WIXLIB

38JADA, Vol. 122, April 1991

IS THERE A LINK BETWEENWATERAND OSTEOSARCOMA?To test the hypothesis thatfluoride is a risk factor forosteosarcoma, a case controlstudy compared thecomplete residentialfluoride histories o fosteosarcoma patients withmatched hospital-basedcontrols. Fluoridation wasnot found to be a risk factorfor osteosarcom a in thestudy population. The trendin the data from this smallsample study suggests thehypothesis that a protectiveeffect may exist against theformation o f osteosarcomafor individuals consumingfluoridated water.S H E IL A M . M C G U IR E , D.D .S.,M .S.E LE A N O R D. V A N A B L E , ED.D.M IC H A E L H. M C G U IR E , M .D .J O S E P H A. B U C K W A L T E R , M .D.CHESTER W . DO UGLASS,D .M .D ., PH .D .Q he “equivocal evidence” of a link between fluoride ingestion and theformation of osteosarcom a in a small percentage of male rats thatreceived high doses of fluoridated w ater (100 and 175 ppm NaF) wasreported in 1990.1Inferences on hum an risk of fluoride carcinogenicityare difficult a t best to obtain from animal studies.Our study reports findings from a hum an population-based analyticalstudy of osteosarcom a patients and matched controls. The relationship ofsystemic fluoride exposure from the ingestion of w ater from residentialsources to the occurrence of osteosarcoma was examined.An incorrect inference im plicating systemic fluoride carcinogenicityand its removal from our w ater systems would be detrim ental to the oralhealth of m ost Americans, particularly those who cannot afford to pay forincreasingly expensive restorative dental care.Osteosarcoma is a rare and painful prim ary m alignant bone tum ormost commonly occurring in children and young adults. Data2indicatethat tum ors of the bones and joints occur in less than one person in100 , 0 0 0 .The American Cancer Society estim ated there w ere 2,100 new cases ofbone sarcom as in 1989.3According to Maeyama,4osteosarcom a incidencevaried with age; 35.7 percent of all the m alignant bone tum ors occurringin the first decade of life were osteosarcomas, rising to a peak of 69.6percent in the second decade and decreasing thereafter to 24.5 percentand 4.3 percent respectively in the third and fourth decades. Malepatients predom inated, with a ratio of three males to two females.Radiation,57viral action,8-9genetic factors10and other exogenous factorshave been implicated in the development of osteosarcom as.11Radiationinduction of osteosarcom a is at present an accepted principle. A recentreview of the epidemiology of bone sarcomas, however, reportedconflicting study results for all other etiologic agents. Fluoride ingestionwas not linked to osteosarcomas in the m ost recent review.11Hoover and others at the National Cancer Institute com paredstandardized bone cancer m ortality rates in fluoridated and nonJADA, Vol. 122, April 199139

fluoridated counties in the UnitedStates.12No change in bone cancerm ortality took place over the 15year comparison. In addition,standardized overall cancerm ortality rates have beencompared for fluoridated and non fluoridated communities.13'14Noneof the studies uncovered asignificant increase in hum ancancer incidence or m ortality as aresult of the im plem entation ofcom m unity fluoridation.According to the CongressionalCommittee on Intergovernm entalRelations and Human Resources,questions lingered about therelationship between fluorideingestion and osteosarcomas.The National ToxicologyProgram was directed to conductan animal toxicity study to test thecarcinogenicity of the fluoride ion.The NTP concluded and the Boardof Scientific Counselors’ TechnicalReports Review Panel concurredth at its two-year studies showedonly “equivocal evidence ofcarcinogenic activity” in male ratsand “no evidence of carcinogenicactivity in female rats or in male orfemale mice.”1(Equivocal evidenceis a category for uncertain findingsdem onstrated by studies that areinterpreted as showing a marginalincrease of neoplasms that may bechemically related.)Findings on the molecularexplanation for the initiation of anosteosarcoma tum or from fluorideexposure could not be found in aTABLE 1F L U O R ID E E X P O S U R E A T OR A B O V E R E C O M M E N D E DO D C LE V E L SControlsMore than 1/3 ofLess than 1/3 oflife at 0.7 ppmlife at 0.7 ppmCasesMore than 1/3 oflife at 0.7 ppmLess than 1/3 oflife at 0.7 ppmO R R R 0.14Z c2“ 3 1D E R IV A T IO N O F A V E R A G E L IF E T IM E F L U O R ID EE X P O S U R E (IN P P M )Example:Case(11 years x 0.8 ppm) (7 years x 1.0 ppm) .87 ppm18Control(15 years x 0.3 ppm) (3 years x 1.0 ppm) .42 ppm18JADA, Vol. 122, April 19911739 5 % C l 0 .0 2 , 1.22refereed journal. However, them echanism of fluoride’s action tostrengthen osteal tissue is relatedto its ionic capabilities. The majorcom ponent of osteal tissue and themajor strengthening material inteeth is structural hydroxyapatite,Ca5(P04)30H.When ionic fluoride enters thehydroxyapatite lattice, a dynamicexchange of F for OH occurs inportions of the structure. Theresulting fluorapatite tightly bindsthe fluoride, strengthening thestructure and rendering it lesssusceptible to dissolution inorganic acids.15Strengthening canoccur either throughincorporation of F ions during thetissue m ineralization phase oftooth or bone development,16orexchange of ions after formationhas occurred.17Because of its strengtheningaction, fluoride has been widelyF IG U R E 14011accepted as the responsible agentfor dram atic declines in the toothdecay rates of U.S. children andadolescents.1819Studies of matchedpairs of fluoridated and non fluoridated communitiesestablished the principle thatdecay rates could be reduced by upto 70 percent through the use offluoridated community water.20Current reduction rates areestim ated to be between 20percent to 30 percent.21Adisruption in the delivery offluoride through municipal watersystems would increase decayrates over time.In one study, fluoride wasremoved from a com m unity watersupply after 11 years offluoridation. The DMFS scoreswere recorded for second-gradechildren at the time ofdefluoridation and second-gradechildren six years later. Sixtyseven percent more children at thetime of defluoridation were cariesfree w hen compared with thesecond graders six years later.22Linkage of fluoride ingestionand cancer initiation could resultin a large-scale defluoridation ofm unicipal w ater systems underthe Delaney clause. (One aspect ofthis Environm ental ProtectionAgency clause prevents theaddition of carcinogenic agents to

the nation’s w ater supply.) To testthe hypothesis th at fluoride is arisk factor for osteosarcoma, thiscase control study compared thecomplete residential fluoridehistories of osteosarcom a patientswith m atched hospital-basedcontrols of similar age, gender andcounty of residence.METHODSOsteosarcoma patients wereidentified from the University ofIowa Cancer Registry and themedical records of the Division ofOrthopedics, St. Josephs Hospitalin Omaha, Neb. Each sourcerecorded birth date, the age atdiagnosis, gender and county ofresidence at diagnosis for eachcase. Eligible cases w ere patientsdiagnosed between 1980 to 1990and younger than 40 at diagnosis.Non-radiation inducedosteosarcom a occurspredom inantly in the first fourdecades of life. Osteosarcomapatients with any prediagnosishistory of the known risk factor,radiation therapy, were excluded.In addition, patients with a historyof kidney dialysis were excluded asthey choose to drink deionizedw ater for medical reasons.The University of Iowa and St.Josephs hospitals identified 44eligible cases. Correct addressescould not be ascertained for sevencases; two patients had receivedpre-diagnosis radiation therapyand one medical record could notbe found. Contact with theTABLE 2A V E R A G E L IF E T IM E E X P O S U R EControlsHigh 0.7 ppmCasesHigh 0.7 ppmLow¿0.7 ppm0 R RR Q.33Low¿0.7 ppm9265X c 11295% Cl 0.04,2.50rem aining 34 eligible patients wasinitially m ade by mail with a letterof explanation of the researchgoals of the study from theappropriate orthopedicdepartm ent chairman.Hospital-based controls frompatients of the appropriateorthopedic departm ent werem atched by age, gender andcounty of residence at time ofdiagnosis. The Midwest states—Iowa, Illinois and Nebraska, inparticular—are geographicallydivided into m any small counties,for example, 99 in Iowa. Mostcounties are rural w ith relativelyfew dom inated by an urbanindustrial area. The advantage ofmatching on the county ofresidence variable includes greaterlikelihood of socio-economic statussimilarity and absence of biasrelated to the complex referralprocess that precedeshospitalization.23Following a localcustom, residents of Iowa andNebraska typically identifythemselves as a resident of aparticular county.Cases and controls wereexcluded with identical criteria:any history of radiation therapyand patients with a history ofkidney dialysis. The factor underinvestigation—ingestion ofresidential fluoridated w ater—isJADA, Vol. 122, April 199141

TABLE 3C H IL D H O O D E X P O S U R EControlsMore than 1/3 ofchildhood at 0.7 ppmCasesMore than 1/3 ofchildhood at 0.7 ppmLess than 1/3 ofchildhood at 0.7 ppmOR RR 0.33Xc2 u 29RESULTS6JADA, Vol. 122, April 199159 5 * 0 1 0.04,2.50not associated w ith any othermedical condition or risk factor.Therefore, no other categories ofdisease were deliberately includedor excluded in the hospital controlgroup.As each osteosarcom a case wasto be m atched to a control patient,a 1:1 ratio of cases to controls wasrequired. Initially, an attem pt wasm ade to contact two controls foreach of the 34 cases with a letterexplaining th e study’s researchgoals from the appropriateorthopedic departm ent chairman.Because there was no age andcounty of residence m atchedcontrol for seven of the cases, 27pairs of cases and controls wereenrolled.The 27 cases were interviewedby telephone forthe followinginformation: allresidentialaddresses of theosteosarcomapatient beforeDr. Sheila M cG uire isdiagnosis;an Oralyear(s) lived atEpidemiology[ Fellow , Harvardeach address;School of D en talaccess toM edicine, 188municipal orLong wood A ve.,Boston, 0 2 1 1 5 .well w ater atAddress req uests foreach address;[ reprints to Dr. M cG uire.and bottled42Less than 1/3 ofchildhood at 0.7 ppmwere provided by mail to theparticipants with access to wellwater. The participants wererequested to retu rn a well watersample for analysis of fluoridecontent by the University of IowaHygienics Laboratory.water use. The same procedurewas followed to ascertainresidential histories from them atched controls.Fluoride levels (in ppm ) formunicipalities supplying w ater tothe cases andcontrols wereascertained fromthe m ost recentCenters forDisease ControlFluoridationCensus in theyears identified.If a case orcontrol lived in am unicipalitybeforeadjustm ent offluoride levelslisted in thecensus or in am unicipality notlisted in thecensus, theappropriatestate’sD epartm ent ofNaturalResources wascontacted for thecorrect pre adjusted level offluoride. Watersample tubesContact by telephone wassuccessful for all 27 osteosarcom acases and their controls. One caserefused to participate and fourcases could not supply sam ples ofwell w ater from previousaddresses. Therefore, the analysiswas conducted using the fluorideexposure inform ation gathered on22 m atched-pair cases andcontrols.Thirteen pairs were males andnine were females, consistent withthe national figures. Additionally,