Long-Term Stability of Inorganic, Methyl and Ethyl Mercury in Whole Blood: Effects of Storage Temperature and Time

Abstract

In this study, we evaluated the effect of temperature on the long-term stability of three mercury species in bovine blood. We used inductively coupled plasma mass spectrometry (ICP-MS) analysis to determine the concentrations of inorganic (iHg), methyl (MeHg) and ethyl (EtHg) mercury species in two blood pools stored at temperatures of −70, −20, 4, 23°C (room temperature) and 37°C. Over the course of a year, we analyzed aliquots of pooled specimens at time intervals of 1, 2, 4 and 6 weeks and 2, 4, 6, 8, 10 and 12 months. We applied a fixed-effects linear model, step-down pairwise comparison and coefficient of variation statistical analysis to examine the temperature and time effects on changes in mercury species concentrations. We observed several instances of statistically significant differences in mercury species concentrations between different temperatures and time points; however, with considerations of experimental factors (such as instrumental drift and sample preparation procedures), not all differences were scientifically important. We concluded that iHg, MeHg and EtHg species in bovine whole blood were stable at −70, −20, 4 and 23°C for 1 year, but blood samples stored at 37°C were stable for no more than 2 weeks.

Introduction

Mercury occurs in the environment in different chemical forms, including elemental (Hg) and inorganic (iHg), and various organic forms, such as methyl mercury (MeHg) and ethyl mercury (EtHg). Speciated metal analysis of clinical samples has become widely accepted for toxicological purposes. Mercury is one of the toxic metals of most concern in the clinical area. Substantial accumulation of this hazardous element in the environment poses health risks to humans (1, 2). The health effects of mercury are diverse and depend on the form of mercury encountered and the severity and length of exposure (3). Our laboratory supports a number of public health biomonitoring studies, such as the National Health and Nutrition Examination Survey (NHANES), of which the measurement of mercury species in human whole blood is a component (4). It is important to understand the stability of mercury species in whole blood between the time that a specimen is collected and analyzed by the laboratory. The preservation and stabilization of mercury species in blood matrix in-between sample collection and analysis is a challenging task because of potential mercury species interconversions (5–7). These changes can compromise specimen integrity and lead to significant errors in the determination of the mercury species concentrations. This in turn will cause incorrect conclusions to be drawn about a patient's mercury exposure (i.e., which species the patient was exposed to and in what concentrations).

Laboratory analysis immediately after sample collection is rarely possible. Time to transport samples to a laboratory for analysis can significantly vary, from hours to days, and storage temperatures during transportation sometimes are not ideal (8). Additionally, large numbers of samples might need to be analyzed as part of population surveys and community exposure evaluations. Those samples undergo many logistic steps and long storage periods before they are analyzed in the laboratory. Long-term storage of biological samples may be necessary for future monitoring investigations involving the comparison of specimens obtained possibly years apart from the same individual (8). For all of these reasons, it is important to ensure that the integrity of samples is not compromised over time.

To the best of our knowledge, there were no long-term stability studies of mercury species in whole blood when this analytical method was developed. A few papers on the stability of total mercury in biologic samples were published in the 1980s and 1990s (9–11). We conducted a long-term stability study of iHg, MeHg and EtHg species in whole blood. The study focused on two factors: the temperature at which samples are stored and the storage period. Two bovine blood pools were prepared and fortified with different concentrations of iHg, MeHg and EtHg. We investigated the stability of the mercury species in these pooled specimens over the course of 1 year. Various aliquots were stored at five different temperature conditions: −70, −20, 4, 23°C (room temperature) and 37°C. At each time interval, the pooled samples were visually inspected for clotting and obvious changes in viscosity followed by quantitative analysis to determine the concentrations of individual species. We used a triple spike isotope dilution (TSID) method employing capillary gas chromatography (GC) and inductively coupled dynamic reaction cell mass spectrometry (ICP-DRC-MS) to quantify iHg, MeHg and EtHg in whole blood during the stability study (12). We used a fixed-effects linear model, performed step-down pairwise comparison and used coefficient of variation (CV) statistical analysis of all the analytical data to get a comprehensive understanding of changes and trends in mercury species concentrations over time at different storage temperatures.

Experimental

A detailed description of the sample preparation procedures and the analytical method used to quantify mercury species in bovine whole blood samples can be found elsewhere (12). Here we provide a summary of the experiment.

Sample preparation

We prepared two sets of pooled samples using bovine whole blood (EDTA disodium salt used as an anticoagulant) purchased from the Wisconsin State Laboratory of Hygiene. We distinguished those as the low blood (LB) pool, which had lower concentrations of mercury species, and high blood (HB) pool, which had higher concentrations of mercury species. The LB and HB pools were fortified with different concentrations of iHg, MeHg and EtHg. We acquired inorganic mercury, 1,000 mg/L in 10% nitric acid, from SPEX CertiPrep (Metuched, NJ, USA). We obtained methylmercury chloride standard solution, 1,000 mg/L, and ethylmercury chloride (powder) from Alfa Aesar (Ward Hill, MA, USA).

The LB pool was characterized by analyzing 144 aliquots over 9 months, on two different instruments and by two analysts. The HB pool was characterized by analyzing 75 aliquots over 3 months, on two different instruments and by two analysts. After analysis was complete, the mean and standard deviation (SD) for each pool were calculated. LB and HB pools contained concentrations of 1.11 ± 0.15 and 4.34 ± 0.81, 1.08 ± 0.12 and 3.99 ± 0.45, 1.49 ± 0.24 and 5.45 ± 0.69 μg/L for iHg, MeHg and EtHg, respectively (Supplementary Table I), the ‘±’ value represents 3 SD. Specific portions of each of the LB and HB pools were reserved to be used as stability study samples and additional samples from each pool were used with each analytical run as bracketing quality control (QC) material samples and will be referred to as QCL (LB material) and QCH (HB material). Note that the supply of bracketing QCL and QCH material was exhausted at 6 months, thus we began using new human whole blood pooled material as bracketing QC samples and it will be referred to as QCL2 and QCH2 (concentrations and limits presented in Supplementary Table II). For this new pool, different spiking material was used: naturally abundant HgCl₂ in 1% (v/v) nitric acid, CH₃HgCl in 1% (v/v) sodium thiosulfate and C₂H₅HgCl in 1% (v/v) sodium thiosulfate from Applied Isotope Technologies (AIT), Inc. (Sunnyvale, CA, USA). Human whole blood was purchased from Tennessee Blood Services (Memphis, TN, USA), which has Institutional Review Board (IRB) approval (protocol number TBS-12-01, expiring 16 July 2015). Our laboratory switched to using QC prepared from human blood pool in all the bioanalytical blood methods in order to get a better matrix match for analyzing patient samples. As of now, we have been using this QC for mercury speciation method for 3 years and have not seen any difference in performance from previous QC (bovine blood pool).

All reagents used for sample preparation were of analytical or reagent grade purity. Samples were prepared for analysis by adding 100 µL of a spike solution containing known amounts of enriched ¹⁹⁹HgCl₂, CH₃²⁰⁰HgCl and C₂H₅²⁰¹HgCl isotopic standards (AIT) to 100 µL of the blood sample and immediately mixing by vortexing for 5 s. Next, we added 500 µL of tetramethylammonium hydroxide, 25% w/w in methanol (Alfa Aesar). The samples were placed in an 80°C convection oven (Precision Freas Model 605, Thermo Scientific, Pittsburgh, PA, USA) for 24–26 h for blood solubilization. We transferred 200 µL of the solubilized sample to a 20-mL glass solid-phase microextraction (SPME) analysis vial (Wheaton, Millville, NJ, USA), into which 7.7 mL of 0.1 M sodium acetate buffer (pH 4.75) was added, bringing the final pH to 5–6. We then added 250 µL of 0.2% w/v of sodium tetra(n-propyl)borate (NaBPr₄; ABCR GmbH, Karlsruhe, Germany) to promote volatility of mercury species. The glass vials were capped and manually mixed by repeated inversions for ∼5 s while derivatization of mercury species occurred at room temperature, then samples were ready for analysis.

Instrumentation

Our laboratory measured the mercury species with an ELAN DRC™ II coupled plasma mass spectrometry (ICP-MS; PerkinElmer Life Sciences, Shelton, CT, USA) operated in DRC mode using argon gas at 0.3 mL/min, which enhanced the mercury signal intensity by collisional focusing (12). We used a PerkinElmer Clarus 500™ gas chromatograph (PerkinElmer Life Sciences) equipped with a GC capillary column (PerkinElmer Elite-5 30 m, 0.25 mm inner diameter, 0.25 µm of 5% diphenyl, 95% polydimethylsiloxane). The GC was coupled with the quadrupole ICP-MS by a heated GC transfer line (Redshift^®, Italy, purchased through PerkinElmer Life Sciences). We used a CombiPAL robotic sample processing workstation (CTC Analytics, Zwinger, Switzerland) to perform SPME fiber (coated with a 100-µm film of polydimethylsiloxane, Supelco, Bellefonte, PA, USA) extractions and injections into the GC (12). Software and data processing are described elsewhere (12).

Study design and statistical analysis

Bovine blood-based QC samples were used to study the stability of mercury species in blood because that is what was in use by our laboratory for this method at the beginning of the stability study. We think that this study will provide a good assessment of the stability of mercury species in human blood as well because after switching to human blood-based QC and using it for 3 years now, no differences in performance were seen.

We stored aliquots of the pooled materials (LB and HB aliquots) for 1 year at five different temperatures: −70, −20, 4, 23°C (room temperature) and 37°C. We chose to evaluate −70 and −20°C because these are recommended long-term storage temperatures and have been known to keep samples preserved for years (8, 11). We evaluated 4°C because it is recommended for short-term storage (analysis within 2–3 weeks). We evaluated 23°C (room temperature) to cover instances where samples are left out for lack of refrigerators or as a result of logistic sample processing that delays the time until final storage. An elevated temperature of 37°C was chosen for evaluation to cover samples that are collected in warm geographical regions and where proper cooling storage is unavailable. Various aliquots of the blood pools were analyzed after 1, 2, 4 and 6 weeks of storage and at 2, 4, 6, 8, 10 and 12 months of storage (10 time events: E₁ = 1 week, E₂ = 2 weeks, E₃ = 4 weeks, E₄ = 6 weeks, E₅ = 2 months, E₆= 4 months, E₇= 6 months, E₈= 8 months, E₉= 10 months and E₁₀= 1 year). At each time event/interval, three LB and HB aliquots were analyzed in duplicate per each temperature condition. After each aliquot was analyzed, it was not chosen for subsequent analysis.

Each analytical run was bracketed by the analysis of QCL and QCH pooled aliquots. QCL and QCH were analyzed to help differentiate between the effects of varied storage temperatures and time on long-term stability of LB and HB pooled material and changes resulting from drift and fluctuations in instrument measurements, variance in sample preparation procedure or any other experimental parameters.

A fixed-effects linear model was used to compare the mean concentrations of iHg, MeHg and EtHg for LB and HB pooled samples under 5 temperatures and 10 time events. Scenarios with significant P-values (<0.05) were further tested for pairwise comparisons. The Bonferroni correction was used to account for the multiple comparisons (13). The significant P-value (0.05) was divided by the number of all possible comparisons (four for the temperature; nine for the time event). Many P-values pointed toward statistically significant differences between different temperatures and time points. However, statistical significance does not necessarily indicate scientific importance, and other experimental factors were considered. We also performed CV analysis. The statistical analysis was conducted using SAS software (14).

Quality specifications

In routine patient samples analysis, as a part of our quality assurance and QC program, samples are bracketed with QC materials. If bracketing QC concentration falls outside of its 3 SD range, QC is marked as ‘fail’ and the sample analysis is repeated. [In practice, we use more sophisticated limits based on modified Westgard rules (15) using SAS software.] These criteria were adopted in this study for monitoring stability samples in addition to statistical analysis. From here on, we will refer to 3 SD range (Supplementary Table I) as ‘quality assurance’ limit. If mercury species concentrations in LB and HB stability samples fall outside of our established ‘quality assurance’ limits, this implies that mercury species are not stable, and if they fall within our ‘quality assurance’ limits, the analytes are considered stable.

Results

Experimental observations

At the end of 1 month, LB and HB aliquots had been analyzed at three separate time intervals (E₁–E₃: 1, 2 and 4 weeks). All mercury species concentrations for all aliquots were within our established quality assurance limits (Supplementary Tables III–V), with one exception. At 4 weeks, one HB aliquot stored at 37°C could not be analyzed because of the fibrin microclot formation and high viscosity of the matrix; the sample could not be pipetted.

Between time events E₄ (6 weeks) and E₁₀ (1 year), the analysis of LB and HB aliquots stored at temperatures of −70, −20, 4 and 23°C yielded mercury species concentration within established quality assurance limits (Supplementary Tables III–V). However, two HB aliquots stored at room temperature solidified at time events 8 and 10 months (Supplementary Figure 1). The solidification resulted from evaporation and has been observed by others (11). Note the referenced study was done with heparinized blood. At 37°C, mercury species concentrations and the appearance of the blood matrix displayed signs of instability. We noted that at 2 months, one LB aliquot could not be analyzed because of high viscosity and one HB aliquot solidified (Supplementary Figure 1). Furthermore, for both 2 and 4 months, iHg and MeHg concentrations in LB aliquots were outside of accepted quality assurance limits (Supplementary Tables III and IV). At 6 months, two out of three LB aliquots were too viscous for analysis, and EtHg concentration in all HB samples was outside of quality assurance limit (Supplementary Table V). At 8 months, all HB and LB aliquots were too viscous for analysis, and we ended further analysis at this temperature. To understand the trends in mercury species concentrations as a function of storage time and temperature, we conducted statistical analysis of the data.

Fixed-effects linear model

We used a fixed-effects linear model to compare the mean concentrations of iHg, MeHg and EtHg for LB and HB aliquots at the 5 temperatures and 10 time events. Three LB and three HB aliquots were included under each temperature/time event, and each aliquot was analyzed in duplicate. Figure 1 presents model-based mean estimate concentrations and the 95% confidence intervals for temperature and time, respectively (concentrations can be seen in Supplementary Tables VI and VII). Tables I and II display calculated P-values representing overall significance of the equality of 5 temperatures and 10 time points for iHg, MeHg and EtHg concentrations in LB and HB aliquots. We found a statistically significant difference among the five temperatures for all three mercury species in the LB pool (P> 0.05). See the italicized values in Table I. There is no statistically significant difference for HB samples. P-values for comparing the time events showed statistical significance for all analytes in both pools (italicized values in Table II).

Figure 1.

Mean concentrations for LB and HB aliquots, based on the fixed-effects linear model. Inorganic—iHg, methyl—MeHg and ethyl—EtHg mercury. Five different temperatures (−70, −20, 4, 23 and 37°C) are shown in (a) for the LB aliquots and in (b) for HB aliquots. Ten time events (E₁= 1 week, E₂= 2 weeks, E₃= 4 weeks, E₄= 6 weeks, E₅= 2 months, E₆= 4 months, E₇= 6 months, E₈= 8 months, E₉ = 10 months and E₁₀= 1 year) are shown in (c) for the LB aliquots and (d) for the HB aliquots. Each point is an average of results from three aliquots. Error bars represent the 95% confidence intervals.

Table I.

Calculated Fixed-Effects Linear Model and Step-Down Pairwise Based P-values (Overall Significance P> 0.05, P > 0.0125 for Pairwise) for Samples from Blood Pools With Low Concentrations of Mercury (LB) and High Concentrations (HB) Stored at Different Temperature Conditions

Temperature (°C)	LB Pool				HB Pool
	iHg	MeHg	EtHg	EtHg → iHg	iHg	MeHg	EtHg	EtHg → iHg
	Fixed-effects linear model P-values
Overall	0.006	0.003	0.006	0.002	0.175	0.821	0.455	0.034
	Step-down pairwise comparison test P-values (standard: −70°C)
−20	0.080	0.095	0.057	0.521	0.680	0.833	0.679	0.720
4	0.609	0.194	0.396	0.420	0.479	0.366	0.925	0.549
23	0.437	0.188	0.352	0.135	0.103	0.812	0.434	0.495
37	<0.0001	<0.0001	<0.0001	0.003	0.891	0.658	0.244	0.009

Inorganic—iHg, methyl—MeHg and ethyl—EtHg mercury. EtHg → iHg represents the conversion percentage of ethyl to inorganic mercury. Italicized values indicate statistical significance.

Table II.

Calculated Fixed-Effects Linear Model and Step-Down Pairwise Based P-Values (Overall Significance P > 0.05, P > 0.0056 for Pairwise) for Samples from Blood Pools with Low Concentrations of Mercury (LB) and High Concentrations (HB) for Different Time Events (E₁= 1 Week, E₂= 2 Weeks, E₃= 4 Weeks, E₄= 6 Weeks, E₅= 2 Months, E₆= 4 Months, E₇= 6 Months, E₈= 8 Months, E₉ = 10 Months, and E₁₀= 1 Year)

Time	LB pool				HB pool
	iHg	MeHg	EtHg	EtHg → iHg	iHg	MeHg	EtHg	EtHg → iHg
	Fixed-effects linear model, P-values
Overall	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
	Step-down pairwise comparison test, P-values (standard: 1 week, E1)
E2	0.301	0.891	0.972	0.001	0.097	0.468	0.467	0.654
E3	0.870	<0.0001	0.851	0.040	<0.0001	<0.0001	0.154	0.063
E4	0.669	<0.0001	0.329	0.249	0.227	<0.0001	0.220	0.659
E5	0.001	<0.0001	0.297	<0.0001	<0.0001	<0.0001	0.811	0.311
E6	0.002	<0.0001	0.012	0.681	<0.0001	<0.0001	0.002	0.011
E7	0.008	<0.0001	0.324	0.214	<0.0001	<0.0001	0.271	0.001
E8	0.009	<0.0001	0.979	0.976	<0.0001	<0.0001	0.226	0.001
E9	0.002	0.028	0.000	0.008	<0.0001	<0.0001	0.001	0.211
E10	0.015	0.013	0.026	0.003	<0.0001	0.001	0.003	0.001

Inorganic—iHg, methyl—MeHg, and ethyl—EtHg mercury. EtHg → iHg represents the conversion percentage of ethyl to inorganic mercury. Italicized values indicate statistical significance.

Step-down pairwise comparison test

For P < 0.05 (calculated with a fixed-effects linear model), we used a step-down pairwise comparison test (16). The temperature −70°C was treated as the ‘standard’ that all other temperature settings were compared against. In this step-down test, our new significance level was 0.05/4 = 0.00125. Figure 1a and b displays mean concentrations of mercury species calculated by the fixed-effects linear model as a function of temperature (Supplementary Table VI, each temperature represents averaged data for all events). The only statistically significant difference was found between −70 and 37°C for all mercury species in the LB pool (italicized values in Table I). Visual analysis of the data (Figure 1a) for all three mercury species in the LB aliquots stored at 37°C showed slightly higher concentrations when compared with data from other temperature settings. This corresponds with the P-values obtained showing a statistically significant difference. Alternatively, mercury species concentrations in HB aliquots at different temperatures did not display any clear trends (Figure 1b), which coincide with the P-values obtained that show no statistically significant difference.

The same step-down analysis was conducted with time. A storage time of 1 week (E₁) was treated as the ‘standard’ and the following events were compared with it. Our new significance level was 0.05/9 = 0.0056. For time events, there were many P-values demonstrating a statistically significant difference (italicized values in Table II). Figure 1c and d displays concentrations of mercury species calculated by the fixed-effects linear model as a function of time events over a 1-year period (Supplementary Table VII, each event represents averaged data for all temperatures). To further understand the concentration trends of LB and HB long-term stability aliquots as a function of time, we examined analyte recovery trends of bracketing QC aliquots (QCL and QCH) over the same time period to determine whether or not instrumental drift/fluctuations, variance in sample preparation procedure or any other experimental parameters are possibly influencing the results (Figure 2). Starting at 6 months (E₇) in Figure 2, we began using a new human whole blood pooled material as bracketing QC samples (QCL2 and QCH2) with concentrations and limits presented in Supplementary Table II. The supply of material from previous low and high pools (except that reserved for the stability study) was exhausted. As we closely examined concentration trends as a function of time events (Figures 1c and d as well as 2a and b), some similarities were noted. For example, similar concentration trends for iHg in HB and QCH aliquots for events E₂ through E₅ (Figures 1d and 2b) and both a concentration increase at time event E₆ and a decline in concentrations at E₈ for EtHg in LB and QCL/QCL2 samples (Figures 1c and 2a) were observed. It is likely that the time effects on mercury species concentration changes were influenced by variations in instrument response over time or other experimental parameters.

Figure 2.

Mean concentrations for QC samples (a) QCL/QCL2 and (b) QCH/QCH2 used for bracketing long-term stability samples over the period of 10 time events (E₁ = 1 week, E₂ = 2 weeks, E₃= 4 weeks, E₄= 6 weeks, E₅= 2 months, E₆= 4 months, E₇= 6 months, E₈= 8 months, E₉ = 10 months and E₁₀= 1 year). Inorganic—iHg, methyl—MeHg and ethyl—EtHg mercury. Each point represents an average of two low QC and two high QC aliquots analyzed at the beginning and end of each analytical run. *Starting at event 7, we began using new bracketing QC pools—QCL2 and QCH2 (Supplementary Table II).

Coefficient of variation evaluation

In addition to the fixed-effects linear model scheme, we applied the CV to evaluate the stability of iHg, MeHg and EtHg under different time and temperature conditions. The mean, standard deviation and CV under each time/temperature combination were calculated based on the data from three aliquots from the LB and HB pools that were analyzed in duplicate (Supplementary Tables III–V). Table III summarizes the results. The mean represents the average mercury species concentrations of all time points per temperature setting. The standard deviation represents the average of standard deviations for all time points per temperature setting. Within CV is defined as the average CV of all time points under each temperature condition. Between CV is calculated based solely on the mean of all time points under each condition (the equations and example for calculating within and between CVs can be found in Supplementary Table VIII). In our laboratory, we have a general rule for acceptable variance of <10% for CVs when it comes to accessing analyte stability; this guideline is based on our analytical experience in the field. The within and between CVs for LB and HB pooled aliquots were <10%, with the majority being <5% (Table III). The only value we found close to 10% was for iHg in the LB pool at 37°C. The within CV was 5.45% and the between CV was 9.55%.

Table III.

CV Evaluation of Mercury Species as a Function of Different Temperature Settings and Time Events from Blood Pools with Low Concentrations of Mercury (LB) and High Concentrations (HB)

Mercury species	Temperature (°C)	LB pool				HB pool
		Mean (µg/L)	SD	Within CV %	Between CV %	Mean (µg/L)	SD	Within CV %	Between CV %
Inorganic (iHg)	−70	1.10	0.05	4.27	3.19	4.45	0.28	6.24	4.75
	−20	1.13	0.05	4.62	4.61	4.42	0.22	4.99	5.88
	4	1.11	0.04	3.87	2.90	4.41	0.23	5.12	5.91
	23	1.11	0.03	2.84	5.31	4.56	0.19	4.24	5.52
	37	1.15	0.06	5.45	9.55	4.42	0.17	3.85	7.16
Methyl (MeHg)	−70	1.09	0.03	2.86	2.37	4.12	0.14	3.38	3.70
	−20	1.11	0.03	2.68	2.72	4.12	0.11	2.70	2.83
	4	1.10	0.03	2.54	3.66	4.09	0.12	2.95	3.09
	23	1.10	0.02	2.04	2.22	4.13	0.11	2.60	4.30
	37	1.13	0.04	3.01	6.42	4.09	0.10	2.38	2.96
Ethyl (EtHg)	−70	1.36	0.05	3.39	2.45	5.11	0.16	3.20	4.44
	−20	1.39	0.05	3.85	2.41	5.13	0.18	3.47	4.36
	4	1.38	0.05	3.85	3.29	5.11	0.17	3.41	2.52
	23	1.38	0.04	2.90	3.17	5.15	0.14	2.78	4.01
	37	1.42	0.04	2.68	4.50	5.01	0.20	4.07	5.16

The mean and standard deviation (SD) are averages of all the time points for the particular temperature setting. Italicized value indicates signs of instability (close to 10% mark).

Mercury species interconversions

During mercury speciation analysis, including our method (12), spontaneous in vitro mercury transformation reactions take place (5–7), especially the dealkylation of organomercury compounds (17). Thus, we use a TSID technique that quantitates the rate of species transformations so that we can apply the proper corrections. To use TSID, the sample preparation has to involve the addition of spike solution (isotopically labeled mercury species) to blood samples (see the ‘Sample preparation’ section). In our mercury speciation method, EtHg to iHg transformation is by far the largest in comparison with other interspecies transformations (others <5%, EtHg → iHg ∼ 35–55%) (12). We used fixed-effects linear model to compare EtHg with iHg species transformation percentages for LB and HB pools stored under 5 different temperatures and 10 time points. Of note, EtHg to iHg conversions that we are able to quantify take place as the samples undergo sample preparation procedures and analysis by GC–ICP-DRC-MS (after isotopically enriched mercury spike is added to the blood sample). If any mercury species conversions do take place during storage, this would be reflected in the incorrect mercury species concentrations of the analyzed samples. We were interested to see if there is any difference in conversion percentages during sample preparation and analysis after storing samples at different temperatures during a 1-year period (Supplementary Table VIX). Tables I and II (column EtHg → iHg) present the overall significance and step-down pairwise P-values for comparison of EtHg to iHg transformation percentages for temperature and time event changes, respectively. For different temperature settings, the only statistically significant difference of EtHg → iHg conversion percentages was found between −70 and 37°C for both the LB and the HB pools (Table I). We noted many statistically significant differences in conversion percentages of EtHg → iHg as a function of different time points (Table II).

Discussion

We compared the mean concentrations of iHg, MeHg and EtHg in LB and HB aliquots for 10 time points. There was overall statistical significance in concentrations for all three mercury species, and the step-down test further confirmed a statistically significant difference in mercury species concentration over time. However, QC samples (QCL, QCH and QCL2, QCH2) used for bracketing long-term stability LB and HB aliquots over the period of 1 year showed similar concentration trends as stability samples. Therefore, we believe that slight changes in instrument response or other experimental parameters are influencing the trends in concentrations of stability samples over time. In this case, statistical analysis by itself does not provide a full picture, thus supporting the necessity to monitor concentrations of independent QC materials in these types of studies.

Next, we examined the effect of temperature on stability of mercury species. At temperatures of −70, −20 and 4°C, all mercury species concentrations fall within established quality assurance limits (3 SD), within and between CV % <10, and we found no statistically significant evidence of mercury species instability. Traditionally, 4°C has been thought of as a short-term stability temperature, but this work suggests that samples remain stable for a period of at least 1 year. At room temperature, two aliquots solidified at 8 and 10 months as a result of evaporation. Otherwise, the concentrations of all non-solidified samples were within established quality assurance limits for the LB and HB pools. We found no statistical difference in concentrations in the step-down test and within/between CV <10%. Therefore, we conclude that the mercury species at 23°C in bovine blood are stable for at least 1 year. However, more air-tight storage containers are required to prevent evaporation at room temperature. Based on visual observations, pooled samples at 37°C started becoming viscous at 1 month, making it challenging to precisely pipette samples for analysis. P-values for step-down analysis pointed toward significant differences between mercury species concentration means at −70 and 37°C. The within and between CV % was under 10 but higher when compared with others. Additionally, some of the samples fall outside of quality assurance limit (Supplementary Tables III–V). Overall, experimental and statistical analyses suggest that mercury species in whole blood stored at 37°C cannot be quantified beyond 2 weeks of storage.

As we examined average conversion percentages of mercury species, EtHg to iHg conversions varied 47.6–52.2% as a function of time and 47.7–50.3% as a function of temperature (Supplementary Table IX). The variation in this study is insignificant. During routine sample analysis, we see conversion percentages varying anywhere from 35 to 55% on average (12).

Conclusion

In this study, we described the first long-term stability study of iHg, MeHg and EtHg species in bovine whole blood (EDTA disodium salt used as an anticoagulant). We focused on two factors: the temperature at which samples are stored and the duration of storage. The statistical analysis was useful in comparing different storage temperatures and durations, but it was only one component of this evaluation. Visual and experimental observations, along with knowledge of the method's capabilities and application, were essential in reaching our scientific conclusions. We determined that the mercury species were stable in LB and HB pools over a 1-year period at temperatures of −70, −20, 4 and 23°C. Blood samples stored beyond 2 weeks at 37°C were unstable, making it impossible to correctly quantify iHg, MeHg and EtHg species in those samples. This long-term stability study will continue for an additional 4 years, which will provide further data on the stability of iHg, MeHg and EtHg in whole blood.

Supplementary data

Supplementary data are available at Journal of Analytical Toxicology online.

Conflict of interest statement

The findings and conclusions in this study are those of the authors and do not necessarily represent the views of the US Department of Health and Human Services, or the US Centers for Disease Control and Prevention. The use of trade names and commercial sources is for identification only and does not constitute endorsement by the US Department of Health and Human Services, or the US Centers for Disease Control and Prevention.