to Fiserova-Bergerova;I' correlation between intensity ofrrposure and biological levels may be disrupted by interferences of another chemical or even its own metabolite^.'^^) These disruptions can occur on the pharmacokinetic'"' or pharmacodynamic"o) level and may result in interactions appreciably different than the additive adjustment currently recommended by ACGIH.'l1' For exyle, kepone, an organochlorine pesticide used tbr fire ant con- trol as been found in the environment.'551 environmentaByre- 3c levels (e.g., 10 ppm in the diet), kepone can produce as high 67-fold increase in lethality compared to an otherwise margindv toxic dose of carbon teuachloride."*~*' Therefore, the concept of chemical mixtures must be addressed in exposuresboth inside and outside the workplace with PBPK/PD models incorporating toxicologic interactions on the mechanistic level. ---Vanability inPBPKModelPtedictedUrinary MetaWie Concentrationsas Comparedwith Current Biological ExposureIndicef -(w9-) t~~ '30 AlHA JOURNAL (57) January 1996 y-18. J i.ndmEm.. QAw~rrJJd. Smp.rPLajrmS.:27p55o-7a2u(1n9l37~).~ u n.d i . c t 19. 1.i n a - r d r in na- dalbjmst~ad~aadrcmFamarbcmporitaa,~cpc- d* 1D & Q&vClkQe xcliabcmvlnxkinacnritics. Arnr pcri.h a d . 1P.343-363(1960). b d q p d crporurr lndiccs H.w.2 ~ ~ 8 (1~9847).2 21. SbeaitT, S.B., &C. sI..rq d L T+oz clioial studyoflivcr blood Boar in man measud by XC-133devanccafnrpornl veio inpion. Gut 161027-1031 (1977). 22.Munck, 0. and NA. Lnaax BiLrcnl cerebral Mood flow and oxygen consumption in man by uoc of&-85. ckr Rrr. 5163-168 (1957). 23. Smith, H.W.: Ibc K a h y Srrvnwrc and Function in H& a d Dircnrr. NcwYort: oxford Univarity Prrss, 1951. 24. rimb by, G.,E. md B. siltin: ~ocaxlenon 133 d m c c - 25 Scirscn, J.: Blood 5ou 111 cutaneous ~~SSUC111 man stu&cd by washout of Hmmn, E.R and J& Petason: Eduarion. in F u ~ ~ t u oLf s radioacavc xenon. CIYCR a 25.215-229 (1969). Marm, D. O h t o , and E Bragnone: ``Dyn;UNc" bi- indcxcshn-hcxanc and 2,5-herulectone,sugecst&iby bzsed pharmacokmcnc model Am Id HB h c J. KW; use OfphVridogicaUy and pcrfuslon in man b\. uptake of Kr-85. J. AppI. -1.23(5):621-630 (1967). 27, saciaa, gH.: Manrung uld +of dB of cdiabonwc tcsls in statrmcalManvalofr6pArrrrclatrm DC:AFswanon ofO&cul AFllytial Chcmis~1, 975.pp. 6688. M C C ~ UOUfW~yFvnmoA 6th cd.NCWYO^: McGnw-Hill, 1994. clderfy Am. J Clin. Nwr. 40S08-819 (1984). in wught ofthc adult h- SplCen. protaction: Repmtofabr w,Aadascn.Phannabbncncs In lnrroductwnto Hodgson and P.E Lcvi,cds. 2nd cd. N~rwlU, T d &up 33. wm&, a R@rcnwA4#n, (No 23).New Yo&. Pcrgamon Press, 1975 J., J.v.GA hunin,R and M.Mahaffy:Influence of muxulv development, obesity, and agc on thc fit-fkcc mass of adults J &I Pbplol 41(2):223-229 (1976) p i m r m a d m e a c / p h a r m a ~ c modding of 34.Pcm, R, R B o c k , P.H. Bcamc, M.Iwasaki, ct al.: Hydroqlanon mirmrcS and paaSiblc appkanons in risk asscssmcnt. T d of chlonozazone as a spetific probc for her cytochrome P-450IIE1 Cbcm. Res Tavrcol 3566573 (1990) H y g k h 6 35.Rcit~,RH.,AL.Mendnh,and F.P. G v I n n m m~coLbojiun~of a d pl?rJrcJ methyknc chlonde m humao and arumal tlssucs use 111 physmlopUyr based phannacokmcuc models T i .qPPl AiarmaMI 972302% (1989) ~ 38. ArrLi, S.: Effects of unnarv vdumc on urinary collfcntrationsof lead, 6 and vnwg anunol-c aad, coproporphyrin, axatinme, and total solutcs Bnt. J. hdstuit' lnd Mcd 3750-54 11980) ethanol ,acetone ,and bcnzcne-mduciblc qochromc P-450 IIEl m rar F.P., D.H. Kim, and M. 1waS;lLi: ROICof human a - 50 IlEl in the oudanon of man1 low molecularweight can Rtr. T d 4.168-179 (1991 1. .C. Krma, and J.M. LzsL;a:Rioamvanon of halogenated b?-. c y ~ o ~ h Pr4o50~2E~1~ Cnt RCP T ~ S C O2~.31I).1-20 AlHA JOURNAL (57) January 1996 31 43 Andersen, ME., Phvsiolopcdlv base methvlene chlonde. W van Haafken, AB. s;w Lhromatographv-IS a 1 3652-58 I 1965). 45 Gargas, M.L., RJ,.-P D S ybird, G.h. Cmm, et al.: Pmuon ioeificients of \owm d r m h r - e vdaik b c a l s m vmous liquds md nssues. Gmcd. =Ippl. R b m u d M87-99 1989). 46 Gargas, M.L.: 'Vmabdkv m punnon coefhents of selected organrc Lompounds." October 21, 1994 (Pmatc cwunurucanon] Or. .Michael Gargas, .WcLaren/Xan,TkmR&.. Counluld Eut Budding, 29225 Chagnn Blvd., Cleveland, OH 441224629 47 Portia,C.J. and N.L. Rjptm:Vuntnlity of& sloac anmateswhen ustng complicated models of d~ c a m m g a x pes. Fwnd. &pi. Tmcol. 13533-544 11989). 18 Naylor, T., J. Balinty, D. &mfkf, aad K cba: Compurn Ssmulanon ,.Gchnrques. New York. JohnU 7 i & Sam. 1968. 19 Fiuiben, probaw plot mfiant test for normal- ity. Technomemcs l'(1 )-I11-117 1975). 50. American Conference of Governmend Industrid Hygienists (ACGIH): Documentation of thc & l o g ~ A exposure indices. In ~ n t n * r t M nof abc Thnabold Limit Valvcs and Biolo@al E.rparun i k ,wi. 3,6rh ed.Cinannati, O H ACGEI. 1991. 51.Md)Ol.yd J.N.,G.V. Jepson, H.J.ClcwcIl, M.G. ~M;rCNa@ton, et d.: A physiological pharmacokinetic modcl hr dermal absorption of vaporr in ttU rat. T A L Appl. Pkrwucd. 85:286294 (19861. 52. MCDOU& J.N., G.W. Jeps~n,H.J. C k ~ d l ,M.L. Cuga~,et 11.: Dermal absorption of mganic chemical vapors in rats and humans. Fund. ApPr. Ti&col. 14299-308 1990). 53. Andersen, M.E. and H.J.Clcwell: `Pharmachetic interaczions of mixtures." Paper presented at 14th Conference on Environmental Toxicology. (AFAMIU-TR-83-499). h Force Aerospace LMedical Research Laboratory, Wright-PattersonAir Force Base, OH. 1984. 54. Mum-, M.M., I.G. Sipcs, H.J.Clewdl, ad RS.H. Y q Rigt assessment of chemical mixtures: biologc and toxicologic issues. Fund. Appl. T&l.21:258-269 (1993). .55. Curtis, L.R. W.L. W i i ad H.M.Mehad& PotcnMon ofthe hepatotoxicity of carbon temcbloride bllowing preexposure to chlordccone (kepone) tn the male rat. T'ol. Appi. Phamrpcol. 5k.283-293 ! 1979). 56. Mehendaie, H.M.: Mecharusm of the tnteramve unplificaaon of halomethane hepatotoxicq and lethality by other chemcais. In Tarercolom o f Cbcrnacal Mrrhrnsr Carc Stwdrcs. Mechanmnr. and Nmcl A~~woL&x'RS.HY.ang,ed. San Diego, CA: Academic Press, 1994. pp. 299-334. -._- ?I' 32 AlHA JOURNAL (57) 1996