Chrysotile Asbestos Consensus Statement and

1 34 Somelimitationsofmeta-analysesinoccupati onalasbestosstudies~ , 'stypicalchrysotileasbestosexposureinCan ada, ,suchascredibilityrangesor,wherepossible , "TypicalexposuresinCanada" (byphasecontrastopticalmicroscopy), (bytransmissionelectronmicroscopy)."Trem olite"meansfibrousformsofmineralsinthetr emolite-ferroactinoliteseries."Amphibole s"meanstremolite, , "tremolite-freechrysotile"whereitisneces sarytodistinguishchrysotilewithouttremol ite, ,participantswereinstructedtoconsidertre molite-freechrysotile, ,whenconsideringtheroleoftremoliteinexpo surerisks, , ,Quebec,Canada, , , , , , , ,byBermanandCrump!,referredtohereasB&C,w ascommissionedbytheUSIB ennanOW, :Technicalsupportdocumentforaprotocoltoa ssessasbestos-relatedrisk; , # , ,byHodgson&Darnton2,referredtohereasH&D, isbytwomembersofstaffoftheBritishHealtha ndSafetyExecutive, ,some60relevant,mostlyrecent,papersweres ubmittedbyanddistributedtomembersofthepa nel, :(1)amphiboles,whichincludecrocidolitean damosite;and(2) , , ,surfacechemistry,andbiopersistence(whic hmodulatescumulativeexposure) , , , ,whichisanamphibole, ,DamtonAI, (2000).

2 (8) ,andthisissupportedbyStayneretaI's(2007) ,theconventionalopticalmicroscopemethodo fmeasuringairborneasbestosgivesequalweig httoallfibreslongerthan5micrometres,does notcountfibresshorterthanthis, ,andmayhaveadifferentrelationwithriskind ifferentstudies, ,participantsreferredoftentotheH&DandB&C reviews, ,orinterprets, ,participantssaidthatthetwostudiesaregen erallyinagreementwitheachother:bothrepre sentthesituationfairlyaccurately, ,andthediscussioncouldonlyreflectwhatisa vailablenow,andmainlydependedonworkavail abletoB&CandH& , ,sothatH&~sriskestimatesassigneffectsto" Chrysotile "whichmaybepartiallyduetothetr emolitecomponent. ,whereasH& ( )anddifferentexposurehistoriesofdifferen tcohorts,butwasmore3 StaynerLT,KuempelE,GilbertS,HeinM, ,on-lineadvancepublication,Dee2007, 'I' ,becausetheycouldincludesomewithlessdeta iledexposure-effectdata. ,butH&Dconsideredcrocidoliteandamositese parately.(H&Dfoundcrocidolitetogivemorem esotheliomathanamosite,withnoappreciable differenceinlungcancer.)

3 ,andH&Dconsideredthemseparately.(H&Dfoun ddifferentrelationshipsbetweenthetwofora mositeandcrocidolite.) ,andH& ,andwhereH&D'smodelpredictedgreaterriskf ormesotheliomaandsmallerriskforlungcance r. ;H& , ,suchas:knowledgeofpotentialcofactorstha tcausediseasesofinterestoftheirown( );effectmodifiersthatmultiplyorattenuate theeffectoftheexposuresofinterest( );thecapacitytoidentifyoutcomesclearly( ,especiallyinstudiesbasedondeathcertific ates);and,particularlyforobservationalhu manstudies,thecapacitytocharacterizeandq uantifyexposuresadequately(seenextsub-se ction). ,butoftencannotspecifysoclearlytheerrors intheestimateofexposure, , ,evennon-differentialexposuremisclassifi cationusuallyleadstounderestimationofthe truemagnitudeofexposure-effectrelationsh ips,andcandistorttheshapeofarelationship , ,fromfibretypetofibretype, ,ratherthanpersonalmonitoring,andthisusu allyunderestimatespersonalexposureandcan missshortperiodsofhighexposure. , ,anuncertainprocess. ,andthedegreetowhichexposuresatunmeasure doperationswereestimated.

4 ,andthisparticularlyaffectsmeasurementsb eforemodemqualitychecksbecamecommoninabo ut1980. ,averyuncertainprocesswhoseeffectwillhav evariedfromenvironmenttoenvironment. ,anditispossiblethatinsomeparticularlydu styenvironmentsasignificantamountofmater ialcountedwasnotasbestos, ,andthismaynotbeapparentinthemeasurement s. , , 'accesifIinformation. ,thefibresizerangecountedisoftennotident icalwiththesizerangeofpathogenicfibres. ,particularlywhensmallnumbersoffibresare counted,butmostdatapointsinanepidemiolog icalstudypoolmanysamples,whichwillreduce therandomerrors. , , , (SouthCarolina)textileworkerstherewassub stantiallymorelungcancerriskperunitofexp osurethanintheotherstudies,andinQuebecmi nersandmillerstherewasless, (TherecentreanalysisoftheCharlestondatab yStayneretaI,referredtoabove,hasshowntha tcorrelationbetweenexposureanddiseaseisi mprovedbyusingadifferentfibresizefractio n, ,andshouldimproveagreementofthedifferent studiesandnarrowtheconfidencelimitsofthe overallrelationship.)

5 "glossover" ,participantsaftermuchdiscussionagreedto offerHealthCanadaageneralapproachusingat hefollowingnarrativestatementDocumentRel easedUndertheAccesstoInformationAct/Docu mentdlvulgueenvertudelaLOIsurI' , )and2) ) ) ) ) (usingconventionalphasecontrastopticalmi croscopy) ( ),butH&Destimatednon-linearrelationships ~.Thepanelpreferredthelinearapproachonth eprinciplethatalinearmodelistobepreferre dunlessdataareavailablewhich'reliablydem onstratenon-linearity.(However,inB&Cmeso theliomariskislinearlyrelatedtoexposurei ntensity,andagivencumulativeexposureresu ltingfromanexposureoflongdurationhassome whatlesseffectthanthesamecumulativeexpos ureresultingfromaveryshortexposure.) , ,wheretheseareavailable, ! ' (anamphibole) ,B&Cfoundthat,forresultsmeasuredwiththen ormalopticalmethod, (ietheamphibole:chrysotilepotencyratiowa sabout2:1),butthehypothesisthatthetwofor msareequallypotentcouldnotberejected(p= ).Thehypothesisthattremolite-&eechrysoti lehadzeropotencywasrejected(p= ).Ifdifferentsizerangesareevaluated,thes econclusionsareunchanged, ' ,theirbestestimatesofriskforthedifferent typesyieldanamphibole:chrysotileratioofa bout40:1,withapossiblerangeof1 ,B&C'smethodallowsdifferentcohortstohave differentbackgroundlungcancerrates,forex ampleduetodifferentsmokinghabits,sothatB &C' ,asnotedabove,H&Ddidnotattempttoremoveth eeffectoftremolite,sotheirestimatesarefo rchrysotileasused, ( ), , (amphibole:tremolite-freechrysotile=300: 1),andthehypothesisthatthetwoformsareequ allypotentwasrejected(p= ).

6 Thehypothesisthattremolite-freechrysotil ehadzeropotencycouldnotberejected(p= ).Ifdifferentsizerangesareevaluated,thes econclusionsareunchanged, "bestestimates"at1 :chrysotileratioof130:1,andamosite:chrys otile18:1,buttheuncertaintyoftheestimate sleadstoaverywidepossiblerangeofratios.( Also,asalreadynoted,H&Ddidnotattempttore movetheeffectoftremolite.)DocumentReleas edUndertheAccesstoInformationAct:Documen tdlvulgueenvertl) , , (B&C),orthe"bestestimates"(H&D).Thefigur esflankingthemarecalculatedconfidencelim its(B&C),or"cautious"or"arguable"figures (H&D).Forlungcancer,H&Dalsogiveahigher"e xceptional"figure(notshown) "riskestimatesforchrysotileincludeanycon tributiontoriskbyfibroustremolitecontami nation( ).Thecumulativeexposurelevelsinthetablea reforoccupationalexposure(assumedtobe8ho ursperday,240daysperyear).Thetypicaloccu pationalexposurestatedtothepanel( )wouldgiveacumulativeexposureof10 ( ) ,equivalenttoalowervaluebyPCOM.)Cumulati veexposureLungCancerMesothelioma(occupat ionalexposurepattern:Deathsper100,000exp osedDeathsper100,000exposed8hr/day,240da ys/yr)B& :Possibly0 VeryH&D:Probably< < :Possibly0 Probably<1H&D:Probably<1"Highest"Cautiou sestimate3"arguable4"B& :Possibly0230H&D:1520B&C:2883210B&C01082 10 :(nofiguregiven)50300H&D:62060 Noteoneffectofsmokingonthesefigures(byDr KCrump).

7 ,butnotmesothelioma, , , , ' , ,itwassuggestedthattocomeupwith"average" ,notingthatinthecaseofmesotheliomarisks, forexample, ,panellists'"bestestimates" , ,onepanellistgavea10%chance,onegaveita36 %chance,andonegaveitan80% ' , , ,theirbestestimatesconcurredaroundanappr oximateSOD-folddifferencebetweenamphibol esandchrysotile,witha95% , ,foreachclassofasbestos,thebest-estimate predictionsofH&DandB&C'smodelswerewithin afactoroften, , , ,suchaschrysotileandlungcancerorchrysoti leandmesothelioma, " "IdonotbelievethateithertheHodsonandDarn ton(2000)ortheBermanandCrump(2008) , (2008a)thehypothesisthatthepotencyforlun gcancerofchrysotilewasdifferentthanthepo tencyofamphiboleswasstronglyrejectedinal loftheirmodels(p= ).InHodgsonandDarnton(2000)theaveragelun gcancerslopeforchrysotilewassimilartothe slopeforamphiboleswhenthestudiesofQuebec minerswereexcluded, (BermanandCrump1995),andofpreviousquanti tativeandqualitativeanalysesoftheepidemi ologicdata( , ). , (2008b) , [RothmanandGreenland1998].

8 ,theheterogeneityinthiscaseappearstobela rgely(butnotentirely)duetothelargediffer enceinslopesforlungcancerriskthathavebee nderivedfromthestudiesofchrysotileexpose dminersandmillersinQuebec[ ],andthetextileworkersinSouthCarolina[ ].However, !easedUndertheAccesstoInformationAct!Doc umentdlvulgueenvertudelaLOIsurlaccesa! [1997]reportedthattheheterogeneityinthef indingsforlungcancerwasexplainedbydiffer encerelatedtoindustrytype,methodsusedfor measuringasbestos, , ,Iamconcernedthatinorderforthemeta-analy sesto"representthesituationfairlyaccurat elyupdated" [ ],whichwouldbehighlyrelevantparticularly forthefibersizespecificapproachthatwasus edbyBermanandCrump[2008a].ReferencesBerm anDW, , ,CrumpKS,ChatfieldEJ,DavisJM, ,shapes, ;15(2) ,StaynerLT, ;64(9) , , , , ,2000 LashTL,CrouchEAC, ;54(4) , ,McDonald, , (1997)The1891-1920 ;41 , ,Philadelphia,1998 , ;86(2) ,KuempelE,GilbertS,HeinM, ;[Epubaheadofprint] ,(r(rl/" , :Thestatementismadethat"tremoliteisanamp hibolewhichisnotdeliberatelymined,butoft enoccursincloseassociationwithchrysotile .))

9 " ,magnesium,silicon, ,magnesium,silicon,oxygenandhydrogen-sod ium,ferrousiron,ferriciron,silicon, (OH)2whilethesilicatecomponentofthechrys otileisShOs.(OH)4 Thestatementismadethat:" ' ,butdonotoverlapwiththosefortheCharlesto ncohort(SeeFig1B&Cpaper2). (Figure3)includesmorecohortsthanB& ' ,thestatementismadethat:"Thisshouldclari fyhowtheoverallrelationshipshouldbeappli edtoparticulargroupsofworkers, ".WhileIamoptimisticthatthismayimproveag reementbetweenstudies, ,Idobelievethatexaminingthelongfibers(>4 0micrometers) (theQuebecminingandtheCharlestontextile) , , , ' ,Liddelletal(AnnalsofOccupHyg41;13-36199 7) (estimatedtobeabout900f/cc-years). , :"Thesefindingsindicatethatexceptatveryh ighlevelsofexposure,severalordersofmagni tudehigherthananypresentlypermitted,adve rseeffectsonhealthwillnotoccur".Itseemsr easonabletoconcludethatwhentherearenoexp osure-relatedincreasesinriskatlevelsorde rsofmagnitudeabovelevelsofconcerntoday, :"Alsomanyworkerswereexposedtoasbestosot herthanthemaintypeusedintheirjob".

10 ,fromlungburdenstudiesthatsomeworkerswer eexposedtofibretypesotherthanthosewithwh ichtheyworked,butwhetherthisis"many" :" (SouthCarolina)textileworkerstherewassub stantiallymorelungcancerriskperunitofexp osurethanintheotherstudiesandinQuebecmin ersandmillerstherewasless,althoughtheQue becstudyiswithintheconfidencelimitsofthe overallrelationship."DocumentReleasedUnd ertheAccesstoInformatlollAct/Documentdiv ul\JueenvertudelaLeisurIaccesaIInformati onBasedonourcurrentknowledge, ,inthefuture,weareabletofindtheparameter /exposure/factorthatisresponsibleforthed ifference,onpresentevidencetheCharleston cohort,inmyopinionmorethan"appears" ,2008.