Evaluating the 'safety' of drugs produced by biotechnology resembles t
he assessment of conventional 'new chemical entities,' but with certai
n major differences. The 'quality' of the product requires careful con
trol because of concern about the carry-over of DNA, immunogenic prote
ins, endotoxin and process chemicals. Equally the potency and purity o
f the product must also be considered, as well as its identity. The to
xicity testing of rDNA-derived proteins, monoclonal antibodies and vac
cines, although increasingly being swept under the umbrella of convent
ional studies, should be empirically devised according to the nature a
nd physiological effects of the substance, taking account of the respo
nsiveness of suitable species for non-clinical testing, the potential
immunogenicity of heterologous proteins and any effect the drug may ha
ve on physiological mechanisms and the immune status of the test anima
ls. Conventional types of single and multidose and reproduction toxici
ty experiments can then be adapted to detect and investigate any hazar
d of the novel drug. Kinetics, metabolism and drug interactions should
be explored and the regulatory demand for genotoxicity data satisfied
. If appropriate, immunological actions, including auto-immunity, can
be sought. 'Safety-in-use' should then be predictable with some confid
ence, because of the extent of the toxicological investigations and be
cause activities that cannot be examined will have been delineated, e.
g. lack of a responsive species or of a suitable laboratory procedure.