Document gbKEGKz2B39520EV7xkrdbBjN

AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 33:282-292 (1 998) Cancer Risks Among Workers Exposed to Metalworking Fluids: A Systematic Review Geoffrey M. Calvert, MD, MPH,*ElizabethWard, PhD, Teresa M. Schnorr. PhD. and Lawrence J. Fine, MD, DrPH Metalworkingfluids (MWFs) are commonly used in a variety of industrial machining and grinding operations. The National Institute for Occupational Safety and Health (NIOSH) estimates that more than one million workers are exposed to MWFs. NIOSH conducted a comprehensive and systematic review of the epidemiologic studies that examined the association between MWF exposure and cancel: Substantial evidence was found for an increased risk of cancer at several sites (larynx,rectum,pancreas, skin, scrotum, and bladder) associated with at least some MWFs used prior to the mid-1970s. This paper provides the evidence pertaining to cancer at rhese sites. Cancer at those sitesfound to have more limited or less consistent evidencefor an association with MWF (stomach, esophagus, lung, prostate, brain, colon, and hematopoietic system) will not be discussed in this paper but are discussed in the recent NIOSH Criteriafor a Recommended Standard-Occupational Exposure to MWFs. Because the changes in MWF composition that have occurred over the last several decades may not be suficient to eliminate the cancer risks associated with MWF exposure, reduetions in airborne MWF exposures are recommended. Am. J . Ind. Med. 331282-292, 1998. 0 1998 Wiley-Liss, I n c t KEY WORDS: industrial oils; pancreatic cancer; laryngeal cancer; rectal cancer; bladder cancer; scrotal cancer; skin neoplasms; polycyclic aromatic hydrocarbons; nitrosamines; occupation INTRODUCTION Metalworking fluids (MWFs) are commonly used in a variety of industrial machining and grinding operations. MWFs (Le., cutting oils, machining fluids, lubricants, and coolants) are primarily used to cool and lubricate both the tool and the working surface. MWFs are also useful for providing corrosion protection, and removal of metal chips and swarf. There are four major classes of MWF: straight oil MWFs (also called insoluble oils), which are made from naphthenic or paraffinic oils and contain no water; soluble oil MWFs, which are naphthenic or paraffinic oils emulsified Division of Surveillance, Hazard Evaluationsand FieldStudies,National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention, Cincinnati, Ohio 'Correspondence to: Geoffrey M. Calvert, National Institute for Occupational Safety and Health, 4676 Columbia Parkway, R-21, Cincinnati, OH 45226; E-mail: jac6@cdc.gov Accepted 21 October 1997 in water; synthetic MWFs, which are chemical fluids containing no petroleum base oils; and semisynthetic MWFs, which are emulsions containing small amounts of oil, sometimes distinguished from soluble oil MWFs by the degree of emulsification. The National Institute for Occupational Safety and Health (NIOSH) estimates that more than one million workers are exposed to MWFs (NIOSH; unpublished data). In recent years, MWF exposures have been the subject of study in several epidemiologic investigations. However, we are not aware of any published comprehensive reviews of the cancer findings from these studies. As such, NIOSH conducted a comprehensive and systematic review of the epidemiologic studies that examined the association between MWF exposure and cancer. The findings of this review can be found in the recent NIOSH Criteria for a Recommended Standard-Occupational Exposure to MWFs [NIOSH, 19981. The purpose of this paper is to summarize the evidence for cancer of the skin, scrotum, larynx, rectum, pancreas, and bladder. o 1998Wiley-Liss,Inc. 'This article is a US Governmentwork and, as cirrh is in the nrihlir: rinmnin in thn llnitpri S t a b s nf AmPrirn Cancer Risks from Metalworking Fluid Exposures 283 METHODS RESULTS In order to be included in this review, an article had to be published in a peer-reviewed journal. Articles were identified from computerized database searches,from recommendations made by reviewers of earlier drafts and from references cited in relevant articles. Studies providing data on the association between MWF exposure and cancer were grouped into three categories based on their study design: (1) retrospective cohort mortality and cancer incidence studies of MWF-exposed cohorts and associated nested case-control studies; (2) proportionate mortality ratio (PMR) studies of occupational groups exposed to MWF and associated case-control studies; and (3) population-based studies (primarily case-control interview studies of specific cancer sites which examined cancer risks associated with MWF exposure, or with occupations likely to have MWF exposure, i.e., metal machinists, grinders, turners, or toolmakers). Of the cohort and PMR studies designed to assess the mortality andor morbidity of MWF-exposed workers, the Eisen et al. [1992] and Tolbert et al. [1992] studies have the most statistical power because the number of subjects with malignant neoplasms in these studiesis an order of magnitude larger than in any of the other cohort studies.The two auto parts manufacturing plants studied by Tolbert et al. were also included in the study conducted by Eisen et al. [1992](Plants I and 11). To avoid reporting the results for workers from Plants I and I1 twice, the findings from these plants are summarized from the report by Tolbert et al. [19921only, and only the findings from the third plant (Plant 111) are summarized from the report by Eisen et al. [1992]. Tolbert's findings for Plants I and I1 were used because her analyses examined the cancer risks associated with exposure to specific classes of MWF. For the purpose of tabulating the number of cohort studies reporting site-specific data for each cancer, Plants I and I1 [Tolbert et al., 19921are counted as a single study and Plant I11 is counted as a second study [Eisen et al., 19921. Tables I-V summarize the data generated to examine the association between MWF exposure and risk of cancer at specific organ sites. In an effort to keep the tables to a reasonable size, all the rate ratio estimates reported by these studies are not included. The Tolbert study provides the risk among those ever exposed to each of the specific classes of MWFs, and the remaining studies provide the risk for all workers with potential MWF exposure and, when available, the risk among those workers with the highest duration of employment.All of the rate ratios are for males, except for the studies by Acquavella et al. [1993],Wortley et al. [1992], and Schifflers et al. [19871 in which the reported rate ratios for males and females were combined. Not included in this paper are hypothesis-generating studies, which examined broad occupational categories based on census or death certificate data. The results of such studies are summarized elsewhere [NOSH, 19981. Skin and Scrotal Cancer Case reports. Since the 1940s, evidence has accumulated to support an association between skin (including scrotal) cancer and occupational exposure to MWFs. Several case reports have identified skin cancer among MWF-exposed workers [summarized in NIOSH, 1998; Kipling and Waldron, 19761. Cohort studies. A cohort study of turners employed between 1960 and 1980 at a Swedish company producing bearing rings found that the straight oil MWF-exposed turners had an increased risk of squamous cell carcinoma of the skin (obs = 5 [four scrotal, and one facial], exp = 0.3, P < 0.001) [Jarvholm et al., 19851 (Table I). Three additional scrotal cancer cases were identified among turners in the 1987 update of this cohort [Jarvholm and Lavenius, 19871. The authors suggest that use of soluble oil MWFs is not associated with scrotal cancer because no cases were observed among the grinders who primarily use soluble oil MWFs [Jarvholm and Lavenius, 19871. Furthermore, it should be noted that changes in refinery methods since the 1950s have reduced the straight oil MWF content of polycyclic aromatic hydrocarbons (PAHs), which have been suggested as the causative agent for MWF-associated skin cancer [Jarvholm and Easton, 1990;McKee et al., 19901.As would be expected because of the high survival rate for nonmelanoma skin cancer, a significantly elevated risk was not observed in the one cohort mortality study that reported skin cancer mortality (Table I). PMR studies. A significantly elevated risk was not observed in any of the three PMR studies (Table I). Population-based studies. In a population-based case-control study in Connecticut involving 45 cases of squamous cell carcinoma of the scrotum, those ever employed in an occupation with potential MWF exposure (toolmaker, setter, setup man, hardener, polisher, automatic screw operator, machinist, and machine operator) had an increased risk of this cancer (odds ratio [OR] = 10.5,95% confidence interval [95%CI] = 4.0-36.9) [Roush et al., 19821. Conclusion for Skin Cancer The large number of case reports, the cancer incidence study, and the case-control study suggest that MWF expo- 284 Calvert et al. TABLE 1. Results for Skin/Scrotal Cancer from Epidemiologic Studies of Populations Exposed to Metalworking Fluids Investigators No. with cancer Type of or no. of Rate %%CI Location study/analysis exposed cases ratio (or Pvalue) Study populatiodwncer site Cohort studies Jarvholm et al. [1985] Sweden Jarvholm and Lavenius [1987] Sweden Eisen et al. [1992] Michigan SIR SIR SMR Proportionate mortalii studies Silverstein et ai. [1988] Park et al. [1988] Vena et al. [1985] Population-based studies Rousch et al. 119821 Connecticut Connecticut NewYork PMR PMR PMR Connecticut Case/control 5 16.6 <0.001 Turners employed between 1960 and 19801 squamous cell cancer of the skin -7 a Turners only1scrotalcancer 10 0.61 0.29, 1.13 White autoworker, Plant Vskin cancer 11 1.06 0.53, 1.89 White autoworkers, Plant Il/skin cancer 7 1.27 0.51, 2.62 White autoworkers, Plant Ill/skin cancer 0.92 0.25, 2.34 1.88 0.51, 4.80 0.60 NS Ball-bearing plant workers, white1skin cancer Ball-bearingplant workers, whitdskin cancer Based on U.S. mortality, engine plant workers, white/skin cancer 26 10.5 4.0,36.9 Ever employed as toolmaker, setter, set-up man, hardener, polisher, automatic screw operator, machinist, or machine operator/ squamous cell cancer of the scrotum CI, confidenceinterval; PMR, proportionatemortality ratio; SMR, standardized mortality ratio; SIR, standardizedincidenceratio; NS = notstatistically significant. There were tw few expectedcases of scrotalcancerto make a reliableestimateof risk. sure, primarily straight oil MWF exposure, has been associated with an increased risk of skin and scrotal cancer. As a result of the changes in MWF composition and reduction of impurities over the last several decades, current straight oil MWF exposures may be associated with a substatidly reduced risk of these cancers. Laryngeal Cancer Cohort studies. confounding by sulfur, but models containing sulfur still had a significantly elevated OR in the highest straight oil M w F exposure category (OR = 1.91, 95%CI = 1.01-3.62). Although unable to adjust for smoking and alcohol, two important risk factors for laryngeal cancer [Austin and Reynolds, 19961, these investigators did not think that these risk factors confounded their results because the risk of lung cancer and cirrhosis did not increase with increasing exposure to straight oil MWFs. Only the studies of the three automobile manufacturing plants conducted by Tolbert et al. [I9921 and Eisen et al. [19921 reported site-specific data for laryngeal cancer (Table 11). Tolbert et al. [I9921 reported a statistically significant standardized mortality ratio (SMR) of 1.98 for laryngeal cancer among whites ever exposed to straight oil MWF and a nearly significant SMR of 1.41 for soluble oil MWF exposure. In a case-control analysis including all three plants, and incident as well as deceased cases, a categorized exposure analysis found an OR of 2.23 (95%CI = 1.25, 3.98) among individuals with > 0.5 mg/m3-yearsstraight oil MWF particulate exposure [Eisen et al., 19941. The authors also examined the association between laryngeal cancer and specific components or contaminants of MWFs (biocides, steel, iron, aluminum, sulfur, and chlorine). There was some evidence for PMR studies. Among the three PMR studies, the overall PMRs ranged from 1.7-1.8, and were not statistically significant [Park and Mirer, 1996; Vena et al., 1985; Mallin et al., 19861. In the study conducted by Vena et al. [1985], a significantly elevated PMR was found for workers employed less than 20 years and those who were employed after 1950 (PMR = 3.95, P < 0.05 for both subgroups). Population-based studies. Among six studies that defined occupational categories in sufficient detail to examine risk associated with exposure to MWF, and that controlled for smoking and alcohol, one Cancer R i s k s from Metalworking F l u i d Exposures 285 TABLE II. Results for Laryngeal Cancer from Epidemiologic Studies of Populations Exposed to Metalworking Fluids Investigators Location Type of study/analysis No. with cancer or no. of Rate E % C I exposed cases ratio (or Pvalue) Study population Cohort studies Tolbert et al. 119921 Michigan SMR 23 1.98 1.26,2.98 Ever straight oil MWF exposure, white 30 1.41 0.95,2.01 Ever soluble oil MWF exposure, white a 1.57 0.68,3.09 Ever synthetic MWF exposure, white 1 0.50 0.01,2.78 Ever straight oil MWF exposure, black 6 0.91 0.70,1.17 Ever soluble oil MWF exposure, black Eisen et al. [1992] Michigan SMR 2 0.77 0.09,2.79 White autoworkers, Plant 111 Eisen et al. I19941 Michigan Nested case/control 28 2.23 1.25,3.98 Autoworkers with highest exposureto straight oil MWFs Proportionate mortality studies Park and Mirer [1996] Detroit-area PMR 1 0.69 0.02,3.83 Engine plant 1, white 4 1.67 0.46,4.28 Engine plant 2,white Vena et al. (19851 New York PMR 3 1.81 NS Basedon US mortality, engine plant workers, white Mallin et al. [1986] Illinois PMR 2 1.76 NS Equipmentmanufacturingworkers, white Population-basedstudies Zagraniski et al. [1986] Connecticut Case/control Wortley et al.[1992] Washington State Case/control Zheng et al. [1992] China Caselcontrol Haguenoer et al. [1990] France Browne et al. [1988] Ahrens et al. [1991] Texas Germany Gase/control Case/control Case/control 22 2.5 1.2,5.2 Ever worked as a machinist 17 2.1 1.0,4.7 Ever worked as a metal grinder NA 1.8 0.5,6.2 Ever employed as grinding, abrading,or b a n g operator 19 1.0 0.5, 1.9 Ever employed in precision metal working 12 1.2 0.5,3.1 Usualoccupation of blacksmith, machine-tool operator, electrician, or other relatedworkers 25 0.8 0.4,1.6 Self-reported exposure to lubricantfumes 7 1.8 NS Employedin metal work or as mechanicfor at least 15 years 5 0.53 0.18,1.58 Ever machinists NA 2.2 0.9,5.3 Ever mineraloil exposure Abbreviations: (see Table I.) found a significant risk of laryngeal cancer. Zagraniski et al. [19861found an elevated laryngeal cancer risk among those with "ever employment as a machinist" (SMR = 2.5, 95%CI: 1.2-5.2) and among those with "ever employment as a metal grinder" (SMR = 2.1, 95%CI = 1.04.7). Another case-control study of 100 laryngeal cancer cases and 100 controls found a nonsignificantly elevated risk among those who self-reported ever having mineral oil exposure (OR = 2.2, 95%CI = 0.9-5.3) [Ahrens et al., 19911. Conclusion for Laryngeal Cancer Several studies suggest that MWF exposure may be associated with laryngeal cancer. In particular, the studies by Eisen et al. [1992, 19941 and Tolbert et al. [1992] suggest that laryngeal cancer is associated with exposure to straight oil MWFs. Rectal Cancer Cohort studies. Only the studies conducted by Tolbert et al. [1992], Eisen et al. [1992], and DecouflC [1978] reported site-specific results for rectal cancer (Table 111). Tolbert et al. [1992] reported a significant association between straight oil MWF exposure and rectal cancer among white, but not black, workers. Poisson regression analyses revealed a trend of increasing rectal cancer risk in relation to years of exposure to straight oil MWFs ( P < 0.0001). The rate ratio for the most highly exposed group was 3.2 (95%CI = 1.6-6.2). 286 Calvert e t al. TABLE 111. Results for Rectal Cancer from Epidemiologic Studies of Populations Exposed to MetalworkingFluids (MWF) Investigators Location Type of study/analysis No. with cancer or no. of exposed cases Rate ratio 95%CI (or Pvalue) Study population Cohort studies Tolbert et al. [1992] Michigan SMR Eisen et al. [1992] Oecoufle [1978] Proportionate mortality studies Silversteinet al. [1988] Park et al. [1988] Vena et ai. [I9851 Michigan SMR Michigan SMR Connecticut PMR Connecticut PMR New York PMR Mallin et al. [1986] Illinois Population-basedstudies Gerhardsson de Verdier et al. [1992] Sweden Siemiatyckiet al. [1987] Montreal PMR Casekontrol Case/control 37 1.47 1.04,2.03 Ever straight oil MWF exposure, white 51 1.09 0.81,1.43 Ever soluble oil MWF exposure, white 9 0.92 0.42,1.74 Ever synthetic MWF exposure, white 1 0.45 0.01,2.53 Ever straight oil MWF exposure, black I) J 0.68 0.14,1.99 Eve: soluble of MWF exposure, black 7 1.70 0.68,3.50 White autoworkers, Plant 111 8 1.25 NS Metal machining plantworkers, white 4 1.29 NS 5+ years of heavy MWF mist exposure, white 14 1.36 0.81,2.29 Ball-bearingplant workers, white 11 3.07 1.54,5.50 Ball-bearingplant workers, white 4 1.38 NS Based on US mortality, engine plant workers, white 4 2.76 P < 0.05 Employed in engine plant >20 years, white 2 0.80 NS Equipment manufacturingworkers, white 25 2.1 1.1,4.02 Ever exposed to cutting oils 13 0.7 9O%CI; 0.4-1.O Ever exposed to cutting oils Abbreviations:(see Table I.) PMR studies. Among the four PMR studies reporting data for rectal cancer, the investigation led by Park et al. [1988] found a significant excess in a cohort with potential exposure to straight and soluble oil MWFs (PMR = 3.07, 95%CI = 1.54-5.50). Park et al. [1988] did not report the risk of each specifictype of MWF exposure. Another PMR study found a significantexcess in a subgroup employed in an engine plant for >20 years (PMR = 2.76, P < 0.05) (although three types of MWF [straight, soluble, and synthetic MWF] were used at this plant, their temporal use was not known to the study investigators) [Vena et al., 19851. Silverstein et al. [ 19881found a slight excess of rectal cancer mortality which was not statistically significant; the risk for various processes was not reported. Mallin et al. [I9861 found a risk of less than 1;however this finding is based on only two rectal cancer deaths. Population-based studies. A population-based case-control study of incident cases of rectal cancer in Sweden found that male workers ever exposed to cutting fluids had an elevated risk of rectal cancer (OR = 2.1, 95%CI = I.l4.0)[Gerhardsson de Verdier et al.. 19921. However, several of the cases had other occupa- tional exposures associated with an increased risk of rectal cancer including exposure to asbestos, soot, and coallcokel wood combustion gases. In an analysis adjusting for these other exposures, the risk of rectal cancer among cutting oil-exposed workers was lower (OR = 1.4, 95%CI = 0.6-3.5). In another population-basedcase-control study that examined the association between several cancer sites and occupational exposure to several petroleum-derived liquids, ever having cutting oil exposure was found not to be associated with an increased risk of rectal cancer (OR = 0.7, 90%CI = 0.4-1.0) [Siemiatyckiet al., 19871. Conclusions for Rectal Cancer Several studies suggest that MWF exposure is associated with rectal cancer. In particular, the findings from the study with the most statistical power suggest that straight oil MWF exposure may be associated with an increased risk of rectal cancer [Tolbert et al., 19921. Pancreatic Cancer Cohort studies. Among five cohort studies reporting site-specific data for pancreatic cancer, two found significant excess mortality Cancer Risks from Metalworking Fluid Exposures 287 TABLE IV. Resultsfor PancreaticCancer from Epidemiologic Studies of PopulationsExposedto MetalworkingFluids (MWF) Investigators Location Type of study/analysis No. with cancer or no. of Rate 9556Cl exposed cases ratio (or Pvalue) Study population Cohort studies Tolbert et al. [1992] Michigan SMR 34 0.80 0.55, 1.11 Ever straight oil MWF exposure, white 61 0.77 0.59, 1.OO Ever soluble oil MWF exposure, white 19 1.03 0.62, 1.61 Ever synthetic MWF exposure, white 8 1.40 0.60,2.77 Ever straight oil MWF exposure, black 19 1.62 0.98, 2.54 Ever soluble oil MWF exposure, black Eisen et al. [1992] Michigan SMR 8 0.87 0.37, 1.71 White autoworkers,Plant 111 Bardin et al. [1991] Michigan Nested case/control 9 3.0 1.2, 7.5 Highest exposure to synthetic MWF Rotini et al. [1993] Ohio SMR 8 0.91 0.39, 1.79 Engine plant, white 7 3.03 1.21,6.24 Engine plant, black Acquavella et al. [1993] Iowa SMR 11 2.0 0.9, 3.8 Metalworkingfactory workers 5 3.6 1.2. 8.3 Factory workers employed>lO years, hired between 1950-1959 Decoufle [1978] Michigan SMR 8 1.05 NS Metal machiningplant workers, white 1 0.27 NS 5+ years of heavy MWF mist exposure, white Proportionate mortality studies Park and Mirer [1996] Detroit-area PMR 10 1.82 0.87, 3.34 Engine plant 1, white 11 1.23 0.62,2.21 Engine plant 2, white MOR 4 3.61 1.04, 12.6 Machiningwith straight oil MWF Silversteinet al. [1988] Connecticut PMR 24 1.43 0.96, 2.12 Ball-bearing plant workers, white MOR 9 3.10 P = 0.05 Employed in grinding 1O+ years MOR 5 3.71 P = 0.05 Employedin machinery 1O+ years Park et al. [1988] Connecticut PMR 8 1.09 0.55,2.18 Ball-bearingplant workers, white VeFa et al. [1985] NewYork PMR 11 1.89 <0.05 Based on US. mortality, white 7 2.32 <0.05 Employed in engine plant >20 years Mallin et al. [1986] Illinois PMR 5 1.19 NS Equipmentmanufacturingworkers, white 5 3.57 C0.05 Equipment manufacturingworkers, black Population-basedstudies Mack and Paganini-Hill (19811 Los Angeles Incidence 21 1.30 NA Machinists, white Inales Abbreviations: MOR = mortalityodds ratio (also see Table II) from this cause of death [Rotimi et al.. 1993;Acquavella et al., 19931 Table IV). Rotimi et al. [1993] found that black men employed at two Ohio engine manufacturing plants had an excess pancreatic cancer mortality (SMR = 3.03, 95%CI = 1.21-6.24, based on seven deaths). However, it should be noted that the investigators found no consistent pattern with respect to time since hire or duration of employment, and no pancreatic cancer excess was observed in white workers. Acquavella et al. [1993] reported that factory workers employed at an Iowa metalworkingfacility, all of whom were white, had an increased risk of pancreatic cancer mortality (SMR = 2.0, 95%CI = 0.9-3.8). The risk appeared to be greatest among factory workers employed 10 or more years who were hired between 1950 and 1959 (SMR = 3.6, 95%CI = 1.2-8.3); however, the authors did not report if workers in a specific occupational group were responsible for this elevation. Among those in the overall cohort, assembly workers, who Acquavella et al. state are unlikely to have MWF exposure, were the occupational group with the highest risk (SMR = 3.0, 95%CI = 1.0-7.5). In contrast, in those departments identified by Acquavella et al. as having potential MWF exposure, there were 2 observed pancreatic cancer deaths while 3.3 deaths were expected. Tolbert et al. [1992] found excess pancreatic cancer mortality among black workers exposed to soluble oil MWFs at Plants I and I1 in Michigan (SMR = 1.62, 95%CI = 0.98-2.54). In a Poisson regression analysis that controlled for race, age, and gender, an increased risk of pancreatic cancer mortality was observed in those workers with the highest exposures to synthetic MWFs (rate ratio = 2.04, 95%CI: 0.88-4.72) [Tolbert et al., 19921. In a case-control analysis that included Plants I, 11, and In, 288 Calvert et al. a categorized exposure analysis found an OR of 2.23 (95%CE = 1.25-3.98) among individuals with >1.4 mg/m3-years grinding with synthetic MWF [Bardin et al., 19971. However, neither synthetic MWF, nor any other measured exposure, was found to explain the previously documented excess pancreatic cancer risk among black workers [Bardin et al., 19971. Although unable to adjust for smoking, an important risk factor for pancreatic cancer [Silvexman et al., 19941, the authors did not think that this risk factor confounded their results because the risk of lung cancer did not increase with increasing exposure to synthetic MWFs. One other cohort study found a nonsignificant elevation in pancreatic cancer mortality; however, this study had limited statistical power [DecouflC, 19781. elevation in pancreatic cancer mortality; however, an MOR analysis was not reported [Park, 19881. Conclusion for Pancreatic Cancer Several studies have found significantly increased risks of pancreatic cancer among MWF-exposed workers. The evidence is strongest for grinding with synthetic MWF [Bardin et al., 19971, and for machining with straight oil MWFs [Silverstein et al., 1988; Park and Mirer, 19961. Although a number of the studies did not have internally consistent findings (Le., excesses in black but not white workers, lack of association with duration of MWF exposure), the number of studies with statistically significant findings suggeststhat exposure to some MWFs may increase the risk of pancreatic cancer. PMR studies. Bladder Cancer Among five studies reporting site-specific data for pancreatic cancer, four found significantly elevated PMRs [Vena et al., 1985; Mallin et al., 1986; Silverstein et al., 1988; Park and Mirer, 19961. White men employed at an engine plant for at least 10 years had an excess of pancreatic cancer mortality (PMR = 1.89,P < 0.05),which was higher for those employed > 20 years (PMR = 2.32, P < 0.05) [Vena et al., 19851. Use of county referent rates resulted in higher PMRs (employed > 10 years, PMR = 2.41, P < 0.05; employed 20 years, PMR = 2.97, P < 0.05). Mallin et al. [1986] found a significant excess of pancreatic cancer among black (PMR = 3.57, P < 0.05),but not white (1.19, not significant),men employed in the manufacture of diesel engines and construction equipment. The PMR for pancreatic cancer was highest among black men who died after 20 years of service (PMR = 4.79, P < 0.01). Another PMR study found an elevation in pancreatic cancer mortality among whites at a ball-bearingmanufacturing plant (PMR = 1.43, 95%CI = 0.96-2.12) [Silverstein et al., 19881. Casecontrol analyses revealed substantiallyelevated risks associated with 10 or more years employment in grinding with various MWFs (OR = 3.10, P = 0.05) and machining with straight oil MWF (OR = 5.31, P = 0.05). The risk associated with grinding was present only for those with early hire dates (during the early 1930s and before) at which time straight oil MWFs were "almost exclusively used in grinding." There were too few deaths among nonwhite men for analysis. A PMR study of workers at two engine plants did not observe significant excesses for pancreatic cancer [Park and Mirer, 19961. However, a mortality odds ratio (MOR) analysis of these workers found an increased risk among those ever employed in machining with straight oil MWFs (OR = 3.61, 95%CI = 1.04-12.6, based on three cases), but no trend was observed with increasing cumulative exposure. One other PMR study found a nonsignificant Cohort studies. Only two cohort studies reported site-specific data for bladder cancer and neither found a significant excess [DecouflC, 1978; Jarvholm and Lavenius, 19871 (Table V). However, both of these studies are limited by relatively small sample sizes. PMR studies. Of the six PMR studies that reported site-specific data for bladder cancer, only one reported a significant excess, which was among white workers employed in an engine plant (PMR = 2.28, P < 0.05) [Vena et al., 19851. The risk was greatest among those first employed during or before 1950 (PMR = 3.37, P < 0.05). A study of bearing plant workers found a nonsignificantly elevated risk of bladder cancer [Silverstein, 19881. A study of workers at two Detroit-area engine manufacturing plants also found nonsignificantly elevated PMRs for bladder cancer [Park and Mirer, 19961. However, an MOR analysis of these workers found a significant association between risk of bladder cancer and cumulative exposures to grinding with `straight oil MWF (MOR for the mean cumulative exposure of exposed cases = 2.99 [95% CI = 1.15-7.771,based on seven deaths), and employment in the machining or heat treat area (MOR for the mean cumulative exposure of exposed cases = 2.86 [95% CI = 1.14-7.181, based on four deaths). Two studies did not find an increased risk of bladder cancer [Park et al., 1988; Mallin et al., 19861; however, these studies are limited by small sample size. Population-based studies. Several case-control studies have also examined the risk of bladder cancer among those whose occupations may Cancer Risks from Metalworking Fluid Exposures 289 TABLE V. Results for Bladder and Lower Urinary Tract Cancer from Epidemiologic Studies of Populations Exposed to Metalworking Fluids Investigators Location Type of study/analysis No. with cancer or no. of exposed cases Rate ratio %%I (or Pvalue) Study population/cancer site Cohort studies Decoufle [I9781 Michigan SMR Jarvholm and Luvenius [1987] Sweden SIR Proportionatemortality studies Park and Mirer [1996] Detroit-area PMR Silverstein et al. [1988] Park et al. [1988] Vena et ai. [1985] Connecticut Connecticut NewYork MOR MOR PMR PMR PMR Mallin et al. [1986] Population-based studies Silvermanet al. [1989a] Illinois us PMR Case/control Siemiatycki et al. [1987] Claude et al. [1988] Montreal Germany Casekontrol Case/control Gonzalez et al. [1989] Spain Case/control Steenland et al. [1987] Ohio Case/control Vineis and Magnani [1985] Italy SchWflers et al. [l987] Belgium Caselcontrol Case/control Howe et al. [1980] Silvermanet al. [1983] Silverman et ai. [1989b] Canada Detroit us. Case/control Case/control Case/control Abbreviations: (see Table I.) 6 1.2 NS Metal machining plant workers, white/bladder 8, lower urinary tract 2 0.8 NS 5+ years heavy MWF mist exposure, white/bladder and lower urinary tract 7 1.04 0.4,2.2 Grinders and tumerdbladder 6 2.16 0.79,4.70 Engine plant 1, white/bladder 5 1.13 0.37,2.64 Engine plant 2,whitelbladder 7 2.99 1.15,7.77 Grinding with straight oil MWFlbladder 4 2.86 1.14,7.18 Machining or heat treat employmenffbladder 14 1.26 0.75,2.13 Ball-bearing plant workers, white/bladder 1 0.24 0.01,1.31 Ball-bearing plant workers, whitelbladder 7 2.28 <0.05 Engine plant workers, whitelbladder 4 2.76 NS Employed in engine plant >20 yeardbladder 2 0.78 NS Equipment manufacturing workers, white/bladder 102 1.3 1.O,1.7 Ever machinist6 months or more/bladder 51 1.4 0.9,2.1 Ever drill press operator 2 6 monthdbladder 47 1.2 9OYoCI: 1 .o-1.6 Ever exposed to cutting oildbladder 18 2.25 1 .O,5.6 Ever turner/bladder and lower urinary tract 43 0.84 0.54,1.3 Ever metal worker/bladder and lower urinarytract 31 0.77 0.5,1.1 Ever toolmaker 2 6 monthdbladder NA 1.86 1.2,2.8 Ever machinery adjuster, assembler, or mechanic 2 6 monthdbladder 11 2.00 NS Ever grinding machine operatorlbladder and lower urinary tract 45 0.69 c0.05 Ever machinisthladder and lower urinary tract 16 1.5 0.7,3.3 Ever employed in machine tools 2 6 monthdbladder 34 2.45 1.28,4.69 All metal workerdbladder 8 2.57 0.92,7.16 Tumedbladder NA 2.7 1.1,7.7 Ever metal machiniwbladder 137 1.1 0.8,1.5 All metal machiniWbladder and lower urinary tract 32 1.5 0.9,2.7 Tool and die workerslbladder & lower urinary tract 26 1.1 0.6,1.9 Metal machineryworkerlbladder involve MWF exposure. Only those studies that controlled for smoking, a known risk factor for bladder cancer [Matanoski and Elliott, 19811, are included in this review. In a large population-based case-control study from the United States, Silverman et al. [1989a] found an elevated risk of bladder cancer among white men ever employed as machinists (OR = 1.3, 95%CI = 1.0-1.7) or drill press operators (OR = 1.4, 95%CI = 0.9-2.1). Furthermore, among drill press operators, the risk increased with increasing duration of employment ( P for trend = 0.008). Among drill press operators who worked 5 or more years, the elevated risk was present in both those who began work before 1950 (OR = 1.7; confidence limits not provided) and those who began work in 1950 or later (OR = 2.9; confidence limits not provided). However, among those who were employed as drill press operators for less than 5 years, the risk of bladder cancer was increased only among those who began employment in 1950 or later (OR = 2.8; confidence limits not 290 Calvert et al. provided). The same study [Silverman et al., 1989bl examined occupational risk factors for bladder cancer among nonwhite men, and reported an RR of 1.1(95%CI = 0.6-1.9) for the summary category "Metal machinery worker," with an identical summary category" Metal machinery worker" for whites of RR of 1.1 (95%CI = 1.0-1.3)[Silverman et al., 1989al. The risks for the subcategories "Machinists" and "Drill Press Operators" was not broken out for nonwhites, nor was duration of employment as a metal machinery worker examined. In a population-based case-control study from Canada, Howe et al. (1980) found that those ever employed as metal machinists had an increased risk of bladder cancer (OR = 2.7, 95%CI = 1.1-7.7). A hospitalbased case-control study conducted in Germany found that individuals ever employed as turners had an increased risk of bladder cancer (OR = 2.25, 95%CI = 1.0-5.6), and the risk was consistently elevated with increasing duration of employment ( P for trend = 0.08) [Claude et al., 19881.This same study also found no increased risk of bladder cancer for the broad category of metal workers (OR = 0.84, 95% CI = 0.54-1.3). A population-basedcase-controlstudy from Belgium found that metal workers had a significantly increased risk of bladder cancer (RR = 2.45, 95%CI = 1.28-4.69). Subgroup analyses of these workers found the highest risk among turners (RR = 2.57, 95%CI = 0.927.16) [Schifflers et al., 19871. Another hospital-based casecontrol study from Italy found that those employed for six months or more in the machine-tool industry had an increased risk of bladder cancer (OR = 1.5, 95%CI = 0.7-3.3)[Vineis and Magnani, 19851. Within the machinetool trade, the risk of bladder cancer was elevated among turners, especially among turners hired before 1940 and employed more than 10 years (RR = 3.1, 95%CI = 0.9-10.5). A population-based case-control study conducted in Hamilton County, Ohio, found that grinding machine operators had an increased risk of bladder cancer (OR = 2.00; not significant), whereas machinists were found to have a significantly decreased risk (OR = 0.69, P < 0.05) [Steenland et al., 19871.Another case-control study did not observe an increased risk of bladder cancer among toolmakers (RR = 0.77, 95%CI = 0.5-1.1) but did observe an increased risk among machinery adjusters, assemblers, and mechanics (RR = 1.86,95%CI = 1.2-2.8) [Gonzalezet al., 19891. Conclusions for Bladder Cancer The association between bladder cancer and MWF exposure is well supported by one large and well-designed case-control study [Silverman et al., 1989a,b], as well as several other studies conducted in different geographic locations, all of which controlled for smoking. Although none of the cohort studies found a significantly increased risk of bladder cancer, it has been observed that mortality studies may not be suitable for detecting elevated risks for cancers with high survival rates [Schulte et al., 1985; Steenland et al., 19881. Route of Exposure Although the route of exposure to MWFs is generally through dermal contact or through inhalation, the large size of many airborne MWF droplets can lead to gastrointestinal exposure. A significant proportion of airborne MWF particles is in the nonrespirable (extrathoracic) range (Le., particles with a mass mean diameter >9.8 pm). Eisen et al. [I9941 report that in their sttidy approximately 20-35% of the total particulate was in the extrathoracic range. Large particles generally result in gastrointestinal exposures since they are filtered out in the nasopharyngealregion and do not reach the airways. In addition, some of the small particles in the thoracic size fraction are captured by the mucocilliary escalator. The mucocilliary escalator transports the particles to the pharynx, where they are swallowed, thereby permitting gastrointestinal exposure. DISCUSSION AND CONCLUSIONS We believe there is substantial evidence for increased risk of cancer at several sites (larynx, rectum, pancreas, skin, scrotum, and bladder) associated with at least some MWFs used prior to the mid-1970s. The inconsistencies between studies with respect to the organ sites that were affected, and the variation in the strength of association between the surrogates of exposure and specific sites are, most likely, related to the diverse nature of MWF mixtures studied, the absence of detailed exposure information, and the limitations of the epidemiologictools with which MWF exposures have been studied. As described earlier, there are four major classes of MWF. The types and amounts of chemical constituents can vary across these different classes of MWF. Furthermore, within each class of MWFs there are many different formulations, which vary in composition, and may contain many different additives and impurities. SomeMWF`constituents are considered carcinogenic in animals, including N-nitrosamines [IARC, 19781, and some PAHs [IARC, 19831. Efforts to reduce these potential carcinogenic expo- sures have been ongoing. Removal of PAHs from MWFs began in the 1950s and U.S. Environmental Protection Agency (EPA) regulations during the 1980s were directed at reducing nitrosamine exposures [Code of Federal Regulations, 19841. Because different epidemiologic study populations may have been exposed to different classes and formulations of cutting fluids, some lack of consistency in site-specific results between studies should be expected when evaluating the carcinogenicity of these substances. Similarly,when comparing studies with limited information Cancer Risks from Metalworking Fluid Exposures 291 on the intensity of exposure, we would expect variation in the strength of association between exposure and the risk of cancer. In our opinion, the consistency among the studies is sufficient to support our conclusions. Given the small number of epidemiologic studies that have adequate exposure characterization, the specific MWF constituent(s) or contaminant(s) responsible for the various site-specific cancer risks remains to be determined. The study with the most statistical power and detailed exposure information [Tolbert et al., 19921 suggests that specific classes of MWFs are associated with cancer at certain sites. However, within these MWF classes, the specific formulations responsible for the elevated cancer risks remain to be identified. Within the Tolbert et al. study, straight oil MWF exposure was modestly associated with an increased risk of laryngeal cancer and rectal cancer, and there was limited evidence that synthetic MWF exposure was associated with an increased risk of pancreatic cancer. Subsequent casecontrol studies based on the original cohort have confirmed the laryngeal cancer-straight oil MWF association [Eisen et al., 19941,and the pancreatic cancer-syntheticMWF association [Bardin et al., 19971.The Tolbert et al. study found less evidence that soluble oil MWF exposure is associated with cancer at any specific site. We believe it is premature to conclude that all members of the soluble class of MWFs were free from any past carcinogenic risks. This is because soluble oil MWFs contain many of the ingiedients found in straight oil MWFs, but in different concentrations. Second, many of the epidemiologic studies with positive findings involved exposures to more than one class of MWF. It is unlikely that non-MWF exposures are responsible for the cancer findings described in this paper. Smoking and alcohol are associated with some of these cancers. However, the case-control studies controIled for these exposures when appropriate or determined that these exposures were unlikely confounders. Although information on these lifestyle factors are not often collected in occupational cohort mortality or PMR studies, it has been demonstrated that smoking is unlikely to account for relative risks >1.3 for smoking related diseases [Siemiatycki et al., 19881.As for non-MWF occupational exposures, it is unlikely that these exposures explain the majority of findings, as the common exposure across all of the studies was MWF. Although some nonMWF exposures may have interacted synergistically with MWF exposure to produce some of the observed risks, the existence or extent of such synergism remains to be determined. The studies that provide the bulk of the evidence suggesting an association between MWF exposure and cancer involved workers employed as early as the 1930sand as late as the mid-1980s. Because there is a latency period of 10-20 years, on average, between initial exposure to a carcinogen and the initial appearance of a solid organ cancer caused by that carcinogen, the excess cancer mortality observed in these cohort studies most likely reflects the cancer risk associated with exposure conditions in the mid- 1970sand earlier.Over the last several decades, substantial changes have been made in the metalworking industry, including changes in MWF composition, reduction of impurities, and reduction of exposure concentrations. These changes have likely reduced the cancer risks. However, since the epidemiologic studies are unable to identify the MWF composition and impurities associated with the cancer risks observed in earlier cohorts, there are insufficientdata to conclude that these changes will have eliminated all carcinogenic risks. The risk of cancer from MWF exposures in the mid-1970's and later remains to be determined because a definitive study has not yet been conducted of workers entering MWF-exposed jobs during this time. The substantial evidence that some MWFs used prior to the mid-1970s are associated with cancer at several sites, and the possible potential for current MWF to pose a similar carcinogenic hazard, support the lowered recommended exposure limit (REL) of 0.5 mg/m3total particulate mass. This new REL is based primarily on evidence for an association between nonmalignant respiratory effects and MWF exposure [NIOSH, 19981. REFERENCES Acquavella J, Leet T, Johnson G (1993): Occupational experience and mortality among a cohort of metal components manufacturing workers. Epidemiology 4:428-434. Ahrens W, Jockel K, Patzak W, Elsner G (1991): Alcohol, smoking, and occupational factors in cancer of the larynx: A case-control study. Am J Ind Med 20:477493. Austin DF, Reynolds P [1982]: Laryngeal cancer. In Schottenfeld D, Fraumeni JF (eds): "Cancer Epidemiology and Prevention." 2nd ed. New York Oxford University Press, pp 619-636. Bardin JA, Eisen EA, Tolbert PE, Hallock MF, Hammond SK, Woskie SR, Smith TJ, Monson RR (1997): Mortality studies of machining fluid exposure in the automobile industry. V A case-control study of pancreatic cancer. Am J Ind Med 32:240-247. Brown LM, Mason TJ, Pickle LW, Stewart PA, Buffler PA, Burau K,Ziegler RG, Fraumeni JF (1988):Occupational risk factors for laryngeal cancer on the Texas Gulf Coast. Cancer Res 48:1960-1964. Claude JC, Frentzel-Beyme RR, Kunze E (1988): Occupation and risk of cancer of the lower urinary tract among men. A case-control study. Int J Cancer 41:371-379. Code of Federal Regulations 40 (1984): Environmental Protection Agency, Prohibition of nitrates in metalworking fluids. Part 747 (49 Fed Reg 2762). Washington, DC: Government Printing Office. DecouflC P (1978): Further analysis of cancer mortality patterns among workers exposed to cutting oil mists. JNCI 61:1025-1030. Eisen EA, Tolbert PE, Hallock MF, Monson RR, Smith TJ, Woskie SR (1994): Mortality studies of machining fluid exposure in the automobile industry 111: A case-control study of larynx cancer. Am J Ind Med 26:185-202. Eisen EA, Tolbert PE, Monson RR, Smith TJ (1992): Mortality studies of machining fluid exposure in the automobile industry I: A standardized mortality ratio analysis. Am J Ind Med 22309-824. 292 Calvert et al. Gerhardsson de Verdier M, Plato N, Steineck G, Peters JM (1992): Schifflers E, Jamart J, Renard V (1987): Tobacco and occupation as risk Occupationalexposures and cancer of the colon and rectum. Am J Ind Med, factors in bladder cancer: A case-control study in southern Belgium. Int J 22291-303. Cancer 39287-292. Gonzalez CA, Lopez-Abente G, Errezola M, Escolar A, Riboli E, Izarzugaza I, Nebot M (1989): Occupation and bladder cancer in Spain: a multi-centre case-control study. Int J Epidemiol 18:569-577. Haguenoer JM, Cordier S, Morel C, Lefebvre JL, Hemon D (1990): Occupational risk factors for upper respiratory tract and upper digestive tract cancers. Br J Ind Med 47:380-383. Howe GR, Burch JD, Miller AB, Cook GM, Esteve J, Morrison B, Gordon P, Chambers LW, Fodor G, Winsor GM (1980): Tobacco use, occupation, coffee, various nutrients, and bladder cancer. JNCI 64:701-713. IARC (1978): N-nitrosodiethanolamine. In "IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Some N-Nitroso Compounds." Vol. 17. Lyon, France: International Agency for Research on Cancer, pp 77-82. IARC (1983): "IARCMonographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, PolynuclearAromatic Compounds." Part 1: "Chemical, Environmental, and Experimental Data." Vol. 32. Lyon, France: InternationalAgency for Research on Cancer. Jarvholm B, Easton D (1990): Models for skin tumour risks in workers exposed to mineral oils. Br J Cancer 62:1039-1041. Jarvholm B, Lavenius B (1987): Mortality and cancer morbidity in workers exposed to cutting fluids. Arch Environ Health 42:361-366. Jarvholm B, Fast K, Lavenius B, Tomsic P (1985): Exposure to cutting oils and its relation to skin tumors and premalignant skin lesions on the hands and forearms. Scand J Work Environ Health 11:365-369. Kipling MD, Waldron HA (1976): Polycyclic aromatic hydrocarbons in mineral oil, tar, and pitch, excluding petroleumpitch. Prev Med 5262-278. Mack TM, Paganini-Hill A (1981):Epidemiology of pancreas cancer in Los Angeles. Cancer 47:1474-1483. Mallin K, Berkeley L, Young Q (1986):A proportionalmortality ratio study of workers in a construction equipment and diesel engine manufacturing plant. Am J Ind Med 10127-141. Matanoski GM, Elliott EA (1981): Bladder cancer epidemiology. Epidemiol Rev 50:588-600. McKee RH, Scala RA, Chauzy C (1990): An evaluation of the epidermal carcinogenic potential of cutting fluids. J Appl Toxic01 10251-256. Schulte PA, Ringen K, Hemstreet GP, Altekruse EB, Gullen WH, Patton MG, Allsbrook WC, Crosby JH, West SS, Witherington R, Koss L, Bales CE, Tillet S, Rooks SCF, Stem F,StringerW, Schmidt VA, Brubaker MM (1985): Risk assessment of a cohort exposed to aromatic amines. J Occup Med27:115-121. Siemiatycki J. Dewar R, Nadon L, Gerin M, Richardson L, Wacholder S (1987): Associations between several sites of cancer and twelve petroleumderived liquids. Scand J Work Environ Health 13:493-504. SiemiatyckiJ, Wacholder S, Dewar R, Cardis E, Greenwood C, Richardson L (1988): Degree of confoundingbias related to smoking, ethnic group, and socioeconomic status in estimates of the associations between occupation and cancer. J Occup Med 30:617-625. Silverman DT, Hoover RN, Albert S, Graff KM (1983): Occupation and cancer of the lower urinary tract in Detroit. JNCI 70:237-245. SilvermanDT, Levin LI. Hoover RN, Hartge P (1989a): Occupationalrisks of bladder cancer in the United States. I. White men. JNCI 81:1472-1480. Silverman DT, Levin LI, Hoover RN (1989b): Occupational risks of bladder cancer in the United States. 11. Nonwhite men. JNCI 81:1480-1483. Silverman DT, Dunn JA, Hoover RN, Schiffman M, Lillemoe KD, Schoenberg JB, Brown LM, Greenberg RS, Hayes RB, Swanson GM, Wacholder S, SchwartzAG, Liff JM, Pottern LM (1994):Cigarettesmoking and pancreas cancer: A case-control study based on direct interview. JNCI 1994;86:1510-1516, Silverstein M, Park R, Marmor M, Maizlish N,Mirer F (1988): Mortality among bearing plantworkers exposed to metalworking fluids and abrasives. occup Med 30:706-714, Steenland K, Burnett C, Osorio AM (1987):A case-controlstudy of bladder cancer using city directories as a source of occupational data. Am J Epidemiol 126247-257. Steenland K, Schnorr T, Beaumont J, Halperin W, Bloom T (1988): Incidence of laryngeal cancer and exposure to acid mists. Br J Ind Med 1988;45:766-776. Tolbert PE, Eisen EA, Pothier LJ, Monson RR, Hallock MF, Smith TJ (1992): Mortality studies of machining-fluid exposure in the automobile industry. 11. Risks associated with specific fluid types. Scand J Work Environ Health 18:351-360. NIOSH (1998): "Criteria for a Recommended Standard Occupational Exposure to Metalworking Fluids." Cincinnati, OH: CDCINIOSH, Pub1 NO. 98-102. Park RM, Wegman DH, Silverstein MA, Maizlish NA, Mirer FE (1988): Causes of death among workers in a bearing manufacturing plant. Am J Ind Med 13569-580. Park RM, Mirer FE (1996): A survey of mortality at two automotiveengine manufacturingplants. Am J Ind Med 306644573, Vena JE, Sultz HA, Fiedler RC. Barnes RE (1985): Mortality of workers in an automobile engine and parts manufacturing complex. Br J Ind Med 4285-93. Vineis P, Magnani C (1985): Occupation and bladder cancer,in males: a case-control study. Int J Cancer 35:599-606. Wortley P, Vaughan TL, Davis S, Morgan MS. Thomas DB (1992): A case-control study of occupationalrisk factors for laryngeal cancer. Br J Ind Med 49937-844. Rotimi C, Austin H, Delzell E, Day C, Macaluso M, Honda Y (1993): Retrospective follow-up study of foundry and engine plant workers. Am J Ind Med 24485-498. Roush GC, Kelly J, Meigs JW. Flannery JT (1982): Scrotal carcinoma in Connecticut methlworkers. Am J Epidemiol I 16:76-85. Zagraniski RT, Kelsey JL, Walter SD (1986): Occupational risk factors for laryngeal carcinoma: Connecticut. 1975-1980. Am J Epidemiol 124:67-76. Zheng W, Blot WJ, Shu X. Ciao Y, Ji B. Ziegler RG, Fraumeni JF (1992): Diet and other risk factors for laryngeal cancer in Shanghai. China. Am J Epidemiol 136:178-191.