Evaluating risk, exposure, and detection capabilities for chemical threats in water

Abstract

The unexpected release of chemicals into the environment requires estimation of human health risks, followed by risk management decisions. When environmental concentrations of toxicants are associated with adverse health risks, the limit for analytical measurement needs to be at or below the risk threshold. The aim of this study was to assess chemical contaminants that have the potential to produce acute adverse human health impacts following oral consumption of contaminated drinking water. The U.S. Environmental Protection Agency’s (EPA) Candidate Contaminant List, version 4 (CCL4) and EPA’s Selected Analytical Methods (SAM) document were screened to identify 24 chemicals that exist as a solid or liquid at room temperature, with acute oral LD₅₀ (lethal dose in 50% of the test population) values < 500 mg/kg-d and water solubility > 500 mg/L at ambient temperature. While these screening criteria were used to identify prioritized needs for targeted research, it does not imply that other chemicals on the CCL4 and SAM lists are not issues in acute and chronic exposures. Of these 24 most toxic and most soluble chemicals, this evaluation identified 6 chemicals (2-chlorovinylarsonous acid, lewisite, N-nitrosopyrrolidine, N-nitrosodiethylamine, 3-hydroxycarbofuran, and triethylamine) lacking either sufficient toxicity value information or analytical sensitivity required to detect at levels protective against adverse effects in adults for acute exposures. This assessment provides an approach for gap identification and highlights research needs related to water contamination incident involving these six priority chemicals.

Introduction

Water systems are vulnerable to chemical contamination through several avenues, such as annual industrial releases, industrial accidents, natural disasters, and terrorist attacks (Erickson et al. 2019). In 2019, 201 million pounds of toxic chemicals were released into U.S. surface water through industrial facility releases alone (U.S. EPA 2019a). That same year, flooding following Hurricane Harvey in and around Houston, TX, produced runoff from industrial and superfund sites to flow into waterways (Kiaghadi and Rifai 2019). Water quality changes and elevated levels of certain metals were subsequently noted in nearby watersheds (Kiaghadi and Rifai 2019). In 2014, an unintentional contamination incident occurred when an industrial storage tank leaked over 10,000 gallons of an industrial solvent mixture (composed of 68–89% 4-methylcyclohexanemethanol) into West Virginia’s Elk River, the region’s only drinking water source (Whelton et al. 2015). The contaminant entered the drinking water treatment plant and distribution system which supplied water for 15% of the State’s population, producing significant disruption of water service and adverse impact on public health (Whelton et al. 2015). Several hundred people were treated for health issues due to exposure to the contaminated water, a federal disaster was declared, and “Do Not Use” orders remained in place for 10 days (Whelton et al. 2015). Over 2200 miles of water mains, 107 storage tanks, and 120 booster stations needed to be flushed to remove contaminated water (Whelton et al. 2015). An intentional contamination incident occurred in 1980 when chlordane was deliberately introduced into a water distribution system that served over 10,500 people in Pittsburgh, PA (Janke, Tryby, and Clark 2014; Welter et al. 2009). In addition, potential contamination of military drinking water supplies with legacy compounds polyaromatic hydrocarbons and chemical warfare agents (CWAs) remains a concern because of past production and use during World War I and the potential for future threats during times of war or other conflicts (Lawrence Livermore National Laboratories (LLNL) 1990; National Research Council (NRC) 1995; DiGuilio and Clark 2015).

In the United States, systems supplying water are typically composed of (1) a surface and/or groundwater source; (2) raw water main to transport the source water to a treatment plant; (3) treatment facility; and (4) distribution system (transportation pipes, local distribution pipes, storage tanks, valves, and hydrants) (Janke, Tryby, and Clark 2014; Winston and Leventhal 2008; World Health Organiazatin (WHO) 2014). Any one of these components may become contaminated, but water reservoirs such as wells, pump stations, and distribution system including water towers, tanks, and fire hydrants are especially vulnerable locations for intentional contamination via contaminant release or injection (Hickman 1999; Janke, Tryby, and Clark 2014). Contamination impacts might be quite vast because the components of a water supply system are interconnected. Contamination of components closer to the point of use by end-users might potentially result in more damage-control, injury, and/or illnesses compared to contamination at a water source due to a decreased opportunity for dilution, decay, partial removal, and disinfection of the chemical contaminant prior to end-user exposures (Janke, Tryby, and Clark 2014; World Health Organiazatin (WHO) 2014). Water contamination detection involves active monitoring, observing noticeable changes in water appearance, taste, or smell, or reports of individuals seeking treatment for adverse health effects (Whelton et al. 2015; World Health Organiazatin (WHO) 2014). Health effects, due to oral exposure to chemical contaminants in source and drinking water, might occur over both short-term (days to weeks) and long-term (months to years) durations (Janke, Tryby, and Clark 2014; Villanueva et al. 2014; Whelton et al. 2015).

Regardless of how an incident is discovered, it is critical to understand the contaminant concentration, potential exposure pathways, and associated adverse human health risks. Several studies’ investigators focused on factors that impact public health exposure outcomes associated with such events. These studies noted that exposure dose, duration, chemical health effects, distribution system, the point at which the contaminant enters the distribution system, agent physicochemical fate (i.e., solubility, hydrolysis, or dilution), and persistence in the distribution system are all important factors to consider (Davis and Janke 2011; Davis, Janke, and Magnuson 2014; Davis, Janke, and Taxon 2010; Janke, Tryby, and Clark 2014). Understanding how these factors affect exposures might inform decisions related to water usage such as do-not-drink orders, evacuation of impacted areas, disruption of water service due to shutting down the service, decontamination measures such as flushing or use of an alternative source water, and potential need for long-term monitoring even after the water system has been returned to normal operation (Janke, Tryby, and Clark 2014; Welter et al. 2009; World Health Organiazatin (WHO) 2014).

Multiple exposure or risk values might be available for reference following an unanticipated chemical release because of potential for multiple human exposure scenarios and the extent to which exposure to some chemicals might be controlled or reduced. For that reason, a comparison across different risk value systems need not be based solely upon the individual numerical values but needs to consider the nature of the exposure and likelihood of a biological response (which is a function of both the exposure concentration (or dose) and exposure duration). The general steps involved in risk assessment are outlined in Figure 1. The nature of a biological response might change with dose and exposure duration, especially in cases where the effects from the chemical exposure may alter as exposure duration lengthens and chemical body-burden increases. Therefore, both dose and exposure duration need to be considered when estimating risk and careful consideration needs to be given for the risk values developed for each available exposure duration (1 d, 10 d, sub-chronic, or lifetime) when selecting the appropriate guideline for the exposure scenario.

Figure 1.

Risk vs. exposure. Side-by-side comparison of the step-by-step process for assessing risk and exposure which lead to the development of distinct drinking water concentrations for health-based risk and analytical detection limits.

The detection of hazardous chemicals in the environment is the primary means of protection against adverse health effects for a number of chemicals. Some water contaminants have monitoring and reporting requirements to prevent exposures from exceeding guidance values. When considering the health protective mandate given to (or assumed by) governments, levels are often designed to be health-protective and conservative and might include the assumption of a continuous exposure over a lifetime. Despite the additional resources and time needed to develop high-sensitivity analytical and monitoring methods, such advances are necessary in order to protect broad segments of the population. Given that both dose and exposure duration contribute to health impacts, parallel consideration also needs to be given to analytical sensitivity and resource requirements for chemical detection systems, especially during the early phase of response to a chemical-specific incident. The general steps involved in detecting and measuring a chemical in water are outlined in Figure 1. Careful consideration is needed to determine the effectiveness of a specific method to detect concentrations of a hazardous chemical in the environment. Having the ability to accurately detect contaminants in water is important, but can be difficult when encountering low concentrations, contaminant mixtures, or the presence of potential interferents in the environment (Villanueva et al. 2014).

The sheer number of chemicals in commerce makes it difficult to ascertain which chemicals pose the highest risk for potential human exposure via contamination through potable water systems following industrial, mining, agricultural, and transportation accidents, intentional contamination acts, or natural disasters. The purpose of this evaluation was to:

• Identify chemicals that might contaminate a water system and produce significant adverse health impacts

• Access risk-based health values associated with those chemicals

• Determine whether there is a capability to detect the chemical contaminants at relevant risk-based levels; and

• Identify potential knowledge gaps associated with the health impacts for any of the chemicals.

Chemicals were identified and separated based upon the following categories:

1. neither risk values nor analytical detection capabilities exist,

2. no risk values exist, but analytical detection capabilities exist,

3. risk values exist, but ability to detect the chemical at levels that reach the risk values do not exist, and

4. risk values and analytical capabilities exist.

The four categories are intended to evaluate chemicals based upon their detection capabilities within an environmental matrix and availability of risk/exposure guidelines associated with the selected chemicals. Several scenarios were selected for evaluation.

• Scenario 1 represents a priority chemical that requires the establishment of both analytical detection capabilities and risk values.

• Scenario 2 represents a chemical for which analytical detection capabilities exist and may inform low doses to be considered for application in toxicity testing.

• Scenario 3 represents risk values that may be used to identify the ideal minimum analytical detection limits for application to contaminated environmental media.

• Finally, scenario 4 represents a chemical for which the analytical detection capability can be compared to risk values specific to the environmental media in cases where a health-protective analytical method with an adequate detection capability exists or where the health risk, health advisory, or cancer risk might be used to prioritize analytical method development for multi-lab testing or refinement

Priority chemicals were previously identified for risk assessment (Thompson et al. 2007) and for analytical method development and refinement (Zulkifli, Rahim, and Lau 2018) in support of risk management decision-making following a chemical release in drinking water. However, chemical prioritization varies depending upon the interest of the groups that compile and categorize them. Each interest group or federal agency may have their own chemical prioritization scheme for research prioritization. This investigation may identify research gaps, which if addressed might potentially inform regulations and/or enhance the ability to prepare for a potential emergency involving chemical contamination. Soluble chemicals that are acutely toxic and pose a significant and acute threat to the general population via oral consumption of drinking water following a contamination incident were the focus of the present review. This assessment might be employed to guide future research and identify gaps associated with analytical detection limit capabilities and/or establishment of levels protective of human health for chemicals of interest to emergency response.

Methods

Chemical selection and prioritization

For this effort, chemical identification and prioritization efforts were selected for contaminants on the following EPA lists: the Fourth Contaminant Candidate List (CCL 4, 81 FR 81099) (U.S. EPA 2016d, 2017d) and EPA’s Selected Analytical Methods for Environmental Remediation and Recovery 2017 (SAM) document (U.S. EPA 2017a).

The primary source was EPA’s Fourth Contaminant Candidate List (CCL 4) (U.S. EPA 2017d). The CCL 4 list uses occurrence and health effects data from the CCL 1, 2, 3, and 4 Universe, utilzing a robust process that includes expert judgment and public input (CFR 2016; National Drinking Water Advisory Council (NDWAC) 2004). EPA is directed by the Safe Drinking Water Act (SDWA) to publish the CCL in order to aid in efforts to set priorities (U.S. EPA 2016c). To be included on the list, the contaminant must:

1. exhibit an adverse human health effect;

2. present or have a considerable potential to be present based on production and use in public water systems at levels of public health concern; and

3. the EPA Administrator determines that regulation could present a meaningful opportunity for health risk reductions for persons served by public water systems.

The CCL chemicals are not currently subject to any of the national primary drinking water regulations, but may potentially require future regulation under the SDWA (U.S. EPA 2017d).

The secondary list utilized for this project is EPA’s Selected Analytical Methods for Environmental Remediation and Recovery 2017 (SAM) (U.S. EPA 2017a) document. SAM was initiated in 2004 by EPA’s Homeland Security Research Program (HSRP), when a group of experts across federal, state, and local agencies, universities, and municipalities were convened to develop a compendium of analytical methods for use during environmental response activities in support of EPA’s Environmental Response Laboratory Network, and EPA’s Water Laboratory Alliance. SAM identifies a single selected method or suite of methods for each analyte and environmental matrix pair. Having a single selected method for each analyte/matrix combination ensures that a consistent analytical approach is used if multiple labs are needed to analyze samples and that the generated data are comparable for decision-makers. Analytes are selected for addition into SAM using the following considerations:

• toxicity (both the parent and main degradation products),

• environmental persistence (both the parent and main degradation products if toxic),

• availability (both commercial and synthetic),

• history of weaponization from published accounts,

• ability to be transported in an environmental matrix that may require EPA remediation,

• presence on Federal lists,

• of public or regional concern, and

• included as a target analyte in method that is currently listed in the SAM.

SAM is included as part of HSRP’s Environmental Selected Analytical Methods (ESAM) Program (U.S. EPA 2020a). The ESAM program and its associated website are a comprehensive resource for field-and-lab-ready documents and web-based tools designed to support EPA’s mission, environmental response activities, and to assist water utilities in preparing for, responding to, and recovering from contamination incidents (U.S. EPA 2020a).

The entire CCL 4 list as well as additional SAM chemicals of interest to the HSRP program provided an initial chemical list to which additional selection factors were applied. Since this evaluation focused on the potential for oral exposure via contamination of a water from public systems, chemicals selected for this effort needed to be a liquid or solid and be moderately soluble in water under ambient conditions. Soluble gases were excluded due to the focus on the oral exposure pathway and the relative likelihood for significant exposure via inhalation route. It is important to note that all routes of exposure need to be considered when assessing risk for chemicals in any physical state. The solubility cutoff for inclusion was defined as ≥ 500 mg/L, representing moderately high solubility. This cutoff was selected to prioritize chemicals that might fully dissolve in drinking water at concentrations that are most likely to be relevant to toxicity, as solids and sediments are more likely to be filtered out before reaching the point of use and more easily detected in drinking water. The selected chemicals also need to represent an acute oral hazard. Oral LD₅₀ (lethal dose in 50% of the test population) values were chosen as a selection criterion because these are common, standard measures of acute oral toxicity that are widely available for individual chemicals. The threshold for this effort was defined as having a rodent oral LD₅₀ ≤ 500 mg/kg, indicating a potential hazard and priority for emergency response preparedness. While these thresholds aid in identifying the chemicals most relevant to acute risk from oral exposure in drinking water, this does not imply that the excluded chemicals on the CCL4 and SAM lists are not issues for acute and chronic exposure. The method used to identify priority chemicals is summarized in Figure 2.

Figure 2.

Chemical selection methodology. Chemicals were screened for inclusion from relevant lists based on defined selection criteria for physical state, acute toxicity, and solubility.

Risk/exposure guidelines

Drinking water exposures might span various exposure durations depending upon site-specific scenarios. Because (1) chronic health benchmarks are expected to be protective of acute exposures, (2) generally require greater analytical specificity and detection limits, and (3) are often available for drinking water contaminant chemicals, chronic (not acute) health benchmarks have been prioritized for further consideration.

Many chemicals have established health benchmarks that were developed by more than one agency. However, there are inconsistencies in the process used in derivation of these values because these were developed by different agencies for various purposes at different times (Holman, Francis, and Gray 2017a, 2017b). In some instances, interagency differences are substantial, especially for assessments that were conducted years apart. In lieu of chemical-specific decisions to resolve these discrepancies, a consistent priority scheme was applied to prioritize the information obtained from these sources. Where available, drinking water concentrations from EPA’s externally peer-reviewed Human Health Criteria (HHC), Human Health Benchmarks for Pesticides (HHBP), and Health Advisory (HA) documents were given first priority on a case-by-case basis (U.S. EPA 2002a, 2017c, 2018d). For substances lacking an EPA established human health benchmark for drinking water, oral reference doses (RfDs), and unit risk estimates from EPA Integrated Risk Information System (IRIS), Provisional Peer-Reviewed Toxicity Values (PPRTVs), and Health Effects Assessment Summary Tables (HEAST) assessments received next preference (U.S. EPA 1997, 2017b, 2018b). Where EPA assessments were not identified, Department of Defense (DOD) assessments were used to calculate drinking water concentrations (NRC 1999). Further information on the sources and applications of these values is provided in Table 1.

Table 1.

Oral health based guideline systems.

Value Name	Duration	Developed for:	Enforceable?	Values derived from?	Developing Program	Reference
HA	1 d, 10 d, longer term, or lifetime	Drinking water priority contaminants	Non-enforceable guidance	NOAEL or lowest observed adverse effect level (LOAEL)	EPA Office of Water (OW)	(U.S. EPA 2012)
HHBP	One day or chronic	Registered pesticides. List does not include pesticides that have existing HAs, Maximum Contaminant Levels (MCLs), or Maximum Contaminant Level Goals (MCLGs).	Non-enforceable	Applying health effects data (PADs) from pesticide registrations to methods used for developing health water advisories.	EPA Office of Pesticides EPA OW	(U.S. EPA 2017c)
HHC	Chronic	National Recommended Water Quality Criteria under Section 304(a) of the Clean Water Act	Non-enforceable exposure guidelines	NOAEL or LOAEL	EPA OW EPA Office of Science and Technology	(U.S. EPA 2000b)
MEGs	Chronic (1 year)	Chemical warfare agents and other select threat agents in field drinking water	Non-enforceable exposure guidelines	NOAEL or LOAEL. If not available, then the no-observed-effect level or lowest-observed-effect level was used.	National Research Council and U.S. Army	(Lawrence Livermore National Laboratories (LLNL) 1990; National Research Council (NRC) 1995, NRC 2004)
MCL	Chronic	National Primary Drinking water regulated chemicals	Enforceable standards	Highest level allowed in drinking water (mg/L)	EPA Office Ground and Drinking Water (OGDW)	(U.S. EPA 2009c, U.S. EPA 2019e)
MCLG	Chronic	National Primary Drinking water regulated chemicals	Non-enforceable public health goals	The level of a contaminant in drinking water below which there is no known expected risk to health.	EPA OGDW	(U.S. EPA 2009c, U.S. EPA 2019e)
Minimal Risk Levels	Acute (1–14 d), intermediate (15–364 d), and chronic (365 d and longer)	Hazardous substances most commonly found at facilities on the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA) National Priorities List	Non-enforceable. Used as a screening tool only	NOAEL/uncertainty factor (UF)	Agency for Toxic Substances and Disease Registry (ATSDR) and EPA	((Agency for Toxic Substances and Disease Registry (ATSDR) 2018)
RfD	Sub-chronic or chronic	Non-carcinogenic chemicals with an assumed threshold of toxicity	Non-enforceable risk assessment guidelines	NOAEL, LOAEL, or benchmark concentration (uncertainty factors applied to reflect limitations of the data used)	IRIS, PPRTVs, and National Research Council and U.S. Army	(U.S. EPA 2019c)
Cancer Slope Factor (CSF)	Chronic	Carcinogenic chemicals with an assumed threshold of toxicity	Non-enforceable risk assessment guidelines	NOAEL, LOAEL, or benchmark concentration (uncertainty factors applied to reflect limitations of the data used)	EPA IRIS and HEAST	(U.S. EPA 2005)

Drinking water concentrations originating from EPA HHC, HA, or HHBP values were reported as the current guidance value (U.S. EPA 2017c, 2018d). Additional drinking water values estimated herein serve no regulatory function and were derived for the sole purpose of identifying potential gaps in detection capabilities (U.S. EPA 2017c). It is important to note that many of the values in the 2017 HHBP Table date from 2012. The 2017 HHBP methodology was utilized in this study to convert the chronic RfD or oral slope factor (OSF) values to drinking water values aimed to be protective of chronic exposure as an initial point of comparison. In an acute exposure incident, there is not likely to be significant outside source contributions from chronic background exposure in the case of all chemicals listed in Table 2. The default relative source contribution (RSC) factor of 0.2, which takes into account other potential sources of exposure, was used in the absence of chemical specific data and clear guidance on judgment-based RSC amendments. For this study, oral RfDs were selected from IRIS, PPRTV, HEAST, or DOD (NRC 1999; U.S. EPA 1997, 2017b, 2018b). The 2017 HHBP formula (U.S. EPA 2017c) used to convert oral reference doses to drinking water values for noncancer endpoints was applied as follows:

Table 2.

Chemical categories.

Chemical	CAS	MDL (μg/L)	Chronic RfD or OSF (mg/kg-d)	Drinking Water Concentration (μg/L)
CATEGORY 1 CHEMICALS
Lewisite	541–25-3	NA	0.000005	NA
CVAA	64038–44-4	41	NA	NA
CATEGORY 2 CHEMICALS
3-Hydroxycarbofuran	16655–82-6	0.029a	NA	NA
Triethylamine (TEA)	121–44-8	0.1b	NA	NA
CATEGORY 3 CHEMICALS
N-nitrosopyrrolidine (NPYR)	930–55-2	4c	2.13d **	0.02e
N-nitrosodiethylamine (NDEA)	55–18-5	2c	150 d **	0.0008 e
1,2,3-Trichloropropane	96–18-4	0.008 f	30 e**	0.0007
Aniline	62–53-3	10 g	0.0057 h**	5.85
VX*	50782–69-9	0.082i	0.0000005 j	0.0032
EA2192*	73207–98-4	0.0130k	0.0000006 j	0.0084 l
CATEGORY 4 CHEMICALS
Methamidophos	10265–92-6	0.017k	0.0001***	0.60 m
Oxydemeton-methyl	301–12-2	0.010k	0.0001	0.60 m
Dicrotophos	141–66-2	0.025k	0.00003***	0.20 m
Acrolein	107–02-8	0.7 n	0.0156	3.00o
Hydrazine	302–01-2	0.003p	3q**	0.01
2-Propen-1-ol	107–18-6	0.5 r	0.005s	32
Acephate	30560–19-1	0.019k	0.0087 t**	2.28
Perfluorooctanesulfonic acid	1763–23-1	0.0011 u	0.00002	0.07 v
Quinoline	91–22-5	0.003 w	3x **	0.007
Nitrobenzene	98–95-3	1.2y	0.00004	10.00o
Oxirane, methyl	75–56-9	0.5 r	0.24 z**	0.87
Perfluorooctanoic acid	335–67-1	0.00053 u	0.00002	0.07 v
Dimethipin	55290–64-7	0.001 w	0.02	140 m
Sarin	107–44-8	0.0788aa	0.00002 j	0.13

^*CVAA and EA2192 chemicals are rapidly forming degradants of lewisite and VX, respectively, in water and should be considered in a water contamination event in addition to the parent chemical

^**Represents a cancer value (OSF)

^***Represents the population adjusted dose (PAD)

References for Table:

^a(U.S. EPA 2001c)

^b(U.S. EPA 2003)

^c(U.S. EPA 2014)

^d(U.S. EPA 2002a)

^e(U.S. EPA 2009e)

^f(U.S. EPA 1990c)

^g(U.S. EPA 1984b, 1996)

^h(U.S. EPA 1990a)

ⁱ(U.S. EPA 2015c)

^j(NRC 1999)

^k(U.S. EPA 2009a)

^l(Army Public Health Center (APHC) 2013)

^m(U.S. EPA 2017c)

ⁿ(U.S. Environmental Protection Agency (U.S. EPA) 1984a)

^o(U.S. EPA 2002a)

^p(U.S. EPA 1996)

^q(U.S. EPA 1988)

^r(U.S. EPA 2018e)

^s(U.S. EPA 1987)

^t(U.S. EPA 1989)

^u(Shoemaker and Tettenhorst 2020)

^v(U.S. EPA 2018d)

^w(U.S. EPA 2015d)

^x(U.S. EPA 2001b)

^y(U.S. EPA 1995)

^z(U.S. EPA 1990b)

^aa(U.S. EPA 2015b).

Risk Level in Drinking Water

$$\mu g∕L=\frac{\text{cRfD mg}∕\text{kg}∕\text{day}\times 1000\mu \mathrm{g}∕\text{mg}\times \text{80kg(adult BW)}\times 0.2(\text{RSC})}{2.5\mathrm{L}∕\text{day(adult DWI)}}$$ (1)

RfD: Reference Dose

BW: Body Weight

RSC: Relative Source Contribution

DWI: Daily Water Intake

For cancer-based values it was assumed that all chemicals exhibited a mutagenic, linear mode of action, with the exception of aniline due to sufficient evidence to the contrary. To account for mutagenic risk, an age-dependent adjustment factor (ADAF) was applied as an additional safety factor for lifetime risk based on the values provided in Table 3–13 of the exposure factor handbook (U.S. EPA 2011). The drinking water concentration derived for initial screening was calculated based upon one in a million (10⁻⁶) risk using the cancer HHBP formula (U.S. EPA 2017c). It is important to note that the agency may report that there is no drinking water level above zero that is without appreciable risk for known mutagenic chemicals. Drinking water concentration estimates for mutagenic carcinogenic chemicals were derived as follows:

Drinking Water Unit Risk

$$(\mu g∕L)=\frac{\text{CSF}\text{mg}∕\text{kg}∕\text{day}\mathrm{X}2.5\mathrm{L}∕\text{day}(\text{adult DWI})}{80\text{kg(adult BW)}\mathrm{X}1000\mu \mathrm{g}∕\text{mg}}$$ (2)

$$\begin{gathered} {10}^{-6}\text{Risk}\text{Level}\text{in}\text{Drinking} \\ \text{Water}(ppb)=\frac{{10}^{-6}}{\text{Unit Risk}\mu \mathrm{g}∕\text{LX}1.6(\text{ADAF})} \end{gathered}$$ Equation 2b.

CSF: Cancer Slope Factor

DWI: Daily Water Intake

BW: Body Weight

ADAF: Age-Dependent Adjustment Factor

In the absence of a more appropriate risk or toxicity value for the general population, military exposure guidelines (MEGs) were employed for CWAs. This study reports the 1-year MEG values for 5 L of water (NRC 2004). It is important to note that MEGs are not derived using the same assumptions typically used for deriving chronic risk values targeted for the general population. For example, the drinking water consumption is assumed to be 5 L per day, instead of the typical 2–2.5 L per day assumption utilized to derive the other chronic risk values reported in this paper. Further, MEGs do not apply any RSC factor whereas the other risk values included an RSC of 0.2. Finally, the toxicity studies employed as the basis of the risk value for MEGs are at times unique to military exposures in the absence of available data from other sources (NRC 2004).

Analytical methods

Analytical methods are important for identifying and quantifying chemical contaminants within environmental media and affected areas. These may be used to initially assess and characterize target chemicals, evaluate the remediation process (as it relates to identifying high levels of contamination and decontamination efficacy), and long-term monitoring after remediation completion. Several factors affect chemical identification and analytical sensitivity. Chemical suppression/enhancement effects associated with matrix interferences (Furey et al. 2013; Gosseti et al. 2010), chemical compatibility with a specific analytical technique including solubility or ionization, and chemical stability such as hydrolysis/degradation and fate and transport (Zwiener and Frimmel 2004a, 2004b) might all affect the ability of a particular method to accurately detect a specific chemical. These factors need to be considered when evaluating target chemicals to determine if analytical capabilities are met when used to evaluate protective human health levels. Analytical methods are readily available across instrument platforms, such that the analyst needs to be cognizant of the proficiencies (and limitations) associated with each instrument and analytical technique relative to a target chemical. Confirming the presence of a toxic chemical in environmental media requires several independent, but interconnected processes, each requiring time for completion. Consequently, contaminated areas may employ more than one analytical capability. While there are many widely available analytical techniques, two readily identifiable analytical techniques used during remediation efforts include field-deployable and lab-based systems. There are multiple attributes to each that may be advantageous toward identifying toxic chemicals and are discussed below.

Field deployable instrumentation have garnered attention because these may eliminate or greatly reduce the need for sample processing and transport (Erickson et al. 2006; Popiel and Sankowska 2011). Field-deployable systems might directly analyze samples within affected areas and provide near real-time data that enables faster decision-making processes during emergency response and remediation scenarios identification of highly concentrated areas that require immediate attention. Real-time analysis of environmental samples may also alleviate sample stability issues associated with sample collection, preservation, and transport. While the development of field instrumentation is rapidly growing, there are still limitations associated with this technique. Field instrumentation capabilities are limited by the ability to miniaturize important components that are necessary to reduce interferents. This hindrance might inhibit the ability to properly identify and quantify chemicals at low levels that are protective of human health. Field detection capabilities may also be accompanied with data uncertainty that might be attributed to false positives/negative results (Sferopoulos 2009). Due to these complications, field-deployable systems should not be considered a stand-alone process and need to be used in combination with additional analytical techniques to confirm analytical results.

Lab-based analytical methods are widely accepted for chemical characterization and might achieve low-level quantitation. However, highly sensitive methods that might reach protective human health levels may result in reduced sample throughput and/or data analysis capabilities. The added benefit of analytical sensitivity may be at the detriment of longer data completion times due to sample collection/transport, processing steps, and potentially longer analysis run times that are needed for lab evaluation. Thus, it is important to identify appropriate sampling, processing, and analysis procedures during initial screening and confirmatory analysis. Sample processing procedures are typically incorporated into lab methods to address matrix interferents associated with environmental media. Despite tedious procedures and potential for delayed analytical response times, sample processing might improve analytical detection limits by eliminating interferences.

Regardless of the selected analytical technique, analytical sensitivity might be directly affected by chemical and environmental conditions. Sample preservation is imperative during collection and processing procedures to ensure that target chemicals do not degrade prior to analysis. Physiochemical properties and chemical hydrolysis or degradation rates need to be evaluated during extraction and analysis processes to ensure sample integrity.

For some chemicals of concern, analytical detection limits still do not reach protective human health levels due to the complexities of environmental media and physiochemical properties of the target compound. It is important to identify analytical capabilities for as many chemicals and environmental media such as soil or water as possible. If human health levels are established, then analytical detection limits need to be evaluated with all available techniques and instrumentation to observe and report any advantages or limitations that may be present.

Analytical detection methods need to be validated and peer-reviewed; however, a validated, peer-reviewed method may not always be possible. As mentioned above, the ESAM/SAM contains a chemical list derived from subject matter experts with a focus on supporting EPA’s homeland security mission and was employed as part of the chemical selection and prioritization process (U.S. EPA 2020a). As part of SAM, one method is selected for each analyte/matrix pair to ensure that all labs involved in remediation activities use the same analytical method, allowing for comparable data to be utilized in decision-making. The ESAM program encountered chemicals that did not have validated analytical methods; however, subject matter experts derived a tiering system to account for differences between a validated method and alternative methods that may not be applicable for a specific chemical or a specific environmental matrix. The objective for this process was to enable a level of confidence in the analytical method for a specific chemical in a specific matrix when being used by the lab. The same process was applied to prioritize analytical methods for selected chemicals and selected methods are referenced in Table 2.

Results

The overall process for selecting priority chemicals for this evaluation is outlined in Figure 2. The number of chemicals remaining after applying each criterion is illustrated in Figure 3. Out of 102 chemicals, 24 chemicals met the defined physical, toxicity, and solubility criteria. These 24 chemicals are listed in Table 2. Risk and exposure are interconnected in emergency response. Both risk and exposure need to be characterized to inform decisions following a chemical release incident (Figure 1). The availability of a toxicity value such as a risk or exposure guidance value and an analytical method are key to integrating hazard and exposure to characterize risk from consuming contaminated drinking water (Figure 4). For each of the 24 chemicals identified in Table 2, available data on oral risk including both toxicity values or exposure guidance values and minimum detection limits (MDLs) for appropriate analytical methods were collected and prioritized, as previously described. In Table 2, available chronic risk values for water were compared to the MDLs for detection in water, and this comparison was utilized to assign each chemical to a category.

Figure 3.

Chemical selection results. Flowchart depicting the results of the chemical screening. “Yes” indicates how many chemicals met the given criteria and were included, while “No” indicates how many chemicals were excluded. Out of 102 chemicals, only 24 met all the defined inclusion criteria.

Figure 4.

Toxicity and analysis relationship. Visual representation of how risk values and analytical detection limits can inform each other for drinking water analysis.

Two chemicals, lewisite and CVAA, met category 1 criteria defined as a lack of both an appropriate risk value and an analytical method. Two chemicals, TEA and 3-hydroxycarbofuran, met category 2 criteria due to a gap in an available risk value. Six chemicals, 1,2,3-trichloropropane, aniline, VX, EA2192, NPYR, and NDEA, met category 3 criteria due to a gap in analytical sensitivity relative to the chronic risk value concentration. Fourteen chemicals exhibited no gap in either risk value or analytical sensitivity, designated as category 4.

The derivation of available chronic and acute risk values was leveraged to aid an informed estimate and discussion of acute toxicity for noncancer chemicals in category 3. For chemicals classified as probable human carcinogens in category 3, the ability to measure concentrations associated with the mid or higher end of the acceptable risk range was examined to determine if analytical sensitivity would be sufficient to inform risk management decisions based upon chronic cancer risk values. The estimated acute risk or higher-end risk thresholds for cancer suggest that the detection methods for four chemicals (VX, EA2192, aniline, and 1,2,3-trichloropropane) possess adequate sensitivity to measure at levels relevant to decision-making in an acute exposure incident (Appendix A). Two of six chemicals in category 3 (NDEA and NPYR) were found to potentially lack sufficient analytical sensitivity to measure drinking water concentrations associated with the upper threshold for cancer risk and potential risk management decisions.

Discussion

It is important to note that in the event of a water contamination incident, all routes of exposure need to be evaluated to assess risk and identify the risk driver. Water use may lead to aerosolization of the contaminant and significant exposure via the inhalation pathway (Davis et al. 2016). At times, inhalation exposure from aerosol-producing water use activities, such as showering, may pose a greater risk than oral consumption of the contaminated water. Further, water use for activities such as bathing and washing dishes might lead to additional exposure via the dermal pathway (Weisel and Jo 1996). Given the acute toxicity of the chemicals presented, it is particularly important to take all potential routes of exposure into account in a contamination incident response.

Given that this analysis is concerned with an emergency response and an unplanned release, including an acute or short-duration exposure, there was no concern with respect to chronic or outside exposures. However, the drinking water values used for direct comparison with detection limits are based upon chronic exposure ideally to the general population. This decision was made based upon the availability of chronic oral toxicity and risk values relative to other exposure durations. Further, the chronic duration of exposure leads to the most conservative risk value. Therefore, the chronic value would be expected to be protective in an acute exposure duration scenario and represents a clear point of comparison with analytical method detection limits. In the case that the chronic risk value is lower than the MDL of the analytical method, it is important to note that an acute exposure value would not require the same level of sensitivity. For example, the default RSC value of 0.2 is conservative and likely overestimates the magnitude of outside exposures in a water contamination scenario with an acutely toxic chemical. In the case where the chronic value is below our ability to detect, the RSC factor may be removed to determine the impact on isolated risk versus analytical capability. For carcinogenic chemicals, the upper threshold of allowable risk was calculated to determine if analytical capabilities were sufficient to inform risk mitigation decisions (Table A2). The lack of readily available acute toxicity and risk values presents a gap in exposure characterization for many acutely toxic and water-soluble agents. While 24 chemicals are presented in Table 2, only select chemicals will be discussed in detail within this study to represent the scenarios described above. For each highlighted chemical, the background, cause, and nature of the identified gap will be further discussed.

Category 1 chemicals: lewisite and CVAA

As previously described, category 1 chemicals lack appropriate risk values for drinking water and analytical capability to measure the chemical at health protective levels. Chemicals within category 1 are concerning because there is a lack of information available to derive appropriate health-based risk levels. Further, it is difficult to determine if analytical capabilities are possible in the absence of risk levels and whether appropriate analytical standards are available to develop analysis capabilities. Lewisite and CVAA are identified as category 1 chemicals, and the complications associated with developing analysis capabilities and risk-based health levels are described below.

Lewisite, 2-chlorovinylarsonous dichloride, is an organo-arsenic compound produced as a vesicant CWA between World Wars I and II (Vilensky 2005). Lewisite is a mixture of lewisite 1 (L1) as the major product and several lesser impurities, such as lewisite 2 and 3 in a 90:9:1 ratio, respectively (Muir et al. 2004). Despite lewisite mixtures, L1 is expected to be responsible for the main source of toxicity (Goldman and Dacre 1989). Improper disposal of former stockpiles from decades ago is creating a new interest in identifying lewisite and its hazardous degradation products. Several incidents include old CWA munitions that are accidently retrieved from ocean depths by fisheries or former military depots that buried unknown quantities of their stockpiles prior to decommissioning the land for public use (Hanaoka, Nomura, and Wada 2006; Ingold 2007; Jaffe 2013). Toxic pollutants that leach from munitions resulted in adverse health effects and environmental contamination of soil, water, and various media (Ingold 2007; Jaffe 2013; Press 2016, 2019; Tucker 2001). The problem remains an important health and environmental concern as more communities are built near previously owned munition depot sites or as vessels continue to fish in old dumping sites.

Lewisite is an oily, amber-colored liquid with a geranium flower-like odor (Goldman and Dacre 1989). Lewisite has a vapor pressure of 0.58 mmHg at 25°C and undergoes hydrolysis when exposed to moisture, even in ambient air (Munro et al. 1999). The Henry’s Law constant for lewisite is 3.2 × 10⁻⁴ atm x m³/mol suggesting volatilization from water, yet lewisite rapidly hydrolyzes making this fate pathway less likely (Munro et al. 1999). The solubility of lewisite in water is 0.5 g/L, but rapid hydrolysis rate results in formation of 2-chlorovinylarsonous acid (CVAA), 2-chlorovinylarsonic acid (CVAOA), and lewisite oxide as major products (Munro et al. 1999). A chronic oral RfD of 0.000005 mg/kg/d (5 x 10⁻⁶) was derived by the PPRTV program for lewisite. To circumvent the rapid degradation of lewisite in water, the chemical was mixed with sesame oil and administered to rats via intragastric intubation for 5 days per week for a period of 13 weeks (U.S. EPA 2015a). The toxicity database for lewisite includes an appreciable number of studies, most of which demonstrate lack of detectable effects below doses that produce frank effects (mortality). These effects or lack thereof were used to identify a point of departure for derivation of a PPRTV for lLewisite (U.S. EPA 2015a). Because the severity of oral effects demonstrated by lewisite exceeds the level of severity selected for determination of acceptable standards for drinking water concentrations, this PPRTV value was not employed to estimate a drinking water concentration for the purposes of this analysis. Effectively, no appropriate risk value for the parent chemical, lewisite, exists or can be derived for drinking water. The value derived may be more applicable to ingestion of lewisite aerosols in a low humidity environment based upon the method of direct internal exposure utilized in the derivation of the value and instability of lewisite in the presence of water. Due to the rapid conversion, the MEGs program derived a drinking water value for its degradation product lewisite oxide to partially account for risk of lewisite introduction into the drinking water supply. This 1 year, 5 L drinking water value of 4.2 μg/L is geared toward military exposures (Munro et al. 1999). Targeted risk values for CVAA and CVAOA are not available at this time.

There have been numerous efforts to develop an analytical method for lewisite, specifically L1, all with varying success (Hooijschuur, Kientz, and Brinkman 2002; Östin 2012). Previous techniques that analyze for L1 include gas chromatographymass spectrometry (GC-MS), inductively coupled plasma-mass spectrometry (ICP-MS), and liquid chromatography-mass spectrometry (LC-MS) (Hooijschuur et al. 2002; Hooijschuur, Kientz, and Brinkman 2002; Kientz 1998; U.S. EPA 2014; Witkiewicz, Mazurak, and Szulc 1990). ICP-MS uses arsenic (As) speciation to identify the presence of L1; however, if any other forms of As are present, such as naturally occurring metalloid in soils or Asladen pesticides, ICP-MS cannot differentiate and identify the metallic source. Lewisite is a highly volatile vesicant (4,480 mg/m³) suggesting that GC-MS is a reliable analysis technique; however, in addition to rapid hydrolysis complications, Lewisite rapidly reacts with GC inlets, columns, and instrumentation, producing difficulties with GC analysis techniques and method robustness (Hooijschuur, Kientz, and Brinkman 2002). L1 analysis by GC-MS generally requires derivatization (Black and Muir 2003; Haas 1998). GC-MS analysis may result in a nonlinear response at low analyte concentration, complicating trace environmental analysis, and detection of thiol-derivatized L1 by GC-MS is enhanced by the presence of excess Ascontaining compounds like AsCl3 and its thiol derivatives (Hanaoka, Nomura, and Wada 2006). Unfortunately, excess of these compounds might also induce damage to GC columns. Derivatized L1 in an organic liquid was qualitatively identified by GC-MS (Hooijschuur et al. 2002) and quantified from munitions using external calibration over 20–200 ng (Hanaoka, Nomura, and Wada 2006). Lewisite detection in water, air, and wipes from surfaces still rely on the conversion of lewisite to a derivatized product (Stan’kov, Sergeeva, and Tarasov 2000).

The derivatization process required for GC-MS analysis of lewisite may be sensitive to hydrolysis; thus, any residual water might result in incomplete transformation of the lewisite-derivatized product and/or formation of hazardous degradation products of lewisite. A major pitfall of the derivatization process is that it does not allow for speciation of any of As products, including lewisite and any by-products that are formed. As described above, examples of hazardous degradation products include CVAA, CVAOA, and lewisite oxide, of which CVAA is known to exhibit vesicant properties (Goldman and Dacre 1989). Therefore, it will be important to distinguish between by-product formation.

Due to the vigorous reactivity of lewisite under normal environmental conditions and rapid chemical changes that occur in the presence of moisture, such as hydrolysis and subsequent oxidation, Lewisite analysis might be extremely complicated. Literature precedent suggests L1 hydrolyzes so quickly that the converted, stable CVAA compound actually is the main material identified in water (Munro et al. 1999). Regardless of the degradation pathway, once lewisite is exposed to environmental conditions or human exposure occurs, its metabolites and degradation products become the main sources of identification (Munro et al. 1999). Data suggest that the detection and analysis of CVAA and/or CVAOA in the environment are indicative of L1 contamination, as there are no known natural sources for these compounds. Lewisite’s presence in biological samples is typically identified by its metabolites because the main degradation pathway converts it to its stable by-products (Logan et al. 1999). There has been extensive work on analysis by liquid chromatography tandem mass spectrometry (LC-MS/MS) of the lewisite metabolites CVAA and CVAOA as indicators of lewisite exposure (Rodin et al. 2011). CVAA and CVAOA analysis by LC-MS/MS was performed with detection limit capabilities of 41 μg/L (U.S. EPA 2015b). Despite the lack of available risk values for CVAA and CVAOA, there remains a clear gap necessary to achieve the sensitivity needed for detection of lewisite in water based upon the oral toxicity and risk value available for the parent chemical and its degradation product, lewisite oxide. The risk value for lewisite oxide exposure via drinking water is reported as 4.2 μg/L. Given the lack of risk values and the ability to detect at the potential health protective levels, it would be challenging to characterize the risk associated with lewisite contamination of the water supply.

Category 2 chemicals: triethylamine

It is important to understand the adverse health risk associated with an exposure at a given dose when assessing and responding to a drinking water contamination incident. The availability of an analytical method enables detection and measurement of the chemical in the water supply but does not provide health context in the absence of a risk value, designated herein as a category 2 gap. TEA falls into the category 2 designation. It is a colorless liquid chemical with a potent ammonia- or fish-like odor (National Library of Medicine (NLM) 2020a; NIOSH 2019; U.S. EPA 2000a, 2017d). TEA is used for many industrial and chemical processes including as a(n): (1) catalyst for synthesizing organic chemicals; (2) inhibitor for corrosion; (3) accelerator and vulcanization agent in the manufacture of rubber; (4) polymer hardening and curing catalyst; (5) stabilizer for herbicides and pesticides; and (6) penetrating, wetting, and waterproofing agent used in production of quaternary ammonium compounds for textile auxiliaries and dyes (National Library of Medicine (NLM) 2020a; NIOSH 2019; Tice 1998; U.S. EPA 2000a, 2017d). TEA has also been used as a propellant, to precipitate and purify antibiotics such as penicillin and cephalosporin and to desalt seawater, was present in carpet cleaners, paint removers, and sealers, and used in photography (National Library of Medicine (NLM) 2020a; NIOSH 2019; Tice 1998; U.S. EPA 2000a, 2017d). TEA is designated as a hazardous substance under the Federal Water Pollution Control Act (sections 311(b)(2)(A) and 501(a)) and the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA- section 201[a]), (CFR 2004, 2011). TEA is regulated by the Clean Water Act Amendments of 1977 (Pub. L. 95–217) and 1978 (Pub. L. 95–676) and is listed as reportable at 5000 pounds (2270 kg) under both the Clean Water Act (section 311) and CERCLA (section 201[a]) (Code of Federal Regulations (CFR) 2003, CFR 2004, CFR 2011). Section 313 of Title III of the Superfund Amendments and Reauthorization Act (SARA) of 1986 requires information to be submitted to EPA’s Toxic Release Inventory (TRI) regarding facility release of TEA and other toxic chemicals. SARA also specifies the associated waste management activities for Section 313 chemicals including TEA (Environmental Assistance Center (EAC) 2018).

Environmental contamination with TEA typically occurs during use and manufacturing of the chemical after it is released via air emissions; however, release of effluents to land or waterways has also been documented (Tice 1998; U.S. EPA 2019d. 2019). For example, in 2019, of the total 108,198 pounds of TEA released, 69% of releases were to air, 23% to land, 6% to off-site disposal, and 2% of releases went into waterways (U.S. EPA 2019d. 2019). Although the majority of releases of TEA occur via air, the focus of this investigation is on contamination of chemicals specifically for water. Therefore, the remainder of this discussion will focus on the impact of TEA present in water and associated methods for detecting TEA in water samples.

In 2018, 23 facilities released a total of 30,429 pounds of TEA into U.S. waterways, with 8 facilities releasing 1000 or more pounds each; the largest releases from a single facility in St. Gabriel, LA totaled 5,793 pounds (U.S. EPA 2019a, 2020b). When released in water, TEA degrades slowly with a half-life of 70 hr and is considered to be miscible (2%) (NIOSH 2019; Tice 1998). However, TEA is not degraded by activated sludge (National Library of Medicine (NLM) 2020a). TEA was added to the U.S. EPA CCL3 list in 2009 and remains on the CCL4 having been given a potency score of 6 (based on the Registry for Toxic Effects Lowest observed adverse effect level of 1 mg/kg-d), a severity score of 5 (based on health effects of degenerative changes to the brain and coverings), a prevalence score of 10 (score based on 2004 TRI release data in 35 states), and magnitude score of 9 (based upon TRI release data from 2004 of 1,167,219 pounds released into waterways) (CFR 2009, 2020; U.S. EPA 2009d, 2016b, 2016c, 2017d, 2018c).

Exposure to TEA most notably occurs in occupational settings through inhalation or dermal exposures as it is being manufactured and utilized; however, ingestion via contaminated foods such as boiled beef and ocular exposure might also occur (National Library of Medicine (NLM) 2020a; NIOSH 2019; Tice 1998). The chemical typically produces irritation in the respiratory system, eyes, and skin but might also affect the liver, lungs, brain, kidneys, and cardiovascular system. TEA might also produce severe toxicity and ultimately death following ingestion or inhalation (NIOSH 2019; Tice 1998). While OSHA has set permissible exposure limits for TEA in air emissions and there is an IRIS RfC, established risk limits for finished drinking water, other nationally representative risk data for contaminated water, and a RfD are lacking (NIOSH 2018; U.S. EPA 1991, 1999). Oral LD₅₀ s are reported for TEA with values of 460 mg/kg in rats and 546 mg/kg in mice (American Medical Association (AMA) 1951; National Library of Medicine (NLM) 2020a). Other available toxicity data for TEA include: slightly reduced body weight in rats given 500 ppm of TEA in drinking water; slight toxicity in rats given 30 mg doses and reports of convulsions in rats and, in some cases, death for female rats given 60 mg via gavage for 6 weeks combined with changes in reflex responses in rats receiving multiple doses of 1 or 10 mg/kg TEA (Tice 1998). In the same study rats that received oral daily doses of 54.4 mg/kg TEA for 2 months did not display any observable effects (Tice 1998).

Currently, there are no quantitative risk estimates, and no exposure guidance values established for orally encountered TEA. The only method to approximate the potential health impact of TEA in water is through EPA’s Risk-Screening Environmental Indicators (RSEI) screening model. RSEI scores provide an estimate of the relative health-related impacts for chemicals reported in the TRI and calculated based upon the (1) amount of chemical released, (2) chemical’s fate/transport in the environment, (3) chemical toxicity data, (4) route of human exposure, and (5) potential hazard to the population at risk (U.S. EPA 2018a, 2019b). The resulting score may be compared to scores for the different facilities releasing TEA as well as compared to other chemicals in the TRI database (U.S. EPA 2018a, 2019b). For example, in 2018, of the 23 facilities that reported TEA releases into water, 4 of these had RSEI scores over 1000 and would be considered to have the most potential health-related impacts compared to the other facilities (U.S. EPA 2019b, 2020b).

Despite lack of established risk values for TEA, this chemical is included as an analyte in EPA Office of Water’s Method 1671. Under the Clean Water Act, Method 1671 is designated as an applicable method for monitoring and surveying non-purgeable volatile organic pollutants used in the pharmaceutical manufacturing industry in water, municipal sludges, and soil (U.S. EPA 1998). Analysis via Method 1671 consists of direct aqueous injection (DAI) GC and detection by a flame ionization detector (FID) (U.S. EPA 1998). The minimum level for TEA in water for this method is 50 mg/L with initial precision and accuracy reported between 70 and 165% (U.S. EPA 1998). Similarly, TEA might also be detected in water samples utilizing EPA Method 8015D which also employs DAI (using a capillary column) and analysis via GC/FID (U.S. EPA 2003). EPA Method 8015D is appropriate for samples containing low μg/L (ppb) concentrations of TEA (U.S. EPA 2003). To determine performance of the method for analysis of TEA, 7 aliquots of reagent water were spiked with TEA at 1 μg/L and found to have a mean recovery of 1.169 μg/L (117%). While purge and trap methods are also listed in this method, TEA is not amendable to analysis via purge and trap techniques (U.S. EPA 2003).

Category 3 chemicals: VX and EA-2192

Category 3 chemicals possess risk values; however, either no analytical value exists or the analytical capability appears to not be protective based upon the associated health-based risk value. VX and its degradant EA-2192 are examples of category 3 chemicals. For these chemicals it is challenging to make decisions regarding the use and decontamination of water systems and to communicate those decisions and associated risks for these chemicals.

VX or ethyl ({2-[bis(propan-2-yl) amino] ethyl} sulfanyl)(methyl)phosphinate is a chemical that falls into this category. It is a Schedule 1 chemical under the Chemical Weapons Convention (CWC), meaning it was developed, produced, or stockpiled and there is little or no use for the chemical outside of that prohibited by the CWC (CFR 2007; Organization for the Prohibition of Chemical Weapons (OPCW) 2020). CWAs and its precursors are on EPA’s SAM list. Because VX is a clear, amber colored, oily liquid that is miscible with water (30 g/L at 25 °C) and is odorless when pure (NLM 2020b), it is viewed as a potential water threat. It is also a persistent threat, with a half-life in water (pH of 7) of 17–42 days at 25°C with the half-life decreasing markedly with increasing pH (Clark 1989). Ingestion exposure to VX produces nausea, vomiting, diarrhea, abdominal pain, and cramping but the time course of the symptoms is not well understood (National Institute for Occupational Saftey and Health (NIOSH) 2011). VX, like other nerve agents, initiates irreversible inhibition of the cholinesterase enzyme, resulting in the accumulation of acetylcholine. This accumulation results in continuous stimulation of the nervous system and failure of autonomic functions which might lead to seizures, coma, and death (Adeyinka and Kondamudi 2020; New York State Department of Health (NYSDH) 2005). While there are no documented uses of VX in water systems, it is believed that VX was deliberately used to poison others, including Kim John-nam, the brother of North Korea’s leader Kim Jong-un (Doyle 2017).

VX degrades in water to produce EA-2192, diisopropylethyl mercaptoamine (DESH) and ethyl methylphosphonic acid (EMPA), with the degree to which these substances are formed depending upon pH of the water (Munro et al. 1999). DESH further reacts to form EA-4196 and bis(2-diisopropylaminoethyl) sulfide. EA-2192 forms at pHs from 7 to 10 and is also a cholinesterase inhibitor. VX has a lowest toxic oral dose (TDLO) in humans of 4 μg/kg (Sidell and Groff 1974). When comparing animal LD₅₀s for intravenous routes, VX (U.S. Department of the Army (DOA) 1974) is less than 2-fold as toxic as EA-2192 in rabbits (Horton 1962). Therefore, when assessing acute toxicity of VX due to a water system contamination incident, EA-2192 needs to be considered.

The NOAEL for significant depression of blood cholinesterase activity is the critical endpoint for defining a maximum acceptable exposure level to nerve agents like VX. Using this endpoint, the U.S. Department of the Army conducted a health assessment for VX in 1996 that was reviewed by the National Research Council (US) Subcommittee on Chronic Reference Doses for Selected Chemical Warfare Agents (NRC 1999). The subcommittee concluded that an acute study based upon human exposure in drinking water would be more appropriate than an animal exposure study for the basis of the VX RfD. The proposed oral chronic RfD was based upon red blood cell cholinesterase inhibition in humans administered VX in drinking water for 7 days (Sim et al. 1964). In terms of acute exposures, there is a MEG for VX that assumes either 5 L or 15 L per day consumption of water for less than 7 days with respective concentrations of 15 and 5 μg/L might not produce effects. Consumption should not impair performance and is considered protective against significant noncancer effects (U.S. APHC 2018; U.S. DOA, U.S. Navy, U.S. AF 2010).

The estimated drinking water risk value derived from the NRC army oral RfD (0.0032 μg/L) is lower than the detection limit of 0.082 to 0.74 μg/L in reagent water described in the Tier 1 method included in the EPA SAM document. This method uses microscale solvent extraction and GC/MS with either full scan Time of Flight MS or Quadrupole MS (U.S. EPA 2016a). In addition to this method, data demonstrate a method employing an immunomagnetic scavenging technique employing magnetic beads coated with butyrylcholinesterase (BuChE) and subsequent detection of the digested VX-BuChE adduct using LC-MS /MS. This method was able to quantify VX in simulated tap water samples (HPLC-Grade Water with sodium omadine (64 mg/L) and sodium thiosulfate (80 mg/L) to simulate the neutralization), with the limit of detection (LOD) of 5.6 ng/L. A lower LOD was observed but it is also slightly above the Oral RfD (Knaack et al. 2013) and requires supplies labs typically do not have on hand isotopically labeled peptides. More sensitive field detection methods would be extremely helpful to rapidly characterize the initial extent of contamination and inform evacuation and do-not-use orders. Currently there is the M-272 Water Test Kit which detects VX in water in 7 min at concentrations above 0.02 mg/L (U.S. Department of Labor (DOL) 2020), which is significantly above the health-based standard.

Because the effect selected for the derivation of the oral RfD is a biomarker for an effect, (60% inhibition of red blood cell acetylcholinesterase activity), rather than the effect, an uncertainty factor of 10 was applied to develop a chronic value. It is likely that the value is overly conservative for making decisions in the days to weeks following a release. Based upon this adverse effect estimation, the analytical sensitivity of the sampling and analysis method described above may be adequate as a surrogate for detection of VX in drinking water related to acute exposures. Although developed for military personnel, the acute MEG value of 15 μg/L for less than a 7-day exposure and an average drinking water consumption of 5 L/day (U.S. APHC 2018) indicates that values developed for acute exposures in the general public may exceed the 0.082 μg/L MDL for VX detection in water.

When dealing with a water system contaminated with VX, measurement of its toxic degradant EA-2192 would be critical. EA-2192 displays an oral RfD of 0.0006 μg/kg/day. This value was not derived based upon EA-2192 toxicological studies but instead reflects an oral RfD derived for VX (U.S. Army Center for Health Promotion and Preventive Medicine (U.S. ACHPPM) 1999) and corresponds to a drinking water concentration of 0.0038 μg/L. The MEG risk assessment program developed a value of 0.0084 μg/L for EA-2192 which aimed to represent a negligible risk to military personnel for up to 1 year of exposure at an average water consumption of 5 L/day (U.S. APHC 2018). The study that served as the basis of this MEG value and the oral RfD was described by Bausum, Reddy, and Leach (1998), where he noted that EA-2192 was a less potent ChE inhibitor than VX. Therefore, the RfD value developed for VX (0.0006 μg/kg-d) (Opresko et al. 1998) might be applied to EA-2192, noting that such an application would ensure protectiveness (Bausum, Reddy, and Leach 1998).

There is a method for detection of EA-2192 in drinking water which utilizes an adapted U.S. EPA Method 538, adding a flow diversion valve to reduce contamination of the ionization source in the DAI-LC/tandem mass spectrometer. Sodium omadine (64 mg/L) and ammonium acetate (154 mg/L) are added to HPLC-grade water to simulate neutralized drinking water (near pH 7). The LOD in this simulated water sample was 13 ng/L, just above the estimated oral RfD (U.S. EPA 2009a). Because the EA-2192 oral RFD is based upon the VX oral RfD and is known to be even more protective because of its reduced ChE inhibition, the analytical sensitivity may be adequate for the detection of EA-2192 in drinking water. Further, the acute MEG for VX of 15 μg/L is well above the LOD for EA-2192 of 0.013 μg/L, suggesting the method sensitivity may be sufficient for protection for acute risk.

Category 4 chemicals: methamidophos

Category 4 chemicals possess both a risk value and an appropriate detection method with adequate analytical sensitivity to measure the chemical at a concentration deemed protective of human health. Chemicals in category 4 may still pose a risk to human health in drinking water due to high solubility and acute oral toxicity; however, there is no gap in the risk value or analytical method sensitivity needed to adequately assess the risk following a contamination incident. An example of a chemical that falls into category 4 is the organophosphate pesticide, methamidophos.

Methamidophos exists as a crystalline solid that is clear to off-white in color and possesses a pungent odor. The pesticide exhibits high solubility in water, low volatility, and acute toxicity via the oral route (University of Hertfordshire 2020). Methamidophos was produced as an insecticide and was registered for use in the United States from 1972 to 2009. All uses of methamidophos in the United States were voluntarily canceled in 2009 (U.S. EPA 2009b). Stockpiles of the pesticide likely still exist within the United States. Further, small exposures to methamidophos may still occur incidentally via the use followed by degradation or metabolism of its precursor pesticide acephate (Chukwudebe, Hussain, and Oloffs 1984; Ziqiu et al. 2020). Acephate is currently still registered for use in the United States. To mitigate the potential risk occurring from inhalation and ingestion of dust or aerosols following acephate use, the chemical is not registered for residential employment indoors or for certain turf applications. EPA’s Pesticide Reregistration and Tolerance Reassessment program determined that acephate did not pose a significant risk for exposure in water (U.S. EPA 2001a). In contrast, the EPA Office of Pesticide Programs determined that screening level modeling estimates for methamidophos indicated that drinking water contamination may pose risk concerns in the Reregistration Eligibility Decision for Methamidophos (U.S. EPA 2002b).

U.S. EPA estimated that only 1.6% of the available acephate breaks down into methamidophos in water, yet data indicate that acephate is rapidly degraded to methamidophos and DMPT in plants, insects, and soil. Acephate hydrolysis is slow but occurs faster at alkaline pHs and elevated temperatures. In distilled water at pH levels 4–6 over 80% of acephate was recovered from water after 20 days, with only 0.2% recovered as methamidophos (Davy, Eckel, and Hammer 2007; Downing 2000). In an acute poisoning case study where a 4-year-old ingested a toxic dose of acephate, low metabolic conversion of acephate to methamidophos is suggested by measurement of urine concentrations. The ratio of methamidophos to acephate recovered in urine was 0.02 (Chang et al. 2009). The half-lives of methamidophos and acephate are low in humans, therefore the chemicals do not bioaccumulate in the body. The half-life of methamidophos is estimated to be 1.1 hr from an adult human study (Garner and Jones 2014). In an NHANES survey of the population, methamidophos was not found at detectable levels in the vast majority of the population surveyed (Centers for Disease Control and Prevention (CDC) 2017). Given the relatively low rate of conversion of acephate to methamidophos in water and in humans, the primary concern for a high dose of acute exposure to methamidophos might be through either intentional or accidental introduction of preexisting stocks of methamidophos into a water supply. Methamidophos is included on EPA’s CCL4 list, receiving a potency score of 7, a severity score of 5, a prevalence score of 10, and a magnitude score of 6 (U.S. EPA 2016b).

Both acephate and methamidophos are known to produce toxicity via inhibition of the acetylcholinesterase enzyme, leading to a toxic and at times fatal accumulation of acetylcholine. However, methamidophos is known to be more potent in humans than its precursor acephate. Symptoms of organophosphate poisoning include excessive sweating, nausea, vomiting, weakness, seizures, and paralysis. The human health risk of the organophosphates at low doses is not well characterized (CDC 2013). The risk of oral exposure to methamidophos was evaluated and a chronic HHBP drinking water value of 0.6 μg/L was derived. The drinking water value is based upon a PAD which takes the RfD and applies an additional Food Quality Protection Act of 1996 (FQPA) safety factor to account for both potential prenatal and/or postnatal toxicity and the completeness of the database with respect to exposure and toxicity for women of child-bearing age, infants, and children (U.S. EPA 2017c). The study used as the basis of the chronic PAD was an 8-week subchronic oral toxicity study in rats which measured brain ChE inhibition. The no-observed-adverse-effect level (NOAEL) dose was 0.03 mg/kg/day to which a total UF of 100 was applied, followed by a FQPA safety factor of 3 to derive an ultimate PAD of 0.0001 mg/kg/day (U.S. EPA 2006).

EPA created a method for the quantitative measurement of methamidophos in water which was vetted by EPA’s ESAM program (U.S. EPA 2009a). EPA method 538 uses DAI-LC/MS/MS to measure selected organic contaminants in drinking water. The LOD for methamidophos in reagent water was determined to be 0.017 μg/L. The method also identifies the minimum reporting level as 0.032 μg/L. Both the LOD and minimum reporting level are well below the risk value of 0.6 μg/L for drinking water. Therefore, there is adequate analytical sensitivity to measure methamidophos and monitor its concentration at levels that are at or below the derived risk value for drinking water. Having a complete risk picture better informs decisions regarding the clean-up and response to a contamination incident.

Conclusions

This study identified research gaps, which if addressed, might yield information relevant to regulations. Filling these gaps might also enhance emergency preparedness for water system contamination incidents. Other organizations may have their own chemical prioritization scheme to inform their research, but this effort provides an approach that may be adapted to meet the needs of other entities. Extremely soluble chemicals that might pose a significant and acute threat to the general population via oral consumption of drinking water following a contamination incident were the focus of the present evaluation. Out of 24 chemicals of greatest concern for public drinking water identified by the homeland security research program, 6 lacked either the risk value or analytical sensitivity required to detect the chemical at levels that are estimated to be protective against adverse effects on human health in adults following acute exposures. Because of differences in drinking water ingestion rates between adults and children, the ingestion rates used in this analysis should not be used to infer adequate protection against the risk to children. Category 1 chemicals (CVAA and lewisite) lacked both an available risk determination value for drinking water and a method with the analytical sensitivity required to detect it at estimated risk relevant concentrations. Category 2 chemicals (3-hydroxycarbofuran and TEA) have an available analytical method but lack a reliable oral toxicity or risk value. Category 3 chemicals (aniline, VX, EA2192, 1,2,3-trichloropropane, NPYR, and NDEA) have analytical methods that are not able to detect at concentrations associated with chronic risk values. Two category 3 chemicals, NDEA and NPYR, are estimated to lack analytical methods with sufficient sensitivity to detect at water concentrations relevant to 100-in-1 x10⁶ cancer risk thresholds. These six category 1, 2, and 3 chemicals identify current gaps in agency response preparedness. By highlighting these analytical and toxicological data gaps, attention is called to the need to focus current research in order to meet our preparedness objectives for response to a drinking water contamination incident involving the chemicals evaluated. Focused research addressing most serious gaps might enable emergency responders to better understand the risks from an oral exposure to contaminated drinking water to make informed risk management decisions. After investigating these six highest priority chemicals, research needs to then identify and address the next tiers of public water supply health threats from CCl 4 and SAM lists of chemicals contaminating drinking water supplies.

Data availability Statement

The data that support the findings of this study are openly available in EPA’s Science Inventory at http://doi.org/10.23719/1521126 reference number [to be assigned after acceptance].