Figure 1. The structural formula of the unsubstituted
biphenyl with numbering of the carbon atoms and
location of the ortho, meta, and para positions
PROCEEDINGS OF THE EXPERT PANEL WORKSHOP
TO EVALUATE THE PUBLIC HEALTH IMPLICATIONS
FOR THE TREATMENT AND DISPOSAL OF POLYCHLORINATED BIPHENYLS-
CONTAMINATED WASTE
Chapter 2 - Expert Panel Report
Health Effects Panel
James M. Melius, M.D., Dr.P.H. - CHAIR
Greg Steele, Dr.P.H., M.P.H. - RAPPORTEUR
Lee M. Sanderson, Ph.D. - CO-CHAIR
Obaid M. Faroon, Ph.D., D.V.M. - CO-CHAIR
As part of the Bloomington Polychlorinated Biphenyls Project, the Agency for Toxic Substances and Disease Registry (ATSDR) assembled a group of health experts to discuss and summarize relevant health issues, to identify data gaps, and to make recommendations regarding the public health implications of occupational and general environmental exposures to polychlorinated biphenyls (PCBs). The 10-member panel met in Bloomington, Indiana, on September 13 and 14, 1993.
Assessing the potential health effects of exposure to PCBs is a complex subject. While considerable research has been done in this area, much remains unknown or uncertain, and new research on this subject continues to expand our knowledge about the toxicity of these chemicals. Recent research has shown that PCB toxicity can occur by a number of mechanisms and that the chemical structures of the different PCB congeners are important in determining their toxic effects. However, these so-called structure activity relationships have not been worked out, and more research on this subject is needed. This issue is further complicated by the fact that nearly all exposure to PCBs in the workplace and general environment involves exposure to mixtures of the congeners and, often, to other toxic chemicals.
To date, a major focus of the health research on PCBs has been cancer. Higher-chlorinated PCB mixtures have been shown to be carcinogenic in laboratory animal experiments. However, studies of humans exposed to PCBs have been equivocal and inconclusive. Further research on PCB-related cancer is needed, including both laboratory studies and epidemiologic followup of exposed populations.
Public health concerns about adverse reproductive and developmental outcomes from PCB exposure have been increasing over the past few years. Recent research indicates that exposure to PCBs may cause reproductive and neurodevelopmental changes in exposed laboratory animals and in some people with environmental exposure to PCBs. Although not conclusive, the studies indicate the need for further research in this area.
Changes described in previous studies of PCB-exposed workers (such as chloracne) are less important for populations with environmental exposures. However, other health concerns related to PCB exposure, including immunologic and neurologic effects, require further study.
The panel briefly reviewed the health implications associated with remediation of PCB contamination. Evaluation of the health risks associated with different remediation techniques needs to focus on all toxicologic substances released by these techniques and should be based on a quantitative exposure analysis of the contaminants.
Finally, the panel made a number of specific recommendations regarding further research on the toxicity of PCBs and recommended development of an information repository on PCB research.
As part of the Agency for Toxic Substances and Disease Registry (ATSDR) Bloomington Polychlorinated Biphenyls (PCBs) Project, a Health Effects Panel of experts was convened to address issues related to the public health implications of occupational and general environmental exposures to PCBs. Members of this Health Panel were not asked to evaluate health issues associated with specific sites in the Bloomington area, but rather to use their complementary backgrounds and areas of expertise to provide an overview that addressed three specific objectives: 1) to discuss and summarize relevant health issues, 2) to identify data gaps, and 3) to make public health recommendations to ATSDR that will support the Agency's mission pertaining to the Bloomington PCB Project.
Potential candidates for the Health Panel were identified through nominations from the public or literature searches for health-related PCB publications. The members were selected so that the panel's composition would be representative with respect to relevant scientific disciplines and affiliations. The 10 panel members included four occupational medicine physicians (including one pulmonary physician from the local university medical center), two epidemiologists, three toxicologists, and a biologist. An epidemiologist and an environmental health scientist from ATSDR served as co-chairs and assisted with facilitating the functions of the Health Effects Panel. Biographical sketches of the panelists are provided in Appendix A.
Before convening the Panel, the co-chairs prepared a background discussion paper for the panelists to review. This background discussion paper identified numerous questions and issues for the panelists to contemplate during their deliberations. These questions and issues were used to develop an agenda for the Panel meeting, which was held on September 13 and 14, 1993, in Bloomington, Indiana. The questions and issues are included in Appendix B. To enhance the clarity of the report, a glossary of technical terms is presented in Appendix C.
During the 2-day meeting, the panelists addressed the questions and issues during their deliberations and discussed other PCB health-related issues as they arose. In their discussions, the Panel members did not attempt to reach a consensus of opinion, but rather expressed a broad spectrum of viewpoints. This report summarizes the highlights of those discussions along with salient information from the background discussion paper.
Commercial PCB products consisted of isometric mixtures of the 209 possible congeners. The structural formula of the unsubstituted biphenyl with the number of carbon atoms is shown in Figure 1. The molecular weights and the formulas for the 10 congeners are shown in Figure 2. The percentage of chlorine in the PCB mixtures generally varied between 18% and 68% on a per weight basis (WHO 1993). Commercial PCB products manufactured in the United States by Monsanto, the producer, were marketed under the trade name of "Aroclor" (Mieure 1975). Several commercial grades of PCBs were available and were designated by numbers such as 1221, 1242, 1254, and 1260; the first two digits stood for the biphenyl structure and the last two digits represented the percentage chlorine by weight in the mixture. Another grade, Aroclor 1016, which consisted of primarily tri- and tetrachlorobiphenyl compounds, contained 41% chlorine by weight and was introduced in 1971 to replace Aroclor 1242 (Mieure 1975).
Figure 1. The structural formula of the unsubstituted
biphenyl with numbering of the carbon atoms and
location of the ortho, meta, and para positions
Figure 2. The Molecular Weight and Formula of the PCB Isomers
The chlorobiphenyl constituents of several commercial PCB mixtures have been studied. In terms of potential exposure of human populations, approximately 130 of the 209 potential PCB congeners are considered likely to have been present in commercial products (WHO 1993). Those congeners that were chlorinated in only one ring were generally not found in commercial mixtures. Many commercial PCB mixtures were also inadvertently contaminated during production with chlorinated dibenzofurans and naphthalenes.
From 1929 until 1971, PCBs were produced in the United States for use in a wide variety of products and processes (in printing ink and plasticizers, as hydraulic fluids, and in transformers, capacitors, and heat exchangers). These compounds were extremely stable; were not hydrolyzed by water, acids, or alkali; and were able to withstand temperatures up to 600º C without decomposition. The major use of PCBs was as dielectric fluid in electrical equipment such as capacitors and in transformers in buildings. PCBs were selected for this application because their flame resistance provided greater safety than other fluids. Their stability and low vapor pressure made them suitable for lubricants, hydraulic fluids, liquid seals, cutting oils, and suppressants for some insecticides. The properties of persistence and stability that made them ideal for commercial use have also contributed to their status as major environmental contaminants.
1. Planar
Planar PCB congeners substituted in both para, at least two meta, and no ortho positions represent the most toxic members of this class of compounds. These congeners are "dioxin-like" in that they are approximate isostereomers of 2,3,7,8-TCDD, bind with high affinity to the Ah receptor, and induce 3-methylcholanthrene-inducible P-450 isozymes. Another set of PCB congeners consists of mono-ortho-coplanar congeners; this group exhibits lower competitive binding affinities for the Ah receptor and, in general, has moderate dioxin-like activity. Toxic effects observed in experimental animals attributable to the dioxin-like PCB congeners include chloracne and related dermal lesions, immunotoxicity, inhibition of body weight gain, and reproductive toxicity (ATSDR 1993; EPA 1991).
2. Non-planar
PCB congeners in this group have two or more chlorine substitutions in the ortho positions. The phenyl rings of a PCB molecule are not constrained through ring fusions and have relatively more rotational freedom than dioxins or dibenzofurans. Chlorine atoms occupying the ortho positions (2,2',6,6') limit the rotational freedom and therefore constrain the planarity of the phenyl rings (WHO 1993).
Non-dioxin-like congeners may also contribute to the overall toxicity of PCB mixtures. Some of these congeners may exert their toxicity via metabolism to reactive intermediates that can form potentially toxic, covalently bound, substrate-macromolecular adducts. Some non-coplanar PCBs are associated with neurochemical changes in the brain (Seegal et al. 1991).
3. Enzymes
The number and position of the chlorine atoms in the PCB molecule cause differences in enzyme induction and toxicity. As a general rule, the more highly chlorinated compounds (those with 48% chlorine or greater) are more potent enzyme inducers, more resistant to metabolism, and, for some endpoints, more toxic than the lower-chlorinated compounds (WHO 1993).
4. Toxicity Equivalency Factors (TEFs)
The public concerns about toxicologic effects resulting from exposure to PCBs and structurally related halogenated aromatic hydrocarbons are well recognized, as are the gaps in available information with which to evaluate the potential for human health effects from such exposures. In response to this problem, in December 1990, EPA held a workshop of approximately 30 experts on PCB-related issues to examine the existing toxicity and exposure database on PCBs as a means of determining the feasibility of developing TEFs for PCB congeners. The panelists concluded that the application of TEFs to dioxin-like PCBs congeners appeared feasible; these TEFs should be used to determine the contribution of dioxin-like PCBs to overall dioxin/furan risks. They pointed out that the toxicities of non-dioxin-like PCBs can be significant. However, they concluded that additional mechanism-of-action data were needed before attempting to develop TEFs for non-dioxin-like PCB congeners (EPA 1991).
Recent data suggest that TEFs for the "dioxin-like" polychlorinated biphenyls would overestimate the potency of these compounds by factors of 10 to 100 (DeVito et al. 1993).
Following the panel discussion, one panelist brought a new concept to our attention. Values similar to TEFs called relative proliferative effect (RPE) and relative proliferative potency (RPP) were derived for estrogen-like chemicals (Soto et al. 1992). The proliferative effect (PE) "is expressed as the ratio between the highest cell yield obtained with the test chemical and the hormone-free control. A value of 100 for RPE% indicates that the compound tested is a full agonist; a value of 0 indicates that the compound lacks estrogenicity at the dose tested" (Soto et al. 1992 ). The relative proliferative potency is the ratio between the dose of estradiol and that of the xenobiotic needed to produce maximal cell yield x 100. RPE% of 7 and PE of 1.4 were developed for Aroclor 1221 (Soto et al. 1992).
Discussions by the Panel indicated that there are at least three classes of PCBs in relation to their mechanism of actions: estrogen-like PCBs, dioxin-like PCBs, and phenobarbital-like PCBs.
1. Estrogen-like PCBs
Estrogen-like PCBs are short-lived PCBs. Some of these PCBs can be converted by the body to sulfate compounds that have an affinity for the lung and uterine tissues. Estrogen-like PCBs exert their action by reacting with estrogen receptors and initiating a cascade of events similar to that of endogenous estrogen. Some recent evidence indicated that this group of PCBs undergoes hydroxylation by the cytochrome P-450 enzyme system. The resulting metabolites then interact with estrogen receptors. Furthermore, some estrogenic effects might be indirect, i.e., involve the estrogen metabolism pathways by the P-450 system. Because PCBs induce P-450, the metabolism rate of estrogen and the metabolites may be different. These metabolites may also have health implications and may indirectly affect growth factor pathways.
2. Dioxin-like PCBs
For the dioxin-like PCB congeners, the expression of toxic responses is believed to be initiated by the binding of the congeners to the Ah receptor. These responses seem to vary, however, among animal species and strains. The responsiveness of a particular organ or cell depends on the presence of a functional cytosolic Ah receptor (EPA 1991). The biochemical processes that follow the initial binding to the receptor have not been extensively studied for individual PCB congeners; most of the existing information has been derived from studies of 2,3,7,8-TCDD. Initial binding to the Ah receptor is followed by a time-dependent depletion of the cytosolic Ah receptor that is paralleled by the appearance of the nuclear 2,3,7,8-TCDD:receptor complex. The formation of this complex is followed by induction of P-450 mRNA and microsomal cytochrome P-4501A1 (Safe 1990). The PCB congeners that exhibit Ah receptor-mediated responses constitute a small fraction of the total number of possible congeners that are routinely identified in environmental samples. The potency of the compounds that bind to the Ah receptor reflects binding affinity and biological persistence.
3. Phenobarbital-like PCBs
Polychlorinated biphenyls are powerful inducers of hepatic enzymes and other drug-metabolizing enzymes found in other tissues. Pretreatment of those PCB congeners with phenobarbital-like inducing activity results in increased levels of liver and other organ Phase I (microsomal) and Phase II (cytosolic and microsomal) drug-metabolizing enzymes in various animal species.
PCBs can be considered mixtures in two distinct ways. First, PCBs were used as mixtures of various congeners, as specified by the percentage chlorination (for example, 1242 or 1248). Second, in both occupational and environmental exposures, individuals are usually exposed to PCBs along with many other chemicals.
1. PCBs and Other Chemicals
a. Interaction
Many of the interactive effects of PCBs with other chemicals are related to the capacity of PCBs for enzyme induction. Simultaneous ingestion by rats of Aroclor 1254 or 1260 and other chemicals, such as the pesticides mirex, photomirex, or kepone or all three, resulted in increased severity of liver lesions (Chu et al. 1980). Interactive effects of PCBs have also been reported with ascorbic acid, 1,1-dichloroethylene, benzo(a)pyrene, acrylonitrile, many solvents, and cadmium (ATSDR 1993; WHO 1993).
b. Antagonism/Synergism
The co-administration of 2,2',4,4',5,5'-hexachlorobiphenyl with 2,3,7,8-TCDD resulted in antagonism of 2,3,7,8-TCDD-mediated teratogenicity (i.e., cleft palate) and immunotoxicity (Biegel et al. 1989). 2,3,3',4,4',5-Hexachlorobiphenyl, when co-administered with 2,3,7,8-TCDD to mice during gestation, resulted in a dose-related enhancement of the TCDD-induced hydronephrosis in mouse fetuses (ATSDR 1993). The genotoxicity of numerous carcinogens is potentiated in vitro by PCBs, but this does not indicate that PCBs should be regarded universally as tumor promoters because of the protective role PCBs play against carcinogenicity of many genotoxic carcinogens in vivo (Hayes 1987).
2. Congener Mixtures of PCBs
Interaction has been observed between individual PCB congeners. Synergistic genotoxicity (chromosome breakage) was reported in human lymphocyte cultures following simultaneous exposure to 3,4,3',4'-tetrachlorobiphenyl and 2,5,2',5'-tetrachlorobiphenyl congeners (Sargent et al. 1989).
1. Ingestion
Exposure to PCBs through food and drinking water has been reported. Small quantities (0.02-0.59 nanograms per liter [ng/liter]) were observed in open ocean water (ATSDR 1993). Surface water may be contaminated with PCBs from atmospheric fallout (ATSDR 1993). Food, primarily fish, but also meat and milk, appear to be the main source of human exposure to PCBs in the environment (Andersson et al. 1988; Mackay 1989). Food may be contaminated from PCBs in packaging materials, or directly as a result of faulty industrial processes (WHO 1993). In Japan and Taiwan, there have been two outbreaks of illness (Yusho and Yu-Cheng) resulting from ingestion of cooking oil contaminated with PCBs and other compounds (Uzawa et al. 1973; Chen et al. 1980).
2. Inhalation
Before the 1980s, PCBs were found in air in various locations around the world at mean concentrations ranging from 0.1-10 nanograms per cubic meter [ng/m3] (Eisenreich et al. 1981). These atmospheric concentrations may have decreased during the 1980s (Jones et al. 1992). Concentrations in the low part of this range were reported for sampling locations near three PCB-contaminated landfills (Hermanson and Hites 1989). Vapor concentrations correlated with temperatures and were higher in urban than rural areas. Higher levels of PCBs have been found in indoor air, particularly in kitchens and offices with electrical equipment. In industrial settings, higher levels (micrograms per cubic meter [µg/m3]) have been reported. For example, in the manufacture of transformers or capacitors, levels of more than 1000 µg/m3 have been measured (Brown 1987). Most of the information regarding the health effects of chronic PCB exposure in humans has been obtained from occupational exposure studies.
3. Dermal
Results from a number of human and animal studies have demonstrated that PCBs can penetrate the skin. Local and systemic effects were observed after dermal applications of PCBs. In animals, hyperplasia, hyperkeratosis, alopecia, edema, folliculitis, and acne were observed after exposure to PCBs (Allen and Norback 1976; NCI 1978; Vos and Notenboom-Ram 1972; WHO 1993 ). It is apparent that dermal lesions in animals exposed to PCBs are similar to the effects observed in humans, depending on the levels of exposure and the presence of contaminants in the PCBs. However, different animal species have different susceptibilities (Safe 1984).
4. Transplacental
Animal studies have indicated that PCBs can cross the placental barrier and accumulate in the tissues of fetuses. Maternal ingestion of PCBs affects serum PCB levels. And, in turn, there is an association between maternal serum PCB levels and umbilical cord PCB levels. Therefore, the exposures of fetuses are probably correlated with maternal exposure (Jacobson et al. 1984; Fein 1984). Transplacental PCB toxicity has been reported in mice, monkeys, and humans (Allen et al. 1974; Haake et al. 1987; Jacobson et al. 1984).
The differences in the carcinogenic potential of PCB mixtures may be based upon the degree of chlorination. The higher-chlorinated congeners have been shown to be more carcinogenic than the lower-chlorinated mixtures. In animal studies, PCBs containing 60% chlorine by weight were carcinogenic (Schaeffer et al. 1984; Kimbrough et al. 1975; Norback and Weltman 1985). Promotion or inhibition of cancer by PCBs has been reported, depending on individual congeners and order of administration of the initiator and the PCBs (Preston et al. 1981; Tatematsu et al. 1979; ATSDR 1993; WHO 1993).
Hepatic neoplastic nodules, trabecular carcinomas, and hepatocellular carcinomas have developed in rats fed a diet containing PCBs (60% chlorine by weight) (Schaeffer et al. 1984; Norback and Weltman 1985). The progression of the lesions appeared to be time dependent. Nodular hyperplasia was twice as prevalent as hepatomas in the animals fed high doses. The incidence of hepatomas was also significantly increased in mice fed a diet containing a PCB mixture (54% chlorine by weight) for 11 months (Kimbrough and Linder 1974). No proliferative lesions (nodular hyperplasia or hepatocellular carcinoma) were observed in groups of 12 mice fed estimated doses of equal to or less than 65 milligrams per kilogram per day (mg/kg/day) of Kanechlor 400 (48% chlorine by weight) or Kanechlor 300 (40%-42% chlorine by weight) for 32 weeks (Ito et al. 1973).
Assessments of an association between PCB exposure and the occurrence of human cancer have primarily involved evaluations of mortality in groups with occupational exposures. To a lesser degree, evaluations of occupational cancer incidence and evaluations of cancer mortality in groups with environmental exposure have been conducted.
1. Occupational
Review of published literature, completed dissertations, and unpublished reports has identified nine different assessments of mortality experienced by occupational subpopulations known or suspected to have been exposed to PCBs (Zack and Musch 1979; Gustavsson et al. 1986; Bertazzi et al. 1987; Brown 1987; Nicholson 1987; Taylor 1988; Liss 1989; Salvan 1990; Sinks et al. 1991 ). All assessments were historical cohort studies, except for one mortality case-control study (Salvan 1990), which had a primary focus on occupational exposures other than to PCBs and little information and findings pertaining to PCBs. With respect to the historical cohort studies, the number of subjects and accompanying person-years at risk varied considerably, from 85 employees and 1,800 person-years at risk (Zack and Musch 1979) to 6,292 employees and 122,782 person-years at risk (Taylor 1988). The male-to-female ratios also varied; two studies had cohorts consisting only of males because of PCB exposures occurring in departments that did not employ women (Zack and Musch 1979; Gustavsson et al. 1986). For all but one of these mortality studies (Zack and Musch 1979), exposure information about the type of PCB mixture was provided.
Excess mortality from liver cancer was assessed specifically in four occupational studies and as a combined category including gall bladder and bile duct cancer in three other studies. Of the seven studies, only one (combined category, Brown 1987) showed a statistically significant excess of liver cancer. With respect to lymphatic/hematologic cancer, a subset of one cohort (Nicholson 1987) showed a statistically significant excess, and two studies (Bertazzi et al. 1987 and Liss 1989) demonstrated excesses that did not reach statistical significance. One study (Liss 1989 ) reported results for cancer of the prostate and showed a statistically significant excess. That study was also one of four that reported results for cancers of the brain and central nervous system; Liss found a statistically significant excess. A second study (Sinks et al. 1991) showed a non-statistically significant excess of brain and central nervous system cancer. The second study also showed a statistically significant excess of malignant melanoma. The largest cohort study (Taylor 1988) (with nearly twice the person-years at risk and more than twice the number of observed deaths of the next largest study) did not show any significant excesses of either total cancer or any anatomic, site-specific cancers. The meta-analysis by Nicholson that involved the aggregation of original data from four different studies (Bertazzi et al. 1987; Brown 1987; Gustavsson et al. 1986; and Nicholson 1987) indicated a statistically significant excess of cancer of the liver/biliary passages/gall bladder; all of the cases occurred after 10 or more years of latency. It is important to note that several occupational studies (Brown 1987; Nicholson 1987; Taylor 1988) focused upon workers at the same plant; these results cannot be compared independently of each other. With respect to occupational evaluations of cancer incidence, a review of medical records of 72 PCB-exposed workers showed a statistically significant increase in the incidence of malignant melanomas (3 observed cases versus 0.1 expected) and of pancreatic cancer (two observed cases versus 0.2 expected) (Bahn 1976; Bahn et al. 1976). Contributions by concomitant exposures and other risk factors were not evaluated (Lawrence 1977). A survey of 55 transformer workers and 56 non-exposed comparison workers found two exposed workers with a reported history of melanoma. No cases were reported in the comparison group (Emmett et al. 1988). This difference was not statistically significant. The historical cohort mortality study by Liss also evaluated cancer incidence and showed a statistically significant excess of brain cancer and non-statistically significant excesses of prostate and pulmonary cancer (Liss 1989). Assessment of cancer incidence among 142 Swedish capacitor workers showed no difference from expectation for total cancer (Gustavsson et al. 1986 ).
2. Non-Occupational
Most studies of non-occupational exposures to PCBs and the risk of cancer have focused upon persons exposed during the Yusho and Yu-Cheng episodes. During 1968, in the western part of Japan, the Yusho episode involved a mass food poisoning of more than 1,850 persons who ingested rice bran oil ("Yusho" oil) accidentally contaminated with PCBs (Kanechlor 400 with 48% chlorine content) (Kamrin and Fisher 1989; Cordle et al. 1978). Although PCBs were initially viewed as the cause of the Yusho poisoning, analysis of samples of the contaminated oil indicated high levels of polychlorinated dibenzofurans (PCDFs) and polychlorinated quaterphenyls (PCQs) (Chen et al. 1985). The majority of the toxicity associated with the contaminated oil was believed to be from the PCDFs (Kamrin and Fisher 1989). Another similar poisoning episode involved approximately 2,000 persons in Taiwan in 1979 (Chen et al. 1980; Chen et al. 1985). This episode was called PCB poisoning or Yu-Cheng (oil disease). The PCB mixtures in the contaminated oil consisted of Kanechlor 400 and Kanechlor 500. It was concluded that PCDFs were the primary causal agent of the Yu-Cheng epidemic (Kashimoto et al. 1985).
A published case series analysis of mortality among 2,061 Yu-Cheng patients showed that by 1983 there were 28 deaths, but specific mortality information pertaining to cancer was not provided (Hsu et al. 1985). An historical cohort mortality study of a subgroup of Yusho patients (n = 1,086) identified a statistically significant excess of liver cancer mortality in males and a statistically nonsignificant excess of liver cancer mortality in females (Amano et al. 1984). Another historical cohort study of mortality among 1,761 Yusho patients showed a statistically significant excess in males, but not females, of deaths from all cancers (Nakamura et al. 1986; Kuratsune et al. 1988). In this study, statistically significant excesses of cancers of the liver and respiratory system were observed for males but not for females. Because of high rates of liver cancer from hepatitis B, regional analysis showed an inconsistent geographic disparity that could not be explained solely by exposure to the contaminated rice oil.
Breast cancer is the most prevalent cancer in American women, and has been shown to be estrogen dependent. Concerns have been raised regarding the estrogenic potential of organochlorine residues such as PCBs. Several studies have attempted to study the relationship between PCB residues in human serum and breast tissue and breast cancer. The results of these studies have been inconsistent. All of the studies have methodologic limitations; none controlled for other known or suspected breast cancer factors, and all dealt with relatively small numbers of cases.
One study compared the levels of PCB, DDE, and DDT residues in mammary adipose tissue from women with malignant and non-malignant breast disease (Falck et al. 1992). Twenty-three fat samples were taken from women with mammary carcinoma, and an additional 27 samples were taken from women who had only benign breast disease. Statistically significant elevated levels of PCBs, DDE, and DDT were found in the fat samples from women with cancer.
A small study (Unger et al. 1984) consisting of only 14 cases and 21 controls, did not find a difference between the two groups in PCB or DDE concentrations in human breast fat.
A large, nested case-control study (Wolff et al. 1993) investigated the serum DDE and total PCB levels in 58 women diagnosed with breast cancer and 171 matched control subjects from the same study population who did not develop breast cancer. In this study population, breast cancer was strongly associated with serum DDE concentrations, but was not associated with PCBs. The investigators, did not, however, control serum levels for lipids.
1. IARC
The available information on the possible carcinogenic risk of humans exposed to PCBs was derived primarily from studies of occupational groups. These occupational studies had a number of limitations; therefore, the evidence of PCB carcinogenicity in humans is limited. Sufficient evidence exists of the carcinogenicity of PCB mixtures (greater than 50% chlorination) in animals. Hence, IARC (1987) classified PCBs as a probable human carcinogen (Group 2A).
2. EPA
Human data are inadequate but suggestive because of confounding exposures or lack of exposure quantification. However, animal studies indicated sufficient evidence of carcinogenicity in 3 strains of rats and 2 strains of mice (EPA 1993a). Therefore, EPA has classified PCB mixtures as a probable human carcinogen (B2) (EPA 1993a).
3. Panel Assessment
Numerous human studies have examined the potential association between exposure to PCBs and the occurrence of specific types of cancer. The majority of the studies have been assessments of mortality experienced by workers known or suspected to have been exposed to commercial PCB mixtures. The findings and limitations of a number of these studies were discussed during the Panel's deliberations.
The Panel concluded that the data on carcinogenicity from human studies are limited and equivocal. Increased incidences of cancers of the hepato-biliary tract have been observed in several studies. However, this association was not judged to be conclusive. The observation of a statistically significant excess risk of cancer was observed in two occupational studies (one plant study and the other a meta-analysis of four cohorts), but not in the other studies. Two mortality assessments of the Yusho episode indicated statistically significant excesses of liver cancer. However, the major contaminant of toxicologic concern in the Yusho episode was PCDFs. Some members of the Panel believe that the association with hepato-biliary tumors appears to be consistent with findings from some animal studies and with results from epidemiologic analyses of dioxin exposures at Seveso, Italy (Bertazzi et al. 1993). One panelist stated that this consistency may suggest that if the association between PCB exposure and hepato-biliary tumors is real, the exposure mechanism involved would most likely include dioxin-like PCBs.
With respect to other anatomic sites, there may be a weaker association between PCB exposure and excess mortality from lymphomas and leukemia. Two studies and a meta-analysis showed non-statistically significant excesses of such cancers, and animal data are limited but supportive of this association (NCI 1978). Among workers exposed to PCBs, single studies showed statistically significant excesses of malignant melanoma and brain cancer, each of which was reported in one additional human study. These associations are less well supported by animal studies. Several recent epidemiologic studies have looked at the association between breast cancer and PCB exposure. The Panel believes that the results of those studies have also been inconclusive.
Existing epidemiologic studies have many common characteristics and limitations, many of which are typical of most studies involving occupational or environmental epidemiology. These common traits include a focus on all occupational or special non-occupational populations; exposures not only to PCBs, but also to PCB-related contaminants such as PCDFs, as well as other contaminants; supposed relatively high exposures and resulting doses (good environmental data on actual exposures were often lacking); small study populations; and small numbers of total and site-specific cancer deaths as a result of limited followup of exposed populations.
Discussion of these traits included consideration of potential contributions of the healthy worker effect and the issue of sensitive subpopulations. Because a number of factors (disease outcome, job-related tasks, occupation and industry of employment, definition of cohort, etc.) affect the magnitude of the healthy worker effect, it is not clear if or how the healthy worker effect might affect the magnitude of any risk estimates for cancer. One panelist mentioned that studies have shown that the healthy worker effect diminishes with sufficient follow up of the cohort. The issue with sensitive subpopulations is whether certain people are at increased risk from the same exposure (dose). There is some evidence of inter-individual variation; for example, some people get chloracne from high levels of exposure and others do not. Individual differences in metabolism and age are important factors for tissue sensitivity. The mechanism for this variability and whether this type of sensitivity contributes to a higher risk of cancer are unclear.
Discussion of the limitations of these studies included suggestions on how to minimize them in the future. First, because approximately 80% to 90% of the members of all occupational cohorts are still alive, additional followup would be beneficial. Use of incidence data and medical records to verify death certificates would be helpful.
In considering populations not yet studied, the Panel discussions focused primarily upon the idea of studying the children of parents with documented PCB exposures. Practical limitations of studying such a population were discussed, and it was suggested that consideration be given to evaluating the feasibility of identifying a sufficiently large second-generation population through one of the larger occupational cohorts.
With respect to cancer etiology, data from animal studies indicate that PCBs are tumor promoters rather than tumor initiators. Initiators damage DNA directly while promoters give a selective growth advantage to already damaged cells. Although human data are inconclusive, PCBs could have a far greater impact on cancer etiology when PCB-exposed persons are concurrently, or have previously been, exposed to chemicals or contaminants that are tumor initiators. This has potential health implications for occupationally or environmentally exposed persons, as well as for persons exposed through remediation strategies that may result in exposure to substances besides PCBs.
In animal studies performed to date, lower-chlorinated PCBs (less than 50% chlorine) have been shown to be less carcinogenic in long-term bioassays (IEHR 1991; IARC 1987). There are a number of possible explanations for these observations, including the fact that the materials may require longer latency periods; that they do not accumulate in the liver to the same degree (e.g., they are removed through metabolism); that they may not contain the specific congeners found in the 60% mixtures responsible for carcinogenicity; that higher doses may be required for these mixtures to cause cancer, etc. Current science cannot distinguish between these possible explanations, and more research is required before these differences can be explained.
Ongoing research designed to address data gaps was discussed. The General Electric Company has initiated a study involving different congener and dose groups to evaluate congener-specific bioaccumulation, enzyme induction, and carcinogenicity in rats. Several panelists believe that this study, which should be completed by the end of 1995, is apparently the only ongoing animal study of PCB exposure. General Electric is also completing a 10-year mortality update of the worker cohort exposed to PCBs. NIOSH is undertaking updates of capacitor worker cohorts in New York, Massachusetts, and Indiana. The National Cancer Institute is funding the expansion of several breast cancer and PCB exposure studies to increase the sizes of the study cohorts and the lengths of followup.
IV. Reproductive/developmental
The fetuses from rats and mice whose dams were exposed to PCB mixtures (Aroclor 1254) during pregnancy did not show malformations or cleft palates (Haake et al. 1987). However, a dose-related increase in embryotoxicity and in the incidence of malformed fetuses, mainly showing cleft palate and hydronephrosis, was observed in mice after daily exposure to the planar 3,4,5,3',4',5'-hexachlorobiphenyl and 3,4,3',4'-tetrachlorobiphenyl, but not to 4,4'-dichlorobiphenyl or 3,5,3',5'-tetrachlorobiphenyl (Marks et al. 1981, 1989).
The available data from animal studies have indicated that PCBs have developmental effects similar to those reported in humans (Fein et al. 1984; Pantaleoni et al. 1988; Taylor et al. 1989). The most sensitive developmental endpoints appeared to be those involving neurobehavioral functions. Learning deficits were observed in offspring of monkeys exposed to low doses (2.5 parts per million [ppm] Aroclor 1248; 1.0 ppm Aroclor 1016) of PCBs before or during gestation and lactation (Bowman et al. 1978; Levin et al. 1988). Behavioral deficits were also seen in offspring of rats exposed to PCBs during pregnancy (Tilson et al. 1979, 1990; Pantaleoni et al. 1988).
Embryos and fetuses may be sensitive to the toxic effects of PCBs because they generally lack the hepatic microsomal enzyme systems that are responsible for the detoxification and elimination of PCBs. With respect to transplacental exposure, PCBs have been detected in umbilical tissues, umbilical blood, and amniotic fluid. PCB concentrations in these tissues and fluids were considerably less than in the mother's blood. Studies that have investigated maternal serum concentrations and cord serum concentrations in women exposed to PCBs by consuming contaminated fish have found higher PCB concentrations in the mother's blood (Jacobson et al. 1984 ; Bush et al. 1984). Elevated umbilical cord serum concentrations of PCBs have been associated with reduced size, shorter gestational age, and neonatal behavioral effects such as delayed autonomic maturity, increased numbers of abnormal reflexes, poor visual recognition memory, and decreased intelligence scores (Fein 1984; Fein et al. 1984; Jacobson et al. 1984; Jacobson et al. 1985; Rogan et al. 1986a, 1986b, 1987, 1988; Gladen and Rogan 1988; Gladen et al. 1988 ). These results should be viewed cautiously. Although increased levels of PCBs in cord blood may be predictors of these effects, the signs or symptoms are not well validated. They may be attributable to uncontrolled confounding factors such as exposure to other chemicals, lifestyle factors, acute illnesses, and genetic predisposition. The maternal serum PCB levels were often within the range of the general population.
The Yusho episode resulted in some fetuses being poisoned at the time their mothers ingested the oil, while other pregnancies were affected in subsequent years from residual contamination in the mothers' bodies. Some of the characteristics of transplacental exposure include intrauterine growth retardation, brown staining of the skin and mucous membranes ("cola-colored skin"), natal teeth, open fontanelles and sagittal suture, and overgrowth of the gingiva (Miller 1985). A study of 128 children who were transplacentally exposed to PCBs from the Yu-Cheng episode showed similar findings (Gladen et al. 1990). Compared to 115 control children, the exposed children at birth had increased occurrences of hyperpigmentation, eyelid swelling and discharge, deformed nails, acne, natal teeth, and swollen gums.
An epidemiologic evaluation of 51 births to women employed in suspected high-PCB-exposure areas of two capacitor manufacturing plants showed that these infants had lower birth weights than 337 infants born to women who worked in low-exposure areas of the plant (Taylor et al. 1984 ). The authors concluded that the difference in birth weight was attributable to a shortened gestation period rather than to retardation of intrauterine growth. Additional reported reproductive effects to offspring of women occupationally or non-occupationally exposed to PCBs included smaller head circumference (Fein et al. 1984) and early impairment of psychomotor development (Gladen et al. 1988). In all three studies, the mean differences between exposed and unexposed groups were small, and observed effects were either not clinically significant or relatively minimal compared with those from other factors that may affect growth and development.
Studies have shown that PCBs can concentrate in human breast milk. PCB levels in breast milk fat are 100 to 200 times higher than in maternal serum (Kimbrough 1987); the concentration of PCBs in breast milk is 4 to 10 times that found in maternal serum (Wolff 1983; Jacobson et al. 1984 ). Several studies have found that the average PCB concentrations are below 2 ppm in milk fat and below 100 parts per billion [ppb] in whole milk (Jensen et al. 1980; Jensen 1983; Rogan et al. 1987 ). Breast-fed infants also may be at increased risk because of a steroid excreted in the breast milk that inhibits glucuronyl transferase activity, and thus glucuronidation and excretion of PCBs (Calabrese and Sorenson 1977; Gartner and Arias 1966).
Three ortho-substituted nonplanar PCB congeners have been observed to decrease both cellular dopamine concentrations and dopamine function in nonhuman primates (Seegal et al. 1990). Decreased cellular dopamine concentrations can alter catecholamine neurochemistry. Catecholamines have been shown to be a factor in the control of attention and learning disorders. Several studies of human populations exposed to PCBs perinatally have shown signs of deficits in visual recognition memory and decreased intelligence scores (Jacobson et al. 1985; Rogan et al. 1986a , 1986b, 1987, 1988; Gladen and Rogan 1988; Gladen et al. 1988). These findings are supported by studies of nonhuman primate infants that have shown analogous behavioral and cognitive dysfunctions (Mele et al. 1986; Schantz et al. 1989).
One of the most extensive studies of human developmental impacts of thermally degraded PCB exposure focused upon children born to Yu-Cheng mothers. Cognitive development of 118 children of Yu-Cheng mothers was compared to controls matched for age, sex, neighborhood, maternal age, parental education, and occupational class (Chen et al. 1992). The evaluation also included 15 older siblings of exposed children, who were born before the poisoning. Cognitive development was measured by using Chinese versions of the Stanford-Binet test and the Wechsler Intelligence Scale for Children, Revised. The exposed children scored approximately 5 points lower on the Stanford-Binet test at the ages of 4 or 5 years, and approximately 5 points lower on the Wechsler Intelligence Scale at 6 and 7 years. Children born up to 6 years after their mothers' exposure were as affected as children born within a year or two of exposure, when examined at 6 and 7 years of age. Cognitive development of older unexposed siblings was more similar to control children than to exposed siblings. The authors concluded that children prenatally exposed to thermally degraded PCBs had poorer cognitive development than matched controls. The effect persisted in children up to the age of 7 years, and children born years after maternal exposure were still affected.
Specific reproductive effects have been described in the scientific literature as being associated with PCB exposures. Dermal and mucous membrane effects have been documented in children born of mothers who were exposed in the Yusho or Yu-Cheng episodes. Decreased birth weight has been observed in not only these children, but also in children of workers who were occupationally exposed to PCBs. A panelist reported that the association of reproductive effects with PCB exposure has been supported by wildlife studies; female seals and other delayed implantation mammals have increased risk of spontaneous abortion following exposure to PCBs. The issue of potential human infertility was also discussed. One panelist mentioned that a recently published study involving a toxicologic evaluation in Rhesus monkeys indicated a significant relationship between endometriosis and doses of dioxin (25 parts per trillion [ppt] for 10 years) (Rier et al. 1993). However, no additional animal or human data are available with which to evaluate whether there is a relationship between endometriosis and PCB exposure. A panelist mentioned that epidemiologic follow up of Yu-Cheng children showed a significant risk of decreased male organ size, decreased height for girls, and decreased IQ for both sexes.
Many issues inherent to these studies affect the ability to make definitive conclusions about increased risks of human reproductive effects. The Yusho and Yu-Cheng episodes involved exposures to high levels of PCB mixtures, including significant levels of PCDFs and other compounds. After adjusting for gestation period, decreased birth weight disappeared in the study involving parental occupational exposures. In extrapolating animal results to humans, it is unclear where human sensitivity falls in the wide spectrum of effects observed among animal species.
Despite these limitations, increased public health concern about potential reproductive effects is justified because of potential measurable effects that result from relatively low doses. There are far more people in the general population who have been exposed at these lower levels than there are workers in occupational groups who have sustained much higher levels of exposures. The extent and importance of these potential reproductive effects have not been fully evaluated.
Studies have suggested that developmental effects associated with PCB exposure include cognitive intelligence and attention behavior decrements. Studies of children of the Yu-Cheng patients and children of consumers of contaminated fish in Michigan have shown IQ decreases of approximately 5 points. In addition to the issue of concomitant exposure to PCDF contamination, an important limitation of the Yu-Cheng study was the fact that the mothers were ill from the exposures, and it is unclear whether the exposures or resulting illnesses might have caused the developmental problems. Limitations of studies with respect to attention behavioral deficits included inconsistent outcomes (hyperactivity versus hypotonicity) and maternal serum levels not statistically different from control populations, or even when statistically significantly different, still within the general population range. Mean maternal blood PCB levels in the Michigan and North Carolina studies were 9.0 and 5.5 ppb, respectively. These studies of developmental effects often lacked ability to control for confounders such as smoking, alcohol use, and level of prenatal care.
More definitive studies need to be conducted. It is often more difficult to document functional, rather than structural, effects given the current level and type of existing medical records and public health surveillance systems.
Neurologic changes (decreased dopamine in caudate nucleus, decreased response to pain stimuli) have been observed in rats exposed to PCB mixtures. At higher doses (6,000 mg/kg/day), ataxia and coma were induced in these animals (Bruckner et al. 1973). Adult male non-human primates were orally exposed to corn oil containing either Aroclor 1016 or 1260 at doses of 0.8, 1.6, or 3.2 mg/kg/day for 20 weeks. Dose-related decreases in dopamine content in the caudate, putamen, substantia nigra, and hypothalamus were observed in monkeys exposed to Aroclor 1016 or Aroclor 1260 (Seegal et al. 1991).
With respect to neurologic manifestations, a study of American, PCB-exposed, capacitor manufacturing employees showed that 23.6% experienced headache and 17.5% experienced acroparesthesia (Fischbein et al. 1979). In the Yusho and Yu-Cheng episodes, headache was reported in 59.6% of a series of Japanese patients and in 34.8% of a series of Chinese patients (Chen et al. 1985). By measuring nerve conduction velocities (NCV), sensory neuropathy was reported in 47.6% of 21 Japanese patients and 44.5% of 155 Chinese patients (Chen et al. 1985). In a study of 110 Japanese patients, patients with higher PCQ blood levels had significantly slower median nerve sensory NCV than those with lower PCQ levels (Chen et al. 1985). In an occupational study of 15 workers exposed to PCBs after a capacitor explosion, the NCV of selected sensory and motor fibers were significantly reduced, but then recovered to normal velocities within 6 months after exposure (Elo et al. 1985). In a case-control study, 14 firemen who were exposed for 15-30 minutes to a fire that involved PCBs were given a battery of psychometric and neurobehavioral tests and scored worse than controls on 13 of the 22 tests (Kilburn et al. 1989), although subsequent analysis implied that scores on only two tests (fatigue and vigor) were statistically significantly lower (Mustacchi 1991).
It is often difficult to distinguish between neurologic and neurodevelopmental effects. Some of the potential conditions presented in the section on developmental effects certainly have a neurologic basis.
Animal studies indicate that neurodevelopmental and neurobehavioral effects are among the most sensitive indicators of toxicity in animals. The same trend may also be true for humans, but there are related issues that hinder definitive conclusions. Existing measures for neurologic effects are imprecise, and very subtle changes are not detectable. The neurologic effects involve a dynamic system for which human testing may not currently be sufficient for epidemiologic studies. Also, it is not clear how much reduction in cholinesterase or dopamine must occur before one can conclude that an adverse rather than a dynamic effect has occurred. A NOAEL, LOAEL, or serious LOAEL for reduced neurochemical levels in the brain cannot be established at this time because of insufficient knowledge of mechanisms and predictive values for health effects from these reductions.
In human studies, potential health effects appear to differ in adults who are occupationally exposed compared with children who are exposed as fetuses or infants. Occupationally exposed workers have exhibited effects on nerve conduction velocities, but the results are reversible and not striking. Public health implications appear to be far greater for children.
During September 1992, the Environmental Protection Agency (EPA) held a workshop for evaluating developmental neurotoxic effects associated with exposure to PCBs (EPA 1993b). The EPA concluded that available scientific data and effects consistently observed in animals and humans are sufficient to conclude that PCBs cause developmental neurotoxicity. Prenatal exposure to PCBs may be more detrimental than postnatal exposure, even though the levels of exposure through ingestion of breast milk may be greater. More research is needed on the neurodevelopmental effects of PCBs. Data needs include the replication of current data in additional human and animal studies and the completion of comparative studies to determine whether the effects of various mixtures of PCB congeners are similar or different.
The available data on rats, rabbits, mice, and monkeys indicate immunological changes resulting from exposure to PCBs. Decreases in natural killer cells and thymus weight were observed in rats and rabbits (Smialowicz et al. 1989; Street and Sharma 1975). Recent data show that the immune system of monkeys appears to be one of the most sensitive indicators of PCB exposure. Monkeys exposed to very low levels (5 µg/kg/day of Aroclor 1254) had a significant decrease in IgG and IgM immunoglobulin levels in primary response to challenge with sheep red blood cells (Tryphonas et al. 1989). Also, elevation of complement (CH50 level) in monkeys exposed to Aroclor 1254 has been observed. Elevated complement levels have been associated with rheumatoid arthritis and systemic lupus erythematosus in humans.
Followup of 30 Yu-Cheng patients and 23 controls showed that at 1 year after exposure to thermally degraded PCBs, Yu-Cheng patients had statistically significant decreases in concentrations of IgM and IgA, but not of IgG; decreased percentages of T cells, active T cells, and helper T cells, normal percentages of B cells and suppressor T cells; enhancement of lymphocyte spontaneous proliferation; and enhancement of lymphocyte proliferation with phytohemagglutinin (PHA) and pokeweed mitogen (PWM), but not concanavalin A (ConA). Follow-up evaluation 3 years later showed decreased PCB levels, normal percentages of total T cells; increased percentages of suppressor T cells; and enhancement of lymphocyte proliferation spontaneously or under the stimulation of various mitogens.
In a study of 15 workers exposed to PCBs, PCDFs, and other degradation products from explosions of capacitors, 6 months of follow-up studies evaluated cellular immune mechanisms by determining the number of T cells, the T-helper/T-suppressor cell ratio, and responses of T cells to phytohemagglutinin, pokeweed mitogen, and concanavalin A (Elo et al. 1985). The total number of T cells was lower in all subjects 5 weeks after the incident; approximately 3 months after the incident, 4 of the 7 subjects were normal. The T-helper/T-suppressor cell ratio was low (< 1.4) in 6 of the 7 subjects 5 weeks after the incident; approximately 6 months after the incident, 5 of the 7 subjects were normal. The response to PHA, ConA, and PWM was lower in 4/10, 3/10, and 5/9 exposed workers, respectively, 5 weeks after the incident; at 6 months, abnormal results were seen in 0/5, 1/5, and 0/5 workers, respectively.
Limited testing of immunocompetency of workers exposed occupationally to "normal" (uncontaminated) PCBs has not identified any evidence of reduced immune function. For example, Emmett found no differences in cutaneous delayed hypersensitivity responses to mumps and to trichophyton antigens between PCB-exposed transformer repairmen and controls (Emmett et al. 1988).
In Seveso, Italy, children exposed to TCDD, an immunotoxicant, had elevated complement levels comparable to those seen in other exposed individuals or in appropriate animal models (Tryphonas et al. 1991).
A Panel of the National Academy of Sciences has reviewed the issue of how much of a decrease in the immune system parameters, at least in IgG or IgM or both, is needed to be considered an adverse health effect. This Panel could not make a determination for individuals. A better approach may involve looking at the population level to gather information on these issues or to look at changes that occurred in wildlife. Various species have different immunosensitivity for PCBs. For example, the mouse is very sensitive; the rat is very insensitive. An important issue in human studies is the difficulty of identifying appropriate comparison groups. The entire general population has been exposed to different levels of PCBs. Some members of the Panel indicated that the immune systems of children and marine mammals who have had high levels of exposure to PCBs and other chemicals may have been compromised. However, in a study of 858 children of exposed mothers in North Carolina who were mostly breastfed and who were followed from birth to one year of age, no harmful effects of PCBs in terms of disease were observed. In fact, there was a decrease in upper respiratory infections related to PCB exposures (Rogan et al. 1987).
In the Yusho population, the incidence of respiratory infections correlated very well with blood PCB level, as well as with reductions in IgM and IgA. One panelist mentioned that, in 1987, sudden major die-offs of seals, porpoises, dolphins, and whales took place across the northern hemisphere. Researchers found high levels of PCBs and other chemicals in their tissues. The animals died from respiratory diseases following viral infections. The viruses identified appeared to be new strains of canine distemper. Preliminary data indicated that the immune systems of the animals were severely compromised; therefore, the animals were more susceptible to infections (Osterhaus and Vedder 1988; Osterhaus 1988; Kennedy 1991). People exposed occupationally to high level of PCBs have not shown any increase in mortality resulting from infections. This Panel concluded that further studies on the offspring of exposed people, and probably animals, are needed to evaluate the impact of PCBs on the immune system.
Dermal effects of PCBs in monkeys are well documented in the literature. Acne, alopecia, erythema, swelling of the eyelids, and hair loss were observed in monkeys exposed to a PCB mixture. The major histopathologic changes in the skin of the treated non-human primates were hyperplasia of the hair follicle epithelium, intrafollicular keratin cysts, inflammatory cell infiltration, and edema (Allen 1975).
Human morbidity assessments of PCB exposures have reported potential dermal effects. These effects have included chloracne, a range of general dermal conditions resulting from Yusho and Yu-Cheng exposures, and malignant melanoma. Chloracne is an acne-like skin disease characterized by follicular hyperkeratosis. It is an occasional human response to certain dioxin-like halogenated hydrocarbons that bind to the cytosolic receptor (AhR), which activates the Ah gene locus (Poland et al. 1979). The first known example of the human health effects of PCB exposure dates from the 1930s when the first cases of chloracne were reported in workers exposed to a variety of compounds, including PCBs and chlorinated naphthalenes (Schwartz and Barlow 1942 ). One of the first outbreaks of chloracne associated primarily with PCB exposure was reported by Meigs, who described in medical terms the development of chloracne in 7 of 14 workers exposed to PCBs from a heat exchanger (Meigs et al. 1954). Thermal degradation of PCBs to PCDFs in the heat exchanger may have accounted for these cases of chloracne.
Several morbidity assessments, which apparently involved both low- and high-chlorinated PCB congener exposures, were not able to document any cases of chloracne. A cohort of 194 workers with a history of prolonged dermal exposure to PCBs and a geometric mean serum PCB level of 431 ppb did not present any cases of chloracne (Lawton et al. 1985). Similar results were reported by Smith and colleagues (1982) and Emmett and colleagues (1988). A community clinical survey including skin examination reported mean serum PCB levels of 17.4 ppb in 89 sludge users, 75.1 ppb in 18 workers with specific occupational exposure to PCB, 33.6 ppb in 19 members of those workers' families, and 24.4 ppb in 22 community residents without unusual exposure to PCB (Baker et al. 1980). Results of skin exams revealed no cases of chloracne. The mean PCB level in the 17 persons with acne was 10.8 ppb; in the 118 without acne, the mean PCB level was 13.1 ppb.
NIOSH conducted serum testing and skin examinations of 60 employees at a PCB storage facility (Bryant et al. 1990). Serum PCB levels ranged from 1 to 23 ppb, with a mean of 6 ppb. None of the 60 participants had skin findings suggestive of chloracne.
Among 81 employees whose serum PCB levels ranged from 2 to 385 ppb, with a mean of 62 ppb (Singal and Roper 1988), 9 employees with dermatological findings suggestive of chloracne had a non-statistically significant higher median serum PCB level (48 ppb) than the 72 employees without skin findings (18 ppb). Four of the nine employees with chloracne-like skin findings had serum PCB levels of less than 20 ppb. This lack of consistency between skin findings and serum PCB levels suggested that either the skin conditions were not chloracne or, if they were due to chloracne, the etiologic cause was not PCB but possibly PCDD or PCDF.
Historically, it has been believed that chloracne occurs only in persons with serum PCB levels greater than 200 ppb (Ouw et al. 1976; NIOSH 1977), except in cohorts where there are exposures to other chlorinated compounds such as chloronaphthylenes or dibenzofurans (Good 1943 ). Based upon TEFs, the human serum level of Aroclor 1254 (ordinary, unoxidized PCBs) necessary to produce chloracne has been estimated at 6,157 ppb (Brown et al. 1991). Chloracne has not been associated with low-level environmental exposures (Shields et al. 1992).
Dermatological findings were prominent in the Yusho and Yu-Cheng episodes and included enlargement and elevation of follicular orifices, acne, and dark-brown pigmentation of the skin, nails, conjunctiva, and oral mucosa (Urabe and Asahi 1985; Fischbein et al. 1982). Approximately 82% to 89% of the Yusho cases reported these dermatological manifestations (Reggiani and Bruppacher 1985). The acne lesions were described as chloracne. In many patients, hypersecretion of the meibomian glands, swelling of the upper eyelid, and irregular lid margins were noted. In a case-control study of Yu-Cheng children who were transplacentally exposed, exposed children had increased rates of hyperpigmentation, eyelid swelling, deformed or dystrophic nails, natal teeth, swollen gums, and acne (Gladen et al. 1990).
With respect to malignant melanoma, a review of medical records for 31 PCB-exposed research and development employees and 41 PCB-exposed refinery plant workers showed a statistically significant increase in the incidence of malignant melanomas, with three observed cases. Contributions from concomitant exposures and other risk factors could not be evaluated (Bahn 1976 ; Bahn et al. 1976; Lawrence 1977). In a survey of 55 transformer workers and 56 non-exposed workers, two exposed workers reported a history of melanoma. No cases were reported in the comparison group (Emmett et al. 1988). This difference was not statistically significant.
A conclusion cannot be reached concerning a relationship between skin effects from PCB exposures and the occurrence of malignant melanomas. In one U.S. plant where an excess of malignant melanomas was observed among workers exposed to PCBs (Sinks et al. 1992), reviews of all medical histories showed no histories of chloracne. Chloracne seldom occurs in workers exposed to PCBs and may result from concomitant exposures to other chemicals (polychlorinated naphthalenes) or contaminants. Panelists mentioned that cases of chloracne have not been observed in occupational cohorts of workers exposed to PCBs at capacitor manufacturing plants that they have studied. Chloracne should not be used as an indicator or marker of PCB exposure.
In experimental animal species, the hepatic effects of PCBs include microsomal induction, increased serum level of liver-associated enzymes indicative of possible hepatocellular damage, liver enlargement, fat deposition, and fibrosis. PCBs also caused increases of hepatic cholesterol levels. PCB-related hepatic enlargement is usually associated with hypertrophy of hepatocytes resulting from an increase in smooth endoplasmic reticulum or increased microsomal enzymatic activity or both. Mild hepatic changes appeared to be reversible (ATSDR 1993).
Results from several human studies of occupational or non-occupational exposures to PCBs have shown a relationship between serum PCB level and the induction of microsomal liver enzymes. Occupational studies have shown statistically significant increases in GGTP (gamma glutamyl transpeptidase) (Kreiss et al. 1981; Smith et al. 1982; Stehr-Green et al. 1986), SGOT (serum glutamic oxaloacetic transaminase) (Fischbein et al. 1979; Smith et al. 1982; Stehr-Green et al 1986 ; B-glucuronidase (Steinberg et al. 1986), and 5-nucleotidase (Steele et al. 1990; Steinberg et al. 1986 ). Although these studies have suggested increased induction of liver enzymes, they did not show any evidence of clinical manifestations of liver damage or disease. Results have shown that PCB exposure-related microsomal enzyme induction subsides after cessation of direct exposure to PCBs (Lawton et al. 1985). Other studies show no association between serum PCB level and liver function parameters. Evaluations of 120 male workers (Chase 1981) and 16 municipal workers (Brandt-Rauf and Niman 1988) found the results of liver function tests to be within normal limits. A recent community-based survey of 1,014 adults residing near PCB-contaminated waste sites (Steele and Richter 1992) indicated no statistically significant association between levels of liver enzymes and serum PCB levels.
Although there may be good correlation between serum PCB levels and hepatic enzyme levels, this clinical enzyme parameter should be viewed as insensitive and non-specific. Changes in these levels can be indicative of a variety of factors (cardiac conditions, alcohol consumption, hepatitis, etc.) other than PCB exposure. When abnormal hepatic enzyme levels are detected, action should be undertaken to determine the source or cause. There is currently no way, however, to attribute an individual's abnormal test results to previous PCB exposure.
Limited data were found regarding the cardiovascular effects of PCBs in animals. Rats exposed to a high level of PCBs (4,000 mg/kg) had no histopathologic changes in their hearts (Bruckner et al. 1973 ). Pericardial edema was reported in monkeys exposed to 12 mg/kg/day of PCBs (Allen et al. 1973). No reliable studies were found that investigated blood pressure changes or other cardiovascular endpoints in animals.
With reference to humans and serum lipid levels, a 3-year followup of 82 Yusho patients demonstrated that serum triglyceride levels remained elevated throughout the entire period of observation (Uzawa et al. 1973). In a study of 148 persons with occupational or non-occupational exposures to PCBs, plasma triglyceride levels for both alcohol drinkers and non-drinkers increased significantly with serum PCB concentrations (Baker et al. 1980). Increased levels of plasma triglycerides have been noted in other studies of groups exposed to PCBs (Okumura et al. 1974; Nagai et al. 1969; Uzawa et al. 1973). Hypertriglyceridemia has been reported at serum PCB levels of 50 to 200 ppb (NIOSH 1977). The authors suggested that PCBs may alter lipid metabolism at levels of exposure and bioaccumulation insufficient to produce overt symptoms.
Although it has been hypothesized that PCBs cause abnormal lipid metabolism, recent reports suggest any correlation simply reflects the affinity of PCBs for serum lipids (Brown and Lawton 1984 ; Lawton et al. 1985; Guo et al. 1987). The major health effect that would be correlated with abnormal lipid levels is cardiovascular disease. However, in occupational studies of workers exposed to PCBs, no significant excesses of cardiovascular morbidity (Gustavsson et al. 1986) or mortality (Brown 1987; Bertazzi et al. 1987; Sinks et al. 1992 ); have been observed.
With reference to hypertension, a community-based clinical survey conducted in Triana, Alabama, found a mean serum PCB level of 17.2 ppb, attributable to the consumption of fish (Kreiss et al. 1981). A correlation was found between serum PCB concentration and blood pressure (especially diastolic), GGTP concentration, and serum cholesterol level. PCB serum levels were found to be statistically correlated only with diastolic blood pressure after controlling for confounders, including body mass, age, and sex. The effect of PCBs on systolic blood pressure was small and of borderline statistical significance. These findings were based upon only one blood pressure measurement per participant.
A clinical study of 106 individuals showed a statistically significant dose-response relationship between serum PCB level and the presence of self-reported high blood pressure (Stehr-Green et al. 1986 ). This association was statistically significant when the effects of age and smoking were controlled for. No blood pressure measurements were made; rather, participants were asked whether they had ever been diagnosed or told by a physician that they had high blood pressure.
In a study of 59 Yusho patients conducted 13 years after exposure, no association was found between serum PCB level and either systolic or diastolic blood pressure (Akagi and Okumura 1985 ). Of these patients, 52.5% still had PCB levels higher than the range found in the general population. The frequency of hypertension in these patients was 16.9%, a value similar to that expected for a general population of the same age and sex composition. For these patients, factors shown to be significantly associated with elevated blood pressure included age, obesity and alcohol consumption, and all known risk factors for hypertension.
In a study of 60 PCB-exposed workers, the association between blood pressure and PCB exposure was evaluated (Steele et al. 1990). Any apparent association between diastolic and systolic blood pressure and serum PCB concentrations was attributable to age.
A community-based survey of 1,014 adults residing near PCB-contaminated waste sites (Steele and Richter 1992 ) showed no association between high-density or low-density cholesterol levels and serum PCB level. With univariate analysis, there were statistically significant trends between increasing systolic and diastolic blood pressures and serum PCB level. However, after multivariate adjustment for age and sex, PCB exposure was not associated with either systolic or diastolic blood pressure.
The major health effect resulting from an association between PCB exposure and hypertension would be observed excesses of cardiovascular or cerebrovascular disease. Again, no significant excesses of cardiovascular morbidity (Gustavsson et al. 1986) or mortality (Brown 1987; Bertazzi et al. 1987; Sinks et al. 1992) have been observed in occupational studies of workers exposed to PCBs.
The association between PCB exposure and total lipid levels, once believed to be causal, is now viewed as artifactual. People with elevated blood lipid levels tend to have elevated PCB levels. A limitation of many studies and lab reports is that the concentration of PCBs in blood is reported in terms of total serum. Results should be standardized for extractable lipid weight as is done with fat biopsies. Also, when one reports effects by specific lipid fraction (for example, HDL, LDL, etc.), the effects of PCBs can be more accurately assessed.
Functional changes were observed in thyroid glands of rats after exposure to PCBs. Serum thyroxine (T4) levels were reduced in rats exposed to PCBs; however, no treatment-related changes in serum triiodothyronine (T3) occurred. Changes in the thyroid gland included thyroid enlargement, reduced follicular size, follicular cell hyperplasia, and accumulation of colloid droplets. The appearance of these changes and their severity depended on the dose and duration of the treatment (Collins and Capen 1980a, 1980b, 1980c; Byrne et al. 1987). Chronic exposure (5 months) of rats to low levels (0.09 mg/kg/day) of Aroclor 1254 resulted in a decrease in thyroid serum T3 and T4 hormones.
The endocrine system affects all the other systems that have been discussed. Some types of PCBs have estrogen-like mechanisms of action, which may potentially affect these systems. Altered hormone balances could be playing a role in some of the health outcomes observed in animal and human studies.
VIII. Health Effects from Remediation Technologies
Both the Incineration Discussion Panel and the Alternative Technologies Panel elucidated many factors that could influence the technology under discussion and affect its ability or efficiency to destroy PCBs. Some of those factors are the volume of contaminated material to be disposed, the physical and chemical state of the material (e.g., sewage, sludge, solids, liquids, inert materials, reactive materials, etc.), proximity of the contaminated material to the public, and potentially, the transportation of the material to the site of remediation, if required. Each of these factors could influence the types and concentrations of the contaminants released.
A majority of the technologies discussed require the excavation and sizing of the contaminated media. During these activities there is a potential for fugitive emissions of PCBs, metals, polycyclic aromatic hydrocarbons (PAHs), volatile organic compounds (VOCs), semi-VOCs, and other site-specific contaminants of concern.
Many of the alternative technologies discussed, such as bioremediation, soil washing, solvent extraction, thermal desorption, solidification, and landfilling, do not destroy the contaminants, making it necessary to further treat or dispose of the concentrated end products, which because of their concentrated state may pose a greater risk.
Inhalation of emissions from the stack of an incinerator is usually a minor contributor to the overall risk to the community. The greatest risk to the community can be an increase in the background concentrations of these chemicals already in the food chain. Of public health concern is the fact that some of the alleged reproductive and developmental effects have been seen at what are generally considered to be background concentrations (see respective sections on reproductive and developmental effects).
In regards to the PCB exposure pathways identified by the Incineration and Alternative Technologies panels and the potential for adverse health outcomes resulting from these exposures, several factors must be considered. First, the exposure pathways identified by these panels have previously been studied either through occupational or non-occupational studies. Although the source of the PCBs or the site of the activity may be different, the exposure pathway is not. Second, the PCB exposure doses associated with published occupational investigations are greater than the doses estimated to occur from incinerator emissions or other alternative destruction technologies. Health effects identified during occupational exposure studies could be used to assess any potential impact on public health resulting from incineration and other alternative destruction technologies.
Because PCB-contaminated materials tend to be a heterogenous mixture of PCBs, metals, and other industrial chemicals, the potential for inadvertent releases of these chemicals exists regardless of the technology used. Fugitive emissions contaminated with other chlorinated hydrocarbons, metals, VOCs, semi-VOCs, or PAHs are also possible. A variety of health outcomes, from no effects to specific disease, have been associated with varying exposure doses of these chemicals. Discussion of exposure dosages and the health effects seen for each of these contaminants is beyond the scope of this report. The members of this Panel suggest that the reader consult the appropriate ATSDR Toxicological Profile or the available literature for the chemical of concern.