Radon risks derive from exposure of bronchioepithelial cells to
highLET alpha particles. Alpha particle exposure can result in
bystander effects, where irradiated cells emit signals resulting in
damage to nearby unirradiated bystander cells. This can result in
nonlinear doseresponse relations, and inverse doserate effects.
Domestic radon risk estimates are currently extrapolated from miner
data which are at both higher doses and higher dose rates, so
bystander effects on unhit cells could play a large role in the
extrapolation of risks from mines to homes. We therefore extend an
earlier quantitative mechanistic model of bystander effects to
include protracted exposure, with the aim of quantifying the
significance of the bystander effect for very prolonged exposures.
Fig. 1. Cartoon illustrating the main results
regarding the interplay of risk between dose and dose rate: The
small boxes represent collective, supracellular targets, defined by
the property that a hit on any target cell nucleus in the collective
target results in bystander signal to all cells in that collective
target. Our estimates suggest ~50 target cells / collective target,
but, for clarity, each collective target is shown as containing just
two cells. In a few cases, a collective target may contain a
hypersensitive cell, shown here as solid. The average number of
alpha particle hits is labeled D in this cartoon to emphasize
its proportionality to dose, and the bystander response â€“ number of
hit hypersensitive cells, is labeled R.
A. Panel represents a dose which is "very low"
in the sense that most collective targets are not hit, and the
chance for two alpha particles in one collective target is
negligible. The bystander response is 1.
B. To illustrate the effect of dose rate on
verylow dose risks, we split the same very low dose into two
separate fractions. The pattern of hypersensitive cells can change
between fractions, but it is seen that a very low total dose will
produce the same average response, R. Thus at very low doses,
inverse doserate effects are negligible.
C and D. Here the dose is twice as large, but
it remains low in the sense that the chance of 2 hits per collective
target is negligible. In agreement with general microdosimetric
arguments, the response is also doubled, i.e. is linearly
proportional to dose, and doserate effects remain negligible.
E. At a high acute dose, the chance of more
than one alpha particle per collective target is no longer
negligible and this panel represents the case where an average of 4
hits occurs per collective target. For acute doses, 4 alpha
particles in one collective target are no more effective, in terms
of the bystander response, than one alpha particle; the bystander
response therefore increases less rapidly than linearly with dose
because of "saturation" â€“ some of the alpha particles are "wasted."
F. If the high dose is split into two
fractions separated by a time interval (long enough for
hypersensitive cells to be replaced), the response is doubled, i.e.
there is an inverse doserate effect at high doses.
Overall, comparing panel E with panel B shows that a
linear extrapolation of risk from a high acute dose to low dose and
low dose rate may underestimate this risk, in this schematic case by
a factor of 4, due to saturation and to inverse doserate effects in
the bystander response.
The model was fitted to dose and
doserate dependent radonexposed miner data (2), and gives a
reasonable fit to the data (Fig. 2). Parameters from the fit suggest
that one directlyhit target bronchioepithelial cell can send
bystander signals to about 50 neighboring target cells.
Fig.
2
The estimated
parameter values from this fit were used to extrapolate the miner
data to lower doses, and for a 60year exposure period. The results
are shown in Fig. 3: For the comparatively short miner exposures
(solid curve; for illustrative purposes, we use a duration of 6 y,
the average time of miner exposure in the data), the doseresponse
relation is linear at very high doses (where the direct effect
dominates). It can be seen, however, that at intermediate doses,
where the bystander response starts to become important, the 6yr
exposure (solid) curve become non linear and curves downwards. At
these intermediate doses the risks from a 6yr exposure (dashed
line) are larger than for a 60yr exposure (solid line) â€“ the
inverse doserate effect. At still lower doses, dose rate effects
become small, so the 6yr exposure and the 60yrexposure produce
the same risk.
Fig. 3
Figure 3 also
shows a linear extrapolation of the miner data in which the effects
of dose rate are ignored. It can be seen that ignoring doserate
effect and simply using a linear extrapolation from the miner to the
domestic situation would result (using our estimated parameters) in
an underestimation of the lowdose radon risk by about a factor of
4.5. This underestimation is comparable to the corresponding
empirically estimated factor in the BEIRVI report of ~3.7.
For comparison,
also plotted in Fig. 3 are results from various domestic radon
casecontrol studies. The spread and uncertainties of the results
are such that they are consistent with both the current
mechanisticallybased lowdoserate / lowdose extrapolation, the
BEIRVI phenomenological lowdoserate / lowdose extrapolation, and
also the "naÃ¯ve" lowdose extrapolation from miner data which
ignores the effect of dose rate. It is important, however, to note
that these data typically represent aboveaverage cumulative radon
exposures, and that, assuming lowdose linearity, most radonrelated
deaths will be at still lower cumulative exposures.
The main
conclusions of this analysis are as follows:

At high doses,
the model predicts saturation effects and inverse doserate
effects in the bystander response. At sufficiently low doses, in
agreement with general microdosimetric arguments, the predicted
response is linear in dose and independent of doserate.

Parameter
estimates based on applying the model to dose and doserate
dependent miner data suggest that a single directlyhit target
bronchial basal cell can send bystander signals to about 50
neighboring cells.

The model
parameter values obtained from this analysis of epidemiological
data, in as much as they can be compared with parameter values
obtained from invitro analyses, are significantly
different. Thus model parameters estimated from analysis of
invitro studies cannot necessarily be applied to the
invivo situation.

The highdose
saturation and inverse doserate effects in the bystander response
suggest that a linear extrapolation from miner data which does not
properly take into account dose rate effects would underestimate
the domestic radon risk by about a factor of four â€“ a value
comparable to the empirical estimate applied in the recent BEIRVI
report on radon risk estimation.
It is important
to stress that we have in no sense "proven" the relevance of
bystander phenomena to lowdose radon risks; rather we have
described a mechanistic model which is parsimonious in its number of
parameters (four parameters, making the model potentially highly
testable), and which is consistent with a large body of
epidemiological and laboratory data.
In conclusion,
bystander effects represent a plausible quantitative and mechanistic
explanation of inverse doserate effects by highLET radiation,
resulting in doseresponse relations which are non linear and which
feature a complex interplay between the effects of dose and exposure
time. The model presented here provides a potential mechanistic
underpinning for the empirical exposuretime correction factors
applied in the recent BEIRVI report on domestic radon risk
estimation.
References

Brenner DJ,
Little JB, Sachs RK, The bystander effect in radiation oncogenesis,
II. A quantitative model, Radiation Research 155:402408,
2001.

Lubin JH et
al, Radonexposed underground miners and inverse doserate
(protraction enhancement) effects, Health Physics 69:494500,
1995.