Arsenic in Rice: Part 8 - How Low Can You Go?

Consumer Reports implies that rice above a 5 ppb standard poses a "troubling," "worrisome," "cause for concern," or "potentially harmful"situation for those who eat it.

Using the 5-ppb standard in our study, we found that a single serving of some rices could give an average adult almost one and a half times the inorganic arsenic he or she would get from a whole day’s consumption of water, about 1 liter.

That threshold of "5-ppb" came from New Jersey:

[t]he 5 μg/l arsenic MCL [was selected] after consideration of the findings and recommendation of the New Jersey Drinking Water Quality Institute (Institute). The Institute reviewed the reports issued by the National Academy of Sciences (NAS) on the health effects of arsenic in drinking water in 1999, as well as an update of this report released in 2001 (Arsenic in Drinking Water: 2001 Update, NAS Press, 2001).  Based on the current NAS analysis, the Institute determined that the drinking water concentration that results in a one-in-one-million excess lifetime risk of lung and bladder cancer for United States populations was an estimated 0.003 μg/l (or three nanograms per liter or three parts per trillion).

Therefore, in view of its concerns regarding reliable removal technology, the [New Jersey Department of Environmental Protection] determined to promulgate the arsenic MCL at 5 μg/l. This determination comports with the NJSDWA mandate to establish the MCL at the most protective level within the constraints of medical, scientific and technological feasibility.

Even at that, some folks, like Jeffrey H. Tittle with the Sierra Club were not happy:

Source

How New Jersey got to "5-ppb" is important because it is what Consumer Reports uses to inform their readers that rice samples found with inorganic arsenic over that threshold of "5-ppb" is "troubling," "worrisome," "cause for concern," or "potentially harmful."  If 5-ppb is okay for Consumer Reports and is implied to be "safe," they are telling the Sierra Club that 3-ppb is too low.  But I digress....

5-ppb, which is 5 μg/l, was the amount settled on because a water system cannot effectively treat water to get the arsenic down below 5 ppb.  If you cannot treat it to less than 5 ppb, then all water systems with arsenic above 3 ppb would be in violation.  Even though the Sierra Club and the NJ Institute believe 3 ppb should be the standard, that unfortunate little devil called "reality" steps in and tells them that you can't always get what you want.

The point here is this:  5 ppb or 3 ppb is not the number that will get you to what New Jersey's law wants - which is an excess cancer risk of one in one million.  To get to that risk level, 1 in 1,000,000 the water must have an inorganic arsenic concentration of  "0.003 μg/l (or three nanograms per liter or three parts per trillion)."

My point is that none of these numbers, 5 μg/l, 3 μg/l, or 0.003 μg/l, are thresholds to which exceeding them by up to 5 ppb would be "troubling," "worrisome," "cause for concern," or "potentially harmful," as Consumer Reports tells its readers.

So lets look at just how conclusive that number of 0.003 μg/l (or three nanograms per liter or three parts per trillion) is.  That number presents an acceptable risk on one excess bladder/lung cancer per 1,000,000.  Basically anything above 0.003 μg/l increases the risk.  When, I will ask, does that increase start to meet the Consumer Reports designation of "troubling," "worrisome," "cause for concern," or "potentially harmful?"

New Jersey came up with the Slope Factor that produced 0.003 μg/l,using the research performed by the National Academy of Sciences (NAS) who put forth the NRC report,  In the NRC report we learn:

Because the EPA did not present theoretical lifetime excess bladder and lung cancer risk estimates [in 2000], the [NAS] used linear extrapolation from the EDs [1% Effective Dose's] presented in Morales et al. (2000) to estimate these risks at 3, 5, 10. and 20 μg/l .

So here we are in 2012.  New Jersey settled on 5 μg/l, they proposed 3 μg/l which would have made the Sierra Club happy, but REALLY wanted 0.003 μg/l.  The EPA, on the other hand, came up with an MCL of 10 μg/l and an MCLG of zero.

So many numbers, so many choices of "safe."

And when you look at how those numbers were derived, one thing starts to pop out at you (well at me anyway).  That is, there is a connection that binds them all:

After the publication of the proposed rule, Morales et al. (2000) published a study in which a risk assessment for mortality from several internal cancers was presented.  The risk assessment was based on reanalyses of the data from southwestern Taiwan.  Risk estimates were calculated for mortality from lung, bladder, and liver cancers, as well as combined cancer deaths, using 10 different statistical models, calculated with and without a Taiwanese comparison population.  [NRC]

Which then tells us this:

EPA was considering those analysis in the final rule making [that established the current 10] and, therefore, published a Notice of Data Availability in the Federal Register summarizing and further analyzing the information from Morales et al. [NRC]

Which leads to the EPA producing a draft document in 2010 called the "IRIS Toxicological Review of Inorganic Arsenic (Cancer)" where we learn this:

In the course of this analysis, EPA has investigated the impact of alternative model forms on the cancer risks estimated for the Taiwanese and U.S. populations for individual endpoints (lung and bladder cancer). Based on the past experience of Morales et al. (2000) and modeling results presented by NRC (2001), this effort was limited to exploring alternative forms for the dose dependence of risks.

These MCLs, these thresholds for what is "safe," are the result of a Slope Factor that is used to estimate excess cancer risk from inorganic arsenic that was derived this way:

EPA investigated a range of model forms for use in the risk assessment, building on previous efforts, including U.S. EPA (2001) and Morales et al. (2000).  The model used in the derivation of the preferred risk assessments employs:

• Poisson regression (of cancer mortality against age and dose) fit by maximum likelihood estimation (MLE).
• A quadratic age model.
• A linear multiplicative dose term.
• Confidence limits on the dose term estimated by profile likelihood.
• Estimates derived for the data set that includes the southwest Taiwan reference population.

That's complicated toxicologist talk.  What I want to focus on is that all of these numbers are based on a Slope Factor that comes from data provided by some guy named "Morales" and his friends "et al."

Morales...Morales...Morales...

Next Post: Arsenic in Rice:  Part 9 - The Broad Shoulders of Morales et al.

.