Health & Environment

PFCs: A case study in favour of the precautionary principle

PFCs are an example of how production and marketing of a substance can outpace scientific research into its safety and placing regulatory restrictions on its use. In the case of PFCs, this has resulted in 3 generations of people being exposed to an unknown hazard while a complex consensus, based on weak data and economic interests, develops around restricting their use.

Here we summarise the emerging issues around PFCs and argue that, had the precautionary principle been applied to their production, then we would not now be facing unknown risks from their continued use.

PubMed citation counts per year for PFCs. These graphs are not a reflection of the absolute number of all PFC studies but are to illustrate the research trend. Note that research is only picking up long after PFC production began. Click to enlarge.

Poly- and perfluorinated compounds (PFCs) are a minor miracle of chemistry, supremely versatile and produced in as many as thousands of varieties. Being soluble in water, yet also oil- and water-repellent, they are useful in a range of applications, from stain repellents such as Scotchguard to layering Teflon onto non-stick cookware.

PFC molecules gain their properties from two basic parts: a water-soluble head and a fluoridated tail. The fluoridated tail is highly inert, resistant to thermal degradation and environmentally stable. The head readily attaches itself to fabric and paper while the tail repels liquids and oils, allowing stain- and water-resistant coatings to be applied to clothing and grease-proof papers.

PFCs also find widespread use as surfactants – agents which allow two liquids that do not normally mix (such as fat and water) to dissolve into each other. This makes them useful as leveling agents in paints, as an intermediate in the production of fluoroacrylic esters and for the manufacture of fluoropolymers (such as Teflon).

PFCs have been produced in industrial quantities since the 1950s, beginning in the USA when firstly 3M and then DuPont began synthesizing PFOA. At that time, companies were under no statutory obligation to either perform rigorous testing or disclose any significant environmental or toxicity data they might have had on their products.

DuPont sounded early internal warnings about the safety of PFOA in 1961 and carried out further studies on workers in 1981. The company’s failure to notify the US Environmental Protection Agency of the results of these studies became the basis of a 2004 EPA case against the company for not meeting its obligations under the Toxic Substances Control Act.

Overall, the occupational and animal testing carried out up to the late 90s was broadly inconclusive. Although DuPont considered it had enough evidence to self-impose a limit of 1ppb emissions into waterways, the consensus at the time appears to have been that the sample sizes in the studies were too small to force regulatory restrictions on the use of the substance. DuPont’s occupational studies indicated an increased risk of prostate cancer among workers exposed to PFOS and some suggestion that PFOA is a teratogen, with an unusually high incidence of female workers giving birth to children with birth defects.

A watershed moment came in August 2001 when the first class-action lawsuit alleging property and health damage from PFC exposure was filed against DuPont by residents living near the company’s Wood County, West Virginia (USA) Teflon plant. DuPont was found guilty of knowingly exceeding its own guideline limits on PFOA release into local waterways and is liable for compensating residents for medical conditions identified as resulting from PFOA exposure (Clapp & Hoppin, 2009) By this time it was also clear that human exposure to PFOA was ubiquitous, with 98% of the US population having detectable levels in their blood (Calafat et al. 2007).

In the last decade we have since seen a rapid increase in the number of toxicity studies on the substances, but before then research was held back by several major challenges in assessing the toxicity of PFCs and their presence in the environment.

The first is determining whether or not they are even there in the first place. Typically present in the environment at parts-per-billion concentrations, until the early 2000s exposure levels were below the limit of detection of equipment typically affordable by laboratories. Prior to then, human exposure could inferred from occupational surroundings, or measured if high enough, but only looking at workers limited sample size and the ability to make correlations between exposure and health outcomes.

Food contact paper, thread sealant tape, stone, tile and wood sealants, and carpet-care liquids have the highest concentrations of PFCs. Source: Wikipedia / Guo et al. (2009). Click to enlarge.

The second challenge is the often almost extreme, yet highly unpredictable, persistence of PFCs in the body, which varies tremendously not only from one species to the next but even one gender to the next. For example, PFOA half-life is 6 days in male rats, only 4 hours in females, but 5.4 years in humans (in whom gender makes little difference at all). The PFC found to be most persistent in humans so far is perfluorohexanesulfonate (PFHxS), with a half-life exceeding 8 years.

Long half-lives complicate epidemiology and understanding of exposure routes of people to PFCs, as modifications to behaviour, diet or occupation can take years before they result in measurable changes in PFC levels in a person’s body. This makes drawing out correlations between behaviour, diet, and occupation and exposure, and then exposure and health outcomes, much more difficult.

Academic research into PFCs really took off when affordable, reliable detection technology became available. This made it easier to perform more fine-grained exposure assessments and epidemiological studies, precipitating a relative flood of toxicological research beginning around 2006.

These more recent studies have in particular found associations between PFOA and higher cholesterol levels, with implications for cardiovascular disease, and elevated uric acid levels, which may increase risk of hypertension and cerebrovascular disease. Steenland et al. (2010) provides a good open-access summary of the epidemiological data.

Exposure routes consist of the usual suspects, with particularly elevated PFC levels seen in people who consume fish or water from contaminated waterways. House dust may be a significant source of exposure, especially for toddlers with their hand-to-mouth behaviours. There is some evidence that crops grown on contaminated soil may accumulate PFCs. This may present an exposure issue for the general population as sewage sludge can be relatively heavily contaminated with PFCs and their degradation products. Overall, food is considered to be the source of 90% of people’s exposure to PFCs, although these figures are produced by modeling subject to the many uncertainties described above.

Evidence of general exposure, toxicity, persistence and bioaccumulation of PFOS has accumulated to the point where regulatory action is beginning to be taken. PFOS and related compounds (numbering 96 in total) have been listed under Annex B of the Stockholm Convention, classifying them as persistent organic pollutants (POPs) and restricting their use to essential applications. The US EPA set a voluntary target of 95% reduction in emissions in PFOA and related compounds by 2010; the 8 companies which signed up to the targets (including 3M and DuPont) have shown a 70-100% reduction in emissions since 2001.

Overall, of the many PFCs it is only the use of PFOA and PFOS which are being subjected to the majority of research and regulatory scrutiny. Other PFCs such as PFHxS and perfluorononanoic acid (PFNA) are being detected in increasing concentrations in environmental and human samples. Perfluoralkyl phosphate esters (PAPs) and phosphoric acids (PFPAs) have recently been found in environmental matrices at concentrations similar to PFOA and PFOS. PAPs have also been found in human serum.

Measures to reduce exposure to PFOS and PFOA are meeting with some success, at least in Western nations, with US NHANES data showing recent declines in levels in humans which are consistent with the reduction in production in the US. Norwegian data show similar declines. However, in China PFOS exposures have dramatically increased, although PFOA exposure remains relatively limited.

We are only just now beginning to develop a picture, impossible 60 years ago, of how people are exposed to PFCs and what the consequences of this may be. What this means, however, is that research and regulatory initiatives are only just beginning to gather momentum at a time when there are already as many as several thousand PFCs in production, with unknown combined toxicity, to which 3 generations of people have now been exposed to an unknown degree.

It may turn out to be the case that PFCs are, for the most part, relatively harmless. However, we are going to determine this by gathering data from the millions of people being exposed to these substances before we have determined whether or not they are safe. We have done this many times already, with PCBs, asbestos, formaldehyde, dioxin and lead – tallying up the damage in human lives before taking action – so we had better hope that PFCs really are safe.

Generic PFC structures. The 8-carbon (8C) structures are shown. The 8C, or "long-chain" structures are generally the most persistent. Click to enlarge.

PFCs are ultimately another illustration of the importance of the precautionary principle in chemicals policy. Currently, the economic benefits of continued PFC production are being weighed against unknown harm they pose to the environment and health. However, even for PFOA and PFOS, thorough research only goes back about 13 years with the majority of studies only being published in the last 8 years, and toxicity studies only in the last 5. This is not the sort of data which can prove that PFCs are harming health, so cannot form the basis of a consensus that harm to health outweighs the economic benefits of continuing to market PFCs.

This mindset, however, results in data on PFCs being generated by measuring people’s exposure and correlating this with increased incidence of ill health. This requires we make people ill in order to justify restricting the use of PFCs – surely an unacceptable way of finding out the PFCs are harmful, yet exactly what has happened with PFOS and PFOA for 60 years and counting.

The alternative is to make sure a pedigree of safety for a substance is established before it is brought to market. This application of the precautionary principle is not reactionary or over-conservative. It simply acknowledges that the slow consensus-building process we see with PFCs, where existing economic benefits and producer interests are balanced against weak data on health effects, is inadequate for protecting the health of the millions of people exposed to PFCs over the decades it takes for that consensus to emerge.

If the precautionary principle had been in play when PFCs were first discovered, we would have been much clearer about the environmental and health risks they pose and been able to make a rational decision based on clear data about whether or not they should be manufactured: if the data was not available, they would not have been produced, and we would face no risks now; alternatively, if the data had shown they were safe, they would have been produced, and the likelihood of our facing a risk now would have been substantially reduced. Instead, we have a muddy process based on emerging data where all we know is, if PFCs do turn out to be harmful, we will regret ever making them in the first place.

Further Reading

Lindstrom AB, MJ Strynar, E Laurence Libelo (2011). Polyfluorinated Compounds: Past, Present, and Future. Environmental Science & Technology 2011 45 (19), 7954-7961

Impasse on BPA: different facts – or different values?

Animal studies suggest exposure to even very low levels of BPA may be harmful to health. Regulators, however, generally describe these studies as being irrelevant to risk assessment. Are they really meaningless? Or is there something the regulators need which the studies aren't giving them?

We are coming towards the end of two years of hearing very mixed messages about the safety of BPA.

On the one hand, in 2009 the US Endocrine Society concluded there is a strong basis for concern about the human health effects of endocrine disrupting chemicals such as BPA (Diamanti-Kandarakis et al. 2009).

In March 2010, Denmark introduced restrictions on the use of BPA in food contact materials for children under 3 years of age. The same month saw Health Canada prohibit the use of BPA in baby bottles.

On the other hand, in October of 2010 the European Food Safety Authority (EFSA) found none of the new research on BPA indicated a need for reducing its tolerable daily intake (TDI) of the substance, or that current use poses a health risk to anyone in the general population (EFSA 2010). The UK Food Standards Agency backed EFSA’s findings.

In 2011 the Advisory Committee of the German Society of Toxicology (ACGST) concluded in a review of the criticisms of EFSA’s evaluation of the safety of BPA that none of the counter-arguments to maintaining the existing TDI are compelling (Hengstler et al. 2011).

These are all bodies with substantial scientific expertise, yet clearly divided in opinion between those who believe BPA is safe, at least at current exposure levels, and those who believe exposure to BPA should be reduced. They all have access to roughly the same data, which they should be able to analyse objectively. So the question is: how has this happened?

The place to look for the reason for the impasse may well be in each side’s conception of what counts as adequate science in evaluating and responding to the possible threat to health posed by BPA. To this end, Health & Environment interviewed two scientists with opposing views on BPA regulation.

On one side of the debate is Professor Jan Hengstler, Project Group Leader for Systemic Toxicology at the University of Dortmund’s Leibnitz Research Centre for Working Environment and Human Factors (IFADo), and corresponding author of the ACGST review of the safety of BPA, which endorsed the existing TDI for BPA.

On the other is Professor Gilbert Schoenfelder, a pioneer of BPA biomonitoring at the German Institute for Clinical Pharmacology and Toxicology (and, like Hengstler, also a member of the German Society of Toxicology), who believes that BPA is inadequately regulated and likely to have a range of harmful effects on health.

Sound science?

Hengstler describes the process by which toxicologists assess the safety of BPA: “You do animal experiments to determine the level at which BPA has no observable adverse effect (NOAEL) and divide that by several safety factors to obtain a TDI for humans.” Regulators then ensure that these levels are not exceeded in humans.

Hengstler repeatedly emphasises how regulatory toxicological studies and reviews are designed to derive a safe dose level for BPA with a very high degree of statistical certainty. “The criteria for scientific validity of a regulatory study is tighter than those required in a peer-reviewed journal,” he says, explaining that in journals, 5% confidence intervals provide enough surety that a result is not a fluke and sufficient for it to be published.

Because regulatory studies are designed to have the last word on an issue, they have to eliminate, as a far as possible, the chance of their findings being false positives. These confirmatory studies have much tighter confidence intervals, with much larger groups of animals being studied.

The confidence of toxicologists in the findings of a study are further enhanced by following agreed protocols for studying harm such as the standardised Organisation for Economic Co-operation and Development (OECD) guidelines (or see e.g. EFSA guidance on oral toxicity studies). More controversially, Good Laboratory Practice (GLP) compliant studies are also preferred by regulators because the exhaustive documentation is believed to further enhance a study’s reliability.

The need for certainty is also manifest in the definition of “adverse effect” as captured in the NOAEL. “It is not enough to see that a compound causes a change, we need to be sure that the change is adverse,” says Hengstler. “For example, glucose alters gene expression but these changes aren’t associated with adverse effects.” Any potential indicators of harm, or biomarkers, therefore “need to be validated if they are to be used for regulatory purposes,” says Hengstler.

Without formal agreement on what the studies mean for humans, a toxicologist cannot use a study to modify the TDI. This is clear in EFSA’s Opinion on BPA: “The Panel… concluded that no new study could be identified, which would call for a revision of the current TDI… The Panel noted that some studies  conducted  on  developing  animals  have  suggested  other  BPA-related  effects  of  possible  toxicological relevance…  At present the relevance of these findings for human health cannot be assessed.”

Of course, statistical certainty means very little if precise studies are looking at the wrong evidence. A study protocol can grant almost absolute certainty that exposure to BPA has no effect on the size or basic function of the liver in a lab animal. This will not, however, tell you anything about BPA’s toxicity in the reproductive or cardiovascular system.

Toxicological testing of BPA is of course more complex than that. Nonetheless, many scientists, of which Gilbert Schoenfelder is one, are unconvinced that the tight standards of regulatory studies have produced accurate assessments of the safety of a substance such as BPA.

These scientists argue that regulatory studies are poorly-equipped for evaluating the safety of BPA, looking for effects which low doses of BPA can’t produce, and missing the toxic effects which low doses of BPA actually produce.

“Independent studies find all sorts of things,” says Schoenfelder, “and try to get clear on what this means for humans. The risk assessors just say they don’t know what this means for humans, so disregard it.”

Schoenfelder agrees about the need for standardised protocols. “We need direction for companies on the basis of what could, at the moment, be the most sensitive test system to exclude a risk to humans. This is why we have study guidelines for industry,” he says. “But we have to ask: how old is the guideline? Are there techniques which are more sensitive? If we have data and information from studies which don’t conform to guidelines, then what is the meaning of those studies?”

Political action

Hengstler believes exploratory research should precipitate discussion of whether more research is needed or study guidelines should be changed: “If you find evidence of an adverse effect it should of course be published, but it does not mean that a single, published exploratory study is a basis on which to ban a compound. Such studies are the start of a discussion process.

“There is nothing wrong with publishing these, but if they aren’t reproduced you have to conclude that the findings are a false positive – which is no basis for regulation.”

Hengstler’s position fails to satisfy the likes of Schoenfelder because they feel that uncertainty in the TDI should not simply start discussion or trigger a more precise research programme, but rather constitutes a reason for doubting the TDI at all.

“What is the meaning of a guideline compliant study if it doesn’t look at things which are being exposed by new research?” asks Schoenfelder. “If there is a good, published paper then it counts as positive evidence, and then we need more research for an explanation or a mechanism, or new research on the topic.”

It is adherence to the exacting standards of regulatory toxicology which toxicologists equate with scientific rigour: where research scientists see uncertainty and a questionable TDI, regulatory toxicologists see a carefully-calculated, validated TDI which could in future be subject to change but only if the data is certain enough to warrant it.

Regulating without certainty, as Schoenfelder advocates, is anathema to the toxicologists’ approach, and may explain some of the resistance to regulation of BPA and description of moves in favour of its regulation as political rather than science-based.

Since both Schoenfelder and Hengstler favour rigour in risk assessment, their difference in opinion can be interpreted as resulting from their attitudes about what regulatory steps should be taken in the face of fresh evidence of possible harm, before guidelines have been developed.

One may not agree with recent moves to restrict the use of BPA. However, there is nothing inherently unscientific about acting early on preliminary findings if these undermine confidence in the accuracy of a TDI, so it is probably not helpful to describe the precautionary moves by Denmark, France and Canada as such.

Regardless, regulation seems to be going the way of Schoenfelder’s precautionary approach. The temporary Danish restrictions on use of BPA, citing continuing uncertainty about the neurodevelopmental effects of BPA, the similarly temporary French restrictions, and also the recommendations of Health Canada, all point to a change in approach to risk assessment.

This is, if new evidence emerges which casts doubt on the TDI of a substance, then use of the substance is restricted until the methods for evaluating the substance have been updated, with restrictions due for reversal if it is subsequently proven safe.

This change matches the shift in who is responsible for the safety of chemical products, formalised in REACH, from the onus being on the regulator to prove that a substance is harmful before it can be restricted, to a manufacturer having to prove a substance is safe before they can introduce it to the market.

REACH is unique insofar as, to a greater or lesser extent, it forces manufacturers to demonstrate that their products have an adequate pedigree of safety before they can be brought to market. The case of BPA demonstrates there is not as much confidence in the existing guidelines and study protocols as some parties to the debate about chemical safety might think there should be.

Controversy surrounds BPA precisely because of the lack of agreement over what counts as an adequate evaluation of a chemical’s safety. This is not politics. It is not even, strictly speaking, just about the safety of BPA. Representing it as being a simple issue of science vs. politics obscures much of the common ground between the two sides on the bigger issue of the modernisation of chemicals policy, of which BPA happens to be the poster-child.

Common ground?

In spite of their differences, during interview both Hengstler and Schoenfelder independently described the need for transparent, reliable testing based on robust data; the need for rapid development of guideline tests to incorporate new research findings; and the need for easy access to all relevant data and transparent evaluation open to the public.

Regardless of how one feels about BPA, there may be a way forward yet.