Document kDw2mwbnz2BGz0Ey0OrozZXDV

f.i Vol 16 pp 6)1 lo 6)1. I6tl Printed m Gtoi Sntam All riftm reserved OOIV4264 XI CoW..|M O WM |ORJDZ.Uft# ^ 1 THE IMPACT OF AIR-LEAD ON BLOOD-LEAD IN MANA CRITIQUE OF THE RECENT LITERATURE* P. B. Hammond. E. J. O'Flaherty and P. S. Gartsde Department of Environmental Health. College of Medicine. Unirerat v of Cincinnati. }22S Eden Avenue. Cincinnati, OH 45267. USA (Received 26 June 1180) PLAINTIFF'S EXHIBIT DOW-1430 Summary--The nature of the relationship between air-lead and blood-lead in man was evaluated using all the available information from published studies in which subjects were in an approximately steady state with regard to lead exposure, and blood-lead could be related to air-lead measurements made in the individual's breathing zone. The observed blood-lead-air-lead relationship was compared with the relationships predicted by applying the two kinetic models which have been proposed to describe lead disposition in man. The models were found to be inconsistent with observed lead disposition kinetics. Each model generates a linear blood-lead-air-lcad relationship with predicted blood-lead levels well - above the actual levels except within the very lowest blood-lead range (<5pg/nr>|. The observed blood-lead-air-lead relationship is curvilinear with slope decreasing as blood- and air-lead increase. The implications of this curvilincanty with regard to the dependence of x. the increment in blood-lead associated with a given increment in air-lead, both on the magnitude of the air-lead increment and on the baseline blood-lead value are illustrated and discussed. Introduction A document published in 1972 (NAS-NRC. 1972) was intended to review and evaluate all the literature bearing on the question of the significance of airborne lead for human health and welfare. The committee that prepared this document was unable to arrive at a firm conclusion concerning the contribution of lead in air to the total body burden. It identified two general exposure routes, the inhalation of lead aerosols and the hand-to-mouth transfer of lead-bearing street dust a potential problem among young children. More recently, a World Health Organization task group (WHO. 1977) and the US Environmental Pro tection Agency (EPA. 1977) have been confronted with the same issues. All three of these study groups considered the possibility of estimating the impor tance of air-lead by calculating directly transfer from air (or street dust) into the body. This approach was found by the NAS-NRC study group to be impracti cal because of the uncertainties surrounding airway deposition and clearance rates for the lead aerosols of interest The importance of hand-to-mouth transfer of fallout lead in street dust was even more difficult to assess because, in addition to uncertainties regarding the gastrointestinal absorption of lead in children, there was absolutely no information concerning how much street dust a child might swallow. This particu lar problem, therefore, was simply cited as being one which required attention. With regard to the impact of the inhalation of lead aerosols, the NAS-NRC committee concluded from the sparse epidemiological data available at the time 'Presented in part at the International Conference on Management and Control of Heavy Metals in the En vironment. London. IS--21 September 1979. that there was no perceptible impact of the concen tration of lead in air (PbA) on the concentration of lead in blood (PbB) below a PbA level of 2-3 jig/m1. The committee recommended that more precise studies be conducted concerning the relationship between ambient-air-lead concentration and bloodlead concentration, perhaps by use of personal moni tors. This specific recommendation was implemented in one study (Azar, Snee & Habibi. 1973a). Between the time of publication of the NAS-NRC document and the preparation of the WHO docu ment, which actually took place in 1975. some new information became available concerning the dispo sition of lead, both inhaled and swallowed. Still more information was available to the US EPA in the prep aration of its 1977 document Yet, in both documents summary judgement as to the impact of inhaled lead on the body burden relied mainly on epidemiologic information concerning the relationship between PbB and PbA. The impact of lead in street dust on the body burden of lead in children was judged in both documents still to be surrounded by great uncer tainly. The present critique will take a fresh look at the impact of air-lead on blood-lead in man. considering all the useful information currently available. The problem still is basically divisible into two. the inhala tion of lead aerosols and the hand-to-mouth transfer of fallout lead in the environment. Consideration will be given only to the first problem. There are two approaches to estimating the impact of inhaled lead aerosols on the body burden of lead. The first is to use measured rates of uptake (or loss) of lead from an experimental study together with a dis position model to project either total body burden or the amount of lead in any of the presumed kinetic compartments at any time. This approach must take * 631 c^YOOl RowVerK O'JM 63: P. B Hammond. E. J. O'Flaherty and P. S. Gaxtside into consideration uptake from all sources by all absorption routes since the concern is with the rela tive importance of inhaled lead to the total body bur den. Two different disposition models have been used to describe lead kinetics. Rabinowitz. Wetherill & Kopple (1976) applied a three-compartment model to human exposure data from their studies of the kin etics of stable labelled lead. Bernard (1977) proposed a five-compartment model to describe human lead kinetics. Both models are shown schematically in Fig. 1. The second approach to estimating the impact of inhaled lead aerosols on the body burden of lead is purely empirical: the observed relationship between PbA and PbB is evaluated mathematically. For this approach to be valid certain criteria must be satisfied. First, steady- state or near steady state conditions must prevail, wherein the amount of lead inhaled per unit time must have been reasonably constant over a period of time long enough for virtual equilibration to occur between PbA and PbB. The input of lead from other sources must also be approximately constant, since it is not negligible relative to the input of lead from air. In this paper we will evaluate the nature of the PbB-PbA relationship within the range of normal ambient-air lead concentrations as observed by Azar et al. (1973a). PbB values predicted by this relation ship both within and beyond the range of PbA levels on which it is based will be compared with individual data points from experimental studies and with those predicted by application of the Rabinowitz and Ber nard models. All the data used apply only to adults. DIET, Al* - * pg /aoy JfttftC BlLC.MAift, SWC*T, NA<L5.*fC Observations: graded exposure levels Azar et al. (1973a) monitored the PbA exposure of ISO adults from five locations in the United States. continuously and individually for 2-4 wk. The subjects were in a steady stale condition with regard to exposure: the average number of yean spent on the current job varied from 16-19-1 at the five different locations. Air-lead exposure, calculated as a timeweighted average, was monitored using personal air samplers and ranged up to about 4 pg total PbA/m1 in the group with the lowest exposures and up to about 9 pg total PbA/m1 in the group with the highest exposures. The concentration of respirable lead was somewhat less. Statistical analysis of these data could be performed using regression models of varying degrees of com plexity. For this calculation it was decided to use the simplest model available: that is. a straight-line fit with PbB depending upon PbA. and also to include a component representing differences among the five geographic locations. Further, the values of PbA and PbB could first be transformed. e.g. by taking their logarithms, or they could remain untransformed. However, any particular choices in this respect might well be criticized as being somewhat arbitrary. In this analysis, therefore, it was decided to use an empirical approach to the problem of transforming PbA and PbB (Box Sl Cox. 1964) to obtain the optimal trans formations. This requires that a variable of interest. Y. be transformed by the expression {/ - I)/). where X is chosen so that the assumptions of normality and homogeneity of variance in regression analysis are best satisfied. In order to avoid unrealistic predicted values for PbB when PbA0. the component Q, where i " 1........3. representing each of the five geographic locations, was incorporated into the model as follows: PbB - A + B(PbA + Q. C, represents the non-air lead contribution to PbB. expressed in terms of equivalent PbA. The Azar et al. data were analysed by applying the above transformation both to PbB and PbA. and the optimal parameters in the model were found to be (PbB)','8,* - -009786 + Ol7904(PbA + C,r ,0* where C, - 6-55627, C, - 1-61865. Cj - 038578. C4 = 5-98683. and C, - 2-76079 for the five lo cations. with R2 -- 049. the proportion of variation explained by (he model. The average relationship between PbB and PbA was found by substituting the average Q value. 3-46186. and solving for PbB: * r f ' i i 4 I * I 1 PbB - [-009786 + 017904<PbA + 346186)"0'10*]-1,1 019 (I) Fig. I. Models proposed to describe lead disposition kin etics by tai Rabinowitz et al. (1976) and (b) Bernard (1977V The amounts of lead in each compartmenu exchanging daily between compartments, and excreted daily are shown for steady state conditions of continuous exposure to food, water, and air lead. The amounts of lead shown entering i he central compartment in each case are the amounts that are assumed to be absorbed daily into the systemic circu lation from all sources. This result can be interpreted as the reciprocal transform for PbB and almost the logarithmic trans form for PbA. The PbB-PbA relationship, shown in Fig. 2. is curvilinear with slope decreasing as PbA and PbB increase. Observatioai: dbertw exposure leveti The criteria for including data points in this analy sis were: first, that the individuals studied were in an 921001 RowVerX 09b.) exposure of 1 ^ States. I The subtn regard to pent on the different * timeP-.^onal air 't*f PbA/mJ 'nd up to the highI able lead Performed ' of com' o use the "ght-line fit to include a the five PbA and taxing their ransformed. : ect might 7 In this t~ empirical 8 PbA and ""nal trarulterest, Y. vhere i. is rmality and analysis are alues for C,, where : geographic s follows: I ion to PbB. ilying the bn. and the md to be j-0 104 , 4 0-38578. he five lo ot variation : ationship s uung the PbB: - ,0,4 (1) e reciprocal t uc transI shown in as PbA and n this analy1 were in an Impact of air-lead on blood-lead in man 633 approximately steady state of lead exposure; and second, that PbB could be related to PbA measure ments made in the individual's breathing zone. Two studies were considered to meet these criteria ad equately. Both involved the exposure of human volunteers ta known concentrations of artificially generated lead sesquioxide. Kehoe (summarized by Gross. 1979) exposed indi viduals singly, and often the same individuals sequen tially. at many different air-lead concentrations for periods of time ranging from 98 to 772 days. Griffin. Coulsion. Wills. Russell & Knelson (1975) exposed groups of volunteers for 14-19 wk at only two air-lead concentrations. Hence, standard deviations were cal culated for the Griffin et at. data. It is assumed that in both of the above studies the PbAs monitored by the investigators were the concentrations actually in the breathing zone of the subjects. Hand to mouth transfer of settled lead-bearing dust probably was minimal. Kehoe subjects dusted the chambers fre quently (Kehoe subject JOS, personal communication 19801 A similar housekeeping program prevailed in the Griffin study (T. B. Griffin, personal communi cation. 1980). fn order to minimize the effect on these data of non-air sources of lead, all the experimental data points were normalized to the same control value. For this purpose, the control period PbA range of 0-1-02 pg/m1 measured in the Griffin tt al. study was used. No control period PbA measurements had been reported for the Kehoe study. From the line of best fit to the Azar et al. data, the PbB values corresponding to PbA values of 0-1-02 (ig/mJ were found to be 16-1-16-2 |ig/dl. Since the Kehoe PbB data were given only to two significant figures, these values were rounded off and a control period PbB of 16 fig/dl was used to normalize experimental data. The adjustment was made by subtracting from individual or mean PbB values, as appropriate, the difference between measured control period PbB and 16 jig'dl. These data are summarized in Table 1 and shown in Fig. 2. Fig 2. Predicted and observed PbB level as a function of PbA. Data are taken from Table I. The Kehoe data points (AAffi) represent 12 individuals, some studied more than once: data points from individuals exposed to large par ticle aerosols (Al: data points from one individual. OH. whose measured PbB from non-air lead sources was excep tionally high (). All data points are adjusted to a PbB from non-air lead sources of l6pg/dL Standard deviations are shown for Griffin er at. points () The line of best fit to the Azar et al. data is shown dashed beyond the range of actual measurement. The Rabinowitz (------ ) and Bernard (--) predicted lines (after Syr) are taken from Table 2: maximum and minimum impact conditions are discussed in the text The differential form of the Bernard model* is dM,/dt (0-133/0-76)Mj + (OOS7/13)M + (00125/27)Mj + (000686/65)M i + (0-2442/04l9)M, dMj/dt - (000686/0-419)M, - (000686/65)M3 dMj/dt - (00125/04191M, - (OOI25/27)M, dM/dt -(0057AM191M, - (0057/13)M dMj/dt - (Ol33/t>4l9)Mt - (0-133/0-76IM, Prediction: linear kinetic models The differential form of the Rabinowitz model (Fig. II* is dM,,dt - <(MX)2/06)Mj + (0007/200IM, - (0015 + 0007 + 0036<T9)M, dMr dt - (0015/1-9|M, - (0002 + 0012/06^ dM, dt -(0007/19)M, - <0007/200)M3 where dMi'dt. dMj/dr and dMj/dt represent the rates of change m the mass M of lead in compart ments 1. 2 and 3 respectively as determined by rates of transfer into and out of the three compartments. In these equations time t is in days and the coefficients of the various Ms have the dimension [day * *]. `Both models are given here for disposition only: that is. it is assumed that intake has ceased. U exposure continues throughout the disposition phase, functions describing the rates of input must be incorporated into the equa tions for dM,<dt. where time t is in days and five equations are necess ary to describe the net rates of change in the five model compartments. The compartments depicted do not have discrete anatomical dimensions. Thus, compartment 1 in both models includes but is not necessarily limited to blood. Therefore, it is not surprising that the amounts of lead associated with specific compartments are dif ferent in the two models. The PbB levels predicted by the Rabinowitz and the Bernard models after exposure for Syr (260wk) and 19 yr (1000 wk) were computer simulated in the following way. The amount of lead absorbed each day into the systemic circulation was entered into the computer program as a single unit. This daily incre ment was allowedio be distributed and excreted in accordance with the magnitudes of the model rate constants (that is. in accordance with the equations given). The initial condition was the complete absence of lead from all body compartments. Table 2 gives the predicted PbB levels at the end of the stated exposure periods as functions of the total amount of lead ab sorbed each day from all sources. 921001 RowVerK n(.) M> 634 P B Hammond. E. J. 0'Flahekty and P S. Gartside Table I. Individual and mean stead v slate PhB let-els in persons exposed to lead sesgutoxide aerosols and values derived from the line of hesi ftI to the data of Azar et al. together with the associated values of Experimental period Subject Control PbB (pg/dl) adjusted or no. of period* PbAt PbB >0 baseline of subjects PbB (pg.dl) (p|/m*l (pg/dl) 16 pgdl >! References MB MOB PB SB FC LD DH - NK HR JOS JUS SS 24 26 26 23 21 22 22 18 18 18 18 30 30 30 30 20 20 20 20 20 20 20 21 21 21 21 21 21 21 21 19 19 19 19 19 19 19 19 19 19 19 19 19 29-44 224 2844 284 274 306 30-U 93 197 27-1 359 56 7-3 76 8-8 06 1-2 19 2-4 33 36 4-0 24 3-7 7-5 94 193 27-1 35-7 28-14 06 1-3 18 24 2-7 34 43 24 3-2 54 02 70 7-2 40 39 32 33 32 39 31 29 36 40 41 31 26 29 28 20 21 23 26 26 30 29 25 21 27 32 37 41 46 32 20 20 18 19 23 24 24 24 21 27 26 26 29 32 055 Kehoe (Gross. 19791 29 058 22 021 26 035 27 040 33 056 25 030 27 1-21 34 092 38 082 39 064 17 018 12 -056 15 -014 14 -023 16 0 17 100 19 176 22 2 73 22 194 26 294 25 2-37 20 (-82 16 0 22 082 27 120 32 084 36 074 41 070 27 039 17 2-5 17 , 091 15 -062 16 0 20 160 21 156 21 1-22 21 2*27 18 067 24 1-54 23 1-17 23 103 26 143 8 20-3 t 3-2 109 36-8 + 4 3 32-5 + 4-3 12 203 4-2 32 260 3-8 21 7 t 3-8 1-54 Griffin et al. 190 (1975) IO 17-108 1-135 Line of best 20 18-133 1074 fit to data of 50 20741 0946 Aaar et al. too 24-285 0825 (I973al 200 30151 0704 300 35-378 0644 400 40386 0608 * The PbA during control periods is not known for the Kehoe studies but was 01-02 for the Griffin et al. 119731 studies. For calculation of (see footnote U baseline PbA was assigned the value 02 pg/m* for the Kehoe studies. tAII PbA levels are expressed as if exposure had been continuous. For example, if a subject were exposed for 8 hr/day. 5 days/week, to a chamber air-lead concentration of (30eg/m1, the value in the table would be [(8hr/day)<5 day wk)/(l6lhr/wkl] 130 pg/tn1 35-7pg/m*. lx is the ratio APbBiAPbA. where APbB is the magnitude of the increment in PbB resulting from an increment in PbA of APbA. The values of APbB are calculated as the PbB values in column 5 minus 16 pg/dL the baseline value. The values of APbA arc calculated as the PbA values in column 1 minus 02 pg/m*. Calculation of x from the line of best At to the Aaar et al. data starts from the baseline assumptions lhat PbA 02 pg/mJ and PbB 16-2 pgdl. The value of i depends on the choice of baseline: tee text for a discussion of this point. f ] (. 1 f-v . `tsrsrjs: 1., 921001 RowVerK u' derived from References IOC Gross. 19791 1 l J 1 1 i l n et al. >5) if best o data of r rl al. 3al 73) studies. 3 hr/day. 5 ayMS day in PbA of !"he values i At to the > aluc of i . impact of air-lead on blood-lead in man 635 Table 2. The impact of PbA on PbB as predicted by the Bernard and the Bahmowitz models under different conditions Toul Pb absorbed daily from all sources Ipgl 35 2 704 105-6 1408 176-0 211 2 PbB level pre dicted by Bernard model (j<g/dl) After Syr (260 wk) After 19 yr (1000 wk) 13 5 305 484 667 85-2 103-9 16 5 34-2 52-4 709 89-6 1083 PbB level pre dicted by Rabinowiiz model (/rg/dl) After 5yr (260 wk) After I9yr (1000 wk) 543 1IOO 166-8 224 5 283-0 342-1 55-4 1119 1694 227-7 2867 346-1 PbA corresponding to daily retention; minimum impact' conditions* (pg/m1) 3-05 8 63 14 22 1981 25-40 3098 PbA corresponding to daily retention: `maximum impact' conditioiut (pg/m*| 128 3-64 599 835 1070 1306 Minimum impact' conditions are chosen to minimize the predicted impact of PbA on PbB. They are: total volume of air breathed daily. 18 m1: 35% of inhaled Pb deposited in lung and absorbed. Lead absorbed from diet (food plus water) is assumed to be constant at 16 pg (1308 x 200mg)/day. t'Maximum impact' conditions-are chosen to maximize the predicted impact of PbA on PbB. They are: total volume of air breathed daily. 23 mJ; 63% of inhaled Pb deposited in lung and absorbed. Lead absorbed from diet is assumed to be 16 jrg/day. The total lead absorbed daily was considered to be the sum of the amounts absorbed from the diet and from the air. It was assumed that total dietary (food plus water) lead was constant at 200pg/day (Rabinowitz et al. 1976) and that 8% of this dietary lead, or 16 jig, was absorbed. The remainder of the amount absorbed was attributed to air-lead. To convert the amount of lead absorbed from air to a concentration of lead in air. certain assumptions must be made about the volume of air breathed daily and about the fraction of inhaled lead absorbed from the lung. Chamberlain. Heard, Little, Newton, Wells St Wiflen (1978) have shown that the fraction ab sorbed is dependent on panicle size. At the same time they investigated the particle size distribution of lead in ambient air. They found that in general the diffusional mean equivalent diameter (DMED) of mature aerosols such as those typical of rural areas and of urban areas away from the influence of major traffic arteries was about 006 pan while the mass median equivalent diameter (MMED) was larger, about 02-03 ton. The DMED of freshly generated lead aerosols measured near major highways was smaller, 003 pm. and the MMED was much smaller, 004 pm. In our simulations two sets of assumptions were evaluated separately: one consistent with maximum impact of PbA on Pt)B. and one consistent with mini mum impact of PbA on PbB. For maximum impact it was assumed that the individual is "reference man" and breathes 23 m1 air/day (ICRP, 197S) and that 63% of the lead inhaled is deposited in the lung and absorbed into the systemic circulation. This absorp tion level approximates the midpoint of the range reported by Chamberlain cf al. (1978) for lead aerosols with very small DMED. 002-094 pm. at breathing cycles of 4-8 sec. For minimum impact it was assumed that reference man breathes 18 m1 air/ day and that 33% of the lead inhaled is absorbed. This absorption magnitude approximates the mid point of the range reported by Chamberlain et al. for lead aerosols having a larger DMED. about 009 pm at breathing cycles of 4-8 sec. The PbB levels pre dieted by the Rabinowitz and Bernard models after exposure for 3 yr are shown in Fig. 2. Results and discussion Figure 2 allows direct comparison of the PbB-PbA relationships predicted by the Rabinowitz and the Bernard models, calculated from the data of Azar et al. and those observed in experimental studies. It should be noted that at steady state the Bernard model will generate a PbB level of 19-0pg/d! for a total daily absorption of lead from all sources of 33-2 pg. corresponding for the conditions of this simu lation to a PbA of between 1 and 3 pg/m1 depending on whether maximum impact or minimum impact conditions are considered. However, because of the very slow return of lead from two of its peripheral compartments (compartments 2 and 3. Fig. 1) the Ber nard model has not yet achieved a steady state even after a 19-yr simulated exposure (Table 2).- Figure 2 shows that the line of best fit to the Azar et al. data provides a good fit to the data points from experimental studies while simulation based on the kinetic models of either Rabinowitz or Bernard pre dicts entirely inappropriate PbB values. One of the reasons for the poor quality of fit is that both kinetic models are linear while the observed relationship between PbB and PbA is curvilinear. Both the Rabinowitz and the Bernard models are linear mammillary models consisting of a central compartment into which the foreign compound enters and from which it is eliminated by excretion or metabolism. Exchange takes place continuously between this central compartment and one or more peripheral compartments into which the compound is distributed. These exchanges as well as the elimin ation processes are first order (kinetkally linear): that is. the unidirectional transfer rate is directly propor tional to the amount available for transfer. Com partments as defined kinetically are not necessarily congruent with specific tissues or organ systems: how ever, the blood is usually considered to be part of the central compartment In general linear mammillary i 921001 RowVerK 636 P. B. Hammond. E. J. O'Flahert'I' and P. S. GaRTside models differ from one another only in the number of peripheral compartments assigned to each and in the values of the transfer rate constants. The utility of linear kinetic models in describing the behaviour of foreign compounds, particularly of drugs, is firmly established. Linear models are . founded on reasonable assumptions about the kinetic nature of transfer across biological membranes, assumptions which have been verified experimentally in many instances. Furthermore, they are the simplest biologically reasonable models. Thus a linear mam millary model is generally the first choice in model ling the kinetics of a foreign compound unless specific information about observed kinetic behaviour dic tates otherwise. The mathematical expression for the time-depen dence of the amount in the central compartment of a linear mammillary model after acute administration (the integrated form of the differential equations given) leads directly to the identification of certain attributes shared by ail linear mammillary models. One of these is a direct proportionality between the amount of material in each compartment and dose or dose rate at steady state. The theoretical proportion ality constant is a function of the excretion rate con stant as well as of the dose or dose rate. One impli cation of representing the kinetic behaviour of a foreign compound by a linear mammillary model is that tissue concentration should be proportional to exposure at steady state. However, it is clear from Fig. 2 that the PbB-PbA relationship is not linear at steady state. The linear models are not accurate rep resentations of lead disposition kinetics. A measure in common use to describe the PbB-PbA relationship is a or APbB-APbA, the increment in PbB associated with a given increment in PbA. As a result of the curviiinearity of the PbB-PbA relationship. 2 is not a constant. To begin with, a is dependent on the magnitude of the PbA increment used in its calculation. This dependence is illustrated in Fig. 3. in which a values for individual or mean data points from the Kehoe and Griffin et al. studies can be compared with a calculated from the line of best fit to the Azar et al. data. For the Azar et al. line baseline PbB was taken to be 16-2 jig/dl the PbB value associated with a PbA of 02 jrg/m3. For the calculation of a from Kehoe data, for which PbA is not known for the control periods, control PbA was arbitrarily assigned the value 02 jig/m3. The calcu lated a values are presented in Table t. It is apparent from Fig. 3 that within the range 0-20 pg Pb/ra3 of air. a is a decreasing function of PbA rather than the constant value which would be expected on the basis of a linear model: that is, as APbA increases, a decreases. Values of 2 calculated from the Griffin et at. and Kehoe data points parallel the behaviour of the Azar et al. line. The wide scatter of a values among the Kehoe sub jects exposed to < 10 jig Pb/m3 is due to variation within individuals as well as between individuals. Thus, for example, subject NK (Table 1) had 2 rang ing from 100 to 2-94 over the narrow APbA range of 1-3-8 jtg/a3 (adjusted). Nevertheless, when individual subjects were exposed to a wide range of PbAs, e.g subjects LD and JOS. there was a clear decrease in a with increasing PbA. 921001 Fig 3. 1 is 1 function of A PbA. Data are taken from Table 3. The line of best fit to the Azar et al. data and the 95*i confidence limits to the line (above and below) are shown dashed beyond the range of actual measurement. Griffin er al. mean values (); Kehoe data points (A). Kehoe data points from individuals exposed to large-parti cle aerosols (A): the only positive value of a from subject DH (Kehoet. whose measured PbB from non-air lead sources was exceptionally high (). The four negative values for Kehoe subjects (see Ttble 1) have not been plotted. a is dependent not only on APbA but also on the baseline PbB value: that is on the amount of lead already in the systemic circulation. PbB0. This dual dependency is illustrated by the simulation in Fig. 4, based on the Azar et at. line. Figure 4 shows that the largest calculated values of a should be obtained for groups of subjects with low baseline PbB levels (PbBo) exposed to small increments in PbA (APbA). In practical terms this behaviour is most marked in the PbB range below about 25 jig/dL the range associ ated with ambient air lead levels. When baseline PbB equals or exceeds about 30/tg/dL 3 is small, about 0-Sjtg/dl for APbA - 10jrg/m3, and is-affected only slightly if at all by the magnitude of the experimental increment in PbA. Fig 4. The relationship of x. simulated using the line of best fit to the Azar et al. data, with baseline PbB for three different arbitrary APbAs APbA - I 0 jigm* (------ 1, APbA - yOMI/m1 (------); APbA IO0jigmJ (l Ro/jVeril ()<)S:l taken from lata and the J below) are leaxuremem. points (A): > large-parti- :rom subject ion-air lead -ur negative e not been i i 4 1 i j Jso on the nt of lead This dual i in Fig. 4, vs that the named for PbB levels 'A (APbA). marked tn nge associseline PbB nail, about fected only penmental * i i t the line elme PbB ! 0 ug m1 100 ugm1 Impact of air-lead on blood-lead in man 637 The curvilineartty of the PbB-PbA relationship in human subjects is supported by reports of a curvi linear relationship between PbB and drinking-water lead (Moore. Meredith. Campbell. Goldberg & Pocock. 1977). and further evidence comes from related observations in experimental animals. Studies by Azar, Trochimowtcz & Maxfield (1973b) and Prpic-Majic. Mueller, Beritic. Stanley & Twiss (1973) suggest that in rats, dogs and rabbits as well as in man. PbB is not directly proportional to lead dose. In the dog and rat studies, the lead salts were added to the diet. In the rabbit study lead acetate was given tv daily for 6 days. It might be expected that the chemical form and particle size characteristics of lead aerosols would influence absorption rate and efficiency and thereby affect the shape of the observed PbB-PbA curve, made-up as it is of data points taken from studies of exposure to air lead having -different particle size ranges and chemical makeup. It is unlikely that the chemical form of the lead aerosol is an important factor in determining airway deposition and clearance characteristics, however. Studies by Chamberlain et al. (1978) indicate that the chemical form of the lead aerosol generally has little influence on the fractional deposition and clearance of inhaled lead. On the other hand these studies (Chamberlain et al. 1978) have established that the particle size of lead aerosols within the submicron particle size range does influence substantially the degree of deposition in the lungs. This point is illustrated in Fig. 2. Many of the subjects in the Azar er al. study were cab drivers exposed while on duty to ambient urban air lead. Chamberlain et al. (1978) found the MMED of air borne lead measured either over London streets or in an enclosed London car park to be 0-3 jim. In general the MMED of the lead sesquioxide panicles intro duced into the Kehoe exposure chambers was 026 pm with 90% of the panicles having equivalent diameters less than 068 pm. but in a few instances large panicle aerosols (MMED of up to 3-98 pm) were prepared. [The geometric diameters reported by Kehoe (Gross, 1979) were convened to equivalent diameters for this comparison by using the Stokes-Cunningham equa tion (NAS-NRC. 1972. Appendix A).] Data points from these experiments are identified as solid triangles in Fig. 2. As would be predicted for exposure to these larger diameter particles, these data points tend to give relatively low s values. Another identifiable group of four data points, shown as solid squares in Figs 2 and 3 also deviates from the behaviour shown by the others. These are the data from Kehoe subject DH. who had an un usually higher baseline PbB. 30 pg/dl. Their position ing in Fig. 2 as a group below the other Kehoe data points is probably an artefact of the procedure used for baseline adjustment with these data, and is a re flection of the curvilineartty of the PbB-PbA relation ship. The APbB associated with a specified A PbA is dependent on PbBo. as discussed above. When PbB0 is low. APbB for a specified APbA is larger than when PbBn is high. Therefore the additive baseline adjust ment. while it is the only practical adjustment to use, is not entirely appropriate. It results in a bias in the adjusted PbB data points toward low values. This bias is probably not serious when PbB0 does not differ greatly from the baseline value of I6jig/dl. but becomes more serious as PbB0 increases, so that its effect is clearly apparent in the group of data points from subject DH. Analogous reasoning suggests that the a values calculated for subject DH should be dis proportionately low. and indeed three of the four negative calculated s values (Table I) are associated with subject DH. It would be of great interest to know how PbB-PbA relationships in children compare with those described here in adults. Unfortunately, there are no data available that would meet or approximate the criteria set forth in the present analysis of the PbB-PbA relationship. In no case has continuous sampling of air lead been done in the breathing zone of children. Only outdoor stationary monitors have been used. Moreover, data in which baseline PbBs were as high as 16pg/dl are scarce. In one study, however, a companson was made of the impact of PbA on PbB in adults and children (Johnson, Tillery & Prevost, 1975). Two populations of females and males were compared in which faecal lead excretions were roughly comparable but PbA was 6-3 jig/m1 in one case and 064 jig/m1 in the other. Using the low PbA group as a baseline, s for the high PbA adult males was approximately l-l as compared to 1-9 for the male children. In females, a for adults was ap proximately 08 as compared to 09 for the children (Johnson et al. 1975). Thus, a may be higher for chil dren than for adults, but probably by a factor of less than 2. It is also interesting to consider the junction of the line of best fit to the Azar et al. data with measured PbB-PbA relationships in populations industrially exposed to lead. For many studies of workers no measures of PbB levels in comparable control popula tions are available, as a result of which the contribu tion of non-air lead sources is not known. Further more. in industrial studies PbA has not always been sampled in the workers' breathing zone. One study in which workers did wear personal air monitors and in which a control population PbB was reported is that of Williams, King & Walford (1969), who studied sub jects working in various different departments of a lead-acid battery factory. When the PbB means for the three worker groups are normalized to a control value of l6jig/dL as described above for Griffin et al. and Kehoe data points, and plotted against the timeweighted average PbA (WHO. 1977), they lie well above the extrapolated line of best fit to the Azar et al. data in the PbA range 33-53 itg/m*. This discrep ancy is unlikely to be attributable to an effect of par ticle size, since the particular jobs on which the men worked are associated with lead dusts of larger par ticle size than the lead sesquioxide generated in the Griffin et al. and Kehoe studies. It may be at least partly attributable to hand to mouth lead transfer in the worker groups. Williams et al. found a significant correlation between' personal working habits and total lead exposure within one of their three study groups. In any case, it should be emphasized that the line of best fit to the Azar et aL data presented here rep resents an empirical fit. As such it cannot be extrapo lated beyond the limited concentration range within which it has been shown to be in agr- jmem with o2,l0;Jl RowVerK 638 P B Hammond. E. J O'Flahcrtv and P. S. Gartsioe experimentally measured data points; that is, beyond PbA values of about 40^jg/mJ. It cannot be applied to industrial data, with PbAs as high as JOO/ig/m1. In general, significant deviation from linearity of the PbB-PbA relationship as evaluated for industrial populations has not been shown (OSHA. 1978). This .observation, which applies to the occupational PbA range 50-300 ng/m\ is not inconsistent with the re lationship reported here. Figures 2 and 3 show that the curvilinearity of the PbB-PbA relationship dimin ishes as PbA increases. This behaviour is also reported by Moore tt al. (1977) in their studies of drinking-water lead. It is interesting to speculate about the basis for the form of the PbB-PbA relationship. By analogy with observations in animals it seems unlikely that nonlinearities in either absorption or elimination are the primary determinants of the shape of the PbB curve, since if this were the case the effect should be seen in other tissues'at about the same exposure level at which it is seen in the blood. In animals, at least most tissues tend to accumulate lead to a degree much more directly proportional to dose than does the blood. When curvilinearity of the tissue-Pb-PbA re lationship is observed, it tends to appear only at ex posure levels higher than those associated with the curvature of the PbB-PbA relationship. That the curvilinearity is not a result of nonlinearities in absorp tion is further supported by its appearance in rabbits administered a lead salt iv (Prpic-Majic er al. 1973). It therefore appears most likely that distributional nonlinearities are primarily responsible for the form of the PbB-PbA relationship. It is possible that these nonlinearities are related to the number, kind, and location of lead binding sites in different tissues. REFERENCES Azar. A. Snee. R. O. Sc Habibi. K. < 1973a). Relationship of community levels of air lead and indices of lead absorp tion. In Environmental Health Aspects of Lead. Proceedinys of an International Symposium held in Amsterdam, 2-6 October 1972. p. 581. Comm. Eur. Communities, Luxembourg. Azar. A. Trochimowicz. H. J. Sc Maxfield. M. E. (1973b). Review of lead studies in animals carried out at Haskell Laboratory--two-year feeding study and response to hemorrhage study. In Environmental Health Aspects of Lead. Proceedings of an International Symposium held in Amsterdam. 2-6 October 1972. p. 199. Comm. Eur Com munities. Luxembourg. Bernard. S. R. (1977). Dosimetric data and metabolic model for lead. Hlth Phys. 32, 44. Box, G. E. P. Sc Cox, D. R. (1964). An analysis of trans formations. J. R. Statist. Soc. B. 211 Chamberlain. A. C, Heard. M. J. Little. P. Newton. D.. Wells, A. C. Sc Wiffen. R. D. (1978). Investigations into Lead from Motor Vehicles AERE, Harwell. England Publication no. R9I98. EPA (1977). Air Quality Criteria for Lead. US Environ mental Protection Agency. EPA-600/8-77-OI7. Washing ton. DC. Gnffin. T. B_ Coulston. F . Wills. H.. Russell. J. C. Sc Knelson. J. H. (1975). Clinical studies on men continuously exposed to airborne particulate lead. In Environmental Quality and Safety. Suppl. Vol. II.- Lead. Edited by T. B. Griffin Sc J. H. K nelson G. Thieme. Stuttgart. Gross S. B. (1979). Oral and Inhalation Lead Exposures in Human Subjects (Kehoe Balance Experiments). Lead Industries Association. Inc. 292 Madison Ave., New York. 1CRP (1975). 1CRP Task Group Report on Reference Man. 1CRP Publication 23. Pcrgamon Press. Oxford. Johnson. D. E,, Tillery. J. B. Sc Prevosu R. J. (1975). Levels of platinum, palladium, and lead in populations of southern California. Encir Hlth Perspect. 12. 27. Moore. M. R. Meredith. P. A, CampbelL B. C_ Goldberg. A. Sc Pocock. S. T. (1977). Contribution of lead in drink ing water to blood-lead. Lancet 11, 661. NAS-NRC (1972). Airborne Lead in Perspective. National Academy of Science. Washington. DC. OSHA (1978). Occupational Safety and Health Standards. Occupational Exposure to Lead. Federal Register 43. 54354. Prpic-Majic. D,, Mueller. P. K, Beritic T, Stanley. S. & Twiss. S (1973). Delta-aminolevulinic acid dehydratase activity, lead blood levels, and the reticulocyte count. In Environmental Health Aspects of Lead. Proceedings of anInternational Symposium held in Amsterdam. 2-6 October 1972. p. 211. Comm. Eur. Communities. Lux embourg Rabinownz. M. B. WetherilL G. W. Sc Kopple. J. D. (1976). Kinetic analysts of lead metabolism in healthy humans. J. din. Incest. SB, 260. WHO (1977). Environmental Health Criteria ). Lead. WHO. Geneva. 74. Williams. M. K,, King E. Sc Walford. J. (1969). An investi gation of lead absorption in an electric accumulator fac tory with the use of personal samplers. Br. J. md. Med. 24. 202 92luu( tfowVer.. 091-1