liver cancer among women in laundry and dry-cleaning work in Denmark.

LYNGEE, THYGESEN L. Primary livercancer among women in laundry and dry-cleaning work in Den mark. Scand J Work Environ Health 1990;16:108-12. Tetrachloroethylenehas been the most commonly usedsolvent in drycleaning in Denmark since the late 1950s. A cohort of laundry and dry-cleaning work ers was identified from the Danish Occupational Cancer Register for the study of cancer incidence of persons exposed to tetrachloroethylene. The Register includes cancer incidence data for a 10-year period for 8567women and 2033 men employed in laundry and dry-cleaning in 1970. A significant excess risk wasfound for primary livercanceramong the women,with7observedand 2.1 expectedcases(standardized incidence ratio 3.4, 95 070 confidence interval 1.4-7.0). No case of primary liver cancer was observed among the men, for whom the expected value was 1.1. The excessrisk of primary liver cancer among the women is unlikely to be explained exclusively by excess alcohol consumption.

An inc reased risk for primary liver can cer among wo men exposed to sol vents was fo und in two casereferent studies in Finland (l, 2) . No increase was fo und a mo ng exposed men. A re view of th e exposure pattern of the case s sho wed that nine of the 13 women had been exposed to chlorinated hydrocarbons, whereas painting was the mo st common solvent exposure for the men. With thi s informat ion as background, data a re reported in this study on the ca ncer incidence of per sons wo rking in Danish dr y-cleaning shops, where tetrachloroethylene ha s been th e mo st commonly used solvent and where the work for ce has included both men and women.

Cohort
A cohort of laundry and dry-cleaning workers was identified from the Danish Occupational Cancer Register for a lO-year follow-up stud y of cancer incidence in the 1970 census population (3) . In the census the ind ustry code 860 specified " la und ries, cleaning and dyeing." A person with th is work code could be either a self-employed per son, a fa mily wo rker (ie, a spouse), or a n employee and then identified either by the occupa tio n code 411 " laundry wo rker , ironer" or by th e occupation code 380 "factory hand." The census code s do not allow a distincti on to be made between laundr ies on the one hand and dr y-cle aning shops on the

Exposure
There were 2886 laundries and dr y-cleaning sho ps in Denmark in 1970 (4). This number is close to the number of 2434 self-em ploy ed persons in th e cens us . The workplaces were small with the average size of the work force being 3.7 (ie, 10 600/2886). Married couples were working to gether in about one-third of the sho ps (ie, 830/2434). Data were not a vailable on the possible division o f labor in these small wo rks ho ps . Out of the 2886 shops, 695 (Danmarks Statistik, per sonal communicat ion) were known to be dry-cleaning and dyeing sho ps , where dyeing was of quantitatively minor importance. White spirit was the main solvent used in dry cleaning in Denmark after World War II. The consumption of white spirit, however, decreased when automatic cleaning machines were introduced in the late 1950s. Tetrachloroethylene then became th e most commonly used solvent , supplemented by trichloroethylene for the cleaning of workclothes, and fluorocarbons II (tri chlo ro fl uo rmetha ne) and 113 (1,1 ,2-trichlo ro-I ,2,2-trifluo retha ne) for the cleaning of fur coats, etc (5). The co nsum ptio n of tetrachloroethylene in Denmark increased to a maximum of 3500 t in 1973 and decr eased to a bo ut 1500 t around 1980. About 80-95 0,70 of the tetrachloroethylene was used in dr y cleaning, and the decrease in consumption wa s due to a decrease in the number of dry-cleaning shops and to the recyclin g of the cleaning fluid in the rem aining sho ps (6). It was a common practice in the dry-clean ing shops to co nt ro l fo r the qualit y o f the drying process by smelling the clothes when they were taken from the mach ine. Measurements from 1979-1980 sho wed con -centration s of tetrachloroethylene of the order of 1000-7000 ppm in 11 sam ples of clothes upon their removal from the dr y-cleaning machine and concentrat ions on th e ord er of 0-100 ppm in 67 sam ples (7).

Cancer incidence
The cancer cases ob served in th e cohort were identified by link age between the 1970 census data and the Danish Cancer Registry data through the use of the personal ident ification number. The numbers of cancer cases expected in the cohort were calculated by multipl ying the per son-year s at risk during the lO-year follo w-up period in each five-year age group with th e site-specific incidenc e rates calculated in th e same way for all per son s economically active in 1970. Th e rat io between th e total number of ob served and the total number of expe cted cases is a standardized incidence ratio (SIR), and the 95 070 two-tailed confidence int erval (95 0" /0 CI ) was calculated on the assumption that th e total number of observed cases up to 30 followed a Poisson distribution (8), and for total numb ers above 30 the distribution was normal (9). Table 2 shows the cancer incidence in 1970-1980 for persons aged 20-64 years and wor king in laundries and dr y-cleaning shops in Denmark in 1970. A total o f 378 ca ncer cases were ob served among the women , for whom 397.3 were expected (SIR 1.0, 95 % CI 0.9-1.1), and 132 cancer cases were observed among the men , for whom 104.5 were expected (SIR 1.3, 95 % CI 1.1-1.5). The tabl e includes all cancer sites for which th ere were at least five observed cancer cases fo r the men and women co mbined . A significant excess risk was found for primary liver cancer among the women with 7 ob served and 2.1 expected cases (SIR Table 1. Number of persons aged 20-64 years and engaged in laund ry and dry-clean ing work in Denmar k in 1970.

Men
Women Tota l 3.4 ,95 % CI 1.4-7.0). No case of primary liver cancer was ob served among the men , for whom 1.1 cases were expected. A significant excess risk of cancer of the pancreas was seen when the figures for the men and women were combined. Twenty-two cases were ob served , and 13.1 were expected (SIR 1.7,95 % CI 1.1-2.6). None of the other cancer sites showed incidence figures for workers in laundries and dry cleaning that differed significantly from the average level of all economically active persons.

Discussion
Tetrachloroethylene has been found to cause hepatocellular carcinomas in mice both after or al administration and after inhalation, and exposure of rats by inhalation produced an increased incidence of leukemia (10). The pot ent ial carcinogenic risk of d ry cleaners has the refore been the focu s of severa l epidemiologic studies.
Blair A statistically non significant excess risk of liver cancer (4 ob served versus 1.7 expected cases) was found in one of th ese studi es (II). Th e observed to expected ratio in the other studies was 4:4.5 (12), I: 1.9 (13), 0:3.5 (14), and no data (15), respectively.
In the Swedish Can cer En vironment Register for 1961-1973 (16) th e inciden ce of cancer of the liver and gall bladder among laundry and dr y-cleaning wor kers was close to the average for the Swedish populat ion within the same region s. The observed to expected rat io was 17: 14.8, and th ere was no difference between the men and the women.
Two proportional mortality studies ha ve also been undertak en of workers exposed to tetra chloroeth ylene, among other solvent s, in metal degreasing. Blair (17) conducted a proportional mortality study of 1292 death s among white men identifi ed from ob ituaries in a union journal for metal workers from 1951-1969. Dubrow & Gute (18) studied the proportional mortalit y of jewelr y workers in Rhode Island in 1968-1978 , where a tot al of 1372 men and 1111 women were included and degrea sing was predominantly a male oc-cupation. A statistically significant excess risk of pr imar y liver can cer was found among the metal degreasers (5 observed versu s 1.8 expected) and among the male jewelr y workers (6 observed versus 2.0 expected) but not among th e female jewelry workers (I ob served versus 2.3 expected).
Exposure to tetrachloroethylene has also been considered in several case-referent studies on primary liver can cer. Sternhagen et al (19) studied New Jersey residents diagnosed with primary liver can cer in 1975-1980, and employment in "laundering, cleaning and other garment service" showed a significantl y elevated relat ive risk (RR) for men (RR 2.5), wherea s employment as a " cleaning service worker" showed a significantly elevated risk for women (RR 4.3). Hardell et al (20) studied men diagnosed with liver cancer in 1974-1981 in no rthern Sweden and found a high level of expo sure to or ganic solvents to be associated with a relat ive risk of 1.8. But onl y one case involved exposure to tetrachloroethylene. Austin et al (21) studied hepatocellular carcinomas in persons aged 18-84 years in five centers in the United State s. In none of their cases was pre vious employment in the laundering and clean ing indu stry reported . Suarez et al (22) studied primary liver cancer deaths among Texas male residents aged 20 years and older in 1969-1980. Employment in " dry-cleaning service" showed an odd s ratio close to unity.
Hernberg et al (I) studied primary liver cancer cases reported to the Finnish Can cer Registry in 1979-1980 a nd found expo sure to solvents to be associated with an odds ratio of 2.3, with the excess of exposure among cases confined entirely to women. Hernberg et al (2) repeated the study, including primary liver cancer cases reported to the Finnish Cancer Registry in 1976-1978 and 1981. In th is study expo sure to solvents gave an odd s ratio of 0.6 for the men and 3.4 for the women, for whom expo sur e to chlorinated hydro carbons was reported for 9 out of 13 of the women.
Our study showed a significantly elevated risk for primary liver can cer among the women employed in laundry and dry cleaning in Denmark with 7 observed and 2.1 expected cases (SIR 3.4, 95 % CI 1.4-7.0) . No case of primary liver cancer was ob served among the men working in this indu stry. The male work force was, however, relati vely sma ll, and onl y 1.1 cases of pr imar y liver ca ncer were expected .
The majority of primary liver can cer cases in Denmark is assumed to be associated with alcohol consumption (23). It is, however, not likely that the excess risk of primary liver cancer found among the women employed in laundry and dr y-cleaning can be explained exclusively by an excess alcohol con sumption . In the Dani sh Occupational Cancer Register no excess risk of primary liver cancer (3 observed versus 2.6 expected) was found among 13 000 women working in hotels and restaurants, a nd no case of primar y liver cancer was observed among 2000 women working in breweries.
In order to pinpoint further the possible reasons for the excess ris k of liver cancer among women in laundry and dry-cleaning work , it would be desirable to undertake a nested case-referent study with a thorough collection of exposure data. Current confidentiality rules do not, however, allow us to retrieve indi vidual records from the linked regi sters .
The proportionate mortality and cohort studies of laundry and dr y-cleaning workers from the United States have pointed to an excess ri sk of kidney cancer. Th e ob ser ved to expected ratios were 2: 1.0 (11), 7:2.7 (12), 7:1.9 (13), and 4:2.0 (14), and no data (15), respe ctively. Our study did not point to an excess risk of kidney cancer among Danish laundry and dry-cleaning workers, as we found II ob served and 12.5 expected cases.
The US st udies have also found a consistent excess risk for cervical cancer among women work ing in laundry and dry-cleaning (11)(12)(13)(14). Neither was this observation supported by our Danish data, in which there were 35 ob served and 41.1 expected cases . The excess risk in the US studies has been suggested by the authors to be due to the low pa y received for laundry and drycleaning work in the United States, and the wom en thus coming from lower socioeconomic groups (II , 14). The situation is different in Denmark with the many small workshops, and with 21 % of the women working in thi s industry being either self-employed or family workers.
Skin cancer was in excess in three of the US studies (11-13). Our Danish data did not support thi s observation either in that both the incidence for malignant melanoma (IQ ob served and 14.0 expected cases) and for nonmelanoma skin cancer (35 observed and 44.8 expected case s) were below the average for the Danish population.
An excess risk of bladder cancer ha s also been observed in three of the US studies (12,(14)(15). Neither did this observation hold true in our Danish data, in which there were 14 observed and 18.9 expected cases.
The only cancer site, apart from primary liver cancer in women, for which we found a statistically significant excess risk in Denmark was pancreatic cancer (SIR 1.7, 95 070 CI 1.1-2.6). This observation is only weakly supported by the results of the US studies, in which the observed to expected ratios for pancreatic cancer were 4:3.1 (11), 9:7.7 (12), 3:5.6 (13), 11:6.4 (14), and no data (15), respectively.
Th e existing studies of the cancer incidence and mortality among laundry and dry-cleaning workers thus do not allow firm conclusions to be drawn about the po ssible cancer risk in this occupational group. However, although the results from these and other studies on primary liver cancer in relation to tetrachloroethylene exposure are inconsistent, they are nevertheless intriguing. The positive results come from studies of metal degreasers with high and mi xed exposure (17,18), from relatively large case-referent studies (1,2,19), and from the present record-linkage study based on incidence data.
Retrieval of data on dry cleaners from other casereferent and record-linkage studies with incidence data, and an updating and collection of exposure data from the existing studies could shed further light on the possible risk of primary liver cancer following exposure to tetrachloroethylene.