Setting food science standards

Health claims are a valuable way of educating consumers about the foods they eat. But what kind of scientific research is needed for a sufficiently validated claim? PER BENGTSSON, MD, PhD, explores the pitfalls of trying to test foods like drugs

International research into the health benefits of various food and nutritional ingredients is prolific and often unequivocally positive. But despite this vast body of work, it remains illegal for companies to extol the healthful virtues of their food products in most countries. This is a peculiar and seemingly unfair situation given the abundant mainstream press coverage most negative food stories receive. The public deserves to have access to both the negative and positive health effects of food consumption.

A more balanced situation may arise when the new European Union claims regulations are finalised and implemented. Yet there are many who doubt these rules will have a liberalising effect, especially as many EU regulators believe a pharmaceutical model should frame the regulations.

There is little to be gained in attempting to regulate foods as drugs. Functional foods are consumer products chosen over regular food products because they are perceived as having an added health benefit. Of course, flavour, texture and taste have to be attractive — the health benefit being only one part of the purchase decision.

A drug is taken for no other reason than its effect on health, and given its wide distribution in body fluids, there is typically a delicate balance between serious side effects and the desired therapeutic effect. A drug without efficacy does not fulfil any useful purpose except for its placebo effect. It does, however, present an unnecessary safety risk to the patient and may delay the initiation of effective alternative treatment.

Drugs and data validity
Because scientific experiments always include an inherent risk of error, conclusions and statements based on clinical trial results have the same weakness. Generalisation of data is another common source of error when drawing conclusions from clinical trials because the results do not necessarily translate to all patients. A selected study population and the protocol should ideally reflect the situation for all individuals using the drug (maximal external validity). However, this is rarely the case since factors like dietary habits, age, gender, pregnancy, genetic disposition, illness and concomitant medication may influence the potential medical benefit. The applicability of the drug label for an individual patient is therefore associated with a respectful degree of uncertainty.

Food/drug interactions
One particular problem is the interaction of food components with drugs. In view of the serious consequences such interactions may cause, it may seem surprising that pharmaceuticals are approved following assumptions based on crude investigations of food interactions.

Usually, if drug plasma concentration is not adversely affected following a standard meal, it is thought appropriate to market the drug without specific food interaction recommendations. Such limited requirements offer threadbare protection against a number of possible interactions with specific food components.

Take experiments that showed olive oil (oleic acid) could restrict breast cancer cell growth by reducing the activity of a gene that stimulates cell cancer growth. If this holds true in a clinical setting, trials conducted in countries where olive oil consumption is high may not produce results applicable for other countries and vice versa.

It is understood that most pharmaceuticals are intended to be taken separately from meals because of the potential for food interactions. If there are specific suspicions that food interactions could be of major significance, drug manufacturers are required to document it explicitly. The problem is, such documentation is rarely made because most interactions are probably unknown.

Documenting functional foods is more complex. Functional food ingredients inevitably mix with innumerable other food components. Since dietary habits vary, it is more difficult for clinical trials to validate functional food products than is often the case for drugs.

This is probably the greatest weakness of functional food data and it is important to realise that this source of error is inherent to the concept of functional food. Its impact could be reduced if consumers behaved in a coherent way and ate the same diet. An alternative but unrealistic approach would be to implement rules stipulating that all types of diets need to be investigated and that the results be reflected in the labelling of the functional food product — the problem being that the cost would approach infinity and there would be too much information to go on a label in order to educate the customer.

The problem of duplicity
In addition to the greater problem of external validity, one other important difference between drugs and foods concerns the multiplicity of similar products in a product range. Our palates are remarkably different, as is our natural drive to seek variation in foods, and therefore the active ingredients in a functional food product need to be presented in a range of products in order to satisfy all customers.

This fact has been a concern for European officials drafting health claims legislation when strict ?product-specific health claims? were made a governing principle in the proposed draft for labelling functional foods. It?s a problem that can be significantly reduced by inference from human and animal experiments. A more radical ?zero-tolerance? approach would require variants of the same product be documented in separate efficacy trials. Surprisingly, this idea has been given serious consideration when outlining appropriate regulations for food products, despite the obvious difficulties and costs of such trials.

It is particularly difficult to defend repeated trials that must be conducted for health claims of only symptomatic relevance rather than disease. If the effect has been demonstrated with one such product in several clinical trials, it should be enough to present surrogate endpoint data in animals showing the plausible mechanism would work without a vehicle or with another vehicle. There is no critical medical issue that needs to be considered, and, in contrast to most pharmaceuticals, the customer has the chance to refrain from buying a product if she perceives better variants.

Seeking compromise
A pragmatic approach that gives credible guidance to the customer but also contains enough leeway for a necessary range of products to emerge on the market is needed. Replication of efficacy trials within a product family cannot be the only way of reaching sufficiently robust ?internal? food interaction conclusions. Is it not possible to rely on other lines of evidence and precautions that would minimise risk in approval decision making?

For example, a chain of evidence relevant for probiotic bacteria could be as follows: First line of evidence should always come from the efficacy proven by one finished product type. Second, studies performed in animals can be used to demonstrate that the active principle (bacteria) affects the plausible mechanism in an expected way without other product ingredients. Having made such inference in animals, human colonisation studies could prove (or disprove) different food matrices can safely deliver live bacteria to the gut.

The implementation of rules stipulating repeated clinical trials for multiple product presentations greatly simplifies the criteria for approval of functional food products, but is financially unrealistic. Alternative approaches are possible and would not significantly change claim validity for the individual customer. Foods are not pharmaceuticals, and that is why it is both possible and necessary to find compromises that satisfy basic scientific principles while giving consumers reasonable guidance.

Per Bengtsson, MD, PhD, is chief executive officer of Probi AB, Lund, Sweden.
Respond: [email protected]

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.