FIFRA Scientific Advisory Panel Meeting
September 5-6, 2001
Held at the Sheraton Crystal City Hotel, Arlington, Virginia
A Set of Scientific Issues Being Considered by the Environmental Protection Agency Regarding:
Preliminary Cumulative Hazard and Dose-Response Assessment for Organophosphorus Pesticides: Determination of Relative Potency and Points of Departure for Cholinesterase Inhibition
|Mr. Larry Dorsey||Ronald J. Kendall, Ph.D.|
|Designated Federal Official||Chair|
|FIFRA Scientific Advisory Panel||FIFRA Scientific Advisory Panel|
|Date: September 11, 2001||Date : September 11, 2001|
This report has been written as part of the activities of the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), Scientific Advisory Panel (SAP). This report has not been reviewed for approval by the United States Environmental Protection Agency (Agency) and, hence, the contents of this report do not necessarily represent the views and policies of the Agency, nor of other agencies in the Executive Branch of the Federal government, nor does mention of trade names or commercial products constitute a recommendation for use.
The FIFRA SAP was established under the provisions of FIFRA, as amended by the Food Quality Protection Act (FQPA) of 1996, to provide advice, information, and recommendations to the EPA Administrator on pesticides and pesticide-related issues regarding the impact of regulatory actions on health and the environment. The Panel serves as the primary scientific peer review mechanism of the EPA, Office of Pesticide Programs (OPP) and is structured to provide balanced expert assessment of pesticide and pesticide-related matters facing the Agency. Food Quality Protection Act Science Review Board members serve the FIFRA SAP on an ad-hoc basis to assist in reviews conducted by the FIFRA SAP. Further information about FIFRA SAP reports and activities can be obtained from its website at http://www.epa.gov/scipoly/sap/ or the OPP Docket at (703) 305-5805. Interested persons are invited to contact Larry Dorsey, SAP Executive Secretary, via e-mail at dorsey.larry@.epa.gov.
Federal Insecticide, Fungicide, and Rodenticide Act
Scientific Advisory Panel Meeting
September 5-6, 2001
A Set of Scientific Issues Being Considered by the Environmental Protection
Regarding: Preliminary Cumulative Hazard and Dose-Response Assessment for Organophosphorus Pesticides: Determination of Relative Potency and Points of Departure for Cholinesterase Inhibition
FIFRA Scientific Advisory Panel Chair
Ronald J. Kendall, Ph.D., Professor and Chairman, Department of Environmental Toxicology and Director, the Institute of Environmental and Human Health, Texas Tech University/ Texas Tech University Health Sciences Center, Lubbock, TX
FIFRA Scientific Advisory Panel
Herbert Needleman, M.D., Professor of Psychiatry and Pediatrics, School of Medicine, University of Pittsburgh, Pittsburgh, PA.
Christopher J. Portier, Ph.D., Director, Environmental Toxicology Program, National Institute of Environmental Health Sciences, Research Triangle Park, NC
Stephen M. Roberts, Ph.D., Professor and Program Director, University of Florida, Center for Environmental & Human Toxicology, Gainesville, FL
Mary Anna Thrall, D.V.M., Professor, Department of Pathology, College of Veterinary and Biomedical Sciences, Colorado State University, Fort Collins, CO
FQPA Science Review Board Members
Rory Conolly, Sc.D., Senior Scientist, CIIT, Center for Health Research, Research Triangle Park, North Carolina
Patrick Durkin, Ph.D., Vice President, Syracuse Environmental Research Associates, Fayetteville, NY
Dale Hattis, Ph.D., George Perkins Marsh Institute, Clark University, Worchester, MA
Ernest McConnell, D.V.M., President, Toxpath Inc., Raleigh, NC
Peter MacDonald, D. Phil., Professor of Mathematics and Statistics, Department of Mathematics and Statistics, McMaster University, Hamilton, Ontario, Canada
Carey Pope, Ph.D., Department of Physiological Sciences,, Oklahoma State University, Stillwater, OK
Nu-May Ruby Reed, Ph.D., Staff Toxicologist, California Environmental Protection Agency, Department of Pesticide Regulation, Sacramento, CA
Kendall B. Wallace, Ph.D., Professor, Department of Biochemistry and Molecular Biology, University of Minnesota School of Medicine, Duluth, MN
Designated Federal Official
Mr. Larry Dorsey, Executive Secretary, FIFRA Scientific Advisory Panel, Office of Science Coordination and Policy, Office of Prevention, Pesticides and Toxic Substances, Environmental Protection Agency, Washington, DC
Oral statements were made by:
Christopher F. Wilkinson, Ph.D., C. Wilkinson, LLC, representing the American Crop Protection Association
Robert L. Sielken Jr., Ph.D., Sielken and Associates Consulting, Inc., representing American Crop Protection Association
Edward C. Gray, Esquire, McDermott, Will and Emery, representing FQPA Implementation Working Group
Written statements were received from:
American Crop Protection Association
FQPA Implementation Working Group
The Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), Scientific Advisory Panel (SAP) has completed its review of the set of scientific issues being considered by the Agency regarding the end point selection and determination of relative potency in cumulative hazard assessment of organophosphorus pesticide chemicals. Advance notice of the meeting was published in the Federal Register on August 10, 2001. The review was conducted in an open Panel meeting held in Arlington, Virginia, on September 5-6, 2001. The meeting was chaired by Ronald J. Kendall, Ph.D., Mr. Larry Dorsey served as the Designated Federal Official.
The Food Quality Protection Act (FQPA) of 1996 requires the USEPA to consider the cumulative effect on human health that can result from exposure to pesticides and other substances. The Agency has addressed the SAP written comments on the approach to cumulative hazard and dose-response for the organophosphorus pesticides (OPs) that were outlined in the September 27, 2000, document entitled Endpoint Selection and Determination of Relative Potency in Cumulative Hazard and Dose-Response Assessment: A Pilot Study of Organophosphorus Chemicals. There are 25 OP pesticides included in the current analysis, nine of which have residential/nonoccupational exposure. Ethoprop, fenthion, chlorpyrifos, and dicrotophos were identified as appropriate for inclusion after the completion of this assessment and will be included in the cumulative risk assessment. The relative potency of these four OPs will be evaluated using the same methodology described for the 25 pesticides.
Relative potency factors of 25 OPs for the oral route of exposure as well as nine chemicals with residential/ nonoccupational exposure have been determined in addition to the endpoints needed for the extrapolation of cumulative risk to humans. The exponential function performed well in the determination of potency for cholinesterase inhibition for the oral studies. The use of NOAELs for potency determination for the dermal and inhalation routes, although not the preferred method, was adequate and generated reasonable estimates.
Background and introduction to the agency presentation were provided by Vicki Dellarco, Ph.D. and Anna Lowit, Ph.D.. Concepts and overview of the methodology was presented by Woodrow Setzer, Ph.D. and Stephanie Padilla, Ph.D. of the National Health and Environmental Effects Laboratory, Office of Research and Development.
The specific issues addressed by the Panel are keyed to the background document, Preliminary Cumulative Hazard and Dose Response Assessment for Organophosphorus Pesticides: Determination of Relative Potency and Points of Departure for Cholinesterase Inhibition, dated July 31, 2001, and are presented as follows.
Issue 1. Selection and Performance of Dose-Response Model
The Office of Pesticide Programs (OPP) used probit transformation to determine the relative potencies of organophosphorus pesticides in a pilot analysis presented to the FIFRA Scientific Advisory Panel (SAP) in September 2000. The SAP recommended that OPP reconsider the model selection and specifically suggested Michaelis-Menton kinetics or the exponential model as alternative methods. Both models were considered by OPP. The exponential model was selected over Michaelis-Menton kinetics because the exponential model provided a better fit of the available cholinesterase data.
An iterative approach was used to fit the cholinesterase data to the exponential function. This iterative approach included setting the y-asymptote to zero followed by sequentially dropping high doses until adequate model fit was achieved (p = 0.05) or only three dose groups remained. A total of 1312 individual cholinesterase data sets from oral toxicity studies were modeled. Although the toxicology studies used were not designed for purposes of dose response modeling, OPP concluded that the estimates derived from the exponential function were sufficient to determine relative potency of the organophosphorus pesticides.
Question 1.1 Does the revised dose-response approach to modeling the cholinesterase data address the SAP recommendations from the September 2000 meeting?
The EPA staff is to be congratulated on a skillful and creative implementation of the basic aspects of the risk modeling approach suggested at the September 2000 SAP meeting on this issue. The Panel consensus was that the major statistical issues raised at the previous meeting have been thoroughly addressed, although many of the Panelists recommended further exploration of modelling issues arising out of additional mechanistic considerations-especially those that could lead to different expectations for low dose relationships between dose and inhibition response. Aspects of the recommendations that EPA did and did not choose to address and some other modelling options that EPA may productively explore follow:
The two most important recommendations made by the year 2000 Panel were to attempt to fit a non-linear model instead of using the probit transformation and to examine numerous studies simultaneously rather than focusing on a single study. Both of these recommendations have been carried out, the first using weighted generalized least squares (as implemented in the R programming language), and the second by a random effects model and the methods of meta-analysis. Because each individual toxicology study has relatively few data points, the strength of the database comes from the large number of studies, hence the meta-analysis is an important contribution.
The Panel also commends the EPA on the use of R (see the main EPA report for references), as it is the best way to ensure portable, open code that is freely available to all interested users, with state-of-the art algorithms for statistical calculation.
The September 2000 SAP Report also advocated testing the "Goodness-of-Fit" of any model and investigating departures from the model that might indicate an inadequacy at extreme doses. This issue was addressed.
The September 2000 SAP Report advocated that data from individual animals be used. In the present analysis, the mean and standard deviation from each dose group was used. If data on individuals can be obtained, it should be made available and the analysis should take repeated measures into account. However individual data would not be likely to change the results using current methods. The use of individual animal data would likely allow the exploration of additional statistical methods, such as bootstrapping. The current approach is a form of meta-analysis as recommended by the previous Panel. However, one member noted that the previous Panel also discussed the use of a single model with individual effects for sex, species, site, and other explanatory variables (e.g. strain, age), which would have allowed for more rigorous testing of assumptions. This Panel member recommended that this approach be explored, in addition to or in replacement of this meta-analytic approach.
The 2000 SAP Report suggested that Physiologically-Based Pharmacokinetic (PBPK) models be explored in analysing these data. It was noted that these models have greater biological support than a simple exponential model and could possibly use both the short term exposure data and the sub-chronic and chronic exposure data. The Panel again recommends that such models be explored, either by the Agency in this analysis or by the general scientific community through release of the data for alternative analyses.
The Panel notes with appreciation that the EPA followed its recommendation to maintain separate relative potency determinations for separate compartments (e.g.,brain, red blood cells). The Panel urges EPA to proceed further to test the predictiveness of these separate indices for the relative potencies for causing specific effects in the central vs peripheral nervous systems.
The EPA staff evaluated the model forms suggested at the last SAP review on the basis of data concerns and chose one on the basis of some analysis (although this comparative analysis does not appear to have been described in detail). The Agency also adapted the previously suggested model form by adding a provision for incomplete inhibition of cholinesterase activity at the limit of high dose (designated by the term "B" in the modified equation used for fitting). This adaptation was generally supported by the Panel, but some members expressed reservations based on concerns that this adaptation focused the modelling effort on achieving fidelity with observations at the high end of the range of doses tested, to the likely detriment of fitting points at the low end of the dose response relationship. This concern is reinforced by an examination of the plots of "Scaled residuals" against "Predicted % inhibition" (for all datasets and sexes combined for each chemical) that were distributed at the meeting. These plots generally show much larger departures of model "predictions" from observations on the left (low % inhibition) than at the right (high % inhibition). The Panel agrees with the EPA comments made during the session that at the very least this observation indicates a need to rethink the weightings used for the data fitting.
A further step that some members of the Panel supported was that EPA explore, in at least preliminary testing, a restructuring of the decision tree for the introduction of "B" into the modelling. Instead of the current approach that includes B in the initial model used for the base analysis, and eliminates B only if the full available data set cannot be fit with the base model, a model could initially exclude B, and then reintroduce B only if the model without B proves statistically inadequate. The current model is primarily an empirical model, without detailed mechanistic justification. Some Panel members offered the view that alternative hypotheses about the mechanistic interpretation of B affect how the desired measure of anti-cholinesterase inhibiting potency at low doses should be defined in relation to the model fit results. Moreover different mechanistic hypotheses could affect the choice of whether to introduce a B term in response to a lack of fit at high doses, rather than successively eliminating high dose points. Three possible mechanisms that could produce the high-dose plateau of inhibition well short of 100% are as follows:
- Some fraction of the measured baseline esterase activity might be produced
by an esterase that is not inhibitable by the cholinesterase agent in question,
or may be the result of spontaneous hydrolysis. Some very limited experimental
observations of apparently decreased sensitivity of remaining enzyme following
inhibition provide some support for this hypothesis.
- At high doses, a substantial degree of inhibition may induce compensatory
synthesis or release of new cholinesterase molecules, thus limiting the degree
of inhibition that can be achieved at high doses. In one case, measurements showed
an increase in the total number of cholinesterase molecules in the brains of animals
treated chronically with chlorpyrifos (Chiappa, S., Padilla, S., Koenigsberger,
C., Moser, V., and Brimijoin, S., "Slow Accumulation of Acetylcholinesterase in
Rat Brain during Enzyme Inhibition by Repeated Dosing with Chlorpyrifos." Biochemical
Pharmacology 49(7): 955-963 (1995).), supporting this hypothesis.
- At high doses, metabolic activation processes needed to produce the most reactive forms of the organophosphate anticholinesterase agents (e.g., the conversion of parathion to paroxon) may approach saturation. To the Panel's knowledge, this possible mechanism has not been studied.
For the first case, the treatment of the fitted "m" as the measure of potency as done in the present analysis may be defensible. The true potency would be measured as a fractional inhibition of the authentic cholinesterases. The influence of any generic esterase activity would be removed.
For the second case, the effect on low dose potency estimation would depend on the form of the relationship between cholinesterase inhibition and the induced synthesis. In order to produce a true flattening of the dose response relationship beyond a specified external dose, the induction would have to be proportional to that extra dose (in order to exactly counteract the inhibitory effects of additional increments of the externally applied anti-cholinesterase agent). If this extra synthesis activity were triggered at a bright-line switch-point, the low dose potency would likely be appropriately modeled by simply subtracting the activity that is uninhibited, as before. However, other relationships between external dose, cholinesterase inhibition, and cholinesterase induction could lead to other results. Introduction of the B in this case may allow EPA to use more data and may achieve lower assessed statistical uncertainty at the cost of introducing phenomenological uncertainty in the determination of relative potency. Similarly, a finite estimate of B may be the result of saturation of metabolic activation. Thus the Panel recommends that EPA explore the implications for relative potencies of (1) dropping high dose data points in response to a lack of fit, rather than introducing B or (2) defining potency in terms of the reciprocal of the ED10. Information presented by EPA indicates a good correlation between potencies defined as slope factors vs the reciprocals of such ED10 "benchmark doses," and the benchmark dose is a more commonly recognized index with ample precedent in previous agency risk analyses.
Question 1.2 Please comment on the performance of the exponential model and the model fitting procedure. Are the strengths of the approach, as well as the defaults and assumptions for the parameters of the model, clearly articulated in the document?
The Panel concluded that the modeling approach, the estimation procedures, and the results were presented in an exemplary manner. Some formal analysis of the residuals as a function of dose is in order, however, as previously recommended.
The Panel expressed concern for the accuracy of the chi square approximation for the "goodness of fit" statistic as used in the EPA generalized least squares analysis. If the right tail of the true null distribution of goodness of fit were longer than a chi-square right tail, then some of the 16% of all fits that failed the chi-square test should have been accepted and the rest could be just Type 1 Errors.
The Panel suggests that the Agency reconsider the confidence interval calculations and perhaps try bootstrapping or some other more robust method; note that almost all the confidence levels for BMD in the simulation study (Appendix 1, p. 16, Fig.3) are much less than the nominal 95%. It might turn out that it suffices to use a nominal level of 99.0% or 99.9% in order to achieve an actual level that is closer to 95%.
The Panel recommends deleting the p and t values in Appendix 4, p. 2.
Finally, the Panel recommends exploration of bootstrapping as a means to generate confidence limits for some of the fitted parameters. Finally, the Panel emphasized that it is important for the final form of the model to be adapted so that there is no correlation between the scatter of the scaled residuals and predicted % chlolinesterase inhibition.
The Panel observes that the parameter estimates vary considerably depending on the weighting that is used, and this needs to be addressed.
The Panel strongly encourages the development of mechanism-based models for these types of analyses.
Finally the Panel recommends that the doses used for evaluation of potencies at various ages within specific data sets should be derived from the actual dietary intake rates observed in the study for those ages where the consumption data are available. Where the study-specific consumption data are not available for particular time points, there could be generic assumptions relating to relative consumption of food per body weight as a function of age.
Question 1.3 Is there a strong basis for considering an alternative model that would be more appropriate for the cholinesterase data on organophosphorus pesticides and could plausibly yield substantially different relative potency values compared to the exponential model recommended by the September 2000 SAP?
No alternative model would be more appropriate at this time. The analyses utilizing higher-order terms to model a nonlinearity at relatively low dose levels presented during the public comment session did not convince the Panel that these nonlinear terms are needed to fit the data because no statistical analyses were offered indicating that the higher order terms were significant statistically. The existing model stays reasonably close to the data. Because of this, and because a large number of the data sets appear to have BMD10's bracketed by exposure doses, other re-analyses that also stay fairly close to the data are likely to yield similar results in terms of relative potencies.
Some public commenters objected to the linear form of the current model at low doses, and this concern is shared by some members of the Panel. Their argument is that the true relationship should be S-shaped, reflecting either a true threshold (an interval of increasing dose at which there is absolutely no inhibition until a defined point is reached where increasing inhibition begins). Alternatively, there could at least be a shallower slope of incremental inhibition with increasing dose than would be observed at higher doses following appreciable saturation of some process.
Some Panel members suggested that the expectation for a true threshold in the relationship between cholinesterase inhibition and external dose (that is, absolutely zero inhibition at a dose range from zero up to a defined point) is without merit. No "first pass" metabolism process can be infinitely rapid (going to absolute completion in the finite time that the blood spends in the liver). Therefore some finite fraction of the absorbed molecules must escape "first pass" metabolism and be available to inhibit cholinesterase enzyme molecules in the blood and the downstream systemic compartments. The Panel was in general agreement that, because of absence of saturation phenomena, at the limit of low dosage in both transport and reaction phenomena some low dose linear slope must exist between dose and cholinesterase inhibition, even though the resulting inhibition may fall below levels that can be directly measured with confidence or where the inhibition is likely to be meaningful and associated with observable biological responses.
Several Panel members noted that some nonlinearity at intermediate doses can be expected, to the extent that "first pass" metabolism or systemic metabolic processes act by a saturable mechanism (e.g., via Michaelis-Menten kinetics). Whether this is indicated by the empirical data in general or for specific pesticides can be evaluated by examining the model residuals in the low region of dosage or by tests using other statistical techniques. This should be done formally in subsequent analyses.
The Panel reiterates its recommendation that the agency pursue the development of physiologically-based pharmacokinetic models utilizing the available data. Such models have been published for some anticholinesterase agents (Gearhart, J.M., Jepson, G.W., Clewell, H.J., III, Andersen, M.E., and Conolly, R.B. 1994. "Physiologically based pharmacokinetic model for the Inhibition of Acetylcholinesterase by Organophosphate Esters". Environ. Health Perspect. 102: 51-60, Supplement 11.) including, most recently, chlorpyrifos. Such models could incorporate any available data on the recovery rates of specific cholinesterases via either regeneration of initially inhibited enzyme molecules or resynthesis. In the near term, extension of PBPK modeling to the whole group of organophosphates may require the use of basic chemical property information (e.g., octanol/water partition coefficients, etc) to estimate partition coefficients and other use of supplementary information. Benefits of such models are that they include parameters that can be measured in subsequent studies, and they are capable of making predictions for a whole range of non-steady-state exposure times.
It would be useful if the guidance document considered how this kind of assessment might be done were it not limited by the availability of data, if, for example, PBPK models were available for the individual compounds and interaction terms could be obtained that would allow the individual models to be combined. This would allow robust descriptions of pharmacokinetic mechanisms to be used in the prediction of cumulative risk of AchE inhibition. This kind of approach is taken in the draft guidelines for carcinogenic risk assessment, which state a preference for use of biologically based models while acknowledging that in most cases the available data do not support the mechanism-based approach. The value of revising the guidance document in this manner would be to provide a clear context and justification for the emphasis on statistical modeling of the inhibition data as opposed to mechanism-based modeling. This revision of the document would also serve to encourage testing and experimental work that would support mechanism-based risk assessments for anticholinesterase agents.
Issue 2. Combining Cholinesterase Data from Multiple Datasets to Derive Relative Potency and Benchmark Dose Estimates
The current dose-response approach used a hierarchical statistical model to combine estimates of potency for the oral cholinesterase data (average absolute potency values) for each organophosphorus pesticide. This approach also was used to combine estimates of benchmark dose for the oral, dermal, and inhalation studies on cholinesterase (i.e., average BMD10s for the index chemical). There are several advantages of combining estimates from multiple datasets compared to using estimates derived from single datasets/points. Combining data increases the precision of estimates, incorporates the variability among data sets into the overall estimate of uncertainty (standard errors or confidence limits), and maximizes the use of the available information.
Question 2.1 Please comment on the present method for combining data to derive estimates of absolute potency for each OP and to derive benchmark dose estimates for the index chemical. Does the document sufficiently describe the methodology used in combining estimates?
In general, the description of methods for combining data to derive estimates of absolute potency for each OP and to derive benchmark dose estimates for the index chemical were reasonable. It is a strength of the proposed approach that all datasets were evaluated, not just a selected subset of data. The Panel had no general concerns about the procedures for combining data.
Issue 3. Approach for Selection of the Index Chemical
The September 2000 SAP report concerning this subject recommended the careful selection of an index chemical with "the best and most complete data for the common endpoint(s)" to minimize the uncertainties in the estimation of cumulative risk. Methamidophos was selected as the index chemical following a comprehensive and systematic review of data availability and data quality on all 25 chemicals.
Question 3.1 Does the document adequately describe the rationale for selection of the index chemical? Should any additional factors be included in the rationale for the selection of methamidophos as the index chemical?
The selection of an index chemical is important for classifying relative potency of a series of toxicants. EPA considered the entire data set of all 25 OP pesticides in evaluating the choice of an index chemical. The document reasonably and clearly describes the rationale for selection of methamidophos as that chemical.
Methamidophos has several characteristics, as outlined in the document, that suggest it may be the best chemical for this purpose. A number of studies with methamidophos covered all three dose routes (oral, dermal, and inhalation), and measures of potency were relatively similar among compartments and across routes of exposure. These data suggested that one index chemical may be used across compartments and routes. Furthermore, as mentioned in the document, methamidophos has been reported to lack some additional actions (e.g., direct binding to cholinergic receptors, alteration of neurotransmitter release) that have been noted for some other OPs and act as a truly "pure" cholinesterase inhibitor. In the case of evaluating insecticides as a class that elicit toxicity through acetylcholinesterase inhibition, this may also be a desirable characteristic. Another factor identified in the document pertained to the relative lack of interactions between methamidophos and other esterases, e.g., A-esterases, carboxylesterases, that detoxify OPs. Thus, methamidophos would potentially interact more exclusively with the "common mechanism" target molecule (acetylcholinesterase) and less with other "non-target" binding sites.
One could argue that characteristics that may make methamidophos a good index chemical could make it a bad choice. For example, if most OPs in fact interact with other macromolecules within the body, it could be argued that a more "representative" chemical would be one that also exhibits these interactions (as do most of the OPs in question). This could be of particular importance when considering the effects of low dose exposures, where some additional binding sites (sites of loss) may be more critical in shaping the dose-response relationship. One would expect that an anticholinesterase that interacts more exclusively than others with acetylcholinesterase would exhibit a more linear dose response relationship at very low to intermediate dose exposures, due to the lesser influence of uncontrolled confounding factors (one of which would be biotransformation) that modify the amount of compound that reaches the target enzyme.
Other considerations include that methamidophos 1) is a small, water-soluble molecule that does not require metabolic activation (although relatively recent data suggest that methamidophos may be activated to an even more potent inhibitor in vivo), 2) does not interact with lipid domains of the ChE molecule, and 3) exhibits a relatively rapid rate of reactivation of the phosphorylated enzyme (as with many dimethyl- compared to larger substitution-analogs). Therefore, although methamidophos may be an appropriate index chemical for analysis under steady state conditions, if analyses change to non-steady state conditions, then criteria used for selecting the index chemical should be reconsidered.
It was generally agreed that while the selection of another index chemical may change the potency scale itself, it would have no influence on the relative rank order of potencies of the 25 OP toxicants using the methodology described. For the purpose of establishing an index for comparison purposes based on the endpoint of anticholinesterase potency under conditions approaching "steady-state" inhibition, considering both 1) the relative completeness of the data and 2) the relative toxicokinetic simplicity afforded by limited additional interactions, methamidophos is a reasonable overall choice for the index chemical.
Issue 4. Use of Steady State Cholinesterase Data for Estimating Both the Relative Potency and the Points of Departure
There are several key general principles for conducting a cumulative risk assessment (EPA 2000 Draft, "Proposed Guidance on Cumulative Risk Assessment of Pesticide Chemicals that Have a Common Mechanism of Toxicity"). One such principle concerns the time frame of both the exposure (e.g., What is the exposure duration?) and of the toxic effect (e.g., What are the time to peak effects and the time to recovery?). Both must be adequately defined prior to performing a cumulative risk assessment so that an individual's exposure is matched with relevant toxicological values in terms of duration. In the case of organophosphorus pesticides, potential human exposure may occur through food, drinking water, or from residential/non-occupational uses of these pesticides. Thus, there will be short-term exposures to organophosphorus pesticides observed as peak exposures via food, residential/non-occupational uses, and drinking water, as well as continuous exposure via food consumption. There are several important considerations with respect to the temporal characteristics of the cholinesterase inhibitory effects of organophosphorus pesticides in estimating their relative toxic potencies and in estimating the point of departures for the index compound.
Relative Toxic Potency: OPP has elected to use data reflecting steady state in the interest of producing relative potency factors (RPFs) that are reproducible and reflect less uncertainty due to rapidly changing, time-sensitive measures of cholinesterase. Also, when the compounds are at steady state, in principle, the differences in toxicokinetics among the individual OPs are less likely to impact the assessment. On average, the inhibitory effects of OPs on cholinesterase activity reaches steady state by approximately 30 days. Therefore, this analysis focused on studies of a duration of 21 days or greater in order to use cholinesterase data that attained steady state so that stable estimates of relative inhibitory capacity (i.e., relative potency) could be calculated among these compounds.
Point of Departure for Index Compound: Cholinesterase data that had attained steady state was also used to estimate the point of departure (i.e., the BMD10) for the index chemical (methamidophos). Thus, OPP has also elected to use steady state cholinesterase data to estimate the cumulative risk associated with different time periods ranging from a single day up to several months. Although toxicology studies of 21 days in duration do not directly reflect the time frame of interest (approximately one day exposures that might occur particularly with residential or drinking water exposures), there are reasons why steady state data are preferred over acute toxicity data for estimating the point of departures for the cumulative risk assessment.
First, recovery from the effects of exposure to organophosphorus pesticides generally ranges from several days to several weeks depending on the dose. Thus, although the exposure may be acute, the cholinesterase-inhibiting effects may persist. Secondly, the exposure scenarios are very complex in the OP cumulative exposure assessment, and there may not be a true acute exposure to these chemicals. Although there may be peaks of exposure via food, drinking water and residential/non-occupational, there is some evidence that a relatively constant baseline of exposure to organophosphorus pesticides may occur. For example, the effects of one day exposures via food to one or more OPs may be extended by several days or weeks due to a home use or an episodic exposure in drinking water of other OPs. Although steady state data may not precisely reflect the one day exposure situations in the cumulative risk assessment of OPs, the use of steady state information to estimate the point of departure(s) for extrapolating human risk is viewed as a reasonable approach to approximate the complex multi-chemical exposure situations that involve the overlapping of different pathways (food, drinking water, residential/non-occupational) having different exposure durations.
Question 4.1 Please comment on the use of steady state cholinesterase data to determine the relative potency of the organophosphorus pesticides.
The approach that the Agency has taken to using steady state cholinesterase data to determine the relative potency of the organophosphorus pesticides is endorsed when the potency estimates are applied to exposure scenarios in which steady-state assumptions are appropriate. The Agency is commended for its effort in compiling and analyzing a large number of data sets. This provides a good foundation for the proposed approach to cumulative risk assessment.
The draft document reviewed by the Panel, however, is not particularly clear on the limitations that will be imposed on the exposure scenarios. During the meeting, the Agency staff clearly indicated that the relative potency parameters derived in the report will not be applied to acute (one day to several days) exposure. The Agency may wish to consider an exploratory analysis of acute relative potencies to support risk assessments for short term exposures in which the assumption of steady state would not be appropriate. This analysis could support the application of the relative potency approach to shorter term exposure scenarios. In any event, the document should clearly state that the relative potency estimates based on steady state data will not be applied to short-term exposures in which the assumption of steady-state is not likely to be valid.
An important factor to consider is the recovery of acetylcholinesterase activity following OP exposure. The relative differences in enzyme recovery could be very important when risk for relatively short-term exposures is estimated using estimates based on long-term dosing studies ("steady-state" conditions). It is well known that inhibition by some OPs is shorter-lived than inhibition with others, e.g., inhibition following exposure to dimethyl- analogs is typically of shorter duration than that noted following exposure to diethyl- analogs. Using steady-state determinations, relative overestimation of the acute potency of an OP that causes longer-term inhibition would be expected under those conditions.
The Agency is strongly encouraged, as it begins the early stages of consideration of other groups of pesticides that share a common mechanism, to explore and articulate the data needs for adopting a biologically based approach to the assessment of cumulative risk. Specifically, the Panel asserts that, with adequate lead time, PBPK/PD models could be developed for the chemicals of concern to EPA. The development and linking of such models would provide the Agency with a powerful tool for assessing cumulative risk for both steady state and dynamic/acute exposures. The Panel recognizes that this would be a labor intensive task and would require a substantial period of time. Thus, the recommendation to explore and pursue the development of PBPK/PD models should not be interpreted in any manner that would divert or delay the basic approach to cumulative risk for OP pesticides that was reviewed and endorsed at this meeting. The Agency should consider the possibility that data in support of the biologically based approach might be submitted as part of, or simultaneously with, the basic data package that is submitted as part of the pesticide registration process. Again, recommendations for the submission of such data should be as clear and explicit as possible.
There was considerable discussion by the Panel regarding the advisability
of basing the relative potency calculations on the average slope estimate (m)
versus the average BMD10 estimate. The debate centered around the basic
question of whether parallelism is being assumed on the scale of the slopes or
on the scale of the BMD
In the absence of such an analysis on the equality of the A's and B's, estimates of relative potency should be based on the ratios of the BMD10's and should not be based on the ratios of the slopes. Moreover, BMD10 being generally recognized as a valid point of departure, it has the advantage in being consistent with the aggregate risk approach, i.e., additive percentage of RfD or the inverse of MOE.
Limiting the analysis to relative potency estimates based on RBC AChE inhibition may be overly and unnecessarily restrictive. To evaluate "Human cumulative risk," the Agency must eventually test and validate the utility of blood cell AChE as a biomarker for human health. Studies are needed on the sensitivity, specificity, and predictive validity of blood cell AChE in relation to more central health-associated measures, such as brain AChE levels and eventually neurobehavioral competence.
In extrapolating potencies and combining doses across chemicals, it is important to take into account all the chemical-specific pharmacokinetic differences when reliable data are available, e.g., absorption rate (or factor), exposure duration, and weekly frequency. Annotation can be made for those scenarios in which default assumptions are used when data are lacking. It is also noted that, in the July 31, 2001, version of the Preliminary Cumulative Hazard and Dose-Response Assessment for Organophosphorus Pesticides: Determination of Relative Potency and Points of Departure for Cholinesterase Inhibition, the inhalation benchmark dose and NOAELs are expressed in the unit of air concentration. Since inhalation dose is a function of both the air concentration and the duration of exposure and activity level, the exposure duration should accompany the inhalation benchmark dose and NOAEL and should be taken into account in the subsequent relative potency calculation. The same information should be considered in estimating the human exposures.
Question 4.2 Please comment on the use of steady-state cholinesterase data to estimate the point of departures for extrapolating human cumulative risk.
In answering Question 4.2, the Panel felt that more information was needed about human exposure patterns. For example, will the Agency exposure assessment mainly be concerned with exposure spikes that are well above typical exposure levels with continuous exposures that do not vary greatly on a day-to-day basis? Little or no information is provided in the document on patterns of human exposure that might provide guidance on how to answer this question. To the extent that it is possible, the Agency should use actual human exposure data to justify the approach used for modeling the laboratory animal data. The comfort level with the present analysis based on steady-state would be greater if the main human concern was with continuous exposures and steady state or near-steady-state AChE levels. The comfort level would be less if the human exposure more likely to be associated with adverse health effects is a spike exposure or some exposure pattern that would not be expected to reach steady-state. If this latter case is in fact the greater concern, then the document should consider as clearly as possible how the steady-state analysis of the laboratory animal data relates to the case of spike exposures or non-steady state exposures of humans. The Agency is encouraged to study the short-term database for this purpose.
Several members of the Panel objected to the use of NOAEL's and LOAEL's for AChE inhibition data by the inhalation and dermal routes of exposure. These Panel members felt that, at the least, the exponential model developed for the oral exposure data should be applied to the inhalation and dermal data to determine what could be done with BMD's.
The Panel also noted that studies of the sensitivity, specificity, and predictive validity of RBC AChE activity as a biomarker for human health are required. Studies of the relationship of RBC AChE activity and central health-associated measures such as brain AchE activity and, eventually, neurobehavioral competence are needed.
One Panel member noted that the BMD and potency estimates described in the document are based upon differences between exposed animals, and AChE levels in reference animals considered to be unexposed. Because of the demonstrated ubiquity of OP's, this assumption needs to be tested and validated. Determining the true reference AChE level will require raising control animals including the dams of the test animals during gestation in sanctuary circumstances and on synthetic pesticide-free chow, filtered air, and distilled water. The AChE levels in the test animals would then be compared to historical and pair raised controls raised under usual conditions. This is the only way to distinguish between "typical" AChE levels, and "natural" AChE levels.
Issue 5. Is the Agency Ready to Address a Draft Dose Response Model for the Cumulative Risk Assessment of OP's by December 1, 2001?
At the end of the session, the Panel was asked to address the time line developed by the Agency for completing the cumulative risk assessment for the organophosphates in light of its comments on the proposed dose-response assessment.
The Panel concluded that it was possible that a draft risk assessment using this hazard and dose-response assessment could be completed by December, 2001, and strongly encouraged pursuit of this goal. However, substantial additional analysis will be required to support the current empirical approach to dose-response modeling, necessitating a significant commitment of resources if this deadline is to be met. The nature of the additional analyses required is discussed in the responses to the foregoing questions.
The empirical approach to assessing dose-response assessments may serve the immediate needs of the Agency, but there is concern that its ability to describe dose-response behavior in the low-dose region and under non-steady-state conditions is limited, particularly given the data that are currently available. Consequently, the use of this approach in the near term should be accompanied by a clear description of its underlying assumptions and caveats. The Panel thought that the interests of reducing uncertainty in cumulative risk assessment of OPs would be served by the incorporation of biologically-based dose-response relationships (e.g., models that incorporate physiologically-based pharmacokinetics). The Panel considered it unlikely that this could be accomplished by December 1, but recommended that the Agency aggressively pursue the creation of these models.
The Panel was very concerned that the dose-response assessment, as currently constructed, might become a prototype for cumulative risk assessments for other chemicals (including non-pesticide chemicals), particularly those for which data sets for analysis are less robust and complete than for the OPs. The Panel considers cumulative dose-response assessments to be nascent methodology that may not yet be sufficiently robust for broad application.