Document rxkaOnqK8adV6YJB1ap4yqVdv

AMERICAN JOURNAL OF INDUSTRIAL MEDICINE 38:295299 (2000) Lead and Cancer in Humans: Where Are We Now? Kyle Steenland, PhD1 and Paolo Boffetta, MD2 Background Lead is only weakly mutagenic, but in vitro it inhibits DNA repair and acts synergistically with other mutagens. Lead acetate administered orally, cutaneously, or intraperitoneally causes kidney cancer, brain cancer (gliomas), and lung cancer in rodents, and acts synergistically with other carcinogens. Most cytogenetic studies of exposed workers have shown increases in chromosome aberrations or sister chromatid exchange, including some studies with positive-exposure response trends. There are eight studies of cancer mortality or incidence among highly exposed workers; most are cohort studies of lead smelter or battery workers exposed decades ago. Methods We reviewed the epidemologic studies with regard to cancer. Results These studies provide some evidence of increased risk of lung cancer (RR 1.30, 1.151.46, 675 observed deaths) and stomach cancer (combined RR 1.34, 1.141.57, 181 observed). However, the lung cancer ndings are not consistent across studies, and confounding by arsenic may affect the study with the highest lung cancer RR. Exclusion of that study yields a combined lung cancer RR of 1.14 (1.041.73). There is little evidence of increased risk of kidney cancer (combined RR 1.01, 0.721.42, 40 observed) or brain cancer (combined RR=1.06, 0.811.40, 69 observed). However, two studies show a two-fold increase in kidney cancer, and one study shows a signicant excess of gliomas. IARC classied lead as a ``possible human carcinogen'' based on sufcient animal data and insufcient human data in 1987. Six of the eight studies cited above have been published since 1987. Conclusion Overall, there is only weak evidence associating lead with cancer; the most likely candidates are lung cancer, stomach cancer, and gliomas. Am. J. Ind. Med. 38:295299, 2000. Published 2000 Wiley-Liss, Inc.y KEY WORDS: lead; cancer; epidemiology INTRODUCTION In 1987 IARC judged the evidence for lead and inorganic lead compounds to be sufcient in animals and inadequate in humans, leading to an overall evaluation of Group 2B, possibly carcinogenic in humans. Since that time, a number of new studies or updates of older studies 1National Institute for Occupational Safety and Health, Cincinnati, Ohio 2International Agency for Research on Cancer, Lyon, France *Correspondence to:Kyle Steenland, National Institute for Occupational Safety and Health, 4676 Columbia Parkway, Cincinnati, Ohio 45226. E-mail: kns1@cdc.gov Accepted 27 April 2000 have been published. We here summarize existing evidence regarding the possible carcinogenicity of lead. ANIMAL STUDIES Lead is not genotoxic in vitro, but increases the mutagenicity of other mutagens, possibly acting via inhibition of DNA repair [Hartwig, 1994]. Early animal studies indicated that lead does cause cancer in animal studies [IARC, 1980], principally kidney cancer in two species by different routes of administration. In addition, rats developed brain cancer (gliomas) in one study, and mice developed lung adenomas in another. Lead acetate and lead oxide acted synergistically to increase kidney and lung tumors, respectively, after administration of known carcinogens (nitrosamines, Published 2000 Wiley-Liss, Inc. y This article is a US Government work and, as such, is in the public domain in the United States of America. 296 Steenland and Boffetta benzo[a]pyrene.) In 1980 and again in 1987 IARC deemed these animal studies sufcient for demonstrating animal carcinogenicity for inorganic lead compounds [IARC, 1980, 1987]. However, almost all these studies used one form of lead (lead acetate, a soluble salt), chosen for ease of administration, given orally, subcutaneously, or intraperitoneally, usually at very high doses. Most human exposure to lead, on the other hand, is to lead oxides or lead fumes which are inhaled. HUMAN GENOTOXICITY STUDIES Winder and Bonin [1993] reviewed 19 cytogenetic studies of workers done from 1969 to 1986 with documented lead exposure in battery manufacturing and lead smelting. Sixteen of these studies showed increased chromosomal aberrations and/or sister chromatid exchange in peripheral lymphocytes, compared to nonexposed workers, although the authors do not report quantitative results or signicance tests. Several studies showed positive exposureresponse trends and three were longitudinal studies comparing workers before and after exposure. While a number of these studies were done in the early 1970s and methods for cyogenetic studies have developed and become more standardized, these studies do indicate that lead has cytogenetic effects in vivo. OCCUPATIONAL EXPOSURE High exposure has occurred among workers in lead smelters and lead battery plants (505000 mg/m3 in air, 40 100 mg/dl in blood) [Fu and Boffetta, 1995]. Moderate exposure has occurred among welders of metals containing lead or painted with lead (lead fumes), lead glass workers, lead miners, workers repairing automobile radiators, printers using lead type, and production workers using lead (e.g., producing lead chromate paint) (501000 mg/m3 in air, 2060 mg/dl in blood). Occupational exposures have decreased markedly since the 1950s in industrialized countries, and today blood levels in most workers in these countries rarely exceed 25 mg/dl, which might be considered high enough to be an occupational exposure by denition. In the US, for example, in 19761980, approximately 9% of U.S. males had blood levels above 25 mg/dl (the overall geometric mean for males was 15 mg/dl), while in 19881991 only 1% had levels above 25 mg/dl (overall male mean 4 mg/dl) [Pirkle et al., 1994]. The U.S. Occupational Safety and Health Administration (OSHA) limits for airborne lead are 50 mg/m3, and blood levels must be kept below 40 mg/dl. EPIDEMIOLOGY STUDIES The epidemiology is mostly based on highly exposed workers who were exposed decades ago. Ambient nonoccupational levels are much lower than occupational levels. This review of cancer epidemiology will focus on occupational exposures, and particularly on documented occupational exposure (e.g., with measurements of exposure levels or blood leads). Other recent reviews of the epidemiology have been published by Fu and Boffetta [1995] and by Hayes [1997]. Human epidemiology of lead-exposed workers has focused on all cancers: lung, stomach, kidney, and brain. We consider each of these, focusing on eight studies with high documented exposures below (7 cohort studies, 1 nested case-control) (Table I). These studies controlled for age, sex, TABLE1. Lead and Cancer: 8 Studies (7 Cohort,1Nested Case-Control) Study Wong et al. (1) [2000] Wong et al. (2) [2000] Steenland et al. [1992] Fanning [1988] Cocco et al. [1997] Lundstrom et al. [1997] Anttila et al. [1995] Gerhardsson et al. [1995] Country USA USA USA UK Italy Sweden Finland Sweden Population 4518 battery workers, 624 cancer deaths 2300 smelter workers, 273 cancer deaths 1990 smelter workers,192 cancer deaths 201cancer deaths among 2073 deaths of battery and other lead-exposed workers 1388 smelter workers,132 cancer cases 2353 smelter workers,172 cancer cases 20741persons w/blood lead across industries, 274 cases 664 smelter workers, 40 cancer cases Mean lead levels, comments Blood Pb 63 mg/dl,1947-72; external referent Blood Pb 80 mg/dl,1947-72; external referent Blood Pb 56 mg/dl,1976; external referent Nested case-control; High ( b 40 mg/dl) vs. low exp 48 mg/m3 air lead,1977; external regional referent Blood Pb 60 mg/dl in1950; external ref; 15-year latency; incident cases Apprx. 26 mg/dl1973; RRs for b 20 mg/dl vs. external referent; incident cases Blood Pb 60 mg/dl in1969; external local referent; incident cases Lead and Cancer in Humans 297 FIGURE1. Lead andall cancerrelativerisks: eightstudies.(*Internal referent;useof RRwith external referent does not change results appreciably.) Meta-analysis results (fixed effects) RR 1.04 (1.00^1.09). and calendar time, but not for smoking or diet. Six used external comparisons, while two used internal ones. Most recent published updates are used for all studies. Eight studies are not too many upon which to base conclusions, despite reasonably large numbers of cancer deaths or cases for most sites (all cancers 1911, lung 675, stomach 181, kidney 40, brain 69). Furthermore, a weakness of most studies is the lack of quantitative data on dose response, although a few studies do divide workers into high and low exposure groups. In the following, the results of each study are reported based on the whole population of exposed workers. While there is a plausible mechanism by which lead might increase all cancers (inhibiting DNA repair), the data do not support such a general increase. Results of studies of lead and all cancers are shown in Figure 1. A meta-analysis using a xed effects model (see below) results in a RR 1.04 (1.001.09). This is a negligible increase, and is not suggestive of a true association. All eight studies under consideration include results for lung cancer risk (Figure 2). The lung cancer results show signicant heterogeneity due to high relative risk in a study of Swedish smelter workers by Lundstrom et al. [1997]. In the presence of signicant heterogeneity, a random effects FIGURE 2. Lead and lung cancer relative risks: eight studies. (*Internal referent; use of RR with external referent does not change results appreciably.) Meta-analysis results (random effects): RR 1.30 (1.15^1.46) with Lundstrom et al. (fixed effects): RR 1.14 (1.04^1.25). model may be used to combine relative risks across studies. This model assumes that there is no single true relative risk underlying the different study ndings, but rather a distribution of true relative risks. Random effects models estimate a kind of average of these true relative risks, using a weighted average of study-specic relative risks in which the weights are the inverse of the variance of each studyspecic relative risk plus the variance of the study-specic relative risks. Random effects models typically result in a larger variance for this estimated relative risk than do xed effects models. Fixed effects models, conversely, are used when different studies yield relatively homogenous relative risks and it can be assumed that there is a single underlying true relative risk. The resulting weighted average of studyspecic relative risks uses weights which are the inverse of the study-specic relative risks without an added component for their variance across studies. The variance of the estimated common relative risk is generally smaller than in random effects models. For lung cancer, analysis using a random effects model yields a RR 1.30 (1.151.46). Without the ``outlier'' Lundstrom study a xed effects model can be used, resulting in a RR 1.14 (1.041.25). None of these studies control for smoking, which may be a confounder, especially for those studies in which workers are compared with the general population. The estimated overall relative risk with Lundstrom et al. [1997] included is within the upper range of what could plausibly be explained by smoking differences [Siemiatycki et al., 1992]. Overall, these studies show only a weak evidence of effect, and heterogeneity and possible confounding limit inference. The lung cancer ndings for lead are also potentially confounded by exposures to cadmium and/or arsenic (known lung carcinogens) in the lead smelter studies. On the other hand, in most of the smelter studies there was documentation that cadmium and arsenic levels were low. However, in the study with the highest relative risk [Lundstrom et al., 1997], there was signicant exposure to arsenic before 1950, a period also of the highest lung cancer risks; confounding by arsenic in this cohort is a possibility. Recent work in this cohort conrms signicant arsenic exposure for the lung cancer cases [Englyst et al., 1999], and the authors state that they cannot conclude whether the observed lung cancer excess is due to lead, arsenic, or a combination of the two. Further case-control studies are in progress in this cohort. Results for stomach cancer are shown in Figure 3, based on all eight studies (Lundstrom et al., [1997] did not break out stomach cancer; we used the results of an earlier followup by Gerhardsson et al., [1986]). A meta-analysis using a xed effects model results in a RR 1.34 (1.141.57). Confounding by diet is possible, although unlikely to explain completely the observed excess. Data for stomach are again suggestive but not conclusive. 298 Steenland and Boffetta FIGURE 3. Lead and stomach cancer relative risks: eightstudies. (*Upper 95% CI for RR for Gerhardsson is 5.50.) Meta-analysis results (fixed effects) RR 1.34 (1.14^1.57). FIGURE 5. Lead andbrain cancer relativerisks: sevenstudies.Meta-analysisresults(fixed effects) RR 1.06 (0.81^1.40). Additional information on stomach cancer is available from four case-control studies. Wong and Harris [2000] conducted a nested case-control study at one battery plant which had 30 stomach cancer deaths (representing half of the 60 stomach cancers in the Wong and Harris cohort study). No doseresponse was found using a variety of measures for estimated lead exposure. This study, therefore, weakens the case for a true association between lead and stomach cancer. On the other hand, three recent populationbased case-control studies using job-exposure matrices offer some support of a true association. Cocco et al. [1998a] found a signicantly (60%) increased risk of cancer of the gastric cardia for men likely to have been exposed to high levels of lead based on death certicate data for disease and usual occupation. Parent et al. [1997] found an odds ratio of 13.7 (2.284.6) for substantial exposure to lead, based on only three exposed cases, in a population-based case control study in Montreal. Wingren and Axelson [1987] showed some association between lead exposure and stomach cancer among Swedish glassworkers, but lead exposure could not be separated from exposure to other metals. Figure 4 shows the results for lead and kidney cancer based on seven studies. A meta-analysis using a xed effects model yields a relative risk of 1.01 (0.721.42). Overall, FIGURE 4. Lead and kidney cancer relative risks: seven studies. Meta-analysis results (fixed effects) RR 1.01 (0.72^1.42). there is little positive evidence for kidney cancer, the cancer found most consistently in animal studies. Although statistically these studies do not show signicant heterogeneity, overall numbers are small (40 deaths or cases total) limiting power to detect heterogeneity, and two studies do show an approximately two-fold excess. This suggests that kidney cancer remains a concern. Figure 5 shows the results for lead and brain cancer relative risks based on seven studies. Meta-analysis results (xed effects model) yield a relative risk of 1.06 (0.80 1.40). Like kidney cancer, there is little evidence for an excess risk of brain cancer, despite one suggestive study [Cocco et al., 1997], and some animal evidence for this site. The number of deaths or cases is again relatively small (n 69). However, there is supportive evidence for brain cancer from two other studies not included in Figure 5 because they are not cohort studies of exposed workers or because they have used a more restricted subset of all brain cancers. A U.S. study of death certicates and occupation found an odds ratio of 2.1 (1.14.0) for men with high probability of high level exposure (job exposure matrix, 27,000 brain cancer cases) [Cocco et al., 1998b]. Antilla et al. [1996] in a nested case-control study of 16 male glioma cases (a subset of the 26 brain cancers included in Figure 5) found an odds ratio of 11.0 (1.06.3) for those with blood lead b 28 mg/dl vs. those with less than 14 mg/dl (p for trend 0.03). These two studies suggest that brain cancer remains a concern. While most of the epidemiology concerns inorganic lead, there are two studies of workers exposed while producing tetraethyl lead that has been and in some countries is still used as an additive in gasoline [Robinson, 1974; Fayerweather et al., 1997]. Tetraethyl lead is metabolized to inorganic lead in the body. The study by Robinson [1974] was based on only 51 total deaths and is not informative. The more recent study [Fayerweather et al., 1997] found a signicant excess of rectal cancer (odds ratio 3.7, 1.310.2) with a positive exposureresponse relation- Lead and Cancer in Humans 299 ship based on estimated exposure. Colon cancer was also elevated, although not as markedly (odds ratio 1.3 (0.7 2.5), 16 exposed cases), and showed a positive exposure response. There is some evidence from one Swedish study of glassworkers who may ingest lead (odds ratio 1.7 (1.0 2.5), 18 exposed cases, Wingren and Axelson [1987]), but the eight principal studies reviewed here show no evidence of an excess of colon or rectal cancer. Differences in ndings for digestive cancer could be due to the absorption pattern of inorganic lead vs. organic lead. CONCLUSIONS Despite the fact that lead is one of the earliest recognized occupational toxins, there exist relatively few studies of cancer among lead-exposed workers with welldocumented high exposures. Such studies are the most informative for identifying whether lead causes cancer. In the eight principal studies with well-documented high exposures, the evidence is somewhat suggestive of an association with lung cancer and stomach cancer, but remains limited. Possible confounding by arsenic is a concern in the study with the highest lung cancer relative risk. In addition, there is weaker evidence of an association with kidney cancer and gliomas. Most of the epidemiologic studies unfortunately do not have data on doseresponse that would be expected to provide a better basis for inference than comparisons of exposed to nonexposed. Future information can be expected from increased follow-up of the newest cohort, 12,000 workers with blood lead levels in Sweden. New developments in measuring bone lead may offer a method of estimating cumulative exposure retrospectively [Borjesson et al., 1997; Hu et al., 1998]. REFERENCES Anttila A, Heikkila P, Mydyri E, Kauppinen T, Pukkala E, Hernberg S, Hemminki K. 1996. Risk of nervous system cancer among workers exposed to lead. J Occup Env Med 38:131136. Anttila A, Heikkila P, Pukkala E, Nykyri E, Kauppinen T, Hernberg S, Hemminki K. 1995. Excess lung cancer among workers exposed to lead. Scan J Work Environ Health 21:460469. Borjesson J, Gerhardsson L, Schutz A, Mattson S, Skerfving S, Osterberg K. 1997. In vivo measurements of lead in ngerbone in active and retired lead smelters. Arch Occup Environ Health 69: 97 105. Cocco P, Fu H, Boffetta P, Carta P, Flore C, Flore V, Onnis A, Picchiri G, Colin D. 1997. Mortality of Italian lead smelter workers. Scan J Work Environ Health 23:1523. Cocco P, Ward M, Dosemeci M. 1998a. Occupational risk factors for cancer of the gastric cardia. J Occup Env Med 10:855861. Cocco P, Dosemeci M, Heineman E. 1998b. Brain cancer and occupational exposure to lead. J Occup Env Med 40:937942. Englyst V, Lundstrom N, Gerhardsson L, Rylander L, Nordberg G. 1999. Determinants of lung cancer risks among lead exposed smelter workers, abstract, International Conference on Lead Exposure, Reproductive Toxicity, and Carcinogenicity, June 79, 1999, Gargano, Italy. Fanning D. 1988. A mortality study of lead workers, 19261985. Arch Environ Health 43:247251. Fayerweather W, Karns M, Nnuwayhid I, Nelson T. 1997. Case-control study of cancer risk in tetraethyl lead manufacturing. Am J Ind Med 31:28 35. Fu H, Boffetta P. 1995. Cancer and occupational exposure to inorganic lead compounds: a meta-analysis of published data. Occup Env Med 52:73 81. Gerhardsson L, Lundstrom N, Norberg G, Wall S. 1986. Mortality and lead exposure: a retrospective cohort study of Swedish smelter workers. Br J Ind Med 43:707712. Gerhardsson L, Hagmar L, Rylander L, Skerfving S. 1995. Mortality and cancer incidence among secondary lead smelter workers. Occ Env Med 52:667672. Hartwig A. 1994. Role of DNA repair inhibition in lead and cadmiuminduced genotoxicity: a review. Env Health Perspect 103, suppl 3: 45 50. Hayes R. 1997. The carcinogenicity of metals in humans. Cancer Causes Control 8:371385. Hu H, Rabinowitz M, Smith D. 1998. Bone lead as a biological marker in epidemiologic studies of chronic toxicity: conceptual paradigms. Env Health Persp 106:18. International Agency for Research on Cancer (IARC). 1980. Some metals and metallic compounds. IARC Monographs Vol. 23, pp 325415. International Agency for Research on Cancer (IARC). 1987. Overall evaluation of carcinogenicity: an updating of IARC Monographs Vols. 1 to 42, IARC Monographs 142: 230232. Lundstrom N, Norberg G, Englyst V, Gerhardsson L, Hagmar L, Jin T, Rylander L, Wall S. 1997. Cumulative lead exposure in relation to mortality and lung cancer morbidity in a cohort of primary smelter workers. Scan J Work Environ Health 23:2430. Parent M, Siemiatycki J, Fritschi L. 1997. Occupational exposure and gastric cancer, Epidemiol 9:4855. Pirkel J, Brody D, Gunter E, Kramer R, Paschal D, Plegal K, Matte T. 1994. The decline in blood lead levels in the US. J Am Med Assoc 272:284 291. Robinson T. 1974. 20-year mortality of tetraethyl lead workers. J Occup Med 16:601605. Siemiatycki J, Wacholder S, Dewar R, Cardis E, Greenwood C, Richardson L. 1988. Degree of confounding bias related to smoking, ethnic group, and socioeconomic status in estimates of the associations between occupation and cancer. J Occup Med 30:617625. Steenland K, Selevan S, Landrigan P. 1992. The mortality of lead smelter workers: an update. Am J Pub Health 82:16411644. Winder C, Bonin T. 1993. The genotoxicity of lead. Mutation Res 285:117 124. Wingren G, Axelson O. 1987. Mortality in the Swedish glassworks industry. Scan J Work Environ Health 13:412416. Wong O, Harris F. 2000. Cancer mortality study of employees at lead battery plants and lead smelters, 19471995. Am J Ind Med 38: 255270 (this issue).