NCRP report no 121 principles and application of collective dose in radiation protection

Some authors trace the origin of the concept back to the term "genetically significant dose," or "population dose" which was proposed to limit radiation induced genetic risk of populatio

NATIONAL COUNCIL O N RADIATION

PROTECTION AND MEASUREMENTS

Issued November 30, 1995

National Council on Radiation Protection and Measurements

791 0 Woodmont Avenue / Bethesda, MD 20814-3095

LEG& NOTICE

This report was prepared by the National Council on Radiation Protection and Measurements (NCRP) The Council strives to provide accurate, complete and useful information in its reports However, neither the NCRP, the members of NCRP, other persons contributing to or assisting in the preparation of this Report, nor any person acting on the behalf of any of these parties: (a) makes any warranty or representation, express or implied, with-resped to the accuracy, completeness or usefulness of the information contained in this Report, or that the use of any information, method or process disclosed i n this Report may not infringe on privately owned rights; or (b) assumes any liability with respect to the use of, or for damages resulting from the use of any information, method or process disclosed in this Report, under the Civil Rights Act of 1964, Section 701 et seq, as amended 42 U.S.C Section 2000e et seq (Title VZZ) or any other statutory or common law theory governing liability

Library of Congress Cataloging-in-Publication Data

Principles and application of collective dose in radiation protection

p cm.-(NCRP report ; no 121)

Includes bibliographical references and index

ISBN 0-929600-46-0

1 Radiation-Dosage 2 Radiation-Safety measures

I National Council on Radiation Protection and Measurements

11 Series

RA569.P6713 1995

616.9'897-dc20

95-26030 CIP

Copyright Q National Council on Radiation Protection and Measurements 1995 All rights reserved This publication is protected by copyright No part of this publication may be reproduced in any form or by any means, including photocopying, or utilized by any information storage and retrieval system without written permission from the copyright owner, except for brief quotation in critical

articles or reviews

Preface

The Committee on Interagency Radiation Research and Policy Coordination asked the National Council on Radiation Protection and Measurements (NCRP) to provide advice on the use of collective dose in radiation protection, particularly as it should pertain to radiation exposures of the United States public

In response to this request, NCRP Scientific Committee 1-3, Collec-

tive Dose, was established Serving on Scientific Committee 1-3 were:

Washington State University Richland, Washington

Members

Richland, Washington

New York, New York

David A Waite

Ebasco Environmental Bellevue, Washington

Scientific Committee 1 Liaison

Eric J Hall

Columbia University New York, New York

iv 1 PREFACE

NCRP Secretariat

The Council wishes t o express its appreciation to the Committee members for the time and effort devoted to the preparation of this Report

Charles B Meinhold President

Contents

Preface 1 Introduction

2 Historical Development ,

2.1 Introduction

2.2 Applications

2.3 Concept Evaluations

3 Scientific Bases for Collective Dose

3.1 Introduction

3.2 Mutagenesis 3.2.1 Cellular Studies

3.2.1.1 Cytogenetics

3.2.1.2 Somatic Cell Mutations

3.2.2 Animal Studies

3.2.2.1 Chromosome Aberrations

3.2.2.2 Germ Cell Mutations

3.3 Transformation and Carcinogenesis

3.3.1 Tumor Induction

3.3.1.1 Leukemia

3.3.1.2 Solid Tumors

3.3.2 Life Shortening

3.3.3 I n Vitro Transformation

3.4 Human Studies

3.4.1 Human Studies of Cancer Risks from Low

Radiation Doses 3.4.1.1 Thyroid Cancer

3.4.1.2 Breast Cancer

3.4.1.3 Leukemia

3.4.1.4 Multiple Myeloma

3.4.1.5 In Utero Irradiation 3.4.1.6 Lung Cancer

3.4.1.7 Other Cancers

3.4.2 Genetic Risks

3.5 Summary

4 Limitations

4.1 Conceptual Limitations

4.2 Practical Limitations

vi / CONTENTS

4.2.1 Tissue Weighting Factors

4.2.2 Population Characteristics 4.2.2.1 Uncertainties in Future Population Size and Location

4.2.2.2 Uncertainties in Future Population

Fertility 4.2.2.3 Uncertainties in Future Medical Technology

4.2.3 Environmental Exposure Pathways 4.2.3.1 Agriculture

4.2.3.2 Resource Conservation

5 Risk Assessment and Management

5.1 Collective Dose as a Surrogate for Societal Risk 5.2 Collective Dose Distributions

5.3 Risk Assessment in Specific Applications

5.3.1 Medical Procedures

5.3.2 Radiation Workers

5.3.3 Special Occupational Groups

5.3.4 Current Exposures to Members of the Public

from Localized Environmental Sources

5.3.5 Indoor Radon 5.3.6 Consumer Products and Other Miscellaneous

Sources 5.3.7 Future Exposures from Long-Lived Environmental Contaminants

5.4 Risk Management

5.4.1 Acceptability of Risk

5.4.2 Categorizing Levels of Risk

5.4.3 Optimization of Protection (ALARA)

5.4.4 Valuation of Collective Dose Avoided

6 Conclusions and Recommendations

Glossary

References The NCRP

NCRP Publications

Index

1 Introduction

Conceptually, collective dose is the summation of all doses received

by all members of a population a t risk, and may thus be expressed mathematically as:

where S refers to the collective dose to the population a t risk, and

Hi is the per capita dose in subgroup i, and Pi is a subgroup i of

population P (ICRP, 1977) Any dose quantity can be used, provided usage is consistent Collective dose is expressed in units of person- dose, using the appropriate dose units for the quantity selected Typically, collective dose to a population is expressed in units of person-sievert

Collective dose is applicable only to stochastic risks Implicit in the concept of collective dose is the assumption that the effect or risk of a given dose is identical whether the collective dose is adminis- tered to a single individual or distributed over a population of individ- uals Application of collective dose in this manner assumes linearity

of dose response, and lack of a n y dose-rate effect While these assumptions may or may not be valid, they are considered to be conservative and have been generally accepted by the scientific com- munity concerned with radiation protection (ICRP, 1977; 1991; NASI NRC, 1990; NCRP, 1987a; 1993)

In recent years, collective dose has been applied ever increasingly

to prospective radiation protection problems, particularly relating

to long-term effects of environmental radiation Such applications lead to questions regarding the applicability of the collective dose concept to large populations with very small individual doses and

to populations that will exist several generations hence This Report seeks to address these and other questions regarding collective dose and its applicability for radiation protection purposes, and to provide practical guidance for the employment of this potentially useful con- cept in consonance with current National Council on Radiation Pro- tection and Measurements (NCRP) philosophy and recommendations

on exposure limitation a s described i n NCRP Report No 116 (NCRP, 1993)

This Report provides a review of the historical development and current applications of the collective dose concept, and attempts to

2 / 1 INTRODUCTION

identify and examine the scientific and other bases that underlie it

It examines the meaning and utility of the concept ofcollective dose in radiation protection and risk assessment for workers and members of the general public Finally, it provides recommendations for applying collective dose based on current scientific knowledge of the health effects and potential risks of radiation

Underlying the consideration of the collective dose concept and the recorrfmendations provided herein is the continuous evolution of radiation protection standards towards a system based on risk For such a risk based system to be practical, it must take account of the uncertainties in the risk estimates which form its basis Additionally, consideration should be given to societal factors, including the will- ingness of society to incur certain risks in view of the perceived overall benefit to be derived

2 Historical Development

The collective dose concept is widely used within the radiological protection community in the estimation of radiological risk, in the optimization or decision making processes, and in the development

of regulations Some authors trace the origin of the concept back to the term "genetically significant dose," or "population dose" which was proposed to limit radiation induced genetic risk of populations

annual collective dose limitation of 100 person-Sv y-' for each nuclear power station imposed by the Canadian Atomic Energy Control Board (Hurst and Boyd, 1972) The concept does appear in the Inter- national Commission on Radiological Protection's (ICRP) Publication

22 (ICRP, 1973), where it was first called "population dose," and evolved to "collective dose" by the time of ICRP Publication 26 (ICRP, 1977)

Modern usage of the collective dose concept originated in the early 1970s within the United Nations Scientific Committee on the Effects ofAtomic Radiation (UNSCEAR) and the ICRP The 1969 UNSCEAR report did not mention collective dose, but by 1977, the use of collec- tive dose by UNSCEAR was prevalent (LTNSCEAR, 1977) The tran- sition seems t o h a v e occurred i n t h e 1972 UNSCEAR report (UNSCEAR, 1972), in which the unit man-rad was introduced Popu- lation doses in units of person-rem were also used in the initial report of the Committee on the Biological Effects of Ionizing Radia- tion [BEIR (NASINRC, 1972)l

No 39 (NCRP, 1971) did not mention the conceptper se But, NCRP

the population of the United States as a whole from all sources

of radiation, including medical and other manmade sources, and background, shall not exceed 14 million rems per million of popula-

that amount in each decade thereafter." The first specific reference

Report No 43, entitled Review of the Current State of Radiation

4 / 2 HISTORICAL DEVELOPMENT

Protection Philosophy (NCRP, 1975), and the concept has been dis- cussed, refined and applied in subsequent NCRP reports and recommendations

The use of the collective dose concept has permeated into many aspects of radiation protection policy making and program imple- mentation worldwide The scientific and technical literature contains numerous examples of collective dose-based estimates of the collec- tive risk for a wide variety of radiation-related activities In its series

of reports assessing the ionizing radiation exposure to the population

of the United States (NCRP, 1987b; 1987c; 1987d; 1987e), the NCRP has made extensive use of the collective dose concept NCRP Report

No 93 (NCRP, 1987b) summarizes radiation exposures from all sources that were individually reviewed in the other assessment reports, and includes collective dose estimates Other NCRP reports, notably Reports No 105, 107 and 116 (NCRP, 1989; 1990a; 1993) consider collective dose in their discussion of radiation protection recommendations

Other examples of considerations of collective dose for radiation protection purposes in other countries include a study by Iyengar and Soman (1987) which examines in detail the collective occupa- tional and public doses from all components of the Indian nuclear fuel cycle Similarly, Hyvonen (1990) evaluates the effectiveness of

medicine, industry, research and nuclear power Early United States examples include the final environmental statements for Pilgrim Nuclear Power Station (AEC, 1972) and for Hanford Waste Manage- ment Operations (ERDA, 1975), both of which discuss impacts and comparative population or overall risks in terms of collective doses,

Relevant guidance documents incorporating collective dose have been prepared by others including the ICRP (19731, and the Organi- zation for Economic Cooperation and Development/Nuclear Energy Agency (OECDLVEA, 1988) In the mid-1950s the principle of main- tainingradiation exposures to the lowest practicable limit was intro- duced into its recommendations by the NCRP (1954) and the concept

able), began to evolve (ICRP, 1955; 1959) This concept is now central

to radiation protection practice and is based on a balancing of risks

ommendation for radiation protection by the ICRP in 1977 (ICRP,

2.3 CONCEPTEVALUATIONS 1 5

1977), although its origins in radiation protection practice go back

a t least to the early 1950s (Kathren et al., 1980)

Regulatory bodies have integrated the collective dose concept into United States radiation protection regulations in various ways In the mid-1970s, the Nuclear Regulatory Commission (NRC) adopted the use of the collective dose concept with a spatial cutoff of 50 miles, see Appendix I to 10 CFR 50 (NRC, 1975) Amplification of the regulation was provided two years later in NRC Regulatory Guide 1.109, Appendix D, which stated that "These doses should be evalu- ated for the population within a 50-mile radius of the site For the purpose of calculating the annual population-integrated dose, the 50-mile region should be subdivided into a number of subregions consistent with the nature of the region" (NRC, 1977) This type of spatial truncation has been widely accepted and utilized in the past, particularly in documents such a s environmental impact statements prepared for regulatory purposes

Such acceptance has not been the case when dose-related and other truncation rationales have been attempted in other aspects of the regulatory framework such a s the NRC proposed adoption of

"below regulatory concern" for determining when individual radia- tion exposures are or will be so low that they do not warrant further regulatory control (NRC, 1988) I n addition to an individual dose criterion, the NRC proposed a collective dose limit as well, stating

in its 1988 policy statement that "The Commission specifically seeks comments on the need for establishing a collective dose limit in addition to a n individual dose criterion" (NRC, 1988) These propos- als have not been adopted

2.3 Concept Evaluations

I n recent years a t least two studies have examined the fundamen- tal validity and utility of the collective dose concept The first was

a study by Lindell(1984) who was commissioned by the OECDINEA

to prepare a report describing the various situations in which ICRP recommendations would require a n assessment of collective dose, the objectives of such assessments, the related methods, and the limits and difficulties of these collective dose assessments Regula- tory aspects were not addressed in this study

I n the Lindell report (Lindell, 1984), the following applications of the collective dose concept are identified a s the most commonly used

in radiation protection:

1 in the assessment of the highest per capita dose in the future from a continued practice which exposes some members of the population to radiation,

2 in the limitation of present use of radioactive material, if it is believed that additional sources in the future may add to the per capita dose in a population so that it might reach unacceptable levels unless all sources are controlled a t a n early stage,

3 as a n input to justification assessments, indicating the total detriment from a certain practice, and

4 a s an input to optimization assessments a s the basis for costing detriment in differential cost-benefit analysis of protection arrangements

His primary conclusion is that while it is often said that for the collective dose to be useful, an assumption of a nonthreshold, linear dose-response relation is needed, in truth, this assumption is not

assumptions on the dose-response relationship at low doses Only

relationship

Lindell also acknowledges that there is some hesitation in using the collective dose, not only due to distrust of the biological assump- tions implied by uses (3) and (4), but also in lack of confidence in the predictiveness of collective doses that have been derived by adding contributions over very long time periods However, none of the four applications is by necessity related to extreme time scales

The second was a study commissioned by the German Radiological Protection Commission in 1985 (SSK, 1985) The objective of this investigation was to determine whether collective dose is suitable

as (1) a measure of the radiation-related detriment and (2) a tool for the optimization of radiological protection and for the comparison

of safeguards, and thus a meaningful measure of radiation expo- sures The study considered both the scientific state-of-the-art and the legal situation that existed in Germany in 1985, and reached the following conclusions relative to detriment, optimization and regulation:

"1 The collective dose is only suited to be a measure of detriment

if there i s a sufficient knowledge of t h e risk coefficients required for the calculation of detriment in the dose range

of interest As far as the relevant dose ranges in practical

radiological protection are concerned, it must be recognized that the required risk coefficients are derived from estimates and not from quantitative determinations This applies in particular to dose ranges t h a t a r e of importance for t h e populations

2.3 CONCEPTEVALUATIONS / 7

"2 With respect to protection implementation, the Radiological Protection Commission considers the optimization of the radio- logical protection of personnel by means of a minimization of the collective dose and the comparison of safeguards suitable, using the collective dose a s a measure of comparison

"3 Although there is a binding obligation for the optimization of

radiological protection, there is no obligation to consider the

collective dose as a suitable tool for reaching this objective

The Radiological Protection Commission recommends that col- lective dose not be included in legal regulations."

made in these earlier studies and reviews these issues in the context

of present circumstances in the United States

3 Scientific Bases for

Collective Dose

The utility of collective dose rests on the assumption that the biological response at low dose and dose rates is both linear and time independent, and that the response of any individual to a given dose is more or less uniform This logically leads to the prediction that at low doses the response will be the same whether the dose is delivered a s a single acute exposure, as multiple small fractions, or

as a protracted low-dose rate exposure The assumption of time independence also implies that the time between each fraction and the time over which the total dose is delivered are not important Whether these assumptions are appropriate have not been deter- mined from human epidemiologic data Some animal studies have shown that both the time between fractions and the time over which the total dose is delivered are important The following sections will review cellular, animal and human studies on the mutagenic and carcinogenic effects of radiation which have bearing on these assumptions of linearity and time independence a t low doses

3.2.1 Cellular Studies

3.2.1.1 Cytogenetics The effects of dose, radiation quality, and dose rate or fractionation on the yield of chromosome aberrations have been extensively studied using cultured lymphocytes from a variety of species including humans Data indicate t h a t similar

results are obtained whether cells are irradiated in vivo or in vitro (Brewen and Gengozian, 1971; Schmid et al., 1974) Pertinent

reviews can be found in NCRP (1980; 1990b) and UNSCEAR (1988;

3.2 MUTAGENESIS 9 aberrations following in vitro exposure of human lymphocytes has

two-break chromosome aberrations, such a s dicentric aberrations, increase with dose according to the linear quadratic function:

imply that the response a t low doses is linear and time independent

At very low doses induced by internally deposited radioactive materi- als, the yield of aberrations was found to be described by the function

a t low radiation doses whether protracted or delivered a s an acute exposure For high-LET radiation, the yield of aberrations increases

a s a linear function of dose over a wide range and is dose-rate inde- pendent This point is discussed in greater detail below

3.2.1.2 Somatic Cell Mutations The induction of gene mutations

in cultured cells by irradiation has been studied by a number of investigators using several different cell lines including those derived from mice, hamsters and humans In addition, specially

engineered hybrid hamster cells containing the bacterial gene gpt

Because of the limited sensitivity of these model systems, most stud- ies have not directly examined the dose response at doses below about 0.5 Gy Although data a t low doses are limited, inferences can

be drawn about the shape of the dose-response curve in the low-dose range based on the effects of dose rate on the response and based

on molecular analyses of the induced lesions The data indicate that the dose-rate dependence of radiation induced mutations depends upon the type of mutation induced (NASNRC, 1990) In general, lesions that can be hypothesized to involve the interaction of dam- aged DNA, such as intragenic deletions, rearrangements and other multilocus mutations, have been found to be dose-rate dependent Because of the apparent involvement of the interaction of sublesions, the prediction of a linear-quadratic dose-response function and dose- rate dependence for such lesions seems reasonable over the range

of doses used In other instances, such as point mutations (i.e., base substitutions), the data suggest no dose-rate effect and also strongly

10 / 3 SCIENTIFIC BASES FOR COLLECTIVE DOSE

suggest a more linear dose-response relationship over a wide range

of doses In either case, the response a t low doses is expected to

be linear and independent of the time course over which the dose

is delivered:

For high-LET radiation, the association is somewhat less certain although the data indicate a n approximately linear dose response following acute exposures Studies examining effects of dose rate and fractionation, however, suggest that reducing the dose-rate results in

dose range (NCRP, 1990b)

The principal focus when considering genetic effects in animals

is on the germ cells, for which information comes primarily from experiments examining chromosome aberrations in these cells and from studies of specific locus mutations in the mouse These studies have been reviewed extensively and most recently in ICRP (19911, NAS/NRC (1990) and UNSCEAR (1988) With respect to low-LET effects a t low doses and dose rates, the most comprehensive reviews

review of high-LET effects from external exposure is presented in NCRP Report No 104 (NCRP, 1990b) and for internally deposited radionuclides in NCRP Report No 89 (NCRP, 1987~)

translocations in spermatogonia has been studied over the dose

linear-quadratic equation followed by a downturn a t higher doses (UNSCEAR, 1986) Reducing the dose rate or the size of the dose fractions reduces the response principally by reducing the beta term (see Equation 3.1) This suggests a more nearly linear response a t dose rates below about 0.1 mGy min-l Further reduction in the dose rate below 0.1 mGy min-I does not significantly affect the slope of this response

Information on chromosomal changes in oocytes is available from studies of Brewen and coworkers (Brewen and Payne, 1977; 1979;

rations following acute exposures increased as a linear quadratic function of dose Chronic gamma irradiation resulted in a linear response function with a slope similar to t h e linear portion of the linear quadratic responses obtained following high-dose rate exposures

3.2 MUTAGENESIS 1 11

3.2.2.2 Germ Cell Mutations Evaluations of specific locus muta-

tions in mice have emphasized the studies of spermatogonia and the resting oocyte Because of the extreme sensitivity of the oocyte, which results in killing and early onset of sterility a t intermediate to high doses of low-LET radiation, studies first concentrated on responses

in spermatogonia While dose-response data for spermatogonia are limited, i t is clear that lowering the dose-rate results in a progressive decrease in mutation incidence down to a dose rate of approximately

10 mGy min-' (UNSCEAR, 1986) Importantly, lowering the dose rate below 10 mGy min-l results in no additional decrease in muta- tion incidence

Although fractionation studies are more difficult to interpret, it

is important to point out the studies of Lyon et al (1972) that com-

pared the effects of a single 6 Gy dose with fractions divided into 60 daily doses delivered a s acute fractions or split into weekly doses of 0.5 Gy delivered a s acute fractions The daily fractionation regimen resulted in mutation frequencies similar to those obtained at low- dose rates while the higher weekly fractions resulted in mutation frequencies similar to those after acute exposures These results,

as well as the dose-rate data described above, are consistent with additivity of effects a t low doses and dose rates

In the female mouse, few if any mutations are observed a t doses

up to several Gy when delivered at low-dose rates (NCRP, 1980; 1990b; Searle, 1974) Because of the extreme sensitivity of the mouse oocyte to killing by x rays, this test system has been called into question a s far as its applicability to humans is concerned

The dose-response relationship for mutation induction following exposure of spermatogonia to neutron irradiation appears to be lin- ear over the 0.5 to 0.9 Gy dose range, but the mutation frequency

is markedly lower a t a dose of 2 Gy (NCRP, 1990b) Dose rate appears

to have little influence on the mutation yield obtained in the 0.5 to 0.9 Gy dose range At higher dose rates and dose, a reduction in yield is observed for many endpoints This is most often explained

on the basis of cell killing More significant for radiation protection considerations is the apparent lack of any influence of dose rate a t lower doses

Considering the small amount of data available in the low-dose range and t h e associated complicating factors, particularly with respect to the female mouse, it appears that most of the data on mutations a r e consistent with a linear a n d time-independent response to radiation in the low-dose region following exposure to low-LET radiation Data for spermatogonia irradiated with high- LET radiations are also consistent with a linear, dose-rate indepen- dent response

12 1 3 SCIENTIFIC BASES FOR COLLECTIVE DOSE

3.3.1 Tumor Induction

Studies of dose-response, time-dose relationships and influence of radiation quality for tumor development in laboratory studies of

(1990b) Despite the large body of data, there are only a few instances

in which the dose response is sufficiently well defined and for which time-dose relationships have been studied

3.3.1.1 Leukemia A large body of data is available on the induc- tion of myeloid leukemia in two strains of mice It has generally been concluded from these data that the dose response is linear quadratic, although a linear response cannot be excluded (Barendsen,

1978) and the data for one of the strains has been described to fit a

dose rate has been shown to result in a reduced response per unit

concluded that the linear component of the linear quadratic model fitted to t h e high-dose r a t e d a t a adequately predicted t h e d a t a obtained for continuous low-dose rate and fractionated doses These data support the conclusion of a linear, time independent, additive response a t low doses For thymic lymphoma induction, the dose response and the effect of low-dose rate are complex and the response

a t low doses has not been sufficiently well characterized to allow any conclusions to be reached

3.3.1.2 Solid Tumors Ullrich (1983) and Ullrichet al (1987) have reported studies on mammary and lung adenocarcinoma develop- ment as a function of dose, dose rate and fractionation Following high-dose rate exposures, the data indicate linear quadratic dose- response functions for both tumor types although the dose range over which the linear response predominated differed markedly For low-dose rate exposures the data were best described by linear functions with slopes similar to the linear portions of the linear quadratic equations obtained following high-dose rate exposures On the basis of these results, the effects of low-dose rate exposures and

of high-dose rate low-dose fractions were compared in a direct test

of the prediction of dose-rate independence a t low doses, i.e., doses where the linear response predominates The data demonstrate that the effects of fractionation were predicted by the linear quadratic regression equations derived from the high-dose rate data When the dose per fraction was on the predominantly linear portion of the

are more consistent with a threshold model However, the apparent extreme sensitivity of the oocyte to killing effects, and the possible role of indirect mechanisms in ovarian tumor development, suggest that this may be a response unique to ovarian tumors in the mouse Data for induction of other tumors are not sufficient to contribute

to resolution of the question Taken as a whole the above data are consistent with t h e concept of a n additive, time independent response for tumor induction a t low doses

All available data for tumor induction following high-LET radia- tion support a linear dose response a t doses below 0.1 Gy (NCRP, 1990b) Further, with the exception of mammary tumor induction,

it appears that the response following fractionated or protracted exposures to low total doses is also linear For mammary tumors the incidence is two- to threefold higher following low-dose rate exposure than after high-dose rates, even after a dose as low as 0.025 Gy The reason for this effect is not known

3.3.2 Life Shortening

Life shortening is one of the most extensively studied late effects

of exposure to ionizing radiation Since it has been shown that virtu- ally all the excess life-shortening effects that occur after an exposure

a t low dose or low-dose rates is due to excess tumor mortality, this endpoint is a useful quantitative tool for the study of the neoplastic effects of radiation a t low doses and dose rates The quantitative nature of this endpoint and the ease of measurement have allowed investigators to examine dose-response relationships and to rigor- ously examine effects of dose rate, protraction and radiation quality

A review of these data is available in NCRP Reports 64 and 104 (NCRP, 1980; 1990b)

Several concepts have emerged from these studies which are of direct relevance to this Report Results obtained with a number of

14 / 3 SCIENTIFIC BASES FOR COLLECTIVE DOSE

different strains of mice, have led to the general conclusion that at low doses and low-dose rates a linear relationship exists between the degree of life shortening and the magnitude of low-LET radiation

sions While there is some evidence of apparent life lengthening a t low radiation doses (Congdon, 1987; Lorenz et al., 1955; Mine et ak., 1990), the mechanisms of this effect may be related more to the low grade stress of the radiation exposure rather than to a true radio-

discussion of these observations and their biologic mechanisms can be found in NCRP (1980), Sacher and Trucco (1962) and Sagan (1989) Investigators have concluded that the dose response for high-LET irradiation is linear at low doses At doses above about 0.2 to 0.4 Gy, there is a bending or deviation from linearity in the dose-response curve While enhanced life shortening effects have been observed following low-dose rate exposures a t total doses greater than 0.2 Gy, fractionation and dose-rate studies strongly support the conclusion that the response is in fact linear and additive a t low doses (NASI NRC, 1990; UNSCEAR, 1986)

Since the first published report of radiation induced transforma- tion in uitro, these model systems have served as useful tools with which to explore many questions (Borek and Sachs, 1966) Particu- larly relevant to this Report are studies of dose-rate and fractionation effects The repairability of low-LET radiation induced transforma- tional damage was one of the first observations made with the C3H 10T1/2 cell system Subsequently, it has been demonstrated that reduction of the dose rate of low-LET radiation results in a reduction

in the transformation frequency in most systems studied Results following fractionation are more complex and depend on total dose, fraction size and time between fractions The most complete data set for studies of dose-rate and fractionation effects for low-LET radiation for in uivo transformation studies comes from the work

of Elkind and Han (Elkind et al., 1985; Han et al., 1984) These investigators reported a reduced transformation frequency after low- dose rate exposures Daily fractions of high-dose rate exposures of 0.5 Gy also resulted in a lower transformation frequency than that from a single acute dose

In contrast to the results with photons, studies of dose-rate effects

effect a t doses above 0.1 mGy when the dose r a t e is less t h a n

3.4 HUMAN STUDIES 1 15

5 mGy min-' (Elkind et al., 1985; Hill et al., 1982; 1984) While these results were initially somewhat controversial, similar results have now been reported by some other, but not all, investigators using different in vitro cell systems (see Hall et al., 1991) This so called inverse dose-rate effect appears to be a complex function of LET, dose and dose rate The reason for this effect is not known Recently, Brenner and Hall (1990) have proposed a model involving a sensitive stage in the cell cycle for transformation which is consistent with the experimental data

Two interrelated questions pertinent to the issue of collective dose can be examined in the human epidemiologic database The first is whether projections of risks from high doses to low doses are correct using a dose-response relationship that is a linear function of dose Alternatively, the question may be whether a linear-quadratic (convex upward) model, for which both components have been esti- mated, projects the risk with reasonable accuracy If the dose- response curve is truly linear quadratic, but is estimated with a simple linear function, then low-dose effects and the estimated risk from collective low-level doses will be overestimated The second question is whether the effects of many small dose fractions or highly protracted doses are additive If so, then the application of collective dose is appropriate, but if not, then collective dose based on many small doses could overestimate the risk

The current risk estimates are largely based on epidemiologic studies involving acute exposures up to a few Gy, such as the Japa- nese atomic-bomb survivors and people receiving x-ray treatments

for a variety of medical conditions A number of studies have docu-

mented the nature of risks in these populations, and there is reason- able agreement among the studies as to the magnitude of risk per unit dose

There is less certainty about the magnitude of risks resulting from exposures at low doses andlor doses at low-dose rates It is intrinsically difficult to assess risk accurately and precisely when doses are below

50 mSv or when they are delivered at rates of a few mSv per year Even among the survivors of the atomic bombings in Japan, a risk from doses below approximately 200 mSv has not yet been precisely demonstrated Lowdose studies tend to be limited for at least two reasons First, with a lowdose study the magnitude of confounding effects may be as large or larger than the exposure-caused effects and

16 / 3 SCTENTIFIC BASES FOR COLLECTIVE DOSE

hence may give "false positive" or "false negative" results Typically, most of the potential confounding variables are either unknown, or

data are not available to control their influence

studies: the smaller incidence of radiation-induced cancers may be drowned out by the much larger spontaneous incidence This means that for low-dose studies, a large sample size will improve the SNR and permit detection of smaller differences between the spontaneous

or background rate and the observed rate In particular, the required sample size is a nonlinear function of the expected size of the effect

To give a hypothetical example; if there is a linear relationship between radiation dose and lung cancer risk, and 1,000 subjects need to be studied to detect an excess risk when the dose they receive

to be studied to detect the same excess risk, and nearly seven million

prohibitively large when doses are small

In the sections that follow, the available human data for several

of the most radiosensitive cancer types have been surveyed to deter- mine whether the data support additivity of effects when the doses are relatively low, fractionated or protracted In particular, the data for multiple myeloma are discussed below This cancer, which has been noted in a few occupational studies, is examined across the range of doses to determine whether its induction is more likely to appear a t low doses or dose rates I t should be noted when comparing human studies that differences in dose and dose rate are not the only confounders, other factors such a s the "healthy-worker effect" should also be considered

3.4.1 Human Studies of Cancer Risks porn Low Radiation Doses

3.4.1.1 Thyroid Cancer There are a number of epidemiologic studies of thyroid cancer following radiation exposures In most of these studies, acute thyroid doses of 0.5 Gy up to 10 Gy or more were involved The available evidence indicates that radiation exposure a t

focus of this review of epidemiological studies is on populations who received exposure before 20 y of age These risk estimates are shown

in Table 3.1

3.4 HUMAN STUDIES / 17 The cohort studies provide the most reliable estimates of risk because of their essentially complete ascertainment of thyroid cancer over time The screening studies should be used with some caution because of possible subject selection biases and incomplete ascertain- ment over time In addition, only two screening studies (Maxon et al.,

1980; Pottern et al., 1990)nhad screened comparison groups for use a s

a baseline For most screening studies there is, therefore, additional uncertainty in the expected number of cancers, since the estimates are from populations that did not have the screening I t should be noted that when the expected numbers are small, as in the screening studies i n Table 3.1, the uncertainty in the expected numbers makes the estimates of excess relative risk especially unreliable, more so than the estimates of absolute risk

A few studies have evaluated the effect of thyroid doses on the

order of 0.1 Gy These include two studies of children epilated with

x rays during treatment of ringworm of the scalp (Ron et al., 1989;

Shore, 1991') (see Table 3.1) The risk estimates in the larger study were a t least three times higher than those from other cohort studies

of thyroid cancer Whether this is due to unusual susceptibility within this population or to other factors is unknown The smaller study showed a n absolute excess risk about six times lower, but the difference in risk estimates between the two studies was only marginally significant (p < 0.10)

The thymus irradiation study reported by Shore (1989) included about 1,500 persons who received <0.5 Gy (mean dose of 0.18 Gy) There was a significant dose-response relationship, although it was based on small numbers (five irradiated cases versus five cases among the 4,800 unirradiated subjects) When the analysis was restricted to less than 0.3 Gy, however, the dose trend was still positive, but no longer statistically significant

The widespread use of 1311 for diagnostic purposes has provided several opportunities to examine thyroid cancer risk following expo- sure to protracted, low-dose rate radiation In the first, Holm et al

(1988) studied 35,000 persons to whom 13'1 was administered, of whom about 1,800 were under the age of 20 at the time of exarnina- tion The average thyroid dose was 0.5 Gy among adults and 1.6 Gy among children Among those whose initial diagnostic examination was not because of a suspected thyroid tumor, there was a deficit of thyroid cancers (Table 3.1) For those under age 20 a t 1311 exposure, two thyroid cancers were observed with 1.1 expected (not statistically

lunpublished data (Shore, R.E., Department of Environmental Medicine, New York University Medical Center, 1991)

3.4 HUMAN STUDIES / 21 significant) A second study, conducted by the Center for Devices

and Radiological Health (Hamilton et al., 1989), examined thyroid

nostic purposes Thyroid doses ranged from less than 0.1 to 20 Gy No significant excess of thyroid cancer was observed Finally, a German

study (Globel et al., 1984) also failed to find a significant dose-

response slope for thyroid cancer following diagnostic I3lI exposure

in nearly 14,000 patients

As part of the studies of the effects of fallout in Nevada and Utah fiom atmospheric nuclear weapons tests during 1951 to 1958, Rallison

disease in 2,600 children located downwind from the nuclear weap- ons test site in Nevada and another 2,100 from relatively unexposed children in Arizona With about 32 y of follow-up after the testing, only eight thyroid cancers were found, and the dose-response trend was not statistically significant Thyroid doses were estimated on the basis of residential history and milk-drinking status

Table 3.1 shows relative and absolute risk estimates for thyroid cancer from the major studies of external x-ray and internal 13'1

irradiation in juveniles The external x-ray and internal 1311 studies are different in dose rate and in the microdosimetric distribution of dose The absolute risk estimates for the external radiation studies are about 10 times greater than those for the 1311 studies If dose rate, and not microdosimetric considerations, is the main difference between 1311 and external thyroid radiation, then at face value lower- ing of the dose rate appears to substantially reduce carcinogenic

niles are very limited and caution should be exercised in drawing conclusions f?om them In particular, the I3lI results for those under age 20 are drawn from only six cases from the diagnostic series and

12 cases from the fallout series Furthermore, the fallout series are

of limited utility since only a small fraction of the Marshall Island

study have been statistically significant (p 5 0.05) only for neoplasms (Brillinger, 1992) The Utah fallout study was limited by the small number of subjects in the highest exposure group and the relatively short follow-up time in comparison to the long latency period for thyroid carcinoma

A few studies of thyroid cancer are available in relation to occupa-

tional radiation exposures Wang et al (1990) conducted long-term

studies of 27,000 diagnostic x-ray workers in China The observed RR

of 1.7 among the x-ray workers was not considered to be significantly elevated The doses to this population are not well characterized,

22 1 3 SCIENTIFIC BASES FOR COLLECTIVE DOSE

but before about 1960 the doses were sufficiently high to depress white cell counts; the mean dose has been estimated to be about

1 Gy (Wang et al., 1988)

Preliminary results of another occupational study have been

United States x-ray technologists who were queried by mail question- naire with subsequent medical documentation and radiation expo- sure estimates Preliminary analyses suggest that thyroid cancer incidence is significantly elevated (208 observed, RR = 2.31, but the dosimetry and full analyses have not yet been completed

Finally, Polednak (1986) reported a small, nonsignificant excess

of thyroid cancer among United States radium dial workers For this group, the radiation exposure to the thyroid was from a combination

of photon and alpha radiation and is estimated to have averaged about 0.7 Sv

In conclusion, several studies have indicated a risk of thyroid

radiation This risk appears comparable per unit dose to t h a t incurred following higher doses The limited evidence for thyroid cancer following acute doses in adulthood suggests that thyroid irra- diation has several times less effect in adults than in children

less carcinogenic than acute exposures, but this conclusion is based

irradiation to adults Attempts to compare effects within the same age group are confounded by the fact that the number of children treated with 1311 is small while few adults have received thyroid doses from external x-ray exposure The consistency among the occu- pational studies in showing suggestive elevations in risk reinforces the idea that the thyroid gland has a high relative risk for radiation induced cancer From the data available a t this time, additivity of

effects can prudently be assumed and, therefore, is not in conflict

with an assumption that collective dose calculations are appropriate

among Japanese atomic-bomb survivors have indicated that the dose-response curve is essentially linear (Tokunaga et al., 1987) (Table 3.2) The low-dose portion of the data set has also been exam- ined In the irradiated group with <0.5 Gy, 179 breast cancers were

dose-response trend was seen over the range 0.2 to 0.5 Gy, and the trend was suggestive, but not statistically significant for the dose range up to 0.2 Gy In a study of infants irradiated for purported enlarged thymus glands, breast cancer risk was assessed among

24 1 3 SCIENTIFIC BASES FOR COLLECTIVE DOSE

1,200 females (Hildreth et al., 1989) There was a significant dose-

response relationship which was consistent with linearity There was also a significant elevation of risk in the subgroup who received less than 0.5 Gy

Breast cancer incidence has been studied among women who received multiple chest fluoroscopic examinations during artificial

pneumothorax therapy for tuberculosis (Boice et al., 1991b; Hrubec

breast dose of 0.8 Gy from an average of 88 fluoroscopic examinations The risk estimate for this highly fractionated exposure protocol was comparable in magnitude to those from studies with unfractionated exposures (Table 3.21, and the dose-response relationship was well approximated by a simple linear model Similar results were found

in a large multiple fluoroscopy study of 31,700 women in Canada

(Miller et al., 1989) where it was reported that a linear dose-response

provided agood fit with a risk estimate comparable with other breast- irradiation studies An excess was evident over the entire dose range, including doses below 0.4 Gy

Women who received multiple diagnostic x rays for scoliosis early

in life have been followed to determine breast cancer risk (Hoffman

of 41 examinations, and a total dose to the breast estimated to

and a suggestive, but not statistically significant, dose-response trend was seen The relatively small size of this study limited its statistical power

In the group of Chinese diagnostic x-ray workers mentioned pre-

1990) The excess tended to be greater among those employed in the earlier years when the exposures were thought to be higher Breast cancer risk is also evaluated among 1,400 women who were

treated with 1311 for hyperthyroidism (Goldman et al., 1988) The

doses to the breast were not calculated, but probably averaged about 0.1 Gy The relative risk for breast cancer in the 1311-treated group was 1.2 (not statistically significant), based on 51 breast cancer cases

Two other studies (Hoffman and McConahey, 1983; Holm et al.,

1991) of breast cancer in women receiving similar doses of 1311 for

134 cancers, RR = 1.0; respectively) In a study of I3lI treatment for thyroid cancer by Edmonds and Smith (1986), 258 patients received

a mean breast dose of 1.0 Gy Six breast cancers were observed (RR = 2.4), which was a significant excess

In summary, there have been a number of studies of women who received breast irradiation from fractionated exposures, protracted

3.4 HUMAN STUDIES 1 25 exposures andlor relatively low doses These results are summarized

in Table 3.2 Most of the studies have either shown, or are consistent with, a significant increase in the risk of breast cancer resulting from irradiation of the breast The data are supportive of an interpretation that low doses or fractionated doses have a substantial degree of additivity of their effects upon breast cancer risk, although this seems less clear with exposures to 1311 The data for breast cancer from x-ray irradiation, perhaps more than any other human cancer, supports the notion of a dose-response relationship that is a linear function of dose and largely independent of dose rate or fractionation, thus supporting the concept of collective dose Animal data on mam- mary cancer (see Section 3.3.1) have shown a linear-quadratic dose- response function and a compatibility between the linear slope and the results with fractionated exposures

ated the Japanese atomic-bomb data, and to a lesser extent the ankylosing spondylitis data, for leukemia and concluded that the dose-response curve was best fit by a linear-quadratic model Since the quadratic component becomes negligible a t low doses, the magni- tude of the linear component is critical in defining risks a t the low-dose levels used in calculation of collective dose The linear com-

small range for those with radiation exposure a t ages 20 or older

dose and were followed for 25 y, their estimated leukemia risk would

be increased to 0.27 percent from the spontaneous or "background" rate of 0.21 percent For a 0.01 Gy dose, the estimated risk would increase to only 0.215 percent Particularly in the latter case, it would bevery difficult to detect an increased risk in an epidemiologic study

A number of studies of leukemia have been reviewed to determine

if fractionated exposures or low-dose rate exposures appear to show significant leukemia risks (Table 3.3) Studies with a high, acute dose to the irradiated volume (e.g., cancer radiotherapy studies) were excluded from consideration

Five case-control studies have been reported of (mostly) adult leu- kemia in relation to diagnostic radiation or fallout from atmospheric weapons tests (Table 3.3) (Gibson et al., 1972; Gunz and Atkinson, 1964; Stewart et al., 1962; Linos et al., 1980; Preston-Martin et al., 1989) Three were marginally positive and the other two were nega- tive These studies should be interpreted cautiously because of the inherent potential for various biases (e.g., reporting bias) A case- control study of leukemia in relation to fallout radiation showed a

32 1 3 SCIENTIFIC BASES FOR COLLECTIVE DOSE

3.4 HUMAN STUDIES 1 33

mia in high- and low-background radiation areas in China (Wei, 1980) showed no differences Caution should be exercised in interpre- ting these studies because of differences in the two populations The three studies of 1311 treatment for hyperthyroidism summa- rized in Table 3.3 (Hoffman et al., 1982; Holm et al., 1991; Saenger

of 1311 treatment for thyroid cancer (Edmonds and Smith, 1986) had

a much higher bone-marrow dose and was suggestively positive A large study of persons administered diagnostic 1311 (Holm et al., 1989) was also positive although the dose to the bone marrow was relatively low Other medical irradiation studies with low dose or protracted

Inskip et al., 1990a; Spengler et al., 1983)

Studies of patients administered the radiologic contrast medium Thorotrast reveal a definite leukemogenic effect, which is not surpris- ing since the protracted bone-marrow irradiation is mainly by high-

1992; Mori et al., 1983; Olsen et al., 1989; van Kaick et al., 1989)

On the other hand, radium dial painters have not shown a n excess

even though the marrow doses are believed to have been appreciable and a substantial fraction of the dose was attributable to alpha radiation

Several studies of radiologists or x-ray workers with relatively

1975; Smith and Doll, 1981; Wanget al., 1990) The studies ofpersons with lower doses did not show a significant excess, although they

and Miller, 1978)

The final 18 citations in Table 3.3 are studies in the nuclear power

or nuclear weapons industries By and large, the bone-marrow doses

in these studies are low as well as highly fractionated Some of the occupational cohorts were exposed to internal radionuclides as well

as external radiation, as noted in Table 3.3 Although only one of the 18 nuclear industry studies shows a significant overall excess of leukemia mortality among the radiation workers, the healthy worker effect makes such overall comparisons difficult to interpret Also, in most of these studies, only a small fraction of the workers were regularly exposed to doses substantially above background, and thus such overall comparisons are greatly diluted

disease in these populations is to conduct dose-response analyses Ten of the studies had individual dosimetry data, and conducted

34 / 3 SCIENTIFIC BASES FOR COUECTIVE DOSE

such analyses Of the individual studies, only the study of Sellafield workers showed evidence of a statistically significant correlation of leukemia with external radiation exposure However, a s indicated

by confidence limits and trend test results, the possibility of positive risks could not be excluded

The most informative assessment of radiation risks based on nuclear worker studies is that obtained from analyses of combined data from several studies International analyses including data from t h e United States, the United Kingdom and Canada were recently conducted to evaluate the risk for leukemia, total solid cancers, and a number of specific cancer sites (Cardis et al., 1995)

I t included some 95,000 workers with a collective dose of 3,843 person-Sv The study included detailed consideration of the compara- bility of dosimetry across studies and over time For leukemia the estimate of excess relative risk was 218 percent Sv-' (95 percent

linear excess relative risk estimate for leukemia of 370 percent Sv-' for males exposed in adulthood The risk estimate from this pooled analysis was thus positive, but somewhat lower than the linear estimate based on the atomic-bomb survivors I t is also notable that the upper bound on the confidence interval excludes a risk more than twice a s great as the atomic-bomb estimate, which indicates that high-risk estimates for low dose and low-dose rate irradiation are not plausible

exposures than following high dose and high-dose rate exposures This concept arose originally Gom the excess seen at Hanford (Gilbert

that among the Japanese atomic-bomb survivors Table 3.4 summa- rizes the available data on radiation and multiple myeloma

low-dose studies (with primarily low-LET radiation) with sugges- tively elevated multiple myeloma rates (e.g., United States radiolo- gists, Hanford workers and Sellafield workers), there are a number

of others which show no excess Several of the high-dose studies

(e.g., the Japanese atomic-bomb survivors, and patients irradiated

for uterine bleeding or ankylosing spondylitis) also show suggestive excesses a s do several of the studies with high-LET internally depos- ited alpha emitters (radium dial painters and Thorotrast patients) Hence, there is no clear pattern of greater or lesser risk per unit dose from acute versus protracted, or high-dose versus low-dose, radiation While the data for myeloma do not argue against the