Urinary mercury excretion in chloralkali workers after the cessation of exposure.

enteenformerchloralkaliworkerswerefollowedthroughthe regulardetermination of urinarymercury for nearly two years after the cessation of exposureto mercuryvapor in a study of the time course of urinary mercury elimination. Their duration of exposure ranged from 3 d to 35.5 years. A one-com partment model for urinary mercuryelimination was applied. The urinary mercury concentrationde clined at a rate indicating a half-time of 91 d. When corrected for an individual base-line level of urinary mercuryresulting from sourcesof mercuryintake not related to work, the half-time was 72.4 [95% confidence interval (95% CI) 63.2-81.7] d, with a mean elimination rate constant of 0.011 (95% CI 0.008-0.013) . d'. The day-to-day variability of the urinary mercury concentration aver aged 22%, expressedas the coefficient of variationbetweenurine samples deliveredon three consec utive days.

There are limited data on the time course of urinary mercury elimination in humans after exposure to mercury vapor. In an experimental study (l), five volunteers were followed for 7 d after being exposed to radiolabeled mercury vapor for 14 to 24 min. The mean elimination half-time for mercury in the kidney region was calculated to be 64 d. An initial marginal increase in the urinary mercury concentration after the cessation of exposure was demonstrated among the same volunteers (2). In a three-week follow-up of six workers with long-term exposure to mercury vapor, a fast initial decrease was observed in the urinary mercury concentration with a half-time of approximately 2 d, followed by a slow excretion phase with an elimination rate constant of 0.01 . d' (3). In a study on nine men exposed to mercury vapor for 20-45 h, an increase in urinary mercury concentration was observed for the first 19 d after exposure, followed by an average urinary mercury elimination half-time of 50 d. The time of followup was 4-37 months (4). Roels et al (5) observed seven workers for about 250-300 d after the cessation of mercury vapor exposure. Their application of a one-compartment elimination model yielded a mean urinary mercury half-time of 90 d.
The present investigation is part of a study on adverse effects among former chloralkali workers previously exposed to mercury vapor (6,7). Its  334 to study the time course of urinary mercury elimination in a postexposure period of two years among former chloralkali workers exposed to mercury vapor and, furthermore, to estimate the elimination halftime and the elimination rate constant.

Subjects and exposure
The study group was recruited from workers at a chloralkali plant which was closed in 1987 and subsequently dismantled. Subjects who had ceased being exposed between 1987 and 1989 were asked to participate in the study. Sixteen men and one woman volunteered to deliver urine samples for the determination of urinary mercury for a period of two years after the cessation of their mercury vapor exposure. Four of the subjects had been involved in the dismantling of the plant, and 13 had participated in regular production, nine as maintenance or mechanical repair workers and four as cell room operators.
All of the participants were in good health, and seven had participated in a comprehensive clinical examination in another part of the present project (6,7). The subjects averaged 38.1 (range 18.7-58.9) years of age when the exposure ceased. The average duration of exposure was 6.2 years (range 3 days-35.5 years). Ten subjects had been exposed for more than one year. The characteristics of the subjects are presented in table 1.
A cumulative urinary mercury "dose" was calculated for each individual on the basis of the urinary mercury concentrations determined during their time of exposure. This dose surrogate was constructed by summing the mean urinary mercury concentration excreted each year. Further details regarding the es-tablishment of this "dose" have been presented elsewhere (6).
One subj ect stopped partic ipating 280 d afte r the cessatio n of expos ure. Two subjects delivered their last urine samples between 500 and 600 d poste xposure, while the remaining subjects were monitored for more than 700 d. The mean time of follow-up was from day II (range 1-30) to day 702 (range 280-8 12) postexposure.

Sampling procedure
All of the subj ects were instru cted to deliver a first void mornin g urine sample, a pro cedure which had been practiced at the plant since 1948 when biological monitorin g of urinary mercury commenced. Urinary mercury was determined by routine analyses at the laborat ory of the study plant (laboratory A) up to 1988. From that time on a noncompany laboratory (lab oratory B) performed the analyses. However, for about four months in 1988, the analyses were carried out in duplicate in both laboratorie s. Whenever possible, the urine samples were collected on three consec utive days from March 1988 on so that we could account for the day-to-day variability in the urinary mercur y excretion,. The samples were collected and stored in separate NUNC®polyeth ylen e tubes.
In all, 376 single urine speci mens were collected from the 17 subjects. The result s of the determined mercur y concentrations in the samples from consecutive days were averaged after anal ysis for the purpose of data presentation. Thus 2 17 individual results of urinary mercury were used to assess the time course of urin ary mercu ry excretion, 53 fro m laboratory A and 164 from laboratory B. Urinary mercury was determined in 72 samples in both laboratories. In the cases for which the urinary mercury concentratio n was determined in both labor atories, the concentrations determined in labor ator y B were used in the study of urinary mercury elimination. Otherwise these results were used to assess the interlaboratory differences in the measured urinary mercury concentrations. All of the samples were determined from fresh urine over the entire period of monitoring.

Determination of urinary mercury
Laboratory A had measured mercury in urine since 1948. Up to 1975 this laboratory applied spec trophotometric detection of mercury after acid mi neralization (dithizone method). In 1975 cold vapor atomic abso rptio n spectrophotomet ry wa s intr oduc ed and subsequently used throu gh 1987. A modified potassium perman ganate/sulfuric acid (KMnO/ H 2 S0 4 ) digestion and stannous tin (Sn''") redu ction procedure (8) and a laboratory data control mercur y spectrophot ometer (model 1235) was used . In the absence of quality assurance materials, both laboratories (A and B) had regularl y determined mercury in the same Scand J Work Environ Health 1993. vol 19, no 5 urine specimens, selec ted randomly since the 1950s unt il the plan t was close d. Laboratory A ge nera lly reported somew hat higher valu es (about 10%) than laboratory B. Thi s difference can be explained by loss of mercury during the two-to three-w eek storage period before measurements in laboratory B.
Urinary mercury was determined by labo ratory B using a co ld vapor batch generation system co upled with a laboratory data control mercury spec trophotometer (model 1235). Two replic ates of 125111 from the preheated urine spec imens (85°C, to redissol ve urine precip itates) were inj ected into a stannic chloride/sodium hydro xide (SnCI/NaOH) rese rvoir and analyzed for mercury. Th e detecti on limit of the meth od is 2 nmol . I-I. The accuracy and precision of the mercury measurements were mon itored with human quality contro l material (Seronorm Trace Element 108, Nycomed Ltd, Oslo, Norw ay). The within-day and between-day variations of this material typically had a relative standard deviation of I%. The average mercury concentration measured in this material (245 nmol . 1. 1 , N = 15, SD =2) was in agreement with the recomm ended value of 250 nmol . I-I. The urin ary mercury concentrations were corrected for urinary dilution by correcting for urin ary crea tinine, which was determined by a Beckm an creatinine analyzer 2 (9) .

Statistics
A one-comp artment model was assum ed after the indi vidual time courses of the urin ary mercur y eli mination were studied. The fast first phase of urinary mercury exc retion reported by Piotrowski et al (3) was not observed in the individual elimin ation curves in our study , nor was the rise in the postexposure urinary mercur y concentration which has been reported by Barregard The result s of the urinary mercur y measurements were log-transformed. After about one yea r of fol- low-up the urinary mercury concentration was low and at a level that could be expected in the genera l population. Hence a cutoff for follow-up was used that was chosen to correspond to the last measured individua l urinary mercury concen tration above 2.5 nmol . mmol' creatinine. That concentration corresponds to the mean plus two standard deviations (1.3 + 1.2 nmol . mmol' creatinine) established for an occupationally unexposed reference population in the main study on clinical effects among workers previously exposed to mercury vapor (7).
In an alternative approach for correcting for the urinary mercury base-line level , attributable to sources of mercury exposure not related to work, the lowest individ ually recorded urinary mercury concentration was subtracted from the measured concentrations.

Results
The urinary mercury concentrations determined in both laboratory A and laboratory B were log-transformed, and a least square regression analysis was performed. The Pearson' s correlation coefficient between these concentrations was 0.90 (P = 0.001). The concentrations determined in laboratory A averaged 96.1% of those determined in laboratory B [95% confidence interval (95% CI) 88.2-103.1] . Figure I presents the decline of urinary mercury in relation to time since cessation of exposure for two of the subjects under study (subject E and I), both monitored for more than 700 d. Figure 2 presents the time course of urinary mercury elimination among all of the subjects under study, the urinary mercury concentrations(log) being uncorrected for the base-line urinary mercury concentration. Data regarding these regression lines are presented in table 2. The mean value of the correlation coeffic ients (Pearson) was 0.93 (range 0.80-1.0). A linear pattern was also apparent among the subjects with lower correlation coefficients. The mean time of follow -up was shown to be 254 (range 114-406) d when the cutoff of 2.5 nmol . mmol' creatin ine was applie d. According to the calc ulated regression lines, the urinary mercury decreased with a half-time of 91.0 (range 33.0-127.5, 95% CI 79.6-102.5) d. Figure 3 shows the median concentrations and ranges related to the time since the individual cutoffs. The bar indicating cutoff shows the last urinary mercury concentrations used in the regression analysis for each participant. Each bar represents the specified time interval since cutoff. It appears that the urinary mercury concentration declined until the time interval of 200-299 d after the cutoff. Because of the low urinary mercury concentrations measured after the applied cutoff value, no individual halftimes were calculated.
An unexposed reference group (N = 5 1) was studied with regard to dental status in another part of the present project (7). Figure 4 shows the regression line between the surface of dental amalgam expressed as amalgam points and the urinary mercury concentrations among the referents in that study. The lowest urinary mercury concentrations determined among the nine subjects in the prese nt study with known amalgam points are plotted in the figure (broken line). Most of the subjects had urinary mercury concentrations within the expected range when related to the amalgam points.
The urinary mercury half-time was recalculated with the concentrations measured in laboratory A (96.1% of the concent rations in laboratory B) standardized to those determined in laboratory B. Only minor changes occurred in the regression equations. Generally the correlation coefficients became somewhat higher, and the mean urinary mercury half-time was calculated to be 89. 8  . .    Table 2. Parameters related to the linear regression equations for urinary mercury excretion, not corrected for the ind ividual urinary mercury base-line concentration, among t he 17 subjects, previously exposed to mercury vapor.     The surface of dental amalgam has been shown to correlate with urinary mercury concentrations among subjects unexposed to mercury at work (10, II). Thus dental amalgam and other sources of inorganic mercury exposure can result in an elevated "base-line level" of urinary mercury among subjects unexposed at work. The lowest measured urinary mercury concentration in each individual during the time of monitoring in the present study was used as an estimate of the base-line level. These concentrations were subtracted from the measured concentrations, and the half-time of the urinary mercury elimination was recalculated. Table 3  rate constants (k· d'). One subject (subject L) was followed for less than 300 d. His base-line level was set at 1.4 nmol . mmol' creatinine, which was the mean base-line level used for correction among the remaining 16 subjects. This procedure yielded substantially lower half-times for the mean urinary mercury concentration, which was 72.4 (range 27.8-95.6, 95% CI 63.2-81.7) d. The elimination rate constant averaged 0.011 (range 0.007-0.025, 95% CI 0.008-0. 013)· d-1 • Figure 5 shows the time course for urinary mercury elimination among subjects A to D with the use of both values corrected and values uncorrected for the urinary mercury baseline level. No significant association was observed between the calculated half-times and the cumulative urinary mercury dose or the number of months exposed. When the subjects were dichotomized according to the duration of exposure (cut point 12 months), the half-time of the urinary mercury decline among the subjects exposed for more than 12 months (N = 10) did not differ significantly from the half-time of the subjects exposed for less than 12 months (N = 7).
Seventy-five urinary mercury concentrations in the present study were based on urine samples delivered for analysis from three consecutive days. Laboratory B determined these samples. These concentrations were mainly observed at the end of the follow-up. The mean mercury concentration was 2.8 (range 0.4-20.0) nmol . mmol' creatinine. The coefficient of variation was calculated for each series of three analyses, and the mean coefficient of variation was 22 (range 4-64)%.

Discussion
In our study urinary mercury was determined in fresh urine. Both participating laboratories took part in in- Table 3. Parameters related to t he lin ear regression equati o ns for urinary mercury excret ion among the 17 subjec ts previou sly exposed to merc ury vapor. Estim atio ns made after correctio n for individual base-lin e urinary merc ury con centr ati on s.  terna l quality contro l linked to external interl aboratory control. Therefore, it is likely that the results are accura te. The results from the ana lyses in the laboratories were in agreement.
Various studies have shown a substantial diurnal varia tion in urinary mercury excretion (12,13). To reduce this variability, we analyzed the mercury concentrations in morni ng urine samp les and corrected for urin ary dilution. In addition, we collected samples from three consecutive days whenever possible and averaged the measured concentrations. The variation coefficient of the mercury concentrations measured in the urine specimens from consec utive days was 22%, a value indicating a substantial dayto-day variation in mercury excretion.
The participants were followed from day 11 postexposure on the average. Some studies have presented conflicting results regarding the first phase of urinary excretion after the cessation of mercury exposure . Piotrowski et al (3) observed a fast washout phase in the first days of postexposure with an elimina tion half-time of 2 d in three chloralkali workers exposed to mercury vapor. During 7 d of postexposure Cherian et al (2) observed a low correlation betwee n urinary mercury and the dura tion of follow-up for five volunteers exposed to mercu-Scand J Work Environ Health 1993, vo119, no 5 ry vapor from 14 to 24 min. Barregard et al (4) observed an increase in urinary mercury excretion which reached a plateau at 19 d postexposure. The results in the present study were too sparse to permit an assessment of the excretion during the first days of postexposure.
The participants were monitored for about 700 d on the average. During this time the concentration of urinary mercury reached a level considered to be in the "normal range." To avoid estimation of halftimes that were too long, a cutoff value for the urinary mercury concentration above the "normal range" had to be determined.
Urinary mercury concentration also reflects exposure to inorganic mercury from sources not related to work (ie, dental amalgam). Elimination half-times which are too long can be calculated if one does not correct for this base-line level. The individual baseline level chosen for correction in the present study was, on the average, 1.4 nmol . mmol' creatinine, which is in agreement with the mean urinary mercury concentration of 1.3 nmol . mmol' creatinine in the previously mentioned study (7) among subjects with exposure not related to work.
The correlation coefficients calculated between the number of days of postexposure and the urinary mercury concentrations were high in most instances (median Pearson's correlation coefficient 0.95), and they were only below 0.90 for three subjects.
In our study the concentration of urinary mercury decreased with a half-time of 72.4 and 91.0 d for corrected and uncorrected urinary mercury concentrations, respectively.
On the assumption that correction for a base-line level is appropriate, our results indicate 72.4 (95% CI 63.2-81.7) d as the best estimate of the half-time of urinary mercury elimination and a corresponding urinary elimination rate constant of 0.011 (95% CI 0.008-0.013) . d'. Hursh et al (1) reported an average half-time of 64 d for the kidney region by using radiolabeled mercury in volunteers followed for 7 d. On the assumption that urinary mercury concentration is an index of the renal mercury burden, as suggested by Cherian et al (2), our results agree with those results. Barregard et al (4) reported an average half-time of 50 d for urinary mercury in nine subjects after an average of 19 d postexposure. The use of a common half-time for all of the subjects in that study yielded an estimate of 59 d.
Correction for a urinary base-line level was not reported by Roels et al (5) for seven chloralkali workers monitored for about 300 d. They reported a mean urinary mercury half-time of 90 d. Skare & Engqvist (14) calculated a median half-time of 41 d in a study among dentists, assuming a one-compartment model with only two data points for each individual. The concentrations of urinary mercury in that study were low. Piotrowski et al (3) reported an elimination rate constant for the slow phase of urinary mercury excretion of 0.01 . d' among six subjects 340 monitored for three weeks; this value is close to our estimate (k =0.011 . d').
We monitored 14 subjects for more than 700 d. Their median urinary mercury level was 1.6 nmol . mmol' creatinine versus 1.2 nmol . mmol' creatinine in the reference group previously referred to (7). The amalgam status was known for nine of our subjects. A comparison of these subjects with the referents showed that their urinary mercury concentrations were slightly higher than expected in relation to the amalgam points. However, no calculations for half-times in a possible long-term compartment were performed due to the low urinary mercury concentrations measured at the end of the follow-up. According to the group results, the level of urinary mercury did not appear to decline for more than 200-299 d after the cutoff level was reached.