PROBLEMS OF PREDICTIVE TESTING 353 negative results should be tempered by this knowledge. He must, there- fore, not only select his test but also his test population. Not every product should have to be subjected to all known tests. For example, there is no reason to subject a vaginal deodorant, nor any other product not normally exposed to light, to a phototoxicity test. Similarly, it is extremely doubtful whether or not a closed patch test of a shampoo at full strength has any meaning since shampoos normally remain on the scalp only for two to three minutes, are diluted, and are almost completely removed by subsequent rinsing. Finally, a 24- or 48-hour closed patch test of a depilatory would be out of the question and of no value. Such use-related considerations alone form a solid rationale for the desirability of in-use testing. These procedures are also valuable for the study of topically applied products because protocol automatically in- cludes exposure to the natural elements, such as sun, wind, heat, and cold. The importance of this test will become more apparent when some clini- cal experiences are discussed later. For the moment, it will suffice to in- dicate that in-use tests employ the product at near normal frequency in and over the area of usual application and can also take into account the possible misuse or even abuse of a product. PROBLEMS OF INTERPRETATION Let us assume for a moment that the investigator has established in a series of animal safety tests that he can proceed with human testing of a new cosmetic product. He has then carefully selected the proper human tests and has taken into account specific modifications to suit the product and its intended use. The results of these tests are available, and the in- vestigator is now faced with the responsibility of establishing whether the product can be "safely distributed." It is at this point that our inad- vertent errors and our lack of complete scientific knowledge can impair clinical judgment. An important source of error is the grading of reactions which may depend entirely upon the patch test material (14) and on the observer who grades the reaction. Occasionally, reactions may be due to improper application of the patch, i.e., friction. Inadequate tissue contact of the material and improper placement (which may lead to loosening of the occlusive covering) can reduce the severity of reaction to the patch test. Climatic conditions are also believed to have an influence on the reac- tivity of human skin, and there is some evidence that there is greater re- activity during winter months (19, 20).

354 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Cross sensitization and what has been termed skin fatigue or cumula- tive reactions add to the problems already facing the investigator. He must now, for example, consider the possible influence of drugs on the reactivity of the test subject or the ultimate user. The age of the patient influences reactivity due to lack of previous exposure or by virtue of the development of the immunological apparatus. Pre-existing skin condi- tions, such as eczema, may or may not have a direct effect on the final re- sults of the patch test or on the consumer. A further problem is gener- ated by the fact that often patch tests are conducted using more than one and sometimes as many as a dozen products at the same time. The ques- tion then arises whether simultaneous patch testing with different mate- rials could have an influence on the reaction to one or more of the prod- ucts tested. This fact could be an advantage or a disadvantage. Our in- ability to be more specific is, in fact, a limitation in itself. In this con- nection it is noted that racial influences and skin color must also be con- sidered, although to date there is relatively little known about the sensi- tivity of different races to any given chemical (21). It is not surprising that heredity plays an important role in allergic contact dermatitis (22) and this may also apply to reactivity to cosmetics. Finally, the investigator must take into account that the sensitivity potential of a population to a product can change with time. This type of latent sensitivity was studied by Baer et al. (23), who attribute this phenomenon to increased opportunity to exposure to allergic sensitizers. It is conceivable that the safety of chemicals or products established some years ago is no longer applicable today. The relationship of these ob- servations to Agrup's (24) conclusion that patch testing can lead to sen- sitization is by no means clearly established. It must be concluded that the interpretation of predictive human patch testing is made difficult by two factors: (a) possible experimental error and (b) the influence of unknown extraneous factors on the test sub- jects and eventually on the population of potential users. The careful experimenter can usually overcome the limitations due to inadvertent error. He cannot make up for the holes in our knowledge, i.e., the un- known. Some known and some unknown parameters, both of which may in- fluence the results of patch testing or the response during use, are shown in Tables I and II. Many of the questions raised above exert an important influence on the results of a well-conducted in-use test. The results from such a test