Document ymXYpqJ97jyEoEvLDrmRbVMe2

/MMtMu/, VoL 4, pp. 157.162, MO to the USA. AU right* tamvad. 01M120/0/Q201$7-06SOtQO/0M'J CopyrightI960 PergamomPms lid. - threshold and dose response estimation USING THE "FILTER MODEL1' David J. Schaeffer nois Enviconmtntal Protection Agency 2200 ChurehiH Rood SpringjWa. UHnoii 62706. USA Konanur G. Janardan M0iSysT 9sngamon State UnfcafOrty Springfield, aiinoia 63706. USA Harold W. Kerster Environmental Studies Center Csfifcrnia Slate universe Sacramento. California 95819. USA (Riartotd 4 April 1980; Acctpitd 3 August 19801 Tbc "filler model" has bees developed to explain the biologic effect* of radiation end ebemicels. W te*e nemliwit neuty 300 w of dose response data. o! -which 50 are presented her*. Responses (Maced by radiation and chertsicaJj) which have bees examined include in titra survival studies on iwi.i ugi plant tiaaucsi induciiOB of aberrations and lime to tumor or death. Similar data from ii woo studies has also been examined. AU of the data appear to fit the model Ro InD+bOaD^+c. there A is the response, a and b are parameters filled by regression to a particular set of data, and c is the response at zero (or lowest) dose. By writing this modal hi exponential form, it can be sacn that the response R results from multistage filtering (by net amounts o and A) of the initial dose. D, The thmhold is obtained from this model as the point br. at which the second derivative becomes sero. This is given by r--exp< t -- a/lb) when a and b arc oppositely signed. Introduction Recent United States federal legislation has brought into sharp public focus questions and scientific con* txoversies about results from studies at high doses *hich are used for extrapolation to low doses. The models used for extrapolation are highly controversial (Groer et a!., 1978; Lewis, 1980), and acceptance of s given result is often based on a legal or nonscientific criterion rather than a technical one. For example, as a result of a law suit brought against the US Environ* eternal Protection Agency (USEPA) by the Natural Resources Defense Council, a .'uMic interest group, the J5 District Court mandated a list of compounds to be studied and regulated by the USEPA. The proposed environmental criterion for carcinogens of an absolute value of 2ero is based neither on scientific data nor on technological feasibility. Totter and Ftnamore (1978) have noted that perhaps because of the controversy ...over tU ?:?~r method of extrapolating to low doses and of comparing the tow-response curves of oae specks with the curves of other species...the dose*respoasc information that wai gradually accumulated was never collected and comprehensively analysed. These authors, began the process of such analysis by comparing the dose-response behavior of 37 data sets 157 AP00009663 151 Sducffn, JiaanUo, tod Kcnur representing a wide range of organisms, exposure mechanisms, and toxicants. For this comparison, they employed a saturation curve {Eq. (1)] suggested earlier by Totter (19741. logO//- 0- -vi(IogZ)-logic), (0 where D is the dose in rem, mg/kg, or other appropriate units. K. the dose required to produce a 5OT& effect, and /. the fraction of subjects affected, are solved for by least squares. Totter and Finamore's (197$) conclusions that there is no response unique to radiation as distinct from chemicals, that simple models containing a few param eters of which (at least) one is related to the genetic material, and "...an unexpected universality of biologi cal behavior that may be helpful in the extrapolation of experimental data to humans** support the conclusions (Jsnardan and Schaeffer. (977)'reached earlier. At that rime, we had developed as a model of the processes leading to the production of chromosome aberrations [Eq. (2)]. the Lagrangian Poisson Distribution (LPD) (Eq- (3)1. A Chromosomes queued for damage B X, Chromosomes ^ queued for B restitution (2) Taking A, as the rate of the damage process and B as the rate of the restitution process, the LPD is obtained u where }K is the survival fraction at the Afth dose and Aj--Xj/B. The LPD is the same as the simple Poisson when Aj0. Based on examination of over 100 data sets representing radiation and chemical damage to plant, animal, and human tissues we concluded from the model alone that ...The Bwcbukiiu of damage appear to be independent of die nature of the. 'insult.' Allb. the same results are found for divergent species. wWch suggests (from the modsl alone) similarities in' which nay Imply similarities in chromosome structure among various organism* (Janardan and Schaeffer. 1977). These conclusions are supported and extended by the results reported here. Data and Oiscussion Schaeffer ei at. (1980a) have shown that the LPD can be used to estimate the energy (4(70) required to produce cell damage by noting that X2 is the ratio of rates for the induction and restitution processes. Hence, at equilibrium. Aa is the thermodynamic equilibrium constant. X^, Because of the relationship between X2 and the LPD, like Totter aoc Finamore's (1978) model, provides a superior way ol tramming dose-response since thermodynamics can be applied in the examination. Although w* have for mally demonstrated the relationship between X3 and Kn, that between X2 (or, in general, response) and dose (>) has not been given. The relationship can be developed empirically by noting that a plot of Aa versus the logarithm of the dose is curved (Janardan at at* 1979). By analogy with Cook et ai (1969) and Ciecka ei at (1979) the relationship is then given by Xs-aln(i>+l)+h[ln(J) + l)]2, (4) where a and b are parameters to be estimated for the model. A biological understanding of this model is devel oped by writing Eq.(d) {or subsequently Eq. (5)] in exponential form. In this form, the response is ex ponential fexp<--Aj)J and the dose a complex exponential function {D'D -WoC). This function is interpreied as the net product of a multistage process which **fiUer$** the initial dose by amounts a and b, where a and b are random variables from an unknown distribution. For this reason we call a model of the Eq. (4) form the "filter model" A formal derivation based on the microdosimetric approach to modelling radia tion damage given by Keilerer and Rossi (1972) and Kellerer and Brenot (1974) has recently been developed (Schaeffer el a/.. 1980b). Since X2 measures the extent of reaction prior to equilibrium, and is the equilibrium constant at equi librium, it can be replaced by other measures of reac tion extent, R : Rc + flIn(D+l) + 6[ln(D + l)]a, (5) where e is usually taken as the response at zero (or lowest) dose. Equation (5) is similar to a lime to death model {Eq. (6)] used by Cook et at (1969) in which the incidence at time t, l,, is given by lg(/,)-C+*[log(0] +rf[log(/)]2. (6) Equation (5), however, expresses the relationship be* tween a generalized dose. D, and a response, R, which is measured on a linear (rather than logarithmic) scale. Threshold Estimation The threshold for a dose effect on survival is the point at which the slope between zero dose to the point in question changes, e.g,, from positive or zero 1(5 negative. The point of slope change for Eq. (5), Df a AP00009664 Tabte I, Dow wlral cw lor ToBgad Ft--iwi* (1978) dwminal agaait> adag Eg. (5).* Substance tasted JtopOQM* 0* b Dr R* Mvnm1 DhniiaibMMK Dibcssamhraeeae DUaettylb--(a)iatlufm gcazc<)pyrm Beatt(c)pyfeM Dibcupyreoe DidbyUtilbesteroi Untbui Urahaae A/Utoxin B| Vitamin D 4-DisMlkyktbyi--fimaiohaaxBl Mmhyfcbolainhrao EtkytafeMMUfM sbQtUBMT Sab Q fibrosarcoma mammary tumor respiratory tract tumor papilloma Sub Q fibrotarcoau mammary carcinoma tumors tumors carcinomas boot aib increase liver tumor Sub Q tumor offspring nervous system turnon a/K* ag/Kg 8/Kg as/Kg -OJ66Q -04151 0.1547 02290 ng/Kg ag/Kg oosot 006096 mg/Kg mg/Kg ag/Kg mg/Kg e 0.04840 -0.0370 --0.1926 01561 0.1 Ilf mg/Kg 04601 mg/Kg -04769 ag/Kg -O02MI 01875 006749 -00644 -00517 -0X0498 --0.1654 -0.01609 -000338 02298 -00788 -0X8025 -005142 002024 --001582 015 0.15 904 18.19 097 098 097 096 Bryn (1943) <TOam(l65) Philips (1973) Pena (1973) 704 0996 Pud (1959) 307 099 Wodiuaky (1965) (I6vnk) I2J0 0.95 Gau(1964) 0204 0996 Sehoafal(l977) OIO* 0997 Schmahl(l977) 702 098 Wagu(!974) 5.70 0489 Waddell (1947) (4.75)9022 057 Dracfcmy<l967) 029 0.98 Bryu(1943) 4.7- 095 Druckray (1969) -Data from Tabtee I aod 2 of Totter aad Rumors (1971) whkk should bt coarulted lor the original nfanaen. kRttp0OM Mend u surviving fraction Si s givsa drwsga sad eomcted for survivsl fraction u lowest don. AH date points usod sslow otherwise specified. *Domjm correspond to thaw given in "Calculated Dougs'* column of Totter and Finamocc (1978). -Thrsshdd calculated using the procedure in Schaeffer f al (1990c). Does a units per 100 g of feed. 159 d by setting the second derivative equal to zero, gives j5r-exp(l-fl/2A). (7) t dung* occur* if cither a or b is negative. If both the lame sign the slope cannot change and the hold cannot be mIt-mU-a in this fashion and the :dure given in Schaeffer f al. (1980c) must be (Table 1). The use of Gq. (5) and Eq. (7) to ate the threshold is most readily discussed in on to radiation experiments, idiation is highly penetrating, and because of this threshold for radiation damage exists it should * at about the same level in fit vivo and in vitro -ns. The experiments in Tables 2 and 3 were cond over doses of several hundreg to tens of thouof rads. The Dr values determined for these data a geometric mean and 95% confidence limits of '24.18<35.63. An additional 80 sets of radiation -lot included here afforded Dr values between S 0 rads. This range is less than 1% of the range of >mnton experients (those with high doses < 5.000 In Conger and Constantin's (1974) experiment rley seeds the lowest dose was 8700 rads. Howthe estimated threshold (272 rads) is reasonably to the values given in Table 2. Whether the .*nce is real and reflects some differences in the ises of animals and plants, or comes from having >oUted nearly 9000 rads (between zero and the measured dose), cannot be ascertained with cer- from our results. However, comparison with Smith et al\ (1974) data on the effects of X-rays on maize seeds in which the low dose was 0.5 rads and ^4.2'rads, and with Bauer and Kaufman's (1938) study of Drosophila sperm where the low dose was 1000 rads and Dr95 rads, suggests that the high Dr value found for Conger and Constantin's (1974) data is from extrapolation over a large range rather than from biologically related response differences between plants and animals. This example illustrates the extremely high predictive power of Eq. (5) to model d ;response effects over the studied range since a 9000 rad extrapolation affords a predicted threshold within a factor of 3 of the other estimates. The ability to use the model to reasonably estimate a threshold from high dose data is of particular importance in assessing the health significance of low doses of environmental mutagens carcinogens, or radiation (Groer et al., J973; Lewis, 1980). In order to facilitate comparisons between models. Tables 1 and 2 summarize the results obtained from Eq. (S) using the data in Totter and Finamore's (1978) Tables 1 and 2. In our work, the response* are the fraction surviving the exposure. In Tables 1-3. c is taken as the survival at lowest dose. Other choices of c, such as c-- 1 when the response is survival fraction, or c0 when scoring chromosome breaks only slightly affects most of the models. Thus, for Druckrey ft ai'& (1959) study on the production of neuroepithelial tumors in the offspring of pregnant female BD1X rats exposed to eihylnitrosourea. the observed survivals are 0.4, 0.22, 0.10. 0.03, 0.02. while the survival fractions predicted from Eq. (5) with c--0,4 and r- 1.0 are 0.30, f SJ^eMaaddo--respooie W` `jii- Table 3. CbuUwd. Substance testad/ayetem X-rays survival fraction of JUmvjvm* 15 days X-ray* mean survival (day?) for 30day daoedeau foutiawm fraction without gross body malformations from imdiatiag ofireut sperm "Coy-rays fractioe without roes body malfotmaiioa from irradiation of trout sperm Average latsecy pariod (weeks); aaginniTomi is Sprsgue-Dawley raw exposed to vinyl chloride Fraction of Sprsgue-Dawley ratt without aepoasrcoau after vinyl eWorld* exposure Average latency period (weeks): lun* tumor* ia Iviu mice exposed to viajrl chloride , ^Fraction of Swiss mice without luag tumors after vinyt chloride exposure " Fraction of CWcodrn gmtarta surviving ethylene oxide Ctposurc -? ... Estimated Dosage Survival survival 440 R 600 <00 1000 400 R 00 800 0.500 0461 0.150 0 134 11.7 104 1000 25 R 50 100 200 400 74 048 0.975 0.969 0460 0402 25fc so 100 200 400 50 ppm 250 500 2500 5000 10000 0492 0494 0490 0.987 0453 135 79 *1 7* 70 64 50 ppm 250 500 2500 6000 041 043 048 0.71 0.78 >0000 0.IS 50 ppm 550 500 51 AS 41 2500 6000 43 3S 10000 36 90 ppm 0465 250 0.43 500 044 2500 0.43 6000 04 10000 04 1.01 ppm 1 U7 048 1.5* 043 - t-66_ 041 1.82 0.74 2.06 0.12 2.10 047 2-30 0.06 2.46 0 043 041 0.147 041 14.1 11.6 94 74 0490 0.97* 0.961 0.941 0417 1400 0.994 0486 0.975 0462 1094 97.1 91.1 754 67.0 61.6 045 041 0.89 044 041 0.79 414 45.8 44.4 40.5 38.0 364 0.70 047 042 0.36 040 0.25 0.96 0.71 0.63 0J6 0.46 048 045 0.09 ooo ft 4* Dr lave*tip*tor 049 55.9 Badr and ftadr (1971) 0.97 56.9 Badr sad Badr (1971) 0.90 115 Newcomb* aad McOtegor(1975) (experiment 1) 041 214 Ntwoonbeaod McGregor (1975) (experiment 4) 0.94 214* Malteot(1977) Table 8 041 3.3 Maltoai (1977) Tabic 1 Q.96 3.0 Maltoui (I9T7> Table 17 043 29.1* Melton (1977) Table 17 0.96 !J**f. R. BusvtM as given in Finney (1971) Table 44 "Corrected for survival si lowest dose. *Thrahold calculated tarn* die procedure ia Scbuffer o a!. (1980c). lil U7,0.07,0.00 (/f?--0.95, 0r-4.74 mg/kg), and 0.23, 0.12, 0.04, 0.007 (/?2-0.9996, /Jr-0.06 o. ie 3 presents several data sets not reported by and Finamore (l'?8V The examples include on survival after irradiation or exposure to als. Additional examples use time to death as the response. Although for the latter we use c as the survival time at the lowest dose, it might be desirable to treat c as a parameter to be fitted if the data set is Urge enough. If c is treated as a parameter, then its value is an estimate of the survival time at zero dose. For the Swiss mice developing lung cancer after exposure to vinyl chloride [Maltoni's (1977) data given in Table 3J. j i A ir AP00009666 142 Schaeffer, Jcnardan, sad th* Mtiaataft suivivil time at uto dose it 67 wssks, lor exampk. [The reader should be aware (hat Pcto (Federal Register, 1960) has suggested that there may be no real UmeHo-tumor dose response effect, and that the interpretation of such data may be more complex.] We think Roger Kaaerva sad Miles Msuzy for encouragement end support Sheik Htakky tad Muit Gregory typtd tbs manuscript Clark Olson brought Psto (1980) to ear stten* den, tad provided helpful comments u our work developed. References Badr, F. M. and Bidr, R. S. (1971) Effect of X-inadistioa on survival of wild htaiwy nut, Rodtor, itas. 4a 256. Buy, H, ud Kiufmas, a P. (IMS). X-ray induced chromosomal tltsntioat is Drooopkita motenoastoo, Ccutb 13,610. Pecks, K PiWaa, IL, and Msrdttt D. (1979) A statistical model for amaB lake wntm quality management, Water Awr. M 14 iaia Conger. B. V. ud Constantin M. }. (1974) The effectiveness of fbtioa neuttoas, 14.7-MeV monoenergciic ueutreas and *Co radiation ia iiudliag growth reduction of cWorophyU-deffcieai mutations in barley. la BioiogiceJ Effects of Neutron Irradiation. pp. 417-432. lAEA-SM-179/IO, International Atomic Energy Agency, Vienne. Austria. Cook, P. J, Doll, R., and FeUingham. S. (1969) A mathematical model (or the age distribution of cancer in ana. /w. J. Cantor 4, 93. Druckrey. H- Preussman. R,, ud lvaokovic, S. (1969) N-Nitroso compounds ia organotropic ud transplacental caretsogens, Ann. N.Y. Acad. Sei. 163, 676. Federal Register (I960) Fart VII Department of Labor, Occupational Safety and Hmdtb Administration: Identification, Classification and Regulation of Potential Occupational Carcinogens, 46,5134. Finney, D. J. (1971) fnbit Analysis, 3rd ed.. p. 74. Cambridge University, Cambridge. Gfoer, P. G,, and Piy. JL J. U, Maatchbergm, M. L, aad Utmu V. R_ (1971) Environmental biological haurda and eomath* riaka, fmtiru. /or. 1, 285. ' Janardea. K. O. and Schaeffer. D. J. (1977) Medela for fee naalyfe of ckromonoasl tbamiioai m koau leukocytes, Biotat J. an 399. ^ Juardu, K. G,, Renter. K. W, aad Schaeffer. E>. 3, (1979) Biotog^ cal applications of (he lagrugiaa Petmon distribution, ifaSm ener 29. 399. Lewis. H. W. (I960) The safety of fission reactors, Set Am. 242. S3. McLaughlin. M. Mh Dacquiato. M. P- Jacobus. D. FM and Horowi^ R. E. (1964) Effects of the germ-free slate on rsspoaim of mice m whole-body irradiation. MW, 8a 23, 333. Maltoai, C. (1977) Vinyl chloride earcmegmiciQu An **p--'timtU model for caidnogmaiii studies, in Origins of Huau **inrat VoL 4, Book A, pp. 119-146. Cold Spring Harbor Lahnton, Cold Spriag Harbor, New York. Newcombs, H. B. aad McGregor, J. F. (1975) Dna< rrtponn tfe tieafeipe tor ike production of body malformations in trout t? exposure of eperm to low doeceof radiation, Redid. Rea 61,519. Schaeffer. D. A, Janardaa, K. G- CSarke, A. C M, aad Kottv, H. W, (1960a) Statistically estimated free energies of chromosome aheriiiion production from frequency data, submitted for pfe tinttiwii Schaeffer, D. J.. Janardan. K. 0,, and Keister. H, W. (1960b) Derrtopment and applications of an equilibrium dosc-rcepoam model, Soairen. Muugtnasis 2.314. Schaeffer, D. J,, Janardaa, K. G,, ud Kcrstcr, H. W. (1980c) Threshold estimation from the linear dose response model: L Method and radiation data, submitted for publication. Smith. H. Hn Rossi, H. K. aad KeUerer. A. M. (1974) Relation between mutation yield aad cell lethality over a wide range of X-ray and fission neutron doses in maitt. ia Biological Effects of Neutron Irradiation, pp. 405-416. IAEA-SM479/2S. Interna- tiooal Atomic Energy Agency, Vienna. Austria. Totter, i. R. (1974) Presented lecture at the annual meeting of the Am. Assoc. Adv. Set. Totter, I. R. aad Finamore, F. J. (1978) Dote response to canccrw- genic ud mutagenic treatment*. Emi/on. Iru. 1.233. AP00009667 An Analytical schema was developed for vinyl chloride, which is applicable to ambient end fn-piant atmospheres. Using computerized gas chromatographs equipped with automatic infection system and flow rate control, high reproducibilities are achieved in the ppb range. Samples from various sources have been analyzed. Improved methods for sampling and analysis of vinyl chloride R. C. LAO. R. S. THOMAS and J. L. MONKMAN Chemistry Division. Technology Development Branch, Air Pollution Control Directorate, Environmental Protection Service, Ottawa, Canada 'Z/0 Introduction Because of increased public health and envi* ronmcmal concern about the effects of vinyl chloride on human health.1*1 Regulations have been promulgated by several countries, ordering reduction in the concentrations which had been present in industrial and ambient urban at mospheres prior to such promulgation.4 Campliunce with these various regulations demands straightforward, reliable analytical methods, which can be applied to a broad range of air concentrations and sampling techniques. In addition, unit analytical time should be very short because of the Large number of samples which have to be processed. Gas chromatogra phy, because of its fast turn around time, wide concentration linearity and high sensitivity for lightweight hydrocarbon molecules, has been selected as the analytical finish for a variety of sample types, including air, water, polymer resins, and solid polymer stocks, filled or un filled, plasticized or unplasticized.' Current processes for the production of vinyl chloride monomer (VCM) fall into one of the following classes.* (1) The direct chlorination of ethylene with subsequent dihydrochlorination. (2) The addition reaction, of acetylene and hydrochloric aeid. (3) A combination of direct and oxychlorina- tion of ethylene with subsequent dihydro chlorination. a. c. Lie. Heed. Special Projects Section. Chemistry OMslon, Air Pollution Con. trei Directorate, received hie B.Se. degree in Cham, ieel Engineering from the University of Taiwan, PM h.e graduate work in Chem ical engineering et the University of Utah and re ceived hie Ph.O.' In Chem istry from the University of Colorado. Denver. R. 3. Thcmat. SUtt Mem ber of Special Projects Section, Chemistry Division, did hie undergraduate work In Chemistry from Queen's University (Canada). His interests include optical spectroscopy, electron mi croscopy and ehrometeg- rapny-mass spectrometry- computer applications and he has authored a number of papers In these areas. J. L. Monkmen, Chief, Chem istry Olvisien, Technology Development Branch, Air Pollution Control Olrsc- terete. Environment Can ada. ceeeived his fl.Sc. degree In Chemistry from the University of Toronto (Canada). Ha has authored more than MX) technical articles In the areas of industrial hygiene and environmental pollution. Amrncaa Industrial Hygiene Association Journal AP00009668