Re: mice v. elephants

[David Scherer <scherer@uiuc.edu>]

I congratulate the biologists who are critically examining the idea of LNT,
but I don't think it will be easily dismissed on simple arguments.  Nor
will it be proven from those arguments.  Not being a biologist, I offer a
few simplistic observations of my own:

>The basis for linear-no threshold theory is that a single gamma
>ray striking a single DNA molecule in the nucleus of a single cell can
>cause a cancer. Since an elephant or other large animal has many
>thousands of times more such DNA molecules than a mouse, what
>explanations have been offered for why elephants don't get a lot more
>cancers than mice from radiation exposure? Why aren't human adults more
>sensitive than small children? Why is radiation dose defined as energy
>deposited per unit organ mass, rather than just total energy deposited
>which determines the number of such hits?

1.  What is described is the basis of a linear model of CELLULAR response.
As I understand the situation, radiation protection standards are based on
a linear response of the total population, not even the response of a
single individual.  I suppose the idea is that  over a large population,
the response works out to be approximately linear.  Some individuals might
be more sensitive, some might be less, but taken on the whole, the
population response is approximately linear.  (Of course, the population
response is illuminated by important features in the individual response.
For example, if there is threshold for an individual, it would be present
in the population response as well.)

2.  I think the specific energy (per cell or per mass, i.e. dose) is
important for assessing acute injury risks.  For example, if 50 percent of
the bone marrow or 50 percent of the intestinal crypt cells are ablated, I
would expect the response might be similar, even if one person is small an
another large.  It would seem that killing 1000 cells would result in
different effects if one person has 1000 and another has 10,000.
(Historically, I believe the R and rad were derived in response to such
nonstochastic effects, such as erythema in radiologists.  Dose may not be
as valuable for assessing latent affects, where the cell is transformed but
not killed.)

3.  As far as the question about sensitivity of adults vs. children, this
issue is clouded because there is a long latent period between exposure and
outcome.  (I assume the outcome is cancer, etc., since the acute effects
are known to be nonlinear.)  By the time 10-20 years have passed, a number
of other exposures might have occurred, further clouding the issue.  To top
this off, the radiosensitivity of a person may change some cells have
different proliferation rates at different ages.  Add in differences  in
family history, gender, race, etc., etc., it is probably hopeless to try to
build a PRECISE model for the population from considerations for the
individual.  Again, prominent features of the individual response are
important in the population response.

>"Yet the simplest considerations strongly suggest that repair, at some
>level, at least of low LET radiation, must exist. With some 2-4E14
>ionizations from background radiation (a large dose from the standpoint of
>the strictly no-threshold linear hypothesis) occurring within the DNA of
>the ~10E13 cells of the human body during the first 30 years of life, it
>is difficult to explain why everyone does not die of cancer unless very
>effective mechanisms for removing the damage are a dominant factor".

It would seem that an overall model is based on the _net_ response (hits
minus repairs).  This might be linear or nonlinear, depending on the
details.  If 9 of 10 molecular lesions can be repaired in a tissue, but
only 800 of 1000 lesions could be repaired in the same tissue, then you
would expect nonlinearity.  (In fact, I believe this happens at very high
doses an dose rates.  Than't why cell survival curves are exponential.)
But I don't know whether the efficiency for repair/removal changes at
relatively low doses.  Do the repair mechanisms handle 1 lesion/cell better
than 5 lesions/cell?  If the repair _efficiency_ is the same regardless of
dose/dose rate, it seems that the overall response should still be linear.

Here's an illustration with numbers:  Suppose a tissue can remove 99.9
percent of the molecular lesions it experiences and that 10 mrad causes
1000 lesions.  At 10 mrad there are 1000 lesions, but 999 are repaired,
leaving only 1 lesion that might go on to produce cancer.  At 20 mrad there
are 2000 lesions, but 1998 are repaired, leaving 2 lesions that might
produce cancer.  And so on.  If only 1996 are repaired at 2 mrad (leaving 4
lesions that might produce cancer), then I can see a basis for nonlinearity.

An aside:  If the repair efficiency for small animals (the mouse above) is
greater than the repair efficiency for a large animal (elephant), than this
might explain why dose rather than total energy is important.  I doubt that
this is the case, though, since mammalian cells are pretty similar across
species.

However, as I said, I think the reason we use dose rather than imparted
energy is historical.  Operationally it would be quite a nuisance to keep
dose values to compare with nonstochastic limits and imparted energy for
stochastic limits.  And since total mass can change fairly rapidly, it
would be a problem to come up with those imparted energy numbers from a
dosimeter reading.  Given all the other uncertainties, I think we are
probably being sufficiently protective by using dose for occupational work.

Regards,
Dave Scherer
scherer@uiuc,edu