Health Risks of Ionizing Radiation: - Clark University

we consider this group of a few hundred children to
be the cohort at risk then a spontaneous case of leukemia
would be unlikely and an excess of the degree
that is seen in other studies would be impossible to
detect.
In conclusion, this analysis supports the idea
that paternal preconception exposure to radiation is
a risk factor for childhood leukemia. Under the traditional
statistical ‘null hypothesis’ that there is no
real relationship, there would be, at most, about a
7% chance of generating data as suggestive of a real
relationship as these in the absence of a true relationship
(based on the second pooled result in Table
C-4). This is, we believe, an informative measure of
the overall state of the science and one that is intuitive
enough to be useful in dealing with the public.
Even without such results it could be argued that,
in the context of a body of literature that has shown
the same effect in mice and has also shown evidence
of increases in solid cancer and adverse birth outcomes,
the standard test of statistical significance
should not be the sole criterion by which we judge
this risk.
It is our interpretation of the discussions accompanying
many of these studies that the scientific
community has been unwilling to consider this as a
serious issue, even in light of convincing evidence,
primarily because a) it seems mechanistically unlikely
that a sperm cell mutation could be a leukemia
risk factor in the child (which would carry such
a mutation in the paternal allele of every cell), and
b) the atomic bomb survivors have not demonstrated
an effect.
One potential solution to the first problem involves
recent animal studies in the area of chromosomal
instability--these indicate that post-concep-
193 Appendix C
tion mutations can occur following preconception
exposures (Niwa 2003), and it is therefore not necessary
to have a germ cell mutation. In vivo and in
vitro studies (not transgenerational studies) have detected
elevated levels of instability for as long as two
years after exposure in exposed mice. If this kind of
persistence is assumed to hold for transgenerational
instability then leukemogenic mutations could theoretically
occur during embryogenesis, or even later,
following preconception exposures. Evidence for
such an effect has in fact been generated in mice.
It is also puzzling that many discussions have
implicitly treated two potential leukemia factors,
preconception irradiation and viruses transmitted
during high population mixing, as independent and
mutually exclusive. Again, animal studies provide
useful information. In an early study, dramatic reductions
in radiation-induced leukemias were observed
in mice that were housed in a microbe-free environment
(Warburg 1968). More recently, preconception
irradiation of male mice has been shown to increase
the sensitivity of offspring to chemical- or radiationinduced
leukemias (Lord et al. 1998). Two risk factors
such as radiation and a virus could be linked in
a two-mutation model of leukemogenesis, through
a model of induced instability and oxidative stress
(Clutton et al. 1996, Limoli et al. 2003), or through
some unknown mechanism; in any case, interpretations
of the Seascale leukemia cluster, where both
factors have been implicated, could consider them
jointly. Finally, the atomic bomb survivor data have
very simply not demonstrated the absence of an effect
and they are in fact not very informative data.
This has not been adequately appreciated in most
discussions.