Cancer Causes and Control 1996, 7, pp. 312-321

Cancer mortality in German male workers exposed to phenoxy herbicides and dioxins

Heiko Becher, Dieter Flesch-Janys, Timo Kauppinen, Manolis Kogevinas, Karen Steindorf, Alfred Manz, and Jiirgen Wahrendorf

(Received 19July 1995; accepted in revised form 23 October 1995)

In an occupational cohort study, the relation between exposure to phenoxy herbicides, and contaminants (dioxins and furans) and cancer mortality was investigated. A total of 2,479 workers from four plants in Germany were included, with a mortality follow-up until the end of 1989 (for one cohort, until the end of 1992). A total of 484 deaths were recorded yielding a standardized mortality ratio (SMR) of 101 (95 percent confidence interval [CI] = 92-111) for total mortality, and an SMR of 119 (CI = 100-141) for all malignant diseases. Avariety of herbicides was produced, including those which are known to have been contaminated with 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). High dioxin and furan exposure (in particular, exposure toTCDD, but also to higher chlorinated dioxins) had occurred in two of the four plants as shown by blood-fat measurements in a sample of workers. Mortality from all neoplasms increased with latency and was highest in the largest plant where the highest TCDD blood levels were recorded. An increased mortality in the total cohort from respiratory cancer (SMR = 154, CI = 115-202), cancer of the buccal cavity and phar- ynx (SMR = 295, CI = 135-560), and non-Hodgkin's lymphoma (SMR = 326, CI = 119-710) was observed. Our findings are consistent with results from other cohorts which showed an increased overall cancer mortality and mortality of respiratory cancer after long-term exposure to these phenoxy herbicides and dioxins. Cancer Causes and Control 1996, 7, 312-321

Key words: Cohort study, dioxin exposure, Germany, lung cancer, males, occupation, total cancer.

Introduct ion

The International Agency for Research on Cancer (IARC) has set up a register of workers exposed to phe- noxy acid herbicides and chlorophenols in 11 countries. These substances are known to be contaminated with di- oxins and furans, often including the most toxic conge- ner, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). TCDD and occupational exposure to phenoxy acid herbicides or chlorophenols have been classified by an IARC working group as possibly carcinogenic to humans (Group 2B).

Recent studies based on occupational c o h o r t s 2-4 and on the population accidentally exposed in Seveso, s con- sistently have shown an increased total cancer incidence or mortality after TCDD exposure.

The register of the IARC includes workers exposed in production or spraying, and has been followed until the end of 1989. The cancer mortality in the total cohort as well as in various subcohorts has been reported. 6 Only a small excess cancer risk was found among workers char-

Drs Becber, Steindorf, and Wabrendorf are with the German Cancer Research Center, Heidelberg, Germany; Drs Flescb-Janys and many are with the Medical Center]or Chemical Workers' Health, Hamburg, Germany; Dr Kauppinen is with the Finnisb Institute of Occupational Health, Helsinki, Finland; Dr Kogevinas is with the International Agency for Research on Cancer, Lyon, France, and lnstitut Municipal D'investigacio Medica, Barcelona, Spain. Address correspondence to Dr Becber, German Cancer Research Center, Division of Epidemiology, Im Neuenbeimer Feld 280, D-69120 Heidelberg, Germany. This study was supported by the German Ministry of Research and Technology under contract 01HK5283.

312 Cancer Causes and ControI.Vol 7. t996 © 1996 Rapid Science Publishers

Pbenoxy herbicides, dioxins, and cancer

acterized as probably having been exposed toTCDD. No significantly increased cancer risk was found. No German cohorts have been included in these analyses because data collection in Germany was still in progress at that time. However, cases with non-Hodgkin's lympho- ma from the German cohorts are included in a nested case-control study performed within the IARC register. 7

In this paper, we present the mortality of male em- ployees in four plants producing phenoxy herbicides and chlorophenols. In three of the four subcohorts, the number of female employees was too small to allow a separate analysis (11, six, and three women, respectively). For one subcohort, the mortality of females was reported previously s and is not presented here. Chemical exposure and a detailed description of the production processes of the German cohorts have been published by Kauppinen et al. 9

Materials and methods

Study population and rnortality follow-up

We aimed to include workers from all factories in the former Federal Republic of Germany which have been involved in the production process of phenoxy-acid her- bicides and chlorophenols or who were likely to have been in contact with these substances through working in neighboring production areas. Five factories were identified of which four agreed to participate. For all co- horts, the identification of the workers was made by the companies under supervision of the first author. For one company (Boehringer, Ingelheim) the identification was made also from union election records and personal in- formation from employees. Only workers with German nationality and at least one month of employment were included. The design and a description of the four subco- horts have been published by Becher et al.I° We excluded non-German individuals because: (i) a large part of for- eign workers return to their home country after retire- ment, making it difficuk to obtain their vital status information; and (ii) the German national mortality rates are calculated on the basis of subjects who had their last place of residence in Germany, thus, for a cohort with a high percentage of foreigners, national rates would not be the appropriate reference.

A mortality follow-up was performed through 31 De- cember 1989 for three subcohorts and through 31 Decem- ber 1992 for one subcohort. In most cases, the cause of death was recorded through death certificate and coding according to the International Classification of Diseases

1 1 • • (ICD-9) was performed by a professional nosologlst. Table 1 describes the study population and follow-up

characteristics. A further description by cohort is given below. Duration of employment is calculated from the

start of employment in a department with suspected ex- posure until end of employment in the phnt. This time may include some few periods without exposure.

Cohort I (Boehringer Ingelheim, Plant Hamburg-Moor. fleet). Cohort I consists of 1,144 male workers with a mean duration of employment of 7.7 years. Of these, 345 deaths were recorded (30 percent of the cohort). Nine men were lost to follow-up (0.8 percent). As possible sources of information for cause of death, we used death certificates, hospital and/or physician reports, and infor- mation of histologic examination. No major discrepan- cies were found when mukiple information was available.

This cohort has been under investigation previously, z Data collection for this study was done under the IARC protocol; however, information on vital status and causes of death was made available by these authors. In compar- ison, Manz et al 2 report results from 1,184 male workers and 313 deaths (2.9 percent lost to follow-up). The analysis presented here differs from previous results in three as- pects: (i) the analysis is based on workers with German nationality; as the percentage of workers with non-Ger- man nationality was low (3.5 percent), the exclusion of those workers could be expected to have little impact on the results; (ii) the cohort has been cross-checked with company records, and a small number of additional workers have been identified and included in this analy- sis; (iii) a further check of vital status and causes of death yielded a small number of additional deaths which were coded previously as alive at the end of follow-up.

All these differences led to a slightly increased mortal- ity compared with previous analyses, although the overall resuks remained virtually the same.

Cohort H (Bayer Uerdingen). Cohort II consists of 135 male workers with a mean duration of employment of 21.5 years. In this cohort, a follow-up was performed until 31 December 1992. Three men (2.2 percent) were lost to follow-up. All 23 recorded causes of death are based on death certificate information. Ten percent of the workers who are non-German were excluded from the analysis. For this cohort, a detailed smoking history was available from the medical department of the factory.

Cohort III (Bayer Dormagen). The cohort consists of 520 male workers with a mean duration of employment of 9.1 years. Twenty percent of workers who are non-German were excluded from the analysis. This is a relatively young cohort as the start of production was in 1965. The percentage lost to follow-up for this cohort is larger than for the other cohorts (15.6 percent). It was not possible to trace a number of individuals with only a temporary ad- dress known to the company. All recorded causes of death are based on death certificate information.

Cancer Causes and Control.Vol 7.1996 313

H. Becher et al

T a b l e 1. Characteristics of the cohorts (males), Germany phenoxy herbicide and chlorophenol manufacturing plants

Characteristics Cohorts

I II III IV Total (No.) (No.) (No.) (No.)

Year of birth < 1920 293 20 4

1920-1929 247 29 51 1930-1939 315 52 131 1940-1949 207 27 126 1950+ 82 7 208

Start of employment(year) < 1955 315 35

1955-1959 279 22 1960-1964 72 19 1965-1969 225 29 1970-1974 139 28 1975+ 114 2

Vitalst~us ~ end offollow-up a Alive 790 109 Dead 345 23 Lost-~follow-up 9 3

Person-yeaBof obsew~ion 27,448.1 3,973.3 Total 1,144 135

88 405 95 422

161 659 126 486 210 507

- - - - 350 - - 271 572 - - 84 175

200 58 512 150 57 374 170 210 496

416 580 1,895 23 93 484 81 7 100

7,783.7 14,857.4 54,062.5 520 680 2,479

a 31 December 1989 (Cohorts I, III, IV), 31 December

Cohort IV (BASF Ludwigsbafen). The cohort consists of 680 male workers with a mean duration of employment of 18.5 years. In this cohort, 31.6 percent of the workers are non-German. All recorded causes of death were based on death certificate information except for seven individuals for whom death certificates were not avail- able because the storage period required by law had ex- pired. Their causes of death were identified by linking the date of birth and date and place of death information with the death records from the Statistisches Landesamt (statistical state office) who keep data in an anonymous format. For data protection reasons, we were not given the individual cause of death for these individuals but only the grouped information. While this is not critical for overall mortality analysis presented here, it precludes subgroup analyses. The reliability of these causes of death can be considered as sufficient as the same data source is used for the official mortality statistics.

The cohort of workers from this company were em- ployees in the synthesis or formulation department of various substances as shown in Table I or in other depart- ments with exposure to these substances. It is important to note that this cohort is different from the accident co- hort described in Thiess et a112 and Zober et al 3 in which a high TCDD exposure occurred.

Description of the plants and production processes Table 2 gives an overview of substances produced in the four plants during different production periods. In the following, we give a short description by plant. The

314 Cancer Causes and Control.Vol 7.1996

1992 (Cohort II).

abbreviations of the names of the chemical agents are explained in the footnote toTable 2.

Cohort I (Boebringer Ingelbeim, Hamburg-Moorfleet). The chlorophenol unit of this plant synthesized 2,4,5- TCP and 2,5-DCP. In 1957-70 the estimated median con- centration of TCDD in these compounds was 5-10 ppm. The rabbit ear tests indicating chloroacnegenic potential and presence of dioxins were generally positive in the 1950s. In the early 1970s, the TCDD content of 2,4,5- TCP was about 1 ppm and < 1 ppm after that. In 2,5- DCP, the measured concentration of all TCDD isomers was < 1 ppm. The reaction temperature of the process was adjusted to avoid the formation of dioxins in the early 1970s.

The herbicide unit of the plant synthesized 2,4,5-T acid and 2,4,5-Testers. The median concentration of TCDD in 2,4,5-T products was about 1 ppm according to measure- ments made in the 1970s and < 0.1 ppm after introduc- tion of a new dioxin extraction procedure in 1981. The estimated median concentration of TCDD in 1955-70 was 5 ppm in the products and 10-50 ppm in some pro- cess and waste streams. The rabbit ear tests were gener- ally positive until the mid-1950s, but negative after that.

No major accidents happened in the 2,4,5-T or chloro- phenol units during their operation. Numerous cases of chloroacne occurred in 1954 among workers of the chlor- ophenol unit when the 2,4,5-TCP synthesis was in its early stage of development, and the hygiene in handling of chemicals and the protection of workers were insuffi-

Pbenoxy herbicides, dioxins, and cancer

Table 2. Data on production of chemical agents in the German phenoxy herbicide and chlorophenol manufacturing plants involved in the IARC study on cancer dsks

Plant and department Cohort Products manufactured or formulated in the plant

Agent a Period Tons/yr b

Boehringer/Hamburg (I) Synthesis

Formulation Bayer/Uerdingen (11)

Synthesis Bayer/Dormagen (111)

Synthesis

Formulation

BASF/Ludwigshafen (IV) Synthesis

Formulation

2,4,5-T 1953-83 840 2,4,5-TCP 1953-54 and 1957-83 160 2,5-DCP 1967-84 2,4,5-T 1960-83 100

2,4,5-TCP 1951-76 400

2,4-D 1965-89 310 2,4-DP 1965-89 1,580 2,4,5-T 1965-76 1,090 2,4,5-TP 1968-75 40 MCPA 1965-88 2,000 MCPP 1965-88 890 2,4-DCP 1970-88 1,910 2,4-D 1965-89 650 2,4-DP 1965-89 610 2,4,5-T 1965-77 270 MCPA 1965-88 560 MCPP 1965-88 380

2,4-D 1957-87 2,880 2,4-DP 1963-85 1,480 MCPA 1956-87 1,210 MCPP 1957-87 2,160 2,4-DCP 1957-87 - - 2,4-D 1958-87 5,400 e 2,4-DP 1963-86 3,010 e 2,4,5-T 1959-84 780 e MCPA 1959-87 2,660 e MCPP 1958-87 4,920 c

a Abbreviations: 2,4-D, 2,4-dichlorophenoxyacetic acid and its derivatives; 2,4-DP, 2,4- dichlorophenoxypropionic acid and its derivatives; 2,4,5-T, 2,4,5-trichlorophenoxyacetic acid and its derivatives; 2,4,5-TP, 2,4,5-trichlorophenoxypropionic acid and its derivatives; MCPA, 4-chloro-2-methylphenoxyacetic acid and its derivatives; MCPP, 4-chloro-2- methylphenoxypropionic acid and its derivatives; 2,4,5-TCP, 2,4,5-trichlorophenol; 2,4- DCP, 2,4-dichlorophenol; 2,5-DCP, 2,5-dichlorophenol.

b Crude estimate of the average production volume during the period of production; unknown if not given.

c Amount of formulated product (solution or mixture).

cient. The process was discontinued in 1954 and re- opened in 1957 after technical improvements that re- duced TCDD-exposure of workers. There was a rather frequent rotation of workers between several production units in this factory that produced also opium alkaloids, tindane, and Bromophos.

Cohort H (Bayer Uerdingen). The chlorophenol plant mamffactured 2,4,5-TCP. No major accidents have been recorded in the history of the plant. The tota ! dioxin level of the product in the 1970s has been on average about 0.01 ppm. Eleven cases of chloroacne occurred in the early 1970s among maintenance workers and their

assistants who repaired, e.g., autoclaves by welding - an operation that may cause the formation of dioxins due to heating of chlorophenols. The chlorophenoi workers of the plant also may have been exposed occasionally to other chemicals and products occurring in the same building. These include alkylated anilines, acetanilide, ethanethiol and alkylated chloroformates.

Cohort 111 (Bayer Dormagen). A variety of phenoxy her- bicides were produced in this plant: 2,4-D; 2,4-DP; 2,4- DCP; 2,4,5-T; 2,4,5-TP; MCPA; and MCPP. The dioxin levels of the products have been in the sub-ppm range, and no cases of chloroacne occurred among the workers.

Cancer Causes and ControLVo[ 7.1996 315

H. Becher et al

Occasional exposure to various other chemicals has been possible because other pesticides also were synthesized in the same building. These herbicides include diuron, Surcopur, propham and chlorpropham, Gokix, Tribunil, and Sencor. Chemical agents used in the synthesis of these products include aromatic amines, isocyanates, hydrazine, and solvents.

Cohort IV (BASF Ludwigshafen). The plant synthesized 2,4-D, 2,4-DP, MCPA, and MCPP. All of them were pro- duced as acids, esters, and amines. In addition, purchased 2,4,5-T esters and amines were formulated in 1959-84. MCPA and MCPP were synthesized directly from o-cre- sol, monochlorocarboxylic acids, and sodium oxychlor- ide. In 1965-85, the syntheses were based on the condensation of p-chloro-o-cresol with monochloro- carboxylic acids. The synthesis of 2,4-D was based on a similar type of reaction between 2,4-DCP and mono- chloroacetic acid in 1957-85.

The plant where phenoxy herbicides were produced is a large building where also many other chemicals (indud- ing, e.g., aniline, o-tolidine, nitrobenzene, ethykoluidine, anisidine, toluene, iso-octanol, and butyl glycol) were synthesized. No major accidents and no cases of chlor- oacne indicating exposure to dioxins have been reported from this plant.

Human dioxin exposure measurements

TCDD-exposure was verified by measurements in blood fat in a subsample of workers in cohort I and cohort II. 2'13'14 In cohort II, dioxin measurements in a sample of eight workers showed TCDD-concentrations in a range of 163-1,935 ng/kg blood fat. These workers had also chloroacne and other clinical symptoms as de- scribed in detail in Jansing and Korff. 14 The workers of cohort I are still under investigation. Previously pub- lished data 14'15 report a range of 3-2,252 ng/kg blood fat TCDD in a sample of 112 male and 18 female workers. High-resolution-gas-chromatography/high-resolution- mass-spectrometry measurements were performed in du- plicate. Taking into account the concentration of other dioxins/furans into account yields a range of 21- 2,733 ng/kg toxic equivalencies (TEQ). For cohorts III and IV no measurements have been performed. Exposure to TCDD was probably lower in these subcohorts be- cause a rather small proportion of the production of the respective plants consisted of products contaminated by TCDD (see discussion; see also Kauppinen et al 9), how- ever, some dioxin exposure is likely to have occurred in these cohorts as well.

Statistical methods

Expected numbers of death were calculated using the program PERSON-YEARS 16 on the basis of the West

316 Cancer Causes and Control.Vol 7-1996

German mortality rates categorized in five-year age inter- vals and five-year calendar periods as supplied from the Statistisches Bundesamt (federal statistical office). Stan- dardized mortality ratios (SMR) and 95 percent confi- dence intervals (CI) were calculated on the basis of a Poisson distribution. 17

For one subcohort, complete smoking information was available. We used Cox regression analysis 17 to inves- tigate the joint effect of smoking and exposure in the synthesis department on total mortality. Both covariables were taken as binary factors. As the basic time variable, we used time since first employment) 6 Age was categor- ized into three groups for adjustment.

We investigated the effect of time since first exposure on the SMR for the total cohort by dividing the cohort in three intervals ( < 10 years, 10- < 20 years, and 20 years or more). 19 We used the total cohort for this analysis because numbers would otherwise become too small.

SMRs were presented for each cohort separately un- der the a priori decision that all cohorts have a specific exposure pattern. For the latency analysis, however, we decided to present a joint analysis because numbers would become too small when presenting it separately by cohort.

Results

Mortality from all causes in the total cohort was around that expected from national rates (484 observed deaths, SMR = 101 [CI = 93-111]). The mortality pattern is differ- ent, however, in the four cohorts, with a relatively high SMR in cohort I (SMR = 116) and lower SMRs in the other three cohorts (SMR = 58 for cohort II; SMR = 76 for cohort III; and SMR = 85 for cohort IV). The overall and cause-specific mortality is given in Table 3 for each cohort separately.

Table 4 also shows SMR-analyses by time since first exposure for all cohorts combined. For total cancer mor- tality, we observe a SMR of 119 (CI = 100-141) which is slightly lower for 0-10 years since first exposure and slightly higher for the other groups. With the exception of cohort III, the SMR for all cancers is about 20 percent higher than the SMR for all causes. An excess risk of lung cancer was observed in three out of four cohorts. A sta- tistically significant excess risk was observed for all co- horts combined with an observed number of 52 cf an expected number of 33.8 (SMR = 154, CI = 115-203). Lung cancer risk did not vary much with time since first exposure. There were six deaths from non-Hodgkin's lymphoma (ICD-9 codes 200, 202), four from cohort I, two from cohort II cf 1.84 expected (SMR=326, CI = 120-711). A fourfold excess risk was observed for 20 or more years since first exposure. A statistically sig- nificant excess mortality was observed for cancer of the

Tabl

e 3

. M

ort

ali

ty

in f

ou

r G

erm

an

c

oh

ort

s

ex

po

se

d

to p

he

no

xy

h

erb

icid

es

a

nd

c

on

tam

ina

nts

, a

ma

les

ICD

-9 b

Ca

use

of

de

ath

C

oh

ort

Bo

eh

rin

ge

r In

ge

lhe

im (

I)

Ba

yer

Ue

rdin

ge

n

(ll)

B

aye

r D

orm

ag

en

(1

11)

BA

SF

Lu

dw

igsh

afe

n

(IV

)

Ob

s c

Exp

d S

MR

e

(CI)

f O

bs

c E

xp d

SM

R e

(e

l) f

Ob

s c

Exp

d S

MR

e

(CI)

t O

bs

c E

xp d

SM

R e

(C

I) f

1-99

9 A

ll ca

use

s 3

45

29

8.1

14

0-2

08

M

alig

na

nt

97

7

2.3

1

40

-14

9

Buc

cal

cavi

ty,

ph

ary

nx

3 1.

7 15

0 E

sop

ha

gu

s 4

1.7

151

Sto

ma

ch

12

9.4

1

52

-15

3

Inte

stin

e 2

5.1

154

Re

ctu

m

6 3

.3

15

5-1

56

Li

ver

an

d b

iliar

y tr

act

0

2.2

15

7 P

an

cre

as

2 3.

1 16

1 L

ary

nx

2 1.

1 16

2 L

un

g

31

20

.7

16

2-1

65

R

esp

ira

tory

CA

3

5

21

.8

170-

171

Bo

ne

+ s

oft

tis

sue

0

0.7

1

72

-17

3

Ski

n

1 0

.8

185

Pro

stat

e 7

4.8

18

8 B

lad

de

r 0

2.2

18

9 K

idn

ey

5 2

.0

19

1-1

92

B

rain

3

1.3

194

En

do

crin

e g

lan

d

1 0.

1 19

5 Ill

-def

ined

CA

3

3.3

2

00

-20

8

Lym

ph

ati

c a

nd

he

ma

to-

11

4.6

po

ieti

c ca

nce

r 20

0, 2

02

N

on

-Ho

dg

kin

's

lym

ph

om

a

4 1.

1 20

1 H

od

gki

n's

dis

ea

se

0 0

.7

20

3

Mul

tiple

mye

lom

a

3 0

.6

20

4-2

08

L

eu

kem

ia

4 2

.2

39

0-4

59

C

ircu

lato

ry d

ise

ase

1

34

11

9.6

46

0-5

19

R

esp

ira

tory

dis

ea

se

10

19.2

5

20

-57

9

Dig

est

ive

dis

ea

se

18

21

.5

78

0-7

99

Ill

-def

ined

g 19

7

.9

91

9-9

20

In

dust

rial

acc

ide

nt

1 0

.8

80

0-9

18

A

ccid

en

t (o

ther

) 2

7

17

.2

95

0-9

59

S

uici

de

24

10

.0

128

39

185

64

19

1 15

0 16

5

118

147

24

5

231

1,2

24

91

2

39

116

(10

4-1

29

) 13

4 (1

09

-16

4)

178

(37

-52

0)

23

7

(64

-60

6)

(66

-22

4)

(5-1

41

) (6

7-4

02

)

(8-2

31)

(23

-69

0)

(10

2-2

13

) 1

15

-23

0)

(3-6

57)

(59

-30

2)

(79

-57

2)

(48

-67

5)

(31

-6,8

20

) (1

9-2

66

) (1

19-4

27)

37

5

(10

1-9

57

)

64

3

(112

-1,5

87)

183

(49

-47

0)

112

(94

-13

3)

52

(2

5-9

6)

84

(5

0-1

32

) 2

40

(1

44

-37

4)

122

(3-6

80)

157

(10

4-2

29

) 2

39

(1

53

-35

6)

23

3

9.6

5

8

(37-

87)

23

3

0.3

7

6

(48-

114)

9

3

10

9.9

8

5

(68

-10

4)

8 1

0.0

8

0

(34

-15

8)

4 7

.0

57

(1

5-1

46

) 2

9

26

.4

110

(73

-15

8)

0 0

.3

--

--

0 0

.4

--

--

6 0

.7

82

2

(30

0-1

,78

9)

1 0

.7

38

7

(10

-2,1

56

) 0

0.2

--

--

0

0.7

--

--

0

1.1

--

--

0 0

.6

--

--

2 3

.2

63

(8

-22

8)

0 0

.7

--

--

1 0

.5

21

8

(5-1

,21

5)

0 1.

8 --

--

0

0.4

--

--

0

0.3

--

--

1

1.1

88

(2

-49

0)

0 0

.3

--

--

0 0

.2

--

--

1 0

.8

12

4

(3-6

91

) 0

0.5

--

--

0

0.3

--

--

2

1.2

169

(20

-61

0)

0 0

.2

--

--

0 0.

1 --

--

- 1

0.4

2

49

(6

-1,3

87

) 2

2.9

7

0

(8-2

53

) 3

1.9

15

8

(33

-46

2)

11

7.5

1

46

(7

3-2

61

) 2

2.9

6

8

(8-2

46

) 3

2.0

1

52

(3

1-4

44

) 12

7

.7

15

5

(80

-27

1)

0 0.

1 --

--

0

0.1

--

--

0 0

.3

--

--

0 0.

1 --

--

0

0.1

--

--

0 0

.4

--

--

1 0

.7

15

3

(4-8

53

) 0

0.2

--

--

1

1.5

67

(2

-37

3)

0 0

.3

--

--

0 0

.2

--

--

3 0

.8

38

0

(78

-1,1

10

)

1 0

.3

30

4

(8-1

,69

4)

0 0

.3

--

--

0 0

.8

--

--

0 0

.2

--

--

0 0

.3

--

--

0 0

.6

--

--

0 0

.0

--

--

0 0

.0

--

--

0 0

.0

--

--

1 0

.5

18

9

(5-1

,053

) 0

0.4

--

--

0

1.3

--

--

2 0

.7

30

3

(37

-1,0

94

) 0

0.6

--

--

0

1.9

--

--

2 0

.2

1,2

04

(1

46

-4,3

49

) 0

0.2

--

--

0

0.4

--

--

0

0.1

--

--

0 0.

1 --

--

0

0.3

--

--

0 0.

1 --

--

0

0.1

--

--

0 0

.2

--

--

0 0

.3

--

--

0 0

.3

--

--

0 0

.9

--

--

12

16.1

7

4

(38

-13

0)

3 8

.9

34

(7

-99)

3

2

41.1

7

8

(53

-11

0)

2 2

.4

84

(1

0-3

03

) 0

1.1

--

--

4 6

.2

64

(1

7-1

64

) 0

2.9

--

--

2

2.8

7

0

(8-2

53

) 6

8.4

7

2

(26

-15

6)

1 1.

0 9

8

(2-5

46

) 6

1.1

56

7

(20

7-1

,23

2)

14

3.0

4

65

(2

54

-78

0)

0 0.

1 --

--

0

0.2

--

--

0

0.4

--

--

0

2.2

--

--

4

3.9

10

3 (2

8-2

63

) 8

8.8

91

(3

9-1

80

) 0

1.4

--

--

2 2

.5

81

(10

-29

3)

0 5

.0

--

--

t~

t~

~.

a A

ll ca

nce

r si

tes

wit

h a

t le

ast

on

e o

bse

rve

d o

r e

xpe

cte

d c

ase

are

in

clu

de

d;

oth

er

dis

ea

se g

rou

ps

are

in

clu

de

d

if E

xp >

20

. b

Bo

eh

rin

ge

r:

1 ca

se w

ith

ca

nce

r o

f th

e t

on

sils

(I

CD

-9 c

od

e 1

46

); 2

ca

ses

of

ph

ary

ng

ea

l ca

nce

r (I

CD

-9 c

od

e 1

41 );

BA

SF

: 3

case

s o

f ca

nce

r o

f th

e t

on

gu

e (

ICD

-9 c

od

e 1

41 ),

2 c

ase

s o

f

can

cer

of

the

flo

or

of

mo

uth

, 1

case

of

can

cer

of

the

to

nsi

ls

(IC

D-9

co

de

14

6).

c O

bs

= o

bse

rve

d.

=~

d E

xp =

exp

ect

ed

. ~

. e

SM

R =

sta

nd

ard

ize

d

mo

rta

lity

rati

o.

f C

I =

95

%

con

fid

en

ce i

nte

rva

l.

g In

clu

din

g u

nkn

ow

n c

au

ses

of

de

ath

. "~

Oo c)

Tabl

e 4.

Sta

nd

ard

ize

d m

ort

alit

y ra

tio

(S

MR

) a

na

lysi

s fo

r th

e t

ota

l co

ho

rt b

y ti

me

sin

ce f

irst

exp

osu

re f

or

sele

cte

d c

au

ses

of

de

ath

, G

erm

an

ph

en

oxy

he

rbic

ide

an

d

chlo

rop

he

no

l m

an

ufa

ctu

rin

g p

lan

ts

ICD

-9

Cau

se o

f dea

th

Tota

l co

hort

co

de

13m

e si

nce

first

exp

osur

e

0- <

10

year

s 10

- < 2

0 ye

ars

20 +

ye

ars

Obs

a E

xp b

SM

R

(CI) c

O

bs a

Exp

b S

MR

(C

I) c

Obs

a E

xp b

SM

R

(CI) c

O

bs a

Exp

b S

MR

(C

I) c

1-99

9 A

ll ca

uses

48

4 47

7.9

101

(92

-11

1 )

67

89.3

75

(5

9-96

) 16

3 14

4.8

112

(96-

131

) 25

4 24

3.8

104

(91-

118)

14

0-20

8 A

ll m

alig

nant

13

8 11

5.7

119

(100

-141

) 15

16

.4

91

(51-

150)

46

35

.6

129

(94-

173)

77

63

.8

120

(95-

151

) ne

opla

sms

140-

149

Buc

cal

cavi

ty-

9 3.

1 29

5 (1

35-5

60)

0 0.

4 0

--

7 0.

9 74

5 (2

99-1

,535

) 2

1.7

114

(14-

412)

ph

aryn

x 16

2 Lu

ng

47

32.9

14

3 (1

05-1

90)

8 4.

4 18

0 78

-355

) 14

10

.1

138

(75-

232)

25

18

.4

135

(87-

200)

20

0, 2

02

Non

-Hod

gkin

's

6 1.

8 32

6 (1

19-7

10)

0 0.

4 0

2 0.

6 36

4 (4

4-1,

314)

4

0.9

425

(115

-1,0

88)

lym

phom

a 20

0-20

8 Ly

mph

atic

and

13

7.

8 16

8 (8

9-28

6)

2 1.

8 11

2 (1

4-40

4)

4 2.

3 17

0 (4

6-43

5)

7 3.

6 19

2 (7

7-39

6)

hem

atop

oiet

ic

canc

er

a O

bs =

obs

erve

d.

b E

xp =

exp

ecte

d.

c C

I = 9

5% c

onfid

ence

inte

rval

.

Pbenoxy herbicides, dioxins, and cancer

buccal cavity and pharynx (ICD-9 codes 140-149). There were nine cases, of which six occurred in cohort IV, and three in cohort I, yielding an SMR of 295 (CI = 135-560) for the total cohort and of 822 (CI -- 300-1,789) for co- hort IV alone. Most of the cases occurred in the group 10-20 years since first exposure. There was no consistent job-category pattern which might indicate a particular exposure as a causal factor for these tumors. Four of the six observed cases in cohort IV were exposed only mar- ginally to the substances of interest. No soft-tissue sarco- mas were observed while 1.2 were expected. Excess risks were observed in various other cancers in specific subco- horts, but were based on small numbers of deaths.

The Cox regression analysis for total mortality in co- hort II indicated an increased risk among smokers cf nonsmokers: relative risk (RR) (smoking)--1.89, CI = 0.75-4.7. An increased risk was observed for work- ers in the synthesis department (where exposures are higher) compared with other production workers: RR (synthesis) = 2.4, CI = 1.0-6.I. These estimates were ob- tained in a model with both variables simultaneously, ad- ditionally adjusted for age. Models with either or both variables yielded similar results. Data were too sparse for the analysis of cancer mortality.

Discussion An important issue within this cohort is the extent of exposure to TCDD and other dioxins or furans, and whether the observed excess mortality, in particular in cohort I, is due to this exposure. For two cohorts (cohort I and II), the exposure to TCDD could have been ex- pected on the basis of the production figures. This was confirmed by blood measurements which showed fairly high TCDD concentration and by the occurrence of chloroacne. However even within the cohorts the extent of TCDD exposure can differ by 1-2 orders of magni- tude. 2 It is interesting to note that the non-Hodgkin's lymphoma excess is confined to these two cohorts with an SMR of 484 (CI = 178-1,055) while in the other two cohorts no case occurred.

In cohorts III and IV, this exposure to TCDD is likely to be lower but exposure to higher chlorinated dioxins may have occurred. Less substances with known TCDD-contamination were produced or formulated. In addition, in cohort III, production started later when more strict industrial hygiene standards were applied. However, blood samples are not available and definite conclusions cannot be drawn.

We observed an unusually high heakhy-worker effect for cohort II. As usual in occupational cohort studies, the completeness of the data source is essential for the valid- @ of the results. However, we are not aware of any selec- tion errors which may have led to the omission of some

workers who died early. It must be noted, however, that the mean employment time in this cohort is very high, which may indicate a possible loss of short-term employees.

Excess risk of non-Hodgkin's lymphoma was observed in cohorts I and II. In these cohorts, exposure to TCDD and other related contaminants of 2,4,5-T and 2,4,5-TCP was probably higher than in cohorts III and IV. The synthesis of 2,4,5-TCP (and 2,4,5-T in I) began in plants I and II in the 1950s when the dioxin levels in raw materi- als and final products were higher than in the mid-1960s when plant III started the synthesis of 2,4,5-T. TheTCDD contaminated products (2,4,5-TCP and 2,4,5-T) were main products in plants I and II, whereas in plant III they were not main products, and in plant IV they were for- mulated to much lesser extent than other herbicides. In addition, chloroacne cases indicating exposure to dioxins occurred only in plants I and II.

The six observed cases of cancer of the oral cavity in cohort IV deserve particular attention. There were three cancers of the tongue (ICD-9 code 141), two cancers of the floor of mouth (code 144) and one cancer of the ton- sils (code 146). However, a common pattern of exposure is not visible from the data available. Three of the six cases had indirect exposure through working in other units in close vicinity to the production areas, one case was a maintenance worker, one case worked in the synth- esis department and one case was a supervisor. In the same cohort, there is a slight excess of bladder cancer which also could be due to exposure to aromatic amines, such as o-toluidine, ethyl toluidine, aniline, and anisidine known to have been used in the plant.

A moderate excess risk of lung cancer was observed in three out of four cohorts. There is no indication of possi- ble asbestos exposure in the plants which could have led to the observed excess in respiratory cancer deaths.

There are advantages and disadvantages in combining the resuks of the subcohorts. As indicated in the statisti- cal methods section, we present the SMR separately by cohort akhough the data were obtained using a standard protocol. It is easily possible, however, to construct the joint SMR from the data given. In Table 4 we present the SMR for all cohorts combined, similar to other mul- ticentric studies, 4 acknowledging that the variation in di- oxin exposure between the cohorts may be large, because numbers would otherwise become too small in this type of analysis. On the other hand, information on produc- tion processes and volumes enables crude classification of worker groups and plants according to occurrence of

9 • • exposure to TCDD. A combmauon of workers from the different cohorts based on job description may lead to false conclusions if the aim was to attribute possible ex- cess risks to TCDD or TEQ exposure and to provide quantitative dose-response estimates.

Cancer Causes and Control .Vol 7.1996 319

I-L. Becher et al

The exclusion of foreign workers may yield an under- estimation of the true SMR if these employees have high exposures. We do not have any evidence that this is the case, and for reasons described above, we decided to omit them from further analysis.

The half-life of TCDD in human tissue has been esti- mated to be between 7.1 and 11.3 years, z°'21 Recently, the

• • 1 5 • data from cohort I yielded an esumate of 7.2 years. It is therefore possible to back-calculate the exposure using current dioxin concentration values for risk assessment. In the United States, a similar investigation currently is being carried out using data from a US cohort. 22 We did not present in this paper further results on subgroup analysis by job title or by duration of employment. We think that a further analysis, based on future dioxin measurements, is more meaningful.

In conclusion, we examined cancer risk in four co- horts in Germany exposed to phenoxy herbicides, chlor- ophenols, and dioxins. We found an overall excess risk of all cancers, and excess risks for non-Hodgkin's lympho- ma, lung cancer and cancer of the buccal cavity. The risk of lymphoma occurred in these cohorts where higher levels of dioxin were expected.

Acknowledgements--We thank Professor Rainer Frent- zel-Beyme for fruitful discussions during all phases of the project and for raising our interest in this topic. We also thank the companies for providing the data and for their continuous help in clarifying data problems. In particular we thank Professor Zober, Mr Messerer (BASF Ludwig- shafen), Dr Steinmann-Steiner-Haldenst~itt (Bayer Uerdin- gen), Dr Schr6der (Bayer Dormagen), Dr Deckers, and Mr Drinnenberg (Boehringer Ingelheim) for their help. We thank Ms Renate Angerer and Mr Michael Hauptmann for effective data management and help with the analysis, Mr Karl-Heinrich Adzersen and Dr J. Ott for helpful comments on an earlier draft of this paper.

References

1. International Agency for Research on Cancer. IARC Monographs 1987. Lyon, France: IARC, 1987; IARC Monogr Eval Carcinog Risks Humans, Vols. 1-42, Suppl. 7.

2. Manz A, Berger J, Dwyer JH, Flesch-Janys D, Nagel S, Waltsgott H. Cancer mortality among workers in a chemical plant contaminated with dioxin. Lancet 1991; 338: 959-64.

3. Zober A, Messerer P, Huber E Thirty-four year mortality follow-up of BASF employees exposed to 2,3,7,8-TCDD after the 1953 accident. Int Arch Occ Environ Health 1990; 62: 139-5Z

4. Fingerhut MA, Halperin WE, Marlow DA, et at. Cancer mortality in workers exposed to 2,3,7,8-tetrachlorodiben- zo-p-dioxin. NEnglJMed 1991; 199: 212-8.

5. Bertazzi A, Pesatori AC, Consonni D, Tironi A, Landi MT, Zocchetti C. Cancer incidence in a population acci- dentally exposed to 2,3,7,8-tetrachlorodibenzo-para-diox- in. Epidemiology 1993; 4: 398-406.

6. Saracci R, Kogevinas M, Bertazzi PA, et al. Cancer mor- tality in workers exposed to chlorophenoxy herbicides and chlorophenols. Lancet 1991; 338: 1027-32.

7. Kogevinas M, Kauppinen T, Winkelmann R, et al. Soft- tissue and non-Hodgkin's lymphomas in workers exposed to chlorophenoxy herbicides, chlorophenols and dioxins. Two nested case-control studies. Epidemiology 1995; 6: 396-402.

8. Nagel S, Berger J, Flesch-Janys D, Manz A, Ollroge I. Mortality and cancer mortality in a cohort of female workers of a herbicide producing plant exposed to poly- chlorinated dibenzo-p-dioxins and furans, lnformatik Biometrie Epidemiol Med Bio11994; 25: 32-8.

9. Kauppinen T, Kogevinas M, Johnson E, et al• Chemical exposure in manufacture of phenoxy herbicides and chlorophenols and in spraying of phenoxy herbicides. AmJlnd Med 1993; 23: 903-20.

10. Becher H, Wahrendorf J, Angerer R. A cohort study on persons exposed to phenoxy acid herbicides and their contaminants in Germany - design and first resuks. Che- mosphere 1992; 25: 1007-14.

11. World Health Organization. International Classification of Diseases, Ninth Revision. Geneva, Switzerland: WHO, 1977.

12. Thiess AM, Frentzel-Beyme R, Link R. Mortality study of persons exposed to dioxin in a trichlorophenol-process accident that occurred in the BASF AG on November 17, 1953. AmJlnd Med 1982; 3: 179-89.

13. Flesch-Janys D, Becher H. Methodological issues in the risk assessment for dioxins and cancer on the basis of epidemiological cohort studies. Informatik Biometrie Epidemiol Med Biol 1994; 25: 283-91.

14. Jansing PJ, Korff R. Dioxinbelastung und Gesundheits- st6rungen - eine Untersuchung 2,3,7,8-TCDD-exponier- ter Arbeimehmer. ZblArbeitsmed 1992; 42: 494-7.

15. Flesch-Janys D, Becher H, Gurn P, et al. Elimination of polychlorinated dibenzo-p-dioxins and dibenzo-furans (PCDD/F) in occupationally exposed persons.J Toxicol Environ Health 1996; 47: 363-78.

16. Coleman M, Douglas A, Hermon C, PetoJ. Cohort study analysis with a Fortran computer program. Int J Epide- mio11986; 15: 134-Z

17. Breslow NE, Day NE. Statistical Methods in Cancer Re- search, Vol IL The Design and Analysis of Cohort Studies. Lyon, France: International Agency for Research on Cancer, 1987; IARC Sci. Pub. No. 82.

18. Cox DR. Regression analysis and life tables (with discussion)• J R Stat Soc A 1972; 34: 187-220.

19. Checkoway H, Pearce N, Crawford-Brown DJ. Research methods in occupational epidemiology. New York, NY (USA): Oxford University Press, 1989.

20. WolfeWH, Michalek JE, Miner JE, et aL Determinants of TCDD-half-life in veterans of operation Ranch Hand. J Toxicol Environ Health 1994; 41: 481-8.

320 C a n c e r C a u s e s a n d C o n t r o l . V o l 7 . 1 9 9 6

Phenoxy herbicides, dioxins, and cancer

21. Pirkle JL, Wolfe WH, Patterson DG, et aL Estimates ofthe half-life of 2,3,7,8-tetrachlorobenzo-p-dioxin in Vietnam Veterans of operation Ranch Hand. J Toxicol Environ Health 1989; 27: 165-71.

22. Salvan A, Dankovic D, Stayner L, Gilbert S, Steenland K,

Fingerhut M. An approach to the quantitative assessment of cancer risk in relation to occupational exposure to di- oxin: limitations and variability to TCDD exposure esti- mates. Informatik Biometrie Epidemiol Med Bio11994; 25: 292-300.

Cancer Causes and Control.Vol 7.1996 321