To better define dose–response associations linking MeHg exposure with deleterious effects, whether the effect be neurodevelopmental or cardiovascular, it is important to reduce uncertainty and address the limitations of exposure assessments. The relationship between dose and response may be obscured when biomarker values are unable to adequately reflect ongoing exposure. Single time point measurements are subject to a range of factors (e.g. variations in consumption frequency, magnitude and duration) that can lead to uncertainty when using single samples to estimate average MeHg exposure over time . In addition, since biological samples integrate MeHg exposure over variable periods of time, the use of such measures can result in uncertainty when attempting to estimate exposure over a specific period of time (e.g., a single trimester of gestation) . These considerations become particularly important when MeHg biomarkers are employed to reconstruct the intake dose associated with an observed biomarker-outcome relationship such as those observed in the Faroe Island study. Such reconstructions are required in order to convert such data into fish consumption advisories . This work suggests that, on an individual basis, blood-Hg levels from separate exposure time periods do not provide a reliable estimate of longer term exposure as represented by a three sample average.
Single time point blood-Hg analysis reflects the aggregate of recent as well as less recent exposure periods. Blood samples were obtained so that the measured Hg levels represented exposure periods largely independent of each other with respect to the half-life of MeHg in blood. The half-life of MeHg in blood is considered to be approximately 50 days and more than two half-lives separated each participant’s samples in this study [17, 18]. Accordingly, less than 25% of an individual’s prior mercury exposure will be reflected in the blood-Hg levels observed between successive samples. This tended to minimize (but not eliminate) auto-correlation among the samples. The similarity in the pooled distribution of blood-Hg values between sampling events, based on group mean values that did not differ over time, is consistent with previous observation made by Tsuchiya et al.  in hair samples. As the individual sample values were not constant over time, the sum of intra-individual variabilities across the sampling periods appear to regress to the sample mean.
In contrast to the temporal stability of the three blood-Hg levels for the group as a whole, there was marked change over time in individual blood-Hg levels. Blood data show that separate and nearly independent samples obtained within a time period of several months can fluctuate by more than 100% and individual sample values exceed the individual’s average across the specified time period by greater than 50%. Further, the proportional variability in blood-Hg was similar for individuals at all levels of blood-Hg with the exception of those above the 90th percentile in blood-Hg. These data may reflect both intra-individual pharmacokinetic variability as well as the change in MeHg intake observed across sample intervals. Accordingly, the temporal blood-Hg variability may reflect short-term changes in MeHg intake, absorption, distribution, metabolism, and elimination.
The temporal changes in blood-Hg levels may aid in explaining the large observed disparity in hair to blood ratios (mg Hg/kg hair: μg Hg/ml blood) found in the literature that range from less than 100:1 to approximately 400:1 . A ratio of 250:1 is generally accepted for individuals having reasonably constant exposure levels. Yet, within this group of Japanese women, the ratios of the proximal 6 cm hair strand mean Hg levels (described previously, ) to the whole blood mean Hg levels across the three visits were 373:1, 242:1 and 300:1. The overall three-sample mean was 305:1.
There are two shortcomings with this work that need be noted. First, comparisons between mean blood-Hg levels, mean blood-Hg levels adjusted for change in blood volume during gestation, as well as minimum and maximum levels led to no observable difference by pregnancy status. Although no data exist from other sources that permit a comparison of these two groups, pregnant women have a decrease in Hg body burden between the 2nd and 3rd trimester possibly due to increased blood volume or transfer of Hg across the placental barrier to the fetus [5, 12]. Our study was not developed to address this issue as participants were accepted into the study at various time periods during gestation with several women becoming pregnant during the study period resulting in only a few women being enrolled for the last two trimesters of pregnancy. Nonetheless, it seems reasonable that any pregnancy-related difference in blood Hg concentration was obscured by the overall intra-individual temporal variability. Further, the small sample does not provide for sufficient power to observe a difference between groups or to see a significant decrease in body burden across the last two trimesters of pregnancy. Second, since samples were analyzed for total blood-Hg, the results do not delineate between organic and inorganic forms of Hg. We cannot exclude exposure through inhalation (Hg vapor from dental amalgams), ingestion (drinking-water and food), or even from medical treatments, as well as possible contribution from indoor air [1, 19, 20]. Also, there is the ritual use of Hg in folk medicine within several populations that may allow for exposure to elemental or inorganic Hg salts .
These additional exposures could have impacted blood-Hg levels and affected correlations between fish consumption and blood-Hg levels. Comparisons between blood-Hg and the FCS reflecting dietary practices across the preceding 14 days provided for low correlations. Although the FCS was a robust instrument with pictorials and species names in multiple languages and used fish models for determining portion size, the estimated intake data did not provide for a precise representation of actual Hg intake [9. 10]. Not all fish species consumed by this population were analyzed for Hg. For the species not analyzed directly, literature based estimates of species specific Hg levels were used. These values may not have accurately reflected the same species or may not have reflected regional differences among the same species. Other factors that could have impacted the relationship between estimated Hg intake and body burden levels include recall bias, variability in toxicokinetics, nutritional status and dietary interactions [1, 21–23].
The results of this work brings into question the use of blood-Hg samples collected at a single point in time to predict average exposure and risk. Physicians have been advised to obtain blood-Hg levels in individuals considered to be exposed to MeHg at elevated levels due to their fish consumption behavior patterns . Although blood analysis may be one tool for examining ongoing exposure, assuming that symptomatic individuals will show elevated blood-Hg concentrations based on a single sample may lead to misdiagnosis and misclassification of exposure. Even within groups or populations that appear to have Hg body burden levels in steady-state, exposure periods largely independent of each other suggest that individual blood-Hg levels can vary markedly. In addition, as a single blood sample may be a relatively poor predictor of future exposure, a blood sample obtained for example at or before conception may not provide the best foundation from which to offer pre-pregnancy advice on ways to reduce exposure during the neurodevelopmentally sensitive stages in fetal development or to provide exposure estimates for use in epidemiologic studies.
If further research endeavors using, for example, a more robust sample size provide results supporting those observed in this study, then average long term Hg exposure values obtained from hair strands may provide an alternative to blood-Hg levels. Hair-Hg values may provide individuals overexposed or concerned about elevated exposures along with pregnant women or women who are considering pregnancy a better means from which to derive subsequent diet alterations so as to minimize MeHg exposure. There are concerns with the use of hair as a metric of exposure such as ethnic differences and impact from various available hair treatments as well as airborne-Hg deposition, but this type of strategy could allow entire populations with potentially elevated levels of exposure to MeHg to be protected through screening as analyses of hair for Hg using atomic absorption spectrometers have come with decreased costs and increased efficiency during the last decade.