Re: New study suggests radon threat may be overestimated

["Strom, Daniel J" <daniel.j.strom@pnl.gov>]

Several of us engaged in a critique of the January 5 PNAS paper by Miller et al.
(1999) on Friday.  It is a well-designed and thought-provoking experiment that
nonetheless has some shortcomings.  This work was reported by Eric Hall at the
NCRP meeting in April, 1998.  

One should consider the following points:

1.  The experimental cell line is C3H10T1/2 mouse fibroblasts, a cell line that
is already far from normal.

2.  For the microbeam experiments, the numbers of transformed cells from 0-, 1-,
2-, 4-, and 8- particle traversals: 4, 5, 7, 5, and 5.  For the broad beam
experiments, the numbers of transformed cells from 0-, 1-, 2-, 4-, 6-, and 8-
particle traversals: 6, 38, 51, 20, 31, 66.  The transformation rates are 0.42
to 13.2 per 10,000 viable cells.  This is hardly hundreds of lung cancers in
uranium miners.  (It is the difference between the exactly 1 and the average of
1 particle hits that is the crux of this paper.  The exactly 1 hit microbeam
experiment yielded 5 transformed cells in 42,700 viable cells for a rate of 1.2
per 10,000, while the average 1 hit broad beam experiment yielded 38 transformed
cells in 124,200 viable cells, for a rate of 3.1 per 10,000.)

3.  There were two differences between the control groups for the microbeam and
broad beam experiments:  i) the transformation rate for the microbeam was 0.86
per 10,000, while that for the broad beam was 0.42 per 10,000; ii) the
clonogenic surviving fraction (plating efficiency) for the microbeam was 0.60,
while that for the broad beam was 0.33.

4.  There are other differences between the microbeam and broad beam
experiments.  For this approach to be valid, one must assume that the broad beam
irradiation of the surrounding cytoplasm (an estimated 2 alpha particles
traverse cytoplasm only for each alpha particle traversal of the nucleus, based
on Fig. 2 of Miller et al. 1999) and of the nearby medium (an estimated 10 alpha
particles traverse medium only for each alpha particle traversal of the nucleus,
again based on Fig. 2 of Miller et al. 1999), neither of which occur in the
microbeam irradiations, do not contribute to the transformation rates.  On an
energy imparted (dose) basis alone, the broad beam irradiations represent
perhaps 3 times as much ionizing energy imparted to each cell as the microbeam
irradiations for the same average or exact number of nucleus traversals.

5.  As I have pointed out in an e-mail to the authors, the microbeam,
exact-number-of-hits data can be used to predict the broad beam Poisson-varying
data.  They do not agree very well, indicating less significant differences than
the authors suggest.  The microbeam data predict not 3.1 per 10,000 in broad
beam experiments but either 2.4 (using microbeam background rates) or 2.2 (using
broadbeam background rates) per 10,000.  Neither of these numbers is
significantly different from the 1.2 per 10,000 observed in the microbeam data.
Thus, the broad beam data point at an average of 1 hit seems to be anomalously
high, and out of line with the Poisson model the authors themselves suggest.

6.  A more convincing experiment would have been to use the microbeam to produce
the Poisson-distributed irradiations, which would have eliminated the fact that
the dose to broad beam-irradiated cells was 3 or 4 times higher than that to the
microbeam-irradiated cells.

7.  Another well-respected researcher also suggested that the irradiation times
are significantly different for the microbeam and broad beam effects.  If there
are bystander effects, or if cells need "permission" from their neighbors to
transform or if their neighbors inhibit them from transforming (as has been
suggested by cell-signaling experiments), then the very fast irradiation times
for broad beam experiments would not permit cells to signal each other during
irradiation, while the slower (3,000 cells per hour) microbeam irradiations
would permit cell signaling effects to take place before all cells were hit.

8.  In the exactly 1 hit microbeam experiment, no cell has an unhit neighbor.
In the average of 1 hit broad beam experiment, 37% of neighbors are not hit.
Again, to the extent there is cell signaling, these are not comparable results.
An interesting experiment would be to hit exactly every other cell or average of
1/2 hit per cell.  This may reveal whether there is any effect of intact
neighbor cells.

These highly-respected authors from Columbia University have indeed broken new
ground and should be commended for carrying out an ingenious experiment.  This,
however, is only one study, and it awaits the critical response of the
radiobiology community.  Before abandoning the linear, no-threshold dose
response model as a basis for radiation protection against the short-lived decay
products of radon, more work is needed. Confirmation of this work in other cell
lines, addressing the points listed above, is also needed.  

Reference

Miller, R.C.; Randers-Pehrson, G.; Geard, C.R.; Hall, E.J.; Brenner, D.J.  The
oncogenic transforming potential of the passage of single alpha particles
through mammalian cell nuclei.  Proc.Natl.Acad.Sci.U.S.A.  96(1):19-22; 1999.

- Dan Strom

The opinions expressed above, if any, are mine alone and have not been reviewed
or approved by Battelle, the Pacific Northwest National Laboratory, or the U.S.
Department of Energy.

Daniel J. Strom, Ph.D., CHP
Risk Analysis & Health Protection Group, Environmental Technology Division,
Pacific Northwest National Laboratory
Mail Stop K3-56, PO BOX 999, Richland, Washington 99352-0999 USA
Telephone (509) 375-2626 FAX (509) 375-2019 daniel.j.strom@pnl.gov
http://www.pnl.gov/bayesian   http://qecc.pnl.gov   http://bidug.pnl.gov

************************************************************************
The RADSAFE Frequently Asked Questions list, archives and subscription
information can be accessed at http://www.ehs.uiuc.edu/~rad/radsafe.html