Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Revised PAG guideline

PAG/UNU revised guideline no. 7: Human testing of novel foods


Novel foods are defined as those that have not been eaten before by the population for which they are intended, or at least not in significant amounts. Similarly, novel processes applied to traditional foods require examination, as do new varieties. After preclinical testing for possible toxic constituents, the features that require examination are nutritional value and components that may cause dietary intolerance.

The tests outlined in PAG/UNU Revised Guideline No. 6, "Preclinical Testing of Novel Sources of Food" (1), should provide adequate information on the safety of a product before commencing any testing with human beings. Human testing, as outlined in the remainder of this Guideline, should only be carried out after a complete examination of the features defined in PAG/UNU Guideline 6. Although the toxicological tests outlined should be performed for very novel foods or novel processes applied to traditional foods, there is a real danger of excessive and unnecessary experimentation with minor variations in formulas using previously tested ingredients or processes.

If animal tests reveal any toxicological problems, these should be resolved before the material is fed to humans. Examples are substances such as lectins (haemagglutinins), trypsin inhibitors, and cyanogenic glycosides.

Other animal tests will reveal the available energy content of the food, digestibility and quality of the protein, and availability of minerals, thereby providing information on the nutritional value of the product that will complement the information from analysis of carbohydrates, lipids, and fatty acid composition, proteins, and amino acid composition as well as vitamins and minerals.

Nutritional measurements on human beings are subject to such individual variation they they rarely contribute as much useful information as the measure meets of nutritional value in experimental animals. They may, however, be desirable in certain instances, such as when comparing the relative value of two sources, or substituting a novel source for a traditional one. In such cases it is necessary to establish precisely what information is required and then to select from the tests described below those most likely to supply the answer to the appropriate questions.

In general, it follows from first principles that any safe food that supplies significant dietary energy, protein, and other nutrients will be of value to the consumer. However, the acceptability of a food, or tolerance to that food once ingested, cannot be foretold from animal experiments alone, and hence human testing is a prerequisite.


Products that require full testing, both preclinically and clinically, are:

1. Food sources not previously consumed by human beings, or at least not in the amounts proposed.

2. Products previously accepted that have been subjected to such different processing conditions that questions can be raised regarding their nutritional or toxicological properties.

3. In certain instances, new varieties of foods and foods newly introduced to a population, even though they have already been consumed safely in other areas.

It is assumed that the following information will be available in full detail before proceeding with human testing:

1. Availability of supplies of the novel food, including aspects of its distribution and shelf-life.

2. The mode of consumption, i.e., whether it is to be incorporated into other foods or dishes, used a supplement, fed alone (as it may well be in the case of infant foods), or used as a substitute for traditional foods, and whether it is to be cooked after distribution

3. Full microbiological data pertaining to public health hazards.

Particular care should be taken to ensure that the material used in the tests is truly representative of the novel food that will ultimately be marketed. Whenever possible, the same large-scale industrial process that will be used to produce the novel food for marketing should be used to produce the food used in testing, since the specification of food produced in a pilot plant may differ significantly from that produced commercially. Where there is an inherent risk of variability of the product-e.g., in yeast products- stringent batch control after general production will be necessary to ensure that the food as marketed does not differ from that tested. Single-cell protein preparations may need to be examined by suitable screening tests that should be applied after each process modification. It is also important to consider the form in which the food will be tested. Food that would normally be cooked prior to consumption should be treated in a similar manner before feeding or testing in vitro.


The approximate nutritional value may be determined by chemical analysis. If the food is to be consumed processed or cooked, then the nutritional tests should be carried out on such cooked foods. For testing nutritional value in vivo, the food should be fed under the conditions it is expected to be consumed in practice, i.e., mixed with the rest of the diet in amounts likely to be used and not at higher levels. Time and effort should not be devoted to measurements on the isolated food-stuff at excessive levels when it is not intended to be consumed alone, and to measuring parameters such as conventional net protein utilization that have little relevance.

Methodologies for determining protein value through analytical procedures and experimental animal and human trials are given in the UNU Food and Nutrition Bulletin Supplement No. 4 on Nutritional Evaluation of Protein foods (2). Approximate protein value can be established by determination of essential amino acid composition and calculation of a protein score. If human feeding studies for this purpose are desired, procedures are available, although they are time-consuming, costly, and add little to the information on protein value obtained from laboratory analysis and feeding experimental animals. It is, however, desirable to conduct a five- to ten-day nitrogen balance trial to measure the product's digestibility by human subjects.

If the food is intended for infant feeding, selected anthropometric parameters can be measured over a period of at least three months. These can be limited to weight and height increment, head and arm circumference, and triceps skinfold. Other possible parameters include serum and urinary urea, prealbumin, albumin, transferrin, ceruloplasmin, cholinesterase activity, aspartate amino-transferase (formely SOOT), hydroxyproline, creatinine-length index; and for vitamins, it may be desirable to measure the usual enzymereactivation tests and other common nutritional indices.


It must be borne in mind that a significant per centage of individuals is intolerant to one or more traditional foods. It has been stated that there is no food that is wholly free of adverse effects in at least an occasional individual. Most individuals have ascertained by trial and error which foods cause them ill effects when these effects follow fairly rapidly after the ingestion of the offending food. Recent reports from the United Kingdom suggest that about 30 per cent of the population sampled in some countries respond adversely to one or more foods, and up to 3/4 per cent avoid one of the major traditional foods widely consumed by the population as a whole (3). The frequency of allergic responses to milk protein alone is variously reported as up to 8 per cent (4), and allergies to chocolate, corn, legumes, eggs, citrus fruit, tomatoes, wheat, and pork are also very common (5). Allergic responses call for special consideration and can be specifically examined and characterized by the presence of antibodies. Gastrointestinal intolerance to ingested food will be indicated by loss of appetite, flatulence, undigested stool contents, diarrhoea, and abdominal pain. Cutaneous rashes are also quite common (6).


When novel foods are developed, it is essential that their acceptability be evaluated in human subjects under controlled conditions before they are used in an unsupervised manner. A variety of adverse reactions may be observed in humans that are undetectable and unpredictable in experimental and farm animals even with extensive testing.

The most common symptoms to be looked for in human subjects are those of gastrointestinal intolerance. In addition, various allergic reactions may occur in some individuals after consumption of almost any common food, especially if it is a protein source. It is the frequency of such reactions that must be evaluated rather than their expected absence. Moreover, symptoms may be due to psychosomatic factors.

Because of these factors, it is important to feed a control group simultaneously. Ideally, the trial is conducted so that individuals are randomly assigned to experimental and control groups, stratified by sex if necessary, in a doubleblind cross-over design. Only after a low frequency of adverse symptoms is assured can additional trials be conducted without the necessity of a control group. Once again, the inadvisability of high-level testing must be emphasized if the product is unlikely to be consumed at such levels; mistaken conclusions may be drawn from reactions at high-intake levels even with traditional foods.


In general, 25 to 50 human subjects can be studied initially. All subjects should be in good health, as determined by medical history, physical examination, and routine blood and urine tests. Individual medical histories should focus on a history of personal or family intolerance to specific foods and manifestations of allergy, such as asthma, hay fever, and urticaria. However, no individual should be excluded on this account, since it is essential to determine the prevalence of symptoms to the novel food in the general population for whom it is intended, The purpose of the study, the cooperation required, and potential risks involved should be explained fully to each subject. Protocols should be reviewed by the multidisciplinary group established by the institutions to evaluate such studies using humans as experimental subjects. In addition, subjects should sign appropriate consent forms. For subjects who are legally minors, written parental or guardian consent must be obtained.

The actual numbers of individuals involved in the study will be a function of the preliminary data. True allergic responses will be characterized by antibody formation and will potentially require fewer participating individuals than in those cases where intolerance is a problem. In the latter case, this should be tested in a manner similar to that used for bacteriological sterility; that is to say, if any subject responds adversely, then the tests should be repeated on larger numbers of subjects to ascertain numbers that may be intolerant. So, while 20 to 50 subjects may be an adequate number for general purposes if intolerance is revealed, this number may need to be increased.


Subjects should consume both the novel and a control food, preferably one that is already commercially available, in a double-blind cross-over study design. Subjects should be allowed complete freedom to consume their usual diet. Since this will vary and may include meals or foods that in themselves cause symptoms, it is essential that each subject keep a diary of all significant departures from the usual in his or her daily diet and activities. These often prove of value in the retrospective interpretation of the results of a tolerance trial.

An appropriate study period would be two four-week periods separated by a one-week interval. In the first fourweek period, subjects ingest daily either a fixed amount of the control or experimental material. After completing the initial four-week phase, subjects will return to their regular diet without supplementation. Thereafter, the second fourweek period will be started, and subjects, unaware of the type of supplement they are taking, will begin ingesting the second protein. Administration of the material for five to six days each week for four weeks is adequate. Experience indicates that individuals not developing symptoms in the first 20 days are not likely to do so thereafter.


Any method of oral administration that facilitates a double-blind trial is acceptable. For example, the material can be incorporated into cookies, cakes, pudding, or other acceptable supplements to a usual diet if it can be introduced so that its presence cannot be specifically identified. Novel foods in powder form can often be tested by allowing the subjects to mix them into bouillon or any of a variety of fruit juices according to individual preference, and they can vary their choice from day to day. In this case, however, there must be available a similar control material for which tolerance is already well established. Experienced nutritionists can usually find an acceptable method of administration.

All subjects recruited should have the opportunity to taste the materials to be consumed before making a commitment to the trial. If this is not done, a number of subjects are likely to confound the trial by dropping out in the first few days because they do not like the taste, texture, smell, or some other aspect of the substances offered.


Tolerance studies should be designed taking into account the intended use of the substance. For example, yeast as a vitamin supplement may be consumed at less than 5 grams daily, as a functional food additive rarely more than 10 grams, but as a significant protein supplement, the amount might be 30 grams or more, providing it does not supply more than 2 grams of nucleic acid. If the amount intended for testing is more than 15 to 20 grams per day, it should be divided into two or more feedings per day. If a supplement is intended to be added to a traditional staple like bread or a rice dish, then it should be given daily or even more frequently, in relatively large amounts, following normal dietary practice so that the experimental material is fed at the upper intended limit of use, which may involve daily administration of 30 grams or higher.

For a totally new food source it may be more prudent to conduct a pilot trial with a smaller number of subjects at a lower level. Materials intended only as minor additives to a mixed diet may be tested at levels as low as 5 to 15 grams daily. When evaluating the data, two contrary effects may be noted: intolerance to which the subject soon becomes adapted, and boredom from eating the same food over a prolonged period. The former (as in flatulence) can be examined by introducing the food in small and increasing amounts over a few days. The latter will always be true even for attractive traditional foods if eaten too frequently.


The level of dietary energy intake should be sufficient to maintain constant weight in adults or adequate weight gain in children.


Before subjects begin the feeding trial, initial blood tests are required. These should include haemoglobin or haematocrit, white blood count, and a peripheral blood smear to determine the percentage of the white blood cell components. In addition, serum electrolytes, blood urea nitrogen, creatinine, and serum uric acid should be obtained. Tests of liver function should be included; i.e., albumin globulin, bilirubin, aspartate aminotransferase (SGOT), alanine aminotransferase (SGPT). Serum calcium and phosphorus should also be determined. A urine sample should be obtained for complete analysis of cells, protein, and pH.

These blood and urine tests should be repeated at the end of the study or sooner in subjects dropped from studies because of adverse reactions. It is important to ascertain if any of these biochemical parameters change during the acceptability study. Examination of the serum immunoglobulin levels is necessary in individuals with possible allergic reactions.

Further, to assure that the novel food is not associated with unpleasant or adverse reactions, all participants in the study should be asked to keep a diary as noted above that will note changes in mood, appetite, sleep patterns, libido, and other subjective reactions.


Much time and effort has been devoted in the past to measuring such parameters as protein quality (by PER or NPU) in human beings. This is not necessary for two reasons:

1. Results obtained in closely controlled laboratory tests are often not repeatable under field conditions. This would indicate that such parameters are unimportant under normal, free-living (field) conditions.

2. Most such measurements have been carried out on the food alone or on protein isolates, whereas it will be consumed mixed together with other foods. Consequently, the protein quality of the food alone is not of any interest.

For example, a protein source completely deficient in one essential amino acid might well be rejected as the result of such tests, whereas if the missing amino acid is not limiting in the mixed diet, then that protein may make a valuable contribution to the diet.


It follows from first principles that any safe food that supplies additional energy, protein, and other nutrients will be of value even if it is not possible to demonstrate such benefit in the trials. While it may be necessary in order to convince policy makers and governments to demonstrate the benefits of the novel food (in which case growth, nitrogen balance, skinfold increases, etc. may need to be measured and reported), it is not necessary for the nutritionist. While he would often like to quantify the benefits of the supplement, such measurements are open to enormous individual variation and rarely have any real meaning. The only instances where such measurements may be of value are when different sources of nutrients are being compared, and even then the variability of the findings usually vitiates the results. Economic and palatability considerations frequently outweigh all such technical detail of nutrient differences.

For example, while weight gain is the best measure of the value of a supplementary food, such measures require prolonged periods of testing (several months) and show the most marked results in children who were previously poorly nourished. Such methods cannot be appiled to adults. Any supplement that provides additional energy and nutrients will be of value when the diet has been poor, and testing the precise effects of such supplements is often quite superfluous and costly.

Finally, once the use of a product has commenced, it is advisable to have a mechanism by which any adverse reactions may be monitored in order that corrective change may be introduced.


1. PAG/UNU Guideline No. 6, "Preclinical Testing of Novel Sources of Food " The United Nations University Food and Nutrition Bulletin 5 (1): 60 (1983).

2. P. L. Pellet end V, R. Young (Eds.), Nutritional Evaluation of Protein Foods, The United Nations University Food and Nutrition Bulletin Supplement No. 4 (The United Nations University, Tokyo, 1980).

3. A. E. Bender end D. R. Mathews, "Adverse Reactions to Foods," Brit. 1 Nutr. 46 (3): 403 (1981).

4. S. I. Bahna and D. C. Heiner, Allergies to Milk (Grune and Stratton, New York and London, 1980).

5. F. Speer Food Allergy (PSG Publishing Co., Inc., Littleton, Massachusetts, 1978).

6. S. Achinewu, "Food Allergy end Its Clinical Symptoms in Nigeria," The United Nations University Food and Nutrition Bulletin, Vol. 5, (3) (to be published, 1983).

Contents - Previous - Next