An informative letter, just published by a team from Novartis, highlights the risk of suitably substituted benzimidazoles acting as latent precursors to reactive ortho-iminoquinone species. The example they report (see structure on left below) was Ames positive and exhibited time-dependent inhibition of CYP3A4.
In this example they found that replacing, or removing one or both of, the metabolically labile ortho-methyl substituents led to Ames negative compounds (disubstitution being postulated to conformationally predispose one of the methyls to metabolism) and similarly, removing one of the imidazole nitrogens (to give the indole shown above), also prevented formation of the putative reactive ortho-iminoquinone species (a species with close analogy to ortho-quinones, which are known to carry a reactive metabolite risk) and hence also resulted in an Ames negative analogue.
The above hypothesis held up in analogues incorporating the minimal toxicophore and the risk could be mitigated in a similar manner. Whilst this is a particular case worth making a note of, there is clearly also a concern that other suitably substituted benzimidazoles could carry a similar risk.