By Aurel Covaci – Scientist, Functional Ingredients Laboratory, Reading Scientific Services Ltd
Good science is fundamental to the production of good food. Whether that science is being applied to address issues of safety or authenticity, provenance or palatability, the industry needs reliable, accurate and robust methods that deliver the correct answer to a wide range of questions. These questions include, but are not limited to: Is this product free from pathogens? Is the ingredient even what it claims to be? Are the ingredients listed on the label actually in the product at the end of shelf-life?
Some of these questions are easier to answer than others. Indeed, science has been answering some of these questions for many years. We know we have reliable methods that can tell us quickly and accurately whether milk contains listeria, or the provenance of a piece of meat i.e. is it chicken, pig etc. The methods utilised may still require specialist expertise to apply them, but we know they can be trusted if carried out correctly.
But what of the new questions about new ingredients and new products/formulations? These new questions require new methods, but, method development has the potential to be time consuming and costly. In a food industry that is anxious to get innovative products to market as quickly as possible, "time consuming and costly" is one phrase nobody wants to hear.
Having developed a method that provides an answer to the analysis required, how can we be sure that we are getting an accurate result? Method development is as much about validation as development.
Product developers, when it comes to using ingredients, are not restricted in their choices. There are, after all, hundreds if not thousands of different ingredients that can be used in a food product, leading to an infinite number of potential formulations.
The situation is made even more complicated by the natural variation that occurs within ingredients. So, whilst some are simple to characterise others are not. Everyone knows that water is H2O, but water from different sources will contain different concentrations of trace elements. So even water, is seldom simply water. However, when we consider other 'single ingredients' such as a strawberry or potato, that are themselves made up of many different chemicals (that can vary between growing regions/conditions, seasons, varieties etc.). It is clear to see that the composition of even the simplest foods can be extremely complex at the level of individual chemicals.
Knowing what goes into a food is not necessarily a guide to what remains after processing or storage. There is considerable potential for chemicals to react, evaporate or change character.
Don't be fooled by the TV
For the team or individual working on product development, it might still be tempting to think it should be simple to determine the chemical content of a finished product. Let's say there's an established product or range to which they intend to add an exciting new ingredient. From the developer's perspective, there is one basic question: Is the ingredient in the finished product still present in sufficient quantity, at the end of shelf life, to satisfy the labelling claim?
What could be simpler? All they need is an efficient analytical test, for the parameter of interest, to substantiate their claim. They have watched the detective programmes on TV, they know there is a black box and a computer that can solve impossible problems, and that results come back within the hour.
Sadly, it's not so simple in real life, and whilst the modern laboratory does indeed have some extremely powerful, sophisticated and sensitive instrumentation, there is no one ‘box’ that gives all the answers.
Moreover, merely in preparing the sample for analysis there is potential for loss and interference, which makes it vitally important to validate the new method and ensure that it works in all cases.
The route to new method development begins in understanding the chemistry of all the ingredients in a product, before working out what testing can and cannot be done. If we use the case of strawberry, we can all see that it's easy to quantify strawberries in a fruit tart if the berry remains whole. But what if they are pulped? There isn't a chemical called strawberry that can be tested. Rather, one or more chemical markers for the fruit will need to be extracted, separated, isolated, detected and quantified.
That same observation ultimately applies to all ingredients, whether we are talking about 'simple', single entities like vitamin C, or more diverse groupings such as fats, antioxidants or highly complex ingredients, such as strawberry. Our new method must be robust enough to extract, separate, isolate, detect and identify the chemicals of interest (and only those chemicals), regardless of the composition of the food product matrix.
This means that it isn't only the chemicals of interest (COI) that need to be considered, since there is potential for chemicals from other ingredients to interfere with any analysis, especially if they are chemically similar. Even where there is not chemical similarity, other chemicals may bind, dissolve or otherwise make it difficult to isolate the COI from the food matrix. Any such interference would mean that our new method over or understates the level of COI in the product, unless our method is capable of avoiding these effects.
It would be easy to become overwhelmed by the complexity of the challenges. How is it possible to develop a method for detecting the COI in any given product, and to ensure that the method is robust and reliable enough to work in all cases? Since there is potentially an infinite variation in product formulations, does this mean that an infinite number of methods have to be developed and validated to be sure that all the bases are covered?
Clearly, it is impossible to test all options, and a more pragmatic approach is required.
Finding a pragmatic approach
One approach might be to generate a model (or a few models) that represent the basic product and to validate the new method against these models. This can significantly reduce the amount of analytical testing required, although it is important to stress there may be regulatory barriers to this approach for some products in some markets. Sometimes there is no alternative to validating the method in the specific product formulation.
This model approach involves creating a high concentration model and a low concentration model. The high concentration model contains only those ingredients that form more than 2% of the actual product formulation. The low concentration model is a mix of all ingredients below 2%, mixed at equal levels (e.g. 1g).
The high concentration model is used in method validation as placebo for specificity, accuracy and method precision while the low concentration model will be used only for specificity purposes and will be analysed with the final method chosen for validation.
In both cases, the models are created after careful consideration of the product’s recipe/formulation and taking into consideration the nature of these ingredients (e.g. sugar, gelatine.) If the model is found to present miscibility and homogeneity problems, it would be modified to overcome those problems.
The same basic approach can be taken if the COI is a marker chemical that is to be identified as indicative of the target ingredient. An example would be using the chemical rutin as a unique marker for hops. However, in order to accurately quantify the amount of hops in the product, the raw material (hops extract), a reference standard (Rutin reference standard) and the finished product (product containing hops extract) must all be analysed simultaneously.
Time to market
It is impossible to describe method development in any detail since so many variables exist depending on the ingredients involved. The technology used to detect and quantify one COI may be different from that are used to detect another. The steps that need to be taken to avoid loss and interference also change according to the ingredients involved. It is easy to be overwhelmed by the complexity of the challenge, especially when dealing with a wide range of products/formulations.
The detail of what is needed only really matters to the scientists who have the job of developing the method. What matters to the product development team is to know that the method is robust and specific, and that there are approaches that can work to reduce time to market and save on the costs. Provided the regulatory boxes are ticked, the high concentration/low concentration models described above provide an opportunity to generate a new method that is robust, reliable, accurate and sensitive, and above all, applicable to every formulation in the company's range.