Minor effects of low exposure to inorganic mercury on the human immune system.

Minor effects of low exposure to inorganic mercuryon thehuman immunesystem . Scand J Work Environ Health 1993;19:405-13. The influence of exposure to inorganic mercury on the immune system was examined in 36 workers occupationally exposed to mercury vapor, 14 individuals with skin hypersensitivity to mercury compounds, 21 sub jects with health disturbances allegedly caused by dental amalgam fillings ("amalgam disease"), and 39 healthy referents. Concentrations of mercury in blood and urine and some parameters judged to mirror different effects on the immune system were determined. The latter included, white blood cell differential counts, serum immunoglobulins and autoantibodies, and in vitro production of the cytokines interleukin I (IL-I) and tumor necrosis factor alpha (TNFa). Virtually all of the immunologic param eters were within normal ranges and did not differ significantly between the groups. In the group sen sitized to mercury, there was a reduction of the in vitro productionof both TNFa and IL-l compared with the reference group's values. No significant correlations were noted between different mercury exposure estimates and the immunologic parameters.

antibodies (7). These autoantibodies and circulating immune complexes are involved in the glomerular IgG depo sits fo und in the mesangium and the vessel wall s, espec ially of mice ca rry ing the H-2' haplot ype (8,9).
Data on human immune reaction s to inorgani c mer cur y e xposure are sca rce . Skin hypersen siti vit y (Coo mbs type IV) to metallic mercury is rare (10 ), but some mercury salts, such as thiomersal and ammon iated mercury, hav e been found to be more common sensitizers (I I) . Cases of membranous glomerulon ephritis have been de scribed among worke rs exposed to inorga nic mercury (12), and nephritis accompa nied by the nephroti c syndro me has been observed in subjects using skin-lightening crea ms and dru gs containing inorganic mercury (1 3,14). However, the degree of exposure has not been cle arly verified in these cases, and no dose-effect relation ships have been establi shed.
In Swed en and some other countries, there has been debate regarding the possible health effe ct s of expos ure to inorganic mercury rele ased from dental amalga m fillin gs (15 ). The relati vely low qu antity of mercury absorbed from this source (16) has been propo sed to indu ce an immune response in sus ceptible indi viduals and hence cause variou s sympto ms (ie, "amalgam di sease").
Th e main obje cti ve of this investi gation was to exa mine possible effects of low exposure to inorganic mercury on the human immune sys tem. The following three targ et groups were studied : (i) workers occupat ionally ex posed to mercury vapor, (ii) ind ividual s with skin hypersensitivity to inorganic mercury, and (iii) subjects with health disturbances attri-buted to mercury release fro m their dental amalgam fillings. Clinica l data were collec ted together with infor mation on a selection of parameters conside red to mirror different effec ts on the immune system. In addition, the plasma sele nium conce ntration and the erythrocy te activity of catalase and glutat hione peroxidase were determin ed. Deta ils regard ing the latter para meters have been described in a separate report (17) .

Subjects
The study population was composed of four groups. The first comprised 36 workers occ upationally exposed to mercury in a ch lora lka li plant (14 male workers randomly selected from a gro up of about 25 daytime workers expo sed to mercury), from a fluoresce nt light bulb factory (all 7 workers exposed to mercury at the plant: 5 men and 2 women), and from the dental service (N = 15,6 dentists and 9 dental nurses, all women, randomly selected at four dentistries) . The average duration of exposure (ie, number of work-years) was 20 (range 1--45) years.
The second group comprised 14 subjects ( 12 women and 2 men) with skin hypersensitivity to inorganic mercury. Thi s "allergy gro up" consisted of individuals who had developed oral mucosal lesions (lichen orale) in conj unction with ama lgam fillings and who had show n a positive reactio n to mercury (metallic mercury, mercuric chloride or phenylmercury) in a patch test. All of these subjects were tested at the same dermatological department dur ing a period of abo ut 36 months. At the time of our study 12 subjects had had all of the ir amalgam fillings removed beca use of mercury hypersensitivity.
The third gro up co mprised 2 1 subjec ts ( 16 women and 5 men) with health distu rbances alleged to mercury release from their dental amalgam fillings ("a malgam group"). All of these subjec ts were referred to the department of occupational medicine by dentists for investigation of possible chronic mercury intoxication. Special selection criteria for thi s group were (i) chronic inconvenience for at least one year from three or more of the followi ng symptoms: muscle or joi nt pain, abnormal tiredness, anxiety, gas- trointestinal dist urbances, oral symptoms (sore mouth, metallic taste , etc , and tooth or jaw symptoms), memory impair ment, loss of concentration, dizz iness or vertigo, headache, or sleep disturbance ; (ii) sympto ms alleged to be caused by thei r OWn ama lgam fillings; (iii) no occ upationa l expos ure to heavy metals or organic solvents; (iv) no overt mental illness or alco holism; (v) primary exam inat ion having been made by a physician and dentist.
The fourt h group was a reference group (30 women and 9 men ) consis ting of hospital perso nnel and office employees, not occ upationally exposed to heavy metals or organic solvents.
Common selectio n criteria for all four groups were age betwee n 20 and 70 years, no clinically appare nt, untreated medical or odon tological disease, no use of drugs known to interfere with the immune system, no acute infec tion at the time of the examination. There were no dropouts (ie, all of the subjec ts who were asked to participate joined the study). Details concerni ng age and some background data of the participa nts are presented in table I.

Methods
Data on the subjects ' smoking habits, alcohol intake, fish consumption, work conditions, health history, and curre nt medication were registered on a que stio nnaire. For 103 of the 110 individual s, the co llected answers were checke d in an interview with a physic ian (SL). In connect ion with the interview (or collectio n of the questionnaire) venous blood samples were drawn for analyses of immu nologic parameters and mercury. In addition, dental ama lgam burden (estimated as 0-5 ama lgam surfaces per each tooth with amalgam filli ngs) was judged by a physici an.
Samples for analyses of lymphocyte deoxyri bonucleic acid (DNA) synthesis and cytokine prod uction, and for the prevalence of interleukin 2 (IL-2) receptors, were collected in sterile, heparinized, metal-free Vacutainer" tubes (Becton Dickinson, Meylan Cedex, France) . Samples for the determination of serum imm unog lobulins and autoantibodies were co llected in metal-free Vacutainer" tubes, whic h were centrifuged to sep arate the blood cells . Samples for white blood cell counts were drawn into Vacutainer" tubes prepared with ethylenediaminetetraacetic acid (EDTA). Blood samples for lymph ocyte transformation tests (LTT) (performed for 18 subjects) were drawn into metal-free, sterile Vacutainer" tubes with polystyrene granules. These tubes were gently shaken for 1D min to prevent coagulation. Sampl es for the whole blood and plasm a mercur y analyses were collected in met al-free, heparini zed Vac uta iner" tubes. Mornin g urin e samples were collected at home by each subje ct in 25D-ml acidwas hed polyethylene bottles which were delivered to the Department of Occupational Medicine within 4-5 h. All of the mercury samples were then frozen at -20°C, and stored for about four months until the analyses .

An alyses of the immunologic param eters
All of the immunologic analy ses except the LTT were performed in the Department of Clinical Immunology of the Huddinge University Hospital. For a more-d etailed description of the methods used, see reference 18.
Whit e blood cell counts and urin ary crea tinine were determined at the Department of Clinical Chemi stry of the same hospital with the use of standard meth ods. In eve ry series of immunologic analyses samples from both "target" subjec ts and referents were included, but the analyses were perform ed without knowledge of the group to which the individuals belonged.
Serum conce ntrations of IgA, IgG, and IgM were determined by nephelometry (Bec kman Array Protein System ), and the IgE conce ntra tion in serum with the Phast sys tem (Pharmacia, Sweden).
Serum autoa ntibodies to various tissue antigens were determined by indirect immunofluorescence using tiss ue sec tions of rat kidn ey and stomach as the antige n. Anti-DNA antibodies were determined in all sera positive for antinucl ear antigens, critidae lucilia e being used as the antigen. Antib odie s to the glomerular basement membrane were determined by the " anti-Goodpasture" enzyme immunoassay (Biocarb Diagnostics AB, Lund, Sweden). Antibodies again st thyroid microsomal antigen and thyroglobulin were analyzed with hemagglutination tests from Wellcome labor atories (Dartford, England ).
The purificati on of mononucl ear cells started within 5 h of the blood sampling. Mon onuclear cells were isola ted by Iymph oprep centrifugation (Nyco med Pharma, Oslo, Norway). In order to study cytokine production in vitro, the mononuclear cells were cultured overnight in conditioned medi a in the absence and presence of lipopolysacarid e.
Spont aneous and ConA-indu ced DNA synthesis (ConA = convanvalin A) were determined as described by Soop et al (19). The cy tokines interleukin 6 (IL-6), interleukin I (IL-I ), and tumor necrosis factor alpha (TNFa) were analyzed in serum (IL-6) and co nditio ned media (IL-I and TNFa ) with commer- For the determination of IL-2 receptors, lymphocytes attached to glass cove r slips were reacted with monoclonal antibodie s directed against CD3, CD4, CD 8, CDI 4, and CD25 (CD = cluster of differenti ation). Th e re activity of the monoclonal antibodies was subsequently detected with a biotin-avidin peroxidase system (Vector Laboratorie s, Burlingame, California, United State s).
The in vitro res po nsive ness of periph eral blood lymphocytes to mercuric chlor ide and phenylmercury acetate was determ ined in 10 subjects occ upatio nally expose d to mercury, in nine subjects with skin hypersensitivity to mercury, and in 10 subjects belonging to the ama lgam group. Th ese persons were randomly selected. The analyses (LTT) were performed at the Laboratory of Immunotoxicology, Astra AB (Sodertalje, Sweden) with the meth od described by Stej skal et al (20).

Mercu ry expos ure estimates
The immunol ogic parameters were related to the following five expos ure indicators: (i) current conce ntration of mercury in blood , plasma and urine (adju sted for crea tinine excretion measured with Jaffe' s colorometric method); (ii) exposure duration (number of work-years with mercury exposure) ; (iii) exposure intensity (low or high level of mercury exposur e, determ ined acco rdi ng to the empl oyee' s work type and their expected mercury expo sure, as judged by one of the researchers); (iv) fish co nsumption (number of fish meals per week and type of fish); (v) amalgam burd en (estimated as 0-5 amalgam surfaces per tooth with amalgam fillings). In the three groups occ upatio nally unexposed to mercur y (refe rence group, allergy group, amalg am group), the immunologic parameters were related to the first, fourth and fifth exposure indicators. Fish consumption was includ ed as an indicator of background exposure to methylmercur y.
The concentration of total mercury in whole blood, plasma, and urine was analyzed in the laboratory of the Divi sion of Medical Chemistry at the Swe dish Nation al Institute of Occupational Health . A modified version of cold-vapor atom ic absorption spectroph otom etry was used . For details concern ing the methods and quality control of the mercur y analyses see referen ce 2 I . The mercury concentrations in whole blood (B-Hg), plasma (P-Hg), and urine (U-Hg) of the four study groups are presented in table 2. The highest mercury levels were seen among the chloralkali workers (median B-Hg 35 nmol . I-I, median U-Hg 7.22 nmol . mmol creatinine"), whereas the dent al personnel and the workers in the light bulb factory had only slightly elevated mercury levels (median B-Hg 16 nmol . I-I, medi an U-Hg 1.86 nmol . mmol creatinine 0 1 ) when they were compared with the referents.
Scand J Work Environ Health 1993, vo119, no 6 Table 2. Current concentration of mercury in whole blood (B-Hg), plasma (P-Hg), and urine (U-Hg) in the four study groups.  Table 3. Total white blood cell counts (in randomly selected subgroups) and differential counts in the four study groups. b Two subjects excluded due to bad smears.

Statistical methods
Comparisons of the examined parameters between the three target groups and the reference group were made with the two-sided Student's t-test or the Mann-Whitney rank sum test (for skewed parameters). Dose-effect relationships were studied with Pearson's correlation coefficient or with Spearman's rank correlation coefficient (for skewed parameters) and by multiple regression. The influence of age, gender, smoking, alcohol intake, and sampling occasion on the immunologic parameters was examined in an analysis of variance (ANOV A).

White blood cell counts
The total white blood cell counts did not differ significantly between the groups. Nor did the percentage of white cells in the differential counts vary between the groups (table 3). In the total population (N = 110), there was a significant correlation (r = 0.32, P = 0.001) between the percentage of eosinophils and the IgE levels in serum. No other nota-408 bIe relationships were seen between the white blood cell counts and the different immune parameters or the mercury exposure estimates.

Immunoglobulins
The serum concentrations of IgA, IgG, IgM, and IgE were normal in all four groups, although the subjects in the allergy group had somewhat higher IgE values (table 4). In the total population, the smokers exhibited slightly higher IgE levels than the nonsmokers (P = 0.063). There were no significant correlations between the immunoglobulin levels and the mercury exposure estimates.

Autoantibodies
The number of sera containing autoantibodies against the different tissue antigens was low in all four groups (table 5). The relative frequency of sera with detectable autoantibodies to one or more of the different antigenic specificities tested was 4.8 % (17 of 357) in the reference group, 5.4 % (18 of 332) in the exposed group, 3.9 % (5 of 129) in the allergy group, and 3. Fisher's test). The workers exposed to mercury did not show elevated titers of autoantibodies against antigens (glomeruli or glomerular basement membrane) associated with mercury-induced glomerulonephritis in animal models (2). None of the subjects with elevated antinuclear antigen titers displayed serum autoantibodies to DNA.

Deoxyribonucleic acid synthesis
The spontaneous JH-thymidine uptake (DNA synthesis) in the four grou ps invest igated is show n in figure I. There were no stati stically significant group differences. In the amalgam group, the DNA synthesis was extremely high in a single individual (27 363 counts · min' ). In the reference group, spontaneous DNA synthesis was statistically significantly correlated to age (r =-0.33, P =0.015) and to the mer-

Cytokines
On ly a few indi viduals showed detect able serum levels of IL-6. The median value was zero in all four gro ups, a nd the ran ges wer e 0-30 pg . ml I in the reference group, 0-30 pg. ml' in the expose d gro up, 0-0 pg . ml I in the alle rgy gro up, and 0-25 pg . ml -' in the amalgam grou p. Serum samples fro m patient s with sep sis (N = 6), analyzed simultaneou sly at the same laborat ory, showe d co nside rably higher co nce ntra tio ns (> 1000 pg . ml') .
The in vitro production of IL-l and TNFa was within normal ran ge s (compared with those of reference per son s and patients examined earlier at the same labo ratory ) in vir tually all of the subjec ts, but the level of both TN Fa and IL-I was notabl y lower (P = 0 .0 I) in the allergy group than in the referents (f igure 3). Within the gro up occupa tio nally exposed to mercur y, there wer e wea k, but statis tica lly sig nificant tend encies towards negative correlations between the mercury concentrations in blood and urin e and the in vitro produ ction of both TNF a and IL-I. However, these tend encies faded whe n the sampling

Lymphocyte surface markers
Th e mononuclear cell s fro m subjects in all fou r groups showed negligible expression of IL-2 receptor s. Neither the number of cells with class II antigens nor the CD4 : CDS ratio diffe red particularly between the gro ups. Further st udie s on various lymphocyte surface ma rkers arc in progress, and the results will be rep orted separately.

Lymph ocyte transformati on tests
The LTT was negative in all of the tested subj ects occ upa tio nally expos ed to mercury (N = 10). In the allergy gro up five subjects (out of the 9 tested ) showed a positive reaction to mercuric chloride. Two of these five persons also had a positive reaction to phenylmercury acetate. In the amalgam group, one subject (out of the 10 tested) showed a positive reaction to both mercury compounds.

Selenium
The plasma selenium concentrations were within normal ranges, and there were no significant correlations between the selenium levels and the immunologic parameters or mercury levels in blood or urine. Details regarding the analyses of plasma selenium and the erythrocyte activity of glutathione peroxidase and catalase have been reported separately (17).
Analyses of the total population (N = 110) by means of multiple regression and ANOV A did not show any significant correlations between the different immune parameters and the mercury exposure estimates. The amalgam burden was not related to the immunologic parameters. Nor did fish intake, alcohol consumption, or gender influence these parameters.

Discussion
In the present study, the hypothesis that persons exposed to low levels of inorganic mercury exhibit immunologic manifestations has been evaluated with an extensive test battery. Thus the tests applied should be expected to detect disturbances of different functional aspects of the immune system, including activation, immunodeficiency, autoimmunity, and allergy. Some of the tests used, especially the determinations of DNA synthesis and in vitro cytokine production, may be susceptible to the transportation and preparation of blood samples. Therefore, the sampling procedure and the preparation of blood samples were standardized as far as possible. Both the "target" subjects and the referents were included in every sampling occasion, and the sampling occasion was included in the statistical analyses (ANOV A). The results from one sampling episode showed a strong deviation in the ConA-induced DNA synthesis; therefore these values were all excluded. A tendency towards decreased in vitro production of TNFa and IL-I was noticed among both the mercury-exposed subjects and the referents on one sampling occasion, which made the negative correlations between the mercury levels in blood and urine and the cytokine production observed in the mercury-exposed group less convincing.
The mercury-exposed workers were selected to include both persons with high and persons with low exposure. According to the observed mercury levels in blood and urine, the degree of current exposure in this group was rather low. However, the duration Scand J Work Environ Health 1993, vol 19, no 6 of exposure had been long (average 20 years), and the exposure levels had been higher earlier, especially in the chloralkali factory. The cross-sectional study design may also have included a "healthy worker" effect (ie, subjects with health problems related to their occupational exposure may have quit). According to the companies' health care units, however, there had been very low worker turnover in the different plants.
The selection of subjects into the amalgam group was a difficult task, as dental amalgam is not generally accepted as an etiologic factor for disease. Some individuals with medically unexplained, widespread symptoms have claimed an exceptional susceptibility to mercury released from their own amalgam fillings and insisted on further investigation of possible mercury intoxication.
The criteria for our amalgam group selected persons with many and severe symptoms. Some of these symptoms are described as the chronic fatigue syndrome. The subjects in our amalgam group did not, however, report fever, lymph-node pain, or pharyngitis, which are primary criteria for the chronic fatigue syndrome. All of the subjects had earlier been examined by physicians and dentists, and therefore evident medical or odontological causes of their illness were excluded. This selection procedure probably excluded persons with rheumatologic disorders and collagenosis and may explain the relatively low (compared with the referents) prevalence of serum autoantibodies to some of the antigens tested. The prevalence of antithyroid antibodies was, however, slightly elevated in comparisons with the referents. The data on health history revealed that two of the subjects with antithyroid antibodies in serum had earlier undergone medical investigations due to goiter. This finding suggested that possibly some persons who attribute their illness to dental amalgam fillings may suffer (or have suffered) from thyroid gland disease.
Other than the decreased in vitro production of TNFa and IL-I observed in the allergy group, our results revealed no marked differences between the three target groups and the reference group regarding white blood cell counts, immunoglobulins, autoantibodies and IL-6 in serum, spontaneous and ConA-induced DNA synthesis, and the expression of lymphocyte activation markers. In addition, the data argue against any specific sensitization against mercury in the group occupationally exposed to mercury or the amalgam group, although more information is required to establish this as a fact.
A suspected effect of dental amalgam and nickel alloys on T lymphocytes, has earlier been reported by Eggleston (22), and elevated levels of serum proteins in mercury-exposed workers was reported by Bencko et al (23). An increased prevalence of antilaminin antibodies among mercury-exposed workers was reported by Lauwerys and his co-workers (24), but it was not confirmed in a later study by the same Scand J Work Environ Health 1993,vol 19, no 6 group of researchers (25). Langworth et al (26) found no differences in the serum immunoglobulin concentration or the prevalence of serum autoantibodies (including antiglomerular basement membrane and antilaminin antibodies) between chloralkali workers exposed to mercury vapor and unexposed referents.
Spontaneous DNA synthesis in lymphocytes has been reported to be elevated in patients with connective tissue disorders (27,28), allergy (29), infections (19,30), and malignant tumors (31). IL-6 in serum also exhibits elevation in acute and chronic inflammation and in septic shock and injury (32,33). The expression of lymphocyte activation markers such as IL-2 receptors and human leucocyte antigen-DR has been applied so that clinical conditions associated with immune activation, for example, in chronic fatigue syndrome, could be evaluated (34,35).
According to our results it seems unlikely that persons with low-level occupational exposure to mercury or with health disturbances attributed to amalgam fillings display any easily detectable immunologic disturbance due to their mercury exposure. However, one individual in the amalgam group exhibitedantinuclear antigen antibodies in serum and a markedly augmented spontaneous DNA synthesis. An extended clinical examination (including laboratory tests) did not reveal any rheumatologic disease, collagenosis, or malignancy in this individual.
The lower in vitro production of both TN Fa and IL-I noted for the allergy group suggests that a suppression of cytokine production may be associated with mercury hypersensitivity. This finding is interesting, as impaired mitogen-induced cytokine production (lL-2) has been described for BN rats exposed to mercuric chloride (36), and it needs further investigation. Furthermore, the DNA response in this group was lower than in the referents' during stimulation with 10 ug of ConA, and this difference may point to a possible impairment in T-cell function. The allergy group, however, differed somewhat in the background parameters (age, alcohol consumption, smoking, amalgam surfaces) when compared with the reference group (table I). Therefore the findings must be interpreted with caution.
In the reference group, spontaneous DNA synthesis, as well as TNFa and IL-l production, negatively correlated with age (P = 0.015, P = 0'()04, and P =0.001, respectively). The noted relations indicate a decline in immune activity with increasing age, which may be reasonable. Such relationships were, however, not seen in the other three study groups and is not, to our knowledge, supported by other reports.
In the total population (N = 110), none of the immune parameters correlated with the mercury exposure indicators. Thus amalgam burden was not related to any sign of immunologic disturbance. Smoking has been shown to influence serum immunoglobulin levels (37) and also the activity of natural killer cells (38). A negative correlation between the in vitro 412 production of TNFa and IL-I and smoking, as seen in our study, has not been reported earlier.
Taken together, the results argue against a doserelated effect of inorganic mercury on the human immune system. The absence of immunologic disturbances among the mercury-exposed workers implies that the immune system of the majority of individuals is insensitive to low-level exposure to inorganic mercury or tolerates it. Nevertheless, not all individuals are insensitive to low-dose mercury exposure. Dose-response and dose-effect estimations in immunotoxicology are complicated (39), and it is well known from animal models that genetically restricted factors predispose a person to the development of mercury-induced autoimmune phenomena (2). Genetic factors also seem to predispose a person to autoimmune diseases induced by drugs or occupational chemicals (40).
It must be considered that the relatively small size of the study groups limited the general significance of our results. Thus it is hard to rule out small effects of mercury exposure on the human immune system. It must also be emphasized that many of the immune reactions seen in rodents are transient and disappear after a few weeks, despite persisting mercury exposure. It has not been studied whether such temporary immune reactions can occur in workers at the onset of their exposure to mercury.
Still, as there may exist certain individuals with a "high" sensitivity to low doses of inorganic mercury, further studies on the possible immunologic effects of this metal are important. Studies of immune effects at higher exposure levels and on larger populations are required for a better evaluation of the potential dose-response relationships and for more extensive conclusions to be drawn.