Non-targeted testing for potential adulterants consists of analytical measurements sensitive to multiple contaminants and can be used as a first-line screening tool to detect adulteration.
USP wants comments from industry experts and analytical chemists using untargeted methods.
EMA-related adulterants are typically present in concentrations consistent with sensitivity limits of non-targeted methods (above 0.1% concentration).
When setting model boundaries there is a trade-off between sensitivity and specificity rate of the method, where the risk willing to be accepted must be balanced against possible increased inventory management and test costs.
Steps behind the criminals
Jeff Moore, senior science director at USP, said food fraud is a complex problem and one of the mitigation measures a lot of organisations use is testing supply chain ingredients to look for it.
“But the criminals that engage in food fraud and economically motivated adulteration ultimately know what people are testing and how they are testing and are always finding creative ways to cheat the system and to create new adulterants,” he told FoodQualityNews.
“That leaves the people doing testing feeling like we are two or three steps behind the criminals in this area. Using melamine as a case study, when that happened, people really quickly developed good methodologies to look for melamine as a known analyte. But the question then became what about the next melamine, how do we know what that is, how can we detect that?
“We’ve seen untargeted methods being used very widely in areas that are fraud prone so juices are a good example, there are a lot of organisations looking at untargeted methods using techniques like NMR to look at juice authenticity. We’ve seen it used in areas like spices as well so untargeted methods looking for foreign materials and adulterants in different spice materials.”
Moore said it has led organisations to a new approach and philosophy which is non-targeted or untargeted testing.
“It’s where you use technology like infrared spectroscopy, or Raman or mass spectrometry to create a signature or a fingerprint for an ingredient,” he said.
“You define what the normal characteristics are for that ingredient and when you test raw materials you look for differences, you look for something that looks unusual that tells you there is something different about that ingredient.
“It could be fraud or it could be a quality issue but it creates a way for you to get that unusual material out of your supply chain and do more sophisticated testing on it but ultimately solves this problem of not knowing what the next adulterant is going to be. It allows you to look for anything unusual in those ingredients.”
In the targeted methods area there are a number of organisations that develop standards and guidance including USP, AOAC, ISO and CODEX.
The guidance is part of USP’s Food Chemicals Codex Forum (FCC) – a free online resource where users can review and comment on proposed standards for food ingredients.
Moore said the guidance is applicable to any type of technology or instrumentation that can be used to do non-targeted testing.
“So whether it is near infrared, mid-infrared, mass spectrometry; the idea was to create a framework in terms of the different steps and to standardise the terminology across the board so regardless of what sort of approach you are using that everyone can come out of it with an understanding of what the performance of that methodology is,” he said.
“Some of the things we’ve included are terminology, so for example defining what a non-targeted method is, there has been confusion in literature around what an untargeted method is, we’ve looked at it through the lens of food fraud. How do you translate performance characteristics used traditionally for targeted methods to their equivalent in the untargeted space?”
Moore said it has a step-wise approach that it suggests people go through when developing a method.
“It’s as simple as before you invest in developing a method, what are the basic questions you should be asking yourself and we call that the applicability statement so what is the problem you are trying to solve and when do you know you are successful in solving that problem?
“How do you procure and test authentic materials or ingredients to make sure you really understand the common characteristics of that ingredient. How you then challenge an untargeted method, so what’s the thought process behind how you would spike an adulterant as to really understand the performance of that methodology.
“Lastly, how you maintain - like any analytical method you don’t create, validate and walk away from it, you have to continually monitor the performance of that methodology. It is very important in the untargeted space when you are looking at complex food ingredients that can change over time.
“So things like a dairy ingredient where the composition depends on the breed of cow and the lactation cycle, so you’ve got to make sure you monitor the characteristics of those ingredients and update your methodology accordingly.”
Apples to apples comparison
Moore said USP has been working in the space for about six years developing and validating untargeted methods.
“[Our work has been] largely around milk ingredients which are a fraud prone area and some of the research that we did led us to understand that there was no guidance, no benchmark standard around the framework of thinking for developing and validating these methods that would allow an apples to apples comparison between different methodologies in terms of performance,” he said.
“If you are talking about how sensitive is a method to adulterants, doing that for an untargeted method requires a different approach than targeted methods, so we developed draft guidance around how to develop and validate these methodologies.”
The 20 page guidance document was published in November and comment closes at end of March.
Publication date depends on the first round of feedback and the comments that come in, said Moore.
“If we generally got it right and just need to make some minor updates we plan to publish in September 2017. If we get feedback that leads us to completely revamp this then it could be a longer timeframe before we publish in the final form,” he said.
“We’ve received more than 60 comments from different organisations so far, we’ve attended two international conferences focussed on untargeted methods and had a lot of good dialogue and feedback so we think we’ve generally got it started down the right path and likely will be some minor updates and enhancements to make it more useful for industry and government agencies.”