FDA-approved PFAS: A serious breakdown in assessing food additive safety

Tom Neltner, J.D., Chemicals Policy Director, and Maricel Maffini, Ph.D., Independent Consultant

This blog is the fourth in a series describing information we discovered in reviewing thousands of pages from the Food and Drug Administration’s (FDA) response to our Freedom of Information Act (FOIA) of the agency’s approval of 31 Food Contact Substance Notifications (FCNs) from 2002 to 2016 submitted by six companies for 19 distinct chemical mixtures of per- and poly-fluorinated substances (PFAS).

In this blog, we identify one company’s serious breach of its obligation to provide FDA with all relevant toxicology data. While hindsight is 20/20, we have reason to believe that if FDA had had all relevant information, it would have demanded more studies potentially revealing risks that are only now coming to light with related chemicals. Though we have not completed a similar review for the other companies, we think this inadequate approach to chemical safety is not unique to a single company, and FDA should reassess all its reviews given what is now known about PFAS chemicals.

Safety assessment requirements for food additives – including food contact substances

When a company seeks FDA’s approval of food additives (including food contact substances), it is required to provide the agency with all relevant chemistry, toxicology and environmental data so it can conduct a safety assessment. While the agency typically conducts a literature search of its own and of public databases, the company that is claiming the chemical’s use is safe is obligated to include any data that is inconsistent with the company’s conclusion.

This requirement is essential because, as part of the safety assessment, the agency must determine that there is a reasonable certainty of no harm from the chemical’s intended use considering three factors: 1) the probable consumption of the chemical due to the use; 2) the cumulative effect of chemically- and pharmacologically-related substances in the diet; and 3) appropriate safety factors. When data are omitted, the agency could miss critical information that would prompt it to raise questions, demand more studies, and possibly refuse to approve the use.

As a practical matter, this review is critical because it is FDA’s best chance to get it right; the agency rarely looks back at the safety of a chemical unless there is a later notice or other request for the same substance. And when it does reassess safety, the agency appears to mistakenly think it must prove the chemical is unsafe rather than show its use no longer meets the standard of safety. Therefore, the agency may never learn of an omission or act on it if discovered. For instance, FDA has never withdrawn an approval for a food contact substance notification (FCN), not even when it raised concerns about the safety of long-chain PFASs.

Since 2000, FDA has reviewed the safety of a food contact substance (FCS) whenever a company submits a FCN. Until the agency approves the notice, the law requires that the agency evaluate it out of public view. If it approves the notice, the agency only posts the final approval and the environmental assessment on its website. A FOIA request is needed to get the notice and the agency’s evaluation – a process that can take months and results in a heavily redacted response. If the notice is withdrawn – which happens about 28% (530 of 1913 notices) of the time – FDA cannot release the notice in response to a FOIA.

Daikin’s failure to inform FDA on an animal study showing liver and kidney effects

From 2008 to 2016, Daikin, a Japanese company that is a major producer of fluorinated chemicals, submitted eight FCNs for PFAS. Our FOIA covered five – FCN# 820, 827, 888, 933, and 1044 – for PFAS polymers made from the same chemical known as 3,3,4,4,5,5,6,6,7,7,8,8,8-tridecafluorooctyl 2-propeonate.[1]

This chemical has six carbons that are fully fluorinated (commonly called C6) and two non-fluorinated carbons. For convenience, we will refer to this chemical as C6SFA, one of several names Daikin uses for it. The other names are perfluorohexylethyl acrylate, 13FA and 13F-SFA. The company’s PFAS manufacturing facility is in Decatur, Alabama and its wastewater is discharged into the Tennessee River upstream of several recreational lakes.

We reviewed the toxicology information Daikin submitted to FDA for each of the five FCNs and compared it to human health effects studies the company listed in its website. We mostly focused on the FCN 820 because it was the first one submitted to the agency using C6SFA. The notice included three genotoxicity studies – two in vitro and one in vivo mouse study.

To our surprise, the company did not submit a month-long feeding study in rats showing that the liver and kidney of the animals treated with C6SFA were heavier and showed pathological changes. Also, the rats’ incisor teeth showed signs of decreased mineralization (mottled teeth) associated with ingestion of excessive fluoride. The study concluded that “it was considered that [C6SFA] had mainly effects on the incisor, liver and kidney” and the no observed adverse effect level (NOAEL) was considered to be 5 mg/kg bw/day based on increased relative kidney weight in males at the next largest dose, 25 mg/kg bw/day (See Table 1 below). This is consistent with the type of harm seen with longer chain PFAS.

Table 1. Comparison between the 2007 and 2014 28-day repeated dose oral toxicity studies in rat

The notice also did not include an animal study that measured the dose necessary to kill half of the tested animals as well as one additional in vitro study.

Beyond the consumer exposure to PFAS uses in contact with food, the issue is important for drinking water systems downstream from the facilities where the chemicals are made and applied to paper and other materials. In a previous blog, we reported that Daikin estimated that as much as 225 pounds of the PFAS made from C6SFA may be released from a paper mill using it to greaseproof paper for food use.

Daikin repeats the missing study in 2014

On Daikin’s website,[ii] we found that the company repeated the rat study of C6SFA seven years later, in 2014, “to confirm the presence or absence of the toxic effects at 30 mg/kg/day and reproducibility [of] the test results.” This typically happens when a company considers increasing potential exposures to levels that may be above the current NOAEL.

The study used only two doses instead of the typical three and the amounts of C6SFA were slightly different from the earlier study. However, the results were very similar to the 2007 toxicity test: the weight of liver and kidneys was increased, the microscopic structure of the liver and its function were altered. Despite this evidence, Daikin selected a new NOAEL of 30 mg/kg bw/day even though that dose increased the weight of the liver. Table 1 below compares the results of the two studies.

Because our FOIA did not include two FCNs approved by FDA in 2014 – FCN# 1360 and 1451 – that used C6SFA, we do not know whether Daikin provided FDA with the studies in those notices.

In addition, we do not understand why in 2014 Daikin would undertake a new animal study and not design it to evaluate the more sensitive endpoints associated with similar PFAS such as reproductive and developmental harm. While these might not have been clear in 2008, there was growing evidence for similar chemicals by 2014.

More-common omissions in industry food additive safety assessments

As noted earlier, a chemical safety assessment must consider the cumulative effects of chemically- or pharmacologically-related substances in the diet. In our eight years of reviewing industry safety assessments we often see this factor narrowly defined or not mentioned at all. Therefore, we reviewed the FOIA response to see whether the companies included studies on other PFAS that were similar in the chemical structure or the organs affected.

We did not find any FCN that assessed the toxicological cumulative effect of similar PFAS. However, because some documents are heavily redacted, we cannot rule out that similar PFAS were identified as members of the same class and that their cumulative effects were discussed.

More significantly, FDA noticed these omissions. In its toxicology assessment memo of FCN 820, the agency’s toxicologists identified five FCNs using PFAS that appear very similar to C6SFA:

• Hercules FCNs 542, 746 and 783: 2-propen-1-ol, reaction products with 1,1,1,2,2,3,3,4,4,5,5,6,6-tridecafluoro-6-iodohexane
• Asahi FCNs 599 and 604: polyfluorooctyl methacrylate

Inexplicably, FDA did not appear to request Daikin or any of the other PFAS manufacturers that submitted FCNs to assess safety as a class of chemically-related substances in the diet.

Potential biopersistence of similar C6 chemicals

As we reviewed Daikin’s webpage, we realized there were studies, beyond toxicology, that were also of interest because they dealt with the breakdown products of C6SFA in the environment by microorganisms, in activated sludge and in water in various pH conditions. For instance, we learned that degradation products of C6SFA are tridecafluorooctanoic acid and 2-(perfluorohexyl) ethanol (aka 3,3,4,4,5,5,6,6,7,7,8,8,8-Tridecafluorooctyl alcohol), a type of fluorotelomer alcohol known as 6:2 (six fully-fluorinated carbons and two non-fluorinated carbons).

These fluorotelomer alcohols are significant because earlier this year FDA scientists published a peer-reviewed journal article stating that these alcohols are components of high-molecular weight polymeric food contact substances such as those approved by the agency. More importantly, the scientists concluded that 6:2 fluorotelomer alcohols have “high biopersistence potential.” We haven’t been able to find any information on biopersistence in the 31 FDA-approved FCNs covered by our FOIA.

Conclusion

Both FDA and companies must apply the legally mandated requirements when performing chemical safety assessments of food chemicals. This includes identifying chemicals that belong to a class because they are chemically-related or because they cause similar biological effects in animals or humans. The agency applied the concept of chemical class only once, when it decided that three long-chain PFAS were unsafe. This should be the norm not the exemption.

Given these problems, we maintain that FDA should demand that Daikin update its notification to include the missing data and reassess whether the use is safe for public health and the environment. While we have not completed a similar review for the other companies, we think a similar reassessment is needed for all PFAS given what is now known about these chemicals.

Lastly, manufacturers should support the agency’s long-promised effort to make information available to the public with the necessary protections for confidential information. Transparency is a first step to reestablish the public trust that has been eroded by decades of excessive reliance on incomplete data and old scientific assumptions.

[1] We have not yet submitted a FOIA for three FCNs submitted after 2011 including two – FCN# 1360 and 1451 – which appear to be based on the same C6SFA monomer.

[ii] Daikin appears to have deleted the webpage, but we tracked down the old page on an Internet archival service.