Late stage failure of candidate compounds in drug development programmes are very expensive to the pharmaceutical industry. Whilst commercial considerations and liver toxicity account for the halting of many programmes, most large pharmaceutical companies will suffer occasional stops or more likely delays in phase 1 or phase 2 trials as a consequence of genotoxicity data. For a potential blockbuster drug these delays equate to losses of at least $1 million per day. It is reasonable to ask why these events occur and how they can be minimized?
Perhaps the main reason is that carcinogenicity and genotoxicity are not uncommon properties of chemicals. The Carcinogenic Potency Database which has data from long term, rodent cancer tests on 1485 chemicals, has positive data for over 60% of all chemicals. This does not mean that over 60% of chemicals are dangerous human carcinogens, because many early studies were carried using doses many orders of magnitude higher than a plausible human exposure. However, this underlines the high prevalence of possible carcinogens. Long term rodent carcinogenicity and germ line mutation studies, taking 2-3 years and costing $2-3 million dollars clearly can not provide adequate and timely risk assessment.
Given the very high rate of attrition for compounds in drug discovery (from tens of thousands to just a couple of prime candidates), a screening programme should be able to reduce the number of genuine genotoxins reaching the later stages of development. However, the current regulatory protocols were not conceived for screening. In practical terms they are too time-consuming and compound hungry. In performance terms they mostly lack specificity. There is now an alternative in GreenScreen.
The Ames test has the highest specificity of the regulatory battery, and for this reason a positive result is a near-certain compound killer. Its relative low cost (~$1k) means that it is often the first line in genotoxicity assessment. Whilst the GLP version is not practical for screening, there are ‘cut down’ as well as microplate fluctuation test versions allowing medium throughput assessment. Unfortunately the Ames test uses bacteria which lack many of the genotoxin targets found in mammalian cells, and it is the subsequent mammalian tests, usually reserved only for Ames negative compounds, which present the huge specificity problem. Furthermore, these are not yet suitable for screening. Only about 1/3 of compounds that are only positive in the mammalian cell tests give positive results in subsequent animal tests and cancer studies, so perhaps inevitably all positives are treated as potentially good drugs. This means they will go forward to animal testing, despite the requirement that any compound carrying a positive in vitro result must have 2 animal studies instead of one. Animal studies cost $20-30K per compound.