Twenty-seven data sets from 12 cellular cytotoxicity assays, intended
to predict ocular irritation, were submitted to the Interagency Regula
tory Alternatives Group (IRAG) for review. These data consisted of pai
red in vivo (Draize) and in vitro responses to individual chemicals an
d formulations. In vivo data consisted of individual tissue scores so
that the predictive value of the in vitro assay could be assessed for
each tissue response normally measured in the standard Draize assay. D
ata were compiled and evaluated according to the IRAG Guidelines Docum
ent. The Pearson's linear correlation coefficient was used as the firs
t step in assessing the relationship between the in vitro and in vivo
responses. The majority of the data sets represented the study of surf
actant-based materials. In many cases, there was good correlation betw
een the in vitro scores and the in vivo tissue responses. Most pronoun
ced were the particularly good correlations between the in vitro score
s and conjunctival redness scores across most of the assays. Based on
the data submitted, a number of the cell cytotoxicity assays show cons
iderable promise as screens for ocular irritancy. None of the submitte
rs recommended that their cell cytotoxicity assay be used as a sole re
placement for in vivo assessment. For almost all of these assays, the
materials being tested should be water-soluble/miscible. The toxicity
of products with reserve acidity or alkalinity or with high reactivity
may be underestimated. A given user may prefer certain assays dependi
ng on the types of materials to be tested, the expected range of toxic
ities Bad the resources available. The cell cytotoxicity assays can se
rve as a valuable component of a tiered or battery testing program. As
with any assay, a sufficient number of replicate values, concurrent p
ositive and negative controls, and a strict adherence to assay accepta
nce criteria are essential to produce credible data. (C) 1997 Elsevier
Science Ltd.