Toxicology of Industrial Compounds

H.-P.GELBKE 367
experimental insufficiencies or to bridge data gaps. Just to give some
examples for such specific considerations:
– reversible versus irreversible effects,
– duration of the study,
– NOAEL versus NOEL,
– LOAEL versus NOAEL,
– local versus systemic effects,
– species-specific effects,
– species differences in anatomy or physiology,
– similar versus different results observed in experiments with various
species,
– biokinetics and metabolism (e.g. metabolic pathways are species specific
or occur only at high doses),
– structure-activity considerations.
This listing certainly not being complete clearly demonstrates that
appropriate AFs cannot be arrived at by a simple cook-book procedure,
but a flexible case-by-case approach is required for each individual
chemical and data set. This is extremely important in order to avoid overconservative
AFs; and in the long run over-conservative risk assessments
are just as prohibitive for an appropriate health protection in an
industrialized world as an underestimation of risk may lead to a more
immediate danger to health. Over-conservative risk evaluations will result
in a wrong allocation of resources, an unjustified prohibition of valuable
chemicals, a wrong or unnecessary selection of alternative materials. etc.
What might be an indication for an over-conservative AF? In principle,
AFs for threshold effects should then be questioned to be over-conservative
if they lead to acceptable human exposures which would also be
appropriate for non-threshold effects (e.g. carcinogenicity) This can be
exemplified by the following consideration:
For a carcinogenicity experiment a ‘LOAEL’ in classical terms would
be equivalent roughly to a dose just leading to a statistically increased
tumour incidence of about 5 per cent. A ‘virtual NOAEL’ in the same
classical sense without a statistically significant increase could then be
at a dose with an actual tumour incidence of 1 per cent, which will
not show up as a substance related effect under usual experimental
conditions. At such a dose level the extra tumour risk would be 1/100.
Applying an AF of 1000 to this ‘virtual NOAEL' would result in an
exposure level with a risk of 1/10 5 , and an AF of 10000 in one with a
risk of 1/10 6 using a simple linear extrapolation without further
default considerations. Exposure levels with a risk of 1/10 5 or 1/10 6 are
under discussion as ‘virtually safe doses’ for the workforce or the