NCRP report no 135 liver cancer risk from internally deposited radionuclides

Risk assessment includesnot only an understanding of dose to a tissue, but also an appreci-ation of biological factors that may impact cancer frequency, such as sensitivity to radiation-

NATIONAL COUNCIL ON RADIATION

PROTECTION AND MEASUREMENTS

Issued March 9, 2001

National Council on Radiation Protection and Measurement s

7910 Woodmont Avenue / Bethesda, Maryland 20814-3095

This Report was prepared by the National Council on Radiation Protection and Measurements (NCRP) The Council strives to provide accurate, complete and use- ful information in its documents 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 respect to the accuracy, completeness or usefulness of the information contained in this Report, or that the use of any infor- mation, method or process disclosed in 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 VII) or any other statutory or common law theory govern- ing liability.

Library of Congress Cataloging-in-Publication Data

National Council on Radiation Protection and Measurements.

Liver cancer risk from internally-deposited radionuclides : recommendations of the National Council on Radiation Protection and Measurements.

p ; cm (NCRP report ; no 135)

“Issued March 2001.”

Includes bibliographical references and index.

ISBN 0-929600-68-1 (alk paper)

1 Liver Cancer 2 Contrast media Carcinogenicity 3 Radiation

carcinogenesis I Title II Series

[DNLM: 1 Liver Neoplasms etiology 2 Radiation dosage 3 adverse effects 4 Risk Assessment WI 735 N277L 2000]

For detailed information on the availability of NCRP publications

see page 83.

Preface

This Report updates the liver cancer risk of Thorotrast® inhumans Thorotrast®, a radiographic contrast medium containingnaturally occurring radionuclides of thorium was widely used inthe first half of the twentieth century An increased incidence ofliver cancer in this population of patients has been known for sometime Utilizing data on the liver cancer risk of Thorotrast® and oflow-LET radiations in animals, the liver cancer risk of low-LETradiation in humans is also estimated

This Report was prepared by Scientific Committee 57-10 onLiver Cancer Risk Serving on Scientific Committee 57-10 were:

Co-Chairmen

University-Tricities Salt Lake City, Utah

Richland, Washington

Members

Colorado State University Instituts des Klinikums

Rhein, Germany

Institut für Nuklearmedizin Institut für Radiologie und

Krebsforschungszentrum Heidelberg, Germany

Heidelberg, Germany

NCRP Secretariat

E Ivan White, Senior Staff Scientist (1979–1993)

Thomas M Koval, Senior Staff Scientist (1993–1999)

Lynne A Fairobent, Staff Scientist (1999–2000)

Cindy L O’Brien, Managing Editor

The Council wishes to express its appreciation to the Committeemembers for the time and effort devoted to the preparation of thisReport

Charles B Meinhold

President

Contents

Preface iii

1 Overview 1

2 Affinity of Radionuclides for Liver Tissue 3

2.1 Characteristics of Liver Tissue 3

2.2 Affinity of Radionuclides in Ionic or Compound Forms in Various Tissues 3

2.3 Distribution of Radionuclides in Colloidal or Particulate Form 8

3 Effective Half-Life 10

4 Radiation Dose from Internally-Deposited Radionuclides 12

5 Radiosensitivity of Liver Tissue 14

5.1 Sensitivity to Morphological Changes 14

5.2 Sensitivity for Cell Killing 15

5.3 Influence of Latent Period on Cancer Risk 15

5.4 Sensitivity for Cancer Induction 16

6 Physical Variables that Alter Response 18

6.1 Cellular and Tissue Dose Distribution 19

6.1.1 Influence of Dose Distribution at the Cellular Level 19

6.1.2 Influence of Dose Distribution on Liver Cancer 20

6.2 Influence of “Particle Loading” on Cancer Induction 23

6.3 The Role of Injury on Cancer 23

6.4 The Role of Mechanistic Studies on Liver Cancer Risks 24

7 Use of Experimental Animal Data 28

7.1 Relative Biological Effectiveness of the Radiation Emitted by Radionuclides 28

7.2 Liver Cancer Risk from Animal Data 28

8 Liver Risk Estimates in Humans 33

9 Uncertainty in Current Risk Estimates 48

9.1 Strengths of the Data 48

9.1.1 Identification of the Thorotrast® Population 48

9.1.2 Defined Dose and Response Relationships 49

9.1.3 Transfer of the Information to Other Populations 49

9.2 Weaknesses of the Data 50

9.2.1 Dose Range and Dose Used in Risk Estimates 50 9.2.2 Uncertainty Related to “Wasted Dose” Estimates 50

9.2.3 Uncertainty Associated with Dose Distribution 52 9.2.4 Uncertainty Associated with Extrapolations from Animal Data 52

9.2.5 Uncertainty Related to Environmental Insults 53 10 Summary 55

References 56

The NCRP 74

NCRP Publications 83

Index 93

1 Overview

This Report provides an update of the cancer risk from clides deposited in the liver The liver has been considered an organwith a low risk for cancer induction from ionizing radiation (ICRP,1991; NCRP, 1993a) This may in part have been because of thelong latency period required to detect increases in radia-tion-induced liver cancer Other estimates have increased the risk

radionu-of liver cancer to a value radionu-of 300 cancers 10–4 Gy–1 (NAS/NRC, 1988;UNSCEAR, 1994) This Report provides a re-evaluation of themolecular, cellular, experimental animal and human liver cancerdata, and an update of the risk of liver cancer from inter-nally-deposited radionuclides

To determine risk, it is first essential to calculate the radiationdose to the liver This is dependent on the affinity of the radionu-clides for hepatic tissue, the radionuclide’s chemical form, the LET(linear energy transfer) of the emitted radiation, and the radionu-clide’s physical and biological half-lives Risk assessment includesnot only an understanding of dose to a tissue, but also an appreci-ation of biological factors that may impact cancer frequency, such

as sensitivity to radiation-induced cell killing and the presence

of liver disease such as necrosis, fibrosis and cirrhosis Otherbiological factors, such as sex, age at exposure, and exposure toother environmental insults, can also alter the sensitivity forradiation-induced liver cancer Obviously, variations in any ofthese relationships will have a significant influence on the risk

of radiation-induced liver cancer and the uncertainty associatedwith such risk Each of these physical and biological variables areconsidered in this Report

For many types of radiation exposures, e.g., chronic exposure to

low-LET radiation, there are no statistically significant humandata (UNSCEAR, 2000; Volume 2, Table 9) Therefore, animal andcellular data must be used for extrapolation of radiation risk tohumans Liver cancer induced in experimental animals by inter-nally-deposited radioactive materials can be used to estimatehuman cancer risks This is done by determining the relative bio-logical effectiveness (RBE) for liver cancer in animals following

exposure to both high- and low-LET radiation This informationcan be used to extrapolate to liver cancer risk in humans.

Of principle concern, in relation to the risk of liver cancer frominternally-deposited radioactive material, are the radionuclideswhich concentrate in the liver and emit alpha particles The majorsource of information on such human liver cancer risk is frompatients injected with a thorium-based contrast media,Thorotrast®.1 These human data are supplemented with animaland cellular data to solve problems associated with usingThorotrast® data as the basis for human liver cancer risk and forthe extrapolation of these risks to other radionuclides Some of theproblems that are addressed in this Report, using animal data,include nonuniform distribution, potential chemical toxicity, dis-ease, and interaction of other biological factors in the cancer induc-tion process Animal studies support the validity of using thehuman Thorotrast® cancer data as a model for cancer risk induced

by other internally-deposited alpha-emitting radioactive materials

(Brooks et al., 1983; Gilbert et al., 1998; Muggenburg et al., 1996; Taylor et al., 1993) This Report updates the risk estimates derived

by the Committee on Biological Effects of Ionizing Radiation [BEIR(NAS/NRC, 1988)] and the United Nations Scientific Committee onthe Effects of Atomic Radiation [UNSCEAR (1994)] The liver can-cer risk for alpha emitters in this Report is calculated to be

560 ± 95 cases 10–4 Gy–1.Extrapolation from animal data makes it

possible to estimate the risk for human liver cancer from

pro-tracted exposures to beta/gamma emitters as 15 to 40 liver cancersper 104 people per gray The uncertainty associated with these riskestimates is high because of the need to extrapolate between differ-ent types of radionuclides, different species and from very high tolow levels of exposure In all these extrapolations, we have used alinear no-threshold model Even with these uncertainties, thisReport concludes that the liver is not a radio-resistant organ

1 Thorotrast® (van Heyden Company, Dresden-Radebeul, Germany) is

a radiographic contrast medium comprised of a 25 percent colloidal tion of yellow dextrin and thorium dioxide It has a mean particle diame-

solu-ter of 9.3 ± 4.3 microns (Riedel et al., 1983) containing 228Th/232Th in a ratio of 0.4 in freshly obtained thorium preparations from natural tho-

rium (van Kaick et al., 1984a) Thorium-228 and its daughter products are

responsible for the majority of the radiation dose from Thorotrast®.

2 Affinity of

Radionuclides for

Liver Tissue

2.1 Characteristics of Liver Tissue

The liver is an organ involved in many complex ships in the body It functions in a large number of endocrine, exo-crine and regulatory capacities It is a major tissue in themononuclear-macrophage system (Roser, 1979; Stuart, 1970) It is

inter-relation-a site of minter-relation-any metinter-relation-abolic inter-relation-and detoxificinter-relation-ation functions inter-relation-and inter-relation-a depot

for iron and a number of trace metals (Burch et al., 1978) In

addi-tion, it is the major production site for plasma proteins involved inthe transport of numerous hormones (Madden and Zeldis, 1958),

lipids, metals, etc (Boocock et al., 1970; Bruenger et al., 1971; Jacobs and Worwood, 1978; Stover et al., 1972) Transferrin, a pro-

tein of liver origin, is the primary iron transport protein and plays

an important role in the hepatic localization of many actinide

ele-ments that are of special concern in this Report (Boocock et al., 1970; Bruenger et al., 1971; Jacobs and Worwood, 1978).

2.2 Affinity of Radionuclides in Ionic or

Compound Forms in Various Tissues

The deposition and retention of radioactive materials in theliver is dependent on both their physical and chemical form Based

on metabolic studies in animals, a number of radionuclides(238/239Pu, 241Am, 252Cf, 144Ce, 210Po, 91Y, 90Y), including some oftheir decay products, are likely to have an affinity for human

hepatic tissue (Brooks et al., 1982; Bruenger et al., 1972; 1976; Durbin, 1972; 1975; Durbin et al., 1985; Lloyd et al., 1972; 1984; Moskalev et al., 1980; Muggenburg et al., 1984; Stover et al., 1971; 1972; Taylor et al., 1993)

Many studies have supported the affinity of the liver for manyradionuclides Well established human data demonstrate that plu-tonium and americium, which move from the site of entry by way

of the vascular system, are retained in the liver in relatively highconcentrations and for prolonged periods of time (Breitenstein

et al., 1985; Foreman et al., 1959; Lagerquist et al., 1969; Magno et al., 1967; McInroy, 1976; McInroy et al., 1985; Palmer

et al., 1985) For example, distribution data derived at the time of

death for five workers who inhaled a mixture of 239Pu and 241Am

provided the following data (McInroy et al., 1989) It was

deter-mined that for 239Pu the liver retained 35.4 ± 13 percent and theskeleton 53.7 ± 12 percent of the systemic body burden From thesedata, the investigators determined that the systemic distribution ofthe material was consistent with the 30:50 division betweenliver and skeleton proposed in International Commission on Radio-logical Protection (ICRP) Publication 30 (ICRP, 1979) andPublication 48 (ICRP, 1986) In three terminally ill patients givensmall amounts of plutonium intravenously, the average organ dis-tribution 5 to 15 months following injection was 31.2 percent inliver and 47.5 percent in skeleton The smaller mass of the liver rel-ative to the skeleton results in a significantly higher concentrationand dose to liver In occupationally exposed workers, McInroy(1976) observed that the distribution of plutonium between liverand skeleton was 30 ± 23 percent and 68 ± 24 percent, respectively

To estimate dose in nonoccupationally exposed humans, he notedthat the concentration in the liver (0.021 Bq kg–1) was higherthan that detected in the vertebrae (0.0081 Bq kg–1) (McInroy,

1976) These data are consistent with that of Fisenne et al (1980)

who found 0.013 Bq kg–1 in the liver and 0.0093 Bq kg–1 in the eton Two healthy humans were injected with tracer levels of 239Puand the levels of activity in the liver and skeleton estimated (Talbot

skel-et al., 1993) From these studies it was dskel-etermined that the liver

retained about 55 to 68 percent of the injected activity This ishigher than the 45 percent used in setting annual limits ofintake for plutonium (ICRP, 1986) A person occupationallyexposed to plutonium over a 12 y period, primarily by inhalation,had average concentrations of 165 Bq kg–1 (9.9 dis min–1 g–1) inliver and 23 Bq kg–1 (1.4 dis min–1 g–1) in skeleton (Foreman et al., 1959) Lagerquist et al (1969) observed a similar distribution with

measured concentrations of 53 Bq kg–1 (0.32 dis min–1 g–1) in liverand 2.1 Bq kg–1 (0.13 dis min–1 g–1) in skeleton in a person whoreceived two contaminated puncture wounds and several inhala-

tion exposures during the 9 y prior to death Magno et al (1967)

found the vertebral concentration to be about 10 percent of that ofthe liver in nonoccupationally exposed persons whose plutoniumwas presumably received from fallout In German fallout studies, a

2.2 AFFINITY OF RADIONUCLIDES IN IONIC OR COMPOUND FORMS / 5

concentration of 19.6 mBq kg–1 (530 fCi kg–1) of 239/240Pu wasobserved in the liver versus 3.4 mBq kg–1 (92 fCi kg–1) in a vertebra(Bunzl and Kracke, 1983), again, suggesting that per unit weightthe liver had the highest concentration and would receive the high-est radiation dose and dose rate On the basis of studies of southern

Finns, Mussalo et al (1980) estimated dose rate received from

fall-out of 0.3 µGy y–1 for liver and 0.04 µGy y–1 for skeleton Fromthese data, it is evident that following contamination, the pluto-nium concentration in the liver exceeds that of the skeleton by afactor of 5 to 10, especially at long times after exposure By con-trast, one case from the U.S Transuranium Registry demonstratedthat in an accidental inhalation exposure, the distribution of

241Am was appreciably less in liver (6.4 percent) than in skeleton

(82 percent) (McInroy et al., 1985) For 241Am, McInroy et al (1989)

determined that at the time of death, the liver had only 6.5 ± 4.8percent and the skeleton had 73.5 ± 12.4 percent of the systemicbody burden These data all suggest that the liver is a major site ofradiation dose from internally-deposited actinides, and its carcino-genic sensitivity should be evaluated

Using the retention and distribution data from animals to dict the behavior of nuclides in human tissues is attractive; how-ever, this has not been completely successful The problems inusing animal data to determine the behavior of nuclides in humans

pre-is related to (1) the different ways nuclides are dpre-istributed andretained by a single species, and (2) the different ways speciesretain and distribute the same nuclide For example, in beagle dogsthere is a wide variation in the percent liver retention for severalactinide elements following both inhalation and intravenous

administration (Durbin and Schmidt, 1985; Guilmette et al., 1994; Hickman et al., 1995; Lloyd and Mays, 1975; Muggenburg et al., 1986; 1996; Stover et al., 1971; 1972) A between-species compari-

son for a single nuclide shows that retention is very species dent Comparing the retention and distribution of 241Am citrate inhumans and dogs following a single intravenous injection illus-trates the impact of species In dogs, the percent of the injectedactivity retained in the liver was appreciably higher than that ofthe skeleton, 50 percent versus 30 percent, respectively Thus, itdoes not appear feasible to predict the relative liver retention ofvarious actinides in humans by direct extrapolation from animalstudies However, this Report does use animal data to evaluate theinfluence of many variables including dose distribution, particleloading, liver injury, and radiation type on liver cancer risk

depen-In spite of these problems, it is necessary to extrapolate fromanimal data when there is little human information on the reten-tion and distribution of radionuclides such as 144Ce-144Pr This isdifficult, as previously explained, because of the differencesbetween species The major source of distribution data that hasbeen extrapolated to humans is the data from the beagle dog(because of its long life and of the beagle’s retention patterns beingsimilar to humans for many radionuclides) For the beagle dog,long-term retention of 144 Ce was about 60 percent in the liver and

40 percent in the skeleton (Boecker and Cuddihy, 1974; Hahn et al., 1996; Stuart et al., 1964) From these data, models were developed

for estimating the retention and distribution of 144Ce in humans Itwas calculated that 30 percent of the deposited activity thatreached the blood would be translocated to the skeleton and 25 per-cent to the liver (NCRP, 1978) In a review of the distribution of

144Ce in man, it was concluded that, “In the absence of data on liverclearance in man, it is considered appropriate to use values derivedfrom studies in dogs wherein the retention of 144Ce was indistin-guishable from its physical half-life This indicates a biologicalhalf-time in man of thousands of days” (NCRP, 1978) Similarobservations were made for the Chinese hamster, regardless of theroute of administration of 144Ce-144Pr For this nuclide, about 70percent of the body burden was retained in the liver and 30 percent

in the skeleton (Sturbaum et al., 1970) Other lanthanides were

evaluated and found to result in less liver deposition and to have ashorter liver retention time than 144Ce (Durbin et al., 1956)

For the alkali metals such as 137Cs, the biological retention wasshort and distribution rather uniform Studies in the beagle dogshowed that 137Cs was uniformly distributed, that is, the concen-tration in the liver was about the same as the average for the body

(Nikula et al., 1995) A rather complete review of the deposition,

retention and distribution of fission products has been published byDurbin (1975) and reviewed in detail by Stannard (1988) Table 2.1illustrates the partitioning of several radionuclides between theliver and the bone where these were the primary organs at risk.Where the effective half-lives are very different for the two organslisted, it might be that the time of sacrifice is not constant whichmay influence the distribution in some cases

2.2 AFFINITY OF RADIONUCLIDES IN IONIC OR COMPOUND FORMS / 7

TABLE 2.1—Partitioning of ionic radionuclides between the liver

and bone in different species.

Nuclide Species

Distribution of Nuclide (%) ReferenceLiver Bone

Pu-239 Human 39 61 ICRP (1979)

Pu-239 Human 31 69 McInroy (1976)

Pu-239 Human 43 57 Durbin (1975)

Pu-239 Human 20 80 Fisenne et al (1980)

Pu-239 Human 55–68 32–45 Talbot et al (1993)

Pu-239 Primate 61 39 Durbin (1975)

Pu-239 Primate 19 81 Brooks et al (1992)

Pu-239 Primate 36 64 LaBauve et al (1980)

Pu-239 Dog 50 50 Muggenburg et al (1996)

Pu-239 Rabbit 69 31 Durbin (1975)

Pu-239 Rat 15 85 Durbin (1975)

Pu-239 Deer mouse 45 55 Taylor et al (1981)

Am-241 Dog 50 50 Taylor et al (1992)

Am-241 Dog 50 50 Stover et al (1971)

Am-241 Grasshopper

mouse

64 36 Taylor et al (1981)

Am-241 Deer mouse 61 39 Taylor et al (1981)

Ce-144 Dog 60 40 Hahn et al (1996)

Ce-144 Dog 60 40 Boecker and Cuddihy

(1974) Ce-144 Chinese

hamster

70 30 Sturbaum et al (1970)

Cs-137 Dog 50 50 Nikula et al (1995)

Y-91 Dog 25 75 Muggenburg et al (1998)

2.3 Distribution of Radionuclides in Colloidal

or Particulate Form

In contrast to radionuclides in ionic and compound form, dal or particulate substances such as Thorotrast® have a highaffinity for the reticulo-endothelial system This includes the liver,spleen, lymph nodes, and bone marrow This affinity is independent

colloi-of the radioactive properties colloi-of the nuclide in the colloidal Thereare extensive studies on the retention of particles by the liver

(Brooks et al., 1974) and by colloids like Thorotrast® (Dalheimer

et al., 1995; Spiethoff et al., 1989; van Kaick et al., 1978),

Hafnotrast® (Riedel et al., 1979; 1983), and Zirconotrast® (Riedel

et al., 1979; Spiethoff et al., 1989).

To evaluate the impact of particle loading and specific activity

on liver cancer induction from Thorotrast®, two other colloids,Hafnotrast® (Riedel et al., 1979; 1983), and Zirconotrast® (Riedel

et al., 1979; Spiethoff et al., 1989) were developed The Hafnotrast®

represented a colloid of a heavy metal, the Zirconotrast® a metalthat was not radioactive The retention data of Thorotrast®,Hafnotrast®, and Zirconotrast® are similar for different species.This made it possible to estimate the behavior of other radioactivecolloids in humans from animal data These extrapolations indicatethat for all of these colloids more than 50 percent of the adminis-tered activity is retained in the liver This results in a high dose tothe liver These animal data also made it possible to determine theinfluence of particle loading and specific activity on cancer inci-dence The particle loading had only a minor impact on cancerincidence, which indicated that extrapolation from animal data tohuman data seems to be possible (Table 2.2)

The same patterns of aggregation are valid for Thorotrast® inboth experimental animals and man This process was studiedwith both light and electron microscopy (Tessmer and Chang,1967) Three distinct patterns of aggregation were reported: homo-geneous distribution (up to four months in man, up to 3 d in labo-ratory animals), formation of aggregates (up to 7 y in man, up to

33 d in laboratory animals), and formation of larger conglomerates(up to 16 y in man, up to 185 d in laboratory animals)

The nonuniform distribution of the activity may cause ical problems In rats, liver fibrosis was not marked, resulting in amore uniform dose distribution and a higher dose per unit of activ-

patholog-ity (Spiethoff et al., 1989) However, in man, fibrosis develops in the

portal tract of the liver at about 7 y after injection Later, the sis becomes widespread This process results in a sequestrating ofmajor Thorotrast® conglomerates leading to a marked dose reduc-tion to the epithelial tissue per unit of activity in the organ

Okazima (1973) Rat 37–75 nda 3.5–19 Riedel et al

(1979) Rabbit 54.1 <43.7 3.2 Kaul (1969) Dog 66.2 nda 4.1 Riedel et al

(1979) Chinese

3 Effective Half-Life

The effective half-life is a combination of the physical half-lifeand the biological half-life Using effective half-life combined withdistribution and retention makes it possible to calculate both doseand dose rate to the tissues of interest

The retention half-time of plutonium in the liver varies ciably among species It is only several weeks in most laboratorymice and rats (Belyayee, 1969; Rosenthal and Lindenbaum, 1969;

appre-Smith et al., 1978; Tseveleva and Yerokhin, 1969) However, in

some rodents, such as Chinese hamsters, grasshopper mice, and

deer mice, retention is prolonged (McKay et al., 1972; Taylor et al.,

1981) Effective half-lives for liver retention have been reported to

be about 180 d in Rhesus and Cynomologus monkeys and in

baboons (Bair et al., 1974; Brooks et al., 1992; Metivier et al., 1974) Durbin et al (1985) summarized the nonhuman primate retention

data and found similar results The effective half-lives are shorter

than those found in dogs (Bair et al., 1974; Park, 1975; Taylor et al.,

1985; 1991) In beagle dogs, given a single injection of 239Pu citrate

of 0.6, 1.8, or 3.6 kBq kg–1 (0.016, 0.048, or 0.096 µCi kg–1),

reten-tion half-times of about 10.3 y were observed (Stover et al., 1971).

The half-time of plutonium in the human body, following nous injection, has been estimated to be very long, 204 y (Durbin,1972)

intrave-It was speculated that 239Pu would be lost from the liver more

slowly than from the skeleton (Durbin, 1972; Durbin et al., 1985; Langham et al., 1950; 1980) The importance of the relationship

among half-times in different tissues is that material from thetissue with a shorter half-time can be translocated to and build

up in tissues with longer retention times This was observed

follow-ing inhalation of plutonium in monkeys (Brooks et al., 1992) and

dogs (Diel and Lundgren, 1982; Mewhinney and Diel, 1983;

Muggenburg et al., 1996; Park et al., 1986) In these animals, the

longest retention time is in the bone As the plutonium is lost fromthe lung and liver, it accumulates in the bone

In interpreting human retention curves, one should be mindfulthat some of the very limited human data were derived from seri-

ously ill people (Langham et al., 1950; 1980) The health status, the

of radionuclides from the liver For example, Stover et al (1971)

has shown that injury from high levels of 239Pu increase the rate

of excretion from the livers of beagles It has also been shown inbeagle experiments that 241Am retention in hepatic tissue isreduced by a factor of approximately six following long-term con-sumption of dietary ethanol Ethanol reduced the radiation dose

through cell injury and death (Taylor et al., 1992) This change in

clearance was related to liver damage In these animals, the cancerrisk was increased, even though the liver dose was decreased Sinceethanol is a ubiquitous part of the human diet, it is important toevaluate its impact on both the dose and the induction of cancer Damage to the liver generally tends to increase excretion rates.Increased excretion rates result in a decrease in cumulative doseper unit of radionuclide deposited If the retention time is overesti-mated in diseased humans using the 40 y standardized retentionfunction (ICRP, 1972), this would increase the calculated lifetimedose above the real dose in a diseased subpopulation The risk iscalculated from the slope of the dose-response relationship Thebiological response or cancer frequency is determined and defined

in any study population Since the dose per unit of deposited ity in a diseased person would decrease and the cancer frequency isconstant, the slope of the dose-response relationship wouldincrease and suggest that the risk would also increase This makes

activ-it important to understand the retention time when performingdose and risk calculations

of the dose to the cells.

If there is a nonuniform distribution of activity as a function

of the physical or biological properties of the radionuclide or tissue,the dose distribution can result in a wide range of cellular doses.For example, for many internally-deposited radioactive materials,the deposition is nonuniform This makes it important to includelocalized distribution, the energy of the emissions, and the fraction

of the energy absorbed in the cells at risk to calculate local or lar dose The mean cellular dose represents the cellular dosecalculated locally in the region of potential target cells for tumor

cellu-development and has been calculated for liver (Dalheimer et al.,

1995)

The usefulness of cellular or local dose has come into questionwith the observations of the bystander effect In studies on thebystander effects, it has been demonstrated that cells not traversed

by alpha particles can be injured, and result in changes in

gene expression (Azzam et al., 1998; Deshpande et al., 1996), mosome aberrations (Nelson et al., 1996), sister chromatid

chro-exchanges (Lehnert and Goodwin, 1997; Nagasawa and Little,

1992; Nagasawa et al., 1990), genomic instability (Seymour and Mothersill, 1997), and cell transformation in vitro (Miller et al.,

1999) The importance of bystander effects in risk assessment hasnot been established, but it may be that total dose to the tissue andnot local dose is the dose of interest in risk assessment This

is especially true for alpha-emitting radionuclides, where therange of the particles in tissue is very short For Thorotrast®, there

is a very nonuniform distribution of the radionuclide resulting in

4 DOSE FROM INTERNALLY-DEPOSITED RADIONUCLIDES / 13

nonuniform dose (Kaul, 1973a; 1973b; Kaul and Noffz, 1978; Kaul

et al., 1979) This is further complicated by the fact that

Thorotrast® particles measure about 5 nm in diameter, andthrough biological processes they accumulate to form conglomer-ates in the liver This process results in large deposits ofThorotrast® in localized regions with much of the alpha-particleenergy being absorbed in the conglomerate This decreases the doseper unit of deposited activity to the target cells at risk forliver tumor development, since many cells would be outside thealpha-particle range However, bystander effects may makethe range of biological influence of the alpha particles much largerthan their physical range

Extensive research has been conducted to characterize the ation dose to the liver cells following the injection of Thorotrast®.The average injection was about 25 mL, which contained approxi-mately 5 g of thorium (22.2 kBq) 232Th, plus its progeny This level

radi-of activity resulted in an average dose rate radi-of about 0.25 Gy y–1 tothe liver of a standard 70 kg man This dose was made assuming aliver retention of 60 percent and that 65 percent of the alpha-particle energy escapes from the Thorotrast® aggregates and isabsorbed in tissue (Kaul, 1973b; Kaul and Noffz, 1978; van Kaick

et al., 1978)

Integration of the energy deposition with time results in themean organ dose For the liver, estimated mean organ doses arehigher by at least one order of magnitude than the mean cellulardoses assessed However, this local dose was calculated by deter-mining the quantity of Thorotrast® particles outside of areas withlarge conglomerates where self-absorption has a major impact Thenumber of these small particles, which yield a more uniform distri-bution of dose, was higher by a factor of 10 when compared to thenumber in tissues without conglomerates The mean organ dose is

an under representation of the dose from the small-size particleswhich may be the dose of importance for estimation of risk for thedevelopment of liver cancer However, the total mean organ dose isused in this and other reports on liver cancer risk estimates(NAS/NRC, 1988; UNSCEAR, 1994), since it is not possible to sep-arately determine the contribution of the localized nonuniform anduniform dose to the risk of liver cancer

essen-5.1 Sensitivity to Morphological Changes

A number of early studies involving both animals and humanssuggest that at short times after large acute low-LET radiationexposures, minimal changes in liver morphology, atrophy, cell kill-ing, and cell proliferation were induced (Case and Warthin, 1924;

Furth and Upton, 1953; Gershbein, 1956; Ingold et al., 1965;

Loletsky and Gustafson, 1952; Patt and Brues, 1954; Weinbren

et al., 1960) The life span of the hepatocyte is long (MacDonald,

1961), and at least some of the acute injury appears to be “masked”

by this low cellular turnover rate This was demonstrated in mice,where pleomorphism and atypical liver-cell nuclei were seen in

only about 13 percent of neutron-irradiated mice, when the cells

were not stimulated to divide In contrast, pleomorphism and ical liver-cell nuclei were present in nearly all of the liver cells inmice when cell proliferation rate was increased by injection of car-bon tetrachloride (Cole and Nowell, 1964) Since carbon tetrachlo-ride does not produce chromosome aberrations, most of theabnormal cells observed after the chemical exposure would bethe result of formation of chromosome aberrations and micronucleithat are not expressed until the damaged cells were forced todivide These data demonstrate that cell proliferation allowsthe cells to express the damage induced by the radiation exposure

atyp-5.3 INFLUENCE OF LATENT PERIOD ON CANCER RISK / 15

The carbon tetrachloride promotion also significantly increased theincidence of primary liver tumors

5.2 Sensitivity for Cell Killing

Cell survival studies have shown that hepatocytes are

radiore-sistant to cell killing (Jirtle et al., 1990) However, in the liver there

is little evidence for repair of radiation-induced potentially lethal

damage (Jirtle et al., 1981) or chromosome damage (McKay et al.,

1974) Thus, little repair or increased survival is found using tionation or protracted exposure Further evidence of radioresis-tance was the presence of normal regeneration in rat’s liver at x-raydoses up to approximately 200 Gy (Gershbein, 1956) There areconsiderable data suggestive of radioresistance in terms of cell kill-ing when evaluations are made at early post-exposure times How-ever, in chronic toxicity studies, in which late effects such asneoplasia are the principal endpoints, hepatic tissue does not

frac-appear to be highly radioresistant (Jirtle et al., 1990; Taylor et al.,

1985)

5.3 Influence of Latent Period on Cancer Risk

Data from laboratory studies on cancer induction in rodentsfrom internally-deposited radioactive materials includingThorotrast®, show very long latent periods, defined as the timebetween the injection of the material and the time that the firstliver tumors appeared In the experimental animal studies, it is dif-ficult to determine the proper time interval to be used for dose cal-culations The radioactive material is injected and after a variableperiod of time there is a change in the liver cancer frequency The

length of time and the amount of dose are important to calculate

liver risk For all protracted radiation exposures, it is difficult todetermine how much of the “dose” was essential to produce theliver cancer The dose that an organ receives after the cancer is ini-tiated and growing may be thought of as being “wasted” and shouldnot be included on the dose axis Selection of the length of time to

be used for calculation of “wasted radiation dose” is one of the mostimportant variables in the liver cancer risk estimates in thisReport For chemical and physical forms of radionuclides thatresult in nonuniform dose distribution, there is another form of

“wasted radiation.” This is the energy and dose that is deposited incells very close to radioactive particles These cells may be killed by

the large radiation doses and are not at risk for the induction ofcancer and the dose may be considered “wasted.” There are manyarguments over both of these two forms of wasted radiation, since

the dose during the tumor development may play a role in tumor

progression Cell killing in a local region may be responsible for ating an environment that is necessary for cells to transform,

cre-progress and be expressed as carcinomas For rodents injected with

a number of different alpha-emitting radioactive materials, ing Thorotrast®, the latent period for tumor induction was about

includ-600 d after the injection of the radioactive material (Brooks et al., 1982; 1983; Ober et al., 1994; Spiethoff et al., 1992; Taylor

et al., 1985) In dogs, it was about 10 y after injection before the first radiation-induced liver tumors appeared (Bair et al., 1980; Gilbert et al., 1998; Muggenburg et al., 1996; Taylor et al., 1986).

Latent periods of more than 20 y for induction of liver cancer in

humans have been observed and reported (Andersson et al., 1994; van Kaick et al., 1995) Because of this relatively long time between

injection of the activity and the development of liver cancer, thereare other radiation related causes of death such as bone cancerwhich eliminate the animals from risk to liver cancer These com-peting risks must be considered for accurate derivation of livercancer risk Studies in experimental animals demonstrated theneed to correct the data for competing risks to insure that the liver

risk is not underestimated (Gilbert et al., 1998; Muggenburg et al., 1996; Raabe et al., 1995).

5.4 Sensitivity for Cancer Induction

Data from dogs that inhaled 238Pu that was translocated to theliver showed an increase in the frequency of liver cancer Studiesconducted at the Inhalation Toxicology Research Institute (ITRI) in

Albuquerque, New Mexico (Muggenburg et al., 1996) showed that

there was a marked increase in the cancer frequency, with 25 livertumors in 144 exposed dogs and only four in 85 control dogs Therewas also an increase in liver cancer observed in the dogs thatinhaled 238Pu at Pacific Northwest National Laboratory (PNNL) in

Richland, Washington (Dagle et al., 1995; Park et al., 1997) Liver

cancer risk seemed to be exposure related However, no risk factorsfor liver cancers were derived from the PNNL studies because ofthe lower doses used The PNNL studies had two of the data pointsthat resulted in less than 1 Gy to the liver The data from ITRI andPNNL have been combined, evaluated and used to derive risk

5.4 SENSITIVITY FOR CANCER INDUCTION / 17

estimates for the liver (Gilbert et al., 1998) These data are used

later in this Report

Some dogs exposed to high doses of beta/gamma-emitting onuclides, 91Y, 144Ce, and 137Cs developed liver cancer Two of

radi-15 dogs with liver doses from 4 to 11 Gy developed malignant livertumors, while there were no liver tumors observed in 18 dogs with

estimated doses from 0.6 to 3.4 Gy (Muggenburg et al., 1996) Hahn

et al (1996; 1997) reported that there was an increased incidence

of liver cancer in dogs that inhaled 144Ce-144Pr at levels thatresulted in estimated liver doses from 6 to 250 Gy However, noliver tumors were observed in dogs whose 144Ce body burdenresulted in estimated liver doses less than 6 Gy Dogs that inhaled

137Cs also developed liver cancer (Nikula et al., 1995) Chinese

hamsters developed liver cancer following large doses (15 to

110 Gy) from injected 144Ce citrate (Brooks et al., 1982) For all

radiation exposures, it is very difficult to extrapolate the risk fromvery large doses of internally-deposited radioactive materials,where there is tissue damage, tissue disorganization, and largeamounts of cytogenetic damage, to levels encountered in the envi-ronment where few cells are damaged and no tissue alterations areevident These extrapolations require the use of the linearno-threshold hypothesis, which for low-LET radiation is more diffi-cult to justify than is the extrapolation from high to low doses forhigh-LET radiation (NAS/NRC, 1999)

6 Physical Variables that Alter Response

Attempts to evaluate the impact of some of the idiosyncrasies

of Thorotrast® with respect to radiation-induced genesis from other alpha-emitting radioactive materials have beenthe basis of various animal experiments These studies weredesigned to differentiate between the possible nonradiation effects

hepato-carcino-of Thorotrast® or the interaction between the chemical effects andthe radiation effects Studies were conducted to evaluate the possi-ble impact of the unique features of Thorotrast®: (1) the possiblechemical effects, (2) the relatively large mass of injected material,(3) the colloidal nature and focal distribution, and (4) the stabiliz-ing dextrans, and the preservative (0.15 percent p-oxybenzoic acid

ethyl-ester in ethanol) (Guilmette et al., 1989; Spiethoff et al., 1989; 1992; Stover, 1983; Taylor et al., 1985; Wesch et al., 1983).

The radiobiological application of these studies requires that thepossible biological impact of these factors be isolated from the radi-ation effects The studies all suggest that the major risk factor fromThorotrast® is the radiation dose and that the other factors evalu-ated are of minor importance

The most significant disparity, which complicates the use ofinjected Thorotrast® as an analogue for predicting the hepato-toxicity of other liver-seeking radionuclides, is its colloidal nature,which results in a nonuniform distribution of the material in thereticuloendothial cells of the liver Environmental exposures toother important elements such as plutonium or americium result

in deposition in the lung or the gastrointestinal tract To move fromthese sites to the liver requires that the radioactive material betranslocated across a membrane in the ionic or monomolecularform by way of the vascular system The deposition of ionic radio-nuclides is much more uniformly distributed in all the cell types ofthe liver, with its primary site of deposition being the hepatocyteswhile Thorotrast® is deposited in the Kupffer cells, the phagocyticcells of the liver There are two major scientific uncertainties whichneed to be addressed to determine if the physical variable, non-uniform dose distribution influences the risk estimated from

6.1 CELLULAR AND TISSUE DOSE DISTRIBUTION / 19

Thorotrast®: (1) Does the nonuniform tissue and cellular tion of dose from Thorotrast® particles influence the level of cellu-lar damage and the cancer incidence? (2) Does particle loadingassociated with Thorotrast® alter the time of onset or frequency ofliver tumors?

distribu-6.1 Cellular and Tissue Dose Distribution

Before it is possible to apply the results of the humanThorotrast® studies to other human populations exposed to eitheralpha or beta particles uniformly distributed in the liver, it is essen-tial to understand how dose distribution influences the induction ofcellular damage and cancer This is especially important sincethere are significant differences in the local dosimetry of colloidalThorotrast® as compared to some of the more uniformly distributedradionuclides Some of the studies that address these issues at bothcellular and whole-animal levels are presented below

6.1.1 Influence of Dose Distribution at the Cellular Level

The frequency of chromosome aberrations produced by the uniform distribution of dose from alpha particles associated withThorotrast® in liver cells has been related to the level of damageinduced by uniformly distributed 239Pu citrate (Brooks et al., 1985; Guilmette et al., 1989) It was observed that when the dose to liver parenchymal cells was calculated (Kaul et al., 1979) and related to

non-the frequency of chromosome aberrations, 239Pu and Thorotrast®were equally effective in producing chromosome aberrations

(Brooks et al., 1985; Cremer et al., 1992; Tanaka et al., 1996) The

level of primary cellular damage was directly related to cellulardose to the parenchymal cells for the two radionuclides, regardless

of dose distribution Many of the tumors that arise in the liver fromThorotrast® are cholangiocarcinomas which arise from epithelialcells lining the bile duct It was calculated that the annual dose tothese target cells could be as high as 8 Gy in regions of the ThO2

aggregates (Fisenne et al., 1985).

To further evaluate the influence of spatially nonuniform dosedistribution, Chinese hamsters were injected with 239Pu-citrate or

239Pu-oxide particles of different monodisperse sizes Using cle sizes ranging from 0.17 to 0.84 µm, the total activity in the liver

parti-was held constant while altering the total particle number and thelocal dose distribution Other groups of hamsters were injectedwith a range of 239Pu activity levels using a single particle size Inthese experiments, the local dose distribution was held constantwhile varying the total organ dose.

When the average dose to the liver was related to the averagefrequency of chromosome aberrations, the amount of damage washigher for uniformly distributed plutonium citrate Using the dose

to the cells in range of the 239PuO2 particles, the aberration quency in animals injected with plutonium-oxide particles

fre-decreased as a function of increasing particle size (Brooks et al.,

1974) When the total dose was used, there was limited influence

of particle size on the total aberration frequency The frequency ofcells with multiple chromosome aberrations increased withincreasing particle size and localized dose The decrease in the totalfrequency of aberrations as a function of increasing particle size,and the increase in the number of cells with multiple aberrations,both appear to be the result of the nonuniform dose distribution.This nonuniform dose distribution may have resulted in increasedlevels of cell killing in regions close to the particle, thereby prevent-ing some of the badly damaged cells from dividing and being scored.Those cells that were close to a “hot” particle and were able todivide resulted in cells with multiple aberrations

6.1.2 Influence of Dose Distribution on Liver Cancer

The effect of dose distribution on the induction of liver cancerhas been measured by comparing groups of rats injected withThorotrast® enriched with different quantities of 230Th Thisexperimental design makes it possible to hold the total number ofparticles and mass of the material constant while changing theradiation dose and dose rate associated with the particles (Wesch

et al., 1983) These experiments demonstrated that the incidence of

both liver and spleen tumors was dose-rate dependent, and that thetumor incidence increased linearly with dose

Additional research was performed to relate the liver cancer quency to radiation dose This was done using mice injected witheither 214Am or Thorotrast® (Taylor et al., 1985) or Chinese ham-

fre-sters injected with 239Pu or Thorotrast® (Guilmette et al., 1989).

Again, Thorotrast® had similar effectiveness to other emitting radionuclides per unit average dose in the production ofliver cancer However, differences were observed when relative risk

alpha-6.1 CELLULAR AND TISSUE DOSE DISTRIBUTION / 21

coefficients were derived for the induction of liver cancer inChinese hamsters following injection of Thorotrast® or plutonium.The data in Table 6.1 represent the natural log of the relative riskfor the induction of cancer These risks were derived for differentlevels of injected Thorotrast® and for a single level of plutonium

(Guilmette et al., 1989) The values reported in the table are the

natural logarithm of the mean and the standard error of the mean

of the relative risk coefficients Following Thorotrast® injection, therelative risk increased as a function of the injected activity Thiswas also observed for rats given the same radiation dose with dif-

ferent levels of injected particulates (Wesch et al., 1983) The

rela-tive risk in Chinese hamsters for induction of liver cancer byThorotrast® injected at an activity level of 1.5 Bq g–1 was similar tothat determined for injection of uniformly distributed plutoniumcitrate at 7.4 Bq g–1

Cancer incidence (Brooks et al., 1983) induced by nonuniform

dose distribution from 239Pu oxide particles was compared withcancer induction by uniformly distributed plutonium citrate inChinese hamsters As discussed above, the experiments weredesigned to change the particle size and number of particlesinjected to result in the same total dose to the liver with markedlydifferent dose distributions Data from this experiment demon-strated that with a constant dose there were no differences intumor response as a function of particle size However, the fre-quency of liver cancer per unit dose was increased by a factor of

TABLE 6.1—Natural logarithm of relative risk coefficients for induction of liver cancer and nodular hyperplasia in Chinese hamster from injection with Thorotrast ® or plutonium citrate relative to combined risk for citrate and dextrin control groups.a

aGuilmette et al (1989).

2.5 in animals injected with 239Pu citrate relative to those exposed

to 239Pu oxide particles Particulate nonuniform dose distributionfrom 239Pu oxide was thus less effective in the production of livercancer than was uniform 239Pu citrate deposition This is in agree-ment with past observations that uniform distribution of alphadose in the lung is more effective than nonuniform radiation fromparticulate materials that emit alpha particles in the production ofcancer (NCRP, 1975)

The above observation is supported by two laboratory ments with rats in which liver tumors were induced by Thorotrast®

experi-and fractionated neutron irradiation (Spiethoff et al., 1992; Wegener et al., 1989) The homogenous neutron irradiation pro-

duced a higher number of liver tumors per unit of dose than thenonhomogenous Thorotrast® No differences were found in the his-tological types of liver tumors and preneoplastic lesions (Ober

et al., 1994).

For dogs and some rodents (Chinese hamsters, deer mice, andgrasshopper mice), nonparticulate 239Pu, 241Am and some of theother actinide elements are deposited principally within the hepa-tocytes and are uniformly retained and distributed at early timesafter injection This distribution changes with time In dogs, theuniform distribution has been shown to be temporary, with a shift

of the radionuclide to the Kupffer cells This shift from the cytes to the Kupffer cells starts at about two to three months postexposure in animals with high activity levels and at more than 1 y

hepato-in the livers of animals with lower levels of activity This shift isseen at the same time that iron-staining pigments are starting toaccumulate in Kupffer cells These reticuloendothelial cells, inturn, migrate toward the portal regions, resulting in an increas-ingly more focal radionuclide burden The nonuniform distribution

of the radioactive material in the liver is further increased at laterpost-exposure times by cell division which resulted in the forma-tion of hyperplastic nodules Since the radionuclides do not migrateinto the new hyperplastic nodules, these cells remain “cold” (Taylor

et al., 1985) The development of hyperplastic nodules also puts

many of the liver cells out of range of the alpha particles makingfurther irradiation and direct radiation-induced cancer changesimpossible Thus, the distinction between the liver distribution ofThorotrast® and the initially nonparticulate forms becomes pro-gressively less at the longer post exposure times These changessupport the use of Thorotrast® as a model for other alpha-emittingradionuclides

6.3 THE ROLE OF INJURY ON CANCER / 23

6.2 Influence of “Particle Loading” on Cancer

et al., 1983) It was observed that larger amounts of Thorotrast®

shortened the time between injection and induction of cancer Thisexperiment suggested that the Thorotrast® particles were impor-tant in tumor production and may act as promoters to shorten thelatent period, the time between injection and the development ofcancer, but did not influence total incidence

Wistar rats were injected with Zirconotrast® (120 µL) and givenfractionated irradiation with neutrons (10 Gy delivered in doses of

0.2 Gy per fraction) during a 2 y period (Spiethoff et al., 1989;

1992) The data showed that neutrons induced a high level of livercancer (40 percent) In the animals treated additionally with ZrO2,the incidence, time of onset, type of tumor, and overall number ofliver tumors was nearly equal to that for the neutrons only Thesame is true for premalignant stages of hepatocellular carcinomathat were examined morphologically and biochemically The addi-tion of ZrO2 burden did not produce a higher number of preneo-

plastic foci or different phenotypes (Ober et al., 1994) This

indicates that the presence of the particles in the liver duringthe animals entire lifetime did not change the liver’s response tothe high-LET radiation in terms of total tumor formation Themajority of the animal data suggest that the presence of ZrO2 par-ticulate material in the liver did not influence the liver cancer fre-quency induced by high-LET radiation Such information suggeststhat the particulate matter does not have a major impact on livercancer risks The risk for the development of liver cancer was, thus,related to the radiation dose to the hepatocytes and not to other fac-tors such as dose distribution, particulate effects, or nonuniformdistribution of cellular damage Such data helps support the use ofThorotrast® to estimate the human liver cancer risks from otherinternally-deposited alpha-emitting radioactive materials

6.3 The Role of Injury on Cancer

The role of cellular response to injury or stress is important inunderstanding risk for tumor induction Animals have been

exposed to either external radiation (Cole and Nowell, 1965) orinternally-deposited beta/gamma-emitting radionuclides, like

144Ce (Brooks et al., 1982) and followed by stimulation of cell

pro-liferation by either carbon tetrachloride or partial-hepatectomy Inthese studies, the combined damage and cell proliferationincreased the liver tumor frequency and decreased the timebetween exposure and the induction of liver cancer This suggeststhat, after genetic damage is induced, stimulation of cell divisionincreases the risk for tumor induction This supports the classicinitiation-promotion theory where the radiation acts as an initiatorand the liver cell division promotes the expression of the initiatedcells to form cancer Liver disease, associated with chronic con-sumption of alcohol, has also been shown to result in an increase inliver tumor frequency in dogs injected with 241Am (Taylor et al.,

1992) In these experiments, the alcohol increased the excretionand clearance of the 241Am resulting in reduced radiation dose tothe liver Since the liver cancer frequency increased and the dosedecreased, the risk for liver cancer per unit of dose increased mark-edly This research suggests a synergistic interaction between cel-lular damage and proliferation induced by alcohol consumption,

altered nutritional status (Kiyosawa et al., 1989) and the exposure

to alpha-particle irradiation from Thorotrast® This interactioncould increase the frequency of liver cancer in Thorotrast® patientsthat had liver disease which was induced by changes in nutritional

status (Kiyosawa et al., 1989) or by other environmental insults

such as excessive alcohol consumption This is an area thatrequires additional evaluation in the human populations exposed

6.4 THE ROLE OF MECHANISTIC STUDIES ON LIVER CANCER RISKS / 25

It has been shown that radiation exposures at very low dosesproduce adaptive cellular responses that seem to be related tothe production of repair proteins These induced proteins alter theeffectiveness of subsequent radiation exposure on the induction ofchromosome damage (Wolff, 1995) There is no evidence, at thecurrent time, that these adaptive responses have an influence onthe cancer risk from chronic low-dose rate exposures such as resultfrom internally-deposited radionuclides (see Annex B, UNSCEAR,1994) However, continued research in this area is important.The interaction of radiation and other physical and chemical

agents produces changes in rates of mutations (Evans, 1991) and

cell proliferation (Taya et al., 1994), both of which seem to be

essen-tial for the multistep process associated with the development of

cancer (Knudson, 1991; Moolgavkar et al., 1990) There is

increas-ing evidence that multiple cellular and genetic changes arerequired to change normal cells to cancer prone cells The initialchanges, associated with high- or low-LET radiation, can be

induced in oncogenes (Hickman et al., 1994; Knudson, 1991), tumor

suppressor genes (Atkinson et al., 1995), or in genes that produce

an unstable genome (Kadhim et al., 1995) It has been

demon-strated that low doses of 239Pu combined with the incorporation ofthe mutated Ki-v-ras oncogene into the liver produce a higher rate

of liver cancer than either of the two treatments given alone

(Brooks et al., 1995) Other studies have demonstrated that

expo-sure of transgenic mice, which are deficient in the tumor

suppres-sor gene, p53, to acute gamma rays, results in both an increased

frequency of chromosomal breaks and an increase in tumor

frequency (Lee et al., 1994).

Thus, the hypothesis that multiple cellular and molecular ations are required to change normal cells into cancer cells is

alter-well supported (Vogelstein et al., 1988) The question remains as

to the relationship between the progression of tumor changes andthe number and time sequence of alpha traversals required

to initiate these changes Recent research has suggested thatradiation-induced genomic instability may play a role in radia-tion-induced cancer and provide a mechanism to explain the multi-ple biological changes needed for normal cells to be converted tocancer cells This radiation-induced genomic instability involves avery large cellular target that is much bigger than a single gene

(Kadhim et al., 1995; Ponnaiya et al., 1997a; 1997b) It has been

demonstrated that the nucleus of a cell seems to contain this target

(Morgan et al., 1996) These genetic changes influence the stability

of the genome, and this leads to a mutator phenotype that produces

the multiple changes observed, especially in solid tumors Genomicinstability can be induced in a range of different cell types from

a range of different types of radiation (Kadhim et al., 1992;

1994; 1995) Thus, a single cell can be altered by a single alpha ticle and undergo changes that make its progeny unstable Thisalteration could result in the cascade of events required for theinduction of cancer or it could put a larger cell population at risk forsubsequent exposures, which could then produce other changesinvolved in the multi-step process leading to cancer

par-Genomic instability has been reported in primary tissue-culture

cells: (1) as a function of time in culture (Cram et al., 1983; Ray et al., 1986), (2) during tumor progression (Bartholdi et al., 1987; Otto et al., 1989), (3) as part of aging and senescence (Hornsby et al., 1992), and, more recently, (4) as a function of alpha particle (Kadhim et al., 1992; 1994; 1995) or gamma-ray exposure (Holmberg et al., 1993; Marder and Morgan, 1993) There have

been many changes associated with induction and progression ofgenomic instability that help to define the mechanisms involved(Cheng and Leob, 1993; Kronenberg, 1994) Cytogenetic instabilitywas induced in mouse bone marrow by alpha particles, but was not

found following exposure to low-LET radiation (Kadhim et al.,

1992) There was little indication that the amount of the chromatiddamage produced by genomic instability was related to the dose

from alpha irradiation that triggered it (Kadhim et al., 1992) For

alpha particles, the lack of a dose-response relationship for theinduction of genomic instability may suggest the possibility thatgenomic instability may be an “all-or-none” phenomenon It hasbeen determined that the frequency of cells that contain genomic

instability increases as a function of dose (Morgan et al., 1996; Ponnaiya et al., 1997a) Such studies demonstrate that molecular

biology is starting to provide additional information on the ular mechanisms involved in radiation-induced cancer Data fromsuch cellular and molecular studies may in the future play anincreasing role in risk assessment Such studies will be very helpful

molec-in selection of models for risk extrapolation, molec-in understandmolec-ing thecancer process, and in defining the risk associated with radiation-induced cancer

The experimental animal studies used to evaluated the ence of dose distribution, particle loading, localized damage, andchemical factors associated with injected Thorotrast® suggest thatall of these factors may have some impact on risk These factors,thus, increase the uncertainty associated with the use of thehuman Thorotrast® data in estimating the risk of liver cancer

influ-6.4 THE ROLE OF MECHANISTIC STUDIES ON LIVER CANCER RISKS / 27

However, the impact seems to be minimal, and the total effectiveradiation dose to the liver parenchyma remains the dominate fac-tor in estimation of the risk of radiation-induced cancer Theseexperimental and mechanistic studies support the usefulness ofthe human Thorotrast® data to predict the risks of liver cancerinduced by other internally-deposited radionuclides

7 Use of Experimental

Animal Data

7.1 Relative Biological Effectiveness of the

Radiation Emitted by Radionuclides

Extensive cellular and molecular data have been derived to helpunderstand the RBE of different types of radiation (NCRP, 1990)including the types emitted by internally-deposited radioactivematerials (Brooks, 1975) Cellular and animal data help improveour basic understanding of the radiation-induced damage in theliver that leads to liver cancer RBE values are dependent on end-point, exposure type, and exposure rate for the types of radiationbeing compared (NCRP, 1990) The RBE value for chromosomeaberrations was derived for liver cells by comparing the effective-ness for inducing chromosome damage from high-LET alpha parti-cles (239Pu, 241Am, 238Pu) with that from low-LET beta/gammaradiation (60Co, 144Ce) There was little difference in the frequency

of chromosome aberration in the liver cells following exposure toprotracted 60Co gamma-ray exposure relative to the beta/gamma-emitting 144Ce With these data, the range of RBE valuesfor the induction of initial cytogenetic damage from alpha-emittingradiation relative to protracted low-LET radiation was from 15 to

20 (Brooks, 1975)

7.2 Liver Cancer Risk from Animal Data

The use of animal data is essential for estimating liver cancerrisk in humans from exposure to protracted exposures frombeta/gamma emitters where there are few or no human data avail-able Some radionuclides deliver their dose over an extended por-tion of the life span One approach is to estimate the risk for theinduction of liver cancer in laboratory animals using differentinternally-deposited radionuclides with a range of half-lives andemission characteristics The tumor responses for these radionu-clides can then be compared to that observed for Thorotrast® It is

7.2 LIVER CANCER RISK FROM ANIMAL DATA / 29

then possible to use the risk-ratio method developed by Mays et al.

(1986) to estimate cancer risks for radionuclides where there are nohuman data In the risk-ratio method, the risk for human liver can-cer for the subject radionuclide is estimated by first deriving theratio of the cancer risks in animals for the subject radionuclide andanother radionuclide for which human data are available Next,this ratio is used to estimate the cancer risk in humans where thereare no human data available for the subject radionuclide Forexample, if we were interested in the risk for a beta/gamma radio-nuclide like 144Ce-144Pr in humans, where there are no data, wefirst establish a toxicity ratio between the liver cancer risk for

144Ce-144Pr and the liver risk from Thorotrast® in an animal tem The toxicity ratio derived is then used in combination with therisk for Thorotrast® in humans to estimate the human risk from

sys-144Ce-144Pr exposure (Equation 7.1):

where 144Ce-144Pr (animal) divided by the Thorotrast® (animal) isthe toxicity ratio This ratio method enables a comparison of theliver cancer response using several different animal species andmakes it possible to minimize the influence of species differences.The ratios approach makes it possible to evaluate the influence ofLET on liver cancer risk Two major data sets were used to evaluatethe role of LET on risk First, Chinese hamster data were used toderive effectiveness factors for induction of liver cancer from bothhigh- and low-LET radiation These studies compared liver tumorinduction from 144Ce citrate (Brooks et al., 1982) with tumors pro-

duced by 239Pu citrate and Thorotrast® (Brooks et al., 1983; Guilmette et al., 1989) Next, data from dogs were used to derive

risk factors for the induction of liver cancer as a function of LET.The data used for determining the risk for the induction of cancer

by high-LET radiation were derived from dogs exposed to 239Pu or

241Am (Gilbert et al., 1998; Muggenburg et al., 1986; Park et al., 1997; Taylor et al., 1991) The data were then compared to those

from low-LET radiation from internally-deposited 144Ce (Gilbert

et al., 1998) The data on the risk from internally-deposited

radio-active materials to be used in making the risk ratios are rized on Table 7.1

summa-The dog data from both ITRI and PNNL were combined and

evaluated for liver risk (Gilbert et al., 1998) The data suggested

that the ITRI dogs may have had a slightly higher response thanthe PNNL dogs, but that the errors associated with the data make

144 Ce-144Pr (animal)

Thorotrast® (animal)

× Thorotrast® (human)

TABLE 7.1—Risk factors for liver cancer derived in experimental

animals: The influence of LET.

Route of Exposure Radionuclide Cancers

7.2 LIVER CANCER RISK FROM ANIMAL DATA / 31

it possible to fit them to a single linear function Gilbert et al.

(1998) used two different approaches to estimate the risk for livercancer from these combined data sets In the first approach, riskwas based on dogs with initial lung burdens (<30 kBq kg–1) Thesecond approach was based on dogs with doses of <1 Gy As can beseen from Table 7.1, these different approaches derived differentrisk factors of 2,500 to 3,300 liver cancers 10–4 dogs Gy–1 for dogsexposed to <1 Gy The 95 percent confidence intervals on theGilbert data were reported to be 1,300 to 4,200 using risk based oninitial lung burdens and 800 to 7,200 using the dose based risk.These data are compatible, but demonstrate the wide range ofuncertainty associated with using these data to estimate liver can-cer risks As an estimate of the risk for the ratios’ extrapolation, acancer risk of 2,500 cancers 10–4 Gy–1 was derived from evaluation

of the data based on activity less than 30 kBq kg–1 (Gilbert et al.,

1998) The liver risk from alpha exposure in dogs was related to theliver risk from low-LET emitting radionuclides in dogs Using theratio’s method, the human liver cancer risk could then be estimatedfor protracted exposures to low-LET radiation

The risk’s ratio, or RBE approach has a large number of tions and uncertainties associated with it It requires that thedosimetry be done in a similar manner, that the differences in lifespan be appropriately addressed, that competing risks be consid-

limita-ered in both species and that the endpoints used, i.e., the types of

cancers, be the same for the two species All of these requirementsare difficult to meet This was illustrated in the dog studies wheresimilar protocols and methods of determining the liver cancer fre-quency and type were used at ITRI and PNNL, but they had differ-

ent cancer incidence (Gilbert et al., 1998) These results

demonstrate that there are large uncertainties associated with theliver cancer risks reported, especially when it is necessary to useanimal data derived at a number of different institutions as theprime source of information for calculating risk

The cancer risk estimate of 2,500 liver cancers 10–4 dogs

exposed to 1 Gy derived by Gilbert et al (1998) (see Table 7.1), was

used as a representative value for inhaled 239Pu and 238Pu andinjected 241Am The risk for the low-LET radiation in the dogs wasderived from studies of inhaled 144Ce and estimated to be

100 10–4Gy–1 (Hahn et al., 1996) The RBE of alpha radiation for

the induction of liver cancer in dogs using 144Ce as the base linewas 25 Combining the rodent data for injected 239Pu in theChinese hamster and injected 239Pu, 241Am, and Thorotrast® inthe grasshopper mouse resulted in a value of 1,080 10–4 Gy–1 This

can be compared to the value of 80 10–4 Gy–1 for Chinese hamstersinjected with 144Ce For the rodents, the RBE for the induction ofliver cancer for high-LET relative to low-LET radiation from pro-tracted low-dose rate exposures was thus estimated to be about 15.These extrapolations to humans from toxicity studies conducted inanimals necessarily assume that the use of toxicity ratios for such

purposes is valid (Mays et al., 1986) Toxicity ratios cannot be used

directly to estimate the risk for humans, but they illustrate thatvalues of the RBE derived from these studies are similar to thoseused by the National Council on Radiation Protection and Mea-surements (NCRP) and ICRP In the past, the liver cancer risk wasestimated to be 300 10–4 Gy–1 (NAS/NRC, 1988; UNSCEAR, 1994)

In this Report, using the more recent Thorotrast® data, it is mated that the liver cancer risk is higher of 560 ± 95 (400 10–4 Gy–1for women to 650 10–4 Gy–1 for men) If this range of values isdivided by the toxicity ratios with the risks for dogs (25) androdents (15), it was possible to generate a range of risks to humansfrom low-LET irradiation for liver cancer of 15 to 40 10–4 Gy–1.These data can be compared with the liver cancer estimatesderived from the exposure to the atomic bomb (UNSCEAR, 1994)

esti-“The risk of fatal primary liver cancer (all cases) was estimatedfrom the life span study mortality data to be about 1.2 10–4(PYSv)–1 or 48 10–4 Sv–1 for 40 y at risk.” This risk is derived athigh-dose rates If the low-dose rate reduction factor of two isapplied the risk would be 24 10–4 Sv–1 for 40 y at risk These valuesare very similar to the information derived by combining the ani-mal data with the human Thorotrast® data

fac-risk from internally-deposited alpha-emitting radioactive als The most extensive human data for estimating the long-termrisk to liver from internally-deposited alpha-emitting radionu-clides are the Thorotrast® cases.

materi-Thorotrast® is a 25 percent colloidal suspension of thorium

diox-ide containing 228Th/232Th The majority of the radiation dose andrisk associated with Thorotrast® is from alpha particles associatedwith the decay of the daughter products of both 228Th and 232Th Itwas developed in Germany and was first used as a radiographiccontrast medium in 1928 by Frick for visualization of the vascularsystem (Muth, 1989) Use of Thorotrast® as a contrast medium wasstarted in 1929 and continued until about 1950 Thorotrast®was also used to enhance the radiological visualization of manyother sites such as the parotid salivary duct, the lachrymal duct,the uterus, and the spinal canal (Ahmed and Steele, 1972; Freilich,

1983; Griffiths et al., 1977; Meyer et al., 1978; Mihatsch and Rutishauser, 1973; Molla et al., 1976; Westin et al., 1973) The

number of people injected with Thorotrast® is unknown, but mates ranging to over 1,000,000 have been made (Abbatt, 1979;Battifora, 1976) In Japanese studies alone, the number of exposed

esti-people is estimated at 20,000 to 33,000 (Mori et al., 1983).

The use of Thorotrast® in vascular imaging, especially cerebralarteriography, gained considerable acceptance because of its excel-lent contrast properties and its freedom from “immediate” toxiceffects (Muth, 1989) However, its late effects ultimately made itone of the most serious causes of iatrogenic disease Two factorswere especially prominent in its toxicity: (1) Prolonged lifetimeretention, principally in the mononuclear-macrophage system

(Hursh et al., 1957; Taylor et al., 1985; Wegener et al., 1976) and

(2) the focal chronic alpha radiation delivered by 232Th and some of

its decay products (Kato et al., 1979; Kaul et al., 1979).

Long-term exposure to Thorotrast® was found to produce livercirrhosis, myeloproliferative disorders and liver cancer in some ofthe patients after a long latent period From the dosimetric, clinicaland the epidemiological points of view, we have to separate the sys-temic, most frequently administered intra-arterial applicationfrom the local administration of Thorotrast® In the following, withregard to liver cancer risk, the focus will be on the systemicapplication

These Thorotrast® patients were injected with known volumes

of Thorotrast® so that they received known levels of alpha-emittingradionuclide The average intra-arterial injection was about 25 mL

of Thorotrast®, which contained approximately 5 g of thorium or22.2 GBq (600 mCi) of 232Th, with additional activity associatedwith its progeny This resulted in a dose rate to the liver for a stan-dard 70 kg man of about 0.25 Gy y–1, assuming a liver retention of

60 percent and that 65 percent of the alpha energy escapes from theThorotrast® aggregates and is absorbed in tissue (Kaul, 1973a;

Kaul and Noffz, 1978; van Kaick et al., 1978).

Many of the individuals were counted to determine the

reten-tion distribureten-tion (Kaul, 1995; van Kaick et al., 1989) And the local and average dose and dose rate to the liver (Dalheimer et al., 1995).

The liver cancer incidence, type of cancers produced and the timebetween the injection and the onset of the disease was well docu-

mented (Andersson, 1997; Andersson et al., 1994; 1995; van Kaick

et al., 1995; 1999) This makes the Thorotrast® records a very richdata set for epidemiological studies Epidemiological studies wereconducted in Germany, Japan, Portugal, Denmark, Sweden

(Dahlgren, 1967), and the United States (Janower et al., 1972) In

addition there have been a number of individual case reports The

German (van Kaick et al., 1989), Japanese (Mori et al., 1985; 1989; 1995), and Danish (Andersson and Storm, 1992; Andersson et al.,

1994; Faber, 1978; 1985) studies provide the most useful data Theyinclude a large number of cases in which the patient follow-up pro-vides information on latency, dosimetry and postmortem evalua-tion including autopsy Four of the studies also established acontrol group (Muth, 1989)

The potential hazards associated with the use of Thorotrast®were cited by the American Council on Pharmacy and Chemistry asearly as 1932 (JAMA, 1932; 1937) and by others (Bauer, 1937;

Janower et al., 1972; Reeves and Stuck, 1938; Taft, 1937; van Kaick

et al., 1986) In 1947, the first probable Thorotrast®-induced liver