430 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Table III Quenching Effects of Benzophenones and Sunscreen Agents for the 7-Hour In Vitro Assay Test Vehicle/Activity in % 8-MOP Test Agent Test 10% Benzo- Concen- . phenone-2 tration Methanol* 90% Methanol 10% Benzo- phenone-4 90% Methanol 10% Benzo- 10% Benzo- phenone-4 phenone-4 10% Octyl 10% Octyl Methoxy Dimethyl Cinnamate PABA 80% Methanol 80% Methanol Heliotropine 5% 0.02 Lyral 5 % 0.01 Lime Oil 5% 0.01 Rue Oil 5% 0.01 Angelica Root Oil 5% 0.08 Bergamot Oil (Expressed) 1% 0.4 Fig Leaf Oil 0.1% 1.0 8-MOP 0.01% 100 0.008 b 0 0 0 0.008 b 0 0 • 0 - 0 0 • 0 - 0 • 0 a 0 0.01 4- 0.007 03 0 0.4 4- 0.03 0.1 • 0.2 • 0.5 4- o o • o • 40-80 + 28.0 50 • 50 • a Non-uniform growth around the perimeter of the treated disc. b Results represented by one data point. 4- = Standard deviation expressed as activity in % 8-MOP. * See Table I for standard deviations. Benzophenones supplied by BASF--Wyandotte Corp., Parsippany, NJ. Octyl Methoxy Cinnamate supplied by Givauden Corp., Clifton, NJ. Octyl Dimethyl Paba supplied by VanDyke & Company, Inc., Belleville, NJ. Table IV depicts published data on 10 fragrance raw materials, comparing the 7-hour in vitro assay to human and animal models (2,7,8-10). For the materials identified, agreement is shown between the methods. In general, the yeast assay is the most sensi- tive, followed by the mouse, guinea pig, and human models regarding the concentra- tion required to elicit a positive response. DISCUSSION The 18-hour in vitro assay modified to reduce the exposure time to 7 hours yielded a comparable 8-MOP concentration curve and relative 8-MOP activity for the materials tested. This modification makes the assay more practical by allowing the investigator to initiate and complete the treatment and exposure phases of the assay within a working day, leaving only the incubation and observation of the responses to be performed two days later. The use of the standard 8-MOP curve, when run concurrently with the assay, allows the data to be relative and a quantitative assay will result. Discrepancies between in vitro and in vivo results have been investigated whereby posi- tive in vitro responses could not be duplicated in the guinea pig model which has been used successfully by a number of investigators (11-13) to demonstrate phototoxic po- tential. When a modification of the animal assay was performed, encompassing mul- tiple applications and exposures to UVA at 10 times the in vitro test concentration, agreement between the assays was achieved. The relevance of these findings to human

IN VITRO PHOTOTOXICITY ASSAY 431 Table IV Comparison of Human, Mouse, Guinea Pig and In Vitro Data Guinea 7-Hour Test Agent Human Mouse Pig Exposure In Vitro Activity* Citralva - (7) • - (7) a Isocyclomone - (12.5) a - (10) a Dimyrcetal - (12) a NR Dihydromyrcenal - (1) a NR Bergamot Oil (Expressed) + (20) b + (10) b Lime Oil (Expressed) + (30) b + ( 15)b 8-Methoxy Psoralen + (0.01) b + (0.0003) b Angelica Root Oil NR + (3) a Fig Leaf Oil NR + (0.1) b Rue Oil NR + (3) e NR -(5) NR -(5) NR -(5) NR -(5) +(50) • +(1) 0.4% +(10) c +(5) 0.01% + (0.006) • +(0.0001) 100% +(25) c +(5) 0.08% +(ly +(0.1) 1.0% +(25) c +(5) 0.01% * Phototoxic activity expressed as activity in % 8-MOP refer to Table I for standard deviations. + = positive - = negative phototoxic response. NR = not reported. • Data reported by Weinberg and Springer (7). b Data reported by Forbes et al. (8). c Data reported by Tenenbamn et al. (2). d Data reported by OpDyke (9). e Data reported by OpDyke (10). (#) = concentration tested. experience, particularly for weak agents, is unclear in the absence of comparative studies. Further research is needed to determine the actual phototoxic potential of these materials. In relating positive in vitro findings to clinical experience, Weinberg and Springer (7) reported positive in vitro responses on heliotropine and lyral which could not be repro- duced in humans. In our previous study (2), we also found positive results in vitro on these compounds and could not reproduce these findings in the guinea pig model. Muller and Mitchell (14), using a similar in vitro method (15), found that a rubber chemical (Tetramethylthiuram-monosulfate) elicited positive phototoxic responses in vitro which was not seen in the mouse or human model. The data from these experi- ments indicate that the yeast assay may be eliciting a false positive response however, the population size used in vivo, which demonstrated negative results, may have been too small to accurately reflect the phototoxic potential of weaker agents. With respect to "quenching" a phototoxic response, published studies indicated that 10% topical application of benzophenone-4 prevents 8-Methoxy psoralen (4) and tetra- cycline (5) induced phototoxic reactions in humans. Additional work by Akin et al. (6) resulted in the elimination of a phototoxic response of 8-MOP in guinea pigs and hairless mice after a topical administration of an experimental sunscreen agent (7% N,N-Dimethyl-2-Ethylhexyl-P-Aminobenzoate and 3 % Benzophenone-3). Data gener- ated from the present in vitro assay indicated a potential to reduce potent and eliminate weak phototoxic test agents, demonstrating the extreme sensitivity of the method. The effectiveness of the benzophenone/methanol vehicle in reducing the phototoxic reaction