PNL-SA--18754
DE91 004737
PHENOMENOLOGICALMODELS
L. A. Braby
m _" ." _" " .w ' m _
0 U .": o ,,,::
,_ _,_ _._ =_- _ c _+k.ev_em_er 1990
,,.,, c= o _0 u_.. ,-
;::) t.. >., ,_.r" 0, _"_ "
>.... _ _ _-= " --'rrepare_ for
._ _ _ _-_ o - : _ DOE Conference on
_ = ,--..-. _ ._ =-,., Molecular and Biological
. _ ._ . . > _ Effects .. ,,a.,a.,on,
f_f & rl.i +-_
,,.,, _ _.u _ _...£ ,..._
,_. _ =_=-- .-_ o Woods Hole MA
>'o_-_ _o_=_:_ _ September 4-7 1990
O •-_ I;_. ,_ ,-, u 0 ,: ..,.., __
.',=_ ".-,' ,-,, Work supported by the Office of
; - .. p .u. _ w) o E
_'- : _ _ _-==s : Health and Environmental Research
"_ _ _.4 _'-._,_'"_ (OHER) U S Department of Energy
_z_ _ _ _ o_ ' " "
_ E_ _ -"._o _ ,_ under Contract DE-ACO6-76RLO 1830
_- _ _ _ ." ,-- -- c: ,. _._
Pacific Northwest Laboratory
Richland, Washington 99352
MASIEB
c:"/]
BIBTRIBUTIONOFTHIS{I[lCU_ENTfS UNLIMITED
PHENOMENOLOGICALMODELS
L. A. Braby
Pacific Northwest Laboratory
Richland, WA 99352
ABSTRACT
The biological effects of ionizing radiation exposure are the result of a
complex sequence of physical, chemical, biochemical, and physiological
interactions which are modified by characteristics of the radiation, the
timing of its administration, the chemical and physical environment, and the
nature of the biological system. However, it is generally agreed that the
health effects in animals originate with changes in individual cells, or
possibly small groups of cells, and that these cellular changes are initiated
by the ionizations and excitations produced by the passage of charged
particles through the cells. One way to begin a search for an understanding
of health effects of radiation is through the development of phenomenological
models of the response. Many models have been presented and tested in the
slowly evolving process of characterizing cellular response. Different
phenomena (LET dependence, dose rate effect, oxygen effect etc.) and different
endpoints (cell survival, aberration formation, transformation, etc.) have
been observed, and no single model has been developed to cover all of them.
Instead, a range of models sovering different endpoints and phenomena have
developed in parallel. Many of these models employ similar assumptions about
some underlying processes while differing about the nature of others. An
attempt is made to organize many of the models into groups with similar
features and to compare the consequences of those features with the act,_al
experimental observations, lt is assumed that by showing that some
assumptions are inconsistent with experimental observations, the job of
devising and testing mechanistic models can be simplified.
INTRODUCTION
Models, ranging from vague conceptual descriptions to detailed
mathematical expressions, play a major part in the development of most
sciences. Models serve many different purposes, and take on different
properties in accordance with those purposes. When previously unknown
phenomena are observed, scientists find it natural to look for their cause and
to devise models which will help in that effort. The process of organizing
observationsimpliessome model which provides a structurefor the resulting
list. This model may only be a vague concept of the relationshipbetween the
observedendpoint and changes in the system,or its environmentwhich may
effect that endpoint, lt may be no more than an effort to fit mathematical
relationshipto the effect and one or more variablesthat are thought to be
related to it, but even the act of selectingthe related variables
necessitatessome thought about the underlyingmechanisms.
The effect of ionizing radiation on biological systems has been found to
be the result of an extremely complicated combination of physical, chemical
and biological processes. Thus the primary function of many models of the
effects of ionizing radiation on cells has been to aid in the organization of
the data so that somepattern can be recognized in the complex collection of
causes and effects. As the data is organized, trends become visible, and the
model can be used to predict effects under conditions which have not
previously been tested experimentally. These predictions, which may be
derived from purely conceptual models but which are more likely to be
quantitative if the model is given a mathematical formulation, can be used to
help in the design of experiments to test the validity of the assumptions of
the model. Thus model building and testing becomes an iterative process.
Refinements in the model provide new predictions, which can be used to devise
new experiments, which produce new data, which provides the information needed
to refine the model.
The use of models to predict effects under conditionswhich haven't been
exploredexperimentallyhas an importantpracticalapplicationas weil.
Radiationexposure limits for the protectionof radiationworkers and the
general publicare based, as much as possible,on real human health effects
data. However,these data are quite limitedwith respect to the range of
doses, the variety of effects cbserved,and the statisticalreliability. Thus
it is necessaryto utilizewhatevermodels are available to guide the
extrapolationto the effects of low doses and to predict possibleeffects
under differentconditionsof exposure. To some extent data obtainedwith
experimentalanimalscan be used to estimatethe effects in humans,but even
animal experimentsare limited to relativelyhigh doses and dose rates by the
cost and experimentaldifficultyinvolvedin trying to obtain statistically
reliable results at doses which are relevantto th_ majority of actual human
exposures. Thus it is frequentlynecessaryto devise models of the effects on
experimentalanimalsand then to devise an additionalmodel that relates the
animal responseto health effects in humans.
The most important use of models may be their inherent relationship to
the mechanisms which are responsible for the effects being modeled. The
health effects of irradiation involve not only the production of biochemical
damage by different radiations under different physical and chemical
environments, but also the modification of thal. damage by a variety of
biochemical systems which may be involved in modifying damage produced by
other environmental insults and the normal metabolic activity of the cell.
Furthermore, both the production of specific biochemical changes and the
response of cells to those changes may depend on the current and past exposure
of that cell to environmental factors which alter the levels of protective
compounds and the activity of the repair systems. In order to interpret or
predict interactions between radiation exposure and other health risk factors
such as smoking and industrial chemical exposure, we must know considerably
more about the mechanisms governing these effects.
BACKGROUND
Limitations on Models
Ali efforts to model natural systems face a dilemma. In order to
describe the complex processes which occur in nature, the model must include
many variables. However, to be testable and useful in any predictive way the
model must not have more variables than the experimer, ts which are used to test
it. For exar_'Jple, a very simple target model is sufficient to describe the
inactivation of phage by low LET radiations, but this is not very helpful when
considering the risk to mammalian cells with their complex capacity to repair
DNAstrand breaks and to attenuate the killing effect during a post
4