EXPERIENCES IN SAFETY TESTING OF COSMETICS 167 The basic procedure for conducting human eye instillation tests is as Follows: Depending on the test material, two drops of a 10% solution of the finished Formulation in water* are placed into one of the eyes of a group oF 10 to 25 volunteers, while the other eye remains untreated to serve as a control during subsequent eye examinations. Next, the concentration of the finished preparation is increased until the product is instilled undiluted. During the first day after instil]ation, both eyes of the volunteers are ex- amined at two, four, six, and twelve hours after instillation. Thereafter, eye examinations are made once daily. Modifications of this test in which the finished product is instilled once daily for a period of seven days have also been used, along with other variations on this basic test procedure. Variations of the basic human eye instillation test are usually established by the nature of the test material, probability of accidental entry into the eyes oF the user, and recommended frequency of'use of the potential product. There is, probably, no universal eye irritation test that can be applied to the variety of products that require this particular type of safety test. There- fore, it becomes necessary to "customize" the test procedure for the various groups oF products to be used, e.g., shampoos, rinses, dyes, eye creams, mascaras, etc. In the three examples presented here, the Draize rabbit eye test er- roneously classified a widely distributed shampoo as irritating, failed to predict the inherent danger in the use of a neutralizer, and could not select the milder of two relatively mild products. It seems almost certain, therefore, that this method of safety testing is not foolproof and may produce results which are not reliable. From the limited number of facts presented here, it would not be proper to disregard the results of the Draize rabbit eye test or to eliminate it from the armamentarium of predictive safety-testing procedures. Instead, the results should be used as an index or guide to justify and provide con- fidence in conducting human trials. Pr. ACT•Ca,• USE or TSE 48-Hour. C,•osEr) PaTes TEST As in the case of the Draize rabbit eye test, the prophetic skin patch test on humans has become a routine procedure for the safety evaluation of cosmetics. Many products have been classified as safe or unsafe solely on the results of these tests. Such results can be very misleading and confusing. Many variables must be taken into account in the interpretation of the results from occlusive skin patch tests. Thus, the complete absence of irritation during the closed patch test strongly suggests that the product has a very low irritation potential. However, the occurrence of positive re- * For safeness it is considered advisable when conducting human eye irritation tests to initiate studies for the test product in decreasing aqueous• dilutions.

168 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS actions, depending on severity and number, may indicate only that further testing is in order. The effect of occlusive dressings on the absorption of drugs has recently been the subject of intensive study. Stoughton and his group (11-13) attribute increased drug absorption from occluded (with Saran ©* ) areas of the skin to hydration of the skin. This may explain why a cosmetic preparation is much more penetrating and irritating under a closed patch than when exposed to normal air circulation. Typical examples of products that might yield positive responses under occlusive patches are detergent-containing preparations, which usually elicit "soap" reactions under these test conditions. These reactions can be differentiated from actual primary irritation by the physical character of the reaction itself. Nevertheless, this type of reaction is almost never re- lated to skin irritation resulting from normal use of the product. Another example involves products containing volatile solvents which are capable of eliciting rather severe reactions under closed patch tests, erroneously suggesting that the product could not be used safely on human skin. In fact, many products are currently being marketed which do not cause adverse effect in common usage but elicit skin reactions under the con- ditions of a closed patch test. Nevertheless, the 48-hour closed patch test for the detection of primary irritation can be a valuable tool when interpreted properly and used as a guide or index for further human testing. In skin patch testing, it is im- portant that the investigator be familiar with the shortcomings of the test and utilize this knowledge in the final analysis and judgment of the safety of the product in question. If the results of a closed patch test of an in- frequently used cosmetic show a minimal response in a small percentage of subjects, the product can probably be considered free of primary irritation potential in normal usage. On the other hand, if the results of the routine test suggest primary ir- ritation liability, the product should be tested under actual use conditions, possibly exaggerating the conditions of normal usage, i.e., the amount and frequency. Perhaps, variations and modifications of the routine skin patch test might prove useful at this point or might even be substituted for the routine test. Recently, Kligman (14) demonstrated a method for routinely screening products for their irritation liability by comparing the test substance with a commercial product. Although this technique requires few patients, it is claimed to predict (with reasonable certainty) that a product is as safe as, not as safe as, or safer than the product used as the standard. Of equal importance during predictive safety testing of cosmetics, or of any compound, are the selection of the test and the interpretation of the * Saran is a trade name of Dow Chemical Co., Midland, Mich.