5. Toxicity

How risk assessors choose the toxicity values may significantly affect results of the risk assessment. Appendix A provides a table that lists some of the common sources of toxicity values as well as links to each for convenient reference. At times, specific state regulations limit the choice of toxicity values, while at other times the risk assessor must use professional judgment to determine the best toxicity values for a specific chemical.

Although the USEPA and some states have developed source hierarchies for use in selecting toxicity values, sources may have varying schedules for reassessing toxicity values. Depending on when new information becomes available or is evaluated, toxicity values for a chemical in one source may be more up-to-date than those in another source.

Guidance documents listed in Section 5.3 provide more detail regarding each toxicity value type, the fundamentals associated with toxicity values, and other key definitions (for example, the difference between acute, chronic, and subchronic exposure durations defined in (USEPA 1989a). Toxicity values should be chosen based on the best available science for a specific chemical at the time the risk assessment is prepared (USEPA 2003d). This approach requires scientifically sound professional judgment by the risk assessor.

5.1.1.1 Options – Use USEPA Guidance

Initially, USEPA (1989a) presented an approach for selecting toxicity values by recommending a hierarchy of sources. Because the field of toxicology evolves rapidly, however, USEPA does not require rigid application of this hierarchy. Instead, USEPA recommended that “EPA and State personnel may use and accept other technically sound approaches.” Protection of human health can only be achieved by using the best available science as the basis for risk assessments, therefore USEPA generally recommends a three-tier hierarchy of toxicity information sources (USEPA 2003d):

• Tier 1 – USEPA’s Integrated Risk Information System (USEPA 2015e)
• Tier 2 – USEPA’s Provisional Peer Reviewed Toxicity Values (USEPA 2013f)
• Tier 3 – Other Sources – additional USEPA and non-USEPA sources, including toxicity values prepared by states and agencies. Priority is given to sources that provide toxicity information based on similar methods and procedures as those used in Tier 1 and Tier 2 sources, and that are peer-reviewed, available to the public, and transparent in the methodologies and processes used to develop the values. USEPA provides examples, including the California EPA’s toxicity value database (CalEPA 2013), the Agency for Toxic Substances and Disease Registry Minimal Risk Levels (MRL) (ATSDR 2013), and the USEPA Health Effects Assessment Summary Tables (HEAST) (USEPA 1997d).

5.1.1.2 Option – Use USEPA Guidance Supplemented with ECOS Guidance

The Environmental Council of the States (ECOS) offers guidance on the characteristics of preferable Tier 3 toxicity values (ECOS 2007). ECOS recommends assigning preference to toxicity values that are developed as follows:

• using a “previously established and publicly available methodology”
• using methods reflecting “…the current best scientific information and practices; new assessment methods should provide reproducible results and meet quality assurance and quality control requirements”
• considering "…the quality of studies used, including the statistical power or lack thereof to detect effects”
• using durations "…consistent with the duration of human exposure being assessed”

5.1.1.3 Option – Use State Agency Toxicity Values

Many states have adopted USEPA’s recommended hierarchy or a modified version of USEPA’s hierarchy (for example, NJDEP 2008; Pennsylvania Revised Code 2011; University of Florida 2005). Also, some states (for example, California) stipulate specific toxicity values within their state regulatory framework and may require that their values be used to supplement or supersede USEPA’s hierarchy of sources. Even if an agency does not stipulate the toxicity values, it may have guidance on this issue.

5.1.1.4 Option – Consult Experts in Toxicology

If experts and agencies disagree about which toxicity value is best when multiple values are available, the Superfund Health Risk Technical Support Center (STSC) (USEPA 2013h) or a professional toxicologist can be consulted. This approach is not an option if the state stipulates the toxicity value source in its code or regulations.

5.1.2.1 Option – Determine whether the Toxicity Value is Needed

It may not be necessary to search for toxicity values if, based on professional judgment, the chemical is unlikely to significantly affect the risk assessment results or risk management decisions (for example, the concentrations of other chemicals colocated in the area would warrant remediation action already). For example, if a volatile compound is missing an oral toxicity value but has an inhalation toxicity value, then the missing oral toxicity value may not be a significant issue because, for volatile chemicals, the inhalation pathway may be more significant than the ingestion pathway. Also, if the frequency of detection of a chemical is low (for example, less than 5%) and if the chemical is detected at low concentrations, then missing toxicity values may not be a significant issue (see Section 4.5.2 for additional information on selecting chemicals for evaluation).

5.1.2.2 Option – Use Chronic Toxicity Values for Subchronic Exposures or Vice-versa

Chronic toxicity values are typically used in risk assessments and are available in the Tier 1 source (IRIS), whereas both chronic and subchronic toxicity valuesToxicity values typically used for exposure durations ranging from 2 weeks to 7 years, and are not used for children ages 0-6. may be available in Tier 2 and Tier 3 sources. When a subchronic toxicity value is needed, a lower tiered source (Tier 2 or Tier 3) can be searched for a subchronic toxicity value or the chronic toxicity value (from the Tier 1 source) can be used. For example, a chronic oral RfD could be used as a surrogate for a subchronic oral RfD. Conversely, subchronic toxicity values could be used as surrogates for unavailable chronic toxicity valuesToxicity values used for repeated or persistent exposures (durations exceeding 10% of a lifetime [7 years or longer] and for exposures by children ages 0-6). (for example, a subchronic inhalation reference concentration (RfC)A concentration specified by USEPA to limit human inhalation exposure to potentially hazardous levels of chemicals in air (Commission 1997a). from a Tier 2 source could be used as a surrogate for an unavailable chronic inhalation RfC). A subchronic toxicity value should not be used as a surrogate for a chronic toxicity value, however, without making adjustments to account for the duration of exposure used to define the subchronic value, and without consideration the sensitivity of the endpoint used to define the subchronic value.

5.1.2.3 Option – Use Toxicity Values for a Similar Chemical

Toxicity values available for other chemicals with similar chemical structure can be used as surrogates (see Section 4.5.3.2) when a toxicity value is not readily available in Tier 1, Tier 2, or Tier 3 sources, and it is not possible to substitute a chronic or subchronic toxicity value (as discussed in Section 5.1.2.2), Currently there is no general consensus on assessing similarity for chemicals used as surrogates. Historically, risk assessors chose surrogate chemicals with little consideration for similarity in chemical structure, but USEPA now provides guidance on surrogate similarity. Consult a professional toxicologist before a chemical is used as a surrogate.

USEPA’s STSC (USEPA 2013h) uses quantitative structure-activity relationship models to compare chemical structures, determine the similarity between chemicals, and provide surrogate chemical toxicity data. USEPA’s regional screening levels (RSL) table (USEPA 2014e) incorporates the surrogate chemical toxicity data that have been endorsed by the STSC. For chemicals not on the RSL table and for chemicals on the table with missing toxicity values, the STSC can be queried directly. USEPA’s RSL table is updated approximately every six months, so the most recent version of the RSL table should be checked for current toxicity values to use as surrogates. Another source of toxicity values is the Oak Ridge National laboratory (ORNL) Risk Assessment Information System (RAIS) database (ORNL 2014). RAIS is updated as new information becomes available.

5.1.2.4 Option – Use Oral Toxicity Values for Inhalation Exposures or Vice-Versa

Toxicity values derived using route-to-route extrapolation may be reasonable in limited cases. For example, it may be possible to extrapolate an inhalation toxicity value from an oral toxicity value and vice-versa. In general, however, route-to-route extrapolation is discouraged because the pharmacokineticStudy of the absorption, distribution, metabolism, and excretion of chemicals and the genetic, nutritional, behavioral, and environmental factors that modify these parameters (Commission1997a). differences between the two routes of exposure may be overlooked. Route-to-route extrapolation also adds additional uncertainty to the risk estimates (USEPA 2009a). The inhalation dosimetry methodology (USEPA 1994b) provides specific examples of situations in which route-to-route extrapolation from oral toxicity values might not be appropriate.

5.1.2.5 Option – Consult Experts in Toxicology

Should none of the potential options above help to address the lack of toxicity information for a particular chemical of interest, the STSC (USEPA 2013h) or a professional toxicologist can be consulted.

5.1.3.1 Option – Use USEPA’s Default Approach, Additivity

The recommendations from USEPA’s first mixtures risk assessment guidance (USEPA 1986) were incorporated into RAGS, Part A (USEPA 1989a), which still serves as general guidance for risk assessments today. While this guidance acknowledges the four different types of chemical interactions listed above, it recommends dose addition (for noncarcinogens) and response addition (for carcinogens) as the default approaches for addressing chemical mixtures in risk assessment.

Dose addition is the most commonly used mixtures approach and is the assumed approach underlying the hazard indexThe sum of more than one hazard quotient for multiple substances and/or multiple exposure pathways. The hazard index is calculated separately for chronic, subchronic, and shorter-duration exposures (USEPA 1989a)., the relative potency factorA scaling factor that represents the relative toxicity of a chemical based on the toxicological dose-response data of an index chemical. (RPF), and toxic equivalency factor (TEF) approaches discussed below. Dose addition assumes toxicological similarity between the chemical components of a mixture (that they have similar mechanisms for exerting toxicity and similarly shaped dose-response curves), and that they differ only in potency. With dose addition, the toxicological effect of the mixture is predicted by the sum of the individual chemical doses (adjusted for potency).

Response addition assumes that the chemical components of a mixture act on different organs and systems or produce effects that do not influence each other (USEPA 2000c). This form of additivity is most commonly used to estimate risks from exposure to multiple carcinogens, where the cumulative cancer risk estimate is the sum of the probabilities associated with exposure to each carcinogenic chemical.

USEPA’s supplementary mixtures guidance provides additional scientific updates and methods for addressing different types of chemical mixtures in human health risk assessments (USEPA 2000c). Although USEPA recognizes the need to better understand the nature and toxicological significance of chemical mixtures, additivity, which implies a lack of interaction, remains the current default approach.

5.1.3.2 Option – Use an RPF Approach

The RPF approach relies on the existence of toxicological dose-response data for at least one chemical of the mixture (referred to as the index chemical), the toxicity of the other individual chemicals in the mixture, and the toxicity of the mixture as a whole. The toxicity of the related chemicals is predicted from the index chemical, and a scaling factor for the toxicity of the related chemicals is used. This scaling factor (the RPF) represents the relative toxicity with respect to the index chemical. For example, if chemical A is considered to be one-tenth as toxic as the index chemical (it requires 10 times the exposure to cause the same toxicity), then the RPF for chemical A is 0.1. USEPA currently uses the RPF approach for assessing cancer risk for PAH mixtures and for cumulative risk assessments of five groups of pesticides (for example, organophosphates).

5.1.3.3 Option – Use a Toxic Equivalency Factor Approach Where Applicable

The TEF approach is a form of the RPF approach, but is reserved for groups of chemicals for which more robust information is available about the mechanisms of chemical toxicity. “TEFs are consensus estimates of chemical-specific toxicity/potency relative to the toxicity/potency of an index chemical. TEFs are the result of expert professional judgment using all available data and taking into account uncertainties in the available data” (USEPA 2010d). Two classes of chemicals for which TEF approaches have been developed by USEPA and other agencies are dioxins/furans and polychlorinated biphenyls.

5.1.3.4 Option – Consult Experts in Toxicology

Should none of the potential options above help to address how to evaluate the particular chemical mixtures of interest, the STSC (USEPA 2013h) or a professional toxicologist could be consulted.

5.1.4.1 Option – Use a General Approach for all Mutagenic Carcinogens as Appropriate

For most carcinogens with a potential mutagenic MOA, chemical-specific age-dependent adjustment factors (ADAFs) are not available (USEPA 2013a). In the absence of chemical-specific ADAFs, USEPA recommends the use of default ADAFs when calculating cancer risk estimates for exposures during early life (USEPA 2005d):

• an ADAF of 10 (10 times higher toxicity) for ages under 2 years
• an ADAF of 3 (3 times higher toxicity) for ages over 2 and under 16 years
• an ADAF of 1 (no adjustment) for ages 16 years and older

USEPA’s Handbook for Implementing the Supplemental Cancer Guidance at Waste and Cleanup Sites (USEPA 2012b) provides details on applying recommended potency adjustment factors in the risk calculations.

5.1.4.2 Option – Use a Chemical-Specific Approach where Chemical-specific Adjustments are Warranted

Currently, USEPA recommends applying chemical-specific modifications to the risk estimates for two mutagens: TCE and vinyl chloride. Nonstandard ingestion and inhalation risk equations apply to TCE and vinyl chloride. For all other carcinogens with a potential mutagenic MOA, the default ADAFs specified in USEPA guidance are used (USEPA 2005d).

Mutagenic MOA Adjustments for TCE and Vinyl Chloride

For mutagenic MOA adjustments for TCE and vinyl chloride, as explained in detail (with example calculations) in the IRIS Toxicity Assessment for TCE (USEPA 2011d), USEPA recommends that kidney risk be assessed using the mutagenic equations and that liver and non-Hodgkin lymphoma be addressed using standard cancer equations. Not applying this modification results in an overestimate of exposure risks.

The following table presents details on calculating cancer risk estimates for ingestion and inhalation of TCE in soil, accounting for contributions to kidney cancer (adjusted using ADAFs) and non-Hodgkin lymphoma (NHL) + liver cancer (not adjusted using ADAFs). For more information, see the Toxicity Review (USEPA 2011g) and the RSL User’s Guide (USEPA 2014e).

Table 5-1. Cancer risk estimates for residential soil ingestion

Residential soil ingestion
	Age group
	0 to 2	2 to 6	6 to 16	16 to 30
Soil concentration (mg/kg)	1	1	1	1
Ingestion rate (mg/day)	200	200	100	100
Conversion factor (kg/mg)	1.00E-06	1.00E-06	1.00E-06	1.00E-06
Exposure frequency (days/year)	350	350	350	350
Exposure duration (years)	2	4	10	14
Body weight (kg)	15	15	70	70
Averaging time (days/year)	25,550	25,550	25,550	25,550
Lifetime average daily dose (mg/kg/day)	3.65E-07	7.31E-07	1.96E-07	2.74E-07
Oral slope factorAn upper bound, approximating a 95% confidence limit, on the increased cancer risk from a lifetime exposure to an agent. This estimate, usually expressed in units of proportion (of a population) affected per mg/kg-day, is generally reserved for use in the low-dose region of the dose-response relationship, that is, for exposures corresponding to risks less than 1 in 100 (USEPA 2013). (mg/kg/day)-1	0.0460	0.0460	0.0460	0.0460
SF for kidney cancer* (mg/kg/day)-1	0.0093	0.0093	0.0093	0.0093
SF for NHL + liver cancer (mg/kg/day)-1	0.0367	0.0367	0.0367	0.0367
ADAF (unitless)	10	3	3	1
Risk of kidney cancer (unitless)	3.4E-08	2.04E-08	5.46E-09	2.55E-09
Risk of NHL + liver cancer (unitless)	1.34E-08	2.68E-08	7.18E-09	1.01E-08
Sum of risks (unitless)	4.74E-08	4.72E-08	1.26E-08	1.26E-08
Total risk	1.2E-07

*From Section 5.2.3.3.2 of the September 2011 Toxicological Review of TCE

5.1.4.3 Option – Determine whether Mutagenic MOA is Appropriate for Site

At sites where receptors under age 16 are not exposed, potency adjustments for early life exposures are unnecessary. In addition, at sites where mutagens are not detected, adjustments are not applicable. Currently some state agencies and programs do not require or incorporate adjustments for mutagenic MOA. In these cases, at sites where receptors under age 16 may be exposed and chemicals with a mutagenic MOA are present, risk estimates for mutagenic carcinogens may be underestimated.

5.1.4.4 Option – Consult Experts in Toxicology

Should none of the potential options above help to address how to evaluate the particular mutagenic chemical, the STSC (USEPA 2013h) or a professional toxicologist can be consulted.

Lead risk assessment is unique because scientific research has linked adverse human health effectsTypically defined as an incremental lifetime cancer risk (for example, exceeding a range of 1E-4 to 1E-6) or a hazard quotient or hazard index (for example, one). to blood lead levels (BLLs) rather than a dose rate. A child’s developing nervous system is particularly sensitive to lead, and increasing BLLs have been linked to a variety of cognitive deficits in children. Once lead enters the body, it can spread into bone and soft tissues, which act as reservoirs that release lead over time, even after the original source of lead exposure has been removed. For these reasons, several mathematical models have been developed to estimate BLLs based on lead intake via various exposure routes. Although a detailed discussion of these models is beyond the scope of this document, the following links are provided for additional information:

• USEPA’s Integrated Exposure Uptake Biokinetic Model (IEUBKwin v1.1 build 11)  (USEPA 2010a)
• USEPA’s Adult Lead Methodology (ALM) Model   (USEPA 2003a)
• California EPA’s LeadSpread 8 Model  (DTSC 2011a)
• USEPA’s draft All-Ages Lead Model (USEPA 2005a)

The issue often arises as to which receptorAn individual (for example, residential adult, residential child, worker, trespasser, or recreator) who has the potential to be exposed to a chemical in environmental media. groups should be modeled for specific land use scenarios. Rather than spend resources modeling all potential receptor groups, options are provided below based on the most sensitive receptor groups, which drive site cleanupThe assessment and reduction, removal, or control of chemicals in environmental media. Cleanup is synonymous with other terms such as "corrective action" and "remediation" used in various state, local, and federal programs. for residential and industrial land uses.

5.1.5.1 Option – Model Child Exposures When Assessing Residential Scenarios

When modeling child exposures for lead, an important consideration is which BLL to use. Long-standing USEPA policy stipulates that the probability of a child’s BLL exceeding a 10 µg/dL BLL of concern should fall below 5%. This criterion is the basis for the 400 mg/kg residential USEPA RSL for lead that USEPA has used for many years (USEPA 1994a).

In January 2012, the CDC’s Advisory Committee for Childhood Lead Poisoning Prevention (ACCLPP) released a report regarding child BLLs. ACCLPP recommended eliminating the use of the term “blood lead level of concern” and replacing it with the term “blood lead reference level” (CDC 2012b; CDC 2012a). This committee also recommended the use of a blood lead reference level of 5 µg/dL (as opposed to the previously recommended 10 µg/dL) in children to trigger medical and prevention actions. The newly proposed BLL is based on the 97.5^th percentile of BLL distributionA distribution describes the probability or likelihood of any potential value. among children ages one to five from the National Health and Nutrition Examination Survey (NHANES) data. In May 2012, CDC concurred in principle with the ACCLPP’s January 2012 recommendations (CDC 2012a).

As of November 2013, USEPA has not changed the risk reduction goal for lead. The USEPA’s current risk reduction goal for contaminated sites is to limit the probability of a child’s BLL exceeding 10 µg/dl to 5% or less (USEPA 2015f).

In 2007, California EPA revised its lead policy as a result of the concerns about potential toxicity to children at BLLs less than 10 µg/dL. California EPA’s Office of Environmental Health Hazard Assessment relied on data from scientific studies that quantified the relationship between BLLs in children and IQ scores. The Office of Environmental Health Hazard Assessment concluded that a 1 µg/dL increase in BLL corresponds to a 1 point decrease in IQ and selected 1 µg/dL as a benchmark BLL change due to a specific exposure source (California Environmental Protection Agency 2007). In current practice, the value for BLL in children that is typically used is 10 mg/dL, unless the state regulatory agency recommends another value.

Another consideration is which model to use. In current practice, USEPA’s IEUBK model for lead in children is the most common model used to evaluate lead exposures in residential settings and to establish lead remedial action levels (USEPA 2007f). This model consists of four components (exposure, uptake, biokineticsMovement of a chemical (for example, absorbed lead) throughout the body by physiologic or biochemical processes., and variabilityA population’s natural heterogeneity or diversity, particularly that which contributes to differences in exposure levels or in susceptibility to the effects of chemical exposures (Commission 1997a). For example, workers may perform different functions that may affect time, frequency, and duration of contact with an environmental medium). Variability cannot be reduced by collection of additional data.) that can be adjusted based on site- or scenario-specific conditions. Through adjustments of these four components, the model estimates a distribution of BLLs for a hypothetical child or group of children.

Although IEUBK is typically used, states may specify other models. California EPA has its own biological model (LeadSpread) for evaluating childhood lead exposures. This model uses the 1 µg/dL change in BLL benchmark with other exposure and variability factors to estimate the highest soil concentration predicted to result in a 1 µg/dL change in the 90^th percentile of the population of exposed children. This soil lead screening value is 80 mg/kg (California Environmental Protection Agency 2009).

5.1.5.2 Option – Model Adult Worker Exposures when Assessing Nonresidential Scenarios

USEPA’s recommended approach for evaluating adult exposure to lead in soil is detailed in USEPA’s Recommendations of the Technical Review Workgroup for Lead for an Approach to Assessing Risks Associated With Adult Exposure to Lead in Soil (USEPA 2003e). In this guidance, USEPA describes a method for assessing potential risks associated with nonresidential adult exposures to lead in soil. The method focuses on estimating fetal BLLs in women exposed to lead in soil. Rather than recommend a single soil concentration that would represent an acceptable risk for adults exposed in nonresidential settings, a range of 750 mg/kg to 1,750 mg/kg is presented; see Figure 2 of the 2003 ALM guidance (USEPA 2003a).

In 2009, USEPA published a memorandum which provided updated estimates for baseline BLLs and the geometric standard deviation using data from NHANES studies that were conducted from 1999–2004 (USEPA 2009c). As a result of these updated inputs, acceptable soil lead concentrations were revised to a range of 750 mg/kg to 2,240 mg/kg (see USEPA 2014g for additional information).

USEPA Adult Lead Methodology for Soil – Model Input and Output

The following table presents the assumptions, inputs, and results of USEPA’s current recommended soil lead model for adults and shows the source of the acceptable soil lead concentration range noted above (USEPA 2003a). Alternative inputs or assumptions from specific agencies can also be incorporated into the model.

Table 5-2. Results of USEPA recommended soil lead model

Input	Variable	Units	USEPA (2003a) Pregnant workers		USEPA (2009b) Pregnant workers
Soil Ingestion – Model Input Values
Gastrointestinal absorption	AF	unitless	0.12	0.12	0.12
Ingestion rate	IR	mg-soil/day	50	50	50
Exposure frequency	EF	days/year	219	219	219
Averaging time	AT	days/year	365	365	365
Typical adult blood lead concentration	PbB0	μg-Pb/dL	2.2	1.7	1.0
Biokinetic slope factor	BKSF	μg-Pb/dL per μg-Pb/day	0.4	0.4	0.4
Geometric standard deviation	GSD	–	2.1	1.8	1.8
Ratio of fetal PbB to maternal PbB	Rf/m	unitless	0.9	0.9	0.9
Fetal blood goal, 95th percentile	PbBgoal	μg-Pb/dL	10.0	10.0	10.0
Soil Concentrations – Model Output Values
Soil Pb concentrations, 95th percentile value	–	mg/kg	749	1,754	2,240

5.1.5.3 Option – Assess Child or Adult Intermittent or Variable Exposures to Lead where Applicable

USEPA’s IEUBK model (USEPA 2010a) and USEPA’s ALM (USEPA 2003a) are commonly used to evaluate standard residential or continuous nonresidential exposure scenarios. These models may warrant modifications when the exposures are not continuous or chronic. Such scenarios may include trespassing, recreational, and daycare exposure scenarios. In these cases, a wider variety of exposure scenarios may need to be considered (such as including exposure from more than one location, or different intensities of exposure). USEPA’s Assessing Intermittent or Variable Exposures at Lead Sites (USEPA 2003b) provides guidance on using the IEUBK and ALM models to assess exposures using a time-weighted exposure. The guidance presents methods, assumptions, limitations and uncertainties associated with this approach, and example calculations are provided.

5.1.5.4 Option – Consult Experts in Toxicology

Should none of the potential options above help to address how to evaluate exposure to lead, the STSC (USEPA 2013h) or a professional toxicologist can be consulted.

Generally, there is a higher degree of certainty in toxicity values from higher tier sources (Tier 1 versus Tier 3 sources, see Section 5.1.1.1), and a higher degree of certainty in chemical-specific versus surrogate chemical toxicity data. In addition, uncertainty factors that accompany the noncancer toxicity values indicate the degree of confidence to which the values are considered protective of human health. For example, if the risk estimates for a site exceed agency-acceptable levels due to only one chemical, no chemical-specific toxicity values were available for that chemical, and toxicity values for a surrogate chemical were used in the risk estimates, then a high degree of uncertainty is likely present in the risk estimates.

Chemicals classified as carcinogens have chemical-specific carcinogenic potential, and the studies on which their carcinogenic classification is determined vary in quality. The USEPA has established weight-of-evidence classifications, with the most recent in (USEPA 2005b). For example, risk estimates associated with benzene may warrant a higher sense of urgency than chloroform. According to USEPA’s IRIS database (USEPA 2000a):

Benzene is a known human carcinogen based upon evidence presented in numerous occupational epidemiological studies. Significantly increased risks of leukemia, chiefly acute myelogenous leukemia (AML), have been reported in benzene-exposed workers in the chemical industry, shoemaking, and oil refineries.

Conversely, USEPA’s IRIS database (USEPA 2001a) lists chloroform as a probable human carcinogen based on observations of increased incidence of cancer in mice and rats exposed to this chemical and inadequate evidence of increased cancer in humans in human epidemiological data and studies.

As another example, hazard index estimates associated with the soil ingestion pathway may warrant a higher sense of urgency for arsenic than thallium due to the large differences in uncertainty in the toxicity values for the two chemicals. The oral RfD for arsenic from the Tier 1 source (USEPA 1993a) is based on human chronic oral exposure studies and has a very low uncertainty factor (3), whereas the oral RfD for thallium from the Tier 2 source (USEPA 2012g) is a provisional screening value based on a 90-day rat oral exposure study and has a very high uncertainty factor (3000). The PPRTV derivation document for thallium (USEPA 2012g)  states the following:

The conclusion reached in the IRIS Toxicological Review of Thallium and Compounds (USEPA 2009a) was that the available toxicity database for thallium contains studies that are generally of poor quality…However, Appendix A of this document contains Screening Values (screening subchronic and chronic p-RfD) that may be useful in certain instances.

5.2.1.1 Option – Review the Risk Characterization to Obtain Uncertainty Information for Chemicals Having the Most Influence on the Risk Estimates

The uncertainty information for the toxicity assessment can be obtained from the last step of the risk assessment (see Chapter 7), which should include the risk estimates for the site and a discussion of the overall uncertainties associated with the risk estimates. In addition, if risk estimates exceed agency-acceptable levels, a discussion of the uncertainties in the toxicity values for the risk-driving chemicals should also be presented in the risk characterizationThe risk characterization integrates information from the preceding components of the risk assessment and synthesizes an overall conclusion about risk that is complete, informative and useful for decision makers (USEPA 2000c).. Section 8.4.2 of USEPA's 1989 guidance (USEPA 1989a) presents the recommended types of uncertainty information to be included in the risk characterization section. Page 8-24 of the USEPA guidance (USEPA 1989a) provides a checklist of the uncertainties that apply to most toxicity assessments.

5.2.1.2 Option – Review Uncertainty Information in the Toxicity Assessment Section of the Risk Assessment

The uncertainty information for the toxicity assessment can be obtained from the toxicity assessment section of the risk assessment, which should cross-reference the specific tables containing the toxicity values for the chemicals evaluated in the risk assessment, including the weight-of-evidence classification for carcinogens and the uncertainty factors for noncarcinogenic toxicity values. When using this information, focus on the chemicals (and toxicity values, cancer classifications, and uncertainty factors) having the most influence on the risk estimates.

5.2.1.3 Option – Consult a Risk Assessor or Toxicologist

If the risk characterization and the toxicity assessment sections do not include uncertainty information associated with the toxicity of chemicals having the most influence on the risk estimates, or if it is unclear where the uncertainty information is, consult with a risk assessor or toxicologist.