Problem Formulation to Dose-Response: Advances via the Alliance for Risk Assessment Beyond Science and Decisions Workshops

Michael Dourson Toxicology Excellence for Risk Assessment

1

Alliance for Risk Assessment

www.allianceforrisk.org

• A collaboration of organizations dedicated working together to solve public health issues – Improve communication among groups – Provide transparency in development of products – Foster harmonization and consistency in risk assessments – Share costs and human resources 2

Collaborators for “Beyond Science” Workshops

3

ARA Science Panel

• • • • • • • • • • • • • Michael Bolger, U.S. Food and Drug Administration James S. Bus, The Dow Chemical Company John Christopher, CH2M/Hill Rory Conolly, U.S. Environmental Protection Agency Michael Dourson, Toxicology Excellence for Risk Assessment *Adam M. Finkel, UMDNJ School of Public Health William Hayes, Indiana DEM R. Jeffrey Lewis, ExxonMobil Biomedical Sciences, Inc.

Randy Manning, Georgia Department of Natural Resources Bette Meek, University of Ottawa (Chairperson) Paul Moyer, Minnesota Department of Health (MDH) *Greg Paoli, Risk Sciences International Rita Schoeny, U.S. Environmental Protection Agency • *On NAS Science and Decisions panel 4

Case Study Process

• • • • • • • Engagement from wide variety of stakeholders Proposed in brainstorming prior to first workshop Initial vetting in breakout groups at 1 st workshop Presentations at 2 nd workshop Additional case studies identified at 2 nd workshop 30+ case studies proposed 24 case studies presented and reviewed by panel 5

Case Study Process & Dose Response Framework

• Organization of methods and ability to identify gaps into an interactive framework based on NAS (2009):

Problem formulation DR method Management decision

• The framework was developed by the panel after review of all case studies in the 2 nd workshop, and was used to prioritize new case studies for 3 rd workshop, focusing on 3 topic areas: – Problem formulation – Mode of action – Endogenous & background exposures 6

Science and Decisions, NAS (2009) Framework for Risk Assessment

• • •

Formulate Problem

Determine problems with conditions Determine options for change What assessments are necessary for risk options?

• • •

Plan & Conduct

Plan assessment Conduct: –

Hazard ID

–

DR assessment

– – Exposure assessment Risk characterization Confirm Utility • • •

Manage Risk

Determine: – – Option benefits How options affect other decision factors Justify decision re benefit, cost & uncertainty Communicate decision

Alliance for Risk Assessment

(ARA)

: Beyond Science and Decisions Workshop Series

PHASE 1: Problem Formulation & Scoping [

Adapted from NAS (2009) Figure S-1] What problems are associated with existing environmental conditions?

If existing exposure conditions appear to pose a threat to human or environmental health, what options exist for altering those conditions?

Under the given decision context, what risk and other technical assessments are needed to evaluate possible risk management options?

Qualitative Decision Quantitative Screening In-Depth Assessment

Health Assessment

Use available data to assist management decision

Qualitative Decision Exposure and Endpoint Assessment

• Identify potential health effects • Consider strengths and uncertainties in data • Identify potential exposure scenarios

ARA: Beyond Science & Decisions

Vulnerable Populations Assessment

Use available data to assist management decision

Exposure Assessment

Use available data to assist management decision

Integration Results Reporting

Quantitative Screening Decision

ARA: Beyond Science & Decisions

Exposure and Endpoint Assessment

• Identify adverse effects and chemical mode of action • Determine strengths and uncertainties in data • Define exposure scenarios; get data on exposed populations

Health Assessment

Use available data to determine critical effect & action mode

Vulnerable Populations Assessment

Use available data to determine potential groups at risk

Exposure Assessment

Use available data to determine upper bound exposures

Exposure & Dose-Response Evaluation

Based on the available information, estimate a health-protective exposure limit

Results Reporting

Quantitative Screening Decision

• Tiered approach case study • Low-dose Extrapolation from BMD(L) • Threshold of toxicological concern/ of regulation • Screening-level safe dose (e.g., RfD) • Structure-activity relationships and read-across Quantitative SAR 11

In-Depth Assessment

ARA: Beyond Science & Decisions

Exposure and Endpoint Assessment

• Identify adverse effects and their precursors and MOA • Identify exposures, endpoints or lifestages under-assessed • Identify probabilistic exposure scenarios focusing on vulnerable populations

Health Assessment

Chose appropriate extrapolation based on MOAs and background disease

Vulnerable Populations Assessment

Identify vulnerable groups, considering exposures, endpoints and MOA

Exposure Assessment

Identify endogenous exposures & conduct probabilistic exposure scenarios

Risk Characterization

Integrated extrapolation with probabilistic exposure based on vulnerable populations

Communicate characterization with uncertainties

Dose Response Framework

The risk assessor is guided to methods that address key issues, such as: – Mode of action assessment – Vulnerable population assessment – Endogenous/background exposure – Dose-response methods reflecting different • Conceptual models • Data availability • Risk management needs 13

Methods Linked to Real World Application

• Summaries that – briefly describe dose response method, – provide references, – outline data requirements, – describe strengths and weaknesses • In depth full case study • Workshop presentation slides 14

The Expert Panel Determined…

• A wide range of problem formulations exist for which different dose-response analysis techniques are needed.

• Risk assessors must explain criteria applied in the choice of a particular dose-response method, and how results will be used in a risk management decision.

• Additional case studies would be useful on topics such as: • Combined exposures • • Value of information Illustrating an entire risk assessment, from problem formulation to conclusion • In vitro to in vivo extrapolation 15

Next Steps

• Framework will be “evergreen” – – Updated with additional methods illustrated by case studies, and Papers developed addressing & resolving cross-cutting issues.

• The National Library of Medicine is hosting the Framework.

• A standing panel will be created to meet twice a year to review additional case studies and issue/resolution papers.

• Additional sponsors/participants will be invited to join in the overall effort.

16

Framework

• •

ARA Dose Response Framework – (working beta)

http://www.allianceforrisk.org/workshop/fra mework/ problemformulation.html

Part 2 of the symposium presents several sample methods and case studies for risk around the RfD 17

0.1

ARA

Cases on Risk Around the

1

RfD

Areas of Uncertainty to Consider in Noncancer Dose Response Assessment

Sub-chronic Animal Chronic Human Chronic Animal Reproductive UF H PBPK UF A

LOAELs

UF L UF D UF S

Dose NOAELs or BMDs

Case Study 17:

Oliver Kroner & Lynne Haber, TERA *This case study is a characterization of the method, and is not intended as endorsement or opposition of linear extrapolation. Methods 1 to 4: extend a straight line from the chosen BMD or BMDL adjusted to the human equivalent dose or concentration (HED or HEC) by default or modified uncertainty factors. Method 5: linearize HED(C) dose-response data using probit transformation in logarithmic space. Fit regression line to the data and extend to the low-dose

Method 1: Linear extrapolation from BMD

0.1

Human Equivalent Dose UF A UF D UF S Animal BMD

Dose

Factor of 10 Enough?

Dourson, M.L., G. Charnley and R. Scheuplein, 2002

Oral Chemical

Acrylamide

Summary of Results

Method 1 and 4 Linear Extrapolation from BMD(L) Risk at RfC/RfD

From BMD(C)L (Method 1)

Risk at RfC/RfD

From BMD(C) (Method 4)

Log-Dose, Probit Risk at RfC/RfD Method 5 Number of Dose Groups (other than control)

1 x 10 -2 3 x 10 -4 1 x 10 -3 3 Chlordecone 1 x 10 -2 1 x 10 -5 2 x 10 -3 4 1,3 Dichloropropene

Inhalation

Nitrobenzene 8 x 10 -3 1 x 10 -2 6 x 10 -4 4 x 10 -3 2 x 10 -12 5 x 10 -1 3 3

Strengths and Limitations

• Strengths • Simple to use and provides risk at any dose • Limitations • No consideration of underlying Mode of Action • Depending on UF, risk could be highly conservative • ARA Science Panel Comments •

Possibly useful for screening, but should not

•

be construed as accurate estimate of risk

Requested exploration of non-cancer linear extrapolation in log-dose, probit space

Probit Transformation

• Linearizes biological data • Requires population data • To allow graphing in log space, response rates were converted to Excess Risk [added risk(d) = P(d) - P(0)] for each dose group • a dataset of at least three test doses above the control From Casarett & Doull 2009

Method 5: Linear Extrapolation in Log-Dose, Probit Space

5

Human Equivalent Dose UF A UF D UF S Animal Data

Log Dose

Acrylamide Effect:

Nerve Degeneration

Species:

Rat

RfD =

0.002 mg/kg-day Log(RfD) = -2.69

Uncertainty Factor:

30 (A-3; H-10)

Probit Response Above Control v. Log(Human Equivalent Dose) (mg/kg)

Log(10) Dose

Probit at RfD =

1.9

Risk at the RfD =

1 x 10 -3

Oral Chemical

Acrylamide

Summary of Results

Method 1 and 4 Linear Extrapolation from BMD(L) Risk at RfC/RfD

From BMD(C)L (Method 1)

Risk at RfC/RfD

From BMD(C) (Method 4) 1 x 10 -2 3 x 10 -4

Method 5 Log-Dose, Probit Risk at RfC/RfD

1 x 10 -3

Number of Dose Groups (other than control)

3 Chlordecone 1 x 10 -2 1 x 10 -5 2 x 10 -3 4 1,3 Dichloropropene

Inhalation

Nitrobenzene 8 x 10 -3 1 x 10 -2 6 x 10 -4 4 x 10 -3 2 x 10 -12 5 x 10 -1 3 3

Strengths and Limitations

• Strengths • Simple, and provides risk at any dose • Consistent with toxicological theory of probit/logarithmic dose response • Limitations • Restrictive data requirements permitted the use of only 4 of 25 chemicals; such restriction could be relaxed • Differing results may be due to expected differences among chemicals in adverse responses, or different amounts of dose response data.

Case Study 11:

Elizabeth Spalt, IDEM & Oliver Kroner, TERA • Straightforward application of Swartout et al. (1998). • A single distribution is assumed for all factors, specifically, lognormal with a median of 10 0.5

and a 95 th % value of 10. • Various probabilities of Swartout et al. (1998) are combined by multiplication. Other combinations may be possible.

• Method indirectly addresses recommendation of NAS (2009) to develop probabilities for RfDs, but probabilities state

whether RfD is correctly identified as a sensitive human NOAEL or BMDL.

Comparison of RfD Values for Three Compounds with an IRIS RfD of 0.03

Strengths and Limitations

• Strengths – Method shows how different factors result in RfDs with different probabilities of being correctly identified as a sensitive human NOAEL.

– Method is straightforward and simple and can be used to judge among given RfDs or used to standardize all RfDs.

• Limitations – Method assumes a similar distribution for all factors; although conservative, it may not represent all chemicals.

– The probability is the likelihood that the stated RfD is a sensitive human NOAEL, rather than describing the probability of a response in a population.

Case Study #21

Robinan Gentry & Cynthia Van Landingham, Environ;Lesa Aylward & Sean Hays, Summit • • • Extension of the Benchmark Dose (BMD) method Development of risk values at doses above the Reference Dose (RfD) Methylmercury – Dose-response information in humans – BMDs estimated using biomarkers (i.e., levels in hair and cord blood) – Multiple BMDs available – Sensitive human subpopulation (children exposed in utero)

4 Approaches

• Approach 1 - A straight line is drawn from both the BMDL and BMD to the RfD, where the RfD is considered to be zero risk; • Approach 2 - The appropriate BMD model is extrapolated to the RfD and then the risk at the RfD is truncated to zero; • Approach 3 - The appropriate BMD model is extrapolated to the RfD and this risk is allowed to stand as an upper bound; • Approach 4 - The appropriate BMD model is extrapolated using a threshold term, where the threshold value is judged to be the RfD, or some higher value.

3 BMDL 3 BMD 1 BMDL 2 BMDL 1 BMD 2 BMD

Strengths and Limitations

• •

Strengths:

– Use of a biomarkers, typically closer to the “target tissue” – Ability to evaluate the potential fraction of people exposed above and below the RfD to assess the likelihood of adverse effects – internal concentration may be extended to an exposure level

Limitations:

– Uncertainties for other compounds as to the relationship between biomarker and effects of concern – Information characterizing the potential shape of the dose-response curve below the BMD/BMDL

Summary

• • • • A wide range of problems exist for which different dose response methods and case studies are needed: – – – Combined exposures Value of information Illustrating an entire risk assessment – In vitro to in vivo extrapolation Assessors must explain choice of a particular method and its usefulness in a management decision.

A standing panel is being formed to review additional methods, case studies and issue/resolution papers.

Additional folks are invited to join the effort.

Extra Slides

38

39