Toxic Curve Ball: Why Outdated Assumptions to Determine “Safe Levels” of Toxicants Forfeit the Game
By now, a large number of consumers are aware of the hazards
of the synthetic compound bisphenol-A (BPA). Effective May 11, 2016, under California state law
Proposition 65, products containing BPA must
possess a warning label indicating that exposure could result in female
reproductive impairment. Independent research on the endocrine disrupting
effects of the chemical, commonly used in plastic bottles, the lining of metal
cans, and customer receipts, among other applications, has consistently
demonstrated toxic effects at low dose exposures. Two recent robust studies
from Denmark
concur, finding
deleterious effects in rats exposed to BPA at doses lower than those considered
safe for human ingestion, yet not
at several higher doses. Nevertheless, regulatory agencies such as the U.S.
Food and Drug Administration (FDA) and the European Food Safety Authority
(EFSA) conclude that BPA is safe at the levels at which it is currently in use.
Clearly, disagreement exists among academic researchers and
regulators about safe levels of BPA, as well as of innumerable other chemicals.
The discrepancy stems both from how data are derived to determine safe levels of
exposure to known toxicants, and from whether safe levels are even derivable
under traditional standards of appraisal.
When citizens inquire about the toxicity of their products,
they are usually met with guarantees that hazardous substances within these
items exist at levels that are too low to produce harm from routine exposures. Likewise,
after accidental releases of hazardous substances into our environment there is
a time, either at the outset or when concentrations of the pollutant subside,
that levels of the contaminant are deemed low enough to be safe for human
exposure. When evidence of contamination in the municipal water system in Flint, Michigan
first emerged, officials
initially declared the water safe to drink. When radiation from the Fukushima nuclear disaster crossed the Pacific and reached
the West Coast of the United
States, the
public was met with assurances that the level of radiation was too low to do
harm. “Safe levels” is a common refrain to assuage fears of chemical
toxicants. Yet accumulating research, like that on endocrine disrupting
chemicals (EDCs) such as BPA, reveals that the foundational principle of safe
levels of chemicals at low-enough concentrations is a flimsily constructed one.
Current
chemical risk assessment operates under the assumption that we can determine a
lowest dose at which a compound produces negligible or no harm to human health
– the Lowest Observable Adverse Effect Level (LOAEL). The presumption is that
at increasingly higher doses, the substance will be increasingly harmful; at
lower doses, harm will be insignificant or nonexistent. (Only a select few
substances are regarded as harmful at any dose.)
Our regulatory toxicological tests are based on this
supposition of positive monotonic dose-response. Monotonic refers to the slope
of the dose-response curve consistently progressing in one direction and never
changing sign along the way. These positive monotonic dose-response curves are
commonly linear, exponential, or sigmoid (Fig. 1). But, this expectation of
monotonicity upon which we base regulation has been strongly challenged not
just by the newest papers on BPA, but by an
accumulating consensus. Indeed, the dose makes the poison, but in unanticipated
and unpredictable ways.
(Figure 1. Some examples of monotonic, positive dose-response curves:
linear, exponential, sigmoid)
Numerous
substances act in non-monotonic dose-response (NMDR) manners (Fig, 2), in
which the sign (positive or negative) of the response can change throughout the
measurement of the dosages. Many essential vitamins and minerals serve as
examples. At too-low doses they are insufficient at providing the necessary
nutritional molecules needed for functioning. At too-high doses, many can be
poisonous. The desirable level of vitamin intake falls at a crucial range in
the middle. Their dose-response curves, in which the response examined is
nutritional benefit, would resemble an inverted U-shape. Indeed, countless nutrients
possess U- or inverted U- shaped NMDR curves.
(Figure 2. Some examples of NMDR curves. Source:
epa.gov)
According to Dr. Pete Myers, founder and chief scientist at
Environmental Health Sciences, “NMDR curves are the default expectation for
endocrine disrupting compounds (EDCs).” As co-author, with the late Dr. Theo
Colborn and Dianne Dumanoski, of Our
Stolen Future, Dr. Myers wrote the seminal book on EDCs. EDCs are hormone
mimics and as such, operate in several complex ways to trigger or suppress
normal hormonal regulatory mechanisms. Consequently, they can produce negative
effects at different doses, often at the very high and very low levels, rather
than in-between. They act in a mode that contradicts the assumptions of low
dose safety.
Dr. Myers estimates that at least 1000 EDCs are currently in
use commercially, but because most chemicals in commerce have not been
sufficiently tested for EDC activity, that number may be much higher. Besides
BPA (and its replacement, BPS), other common EDCs of concern include:
phthalates found in plastics, cosmetics, and fragrances; PCBs formerly used
(and still found) in industrial applications as coolants, lubricants, and
insulators; brominated flame retardant chemicals (PBDEs) in furniture and
electronics; and the ubiquitous pesticides glyphosate and atrazine. Dr. Tyrone
Hayes of the University
of California Berkeley
has conducted numerous studies on atrazine demonstrating the endocrine
disrupting effects on various frog species. Perhaps the most alarming of all
his findings may be the hermaphroditism
and feminization of
male frogs after exposure to atrazine at environmentally relevant
doses – doses at or below those found routinely in rivers and streams in
the United States.
While synthetic endocrine disruptors are the most commonly
discussed examples of chemicals that exhibit NMDR patterns of toxicity, they
are not the only substances that do. Heavy metals such
as lead, cadmium, selenium, arsenic, and manganese show NMDR patterns as well.
In fact, even though the presumption of monotonicity pervades all of risk
assessment, it is unclear whether even the majority of compounds actually do
act in that simplistic manner. What is clear, as Pete Myers states, is that “by
ignoring NMDR curves, risk assessment as currently practiced is deeply flawed
and unquestionably allows people to be exposed to harmful chemicals at
dangerous doses.”
One of the major flaws lies in the methods of chemical
toxicity testing. Most toxicity tests utilize a maximum of three doses as
reference points. As we know from basic algebra, plotting three points cannot
possibly lead to an accurate estimation of any curve. In order to determine the
level at which negative health effects might emerge, says Myers, “You need to
have tested an extraordinarily wide range of doses and have, preferably, at
least five doses across that range." He adds, "You can't say anything
about the absence of (NMDR) with just three doses.” Thus, with such a small set of reference
points, many substances could appear to follow monotonic dose-response with the
attendant fall-back assumption of safety a very low levels of exposure. But,
untested low doses could actually be the most harmful.
Further complicating determinations of safe levels of
chemicals, dose-response curves are specific to precise endpoints. Endpoints
are the biological outcomes – such as cancer, reproductive toxicity, or
neurological impairment for which toxicologists test. Even if all of the possible
endpoints could be or were tested for each chemical (which they are not), each
chemical may follow a different curve for each endpoint assessed. For example,
arsenic acts monotonically for cancer risk, but inflammatory
markers in the umbilical cord of pregnant women are lowest at intermediate
levels of arsenic exposure, demonstrating a NMDR curve for that endpoint.
Hence, the same chemical may be both safe and unsafe at the same exact level of
exposure, depending
upon which health effect one examines.
Another issue with establishing safe levels of any single
chemical through traditional toxicity measurements stems from the fact that
cumulative exposures are not accounted for, nor are aggregate exposures.
Chemicals in combination may act synergistically. Roundup herbicide, for
instance, causes cell
cycle dysfunction (which can lead to cancer) and apoptosis (programmed cell
death) in certain product formulations which contain different “inert” (yet
toxic) ingredients. These toxic effects are either not produced or produced to
a much lesser degree from glyphosate (the “active” ingredient) alone.
Additionally, the time of exposure within the lifetime of an
organism can determine whether or not the chemical produces toxic effects and
at what dose. Early development and puberty/adolescence are critical stages of
life (“windows of vulnerability”) at which exposure to toxic substances may
generate greater harm than at other life stages. Lead exposure in
children, particularly during embryonic, fetal, and postnatal periods, produces
neurological deficits that do not occur in equivalent adult exposures. By
overlooking additional complexities such as these in deriving safe levels,
chemical testing protocols as they stand are greatly in need of repair to
adequately reduce health risks.
In the face of such evidence that our notion of “safe
levels” of toxicants is outdated, why are such antiquated modes of risk
analysis still utilized to determine regulations? “Because too much money is at
stake” says Dr. Myers. “Using procedures capable of detecting NMDR curves would
be likely to require lowering a large number of reference doses so much that
the chemical would be required to be removed from the market.” The removal of
so many chemicals would more reliably ensure safety, but would impede
commercial and industrial profits.
Given the inadequacy of the current risk assessment paradigm,
changes are warranted to better protect public health. Tony Tweedale, founder
of RISK (Rebutting Industry Science with Knowledge) Consultancy, suggests
that studies must “test for the effects of real world doses” and “test the
whole dose response curve,” rather than simply a few high dose points. He also
advises drawing from the thousands of peer-reviewed academic studies for policy
decision-making, because “tens of thousands more experimental and supporting
etiologic and epidemiologic papers (are) being tragically ignored.”
Chemical regulations based on current unsound testing
practices cannot possibly be considered adequate. In fact, in 2014, the
National Academies of Sciences (NAS) offered updates to the EPA's traditional
risk analysis methods to better address NMDR and other deficiencies in chemical
risk assessment. Among their proposals is augmentation of risk evaluations to
include “statistical
considerations, uncertainty analysis, life stage or susceptibility issues, and
modes of action.” The EPA has yet to act on these recommendations.
Because of the faulty paradigm under which current risk
assessment and regulation proceed, one cannot confidently dismiss the
contribution of the innumerable commercially utilized chemicals toward human
diseases and negative health outcomes. As such, assertions by the FDA and EFSA
about the safety of BPA or other toxicants at current levels should be taken
with a note of skepticism. Cautions such as those now abundant in California should be
heeded.
A society that values human health and safety over
commercial growth would acknowledge the tremendous defects and scientific uncertainty
implicit in our current paradigm of assessing chemical toxicity. We cannot even
begin to approach a valid judgment of “safe levels” within the context of the
more than 85,000 chemicals currently in commerce (of which only a small
percentage have been tested for safety even under current protocols). Chemical
regulation based upon the precautionary principle would not only be relevant under
such conditions of uncertainty, it would be the most prudent option for the
benefit of public health.
Kristine Mattis holds
a Ph.D. in Environment and Resources. Email: k_mattis@outlook.com.
Comments