Considerations for Stability of Environmental Samples in Storage for Long-Term Studies

It is often advantageous to store collected environmental samples for future retrospective analyses. However, information about sample stability is necessary to determine if there will be analyte loss or gain or degradation under the specified storage conditions and storage period. Failure to evaluate stability could result in inaccurate results and biased exposure assessments. As part of the National Children’s Study pilot, we considered which types of environmental samples could be stored for extended periods of time. We conducted an extensive literature review and considered the conduct of long-term stability studies for environmental samples. We present our findings and experience below as guidance for consideration by the environmental research community.


Introduction
For long-term environmental studies, such as for prospective epidemiology studies, it is often advantageous to store collected environmental samples for future retrospective analyses. Traditionally, stored environmental samples have included human tissue and fluids, animal and plant tissues, soils, sediments, and ice cores [1]. Such samples can be used to evaluate results of government policies, health of an animal population, or temporal trends in ecosystems or exposures [2]. In longitudinal studies, this also permits spreading costs of analyses over time -an important consideration as analysis for environmental contaminants can be expensive. Additionally, it provides more flexibility to analyze subsets of samples in nested case-control studies for specific health outcomes or for inclusion of new target analytes or analysis methods. For instance, concern about the presence of pesticides, pharmaceuticals and personal care products (PPCPs) has heightened over the past several years due to the presence of these chemicals in wastewater, groundwater, surface water and drinking water [3][4][5][6][7][8][9][10].
Information about sample stability in long-term studies is critical to determine, if there will be analyte loss or gain or degradation under specified storage conditions and storage period. Failure to evaluate stability could result in inaccurate results and biased exposure assessments due to partial or complete analyte decomposition, chemical transformation, or loss/gain.

Literature survey
The literature was surveyed through 2020 in PubMed, Web of Science and other commercial sources to identify published sample preparation and storage information of contaminants in any environmental matrix. Stability of analytes in analytical standards and in stored biospecimens were not included in this review. Table 1 summarizes the 63 peer-reviewed articles and 8 reports and book chapters selected after initial screening that discussed stability of anthropogenic chemical compounds in environmental matrices. The most common analytes studied were pesticides and trace elements. Various environmental matrices were considered, including air, water, soil/sediment, dust, and food (plant and animal tissue).
Two major scopes for stability studies were identified: (1) long-term stability in storage (retained samples, environmental specimen banking), and (2) short-term stability of samples in transport conditions. There was no generally accepted procedure for performing a stability study. Stability samples were sometimes prepared and stored all at the same time with some samples removed for analysis at various time periods. Alternatively, samples were prepared and stored at different time points and analyzed all at the same time. The former procedure is subject to dayto-day and longer term variability and changes in analysis procedures; the latter is subject to variability and changes in the preparation and analysis procedures when the time points are far apart.
Another sample stability test procedure, generally used for reference standards and sample transport, is isochronous testing, where samples are prepared at the same time and stored in conditions that offer the best stability, or the least degradation, until they are moved to the storage conditions to be tested for the specified time periods [43]. At the end of the testing period, each sample set is moved back to the original storage until all samples are removed for analysis at the same time. This procedure avoids the challenges of varying sample preparation and analysis conditions.
The determination of stability of an analyte in a sample is measured as the ratio of the concentration measured at time point t, compared to that measured at time point 0. The sample size required to make a decision with specified confidence for each analyte depends on the precision of the analysis method at each concentration level and the amount of change at which samples are considered to have degraded. The change in concentration over time that indicates significant degradation is often set to 5 or 10 percent, particularly for biological samples. This level is likely too stringent for environmental samples where, for instance, sample processing may itself introduce a reduction of 5 to 10 percent and, therefore, a decision criterion of 20 percent change is most often used. Despite the widespread practice and numerous and strong benefits of long-term environmental sample storage, we found very little documentation to support preservation of analytes during long-term storage. The findings of the literature review were disappointing in that many questions went mostly unanswered for many sample matrices and analytes of interest, in particular:

• What are acceptable long term storage times for various sample types?
Storage times in the identified studies varied from a few days up to five years. One study extended the storage study to 12 years for air samples [82].

• What are acceptable long term storage conditions for various samples types?
Storage conditions in the identified studies included ambient or room temperature (generally around 20°C), refrigerator (4°C), freezer (−20°C), and/or cryogenic temperatures (−60-80°C). Environmental specimen banks store samples for 50-100 years in cryogenic conditions [79,80,87], but do not document the rationale for sample stability under these conditions and time periods. For longer-term storage, some researchers have extracted water contaminants through solid phase extraction (SPE) cartridges and stored frozen for up to a year. Interestingly, colder storage does not always equate to less analyte degradation, as commonly assumed [74].
We also reviewed standard analysis methods for the samples and analytes of interest in the NCS. We found that standard methods are generally focused on regulatory compliance and do not consider long term storage. For example, for ambient vapor and gas sampling, updated storage studies of SUMMA and whole air canisters are needed (US EPA method allows storage of only 30days).

NCS stability study experience
The overall objective of the planned NCS storage studies was to determine the stability of target analytes in NCS environmental samples stored in collection containers at specific storage conditions, including reconstitution after thawing and re-freezing of samples, as applicable. For instance, the objective of the tap water study was to evaluate the effect of prolonged storage at −20°C on the stability of the pesticide and pharmaceutical target analytes spiked in analyte-free water. The results of this study were to be used to extrapolate the observed changes in the target analyte concentrations to the stability of those compounds in the NCS study samples.

Planning
The stability study plans included stability samples similar to environmental field samples, sample processing, analytes or classes of analytes of interest, and analysis methods. The in-depth literature review addressed the prevalence of potential analytes and classes of analytes of interest in an indoor environment, concentrations measured, sampling and analytical methods, reaction or degradation products in these matrices, and storage conditions and stabilities. In addition, experts, agencies and groups with relevant experience, e.g., National Institute of Standards and Technology (NIST), the U.S. Environmental Protection Agency (EPA), the Children's Center Study, National Human Exposure Assessment Survey (NHEXAS), Children's Total Exposure to Persistent Pesticides and Other Persistent Organic Pollutants (CTEPP), and New York State Department of Health/Love Canal project researchers, were contacted for information. This information was used to develop the plans for the preparation and analysis of the stability samples and archiving of both the field and stability samples.
Storage effects due to freeze-thaw cycles, sample pretreatment, and reconstitution were also considered. For example, freeze-thaw cycles can affect the volatile and semi-volatile compound concentrations on a filter or in bulk dust. The pretreatment of samples before storage is performed when metabolism or reaction after collection is known to be of concern. For example, water samples for metals analysis are usually acidified to pH < 2 [86], and dust samples are sometimes irradiated to stop microbial growth and metabolism.

Acquisition of samples
Stability study samples with known concentrations of analytes were planned to be acquired in similar matrices to the NCS samples, i.e., particulate filters, wipes, and vacuum filters. These samples were to include (1) newly-prepared samples that have been spiked with known concentrations of analytes, and (2) pre-tested 'realworld' samples.
Surface dust wipe and air diffusive samples could be made by spiking study sampling media and tap water samples by spiking analyte-free water with known amounts of the target analytes. These kind of samples can be purchased from reference material producers, proficiency testing sample providers, and/or accredited laboratories. To allow for a possible difference in reaction/degradation rate of analytes with their concentration in matrix, each matrix was to be spiked at environmentally relevant concentration levels, based on the literature review, and at concentrations that can be most reliably quantified by the analytical method employed, preferably at concentrations near the midpoint of the instrumental calibration range. Another benefit of this approach was that the concentration of the stability samples could be set to match the concentration used in the laboratory QC samples for increased comparability [7]. Some samples cannot be spiked onto matrices in a manner representative of the corresponding environmental samples, or it would be very difficult to do so. There may be a difference in behavior of analytes in environmental samples and in similar spiked matrices due to pH, microbial enzymatic activity of the sample matrix, physicochemical properties of the analytes themselves and their reactions with the matrix, and other analytes present in the environment. For example, for vacuum filter and deposition plate stability samples, dust from non-participant homes could be obtained, homogenized, and aliquoted before storage.
Each spiked and real-world sample should represent the environmental sample matrix as close as possible to account for inter-analyte and analyte-matrix interactions. If it is not possible to include all analytes of interest in one matrix sample, they can be classified into groups by their chemical characteristics, and include representative analytes of each group deemed most unstable. Because sample homogeneity is very important, for example, surface wipe and dust samples should be aliquoted before spiking and the whole sample used for analysis. DOI: http://dx.doi.org /10.5772/intechopen.97445 All samples were planned to be collected on exactly the same collection media using the same samplers and stored in the same containers as used in the field study and one manufacturer's lot of each media and containers used, whenever possible. Blank media was to be stored together with the samples for each sample type and analysis time point and unused sample media provided to the analysis laboratories, as required. There should be additional samples stored as back-up for sample loss due to, for instance, filter or container breakage, and for repeat analyses. Additional samples should also be stored to account for the possibility that laboratories or analysis methods may change during the study and simultaneous measurements are necessitated. For each sample type, to the extent possible, samples should be analyzed by one laboratory in order to minimize inter-laboratory bias. For NCS, standard laboratory methods were specified for each analysis, identical to those planned for the field sample analysis.

Number of samples and analysis time points
Samples are usually analyzed before storage (time point 0) to determine analysis method recovery, concentrations at the beginning of the study (time 0), and to verify the homogeneity of the aliquots. This will take into account any analyte losses during pre-analysis, such as extraction and processing. The frequency of future time points and the number of samples required for statistical significance at each point should be planned carefully before start of the study.
The frequency of time points is chosen to detect the instability of samples and analytes, such as VOC, as early as possible. It is assumed that in most cases, particularly for the organic analytes, the rate of degradation would be highest at the beginning, that is, first order kinetics; this corresponds to a linear change in the log transformed measurement over time [65,82,88]. In the NCS stability plans, at each test time t, it is tested statistically with a t-test if a specified percentile of the concentration measurements (the target percentile) will be greater than the specified limit of 80 percent at some future time, assuming stability decreases exponentially (or the log transformed stability decreases linearly) with time. For calculating sample size, it was assumed that the log transformed ratios have a normal distribution and the mean log transformed recovery decreases linearly over time.
As the study proceeds there will be accumulating data from prior test times that can also be used to get a more precise estimate of the slope or fit a nonlinear trend. Calculating samples size when using all prior data is more complicated than calculating sample size for a t-test. The calculations depend on the analysis that might be performed. The number of samples depends on: the alpha and target power for a slope of zero; the standard deviation of the measurement error; the standard deviation of the slope factor; the target percentile (80%); and, the timing of the time points. The test-wise alpha and power for no degradation were set to achieve an overall mean alpha of 5% and an overall power of 95%.
The standard deviation of the measurement error can be estimated from the variation among replicate samples or aliquots, typically measured as a relative standard deviation or, equivalently, a coefficient of variation (CV). The sample sizes were calculated using the error term calculated from the estimated coefficient of variation of the measurement methods. This model was assumed to be adequate for calculating sample sizes and for a variety of sample collection designs, analyses, and distributional or statistical model assumptions. The planned frequency of time points for the NCS was 0, 1, 2, 3,5,8,13,21,32,48,72,108,160, and 240months, altogether 14 time points (or test times). The numbers of planned stability samples varied by sample type and were mostly around 300. The variation was due to the variation of the CV of different analysis methods.

Conclusions
Collection and analysis of environmental samples for various analytes in largescale or longitudinal cohort studies is useful to investigate the contribution of the environment on health outcomes. Storing samples for long periods of time is necessary but expensive in these kinds of studies. Information about preserving sample quality and analyte stability in stored environmental samples is limited. Design and implementation of sample stability studies like those described here are recommended to ensure that samples are stored properly and generate reliable analytical results when required. Publication of environmental sample stability study results will likely provide valuable information to investigators who design and implement large-scale longitudinal environmental health studies.

Considerations for the environmental research community
Storage of environmental samples is an important component of large-scale and prospective studies. The environmental research community must address and document the answers to the questions above by conducting and providing data on stability of samples stored over long periods.
• If more stability data are available than presented here, these data should be provided in technical guidelines, study manuals, or published papers.
• Conduct and report on environmental sample storage stability studies, especially for new analytes of interest.
• In the interim, based on our research, a sample stability program should be integrated with sample collection as a part of the quality assurance procedures for the study.