Comparative study on prediction performance of photosafety testing tools on photoallergens.

Abstract Several testing methods have been established to identify potential phototoxins. The present study was undertaken to clarify the predictive ability of in vitro photosafety assays for photoallergenicity. On the basis of animal and/or clinical photosafety information, 23 photoallergens and 7 non-phototoxic/non-photoallergenic chemicals were selected and subjected to UV/VIS spectral analysis, reactive oxygen species (ROS)/micellar ROS (mROS) assays, and 3T3 neutral red uptake phototoxicity testing (3T3 NRU PT). Of the photoallergens tested, ca. 96% of chemicals had intense UV/VIS absorption with a molar extinction coefficient of over 1000 M − 1  cm − 1 , and false-positive predictions were made for 3 non-photoallergenic chemicals. In the ROS assay, all photoallergens were found to be potent ROS generators under exposure to simulated sunlight. In the photosafety prediction based on the ROS assay, the individual specificity was 85.7%, and the positive predictivity and negative predictivity were found to be 95.8% and 100%, respectively. Most of the photoirritant chemicals were correctly identified by the 3T3 NRU PT; however, it provided false predictions for ca. 48% of photoallergens. The orders of sensitivity and specificity for photoallergenicity prediction were estimated to be: [sensitivity] ROS assay > UV/VIS absorption ≫ 3T3 NRU PT, and [specificity] 3T3 NRU PT > ROS assay ≫ UV/VIS absorption. Thus, photochemical assays, in particular the ROS assay, can be used for assessment of photoallergenicity, although there were some false-positive predictions.