Modeling the antioxidant properties of the eye reduces the false-positive rate of a nonanimal eye irritation test (OptiSafe)

Abstract

We recently identified a group of chemicals that are misclassified by most, if not all, in vitro alternative ocular irritation tests, suggesting that nonanimal tests may not fully model the ocular environment in which these chemicals interact. To address this, we evaluated the composition of tears, the first defense against foreign substances, and identified the presence of antioxidants that could detoxify reactive chemicals that otherwise may be falsely identified as irritants in alternative irritation tests. In this study, we evaluated the effects of tear antioxidants on the ocular irritation scoring of commonly overclassified chemicals (false positives) using the OptiSafe™ ocular irritation test. Six tear-related antioxidants were individually added to the OptiSafe formulation, and the effects on test outcome were determined. Ascorbic acid, the most abundant water-soluble antioxidant in tears, specifically reduced the OptiSafe false-positive rate. Titration curves showed that this reduction occurs at in vivo concentrations and is specific to chemicals identified either as producing reactive oxygen species or as crosslinkers. Importantly, the addition of tear antioxidants did not impact the detection of true negatives, true positives, or other false positives unassociated with reactive oxygen species or crosslinking. These results suggest that the addition of tear antioxidants to in vitro alternative test systems may substantially reduce the false-positive rate and improve ocular irritant detection.

Keywords: Antioxidants; Eye irritation; False positive; Tear.

Figure 1.

Screening study of antioxidants at in vivo concentrations (control = no antioxidant, L-tyrosine at 78 μM, uric acid at 43 μM, ascorbic acid at 530 μM, L-cysteine at 14.3 μM, and glutathione at 5.5 μM). Three representative chemicals, (A) cyclohexanol GHS category 1; (B) 2,4-pentanediol GHS NC; and (C) triethylene glycol GHS NC, were used to screen the effects of each antioxidant. Dashed lines show the NC and category 1 cut-offs for the OptiSafe prediction model.

Figure 2.

Control titrations. (A) Positive control (cyclohexanol). (B and C) Negative controls (B) 2,4-pentanediol and (C) dodecane. (D–G) FP controls (D) triphenyl phosphite; (E) ethyl acetate; (F) 2,4-pentanedione; and (G) 2,2-dimethyl-3-pentanol. The dashed lines show the GHS NC and category 1 cut-off values.

Figure 3.

ROS and CL antioxidant titrations. (A–D) ROS chemicals (previously classified as FPs) (A) 2-(2-ethoxyethoxy)ethanol; (B) triethylene glycol; (C) ethylene glycol diethyl ether; and (D) styrene. (E and F) CLs (previously predicted as FPs) (E) 1,9-decadiene and (F) 2-ethoxyethyl methacrylate. The dashed lines show the GHS NC and category 1 cut-off values.