Aplastic anemia, leukemia and other cancer mortality in a cohort of shoe workers exposed to benzene.

Aplastic anemia, leukemia and other cancer mor tality in a cohort of shoe workers exposed to benzene. Scand J Work Environ Health 1989;15:313-318. Benzene is a well documented carcinogen for the hematic and lymphopoietic system, and experimental research confirmsits carcinogenicity for tumors of other sites. This report presents the results of a histori cal cohort study in a shoe manufacturing plant in Florence where cases of aplastic anemia and leukemia were reported in the 1960s.A total of 1008 men and 1005 women were considered eligible members of the cohort. For total mortality, comparing the rates of the cohort with the national rates, the standardized mortality ratio (SMR) was 79 for the women and 95for the men. For the men excessesof risk for aplastic anemia [SMR 1566;95 % confidence interval (95 % CI) 547-3264] and leukemia (SMR 400,95 % CI 146-870) were observed. The increased risk occurred among workers first employed during the period in which benzene was used, but the expected number of cases in the subsequent period was too small to evaluate whether any reduction in risk had occurred. No increasing pattern with duration of employ ment was discernible.

One o f the monogr aphs of the Int ernati on al Age ncy for Resea rch on C a ncer (IARC) has evalua ted as "su ff icie nt" th e epidemiologic evidence on the ca rcinogenicity of ben zene (I) . The o bse rve d to xic a nd ca rcinogenic effec ts o f exposure wer e mainl y related to th e hem atopoieti c sys tem , including pan cytop eni a and leukemia, mostl y of the acute myelo id typ e. A majo r co nt ri butio n to the evide nce was from case re po rts of a plas tic anemia and leukemia am on g wo rker s who were occupationally exposed to ben zene in Milan a nd Pa via (2) and from cas e series from Turkey (3). Prominent among the industries involved was shoe manufacturing, in which benz en e was widely used as a so lvent fo r glues.
The IARC monograph on benzene (I) reported that there wa s no clear evidence of an increas ed risk for other malign an cies among exposed humans. However, experiment al dat a have shown evidence that benzene is a mul tipotential ca rcinogen in rats (4 , 5). An IARC monogr aph on shoe workers (6) reported a n excess of risk fo r na sal adenocarcinoma and suggest ed an associatio n with bladder cancer. Some other ca ncer sites we re a lso co ns idered in th e monograph , but fo r none of these was the evidence considered to suppo rt a clear ca usa l rel at ionship . I  In Floren ce (T uscany), shoe manu factu rin g is a tradition al industri al acti vity and, follo wing th e repo rts on leukemi a , a n increased effort ha s been m ad e to monitor th e exp osure level s and health sta tus of shoe workers. At th e end of 1963, a national law lim ited the use of ben zene in shoe glues to less than 2 % o f the total so lvent.
Although there ha ve been reports of leukemia among shoe wo rkers in Flo rence and other Italian citie s, no epid emi ologic stud y has ever been ca rried o ut on thi s su bject.
In th e pr esent paper, we report on a historical cohort study of workers employed in the lar gest shoe manufacturing plant in Florence. It was alr eady known that during the I960s six deaths from aplastic an em ia or leukemia had occurred among the workers. Th e study had th e follo wing two goals: (i) to describe aplastic anemia and leu kem ia mortality among shoe workers during a nd aft er the period of benzene exposure and (ii) to eva lua te the mortality risk fro m other neoplasm s following exposure to ben zene.

The plant
The so urce o f info rmatio n was the records of th e lar gest shoe manufacturing company in Florence . This company existed prior to the Second World War in four separate lo cat ions and in 1954 was consolidated at the pr esent locat ion , where all the wo rk is now ca rried out.
The register of employees has always been centralized for the entire company so that now workers cannot be identified according to specific buildings in which they worked from the beginning of the plant until 1953. From the beginning of 1954 on, the main activities have all been carried out on one floor (the ground floor) of the main building.
Between the period 1961-1962 the main building was completely restructured. During this time, characterized by the highest levels of exposure to benzene, all activities were carried out in two communicating departments. The first consisted of the storerooms and cutting section, while in the second the main shoemaking activity was performed.
At the end of 1961, the building was considerably extended, and since then the activities have been subdivided as follows: ground (or first) floor: shoemaking department, consisting of five workshops (about 400 employees), storage department; second floor: finishing department (60 employees); third floor: assembling department (150 employees); fourth floor: cutting department (90 employees); fifth floor: offices (10 employees).
Until the renovations at the end of 1961, the only ventilation in the plant was supplied by windows and fans. In April 1962 the installation of mechanical air conditioning and local ventilation systems was completed.

Assessment of benzene exposure
Original plant production records, factory inspection reports, and information from the glue suppliers were reviewed for this study. In these documents it was reported that in this shoe factory the use of glues containing benzene began between 1950 and 1955, and one report specifically referred to 1953. This information can be considered reliable because during World War II and immediately afterwards coal-based chemical products such as benzene were not easily available in Italy, and the usc of "latexes" in the glues was prevalent.
Data available for the years 1957-1960 (probably a good estimate of the situation from 1953 to 1960) show that the three main glues used in shoe manufacturing contained more than 70 070 benzene by weight. It was estimated that 20 g of glue was used for each pair of shoes for a total of 2200 pairs per day, corresponding, on the average, to 42.5 kg of glue used each day in the plant. The glues were kept in open containers with an opening of 15 em; only after 1962 were these containers replaced by cans with a small opening (about 8 em in diameter).
In 1963, when the law limiting the use of benzene came into effect, the commercial product formula was changed. Hexane became the most important solvent, and the glue contained solvent with less then 2 % benzene. Nevertheless, old products, including those which were benzene-based, were still used until 1964. Glues were primarily used in the shoemaking department, sel-314 dom in the finishing department, and no usc was reported in the cutting and storage departments.
As the main part of the production process occurred on the ground floor, indirect exposure of the workers was possible. Using the plant history, we defined the following four periods of benzene exposure: (i) before 1953, prior to the introduction of benzene (considering the uncertainty of the exposure information, members of the cohort who left the plant before 1953 were classified as "unlikely" to have been exposed); (ii) the subsequent years 1953-1962, which were considered to comprise the high exposure period; (ii) after 1962 until the end of 1964, period of "intermediate" exposure because benzene usc was declining substantially; (iv) after 1964, the time of very low levels as a result of the 1963 law. Subjects employed after 1964 were considered unexposed.

Study population
All individuals ever employed at the plant from 1939 to the end of the study period (31 December 1984) and still employed in or after January 1950 were considered members of the cohort. From the plant register, demographic information and the dates of starting and leaving work in the plant were obtained. The vital status of the cohort members was ascertained through 31 December 1984 from the official register of the municipality of residence. From the same register copies of death certificates were obtained. Causes of death were classified from the death certificates according to the eighth revision of the International Classification of Diseases (ICD) (7) by two doctors, one of whom carries out coding for the Tuscan Mortality Registry. The national Italian mortality rates, specific for cause, age, sex, and calendar period and computed with the use of the mortality data bank of the World Health Organization, were used as reference.

Additional information on cause of death
For death certificates mentioning leukemia, aplastic anemia, buccal-pharynx neoplasm, or nasal cancer, more-detailed information was sought from other sources, including the records of the main hospital of the town of Florence and previous case reports or casereferent studies carried out in the Florence area.

Analysis
The statistical analysis was performed with the personyears program (8). We obtained the number of deaths expected in the cohort by multiplying the accumulated person-years by the national reference rates specific for cause, age, sex, and calendar year, and the standardized mortality ratio (SMR) values were calculated. The 95070 confidence interval (95 % CI) of each SMR was computed on the basis of the Poisson distribution according to Byar's formula (9).
An analysis by duration of exposure and time since the date of first exposure was carried out for the subjects who worked in 1953-1964. Three ca tego ries of duration of exposure were considered, ie, < I year, 1-5 years, and > 5 year s.

Results
A total of 3104 employees were regi stered in the factory records from I January 1939 to 31 December 1984. Restriction to employment on or after I Jan uary 1950 left 1008 men and 1005 women in the st udy cohort. Table I shows the vita l status of the cohort members, whether alive at the end of the study, known to be deceased, or lost to follow-up .
Results were o btained both wit h the subjects lost to follow-up being considered alive at the last know n da te and with them consi dered alive at the end of the study. T he di fference in person-years was 8.5 ala for the men and 11.8 ala for the women. Th e overall results differed litt le between the tw o met hod s, and the first was used in the resu lts that follow.
Observed and expected deaths and the SMR va lues for selected causes are presented in table 2 for both sexes. Among the women a total of 5 1 observed deaths occurred versus the 64.3 expected (SMR 79) . For four subjects the cause of death was not available. The case defined as a buccal-pharynx neo plasm was a rhinopharyngeal neoplasm (lCD 147). A case of mu ltiple myeloma (IC D 203) was included in the group of "other lymphatic and hem a to po ietic neoplasms." One case of aplastic anemia occurred among the women versus 0.2 expected . T he woman involved worked in the plant from 1951 to 1962 and died in 1962.
Among the men a total of 171 deaths was observed versus 180.7 expected (SM R 95). The cau se of deat h was unknown for 22 men . Fewer observed than expecte d deaths for respiratory and digestive tract diseases and accidents occ urred among the men . An SM R of 1566 was observed for blood diseases (95 tJl o C I   In cluded in "other lymphati c and hematopoietic neoplasm s" was a case o f non-Hodgkin ' s lymphoma . Among th e men th e digesti ve tra ct neo plasms show ed an SMR of 227 (95 alo C I 147-335), bUI o nly th e sto mach ca ncer excess was sta tistica lly significa nt (SMR 240, 95 % CI 137-3 78).
Further a na lysis was ca rr ied out fo r th e subco hort o f men with "high" and " intermed iate" levels of exposure. Th ese subject s, at work in the peri od 1953-1964 a nd co nsidered exposed to ben zene , were cla ssified int o th ree gro ups by durat ion of exposure . A ll the ca ses o f a plas tic an em ia a nd leukemia wer e o bserve d in th is subco ho rt , The res ults for leuk em ia a nd a plastic a ne m ia a re show n in ta bles 3 and 4 , subdiv ided by year s of duration of exp osure to ben zene a nd by risk inte rval (time interval since first expo sur e), where I Januar y 1953 was consid er ed the date of first exposure for those who were employed before thi s date. The same a na lysis was ca rrie d o ut for stomach ca ncer. T he SM R for sto mach ca ncer was 425 (95 alo C I 114--1089) for th e un exp osed wo rke rs versus a n SMR of 2 12 (95 ala C l 112-362) for th e expose d subco ho rt .
Add itio na l evide nce a vail abl e from a case -refere nt st udy· carried o ut in the Pro vince o f Floren ce (11) a llowed us to classify one ca se a mo ng the men as a " nasa l adenoca rcino ma ." From clinical records two ap lastic anemia dea ths and on e leukemia de ath wer e reclassified as ac ut e er ythrorn yelosis, co ns idered as typ ically related to benzene expos ure (2). The two mo re recent death s fr om hernol yrnph opo ieti c tumors (on e in 1979 a nd o ne in 1982) were du e to non-H od gkin 's lymphom a an d ch ro nic lymphati c leukem ia .

Dis c ussi o n
T he nu mber o f su bje cts lost to fo llo w-u p and the lack of information o n some causes o f death are limit ations of th is cohort study . The losses were due to incompleteness o f the a rc hives o f the municipalities invo lved in the stud y.
Th e tot al mort alit y o f the shoe wo rke r co ho rt sho wed a deficit fo r both sexes . Thi s frequen tly o bserved result is usu all y expl ain ed as bein g du e to th e "healthy worker effect." For the men the low SMR va lues for the major causal gro ups cou ld a lso be partiall y expla ined by a lack of information on th e ca use o f death for 22 dece ased subjec ts .
Regional mortality rates are not available in Ital y for the whole study period and th erefor e were not used as th e reference.

A plastic anemia and leukemia
Thi s study co n firms the stro ng tempor al relati on ship betw een th e occurrence o f aplastic anemia and leukemia a mo ng shoe wo rkers and th e use of glues co ntaining benzen e.
Becau se of the expected number o f cases of leuk emia and aplastic anemia among th e subjec ts beginning work a fter 1963, it is not pos sible to mak e a conclusive sta te ment about the risk remo val and its relationship to th e possible effectiveness o f th e Italian regulation o f 1963. Nevertheless, it is worthwhile to note that, out of the 14 cases (involving 12 men and 2 women) ob served , none occurred among th ese subjec ts.
Th e abse nce of an excess of risk for leuk emia and aplastic a nem ia among the women sugges ts either a di fferential susceptibility to ben zene by sex or lower expo sure levels fo r the women th an th e men. Th e po wer o f th is study and th e info rm at ion on expos ure levels associa ted with specific jobs are not su fficient to clarify thi s issue. Risk did not incre ase clearly with durati on o f work, con side red as an ind icator of do se, and there was th erefore no evidence of a do se-response relati on ship. This result could be du e to the poor classifica tion of the dose of expo sure . A mor e-de tailed job histor y descript ion will be co llected in view of the semiqu antitati ve analysis of the exposur e levels.
An oth er reason for the lack of relationship between risk and exposure co uld be that the expo sure levels were high eno ugh to produce sub stantial excesses even a fte r shorter expos ur e intervals.

Digestive tract neoplasms
An association between stomach , rectal, and digestive tract can cers in the leather and shoe industry has been suggested in previous epidemiologic studies (12,13,14). In th e pr esent study an elevated SM R fo r digestive tract ca ncers and, spec ifically, for sto mach ca ncer was sho wn both for the whol e coho rt and sepa ra tely fo r th e two subco horts (expo sed a nd une xposed).
In th e Tu scan region , where the st udy was ca rried o ut, there is a higher mortality rat e for sto mac h cancer than for th e co untry as a whol e. Furth erm ore the SMR for st omac h can cer was high er in th e un exposed subcohort than in the exposed gro up . Therefore th e excess of sto ma ch ca ncer doe s not seem to be relat ed to benzene exposure. Theinformation from the present study is not su ffici ent to evaluate whethe r or not th e excess was due to occup at ional or related risk factors or simply du e to an inappropriate choice of a reference populati on .

Nasal cancer
An excess of risk of na sal adenocarcinoma has been well described for shoe workers (5,15). Onl y one ca se occurred among the members of our cohort. Two more cases of this rar e tumor occurred among th e work ers registered in th e plant in the period befor e 1950, a nd the y ar e not included in this cohort.

Concluding remarks
The se results con cern the first true cohort study of shoe mak ers in Italy, a co untry in which a well-documented use of benzene occurred in shoe production a nd severa l case repo rts had suggested a co nseque nt risk of hem atopoietic disea ses. Our results con firm the very high SMR va lues for leukemia and apl asti c a nemia among Italian shoe workers and are fully co nsistent with an association with exposure to benz ene in glues. The effecti veness of the Italian law of 1963, which inhibited the usc o f glues with a benzene content of o ver 2 Ufo, in preventin g the occurrence of new cases o f these diseases is also sugg ested by our results, but the conclusion ca nno t be co nclusively evaluated .