In 2003 we recruited 28 volunteers, 23 women and 5 men, 28–60 years of age (mean 48 years) for measurement of Hg biomarkers. Sampling comprised venous blood from the cubital vein (5 mL, Venoject II, EDTA(K2), VP-050SDK), red blood cells (RBC) and plasma (5 mL, Venoject II, EDTA(K2), VP-050SDK; Terumo Corp., Leuven; Belgium), hair (a hair sample was tied with a cotton thread, cut close to the scalp from the back of the head and put into a plastic bag), and urine (a spot sample collected in acid washed plastic containers). Information regarding fish intake (usual number of meals/month) and number of dental amalgam fillings was collected via self-reported questionnaires. A usual number of 0–22 fish meals/months and a total number of dental amalgam fillings between 0–15 were reported. For evaluation of Hg distribution in hair, and blood to hair ratio, we also used data previously collected in a study of women with a high fish intake (N = 145, 20–50 years of age; 31). The study was approved by the Ethics Committee of the Karolinska Institutet, Stockholm.

Whole blood, RBC, plasma and urine samples (1.0 mL) were treated with 1.0 mL L-cysteine (0.012 M), 1.5 mL NaOH (11 M) and 0.5 mL deionised water, and stored in the dark over night at room temperature to complete the solubilisation. Hair samples (3 cm from the scalp end; approximately 20 mg) were treated with 2.0 mL L-cysteine (0.083 M), 4.0 mL NaOH (11 M) and 14 mL NaCl (0.17 M). The mixture was heated to 90–95°C for 20 minutes to complete the solubilisation.

Total mercury (THg) and inorganic Hg (IHg) were analyzed in whole blood, RBC, plasma, hair, and urine using cold vapour atomic fluorescence spectrophotometry (CVAFS, Merlin, PSA 10.023; P.S. Analytical Ltd., Orpington, Kent, UK), following reduction to Hg⁰ in a reaction tower, using an automatic multiple-injection analysis (MIA) system, with a Tefzel^® 13-channel selector valve (Analys Modul Sweden AB; Figure 1). In this system, a motor-driven pump (Microlab 900, Hamilton Bonaduz AG, Switzerland) is connected to the central port of the selector valve, which opens to one of 13 peripheral ports at a time. The pump dispenses volumes with low variation which enables low volumes of chemicals to be used. Deionised water is used as an extended syringe piston and washing medium. No chemicals or samples reach the syringe at the base of the tube loop (Figure 1). The sampler, the reaction tower and the reagent bottles are connected to the peripheral ports by Tefzel^® tubes. The sampler and the selector, with an AD converter, are controlled by a PC using a control program (EASYLAB, Analys Modul Sweden AB). The program executes the commands from a command list for THg or IHg in a sequential or logical order. In order to reduce blank values, all reagents are initially mixed in the reaction tower to eliminate any Hg impurities. Any Hg⁰ formed in this initial cleaning step is transported with argon (Ar) gas through the detector.

For determination of IHg, 800 μl L-cysteine solution (0.1% w/v L-cysteine in 1.5% w/v NaCl), 200 μl 8 M H₂SO₄ (p.a.) with 0.4% antifoaming agent (Antifoam 204; Sigma Chemical Co., S:t Louis, MO, USA; soluble in acid but not in alkaline solution), 2000 μl 11.25 M NaOH, 100 μl deionised water, and 100 μl 10% w/v SnCl₂ in 2.4 M H₂SO₄ were delivered to the reaction tower for elimination of Hg impurities, after which 500 μl sample solution and 200 μl cysteine solution, followed by 1200 μl of deionised water (for rinsing) were added. For determination of THg, 800 μg L-cysteine solution, 600 μl 8 M H₂SO₄ with 0.4% antifoaming agent, 2000 μl 11.25 M NaOH, 200 μl cysteine solution, 1000 μl deionised water, and 100 μl of a mixture of 10% CdCl₂ and 50% SnCl₂ in 8 M H₂SO₄ were delivered to the reaction tower for elimination of Hg impurities. Then 500 μl 8 M H₂SO₄ (supra pure) was added (in order to increase the temperature), followed by 500 μl sample solution and 500 μl deionised water.

The Hg⁰ released from the sample was transported by Ar gas (0.087 L/min) through a moisture trap, chilled with ice, followed by a tubular permeable membrane (Perma Pure mini-dryers, model MD-125-12S, Perma Pure Products, Inc, Farmingdale, USA) before reaching the AFS detector. The signal was stored by the PC and also recorded on paper for process control (Perkin Elmer Model 56 recorder). The area under the curve was integrated by the computer and used for evaluation of the amount (ng) Hg in the sample. The shield gas flow (Ar) for the detector was 0.099 L Ar/min. The standard solutions (0.1–2.0 ng Hg/mL) were made in 0.1% L-cysteine. Furthermore, a MeHg standard of 0.4 ng/mL was included in the standard curve in order to control the degree of demethylation in the IHg analysis and for recovery in the THg analysis. The time interval between reactions was 15 minutes. All samples were analyzed in duplicates.

The concentrations of the organic mercury fraction (OHg) were calculated by the subtraction of the IHg concentrations from that of the THg concentrations. OHg is assumed to be mainly MeHg as the only other known exposure sources of organic mercury compounds in Sweden are a few vaccines containing thiomersal (ethylmercurithiosalicylate), a seldom-used preservative containing ethylmercury. Hg concentrations in urine were adjusted to specific gravity (1.019 μg/mL; urine specific gravity refractometer, Uricon-Ne, Atago Co., Ltd., Tokyo, Japan) and creatinine (analyzed at the Department of Clinical chemistry, Karolinska University Hospital, Stockholm). Haematocrit and haemoglobin (Hb) concentrations were measured (Department of Clinical chemistry, Karolinska University Hospital, Stockholm).

For evaluation of potential demethylation during solubilisation, a purified (> 99%) radiolabeled methylmercury-chloride (²⁰³Hg, Amersham Laboratories, Amersham, UK) solution was solubilised according to the method described above, at room temperature and at 88°C for one hour. The total Hg concentration was measured by gamma-counting (Searle 1195, Searle Analytical Inc.). The loss of total Hg during solubilisation was < 1%, i.e. not detectable. An aliquot (10 g) of each of the solubilised solutions was acidified by 3 mL 6 M HCl, stored over night at +4°C, extracted with chloroform three times (30+20+10 mL) to separate MeHg and IHg 32. The water phase, containing the IHg from MeHg demethylation, was then measured by gamma-counting to calculate the degree of demethylation during solubilisation.

In order to quantify the degree of demethylation in the reaction tower during the analytical step, samples with and without addition of IHg and MeHg were solubilised and then acidified and extracted with chloroform, following the procedure described above. THg and IHg were then measured by the analytical method described above, and IHg from demethylation of MeHg was calculated as the percentage of the initial amounts.

The blood sampling material was tested for Hg contamination. Simulated blood sampling using a weak acid (0.03 M HNO₃) was performed. The acid solutions were analyzed for Hg content by the method described above. The material was found to be essentially free from Hg contamination (all acid solutions were below the limit of detection, LOD, i.e. 3 standard deviations of the mean of the chemical blanks). All other materials used for analysis were acid washed. Appropriate reference materials for Hg in blood, serum, urine and hair were analyzed in each analytical run, respectively (see Results and Table 1).

We used spearman correlation (r_s) test to test for associations between parameters, and linear regression analysis for evaluation of association between parameters when the requirements for normally distributed residuals were met. Statistical analyses were conducted using SigmaStat^® (Version 2.03 for Windows (Systat Software GmbH, Erkrath, Germany). Statistical significance was set to p < 0.05.

The accuracy of the Hg speciation method, as evaluated by repeated analyses of reference materials, was satisfactory (Table 1). There are no commercially available reference materials for IHg. However the obtained values for IHg in blood were well in agreement with our results from previous analytical runs of the same Seronorm sample 3133. The analytical variability, as calculated by coefficients of variation (CV%) of duplicate analysis of collected samples and reference materials, was low (Table 2). The detection limits (LOD) were lowered significantly by the introduction of the cleaning step of the reagent chemicals in the reaction tower prior to the sample addition (Table 3). As a result, very few samples of whole blood (n = 3), plasma (n = 2), and RBC (n = 3) had IHg concentrations below LOD.

A summary of the concentrations of Hg species in whole blood (B), red blood cells (RBC), plasma (P), urine (U) and hair (H) is given in Table 3. The correlations between the Hg species in the various media as well as the exposure variables fish consumption (number of meals per month; range 0–22 meals per month) and number of dental amalgam fillings (range 0–>15), are given in Table 4. Fish consumption was positively correlated with THg in blood (r_s = 0.74, p < 0.001), RBC, and hair, and with OHg in blood, RBC, plasma and hair (Table 4). Fish consumption was also correlated with IHg in hair. Number of dental amalgam fillings was positively correlated with THg in plasma (r_s = 0.46, p = 0.01) and urine (r_s = 0.49, p = 0.009), and with IHg in blood, plasma and urine (Table 4). Mercury levels in blood (THg, IHg and OHg in whole blood, RBC or plasma) were not associated with haemoglobin and haematocrit.

The distribution of OHg and IHg in whole blood between RBC and plasma was calculated as the percentage of total OHg (or IHg) in whole blood according to equation 1 and 2 (below), using individual haematocrit values (B-EVF, %). The range of B-EVF was 35–47% (mean 42%). Data below LOD were not included because of their uncertainty.

1) RBC-OHg * (B-EVF/B-OHg) * 100

2) P-OHg * ((1-B-EVF)/B-OHg) * 100

On average 87% of OHg in whole blood was localized in the RBC (95% CI of mean ± 3.7; range 76–104%; n = 20) and 9.6% in plasma (95% CI of mean ± 1.6; range 5.1–20%; n = 22). On average 34% of IHg in whole blood was localized in RBC (95% CI of mean ± 4.0; range 15–54%; n = 22) and 64% in plasma (mean; 95% CI of mean ± 5.4; range 30–81%; n = 22). The distribution of OHg or IHg between RBC and plasma did not change with increasing concentrations of the respective Hg form.

The concentration of IHg in RBC was positively correlated with both the concentration of OHg in RBC and the concentration of IHg in plasma (Table 4) indicating that IHg in RBC is a function of both IHg and OHg exposures. RBC-IHg was on average 6.8% of RBC-THg (median; range 3.3–24%), and increased with increasing concentrations of RBC-OHg (r_s = 0.46; p = 0.03) and increasing consumption of fish (r_s = 0.60; p = 0.003), but not with increasing number of dental amalgam fillings. In a person with no dental amalgam fillings RBC-IHg was 4.6% of RBC-THg.

The average RBC to plasma ratio of IHg concentrations was 0.90 (range 0.25–2.5), or as evaluated by linear regression, 0.50 (RBC-IHg = 0.11+0.50*P-IHg; R² = 0.56; Figure 2). The ratio increased with fish consumption (r_s = 0.52; p = 0.008; n = 25), but not with the number of dental amalgam fillings (r_s = -0.23).

The average RBC to plasma ratio of OHg concentrations was 14 (range 3.1–28). When evaluated by linear regression the ratio was also 14 (RBC-OHg = 0.20+14*P-OHg; R² = 0.72). The ratio did not increase with fish consumption (or number of dental amalgam fillings).

The total mercury concentration in hair (H-THg) was positively correlated with B-OHg and P-OHg, as well as with fish consumption, but not with B-IHg or the number of dental amalgam fillings (Table 4). Speciation of Hg in hair showed that on average 91% of THg was OHg (CI of mean ± 1.2; range 79–95%; n = 28), and 8.9% was IHg (CI of mean ± 1.1; range 4.9–21%; n = 28). In our previous study of women with a high fish intake, the distribution was approximately the same, i.e. 91% of THg was OHg (CI of mean ± 1.2; range 82–97%) and 8.7% of THg was IHg (CI of mean ± 1.2; range 3.2–18%; n = 144). The percentage of IHg in hair was not associated with the number of dental amalgam fillings. The average percentage of IHg in hair was 8.3% (range 4.4–13%; n = 23) in individuals without dental amalgam fillings, and 8.8% (range 4.6–18%; n = 48) in individuals with 10 fillings or more (including data from our previous study). The difference was not statistically significant (Student's t-test, p = 0.4). The concentration of IHg in hair was highly correlated with OHg in hair, and with OHg in blood (B-OHg, RBC-OHg and P-OHg; Table 4), but not with IHg in blood (B-IHg, P-IHg or RBC-IHg; Table 4). The concentration of IHg in hair was also positively correlated with fish consumption, but not with number of dental amalgam fillings (Table 4).

The average hair to blood ratio, (H-THg (mg/kg) divided by B-THg (μg/L)) was 0.366 (median 0.373; 95^th percentile 0.552; range 0.185 to 0.673). If H-THg was divided by B-OHg, the ratio was 0.465 (95^th percentile 0.670). If we included the hair and blood mercury data from our previous study of women with a high fish consumption the average hair to blood ratio was 0.341 (median 0.330; 5^th percentile 0.168; 95^th percentile 0.563; range 0.066–0.824; n = 173). The hair to blood ratio seemed to decrease with increasing B-OHg (Figure 3).

The ratio as determined by linear regression of H-THg versus, B-THg was 0.264, i.e. H-THg = 0.179+0.264*B-THg (R² = 0.83; p < 0.001; n = 28). Replacing B-THg with B-OHg only resulted in an increased intercept, to 0.282 (R² = 0.80; n = 25). Inclusion of data from our previous study in the linear regression analysis resulted in H-THg = 0.169+0.254*B-THg (R² = 0.62; p < 0.001; n = 173; Figure 4).

Essentially all Hg in urine (> 98%) was IHg. IHg in urine, adjusted to specific gravity (1.019 g/mL) or adjusted to creatinine (g creatinine/L urine) were highly correlated with IHg in blood, plasma and RBC, but not with OHg in the various media (Table 4). IHg in urine was moderately associated with the number of dental amalgam fillings, but not with fish consumption (Table 4).