..... . ...... .. .... .I —%Ai-- ~,..-..J.L+ ;:.+’:., -,,: ;, -.,., ,,. . .:...,. .,+ ,::;, ~“ ~~~‚¸÷•‚..), - ., . ,

!.. ,- . . .,

,. . ,, ., - . .

. . . .T ~!~#. k. ~ ~. ; .!-..,** --<?. >:%..-.-:7, ..+, ~’ + - - ..,---,. -., * :..-.. : ~ .--=. W. .+---.-.7 - +.. . . ... . . . . . ,. .

-w ‘ + :. ., “ .4’ :“ “’ . -. . . ,.

~ ‘ -Tt---l ------ .:- ..,- , , , ~ L, .- ~ . .

nla for the United Stat& be’pa~ent of Energy under contract

,. .-

.,—-- - .. . .. .. .. . _ !

. .W-7405 -ENG.36.; ..,==. “ . .. ... :?,, : . . —.

--— . . . .. . . ....,. –,.>.. , ,..,, ,.,, ,.

. .- . .... ._.__&.-+--...& A-+i+.4u.+..4

,.. -! .

.—.-.

.... ... .. .. ;.

For ReferenceNot to be taken fromthisroom

I

~,.,,.

;.-.1

‘yak+-:’?.. . .. rims!LosAlamosNationalLaboratory.. LosAlamos,NewMexico87545,,,,..+...,, .,.,.., ,,.--, . : ..........,, ++::;::: ‘;:---+=++“i’”:.”‘; G‘--;”-“”:“w’”“,:“ ‘- ““‘ ‘-‘-”‘““’---”.“:“‘-”,.”....., 1\...... ?.. ...,,,..;:.,,.”,..--..,..:,.,/ .:,. :- ., ,, ., ,.,,-.!..-,W.. -,,,.:..-.i ..,,. -<.,.,7,......!: .“..,’:,“.,JL.f:,, ....s,-.,-*... ;-i+iq.,,?*,?.,;?,-;>”:.r*.&,.w.:>(..-.:!,- ~;. 2.,. T.J-,?,?:; : ;7;‘-.: ‘K.., “YAJ::,::;;,:~:~::’’”;.,..:;!:::’:”::’-{i‘.:’”.:’,.;:,!;‘“.”.:’:‘“ -“.::,“ ‘- . ;“ “ ‘::+..7...,,:$%:.<}!$;’l:.......‘:~,-,.,>:,‘,..?, : .. .-.,...’,-.....1 . ...:. ., ,.-.

- ..,—., w

‘.f 1.. ., ~ ,,.. ‘, ,,..’‘., ,:.

,., AnAthraatitiAcrios@ualoppc+sity ~pl+r

.’,. J,

. .

.. -, .,. . - . !, . .: .,

. . .

,.;.

.>

,.

,,

.,. . .,/. .

‘.,. -, . -,:”“’ .. f

\

. .,..- \ ‘,

,. . .,. .. -.

? ,“ “~ & .- ,. ;

- . ., .“ ;

., .’

,

,.. .

. .. . ...

., ..4 +. :

,-

. .

..,, . ,,. .

. ., .,... ,.“.).

.>.

.“.- ‘t

.,. !

,. ,.,,

..‘.

, ... . , ), ~~•ˆp)••¤,••à,•••4••˜D•ˆ

., .

.

.. ~. . . . . . . . ., . . . .- ,., , . “ ,, , . : . ,

Thisrep& MSpieparedasanaccountof worksponsored”byb“agerrtyof the UnitedSt.?atesCoverrrrnent.!. ~-crsn%ther the United!jra~esCover~ent nor+rryagencythereof,norsrry,of theiremployees,makesany

.’ -“~warranty-,expressor”iMpfied,ora.wmes any Iegaffiability or resportkibtiety?orItk acciaricy,mmpleteness,or rasefuktessofanyinformation,apparatus,product,or pro&s disclo”wd,or ;eprescrrtithat itsusewouldMl Infringeprivatelyownedrights.Referencehereinto anyspecill commercialproduet,process,orserticeby traderqme,tti.dqrrar~l~nuracltier, or otheryk, dog not WXX@y rn~itu~e or~~PfY itsendorserrient,recornmendaiiori,or faioringtiy~heUm’ted$ratesGov&r6eriior anyagericytheieof. TheviewsandopinionsofaurhorsexpieKed hereindonornecessarilystareor reflect thoseof the UrwedSutfi ~wwnt or any’agerrcy~hered.- “- “L “ “- “ ~

.. ., F

,. ...

LA-98(24-MS

UC-41Issued: June 1983

.

Case-Control Study of Brain Tumors AmongWhite Males Employed at the Rocky Flats Plant

●Rocky Flats Plant, P. O. Box464, Golden, CO 80401.

LOSI!NI~m~~Lo..lam..,NMeMi.87~Q~~Q~LosAlamos NationalLaboratory

.

CASE-CONTROL STUDY OF BRAIN TUMORS AMONGWHITE MALES EMPLOYED AT THE ROCKY FLATS PLANT

by

M, Reyes, G. Wilkinson, G. Tietjen,G. Voelz, J. Acquavella, and R. Bistline

ABSTRACT

We conducted a case-control study of 16 primary brain tumor deaths occurringamong white males employed at the Rocky Flats Plant during 1952-1980 to investigatethe relationship between these tumor cases and occupational radiation/nonradiationexposures. Three mutually exclusive control groups were selected from the white maleemployees: matched (individually matched on age and period of employment),deceased, and random. Each group consisted of a 4:1 control-case ratio. Analyses forexposure to internally deposited plutonium and external penetrating radiation werebased on health physics dosimetry measurements. Potential exposure to nonradiationhazards was analyzed by job title and work area. Because evidence of exposure toberyllium and uranium existed in only one case, we did not analyze for these exposures.Our study showed no statistically significant association between exposure to internalradiation and brain tumor. We found no statistically significant differences in externalradiation exposure between the tumor cases and the three control groups. In addition,no nonradiation occupational risk factors were identified.

.—————————————————

1. INTRODUCTION

Recently, two independent studies found an excess ofbrain tumors (BTs) among workers at the Rocky Flats(RF) Plant located in Jefferson County, Colorado. ’-’ Inthe RF cohort mortality study of white males,2$3astatistically significant excess of observed versus ex-pected deaths occurred only in the benign and un-specified neoplasms category [8th Revision InternationalClassification of Disease (ICD) codes 210-239]. Ex-amination of the death certificates for individuals in thiscategory revealed that these neoplasms were BTs. Toinvestigate this finding, we conducted a case-control

study of the relationship between BTs and occupationalradiation/nonradiation exposures, nested within the RFwhite male cohort.2’3

Since RF opened in 1952, employees at this plant havefabricated nuclear weapon components containing pluto-nium, beryllium, and uranium. The plant is geo-graphically divided into four major work areas. Area Ais primarily used for casting and machining depleteduranium. In 1958, beryllium and stainless steel castingand machining became an additional part of operationsin the area. Other minor processes in Area A includeincineration disposal of machining chips, stainless steelwelding, mold coating, and graphite machining. In area

1

B, major activities include casting and machining depleteduranium, as well as rolling and pressing beryllium,depleted uranium, and until 1965 enriched uranium. Inthese two areas, workers may be exposed to radiation,decreasing agents, and solvents. Most plutonium-relatedwork takes place in Area C and includes fabrication,casting, machining, inspection and assembly of pluto-nium weapon components, and recovery and processingof plutonium waste. Until the 1970s, the final assemblyof all weapon components took place in Area D. Inrecent years, the activities in this area have includednondestructive testing research and the storage of plantmaterials. The enriched uranium work in Area B and theplutonium operations in Areas C and D are conducted insealed glove boxes to prevent release of materials into thework environment, Exposures to radiation and thechemicals in these areas occur primarily as a result ofaccidents. Workers at RF may be exposed to externalpenetrating radiation in the form of gamma and betarays and neutrons, or they may be exposed to internalradiation from plutonium particulate in various chemi-cal forms that are inhaled, or from metallic 239Puthat isintroduced through puncture wounds. Because of thelong, effective half-life of plutonium, internally depositedparticulate irradiate adjacent tissues with alpha raysthroughout an individual’s life.

II. BACKGROUND

Although neural tissue is relatively resistant to bothexternal and internal radiation,4 BTs have been inducedin primates following >400-rem doses of 55-MeVprotons. 5-7 Human studies have also demonstrated as-

sociations between BTs and external radiation exposureof > 11)0radsos-lzThere is some evidence that BTs are

associated with lower levels of radiation. A case-controlstudy of central nervous system cancer among childrendemonstrated statistically significant age-adjusted rela-tive risk of 1.3 among those antenatally irradiated(dosages were estimated to be <5 rad).’3 A study of 904persons treated with nasopharyngeal radium implantsfound 3 malignant BTs in the exposed group (estimatedexposure: 44-78 rads) and none in the 2021 controls.14

Studies of persons employed in the nuclear industryhave resulted in conflicting findings. Among white malesemployed at least 2 years at Hanford, no positivecorrelation of radiation exposure and the 16deaths frommalignant BT was demonstrated.ls Nuclear workers atPortsmouth who received <5 rem annually or 100 rem

cumulatively showed no increased mortality from malig-nant BTSO115Male employees at the Oak Ridge ‘-12

Plant demonstrated a significantly elevated standardizedmortality ratio (SMR) for brain cancer among thoseemployed 5 to 10 years (SMR = 589, p-value < 0.05).Although detailed exposure analyses were lacking, ex-posures to uranium dust, the major radiation hazard atthe plant, were reportedly low.” In a recent study of3508 males employed at a nuclear fuels fabrication plant,the incidence of brain cancer was significantly morefrequent than that expected [Standardized IncidenceRatio (SIR) = 2.67, 95% Confidence Interval (CI) =1.15-5.26]. However, the excess could not be attributedto any specific radiation or other occupationalexposure.la

Our primary objective in this case-control study wasto investigate the following hypotheses: (1) an associa-tion exists between exposure to internally depositedplutonium and death caused by BTs among RF whitemales, and (2) BT cases have higher external penetratingradiation exposures than do controls. In addition, wecompared the work histories of the cases and controls toinvestigate potential nonradiation occupational hazards.

III. METHODS

The distinction between malignant and benignneoplasms of the brain is difficult to make in epidemio-logic studies. Placement and size of the tumor mass oftenmake benign tumors as life-threatening as malignanttumors, and this characteristic makes case identificationdifticult. The inoperability of such tumors and frequentlack of autopsies also hinder rigorous case identification.For this study, cases of primary BTs included allmalignant and benign tumors of the brain, cranial nerves(intracranial portion), and cranial meninges, includingtumors of the pituitary and pineal body. ’9 Case ascer-tainment was based on the vital status follow-up of theRF white male cohort. Death certificates through 1980were reviewed for any mention of BT or other centralnervous system tumor or disease. For the 22 potentialcases thus identified, we obtained medical, autopsy,and/or pathology reports and reviewed them to deter-mine tumor classification and to exclude metastasis fromelsewhere in the body (Table I). This investigationresulted in 16 cases of primary BT including 14 gliomas(4 astrocytomas, 8 glioblastomas, 2 gliomas), 1 acousticneuroma, and 1pituitary tumor.

.

.

2

.

Three mutually exclusive control groups were selectedfrom the white male cohort: matched, deceased, andrandom. Each group consisted of a 4:1control-case ratio.The controls were selected as follows.

(1) Potential matched controls were first selected foreach case based on year of birth (+5 years). Minimumdistance matchingzo was then used to match on hire andtermination dates; the requirements were that the hiredate for each control must precede that of the case (s2years) and that the termination date for each controlmust follow that of the case (s2.5 years), The matchingcriterion for year of birth was expanded to ~ 10 years forone case.

(2) The random controls were selected at random fromall white male employees.

(3) The deceased controls were selected at randomfrom the pool of deceased persons within the white maleRF cohort after excluding all persons who died of canceror of central nervous system diseases.

TABLE 1. Detailed Information: Potential BT Cases

SupplementalStudy 8th Revision Information

Number’ ICD Code Diagnosis

123456789

10111213141516171819202122

438.9191225.1238191191191192.9191238.1192238191191238226.2191238238199199162

AstrocytomaGlioblastoma multiformAcoustic neuromaGlioblastoma multiformGlioblastoma multiformAstrocytomaGlioblastoma multiformGlioblastoma multiformGliomaGlioblastoma multiformAstrocytomaAstrocytomaGlioblastoma multiformGlioblastomaPontine gliomaPituitary tumorMetastatic, 1° lung10 kidney, 1° thalamusMetastatic, 10 unknownMetastatic, 1° lungMetastatic, 10 unknownMetastatic, 1° lung

‘Study numbers 1-16were used in the case-controIstudy.

Demographic data, date and cause of death, anddetailed work histories were obtained for cases andcontrols from employment records and death certificates.We divided job titles into nine categories (machinist,craftsman/maintenance, service worker, office worker,laborer, professional, technician, and manufactur-ing/production) according to a strict protocol developedafter consultation with an RF industrial hygienist. Eachworker’s location within each of the four major workareas of the plant was abstracted also. We then calcu-lated the amount of time spent in each job and area.Health physics records for all study subjects werereferenced as the source of internal and external radia-tion exposure data. All information was computerized,and the accuracy of the data was verified.21

Because of the matching criteria used, the matchedcontrols worked longer than the BT cases and thereforehad more opportunity for exposure. To correct for thispotential “bias towards the null,” each control’s durationof employment was truncated to equal that of the case.We then calculated exposure for each control from hisdate of hire through the time period equal to the durationof his case’s employment.

The percentage of maximum permissible body burden(0/6MPBB) was used to measure internally depositedplutonium.22 The MPBB is a systemic deposition ofplutonium tha4 when distributed and maintained in thebody for 1 year, gives a dose equivalent to the allowableannual radiation dose from 40 nCi of plutonium.23-24Exposures of <5V0 of MPBB (2 nCi) are indis-tinguishable from background because samples frompresumably unexposed persons can lead to calculatedbody burdens of as much as 3?40 of MPBB.* Thedeposition values were computed from the Langhamequation applied to urinalysis data collected periodicallyduring an individual’s employment at RF.2S

In this study, we used annual exposures to penetratingexternal radiation (x rays, gamma rays, and neutronscombined), recorded by the employee’s film badge orthermoluminescent dosimetry (TLD) measurements.These measurements reflect only occupational exposurereceived at RF.

For the deceased and random controls, the agestandardized maximum likelihood estimate of the oddsratio (OR) was used as a measure of relative risk.2c Forthe matched control group, matching was maintainedand Miettinen’s maximum likelihood procedure was———*This information was supplied by J. Langsted, RF Plant,August 1981, and C. Lagerquist, RF, February 1981.

3

used.z’ The statistical significance of the resulting ORSwas evaluated according to the associated lower 9570test-based confidence limit.

External exposure distributions for both cases andcontrols were extremely skewed toward the high-ex-posure values; therefore, analyses based on the normaldistribution assumption were inappropriate. Because ofthe uncertainty inherent in film badge and TLD measure-ments, it was impossible to classify some of our subjectsas exposed or unexposed, particularly those with re-corded cumulative exposures between 1 and 100 mrem,As a consequence, we did not consider the usual questionof whether the proportion of exposed persons in the twogroups was equal; instead, we asked whether casestended to have higher exposures than controls by using anonparametric Mann-Whitney procedure,28 For thisanalysis, the doses (rem) for BT cases and each controlgroup were put into a single combined array, and rankswere assigned from 1 to 80. The test statistic wascalculated as the sum of these ranks for the cases.Approximate p-values were calculated for each teststatistic.

For matched controls, a similar procedure was de-veloped. Each case was ranked with its associatedcontrols. The sum of the case ranks, across the 16 cases,was calculated and called T. We obtained the distribu-tion of T for the null case (no difference between casesand controls) by Monte Carlo simulation. We began bygenerating five uniform (O,1) variates, the first represent-ing the case and the other four representing the controls.The case was then ranked among the controls, and thesum of the ranks T1 was calculated. Approximate p-values were calculated from 1000 values of T,.

To evaluate job title and work area, we judgedcomparisons with matched controls to be the mostappropriate because of the many changes in plantoperations over time. For these analyses, we assumed

that the greater the time spent in a particular category,the greater the potential for exposure. Each case and hiscontrols were then ranked by time spent in i!ach job andwork area. Our matched version of the Mann-Whitneytest was used to test whether the cases had spent moretime in a particular category than did the controls.

Solid tumors may take years to develop after enoughexposure to induce cancer.9’” To focus on relevantexposures, supplementary analyses excluded exposurereceived within the 10 years before death (or for livingcontrols, the 10 years before the last date of follow-up).

IV. RESULTS

Characteristics of the BT cases and controls arepresented in Table II; a more detailed description of thecases is given in Table III. Because the data wereskewed, median values are reported. As expected be-cause of the matching, the median years of birth for thecases and matched controls were very similar. Deceasedcontrols were slightly older (6 years) than were cases,whereas random controls were considerably younger (15years). The median age at death for the cases was 51years compared with 58 years for deceased controls.

The median age at hire for the cases was 41 years. Weselected matched controls that had been employed forapproximately the same period and at the same age. Thedeceased controls began working 4 years earlier than thecases and at slightly older ages, whereas the randomcontrols began employment later than the cases did butat much younger ages. The cases worked a median of 7.1years, the deceased controls worked a median of 9.1years, and the random controls worked a median of 5.5years.

We noted no temporal clustering of employmentperiod among the cases (Fig. 1). Among both the cases

TABLE H. Characteristics of BT Cases and Controls (Median Values)

Matched Deceased RandomVariable Cases Controls Controls Controls

Year of birth 1921 1920 1915 1936Year of hire 1961 1961 1957 1963Age at hire (years) 41 40 45 28Year last employed 1970 1971 1968 1974Year of death 1972 --- 1973 ---Duration of employment (years) 7.1 8.4 9.1 5.5

4

.

and deceased controls, 750/0of the deaths occurred in1970 or later. However, this finding would be expectedbased on the age structure of the cohort.3

For analytic purposes, we defined exposure to internalplutonium as any positive-value body burden. As shownin Table IV, no association between BTs and theseexposures was found. ORS were <1 for all case-controlcomparisons. Of the 12 cases who had been tested forplutonium, only 4 had positive body burdens. Whenplutonium exposure was restricted to that received beforethe 10 years preceding death (or for living controls, the10 years before the last date of follow-up), severalindividuals had to be dropped from the analysis forfailing to meet this 10-year latency restriction. Hereagain, all case-control comparisons produced ORS <1(Table V).

Uranium and beryllium exposures, as recorded foreach employee in his plant medical record, were re-viewed. Evidence of exposure to beryllium or uraniumexisted in only one case; therefore, we did not analyze forthese exposures,

The distribution of cumulative external radiation forthe BT cases and controls is illustrated in Fig, 2. Eachdistribution was extremely skewed toward the higherexposure values, and the range of exposures was quitevaried. None of the cases had >16 rem of cumulativeexposure; exposures for the matched controls rangedfrom Oto 32 rem, for deceased controls Oto 52 rem, andup to 60 rem for the random group. All exposures wereless than the maximum permissible dose equivalent forradiation workers of 5 rem/year.24 The median exposurevalues in rem for the cases and the matched, deceased,and random control groups were 1.4, 0.4, 0.6, and 0.8,respectively.

Table VI shows the test statistics and their associatedp-values for the nonparametric Mann-Whitney analysesof exposure to external penetrating radiation, Althoughthe cases did appear to be more exposed to this form ofradiation than were the controls, these differences werenot statistically significant. When external exposuresaccumulated up to 10 years preceding death (or for livingcontrols, 10 years before the last date of follow-up) wereconsidered, no statistically significant difference in ex-posure levels existed between the cases and controls(Table VI).

To check on the results of the Mann-Whitney analy -ses, we also calculated ORS with exposure dichotomizedat different levels (that is, the case median and values >1rem) for both O and 10 years latency. None of theaccompanying 95?40CIS were statistically significant.

Table VII presents the Mann-Whitney scores and theassociated p-values for each job category. The cases didnot appear to have spent significantly more time in anyparticular job than did the controls, In analyses restric-ting exposure to that received 10 years before the deathof the case, the cases and controls demonstrated nodifferences. Analysis by work area showed similar results(Table VIII).

These factors were further examined in an OR analy-sis based on usual job title and usual work area for bothO and 10 years latency. No significant associationbetween the cases and controls was noted for anycategory. To evaluate possible dose-response rela-tionships for work area and job title, standardized rateratios (standardized to the lowest exposure level) wereexamined using the Mantel extension procedure;29 dura-tion of employment was broken into three mutuallyexclusive categories (never employed, employed up to 2years, and employed 2 or more years). In analyses forboth O and 10 years latency, no statistically significantdose-response trends were observed for any job title orwork area.

Because different histological types of BTs havedemonstrated different epidemiologic patterns,19i30 weeliminated the acoustic neuroma and the pituitary tumorand repeated the analyses for the more homogeneous setof glial-type tumors.altered.

V. DISCUSSION

Our findings, however, were not ‘\

Our primary purpose in this study was to investigatethe relationship between BTs and occupational radiationexposure. We found no significant association betweenexposure to internally deposited plutonium and BTsamong the RF cohort members. The health physicsexperience at RF suggested that only estimated deposi-tions of >5.0’%0MPBB be considered evidence of positiveexposure.* Because none of the cases demonstrated bodyburdens of >2.5V0MPBB, it is questionable if they wereexposed at all. Although the cases were exposed tohigher levels of penetrating radiation than were controls,these differences were not statistically significant.

Another objective of this investigation was to identify,potential nonradiation occupational hazards. No

—————————*This information was supplied by J. Langsted, RF Plant,August 1981.

5

I1=+462 -+533 -+304 -+505 ~ =....+ 496 =========+56

111 7 ~=====+&3:

9PERIODOFEMPLOYMENT

EPERIODBETWEENTERMDATEANDDEATHDATE

+ AGEATDEATH

~8 - ~+ 31ag -0 10-

~+ 39~+ 5 I

II - := ❑ ==========--=--~+ 4712- ~+3713- ~+5614-15- -=:55’ +6416- —————-- —- —————- - - ——————————-+77

1 1 I I 1 II950 1955 1960 1965 1970 1975 1980

YEARFig. 1. Age at death and work historiesfor BT cases.

6

.

TABLE III. Characteristics of BT Cases

Year Prehire

Study of Work

Number Birth Yearsi

Year Years Year

of Employed of Prehire Informations RF InformationHire at RF Death Emrioyment Years Emtslovment Years

I 1908 25.0

31.1

8.0

27.2

24.4

1953 1.5 1954 PostmanTransfer freightFillingstationParts mover

4 Patrolman 1.51341

Inspector 1.3

2 Laborer 0.5Clerk 1.5

5 Sheetmetalworker 3.27

2 Stationary operator engineer 193

83

25

1

2 1909

3 1935

4 i924

5 1924

1960 1.3 1962

1963 2.0 1965

1971 3.2 1974

1969 1.0 1973

Unknown

Air-filterserviceman

SheetmetalforemanProductionforeman

Stationary engineerSteam plant operatorMilitary

TechnicianMechanic

Electrical construction

6 1919 30.9 1970 0.4

7 1904 38.6 1963 6.4

8 1949 1.3 1970 8.3

Stationary operator engineer 0.41975

1972

1980

Electrician 6

Janitor 1Chemicalprocess operator 6Metallurgyoperator 1

Accountant 11.3

Stereo installer

9 1924 8.0 1952 11.3

10 1922 19.1 1961 11.7

1963

1973

53

6103

4214

1.5

Accountant

Military

Tool grinder

Machinist

Military

Machinist 11.7

11 1925 11.3 1957 2.0 1972 DevelopmentalengineerRadio repairmanLinesmanMilitary

Physicist 2

12 1940 2.0 1962 14.7

13 1914 19.2 1953 16.7

1977

1970

Grocery clerk Quality control chemist

Laboratory supervisor

Standards engineer

Machinist

Machine foreman

Supervisor

663

457

MaintenanceMachinistForemanMilitary

4484

2227

30.5

16

14 1910 25.7 1956 12.1 1974 Foundry foremanGrocery clerkShop analystProductionforeman

Productionforeman 12.1

15 1916 15.7

38.0

1952 19.4 1971

1952 7.0 1971

MilitaryLaborer

Inspector 19.4

Janitor 716 1894 :—‘Assumeswork began at age 20 years.

Motorman

7

TABLE IV. Comparison of Plutonium Body Burdens Between BT Cases and Controls

Matched Deceased RandomCases Controls Controls Controls

Pu Measurements (N = 16) (N = 64) (N = 64) (N = 64)

Number with positive Pu measurementsat date of termination 4 28 17 28

OR 0.29’ 0.80b 0.60b(lower 95% confidence limit) (0.09) (0.26) (0.19)

‘Miettinen’smaximumlikelihood estimate of the OR for matched case-control analysis.‘Gart’s maximum likelihood estimate of the pooled OR.

TABLE V. Comparison of Plutonium Body Burdens Between BT Cases and Controls: ExposureRestricted to 10 Years Before Death

Matched Deceased Random

Cases Controls Controls Controls

Pu Measurements (N= 8) (N= 32) (N= 47) (N= 50)

Number with positive Pu measurements10 years before death 2 14 12 21

OR 0.86’ 0.82b o.73b

(lower 95% confidence limit) (0.14) (0.17) (0.15)

‘Miettinen’s maximum likelihood estimate of the OR for matchedcase-controlanalysis.‘Gart’s maximumLikelihoodestimateof the pooledOR.

statistically significant differences in exposure by job titleor work area were demonstrated among the cases andthe matched controls.

Power calculations for this study were based on thethreefold excess of BTs reported in the cohort study.2’3The power to detect three times as much externalexposure among the cases was 70%, whereas the powerto detect a threefold difference in length of employmentfor job title or work area was approximately 40!L0.Thesame conclusions were obtained in the supplementaryanalyses based on the slightly more powerful parametricstatistical tests.

Each control group demonstrated certain strengthsand weaknesses. Unlike the cases, many of the randomcontrols were recently hired employees whose durationof employment was truncated at the end-of-study date.They appeared to have begun work at younger ages thandid the cases and to have worked for shorter periods of

8

time. Therefore, crude comparisons with this groupwould have been biased. However, age standardizationminimized these differences.

The deceased controls were older and tended to havebegun work at older ages than did the cases. However, inboth the crude and age standardized assessments ofemployment duration, the deceased controls workedlonger than the cases did, thus providing no evidence thatlength of employment was a BT risk factor. It is possiblethat death was an exposure correlate, thereby decreasingthe likelihood of detecting any real difference that mightexist between the cases and these controls.

The matched controls (matched on age and period ofemployment) were selected to minimize the potentialconfounding effects resulting from changes in plantoperations, industrial processes, industrial hygiene reg-ulations, and health physics standards. However, wemay have overmatched on period of employment, thus

40 “

7/

20 - //

mm u m mo I I I I I 1 T

40r

7

/

20 - / .//

o - Pr1 I I [ I I I 1 I 1 I 1

40-7//

20 -/////

o .-n[ I I I I I 1 I 1

mo I I I I I I I I I I I 1

0 10 20 30 40 50 60

rem

Fig. 2. Distributionof cumulativeexternal radiationfor BTcases andcontrols.

hindering our ability to recognize differences in theoccupational exposures of interest.

The potential for overmatching was examined brieflyin additional case comparisons with control poolsmatched on the following criteria: (1) date of birth only(+2 years); (2) birth date (+2 years) and hire date (s2years); and (3) the original matching on birth date (+5years), hire date (<2 years), and term date (s2.5 years).Each pool contained all potential controls. For a givencontrol pool, the cumulative external radiation exposurefor each case was ranked among his controls. Rankswere converted to percentiles, and the sum of the casepercentiles was calculated. The respective scores for the

three pools were 8.6, 8.9, and 8.5. None of the scoreswere statistically different than the expected value of 8.0.On the basis of this examination, we have no reason tobelieve that overmatching was a major problem in ourstudy design.

Comparing death certificates with the supplementalautopsy, hospital, and/or pathology data allowed us toevaluate misclassification resulting from inaccuratecause-of-death statements. Among the 16 BT cases, wefound a 36?40discrepancy between the cause-of-deathstatement and additional information. This discrepancyrate was similar to that reported in the literature, whichdocuments discrepancies ranging from 39 to 57cz0.31’32

9

TABLE VI. Comparison of Cumulative External Radiation Exposure Between BT Cases andControls

No Latency Exposure IO Years Before Death(N = 16) (N = 8)

Matched Deceased Random Matched Deceased RandomControls Controls Controls Controls Controls Controls(N= 64) (N= 64) (N= 64) (N= 32) (N= 47) (N= 50)

Mann-Whitney ScoreObserved 54 579 513 23.5’ 227.5 178.0Expected 48 512 512 24 188.0 200.0

Approximateone-sided p-value 0.15 0.21 0.50 0.55 0.17 0.69

‘Because the Mann-Whitney procedure is not appropriate for matchedcontrols,a similartest wasdevisedusingMonte Carlo simulation.

Comparing ICD codes with the cause-of-death state-ments identified another source of error: one caseappeared as a false negative. These exercises demon-strated the importance of careful review of all deathcertificate data used for case definition in case-controlstudies.

Because of the paucity of data on BT etiology, we canonly speculate about other factors that may have beenresponsible for the excess of BTs among the RF cohortmembers. On the basis of animal studies, numerouschemicals and viral agents have been recognized as braincarcinogens. 33.34population-based epidemiologic studies

of humans have been less definitive. The possibility thatenvironment al factors are involved in braintumorigenesis has been suggested based on geographicvariations in disease rates.30 Genetic factors have alsobeen implicated as BT risk factors.19s30Within the pastdecade, excess deaths from BTs have been reportedamong a variety of occupational groups including rubberworkers, aluminum workers, chemists, veterinarians, oilrefinery workers, electricians, pattern makers,machinists, and asbestos insulators.3s$36However, wefound no association at RF with specific job titles oroccupational exposures.

Another possible explanation for our observed excessof BT deaths is a “diagnostic sensitivity bias.”37This biasis purported to result from comparing the generalpopulation to an employed population that may havebetter medical care and di~gnosis.

Investigation of these suspected risk factors wasbeyond the scope of our study. Also, we could not

10

evaluate radiation and other potential carcinogenic ex-posures received before employment at the plant orreceived nonoccupationally. [n reviewing the pre-RFoccupational histories of the cases (Table III), note thatthe median for pre-RF employment was 21.8 years.Eleven cases had been previously employed in industrialoccupations; nine of these were employed for more than10 years. However, no common occupation wasidentified.

VI. CONCLUSIONS

Although limited by a small number of cases, wefound no statistically significant differences in exposurebetween BT cases and the three control groups forplutonium or external penetrating radiation. In addition,no potential nonradiation occupational hazards wereidentified. These same conclusions held in our sup-plementary analyses by latency and for glial-type tumorsalone.

ACKNOWLEDGMENTS

We acknowledge the valuable assistance of everyoneinvolved in the collection and organization of the datafiles used in this study. We especially thank LaurieWiggs, Jeannene Masterson, and Joan Essington; RoyTally of RF for providing the radiation dosimetry data;Eva Woods for computer support; and Stan Ferguson of

. .

TABLE VII. Com~arison of Job Titles Between BT Cases and Matched Controls

Ever Exposed 10 Years Before DeathEver Exposed (N = 16) (N = 8)

Number of Approximate Number of ApproximateExposed Mann-Whitney Score One-Sided Exposed Mann-Whitney Score One-Sided

Occupation Cases Observed Expected p-Value Cases Observed Expected p-Value

Professional 4 49.0 48.0 0.43 3 27.5 24.0 0.20

Technician 1 43.5 48.0 0.77 0 --- --- ---

Craftsmen/Maintenance 2 37.5 48.0 0.97 0 --- --- ---

Machinist 3 48.5 48.0 0.50 3 28.0 24.0 0.17

Manufacturing& Production 6 53.0 48.0 0,20 3 27.0 24.0 0.23

Service Personnel 4 48.0 48.0 0.50 0 --- --- ---

OffIce Workers 3 46.5 48.0 0.59 2 25.0 24.0 0.40

Laborers 3 46.0 48.0 0.62 1 26.5 24.0 0.27

TABLE VIII. Comparison of Work Area Between BT Cases and Matched Controls

Ever Exposed 10 Years Before DeathEver Exposed (N = 16) (N =8)

Number of Approximate Number of Approximate

Exposed Mann-Whitney Score One-Sided Exposed Mann-Whitney Score One-Sided

Work Area Cases Observed Expected p-Value Cases Observed Expected p-Value

A 8 49,0 48.0 0.43 3 23.5 24.0 0.55

B 9 51.5 48.0 0.26 2 22.0 24.0 0.69

c 12 53.5 48.0 0.19 3 23.5 24.0 0.55

D 6 44,0 48.0 0.74 2 20.0 24.0 0.84

.

me Coloracto State Health Department. Uur thanks goto all reviewers and the Epidemiology Advisory Commit-tee for their valuable suggestions and advice, with specialappreciation to Strother Walker, Roy Shore, and Rich-ard Waxweiler.

REFERENCES

1.

2.

3.

4.

5.

6.

7.

12

C. J. Johnson, “A Preliminary Evaluation of BrainCancer, Melanoma and Respiratory Cancer of Em-ployees of the Rocky Flats Nuclear Weapons Plantin Jefferson County, Colorado,” American PublicHealth Association Meeting, Detroit, Michigan, Oc-tober 22, 1980.

G. L. Voelz, G. S. Wilkinson, J. F. Acquavella, G.L. Tietjen, R. Brackbill, M. Reyes, and L. Wiggs,“An Update of Epidemiologic Studies of PlutoniumWorkers,” Proceedings of the International Con-ference on the Radiobiology of Radium and theActinides in Man, Lake Geneva, Wisconsin, Octo-ber 11-16, 1981 (to be published in Health Phys.).

G. S. Wilkinson, G. L. Voelz, J. F. Acquavella, G.L. Tietjen, M. Reyes, R. Brackbill, and L. Wiggs,“Mortality Among Plutonium and Other Workers ata Nuclear Facility,” 16th Midyear Topical Meeting,Health Physics Society, Epidemiology Applied toHealth Physics, Albuquerque, New Mexico, January9-13, 1983.

Committee on the Biological Effects of IonizingRadiations, The E&iectson Populations of Exposureto Low Levels of Ionizing Radiation: 1980 (NationalAcademy Press, Washington, DC, 1980).

W. Haymaker, L. Rubinstein, and J. Miquel, “BrainTumors in Irradiated Monkeys,” Acta Neuropathol.(Berlin) 20, 267-277 (1972).

G. Splitter, J. Kirk, and H. Casey, “Radiation-Related and Spontaneous Tumors in Primates,” inRadionuclide Carcinogenesis, C. Sanders et al., Eds.(US Atomic Energy Commission, Washington, DC,1973) pp. 55-89.

J. Traynor and H. Casey, “Five-Year Followup ofPrimates Exposed to 55-MeV Protons,” Radiat.Res. 47, 143-148 (1971).

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

B. Modan, D. Baidatz, H. Mart, R. Steinitz, and S.Levin, “Radiation-Induced Head and NeckTumors,” Lancet I, 277-279 (1974).

R. E. Shore, R. E. Albert, and B. S. Pasternack,“Follow-up Study of Patients Treated by X-RayEpilation for Tinea Capitis,” Arch. Environ. Health ,

31, 17-24 (1976).

M. Colman, M. Kirsch, and M. Creditor, “TumorsAssociated with Medical X-Ray Therapy Exposurein Childhood,” in Late Biological Eflects of IonizingRadiation, Vol. I (International Atomic EnergyAgency, Vienna, 1978).

S. Seyama, T. Ishimaru, S. Iijima, and M. Kazuo,“Primary Intracranial Tumors Among AtomicBomb Survivors and Controls, Hiroshima andNagasaki, 1961-1975,” Radiation Effects ResearchFoundation report RERFTR 15-79.

G. Matanoski, R. Seltser, P. Sartwell, E. Diamond,and E. Elliott, “The Current Mortality Rates ofRadiologists and Other Physician Specialists: Spe-cific Causes of Death,” Am. J, Epidemiol. 101,199-210 (1975).

J. F. Bithell and A. Stewart, “Prenatal Irradiationand Childhood Malignancy: A Review of BritishData from the Oxford Survey,” Br. J. Cancer 31,271 (1975).

D. P. Sandier, G. W. Comstock, and G, M.Matanoski, “Neoplasms Following ChildhoodRadium Irradiation of the Nasopharynx,” J. Nat.Cancer Inst. 68, 3-8 (1982).

E. Gilbert and S. Marks, “Analysis of Workers in aNuclear Facility,” Radiat. Res. 79, 122-148 (1979).

R. A. Rinsky, R. D. Zumwalde, R. J. Waxweiler, W.E. Murray, Jr., P. J. Bierbaum, P. J. Landrigan, M.Terpilac, and C. Cox, “Cancer Mortality at aNuclear Naval Shipyard,” US Department of Healthand Human Services, Centers for Disease Control,National Institute for Occupational Safety andHealth, Cincinnati, Ohio (1980).

C. C. Lushbaugh, “The Development and PresentState of the DOE Health and Mortality Studies,”

Proceedings of the DOE Radiation ContractorsWorkshop, Rockviile, Maryland, April 13-14, 1982(United States Department of Energy) pp. 1-36.

18. 0. C, Hadjimichael, A. M. Ostfeld, D. A. D’Atri,and R. E. Brubaker, “Mortality and Cancer In-cidence Experience of Employees in a Nuclear FuelsFabrication Plant,” J. Occup. Med. 25, 48-61(1983).

19. B. S. Schoenberg, B. W, Christine, and J, P.Whisnant, “The Descriptive Epidemiology of Pri-mary Intracranial Neoplasms: The Connecticut Ex-perience,” Am. J. Epidemiol. 104,499-510 (1976).

20. A. H, Smith, J, C. Kark, J. C. Cassell, and G.Spears, “Analysis of Prospective EpidemiologicStudies by Minimum Distance Case-ControlMatching,” Am. J. Epidemiol. 105,567-573 (1977).

21. M. Reyes, G. S. Wilkinson, and J. F. Acquavella,“Quality Control Activities of Coding Consistent forData Collected at Rocky Flats,” LA-9462-MS (tobe published).

22. “Maximum Permissible Body Burdens and Max-imum Permissible Concentrations of Radionuclidesin Air and in Water for Occupational Exposure(includes Addendum 1 to NBS Handbook 69 issuedin August 1963),” National Council on RadiationProtection and Measurements report 22, Washing-ton, DC (1959).

23. International Commission on Radiological Protec-tion, “Report of ICRP Committee 11on PermissibleDose for Internal Radiation (1959),” Health Phys. 3,1-380 (June 1960).

24. John W. Healy, “Los Alamos Handbook of Radia-tion Monitoring,” Los Alamos Scientific Laboratoryreport LA4400 (March 1970).

25. J. N. P. Lawrence, “A History of PUQFUA(~lgtonium Body Burden Q) from Urine$ssays),” Los Alamos Scientific Laboratory r~portLA-7403-H (October 1978).

26. J. J. Gart, “Point Internal Estimation of the Com-mon Odds Ratio in the Combination of 2 x 2 Tableswith Fixed Margins,” Biometrics 57, 471-475(1970).

27. 0. S. Miettinen, “Estimation of Relative Risk fromIndividually Matched Series,” Biometrics 26, 75-86(1970).

28. J. Conover, Practical Nonparametric Statistics(John Wiley & Sons, Inc., New York, 1971).

29. N. Mantel, “Chi-Square Tests with One Degree ofFreedom: Extension of the Mantel-Haenszel Proce-dure,” J. Am. Stat. Assoc. 59, 690-700 (1963).

30. E. Gold, “Epidemiology of Brain Tumors,” inReviews in Cancer Epidemiology, A. Lilienfeld, Ed.(Elsevier-North Holland Publishing Company, NewYork, 1980), Vol. I.

31. A. Gittlesohn and J. Senning, “studies on theReliability of Vital Health Records: I. Comparisonof Cause of Death and Hospital Record Diagnoses,”Am. J. Public Health 69,680-688 (1979),

32. G. J. James, P. E. Patton, and A. S. Heslin,“Accuracy of Cause-of-Death Statements on DeathCertificates,” US Public Health report 70, 39-51(1955).

33. J. Swenberg, “Chemical and Virus-Induced BrainTumors,” J. Nat. Cancer Inst. Monograph 46, 3-10(1977).

34. D. D. Bigner and C. N. Pegram, “Virus-InducedExperimental Brain Tumors and Putative Associa-tions of Viruses with Human Brain Tumors: AReview,” Adv. Neurol. 15, 57-83 (1976).

35. V. Alexander, S. Leftingwell, W. Lloyd, R. J.Waxweiler, and R. L. Miller, “Brain Cancer inPetrochemical Workers: A Case Series Report,”Am. J. Ind. Med. 1, 115-123 (1980).

36. R. Waxweiler, “A Review of Occupational RiskFactors of Brain Cancer,” Nikko Brain CancerTherapy Conference, Nikko, Japan, October 14,1981.

37. P. Greenwald, B. R. Friedlander, C. E. Lawrence, T.Hearne, and K. Earle, “Diagnostic sensitivity

Bias—An Epidemiologic Explanation for an Ap-

parent Brain Tumor Excess,” J. Occup. Med. 23,690-694 (1980).

13

Fj’)

PageRange

001.0250264500s1.07s076.100101.125126.150

NTISPriccccde

A02A03A04A05A06A07

PrinkdintheUnitcd Ststesof AmericaAvailablefmm

Nstiond TechnicalInformationServiceUS Depmtmmtof Commeme

S28S Port Royal RoadSpringfield,VA 22161

Microfiche(A131)

Psge RangeNT2S

Pric4Code PageRangeNTIS

Price code PageRange

Is 1-175176.200201.225226-2so251.275276.3oo

A08A09A1OAllA12A13

301-325326.350351-37s37640401.42s426450

A14A15A16A17AMA19

451473476-500.$01.525526.350551.575S76.6006ol.up”

NnsPrice Code

A20A21A22A23A24A25A99

“ContactNTIS fora pricequote.