The pollution of the environment with per- and polyfluoroalkyl substances (PFAS) is an unprecedented societal challenge. The spread of these long-lasting, bioaccumulative and potentially toxic compounds through the environment has resulted in food and drink becoming an important vector for human exposure.
To properly protect public health and monitor PFAS exposure routes, a comprehensive suite of analytical tools is needed.
Speaking at the 11th International Symposium on Recent Advances in Food Analysis (RAFA), Dr. Stefan van Leeuwen, senior scientist at Wageningen Food Safety Research, presented his vision for a “comprehensive toolbox” for investigating PFAS in food.
The challenge of analyzing PFAS in food
Depending on the data source, the estimated number of PFAS compounds in existence ranges from approximately 5,000 compounds under the OECD definition to the 7 million listed in the PubChem database.
“It’s a wide compound class, varying in chain length and functional group,” van Leeuwen told the RAFA audience. “They have a wide range of physicochemical properties – some are very water-soluble while others accumulate in fish, for example, or in liver tissue. Others are very volatile. There is a lot of variability among these compounds.”
In addition to the sheer number of possible PFAS compounds that must be tested for, food is inherently a very complex matrix for analysts to work with.
“One other thing — if you would like to do an exposure assessment, we know that the European Food Safety Authority (EFSA) opinion has set very low benchmark levels,” van Leeuwen said. “So in order to be able to do a meaningful exposure assessment, we need to go to very low detection limits — in the low nanogram per kilogram LOQs”
The diverse nature of PFAS and the matrices that they are being analyzed in necessitates a multi-analytical approach, van Leeuwen explained. Using a “toolkit” of different analytical methods, all aiming to analyze PFAS levels in a slightly different way, researchers can build up a better picture of PFAS contamination and the associated exposure risks.
“There is no one-size-fits-all solution. We need to have multiple analytical approaches, rather than a single method,” van Leeuwen urged.
The PFAS analysis toolkit
The analytical toolbox that van Leeuwen presented to the RAFA audience contains three central pillars:
Targeted analysis, for the determination of known PFAS in food.
Mass balance analysis, to determine total PFAS contamination levels.
Unknowns identification (i.e. non-targeted analysis), to screen for and assign identities to any unknown PFAS in the sample.
Targeted analysis focuses on the analysis of specific PFAS compounds. This will often be the PFAS that are named in relevant national or international regulation, or individual PFAS that are of interest to the analyst.
Targeted analysis techniques – such as liquid chromatography tandem mass spectrometry (LC-MS/MS) – also require each PFAS analyte of interest to have an existing analytical reference standard that can be used for quantitative analysis.
“One of the tools that we use, and that I think many other laboratories have, is a triple quad mass spectrometer (TQMS) approach,” van Leeuwen said. “It is very selective, very sensitive. We use that for routine analysis and it works very well for dietary exposure assessment.”
“Things may vary a bit here and there, but this is really our workhorse for targeted analysis,” he continued.
van Leeuwen and colleagues recently published an analysis of PFAS levels in fruits and vegetables grown in allotments up- and down-wind from a fluorochemical production plant near the Dutch city of Dordrecht. Using a targeted ultra-performance UPLC-MS/MS analysis approach, they found that PFOA and GenX were the most prominent PFAS observed. The highest PFAS levels were observed in brassicas (i.e. cabbages, cauliflower, broccoli), leafy greens and root vegetables, with contamination levels being highest directly down-wind of the fluorochemical plant.
“Targeted analysis helps us with the determination of the knowns. But what about the unknowns?” van Leeuwen said. “In my view, that breaks down into two questions: what is the identity of the unknowns, and what is the mass balance — how big is the contribution of the unknowns to the totality of PFAS present?”
Continue reading below...
Exploring PFAS Analysis Techniques
Exploring the base of the PFAS iceberg
Known PFAS are just the tip of the iceberg, as far as PFAS analysis goes. But to uncover the substances that might be lurking underneath the waves, researchers first need to know how big the “iceberg” they are dealing with is.
“If you were able to analyze the total amount of PFAS in a single way, and if that balanced with [the amount seen in] your targeted analysis, then you know that you're fine — you’ve found all the PFAS that were in your sample,” van Leeuwen said. The implicit conclusion is that if your PFAS totals do not balance, then you have more work to do.
To analyze the total amount of known and unknown PFAS in a food sample, researchers use techniques that can determine the total amount of extractable fluorine in a sample.
“We use combustion ion chromatography (CIC) for extractable PFAS analysis,” van Leeuwen said. “You get your sample or your extract, you burn it down and mineralize it, then you determine the fluorine content using ion chromatography. That gives you a feel of the amount of organic fluorine that was in your sample.”
Total extractable fluorine and the total PFAS burden are not exactly equivalent, as there may be other organofluorine compounds present depending on the sample and matrix composition. However, it is widely accepted as an acceptable proxy measurement.
If the PFAS levels measured during targeted analysis do not closely correlate with the total extractable fluorine levels, then the next step of the “analytical toolbox” dictates that non-targeted analysis should be done.
“The third technique, to try to identify these unknowns, is the non-targeted analysis and advanced data filtering that we apply,” van Leeuwen said. “That helps us figure out which unknown compounds are there in a sample.”
There are many different ways to conduct non-targeted analysis for PFAS. In van Leeuwen’s lab, they have developed a non-targeted screening approach called fragment ion flagging, or FIF.
“If you run your extract on [the mass spectrometer], they get a lot of signals. We can get over 10,000 features that are detected. And then, of course, you need to figure out which of these are relevant. So that brings you to data filtering, because you would like to extract the data of the PFAS from the bulk of the data,” van Leeuwen explained.
“A couple of years ago, we worked on something that we called fragment ion flagging. What that is, is you look at specific fragments for these PFAS – because they do fragment in a specific way – so that can help in the identification of the unknowns.”
The FIF data filtering procedure can be used with any high-resolution mass spectrometer (HRMS) system, coupled with either gas chromatography (GC) or liquid chromatography (LC), to screen for and discover novel PFAS.
“What I would like to tell you, and what I hope I have been able to show, is that we have three complementary methods that are available to detect the known, the total or extractable amount of PFAS, and finally, also the unknown PFAS,” concluded van Leeuwen, addressing the RAFA audience. “And of course, once you have found an unknown one and you're able to identify it, you can add it to the known ones in your targeted methodology.”