Banning or Applying ALARA to “Evil” Chemicals

According to the National Research Council’s Glossary: ALARA is an acronym for controlling exposure to a chemical "as low as (is) reasonably achievable,”.  It is just slightly higher on the scale of ultimate controls that culminates in outright banning.   The word “reasonably” as used here is clearly open to interpretation. 

I recently wrote a blog about using dose-response modeling to assess the risk of contact allergy.  This particular blog resulted in quite a bit of spirited discussion/debate over whether the technique had any value at all.    One responder suggested that exposure to any and all allergens should be controlled using ALARA especially for highly induced or sensitized individuals. 

Indeed, on the face of it, it is hard to argue that the exposure to the specific allergens for these folks should be keep as low as “reasonably” achievable.   Indeed, there’s that  r-word again! 

Let us take the case of Type IV contact allergens (of which poison ivy is a well-known example of a “natural” allergen of this type).  I have worked in this area for many years and my understanding is that many of the preservative employed to keep our water-based products fresh (including cosmetics) is a Type IV contact allergen.   Is the amount of these preservatives in products as low as reasonably achievable?   Some would argue that it is not.   These products definitely deliver a dose of allergen every day to any person with dermal exposure to them.   It also gets into the air from these products and other treated products (like treated wood or treated water) which also exposes people.

We could BAN the use of allergens in these products so that the amount of these clearly identified allergens could be ZERO and the risk of Type IV contact allergy from their exposure would be ZERO as well.  But is that truly reasonable?   What are the economic implications of having all of these products with dramatically and perhaps unreasonably reduced or essentially eliminated useful shelf-life?

My sense is that this is exactly why we do human health risk assessment and why we use dose-response modeling to predict the risk of common environmental exposure levels.   If we are smart, we do not rely on modeling alone but also on historical, clinical and epidemiological data to tell us if these models are an appropriate portrayal of reality.   In the case of the model I discussed in the blog, it does not have or assume a threshold of response or risk.   There is always some finite risk predicted at any exposure above ZERO.   Some number of folks on the hypersensitive end of the population curve would be predicted to respond.

How large this subpopulation of hyper-susceptible persons might be is clearly a matter for debate and deliberation by the stakeholders (i.e., a proper democratic/political process).   What I believe is not open to debate is that it should not be ZERO; that is, the product should not be banned.

In my opinion banning any product is equivalent to declaring it to be evil.   That is, attributing a malevolent characteristic to an inanimate object.    This is a form of animism or a belief that inanimate objects have spirits which has generally been rejected by modern Western Society.

 

My sense is that we should be using the rational assessment of risk by applying the best tools available to us while considering the uncertainty.  If the risk is considered too high politically at any particular concentration then it should be used at a lower concentration determined to present a risk which is not unacceptable.  In the example we have been discussing, if that concentration is not effective as a preservative then the material would not be used in that application.  This is rational, analysis-based restriction, not banning.

Clearly, folks on the hypersensitive end of the spectrum – those who might be expected to react at concentrations allowable for the general population – will want to take steps to lower their exposure even more.   I would certainly support product labeling requirements to facilitate this process.   

I believe that it all comes down to doing the best for the most and drawing the lines (as risk managers) or informing the position of the lines as risk assessors. 

Question for Group Discussion:  What is your “gut check” for an exposure/ risk that is not unacceptable?   My answer:   Any exposure that I would allow my children to experience.

ConsExpo: A Valuable Modeling Tool

Those of you who read this blog regularly know that I am pretty high on IH MOD as a first-rate modeling tool for the estimation of inhalation exposure potential.   This week I want to talk about another important modeling tool:  ConsExpo which is short for Consumer Exposure modeling tool.   It has been developed for a number of years by the good folks at the Netherlands National Institute for Public Health and the Environment (its the initials from the Dutch name are RIVM).   According to the RIVM web site (http://www.rivm.nl/en/Topics/C/ConsExpo):

The software model ConsExpo is a set of coherent, general models that enables the estimation and assessment of exposure to substances from consumer products that are used indoor and their uptake by humans.

ConsExpo is a relatively “high level” model especially compared to IH MOD in that there is a considerable amount of information and structure related to consumer exposure is already built into it.  The advantage is that is that it is relatively easy to use and is specific to exposure from consumer products.   The disadvantage is that it may not be as flexible as IH MOD in addressing a wide variety of human exposure scenarios including those often presented in the work place.  

ConsExpo provides modeling estimates for various exposure routes (inhalation, dermal or oral route).   To an extent the software guides one to the most appropriate exposure scenario and uptake model is chosen for each route.   It differentiates between applied exposure or the amount available for update and the actual uptake of the chemical of interest.    You as the modeler are required to fill in the model input parameters.    Another excerpt from the RIVM web site is below:

To enhance transparency and standardization, for a number of product categories, default parameter values have been compiled in so-called fact sheets. The default-models and default-parameter values are available in the form of a database with the ConsExpo software. Currently, the most recent version is ConsExpo 4.1. A beta-version of ConsExpo 5.0 is available, in which the methods for probabilistic exposure evaluation have been improved.

The last sentence above highlights a feature of Consexpo that IH MOD does not have at present; namely, built-in probabilistic exposure evaluation in which you provide the input parameters as probability distribution functions (PDFs) rather than single deterministic values.   My friend and colleague, Tom Armstrong has studied and worked on the integration of IH MOD with various probabilistic tools for some years but to date the the capability of including PDFs have not be directly integrated into the tool.   Perhaps I (or you) can one day persuade Tom to write a guess blog here that would help us with this subject.

The fact is that ConsExpo is a valuable resource when estimating exposures to chemicals from the use of consumer products.   You can get the program at:

http://www.rivm.nl/en/Documents_and_publications/Scientific/Models/Download_page_for_ConsExpo_software

The above mentioned “fact sheets” from RIVM provide the modeler with general background information.   Indeed, they provide specific exposure parameters within the context of an exposure scenario.

The following fact sheets are available for download at

http://www.rivm.nl/en/Topics/C/ConsExpo/Fact_sheets

·         General fact sheet

·         Children's toys fact sheet

·         Cleaning products fact sheet

·         Cosmetics fact sheet

·         Desinfectant products fact sheet

·         Do-It-Yourself products fact sheet

·         Paint products fact sheet

·         Pest control products fact sheet 

Here is our LinkedIn Group discussion question for the week:

How might you see ConsExpo fitting in with your world?

Also, I would like to express my thanks to my friend and colleague Bert Hakkinen for suggesting ConsExpo as a blog topic.  If you have any topics you would like to see covered in line with this blog please drop me a note at mjayjock@gmail.com.   Some of the best blogs have come as a result of your suggestions.

Low Dose Response Modeling of Contact Allergy

Some folks think that, once you are sensitized, that there is really no level of exposure to an allergen that is safe.  You must simply eliminate your exposure to it.   For quite a few years, the company I worked for sold a Type IV contact allergen that was used in both industrial (e.g., Metal working fluids) and consumer products (e.g., cosmetics and house paint).   Let’s call it product A.   I was asked to do a risk assessment in which it was determined that the typical dose encountered during Product A's use when compared to the anticipated dose-response would result in a risk that was not unacceptable to the dermatological community. 

It is a fine but important distinction that “not unacceptable” is not the same as acceptable.  It is like the statistical test for differences.  We do not accept the “null hypothesis” we simply fail to reject it. 

For those of you who are unfamiliar with allergic response, it starts with a period of induction in which the body learns to “recognize” the allergen and then marshals its immunological agents to respond to it on subsequent exposures or challenges.   Thus, the first time you contacted “poison ivy” (another Type IV contact allergen) you probably did not get the characteristic rash; however, if you became induced, as about 80% of us in North America are, then you responded with a rash on subsequent exposures to this shiny 3-leaf weed. 

I started this analysis with the following assumptions:        

•  Everyone who comes in repeated contact with Product A in normal use would become induced; that is, they would be sensitized.    This may not be true but it would be worst case.
•  That there was no threshold of allergic response to sensitized individuals.  Again, probably not true but worst case. 
•  That the dose-response of sensitized individuals was described by the standard log-probit dose-response curve.
•  Guinea Pigs are a good model for human response.   That is, we can use the dose-response from these animals as a reasonable surrogate for human response.
• That a risk of getting a skin rash from exposure to Product A of 1 in 1,000,000 would be considered de minimus or at least not unacceptable.

A final assumption is that the appropriate dose metric for contact allergy is the maximum amount of allergen applied to any particular square centimeter of skin.   Indeed, dermatologists have shown us for many years that they can both induce dermal sensitization and elicit an allergic skin rash through relatively small patches on the skin of about a square cm or so of area.   Exposing the skin over larger areas for the most part did not result in significantly different responses.

Well, it turns out that the model did a pretty good job of predicting human response for both induction and elicitation of an active skin rash.   The work was presented to the dermatolgical community and published in a peer-reviewed journal of this group.   Anyone wishing a copy of the original paper should ask me at mjayjock@gmail.com and I will send it to them.

This blog gets posted mostly to LinkedIn Discussion Groups.   As their name implies, these groups are set up to foster discussion.  I have promised to try and make this blog more interactive.  Thus, I will be asking some questions at the end of every blog from now on to encourage some discourse within the various groups or directly to the blog as you wish.

This question this week:  In general, how do you view low-dose extrapolation of dose-response models for the purpose of risk assessment?

Evaluation of IHSkinPerm with In vitro Data

It is always exciting to meet a new (to me) colleague and come across a piece of work that materially adds to our understanding of dermal exposure assessment.  This happened for me at the recent AIHce in San Antonio.   I had the good fortune of hearing a presentation by Dr. Deborah Lander entitled:  Are Dermal Absorption Models Effective in Refining Dermal Exposure Assessment for Product Stewardship and Regulatory Risk Assessments?   

Debbie and her colleague William Fasano from DuPont’s Haskell Labs posed this question and set about to answer it.    Some of the salient points listed in Debbie slides include:

•       Regulatory requirements increasingly require assessment of     dermal exposure

•       There are ways to estimate the loading to the external skin      surface (measurement, and increasingly models)

•       Old conservative rule is use 100% dermal absorption unless      MW greater than 500 and Log Kow is greater than -1 and less than 4, then 10% is conservative (De Heer 1999).

She then asked “What is Plan B”?   Which is listed below:

Assuming there is no animal dermal toxicity testing for the substance of interest or similar analog we:

•       Could use steady-state permeability (Kp) from an aqueous        solution of infinite volume if applicable

•       Could perform in vitro testing on human skin

•       Could perform bioavailability studies such as sweat                   extraction (reasonable if water soluble)

•       Could use a model to predict dermal absorption

Ah ha!  They looked at the last bullet and considered AIHA IH SkinPerm with the question:  How do the predictions from this model differ from the results of well conducted in vitro studies of 15 neat organic chemicals for 60 minute flux time?   She reports a total range for the ratio of predicted to measured  values of 0.03 to 8.95.    In Dr. ten Berge's work he reports a ratio of 0.1 to 30 for 26 chemicals.   This data set shows that 27% of the model predictions were within a factor of 2,  73% where within a factor of 10 and all of them were within a factor of 30 fold.   Dr. Lander calls these MUs or model uncertainty factors.

Dr. Lander’s results were for VOCs with MW well below 500 and LogKow less than 5.5.   She then examines the situation for neat lipophilic compounds with LogKow greater than 5.5.    Her data indicated that IH SkinPerm is conservative for these compounds and one could probably exclude the application of any model MU.

Clearly an MU of 30 is a relatively large factor but my sense is that it can be brought down by future experimental data in which the consistency of the quality of the experimental data are carefully monitored and assured.   Another approach would be to assign MUs based on the grouping of compound families related to empirical factors such structure.  However, all of this requires more data and good data which is something for the future and because it will need funding it could be the distant future.   What about today?

One might think that the application of an MU of 10 for 73% inclusion or 30 for 100% inclusion would not be very useful but the reality is that in many situations multiplying the model predicted value by 30 will still provide a significantly lower estimate of dermal absorption that the assumption that all of the material contacting the surface of the skin penetrates to the circulatory system of the body.    Also we now have some analysis that indicates IH SkinPerm is conservative for lipophilic (LogKow greater than 5.5) compounds.

The above is what I found most interesting and informative about Debbie’s presentation; however, she has given me permission to send the slides to whomever asks me for them at mjayock@gmail.com.   They cover evaluation of the NIOSH dermal model and some other issues and they have all of the references.   She is also planning on submitting this work for publication.