|Home | About | Journals | Submit | Contact Us | Français|
There is limited literature concerning the effect of urinary flow rate on mercury excretion at low-level exposure. The aim of the present study is to examine the influence of urinary flow rate on mercury excretion in children. Also of interest is the influence of flow rate on creatinine excretion and creatinine-corrected mercury, which arise with spot urine samples.
A substudy of the New England Children's Amalgam Trial collected pairs of urine samples from children aged 10-16 years: a timed overnight collection and a spot daytime sample collected the following day. These samples were analyzed for mercury and creatinine concentration. Regression analysis was used to model the effect of urinary flow rate in the timed overnight samples. A paired t-test compared concentrations and creatinine-corrected mercury between overnight and daytime samples.
Creatinine excretion rate (mg/hr) increased significantly with urinary flow rate (mL/hr), whereas creatinine concentration (g/L) decreased with flow rate. We found a non-significant increase in mercury excretion rate (ng/hr) with flow rate, and mercury concentration decreased with flow rate. Mercury and creatinine concentrations were significantly higher in the overnight compared to daytime samples. For creatinine-corrected mercury, no significant impact of urinary flow rate was found.
Although the creatinine excretion rate, and probably the mercury excretion rate, increased with urinary flow rate, the mercury/creatinine ratio seemed relatively unaffected by urinary flow rate.
Mercury (Hg) is a toxic heavy metal, occurring in several physical and chemical forms. The most important from a toxicological point of view are elemental mercury vapour Hg0, and methyl mercury (MeHg). The major source of Hg0 in the general population is Hg released from dental amalgam fillings. Urinary mercury excretion is widely used to assess long-term exposure to and/or body burden of inorganic mercury.
Ideally one would like to measure the average excretion rate of mercury, e.g. the mass excreted per 24 hours or per hour. In practice, it is often difficult or inconvenient to collect 24-hour urine samples, or other timed samples, which are necessary in order to calculate the excretion rate. Instead spot samples are used, either first morning urine or daytime samples. However, these urine samples could be more or less concentrated, depending on fluid intake and urinary flow rate. Therefore, spot samples are usually ‘corrected’ for such effects using creatinine or specific gravity. Ideally, the mercury/creatinine ratio should be constant, irrespective of urinary flow rate.
Since spot samples are collected at various times of the day, and urinary flow rate (mL/hour) may vary substantially over the course of the day (or night), any effect of urinary flow rate on the excretion rate of mercury would cause problems in interpretation of the results. Additionally, when a correction for creatinine is used, it is also important to know the influence of flow rate on creatinine excretion rate, and in the end, on creatinine-corrected mercury.
There is limited literature on the effect of urinary flow rate on mercury excretion, and all such studies were performed on small numbers of male workers. Piotrowski studied 5 male mercury workers and found that mercury excretion increased with urinary flow rate and mercury concentration decreased with flow rate. Araki[5-7] studied 19 male metal workers and found no effect of urinary flow rate on mercury excretion, but a decrease in creatinine-corrected mercury with increased flow rate.
The Piotrowski study  also found diurnal variation in mercury concentration in 32 adult mercury workers, with higher levels at night and in the morning, and lower levels in the afternoon. Two additional studies [8, 9] (N=19 male metal workers and N=35 dentists) also found diurnal variation in mercury levels.
Furthermore both the Araki study [5, 6] and two others [10, 11] (N=20 adults and 19 men, respectively) found that creatinine excretion rate increased with flow rate. Therefore, in order for the creatinine correction not to introduce a flow rate bias, mercury excretion would need to increase with flow rate at the same rate as creatinine.
The aim of the present study is:
This study was performed as part of the New England Children's Amalgam Trial (NECAT)[12, 13]. The NECAT study was designed to examine effects of amalgam dental fillings in 534 children in Boston and Maine, aged 6-10 at the beginning of the study, for 5 years. The study sample was gender balanced and racially diverse. Mercury and creatinine concentrations were measured yearly.
The initial NECAT protocol called for yearly timed overnight urine samples for five years, with spot samples collected from those children who did not provide overnight samples. However, it became clear that too few children were providing timed overnight samples, and thus we switched to collecting only daytime spot urine samples in the middle of the trial.
Therefore, in order to determine how to analyze and interpret our longitudinal findings, we embedded also a small paired data collection substudy at the 4-year visit to examine the effects of overnight vs. daytime samples. These samples allowed us to compare the creatinine-corrected measures between overnight and daytime spot samples, as well as to examine the influence of urinary flow rate on the timed overnight samples. Since the substudy required quantification of low-level urinary mercury, these analyses were performed using a more sensitive method than in the main study.
Children in the substudy were instructed to collect a timed overnight sample, as well as to provide an un-timed spot sample on the next day. Of the specimens available, 42 pairs were of sufficient volume to allow analysis of mercury and creatinine concentrations. Of these 42 pairs, 6 pairs from Maine with sample volume <100 mL were excluded because of increased inaccuracy in measurement of urinary volume through use of a measuring cup with larger increments than in Boston, and thus with larger, albeit random, rounding error. Although this cut-off was somewhat arbitrary, it was necessary to exclude samples with small volumes, as these samples would have too large an impact on the variability (in terms of percent discrepancy) in calculation of excretion (excretion = concentration * volume / collection time). Of the 36 remaining pairs, 15 were not collected on the same day – 13 children failed to collect their overnight sample on the assigned day, and 2 did not provide a spot sample on the following day, due to scheduling difficulties. Of these 15 pairs, 13 were collected within 2 weeks of each other, thus providing what we believe to be useful information. The remaining two were collected far apart (4 months and 1 year) and thus excluded from comparisons between overnight and daytime samples, as they may not contain comparable concentrations of mercury or creatinine. Thus, the final sample size is N=36 for analysis of urinary flow rate (aim 1) and N=34 for comparison of overnight and daytime samples (aim 2).
Both mercury and creatinine were measured at the Department of Occupational and Environmental Medicine, Lund University, Goteborg, Sweden. For mercury an automized cold-vapor atomic fluorescence technique was used. Samples were digested using a mixture (5:1) of concentrated perchloric and nitric acids for 1 hour at room temperature. The detection limit was 0.1 μg/L for urine. Reference samples were included in all analysis series (Centre de Toxicologie du Quebec Interlaboratory Comparison Program, batch D-04-03, certified; target value 2 μg/L, obtained value 2.18 μg/L, SD 0.16 μg/L, N=4). The precision for duplicate analyses was good with a relative standard deviation of 8% at levels of 0.5 – 1 μg/L, and lower at higher mercury concentrations. Creatinine was determined by the photometric ‘Jaffe’ method (reagent from Roche diagnostics, detection limit 0.1 g/L).
Urine samples were stored frozen at −20 °C for 8 - 25 months until analysis, with a median storage time of 15 months.
Urinary flow rate (ml/hour) and excretion rates of mercury and creatinine (units/hour) were calculated in the timed overnight samples. To address aim 1, in the timed samples, regressions of excretion rates, concentrations, and creatinine-corrected mercury on urinary flow rate were performed. A log-transformation was used for all mercury measures. Analysis of covariance models (ANCOVA) also controlled for age, sex, race, lean body mass, and number of dental amalgam surfaces. Lean body mass was calculated as weight*(1 - % body fat), with body fat measured by a body fat scale (model TBF-551). To verify overall findings, a sub-analysis for models of excretion rates of mercury and creatinine was performed using only subjects from Boston (N=23), given that some urinary volumes in Maine were more innacurately reported (rounded) from larger calibrations on the measuring cup.
To address aim 2, paired t-tests were used to compare mercury and creatinine concentrations and creatinine-corrected mercury between overnight and daytime samples. A log-transformation was used for mercury concentration and its creatinine-corrected level. To verify overall findings, sub-analyses were performed (1) including only subjects whose overnight and spot samples were collected on the same day (N=21), and (2) excluding samples where we believed the timed overnight collection might have been incomplete (creatinine concentrations at least 20% under the regression line for creatinine concentration vs. urinary flow rate, and in the lowest quartile of creatinine excretion per hour per kg) (N=7 excluded and N=27 included).
Background demographic data on the 36 children are shown in Table 1. Descriptive analyses of urinary collection times, volumes, and calculated flow rates are shown in Table 2, as well as mercury and creatinine concentrations, excretion rates, and creatinine-corrected mercury.
As could be expected, creatinine concentration is significantly higher in the overnight compared to the daytime samples (table 2). The trend for mercury concentration is similar, with significantly higher mean concentration in the overnight samples. Mercury excretion in ng/mg creatinine was similar in overnight and daytime samples. Both subgroup analyses produced consistent findings.
Figure 1 shows plots of excretion rates and concentrations of mercury and creatinine and creatinine-corrected mercury, all vs. urinary flow rate. Clearly creatinine excretion rate increases with urinary flow rate. However, mercury excretion appears independent of urinary flow rate.
Table 3 shows results of regression/ANCOVA models of urinary flow rate on these five measures. Creatinine excretion significantly increases with urinary flow rate. Creatinine concentration appears to decrease with urinary flow rate, as could be expected, with significance reached when 7 potentially incomplete collections are excluded. There were no significant covariates in the ANCOVA models for creatinine excretion and concentration.
Mercury excretion does not vary significantly with urinary flow rate. However, the variability is large and the slope is positive. Mercury concentration decreases with flow rate, as could be expected, with significance reached when 7 potentially incomplete collections are excluded. For creatinine-corrected mercury, no significant impact of urinary flow rate was found. All measures of mercury significantly increased with the number of amalgam surfaces (p<0.001).
No conclusions are changed when analysis is restricted to the Boston sample (N=23).
We have found that the creatinine excretion rate increases with urinary flow rate in overnight samples, but as expected, the creatinine concentration decreases with flow rate. The latter is in accordance with the finding of significantly higher creatinine concentrations overnight than in the daytime (when flow rate is higher). This indicates that the increase of the creatinine excretion rate with urinary flow rate in daytime only partly outweighs the ‘dilution’ effect on creatinine concentrations of the increased daytime urinary flow. The theory behind adjustment for creatinine, commonly used for many biomarkers, assumes the excretion rate of creatinine to be constant, but our results show that this is not the case in children. Similar results have been shown previously in adults, where the creatinine excretion rate increased with urinary flow rate [10, 11]. Part of this increase may be explained by a higher glomerular filtration rate in daytime, when urinary flow rate is higher . It is often recommended that very concentrated (low flow rate) or dilute (high flow rate) urine samples be excluded from studies where metal concentrations are assessed in urine . However, all samples showing a clear association between creatinine excretion rate and urinary flow rate (Figure 1D) are within the recommended “acceptable” range of 3-30 mmol/L creatinine (0.3-3.4 g/L) , as shown in Figure 1E.
For mercury, we found no significant increase in mercury excretion rate with flow rate. There was a `dilution effect' with lower concentrations when urinary flow rate increased in overnight samples and also lower concentrations in daytime samples than overnight, when urinary flow usually is lower. As shown by the point estimate in Table 3, our data are, however, compatible with a moderate increase of mercury excretion rate with urinary flow rate, but if so, the relative effect is probably smaller than the clear effect of urinary flow rate on the creatinine excretion rate.
An apparent contradiction is indicated in Table 3; if an increased flow rate increases the creatinine excretion rate but not the mercury excretion rate (or less than the creatinine excretion rate), the effect should be an inverse relation between urinary flow rate and creatinine-corrected urinary mercury excretion. Such an inverse relation was weakly indicated in the regression models (Table 3), but not significantly so. On the contrary, the creatinine-correction seemed to work fine; also there was no significant difference in creatinine-corrected mercury concentrations between overnight and daytime samples, in spite of differences in urinary flow rates in day and night. Urinary flow rate, which depends on fluid intake and is also strongly related to the time of day, thus affects mercury and creatinine concentrations mainly by the ‘dilution’ effect. Additionally, despite lack of statistical significance, there is evidence of a small but real increase in mercury excretion rate with urinary flow rate.
One limitation of this study is that these children, who met stringent eligibility criteria for a clinical trial, by no means constitute a community sample. The eligibility criteria for the NECAT study, including the presence of untreated dental caries, resulted in a study population with lower socio-economic status than the general population. Secondly, mercury levels in this study were typically low. Conclusions cannot be extended to high mercury levels as might be found in adult industrial workers. Therefore, this study does not necessarily contradict the results found by Piotrowski. Finally, we only collected timed samples overnight, when urinary flow rate is typically lower. Thus, we are only able to make conclusions that hold on the lower end of urinary flow rates. We cannot extrapolate to higher flow rates, and further research on timed daytime samples would be valuable.
Epidemiological studies, in which valid data on mercury excretion are needed, should ideally control for urinary flow rate. When this is not possible, creatinine-corrected mercury levels should be appropriate, with control for variables that are related to flow rate, such as time of day, age, and sex. Fortunately, in a randomized trial, variability in urinary flow rate should be approximately balanced in the treatment arms, and thus should not affect any conclusions of the study; however, variances will be increased, causing lower power.
On the other hand, epidemiological studies should be aware that mercury and creatinine concentrations are not comparable in the daytime and overnight. Therefore, a single time of the day should ideally be selected for urine collections. If samples have already been collected at varying times of day, as with NECAT, it is best to control for time of day in statistical models. In the planning stages of a study, however, if compliance is expected to be low for collections of timed overnight samples, a reasonable alternative may be an un-timed first morning collection.
This study was supported by a cooperative agreement (U01 DE11886) between the New England Research Institutes and the National Institute of Dental and Craniofacial Research, National Institutes of Health.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.