310 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Table II Validation Experiments: Negative Validation For Fragrance Chemicals I Results 2 Expected UVA No UVA Phototoxic Methanol Neg. 3 Neg. Neg. No 6-Methylcoumarin Neg.4 Pos. Neg. Yes Galaxolide Neg. Neg. Neg. No Isoeugenol Neg. Pos. Pos. No D Eugenol Neg. Pos. Pos. No D Methylsalicylate Neg. Pos. Pos. No D Octahydrocoumarin Neg. Pos. Pos. No D Transoctahydrocoumarin Neg. Pos. Pos. No D 6-Methyloctahydrocoumarin Neg. Pos. Pos. No D •Correlation with expected: 8995. 2Maximum concentration tested: 10%. 3From experiments in vivo or analogy. 4Shown to be a contact photoallergen--literature data and personal communication with D. Forbes. •Zone in dark equal in diameter of zone in light (inhibition non-specific). mammalian skin. Although several chemicals were directly inhibitory to yeast, the effects were not light-dependent and therefore were negative for phototoxicity. One chemical, 6-methylcoumarin was phototoxic to the yeast at 0.1% concentration. Although this was a false positive according to available biological data, 6- methylcoumarin has been incriminated as a photoallergen (7). Concurrent with the in vitro validation process, UV absorption data were generated for the pure chemicals. This attempted to answer the question of whether the physical data would preclude the need for tests in yeast. As illustrated in Figure 4, even a small tail of absorption in the UVA spectrum range (1% of total absorbance) was enough for 37O NM •PHANTOLID • -,•x'x'x,. UVB • •UVA . \•-VISIBLE--• INRFER 280 NM 315 NM 400 NM 770 NM OUTPUT RANGE USED IN PHOTOXlCITY SCREEN IN VITRO Figure 4. UV absorbance pattern of phantolid relative to the emission spectrum used in the yeast essay system. Figure is intended to illustrate only minor (1%) overlap.

PHOTOTOXIC ACTIVITY OF FRAGRANCES 311 Phantolid to produce inhibition in the yeast system. Our consultants called Phantolid a "non-absorber in the UVA range." Yet, Phantolid was positive in hairless mice at 10% and at 1% in our yeast assay. Predictions on the reaction of Phantolid based on absorption performance would have led to an error in prediction of the phototoxic potential and therefore suggest that UV absorption alone is not a sufiqcient predictor of phototoxicity. Table III shows post-validation test results in the yeast system compared to results on Table III Validation Experiments: Fragrance Chemicals--No Prior Knowledge Of Phototoxic Status Test Concentration In Percent (Response: +, -) In Vitro Human Mouse Skin Skin Yeast Screen Coumarin Replacement Composite 15 (-) 15 (-) 10 (+) 1 (-) Alpha-Terpenyl Methyl Ether 6.25 (-) 5 (-) 10 (+) 1 (-) Isocyclamone E 12.5 (-) 10 (-) 10 (+) 1 (-) Methyl Carbopolycylic Substituted Dimer 1 (-) 5 (--) 10 (--) 1 (-) Citralva 7 (-) 7 (-) 10 (+) 1 (-) Fig Leaf Absolute NT • 0.1 (+) 0.01 (+) 0.001 (-) 8-Methoxypsoralen NT 0.005 (+) 0.0001 (+) 0.00001 (--) •Not tested. human and mouse skin. These data show the yeast assay has at the very least, a built in safety margin of sensitivity relative to results in vivo. However, the data are inadequate to establish the extent of this margin because only two of the concentrations tested were positive on skin. Fig leaf absolute was positive in mice at 0.1% and in vitro at 0.01%. 8-MOP was positive in mice at 0.005% and in vitro at 0.0001%. The data indicate a sensitivity differential of 10 to 50X in favor of the in vitro system. The testing of additional chemicals (Table IV) also showed that in these cases the screen was more sensitive than the human model. Studies are continuing to quantirate this sensitivity differential. Table V shows the utility of the test in analyzing the effect of structure modification on photoreactivity. 6-Methylcoumarin, being an unsaturated cyclic molecule known to be photoreactive was readily detected in the in vitro system. The saturated analogues methyloctahydrocoumarin, transoctahydrocoumarin and 6-methyloctahydrocoumarin were not reactive at the highest concentration tested, 10%. DISCUSSION In commercial usage, the incidence of significant dermal reactions to fragrance materials is low. This is partly due to the prior identification of active molecules such as 5-methoxypsoralen (the cause of Berloque dermatitis) and partly due to the systematic screening by RIFM of new or suspect materials in the hairless mouse