Document b58nRvR8kXXKYd5wOpaXd3RJD

mAmeri=n !Petroleum ~ 1220 L Street, Northwest Washington, D.C. 20005 202-682-8000 ~ Date: To: From: Subject: October 16, 1990 Hugh Spitzer Cosmo DiPerna Peter Craig Mark Swanson Pat Beatty Chuck Lambert Don Stevenson Carol Houth Paul Price rfe-A. ' f~ Summary of first meeting of the Biologically Based Risk Assessment (BBRA) Advisory Workgroup with Dr. Tony Cox The first meeting of the BBRA Advisory Workgroup was held on September 24 in Denver. This group was created at the September 13 meeting of the Benzene Task Force. The purpose of the meeting was the r-eview ef Dr. Cox's work on the development of a PBPK model, as the first stage of his biologically based model. For background information, find attached a copy of-Or. Cox's contract. The group heard a detailed presentation of the work which has been performed up to this time,"""See attached outline from Dr. Cox, and discussed two recent written products, "BIOLOGICAL BASIS OF CHEMICAL CARCINOGENESIS: BENZENE AS AN EXAMPLE" and "BIOLOGICAL EVIDENCE THAT THE ONE-HIT MODEL AND LOW DOSE LINEARITY ASSUMPTIONS DO NOT HOLD FOR BENZENE". An equal opportunity employer BP-00016005 In general the group was pleased with the progress that Dr. Cox had made on the PBPK project and agreed with the general directions of his efforts. Several specific comments were made on tile PBPK-project o PBPK in mouse, rat, and man will play an important role in the new benzene risk assessment. o Efforts to incorperate non-steady-state assumptions into the PBPK model should be reconsidered. Modeling_of microvascular events is perhaps an unnecessary complication. Dr. Cox should--perform a sensitivity analysis on the benefit of using non-steady-state equations and demonstrated that the additional complication was warranted. o The Work Group agreed to provide additional help in obtaining additional informatio:n for the model: Peter Craig will provide a copy of a recent paper by Dr. Ross. Pat Beatty will speak to Dr. Mike Gar:gas over recent low dose work on benzene metabolites and metabolism. o Dr. Co)( will meet with lreAs and Ross during October to discuss the PBPK modeling project. There was considerable discussion on the paper BIOLOGICAL BASIS OF CHEMICAL CARCINOGENESIS: BENZENE AS AN EXAMPLE". Several members e)(pressed surprise ihat Tony would author a paper summarizing metabolism and pharmacokinetics of benzene since his area of e)(per:tise is Operations Research. This concern was augmented by the fact that the paper had already been submittedto the Joumal- of the Society of Risk Assessment ana -was tentatively accepted for publication. Other members found the paper to be, outside of the nomenclature and other problems-of phrasing,-to be an excellent summary of the current state of understanding of the biology of benzene's leukemogenic effects. BP-00016006 Dr. Co)( e)(plained that he- had been approached to submit a paper on BBRA for benzene by the Journal's editor and after some discussian they had concluded that the draft report specified in his API contract would be appropriate basis for such a paper. He agreed that he had been in error in not notifying API of his submission and not providing opportunity for review prior to submission. However, the paper's submission was unusual in that the Editor requested an early copy of the paper and agreed to allow Dr. Cox to simultaneously distribute the paper to external reviewers and to incorporate their comments. Copies were thus sent to API and to an number of e"perts in benzene toxicology, including, Dr. Ross (U. of Col.), Dr. Michael Byu:i (Enon), and Dr. Ben Thomas (ENSR). Because of this simultaneous review there will be no problem in incorporating API's comments in to the document before it is published. Dr. Cox agreed that if API and he could not come to agreement over tbe paper then the document would not be published. The group a9reed to review the fjaper and prepare comments by the Santa Monica meeting of the benzene Task Force~ These comments will be incorporated by Dr. Co)( along with comments from other reviewers. The Task Force will review the revised document before it is resubmitted to the Journal. The discussion on the second paper "BIOLOGICAL EVIDENCE THAT THE ONE-HIT MODEL AND LOW DOSE LINEARITY ASSUMPTIONS DO NOT HOLD FOR BENZENE" focused on whether the ar-suments presented actually demonstrated that non-linearities wer-e observed-in the range where the current models predict linearity. Most of the work cited in the paper clearly fell in the range where current cancer models demonstrate non-linear behavior. The group agreed that the current data was not sufficient-to compellingly demonstrate that dose responses wer-e non-linear in the nanogram/kilogram range. The group decides that paper served a legitimate-function as a summary of non linear processes but should not be taken any further at this time. The group agreed to meet again in Denver in November either the day before of after the meeting with Dr. Irons. BP-00016007 / Fa. LL I pr , C~-' -A p L '11'101:~ lF'rom: Tony Cox, Cox Associates ~e: Statement of Work for 1990 Research on Benzene Dlil~e: 1215/1939- z/2 Z/ Of .Rf In response to your recent request, I-am pleased to submit the following statement of work for applied research on quantitative risk assessment of the human health effects from benzene, to be undenaken in 1990. A. llhclkgroun~ In 1989, Cox Associate_s_ developed for the American Petroleum Institute a biologically-based cancer risk assessment framework for benzene. The emphasis in that work was on integrating mathematical models. of (i)-benzene pharmacokinetics and metabolism (PB-PK lllGdels); (ii) celllcinetics and-cytotoxic effects; (iii) leukemia induction; and (iv) leukemia progression, into a complete, computationally viable, risk morlel. The methodologydeveloped involved a m:i-x"ttlre of companmental flow morleling and stochastic simulation for cell popuiations. The results of this modeling work were documented in chapter 4 of a repon on New Directions in Cancer Modeling arzd Risk Assessmenr for Benzene. submitted jointly to the American Petroleum Institute and the Western States Petroleum Association for review and comments in August of 1989. This chapter also identified needed..e_llhancements in current risk modeling techniques for benzene in the areas of PB-PK modeling (e.g., adding bone marrow as a separate compartment), cell kinetics (e.g., creating a disaggregated companmental model of the hematopoietic system) and leukemia induction--{e.g., representing the interaction between cytotoxic and leukemogenic effects in the bone marrow.} The remainder ot the report identified methodological flaws in statistical risk assessments of benzene underlying current regulatory risk analyses, and concluded that current quanti-tative risk estimat-es for benzene health risks at low exposure concentrations are biased upwards by at least a factor of 20 (and probably much more.) This large error in current risk estimates motivated the search for more biologically realistic benzene-risk models. Finally, a demonstration computer program was develoJre(l to illustrate the modeling-framework and to prove its computational practicaliry. The modeling work completed in 1989 established a -framework for more realistic quantitative assessment of ilie hwnan health risks from inhalation of benzene;- However, this framework has yet to be applied. The puJJX>Se of the work proposed for 1990 is to begin BP-00016008 applying portions of the framework for which adequate data exist to the quamiitication of human health risks from ~nzene. The goal is to use biological knowledge, data, and models to suggest quantitative correction factors for-current risk estimates. We propose to conduct three tasks spread over eight months. Descriptions. deliverables, and estimated resource requirements f-or each task follow. 7/'as!t 1: M.~view and incorporate e:risling data Oii lJenz.eYile 8va8o She wsodd. Considerable experimental data have now been collected in each of the relevant areas of (i) Benzene pharmacokinetics and metabolism; (ii) CytotO)(JC effects and effects of continued exposure on short-run and long-tenn celllrinetics; (iii)_ Genotoxic effects; and{iv) ~ossib1e epigenetic mechanisms of benzene-induced leukemogenesis (including protein lcinase C activation as well as more lraditional stem cell proliferation effects_)_ The state of each of these areas of lmowledge about benzene biology as of-early 1939 is well portrayed in a special issue of Environmental Health Perspectives on "Benzene Metabolism, Toxicity, and Carcinogenesis," (EHP, g2, July, 1989.) Task I will be to review and integrate these data in the context of our biologically-based risk model. The purposes of this task are as follows: (i) To identify data elementS required by- the model that can already usefully be supplied from current knowledge;- (ii) To identify critical data gaps that still need to be fllled to make biologically-based risk modeling for benzene practical; -and (iii) To provide a unified summary of c\ment, practically applicable, !mowledge of benzene biology as it applies to risk assessment. Deliverable from Task 1: A technical paper.-suitable for submission to a peer-reviewed journal, discussing existing quantitative data on benzene bioiogy \l.ithin the framework of the biologically-based stochastic simulation risk model of benzem: developed for API in 1989_ Estimated Resowce Requirements for Task. I: 150 hours, 15k, three months. 7/'g,sf! 2: JD<rv~lop t1 dei~Ailed l!'JB-P/JLcomputer model for benzene -and populate it with ~existing dat~&. BP-00016009 The most prorluctive research strategy for developing defensible quantitative correction factors for benzene risk estimates in the short run is likely to be improved PB-PK modeling. Task 2 will build state-of-the-an computer model!;_of benzene pharmacokinetics and metabolism in rats, mice. and to the extent now possible, man. The models will incorporate recent developments (e.g., by Bois er a/ at the University of California at Berkeley on detailed phannacokinetics for Fischer-344 rats). including addition of bone marrow as a separate metabolizing compartment and target site. They will be-validated by comparing numerical predictions with published data. The PB-PK work described here has practical value for two_r_easons: (1) It can be used to interpret past animal bioassay data by more accurately quantifying the bUe internal doses received under different exposure regimens in different species; and (2) It can potentially yield more accurate projections of the true internal doses received by humans under assumed occupational exposure conditions. In similar modeling work on butadiene conducted for the Chemical Manufacturer's Association in 1989, we have feund that regulatory risk models have underestimated animal internal doses in bioassays and overestimat~ projected human internal dose levels from occuparionaLexposures, leading to an overestimate of human risks extrapolated from animal data of at least two orders of magnitude. A similar methodology applied to benzene may yield similar results. The PB-PK models developed in this task should also help to clarify some of the surprising anomalies of recent animal experiments (e.g., that increased exposure concentrations for increased exposure durations can_produce dramatically fewer leukemia responses in mice.} Finally, the results of this task should also shed light on the relevance of different epidemiological exposure conditions and data bases (e.g_, eight-hour shifts with relatively low, variaole concentrations in Pliofilm workers in Ohio vs. 24-hour, hig_h-concentration -exposures for Turkish s!foe workers) for occupational and public health rislu.ssessments. Deliverable from Task 2: Computer PB-PK models for mouse, rat, and huma.~. The mouse and rat models shall be populated with available data from the literature. The human model will use plausible guesses for parameter values where data are not available. All models will be delivered on one or more floppy disks, and all--data values used and assumptions made will be identified as part of the -model documentation. Estimated Resource Requirements for Task 2: 200 hours, 20k, six months. BP-00016010 .. i!'I!!Ji>lr :J: IJ.ppKy OCae JPJ!).f?C{ waodld$ ti111w:Kop(uff iva 'il'1!!1$fs 2 P 1!!1/liJtil r!POfuF vrz/ia'Cfl!iiiiP.O biological lusowl~dg~ d:md data summtuiz.ed in 7!'txs& J. 11o sBDggest mvuJ justify ~BOavatitative COITrnu:tiow. factors foil' exiseing fo(tvazeVBIZ vis& ~stiwuztes. A primary reason for doing biologically based risk modeling is to obtain more accurate and more defensible risk estimates. The data to support a full biologically-based risk assessment for benzene are not yet available. Use of simple ad hoc "biologically motivated" modeling assumptions for benzene in place of solid data have recently drawn much criticism. But there is another way to use the considerable panial data that are available to improve benzene risk estimates. By quantifying the differences in biologically effectivedoses received (i) between animals and man; and (ii) between worker populations exposed to -different exposure conditions, it should be possible to suggest numerical correction focxors for previous quantitative risk estimates. The logic is as follows: using the regulatory assumption that cancer risk is approximately proportional to biologically effective dose. estimated bounds on --the ratio of internal doses between alternative exposure conditions and/or species are also bound-s for relative risk ratios. Thus, the available PB-PK data can ~ used to correct previous relative risk estimates using internal dose information alone, without requiring speculative--assumptions about the biological damage mechanisms involved (except the assumption of proportionality between internal dose and health response, which is already pan of the linearized multistage model endorsed by the agencies.) Malcing these corrections leaves open the possibility that the linearized multistage model for biological damage (via genotmdc "hits") will eventually tW'll out to~ inappropriate for benzene. Since corrections are based only on the PB-PK portion of the complete cancer risk morlel. for which relevant data are comparatively available, they do not foreclose funher com:ctions at a later__date in ihe response ponions of the model (cytotoxic, cancer induction, and cancer progression submodels) when the-data for these moduies become available. This correction factor approach to revision of quantitative risk estimates based on panial biological data grows out of work performed by us for the Che-mical Manufacturer's Association in 1989. It has already been successfully applied to butadiene, for which -comparatively little data and primitive models are available. The goal of Task 3 ino apply it to benzene, to discover the extern-to which existing knowledge of benzene biology can already suppon quantitative revisions in human health risk estimates. This analysis should also-highlight the most critical areas for new empirical research to improve quantirative risk estimation. BP-00016011 Deliverable from Task 3: A technical paper. suitable for submission to 11 peer-reviewed journal, on proposed quantitative revisions in regulatory risk estimates for benzene based on cwrent biological-data and on improved internal dose estimates. Estimated Resoldrce Requirements forTask 3: 130 hours. 13!t:, three months. The resource requirements for all three tasks total to 480 hours and $48,000, spread over a proposed eight-month interval. We suggest that an additional budget not to exceed $2,000 be allocated to cover travel and preparation expenses for a meeting at the end of the project, probably in Washington, D.C., to present and discuss the results. BP-00016012 ris!A ~s~s.-as a potential benefit. The purpose of this wk is to help malce that potential real by providing explicit methods for using biologic&llmowledge (represented, (;.g. by pmtial mOOels)-te guide applied risk-research. 'i{'~sl!! 5: J?_v<eptEI?tt Buhuaic~El !Ntpovt! (1}Jii8d worrl!!lttioOB4E !EE~Empl~ts (iO&clMriiliBg 4illi8J! li81tCitU(!,YJ "ir;ompMtltV' JPPDfYtlWU) UlMsRr~KiliBE 8/J81t llH'BCitY8tEini;yuWB&H'i4EflltMI11iil0 O<echnitJMitS tiiscv.ssed emd devdopced Bliil 1ras&s K,fJ. The deliverables from this task-will document the research and will provide examples (and. .where necessary. simple computer programs} to help illustrate and transfer the technology being developed here. As described above, each technical task- will result in a technical report; these will be revised, finalized, and submitted for publication at the-end of the project. All work outlined here will ~ perfonned on a ~st-effort basis by Cox Associates in a timely and cost-efie_ctive fashion. Dr. Cox will lead all technical work on behalf of Cox Associates and will be responsible for the technical content and quality of all deliverables. Dr. Paolo F. Ricci of the UCLA School of Public Health will act as designated alternate project manager should the need arise. BP-00016013 'Irll!& d.-_~ JI)e"6'elop eavu8 mustrate IMJetUatu8s foT? <lieri"6'ivag afficnrtlifJO f?&St&f9IT?e0a 5i0P~I1~Eict~- 06 flrtti&u:~ MlifJC<rT?O~BiVBgitEs &boMg ~risfr hy fi:OKl~Er:EBiiBif @&&itiomd afii18JPirlcfJU &&e~. This task will create methods for using partial knowledge of a carcinogen's mechanisms of action io gUide-the acquisition of new empirical data, taking into &ccouni the costs, durations, and reliabilities of different tests and experiments and their relative eltpected tocontributions reducing uncertainty about risk. We will use operations research techniques to study the problem of minimizing the resources (time and cost) required to achieve a given level of certainty about risk by achieving both greater understanding of causal mechanisms and greater confidence and precision in statistical estimates. The purely statistical issues (e.g., sample size and test selection) are well addressed in the literature on "expected value of information" in statistical decision theory. The contribution of this task will 1;(; to create analogous methods for quantifying the expected value of better knowledge of-specific causal mechanismsJn reducing uncertainty about health risks. The intent is to provide an analytic resource-management framework for mddressing questions ofresem;h strategy such as "Which of the following applied research options would~ most valuable for---reducing uncertainty about the human health ri-s!cs from occupational enposure to butadiene: (i) identification and isolation of the specific target cell populations gfi"ected; (ii) determination of whether the diepoxide metabolite, as well as the monoepoxide, is carcinogenic in B6C3Fl mice in vivo; (iii) more accurate measurement of detoxification :rates in monkeys as a function of exposure concentration and duration; or (iv) more accurate modeling of the dynamic response of the human hematopoietic system ao mitogens?" We do vwt propose to address the problem of evaluating competing basic research investigations that potentially provide benefits cutting across current &nd future applied research projects. Rnstead, we will focus on research strategies for efficiently reducing uncertainties about the human cancer risl<s from a specific chemical carcinogen. Finally, we will address the problem of selecting an applied research-portfolio when different rese&reh options carry different costs and durations as welll!.s uncertain-outcomes. Throughout-.-current research O}i)rions for benzene and butadiene will be used to_provide realistic examples of risk research decision problems. 'iask ~ is m<Aivated by a belief that one of the greatest potential benefits from- mebioJogicallybased risk modeling is ihe information ihat it provides about how reselution of current scientific uncertainties can affect quantitative rislt gstimates. Enthusiasts of biologically-based risk assessment often mention unceru.inty lreduction for BP-00016014 ,' The following sources providt;; excellent technical background on rome unconventional ~md llilOiH\i&tisdcal aechniques of uncertainty malysis that are potentially useful foY mpplied Gmtcei' rislc mnalysis. Etherington,,lD.W., Reasoning with Incomplete Information. Morgan Kaufmann, San Mateo, CA. 1988. (Requires some familiarity with contemporary mathematical logic.) Karp, l?.D., Hypothesis Formation and Qualitative Reasoning in Molec&dar Biology. Ph.D. thesis, Stanford University Department of Computer Science. Report No. STAN- CS-89-1263, Stanford Universily, June, 1989. Martins, J.P. imd S.C. Shapiro, "A model for ~liefrevision," Anificiallntelligence, 3~. 25-79, 1933. . Pearl, J., Probabilistic Reasoning inlnrelligent Systems: Networks of Plausible Inference. Morgan Kaufmann, San Mateo, California, 198ft Smets, P. et al (eds), Non-Standard Logics for Auromcued Reasoning, Academic Press, New York, 1988. (Requires some background in mathematicallog.ic Md probability.) BP-00016015 .. P.2 2. Otlbat API C~Xl~C!11t 3. Dste needs fo:- the !)rope~ ~- ~urure directions ror t.h~ projacw S. Meeting with EPA (6. STI:.LLA ti!IOml) Empirical ffl!Uivrmrm: O!;cupation&Jy ta)(poo00 WotllP..fll m U~ 1M IWr ~ CUi'it\nt l'illoile!.B ~t Iftwma! dore D AVC of ID@I800lil@ 00 ~ -rechn!ca/ approach: (I) Ellperimeii~ with "Q!Get?onlcp rruee, iQW, tm!ll hufilM!!. (2) ~@ ~\!Wiea m~. a This ft!i'Oj"l' will ll'M~f'o.r 10 API l\IOrlciilg n'llll..A moOOl!! f'011 J:A~, mnoo, DMII'lumn~iiG Model!i should ~ reflnec:i ru1 mOR ~ene da&a ~ome svlillab!e (eiip<iCi&Dy an ~~ illle~res) 3, Ezwu! P0 PK ~11F18 leC/mifl~S W ~iltW'e full!re b:a~ @Ill. Be prepnred to u!l9 f\mheomlna ~ from iUc:h Irons and other~ Improve ~wee}' ofin~ ~ pr~n obl!linOO by~-~ nw:delin~ ~. It~ G~Rr P'!tPir r&to f"r /;:gnf?&iif? wimin context of Bl!IR A. i"mrnewt!1!t .PrepNe m integme ~~PIC delt!l !lite! fflOfiel im.o ihe.-BBR.~ fmmewc:rfi 11eroul ~~ct CQlJCef ria!t. Mvt;mQe r11e BBR~ fulmewarn BP-00016016 F.. --= ~i'OBI"&M 06 bli' ~/~(1: la:aalllle-~I'IDf.'l 8(Qio 1. ~t!:U4>fo~-m." PB-f?K modem Mlve ~ IMh (O'l ~. ffit!J, &m RllMDtm!l. '!Jt.:go ~ (TI'tlvi&) ~- IL<D~ ~to~~ if~. ~ity ~ mmt is 00~/S Final SiElJ.A versions to b2 delivered 10 API 011 dis!ts AJ>I should piM to pu."Cilase S'i'Ell.A ~!ely. l.hniration5 or 9tate-of-!he&ft motle~~ ~ All melli.OO!i~ @Te ~~. w:1> Diffu.!lion-limited flol':l iB~ :t> Ignore..s intef'lllctions wir.h iMITOW macta;l)Mge!l. <Mer Met.ahnliM:s, e(IC. :r> Key me.tabolic par-arne~ vl!l!ucs ere estimaled, not m~ :~> Jii'rad.ic:.ions mre not !!lw!lyc ~. ~I iml;)licc.tiros fO? low-h~~we i!l.leiilml r:bes fJJld m.!!rliMOOI: cam~~ ~B ~e!~ a. CU~nlntly ell~rimenLina witb Bimple models ofhooe iiW'i'OW memlmli9ffL o Multiple ~fllholitoff llfld !Mif lnto~ticMta tro l:ii::cl!:(')d Xnlemetioi'lS with macrophc;et not y~ ~100. ~ md tailed c-ell kineticO/t:ywtoxiciry mo&!lm n~ w~Wfi ~Q'ill iW1!123. 3. ~themll.tU:al fmmcworfr 1'1&3 OOzft suageilte!l to improve ~ ~llmcy cf ~tuooe Jil'~-PI'~ i>'tttdellilg. oota!orne ne\!:f requ~rs creti~IOO for (i) Diffusioolimi~ Oowstln!'i (ii)..J!..iel!llmli!: li'Ores i'oi ~ and oiks:r metslrolileS. ~. Simpls, MI!TfB&te rnotlels tMI wAISfy cUfiei!l <!!Ita ctwll'!'!iilt.s m SliD ~fl~ &:'!!@~. TOOory is bein; devel~ under WSPA uncc:wjnry reaeM:b ~mj<;ct. G:nt!!l tl/91) BP-00016017 '. P.4 M~Poltat: ~~ o!~'I@~~~~!Ml~c~iEnp~tilu@::s~~n ~ib~~~- ~~ I. Some qWlllttltive 8Jr8Uments have been documenti.'4 more thorolllJhl)' and l'lclu~~ b1 the JUs~ AM/ysia remew ~- 2. Aile mor:tconvinclnt quantimtive w-gumenu probAbly roquire ~nintt ind/vidwz! IYilmal fi/RWP tM!ta. ~ o rgumenrs ha'e y~ 10 ~ I"M<te .Most datil to ~uppo:rt irrelevAAce of dflta t:ll expoSW"CS ovc SO ppm for preciictift~ ~bgLOW i.O Wffi ano cifcumlltafttlal, b-.11 ~ PB-'K <1rus &<! !Wo..imble. 3. Oenaralwguments !'!nd strong evidence fm nonline.'uilloa wfJI oo presented mt the S~A mooti:tg lil Octoh<:!r. S!!lturat.!oil, ()vetwblmins. ;nd &plet.ioo ofintrn:ellul& @!l;Zymc< gy~!9m~; VIOlllin<m ~lcin@~, aie. 1. Met!!lhtllic l'!llie cooatants for llvf1! Dom YOr amlemng OfW!IIIJ" mM~ '" sl ~ C8l6Chol p1'Clduction md c:onv~ rof!!lS AlifcbrciG femur ol inunction~ b~ on ~me l!inel!ieu Hydf'Cquhtclle to ~uincm ln~~io:rllfi.:!800 "Cruechol to o-bmzCJl!uino:i!e C!!I'IYGMon rote iR W1lc Soole GOnllfMit miuht hg ~maiOO from l!ver ~if rt&dos ofsprsctftr: ~fll1ii'l2 tiJctMi)' i~els t:~ere !mG'WU. (BYt myel~tr(i~ ~::te.m i.!! tmique.) What MpjitOO 10 ~Wt.<me? 3. ~~ts Plrctein binding re:oo Bnd pa"liti:m cnaffic:lents ,. Compt'te&ivt' Wid other ir.terliCUOilll Iii vivo Metib:oliUJ ~ticm ccaMdents ~lla"-CMS10nts for ~e tp'C)IJp cl~'lte BP-00016018  P.5 2. l'luild il'litbt.J detailed COmf'l!MI~W now model of bone mmow inetabolism ~d tolt.ld)'. IHClude emerginz quantitati\e Mu!, qualirat!\'e unde.rmnding, Md cnzym~ kineries :alcuW!:.ic.mJ 3. Develop/liSt sf.IPiufati~fl te.-h.'if ;so:s u3ing modern ~.ompanrnemlll tlow met."loel!. &aer approll.!moticm th:am}l Suppor~ ncmllnearlty 11rgt!ments m.uggem mles fer dl!iei'Mt met&bo!ite4, 1'1\!lSh!!!n.i.mls, ~d .Appr~h: 'i'ime-10-!Wi)Qf ooclyais (or muldple l!!MO!' gi@OO. 2. lii}uiJd initial version or cytm.oxiciry W1d Mlzene myeboltlcity model. PH!('AT/HQ!rM:rrnpMge dynamlc~ DWiggY'!'p MOdel of hewuopoie6is Ineericce !0 PB-PI( mm~t 3. Uevelop ~.,s~ie rum!!hili.111 model fnr cell kinericr;, C)'tme:t-!!:iry, mtl CM~ ~!!ltioos. Ma!Jiams!fr:&.l !ramewom deve!~ In 'WSRtA )'ti'Oject. "Soci!izinK:MM; l. ~wis PYbliahe~C' imeregred in t:n~oolr t)ij DBRA ~ on Sltll. A!xm cowss. ~Olltlh oolid m11teri8! far 11 good memo~ or te~:.t Key~ fOT malting BBRA more ~rul is to iniep-are It-with &liB am.l)'ai6 metllods3~ l:mcue&iom wilb EPA (?) BP-00016019