Further optimisation of a macromolecular ocular irritation test (OptiSafeTM).

Purpose: OptiSafe^TM (OS) is a shelf stable, nonanimal test for ocular irritation. A recent database search found that half of the OS false positive (FP) materials were associated with reactive oxygen chemistries but were not eye irritants in vivo (based on historical rabbit studies by other groups). We hypothesized that naturally occurring tear antioxidants protect the eye from reactive chemistries in vivo and that specific tear chemistries might help explain why some materials are FP for nonanimal tests but are reported as nonirritants in the live animal. To test this hypothesis, a prior study evaluated tear antioxidants and found that the tear antioxidant ascorbic acid, added at human physiological tear levels to the OS test, specifically reduced the measured values for these FPs but did not reduce the true-positive rate. Based on these findings, the OS test method was further optimized. The purpose of the current study was to comprehensively evaluate the performance of the further optimized test method for detection of ocular irritants.

Materials and methods: The OS test measures chemically induced damage to macromolecules and relates these measured values to ocular irritancy. To improve the performance of OS, we made updates to the material to be tested physiochemical handling procedures, prediction model, and test method to include tear-level concentrations of ascorbic acid. We then retested the 78 chemicals from the prior OS-coded validation study in triplicate and compared the accuracy of the 'nonirritant versus irritant' prediction for the further optimized method with the prior results.

Results: We report that for the detection of 'nonirritant' versus 'irritant' (GHS NC versus categories 2B/A and 1) test substances, the further optimized OS test with ascorbic acid compared with the original version has a FP rate that is reduced from 40.0 to 22.2%, the false-negative (FN) rate remains at 0.0%, and the accuracy improved from 80.3% to 89.2%.

Conclusion: A comparison to OECD-adopted tests demonstrates that the further optimized OS test, like the original method, has a higher accuracy and lower FN rate for the detection of 'nonirritants' versus 'irritants' (GHS Category NC versus 2B/A and 1) than the other eye irritation tests (BCOP, EpiOcular^TM Eye Irritation Test, ICE, Ocular Irritection^®, and STE).