adequate uncertainty factor for exposure limit

Alan Stern ahstern at EOHSI.RUTGERS.EDU
Fri Oct 6 15:16:25 EST 1995

In response to Lawrence Segal's query regarding the advisability of a 25 fold
uncertainty adjustment to derive a TWA occupational exposure limit for a
putative carcinogen:  In regulatory standard setting and guidance for
environmental exposures (at least in the U.S.) the LOEL/NOEL approach, with
its assumption of a threshold, is not employed for carcinogens.  Rather,
exposure standards and guidlines for carcinogens are derived on the basis of
cancer slope factors (also referred to as cancer potency factors) generally
employing the linear multi-stage model of cancer dose-response.  This
approach is based on the assumption that there is no demonstrable threshold
for carcinogenesis (either genotoxic, or epigenetic).  Nonetheless, if we
were to start with the assumption of a threshold and use the NOEL/LOEL
approach, the the standard use of uncertainty factors (UFs) for dieriving a
Reference Dose (RfD) for chronic exposure is as follows:

UNCERTATINTY                                  UF

extrapolation from average                    10
human population to 
sensitive human population

extrapolation from animal                     10
dose-response model to human

extrapolation from short term                <=10
exposure study to long term human

estimation of NOEL dose from                 <=10
LOEL dose

incomplete database                          <=10
and other considerations of
scientific integrity

The formula for calculating the RfD from a LOEL (such as is presented form
the current study) is:

LOEL/(UF1 x UF2 x ....UFi)

with a maximum value of 10,000 in the denominator.

For a further explanation of this approach see Barnes and Dourson.  Reg. Tox.
and Pharm. 8:471-486 (1988).  These UFs are somewhat arbitrary, although they
do have some (ex post facto) support in empirical toxicology (see Dourson and
Stara.  Reg. Tox. and Pharm. 3:224-238 (1983)).

The information Lawrence Segal provides is incomplete (e.g. was the study a
lifetime exposure, what was the size of the exposure groups, how many
exposure groups, could the animals be considered to be representative of a
senstive population)  Each of these factors could potentially influence the
application of UFs in this approach.  Nevertheless, using this approach, it
is clear that we would at least employ a safety factor of 10 for the
esitmation of a NOEL from a LOEL and an additional factor of 10 for
extrapolation from animal data to a human population.  That would give a
minumum UF product of 100.

Alan Stern, Dr.P.H., D.A.B.T.
New Jersy Dept. Environmental Protection
ahstern at

More information about the Toxicol mailing list