Abstract
Editorial
Stem Cell Radiobiology: Fundamental Science for Radiological Protection
The International Commission on Radiological Protection (ICRP)’s system of radiological protection is built on three pillars: scientific evidence, ethical values, and experience. The present publication is the most recent to examine aspects of the scientific evidence base. Other recent examples have included Publications 115 and 118 (ICRP, 2010, 2012).
It is imperative that ICRP continues to examine the latest developments in the science fundamental to the system of radiological protection in order to assess whether changes may be necessary in the light of deeper understanding. One of the five standing committees of ICRP, Committee 1 on Radiation Effects, is dedicated to this task, and this is also an important element of several of the other committees, most notably Committee 2 on Doses from Radiation Exposure, and Committee 5 on Protection of the Environment.
Although the present publication has a larger component of review and consolidation of primary scientific evidence than is often encountered in ICRP publications, it does consider the implications of the emerging evidence from stem cell biology for radiological protection. These implications touch on some of the fundamental aspects of the system of radiological protection concerning the stochastic effects of ionising radiation on health including: defining the populations of target cells for radiation carcinogenesis in different tissues, the use of linear-no-threshold extrapolation and relative risk models for carcinogenesis, the relationship between high and low dose and dose-rate effectiveness, and age dependency of the risk of radiation cancer.
Some of the evidence and concepts relating to adult stem cells (long-lived cells that maintain body tissues through life) have their origins in radiobiology (e.g. Till and McCulloch, 1961; Bedford and Dewey, 2002). Stem cells and their immediate progeny have long been considered as likely target cells for the initation of cancer due to their characteristics of long life span in the body and the requirement for careful control of the fate of progeny daughter cells following division, which enables continued long-term tissue maintenance in the absence of excessive and detrimental proliferation. Fundamental research on stem cells has progressed dramatically in recent decades, driven mainly by the potential use of stem cells in therapy for a range of diseases, and recognition of the importance of stem cells in pathogenesis. The identification of a very limited number of genes that determine ‘stemness’ has been critical to the progress of stem cell research. This was recognised in the award of the 2012 Nobel Prize in Physiology or Medicine to Sir John Gurdon and Sir Shinya Yamanaka for their work on the control of pluripotency. In addition to a greater depth of knowledge in the control of stemness, there have been major advances in the ability to characterise and isolate specific stem cell populations in humans and laboratory models. Given this expansion of knowledge, and the recognition of stem and early progenitor cells as being important target cells in radiation carcinogenesis, it is timely to see a comprehensive review of tissue stem cells in relation to radiation carcinogenesis.
Clearly, we do not yet have a complete understanding of the characteristics of stem cells, and how they react to ionising radiation exposure over dose ranges relevant to radiological protection. Nonetheless, this publication raises some interesting and potentially important new issues and concepts; for example, the radiation-induced modification of specific tissue microenvironmental ‘niches’ in which stem cells reside, and how competition for niche occupancy may impact upon the risk of cancer. In addition, new insights into the possible dedifferentiation of cells to acquire stem cell characteristics may be important in developing a complete understanding of the ‘at-risk’ target cell populations. Given that the knowledge base will continue to expand and develop, one should anticipate further development and refinement of the implications of the evolving understanding of stem cell radiobiology for radiological protection.
In an overall sense, this publication highlights the need to gain detailed insights into fundamental biological processes, particularly in stem cells, to inform judgements on risk extrapolation and risk estimation for radiological protection. In the near future, it is difficult to envisage that radiological protection will move away from drawing on epidemiological evidence as the fundamental basis for evaluation of the risk of radiogenic diseases. However, increasing knowledge of the biological basis of radiogenic disease pathogenesis will serve to strengthen confidence in approaches to risk estimation at doses and dose rates beyond those where epidemiological evidence is available. In this respect, the present publication should be welcomed as an important move to further integrate biological knowledge into radiological protection recommendations.
SIMON BOUFFLER
VICE-CHAIR, ICRP COMMITTEE 1
CHRISTOPHER H. CLEMENT
ICRP SCIENTIFIC SECRETARY
EDITOR-IN-CHIEF
Stem Cell Biology with Respect to Carcinogenesis Aspects of Radiological Protection
Approved by the Commission in February 2015
Knowledge of the biology and associated radiation biology of stem cells and progenitor cells is more developed in tissues that renew fairly rapidly, such as haematopoietic tissue, intestinal mucosa, and epidermis, although all the tissues considered here possess stem cell populations. Important features of stem cell maintenance, renewal, and response are the microenvironmental signals operating in the niche residence, for which a well-defined spatial location has been identified in some tissues. The identity of the target cell for carcinogenesis continues to point to the more primitive stem cell population that is mostly quiescent, and hence able to accumulate the protracted sequence of mutations necessary to result in malignancy. In addition, there is some potential for daughter progenitor cells to be target cells in particular cases, such as in haematopoietic tissue and in skin. Several biological processes could contribute to protecting stem cells from mutation accumulation: (a) accurate DNA repair; (b) rapidly induced death of injured stem cells; (c) retention of the DNA parental template strand during divisions in some tissue systems, so that mutations are passed to the daughter differentiating cells and not retained in the parental cell; and (d) stem cell competition, whereby undamaged stem cells outcompete damaged stem cells for residence in the niche. DNA repair mainly occurs within a few days of irradiation, while stem cell competition requires weeks or many months depending on the tissue type.
The aforementioned processes may contribute to the differences in carcinogenic radiation risk values between tissues, and may help to explain why a rapidly replicating tissue such as small intestine is less prone to such risk. The processes also provide a mechanistic insight relevant to the LNT model, and the relative and absolute risk models. The radiobiological knowledge also provides a scientific insight into discussions of the dose and dose-rate effectiveness factor currently used in radiological protection guidelines. In addition, the biological information contributes potential reasons for the age-dependent sensitivity to radiation carcinogenesis, including the effects of in-utero exposure.
© 2015 ICRP. Published by SAGE.
Keywords:: Stem cells; Normal tissues; Radiation sensitivity; Chronic radiation; Carcinogenesis; Radiation risk
AUTHORS ON BEHALF OF ICRP O. NIWA, M.H. BARCELLOS-HOFF, R.K. GLOBUS, J.D. HARRISON, J.H. HENDRY, P. JACOB, M.T. MARTIN, T.M. SEED, J.W. SHAY, M.D. STORY, K. SUZUKI, S. YAMASHITA
PREFACE
This report was prepared by a task group of International Commission on Radiological Protection (ICRP) Committee 1 under the following terms of reference: (a) to review stem cell biology and radiobiology with reference to mechanisms of radiation carcinogenesis; (b) to compare the response of stem and associated cells in different tissues with the respective risks of cancer, and elucidate the likely roles of stem cells, progenitor cells, and the stem cell niche; and (c) to use knowledge of stem cell/progenitor cell biology, radiation responses, and carcinogenic risks from homogeneous acute exposures in a discussion of the guiding principles for the projection of stochastic risks for short-range radiation and chronic-irradiation scenarios.
In its recommendations and various reports, ICRP has made various judgements and assumptions about the location and radiation response of the target cells responsible for carcinogenesis in various tissues. In most cases, the target cells are considered to be the tissue stem cells, and, in some cases, their daughter progenitor cells. The renewal and radiation response of these cells change with age and are governed by signals from their ‘niche’ residence. The fundamental evidence for stem cells as target cells has increased in recent years. This evidence contributes to the understanding of the biological basis for carcinogenesis, and helps to support modelling of human responses. It was considered that a report on this subject of target/stem cells would be topical and valuable in order to put all the target cell evidence for carcinogenic radiation risk in different tissues into a common framework and perspective for the first time.
In order to address stem cell knowledge with respect to particular issues of
continuing importance to ICRP, such as the linear-no-threshold model, dose-rate
effects, location of target cells, tissue risk factors, and age-dependent
sensitivity to radiation, it was necessary to review evidence for different organ
systems. This was undertaken by compiling a series of annexes as separate reviews
using a common template of topics for each of a chosen series of seven organ systems
with different characteristics. The annexes (and authors) are as follows: Annex A. Haematopoietic tissues: role played by stem cells and
lineage-committed progenitors in radiation-induced leukaemia (T.M.
Seed); Annex B. Mammary gland stem cells (M.H.
Barcellos-Hoff); Annex C. Thyroid stem cells (K. Suzuki and S.
Yamashita); Annex D. Digestive tract stem cells (J.H. Hendry); Annex E. Lung stem cells (J.W. Shay, M.D. Story, and P.
Jacob); Annex F. Skin stem cells and radiation carcinogenesis (M.T.
Martin); and Annex G. Bone stem cells (J.D. Harrison and R.K.
Globus).
Information on both humans and experimental animal systems was reviewed, and projections were made of the possible role of various stem cell processes in cancer risk.
Many helpful comments were received, in particular from R. Cox and C.S. Potten (deceased 3 August 2012); Committee 1 members: S. Bouffler, D. Laurier, A.J. Sigurdson, M. Tirmarche, R. Wakeford, and W. Doerr; Main Commission members: D. Cool, C. Cousins, C-M. Larsson, and J. Lee; and C. Land, N. Nakamura, A. Noda, D. Preston, J. Preston, R. Shore, and C. Streffer.
The membership of Committee 1 during the period of preparation of this report was:
EXECUTIVE SUMMARY
(a) The International Commission on Radiological Protection (ICRP) issued
new recommendations for the system of radiological protection in
Publication 103 (ICRP, 2007). Stochastic
effects of radiation, which are cancer and heritable effects, were also
considered. Previously, the Commission reviewed various aspects of
cancer induction from radiation, such as the risk of skin cancer in
Publication 59 (ICRP, 1992), genetic
susceptibility to cancer in Publication 79 (ICRP, 1998),
biological effects after prenatal irradiation (embryo and fetus) in
Publication 90 (ICRP, 2003), and low-dose
extrapolation of radiation-related cancer risk in Publication
99 (ICRP, 2005). More recently, the Commission also reviewed the
risk of lung cancer from radon in Publication 115
(ICRP,
2010), and the threshold doses for tissue reactions
(deterministic effects) in Publication 118 (ICRP,
2012). (b) Cancers arise as a result of mutational changes in single target
cells, and an aspect of fundamental importance is the nature and
characteristics of the target cells responsible for cancer induction.
Previously, it has been assumed that, in most cases, the target cells
are the ‘stem cells’ in the tissue in question, without further
elaboration. In recent years, there has been an enormous increase in
knowledge about the lineages of stem cells, daughter progenitor cells,
and the differentiated functional cells. Knowledge has also accumulated
about the regulatory signals in different tissues, and how the tissues
respond to injury. Hence, it was considered appropriate at this time to
review the aspects of the field of stem cells and their regulatory
processes that are pertinent to radiation carcinogenesis models.
However, there are no recent comprehensive reviews of carcinogenic
mechanisms after exposure to ionising radiation regarding stem cells as
target cells in different tissues and including age effects. (c) The main text of the report consists of Chapters 1–3. These are
followed by Annexes A–G which comprise detailed reviews on stem cells in
the haematopoietic system, mammary gland, thyroid, digestive tract,
lung, skin, and bone. Selection was made on the basis of importance for
radiological protection purposes and of the extent of available
radiobiological knowledge and interest. The main implications of the
present publication for the ICRP framework of radiological protection
are as follows.
GLOSSARY
α/β value or ratio A measure of the curvature of the cell survival curve. The α/β value is also
the dose at which the linear and quadratic components of cell killing are
equal. For tissues, the α/β value is a measure of their sensitivity to
changes in dose fractionation. In vivo, the α component describes the
dose–response slope at low doses, which is often considered independent of
dose rate, but it is likely that it can be modified in chronic radiation
scenarios by cell renewal and cell competition processes. The β component
describes the increase in slope at higher doses due to cumulative damage,
which is repairable during fractionated or low-dose-rate
exposures. The risk of an adverse health effect, i.e. the probability or rate of the
occurrence of a particular health event (e.g. disease incidence) over a
specific period. The energy imparted per unit mass by ionising radiation to matter at a
specific point. The SI unit for absorbed dose is joule per kilogramme (J
kg−1), and its special name is gray
(Gy). Bone marrow contains the cell systems for the formation of blood cells,
starting from the pluripotent haematopoietic stem cells (HSCs) to the mature
blood cells. Increased resistance of cells or tissues to radiation following a priming
dose, or adjustment to radiation exposure that enables an organism to retain
viability, and maintain fertility and normal functional stability of all
tissues, organs, and systems under the conditions of chronic exposure. The
principal criterion of radiation adaptation is increased radioresistance
(tolerance) of the organism and the cells of its critical
organs. A benign tumour of glandular origin. Adenomas can grow from many organs
including the colon, adrenal glands, pituitary gland, thyroid, prostate,
etc. Although these growths are benign, they may progress to become
malignant over time, at which point they are called
‘adenocarcinomas’. Acute myeloid leukaemia (AML)-associated chimeric gene/gene product resulting
from reciprocal chromosomal translocation t(8,21)(q22,
q22). A mode of cell death in which the cell nucleus displays characteristic
densely staining globules, and at least some of the deoxyribonucleic acid
(DNA) is subsequently broken down into internucleosomal units. Sometimes
postulated to be a ‘programmed’ and therefore potentially controllable
process. Cell division producing two different types of a daughter cell (e.g. a tissue
stem cell producing both a stem cell and a progenitor cell) (see ‘Immortal
strand hypothesis’). The annual disease incidence observed in a population in the absence of
exposure to the agent under study. A unique chronic myloid leukaemia (CML)-associated genomic
re-arrangement. Structure formed in the early gestation of vertebrates. It is preceded by the
morula. It possesses an inner cell mass (embryoblast) that subsequently
forms the embryo, and an outer layer of cells (trophoblast) surrounding the
inner cell mass and a fluid-filled cavity known as the ‘blastocoele’. The
human blastocyst comprises 70–100 cells. Abbreviations for an in-vitro assay of primitive, self-renewing
haematopoietic progenitor cells in marrow. Caretaker genes encode gene products that stabilise the
genome. In the context of radiobiology, cell death is generally equated with any
process that leads to the permanent loss of clonogenic capacity, often
termed ‘loss of reproductive integrity’. Cell death can also refer to
physical death through a variety of processes such as apoptosis, necrosis,
and autophagy, and also, sometimes, premature senescence and premature
differentiation. A point in the cell cycle at which injured cells are arrested and then
released after recovery to progress to the next phase of the cell
cycle. Multiple genomic rearrangements with sharply circumscribed regions of one or
a few chromosomes, crisscrossing back and forth across involved
regions. Cells that have the capacity to produce an expanding family of descendants
(usually at least 50). Also called ‘colony-forming cells’ or
‘clonogens’. The fraction of clonogenic cells that survive exposure to, or treatment with,
an agent that causes cell death. Only cells that are able to form colonies
(clonogenic cells) are considered to have survived the treatment (see ‘Cell
death’). Can be used to check whether or not the segregation of sister chromatids is
random. The family of cells derived from a single clonogenic
cell. Cell populations in which both function and proliferation can take place
alternatively in the same cells. An interval giving the lowest and highest estimate of a parameter that is
statistically compatible with the data. For a 95% confidence interval, there
is a 95% chance that the interval contains the parameter. Organic molecules with biological function, originally defined as being
polypeptides released from lymphocytes and involved in maintenance of the
immune system. These factors have pleiotropic effects on not only
haematopoietic cells, but also many other cell types. Often synonymously
termed ‘growth factors’. The total absorbed dose resulting from repeated exposures to ionising
radiation over a period of
time. A parameter in the multitarget equation for cell survival; the radiation dose
that produces, on average, one lethal event per cell and reduces survival to
e−1 (i.e. 0.37) of its previous value on the exponential
portion of the survival curve. A judged factor that generalises the usually lower biological effectiveness
(per unit of dose) of radiation exposures at low doses and low dose rates as
compared with exposures at high doses and high dose rates; includes dose
effectiveness factor and dose-rate effectiveness
factor. The absorbed radiation dose delivered per unit time and measured, for
example, in Gy h-1. Decreasing radiation response with decreasing radiation dose
rate. The lowest dose given by a single track of radiation to the nucleus of a
cell. Cells in the inner cell mass of the blastocysts, responsible for further
development of the entire embryo proper. Epigenetic changes consist of changes in the properties of a cell that are
inherited but do not represent a change in genetic information (e.g.
methylation effects). They influence the phenotype without alteration in the
genotype. Membranous tissue composed of one or more layers of cells, forming the
covering of most external and internal surfaces of the body and its
organs. Cytokine that regulates erythrocyte levels and stimulates late erythroid
progenitor cells to form small colonies of
erythrocytes. The additional risk (or rate) from radiation exposure above the underlying
(baseline) risk (or rate) of the disease. This is often expressed as EAR per
Gy or per Sv. The excess proportion (or percentage) of the rate of radiation-induced
disease in an exposed population divided by the rate of disease in an
unexposed population that has the same background risk factors (age, sex,
race, etc.). This is often expressed as ERR per Gy or per
Sv. A survival curve without a threshold or shoulder region, which is a straight
line on a semi-logarithmic plot. Can be used to identify stem cells using particular cell-surface
markers. The dose per fraction of radiation is the total dose divided into a
particular number of fractions. A very large number of extremely small dose
fractions becomes equivalent to low-dose-rate exposure. Very low dose rates
protracted over long durations are called ‘chronic
exposures’. The dependence of the iso-effective absorbed radiation dose on the dose per
fraction. Usually quantified by the α/β value – a high fractionation
sensitivity is characterised by a low α/β value (see ‘α/β
value’). Identification of the broken ends of DNA caused by ionising radiation. H2AX
is one of several genes coding for histone H2A. H2AX becomes phosphorylated
on serine 139, then called ‘γH2AX’, as a reaction on DNA double-strand
breaks (DSBs). γH2AX is a sensitive target for looking at DSBs in
cells. A congenital defect characterised by a defect in the anterior abdominal wall
through which the abdominal contents protrude freely. Early phase in the embryonic development of most animals, during which the
single-layered blastula is re-organised into a three-layered gastrula
comprising the ectoderm, mesoderm, and endoderm. Master regulatory gene encoding for zinc finger DNA binding domain
transcription factors/regulatory gene/gene product for early haematopoietic
progenitor cells. Gatekeeper genes encode gene products that act to prevent growth of potential
cancer cells and prevent accumulation of mutations that lead directly to
increased cellular proliferation. Preservation of the structural and functional content of the genome of
cells. Cytokine that stimulates proliferation and differentiation of progenitor
cells into granulocytes. Cytokine that stimulates proliferation and differentiation of progenitor
cells into granulocytes, macrophages, and eosinophils. The special name for the SI unit of absorbed dose:
1 Gy = 1 J kg−1. An organic molecule that stimulates cell proliferation when it binds to its
cell-surface receptor. Often synonymously termed
‘cytokine’. Proportion of viable cells in active cell
proliferation. Tissues comprising a lineage of stem cells, transit (amplifying) cells, and
postmitotic (differentiated or mature) cells. Radiation with a high LET; for example, α particles, heavy ions, and
interaction products of fast neutrons. The ionisation density along the
trajectory is high. HR takes place in S and G2 phase cells to repair a damaged region
of DNA by copying the intact counterpart of the sister DNA strand. HR is
potentially error-free. Reduction in cell numbers in a tissue (e.g. due to radiation-induced
impairment of proliferation in early responding
tissues). The Hprt assay is an in-vitro mammalian cell gene mutation test. The
estimation of mutant frequency in the reporter gene, called ‘Hprt’ located
on the X chromosome, can provide information on the biological effect of an
absorbed dose in the cell type studied, and hence is a useful biodosimetry
tool. Asymmetric segregation of DNA strands to minimise the replication error in
the stem cells. The stem cell retains the template DNA strand after a round
of DNA synthesis, while the progenitor cells inherit the daughter
strand. The rate of occurrence of a disease in a population within a specified period
of time, often expressed as the number of cases of a disease arising per
100,000 individuals per year, or per 100,000
person-years. The steepness of the initial part of the cell survival curve, usually
indicated by the value of α in the linear-quadratic
model. The death of irradiated cells before they reach mitosis. Sometimes used as a
synonym for ‘apoptosis’. Mice in which one (or more) gene has been
inactivated. Cells that retain a DNA label through multiple rounds of cell
division. The cumulated risk of morbidity or dying of some particular cause up to a
given age. The rate of energy loss along the trajectory of an ionising particle, usually
expressed in keV µm−1. A dose–response model which is based on the assumption that, in the low dose
range, any radiation doses greater than zero will increase the risk of
excess cancer and/or heritable disease in a simple proportionate
manner. A statistical model that expresses the risk of an effect E
(e.g. disease, death, or abnormality) as the sum of two components: one
proportional to dose (linear term) and the other proportional to the square
of dose (quadratic term).
E = αD + βD2,
where D is dose. For cell survival: S
= exp − (αD + βD2). Radiation with low LET; for example, electrons, x rays, and γ
rays. A network of lymphatic vessels of varying calibre that collects tissue fluids
from all over the body and returns these fluids to the blood. Accumulations
of lymphocytes, called ‘lymph nodes’, are situated along the course of
lymphatic vessels. Cytokine that stimulates formation of macrophages from pluripotent
haematopoietic cells. A spheroid of cells derived from single mammary gland cells. A single cell
from a mammosphere can regenerate an entire mammary gland when transplanted
into a mammary fat pad. Min mice are genetically heterozygous for a germline truncating mutation of
the adenomatous polyposis coli (Apc) gene (i.e. ApcMin/+), and
develop multiple intestinal tumours and sporadic colon tumours in their
intestinal tracts within several weeks of birth. The Min mouse provides a
sensitive model for the study of tumourigenesis in irradiated
mice. Carcinogenesis model associated with a stepwise acquisition of mutations of
oncogenes and tumour suppressor genes, associated with a progressive loss of
external proliferative factors. Cell death associated with loss of cellular membrane integrity. Occurs, for
example, in anoxic areas of tumours, and is also a mode of cell death after
irradiation. People who have never smoked. Specific microenvironment in a tissue where stem cells reside and are
maintaind by various signals controlling proliferation and
differentiation. NHEJ repair takes place in non-cycling cells and variously in all cycle
phases, and is dependent on the repair proteins Ku70, Ku80, and
DNA-dependent protein kinase catalytic subunit (DNA-PKcs). People who do not smoke. Effects arising in cells that are not irradiated themselves but their
neighbouring or parental cells are irradiated, including bystander effects
and the induction of genomic instability. A gene that, when mutated or overexpressed, contributes to converting a
normal cell into a cancer cell. A slowly growing, benign skin tumour originating from primitive cells of the
hair matrix and hair shaft. A cytokine that induces growth of fibroblasts and is involved in wound
healing. Also acts on some epithelial and endothelial cells, and on
mesenchymal cells. Distribution applicable when the probability of an event happening is small
but the number of observations is large. The distribution of probabilities
runs from zero to infinity, and an important characteristic of the
distribution is that the mean equals the
variance. DNA repair occurring in a delay period after irradiation, before cell
division occurs. Cell death that occurs as the result of an active process carried out by
molecules in the cell. Examples include apoptosis, autophagy, terminal
differentiation, senescence, and even necrosis. Dose quantities which the Commission has developed for radiological
protection, and that allow quantification of the extent of exposure of the
human body to ionising radiation from both whole- and partial-body external
irradiation and from intakes of radionuclides. Any agent that increases the sensitivity of cells and tissues to radiation.
Commonly applied to electron-affinic chemicals that mimic oxygen in fixing
free-radical damage, although these should more correctly be referred to as
‘hypoxic cell sensitisers’. The sensitivity of cells to surviving exposure to ionising radiation. Usually
indicated by the surviving fraction at 2 Gy (i.e. SF2) or by the
parameters of the LQ or multitarget equations. Rat sarcoma proto-oncogene/small guanosine triphosphate hydrolase (GTPase)
protein/a basic cell signal transduction factor involved in cell growth,
differentiation, and cell survival. Retinoblastoma proto-oncogene with tumour suppressor function; limits
excessive cell growth. Molecular species such as superoxide, hydrogen peroxide, and hydroxyl
radicals. These species may function in cell signalling processes. At higher
levels, these species may damage cellular macromolecules (such as DNA and
RNA) and participate in cell death processes. At the cellular level: an increase in cell survival as a function of time
between dose fractions or during irradiation at low dose rates. At the
tissue level: an increase in tissue equi-effective total dose with an
increase in time interval between fractions and a decrease in dose per
fraction, or with irradiation at low dose rates. The ratio of a dose of a low-LET reference radiation (usually of
60Co γ rays or kilovoltage x-ray quality) to a dose of the
test radiation considered that gives an identical biological effect. RBE
values vary with the dose, dose fractionation, dose rate, and biological
endpoint considered. An expression of overall risk (i.e. including the radiation-induced risk)
relative to the underlying baseline risk. If the total risk is twice the
underlying baseline risk, the RR is 2. Ability of cells to divide many times (usually more than five) and thus be
‘clonogenic’. A permanent arrest of cell proliferation associated with differentiation,
ageing, or cellular damage. In flow cytometry, a subpopulation of cells that is distinct from the main
population on the basis of the markers employed; often cells that show
higher efflux of DNA-binding dye Hoechst 33342. By definition, SP cells have
distinguishing biological characteristics (e.g. they may exhibit
stem-cell-like characteristics), but the exact nature of this distinction
depends on the markers used in identifying the SP. The special name for the SI unit of equivalent dose, effective dose, and
operational dose quantities in radiological protection. The unit is joule
per kilogramme (J kg−1). SKY of the chromosomal content of cells using metaphase cells pretreated and
hybridised with a SKY probe mixture containing uniquely labelled
chromosome-specific probes. Long-term restoration of radiation tolerance that takes place on a time scale
of weeks to years, often associated with long-term intracellular
repair. Aggregates of cells produced in culture by multiple divisions of a single
cell. Spheroids can be produced from both normal and malignant cells, the
latter often being used as a model of tumour
metastases. Cells with an unlimited proliferative capacity, capable of self-renewal and
differentiation to produce all types of cells in a lineage system. Stem
cells are described as totipotent (producing all lineages), whereas their
daughter progenitor cells can be pluripotent (producing many lineages),
multipotent (producing several lineages), or unipotent (one
lineage). Stem cell characteristics that underlie self-renewal and the ability to
generate differentiated progeny. There are differing degrees of stemness
among more primitive and less primitive stem cells in some hierarchical
lineage systems. Malignant disease or heritable effects; the probability of an effect
occurring, but not its severity, is regarded as a function of dose without
threshold. DNA repair occurring during low-dose-rate exposure or between dose fractions,
resulting in less cell kill or less tissue reaction than if the total dose
was delivered acutely. Effects occurring in irradiated cells. The ends of chromosomes. One of the determinants of cellular senescence is
the loss of telomeres through DNA replication. Rapidly replicating cells
usually have telomerase activity to avoid telomere shortening. Shortening of
telomeres is also associated with genomic instability and
carcinogenesis. A malignant neoplasm consisting of elements of teratoma with those of
embryonal carcinoma or choriocarcinoma, or both; occurring most often in the
testis. The factor by which the equivalent dose in a tissue or organ
T is weighted to represent the relative contribution of
that tissue or organ to the total health detriment resulting from uniform
irradiation of the body. A cytokine that regulates many of the biological processes essential for
embryo development and tissue homeostasis, and which therefore plays a role
in the healing of some tissues. The effects of TGFβ may differ depending on
the tissue involved; for example, TGFβ inhibits the proliferation of
epithelial cells but stimulates proliferation, differentiation, and collagen
synthesis in fibroblasts. Differentiating proliferative cells that amplify cell production in a
hierarchical tissue. Chromosomal abnormalities that occur when chromosomes break and the fragments
rejoin to other chromosomes. There are many structurally different types of
translocations. The outer layer of the mammalian blastocyst after differentiation of the
ectoderm, mesoderm, and endoderm, when the outer layer is continuous with
the ectoderm of the embryo. A tumour suppressor gene, or anti-oncogene, is a gene that protects a cell
from one step on the path to cancer. When this gene is mutated to cause a
loss or reduction in its function, the cell can progress to cancer, usually
in combination with other genetic changes. The cumulative exposure from breathing an atmosphere at a concentration of 1
working level (WL) for a working month of 170 h. In the case of radon,
1 WL = any combination of the short-lived progeny of radon in 1 l of air
that will result in the emission of 1.3 × 105 MeV of potential α
energy. 1 WL = 2.08 × 105 J m−3.
ABBREVIATIONS
2D Two-dimensional
3D Three-dimensional
8-OHdG 8-oxo-2′-deoxyguanosine
53BP1 p53-binding protein 1
A bomb Atomic bomb
ABC ATP-binding cassette
ABCG2 ATP-binding cassette subfamily G member 2
ADC Adenocarcinoma
AGM Aorta/gonad/mesonephros region of the developing embryo
AKAP9 A kinase anchor protein 9 gene
AKT v-akt murine thymoma viral oncogene homologue
ALL Acute lymphoblastic leukaemia
ALP Alkaline phosphatase
AML Acute myeloid leukaemia
Ang-1 Tie2/angiopoietin 1
APC Adenomatous polyposis coli
AR Absolute risk
AT Ataxia telangiectasia
ATM Ataxia telangiectasia mutated
ATP Adenosine triphosphate
BADJ Bronchioalveolar duct junction
BAR Background absolute incidence rate
Bax Bcl-2 associated X protein
BASC Bronchioalveolar stem cell
BBD Benign breast disease
BCC Basal cell carcinoma
BCL-2 B-cell lymphoma-2 gene product encoded by like-named specific gene
BCNS Basal cell nevus syndrome
BCRP1 Breakpoint cluster region pseudogene 1
BEIR US Committee on Biological Effects of Ionizing Radiation
bFGF Basic fibroblast growth factor
BFUe Burst-forming unit, early marrow progenitor within erythroid lineage
B-lymphocyte Bursa of Fabricius-related, humoral-mediating lymphocyte
Bmi1 B-cell-specific Moloney murine leukaemia virus integration site 1
BMP-4 Bone morphogenetic protein 4
BRAF v-raf murine sarcoma viral oncogene homologue B1
BRCA Breast cancer susceptibility
BrdU Bromodeoxyuridine, synthetic nucleoside of thymidine
C57BL/6 Common inbred mouse strain (CC Little strain 57) with black coat (substrain 6)
CAFC/CACA Cobblestone area forming cell/cobblestone area cell assay
CALLA Common acute lymphoblastic leukaemia antigen
CaM kinase Calmodulin-dependent protein kinase
Car-S Carcinogen-susceptible
CBA Common inbred mouse strain (strong strain), myeloid-leukaemia-susceptible strain (other strains include RFM and C3H)
CBCC Crypt base columnar cell
C cell Calcitonin-secreting cell
CCSP Clara cell secretor protein
CD34+ Cluster differentiation surface molecule bearing cell (numbered 34)
CD34+Lin− Haematopoietic stem cell subtype, bearing CD34+ surface antigen, but lacking specific lineage-specific surface markers
CD49f Cluster differentiation surface molecule bearing cell (numbered 49)
CFU-C Colony-forming unit in culture (progenitor cell) assayed in vitro
CFUe Colony-forming unit with restricted erythroid differentiating potential
CFU-F Fibroblastoid colony-forming unit
CFU-gm Colony-forming unit with restricted granulocyte and monocyte differentiating potential
CFU-m Colony-forming unit with restricted monocyte differentiating potential
CFU-meg Colony-forming unit-megakaryocyte (megakaryocyte committed progenitor)
CFU-Ob Colony-forming unit-osteoblast
CFU-Sday 7 Colony-forming unit in spleen (progenitor) assayed in vivo (subscript = sampling day)
CHEK2 Checkpoint kinase 2
CI Confidence interval
cKit+ Cellular homologue of the feline sarcoma viral oncogene v-kit, tyrosine-protein kinase Kit, with binding affinity to stem cell growth factor receptor
CLL Chronic lymphocytic leukaemia
CLP Common lymphoid progenitor
CML Chronic myeloid leukaemia
CMP Common myeloid progenitor
CNS Central nervous system
COL1 Collagen type Iα1
CO-FISH Chromosome orientation-fluorescence in-situ hybridisation
COX Cyclo-oxygenase
CS Carnegie stage
CXC C-X-C motif chemokine – integral membrane receptors, specifically binding and responding to cytokines
CXCL12 Haematopoietic stem cell (HSC)-homing C-X-C motif chemokine ligand 12
DC Dyskeratosis congenita
DCAMKL-1 Doublecortin and calmodulin-dependent protein kinase-like-1
DCC Deleted in colorectal cancer
DDREF Dose and dose-rate effectiveness factor
DEF Dose effectiveness factor
DMBA 7,12-dimethylbenz(a)anthracene
DNA Deoxyribonucleic acid
DNA-PK DNA-dependent protein kinase
DNA-PKcs DNA-dependent protein kinase catalytic subunit
DREF Dose-rate effectiveness factor
DSB DNA double-strand break
Dsh Dishevelled
DUOX1/2 Dual oxidases 1/2
Dusp 2 Dual specificity protein phosphatase 2
E Embryonic day
EAR Excess absolute risk
EB Epidermolysis bullosa
EGF Epidermal growth factor
EGFP Enhanced green fluorescent protein
EGFR Epidermal growth factor receptor
EMA Epithelial membrane antigen
EMT Epithelial-mesenchymal transition or transformation
EpiSC Epidermal stem cell
EPU Epidermal proliferative unit
ER (O)Estrogen receptor
ERR Excess relative risk
ES Embryonic stem
ESA Epithelial-cell-surface antigen
eto Eleven twenty-one
FACS Fluorescence-activated cell sorting
FAP Familial adenomatous polyposis
FasL Fas ligand
FISH Fluorescence in-situ hybridisation
Flk1 Fetal liver kinase 1
FLM Fraction of labelled mitosis
Flt3 McDonough strain of feline sarcoma virus (fms)-related tyrosine kinase 3
FOXA2 Forkhead box A2
Fzd Frizzled
GATA-2/5 GATA binding protein 2/5
GBP10 Growth factor receptor-bound protein 10
G-CFU Granulocyte colony-forming unit (progenitor cell) within bone marrow
G-CSF Granulocyte colony-stimulating factor
GEMM-CFU Granulocyte, erythrocyte, monocyte, megakaryocyte colony-forming unit
GFP Green fluorescent protein
γH2AX Phosphorylated histone H2AX
GLI Glioma-associated oncogene homologue
GM-CFU Granulocyte/monocyte colony-forming unit (progenitor cell)
GM-CSF Granulocyte-macrophage colony-stimulating factor
GMP Granulocyte/macrophage progenitor
GNAS Guanine nucleotide-binding protein, alpha-stimulating activity polypeptide 1
GPA Glycophorin A
Gpr49 G-protein-coupled receptor 49
GTP Guanosine triphosphate
GTPase Guanosine triphosphate hydrolase
Gy Gray, a unit of ionising radiation absorbed dose
H3 Histone chromatin 3
HD Hodgkin’s disease
HER2 Human epidermal growth factor receptor 2
HH Hedgehog
HLX1 H2.0-like homeobox 1
hMSC Human mesenchymal (stromal)/stem cell
Hopx Homeobox only protein homeobox
Hoxb4 Homeobox b4
HPC Haematopoietic progenitor cell
HPP-CFU High-proliferative-potential colony-forming unit
Hprt Hypoxanthine-guanine phosphoribosyltransferase
HPV Human papilloma virus
HR Homologous recombination
HSC Haematopoietic stem cell
HSPC Haematopoietic stem and progenitor cell
hTERT Human telomerase reverse transcriptase
HZE High charge, high energy
IGF1 Insulin-like growth factor 1
IL Interleukin, lymphoid tissue cytokines
IL2Rgc Interleukin 2 receptor gamma chain
Ink4a/Arf Inhibitor of kinase 4 a/alternative reading frame
ISC Intestinal stem cell
iPS Induced pluripotent stem
JUN Regulatory gene/gene product; in combination with cFos forms an early-response transcription factor, AP-1
K Cytokeratin
KDR Kinase insert domain receptor,VEGFR2 surface receptor
KGF Keratinocyte growth factor
Klf4 Kruppel-like factor 4
K-ras Kirsten rat sarcoma viral oncogene homologue
KSC Keratinocyte stem cell
LacZ β-D-galactosidase
LCNEC Large cell neuroendocrine cancer
LEAR Lifetime excess absolute risk
LET Linear energy transfer
Lgr5 Leucine-rich repeat-containing G protein-coupled receptor 5
LI Labelling index
LIF Leukaemia inhibitory factor
Lin+/Lin− Haematopoietic cells bearing (+), or not bearing (−), cell lineage-specific surface antigens
LKB1 Liver kinase B1
LNT Linear no threshold
LOH Loss of heterozygosity
LQ Linear quadratic
LRC Label-retaining cell
Lrig1 Leucine-rich repeats and immunoglobulin domains 1
LRP5/6 Low-density lipoprotein receptor related protein 5/6
LSK Lin−/Sca-1+/c-Kit+
LSS Life Span Study
Lt-HSC Long-term renewal haematopoietic stem cell
LTR-Ic Long-term repopulating-initial cell; in-vitro, primitive, self-renewing progenitor cell
LYrs/TLSR5 Recessive maternally transmitted lymphoma resistance 2/thymic lymphoma suppressor region 5
MAPK Mitogen-activated protein kinase
MaSC Mammary gland stem cell
MCL-1 Myeloid cell leukaemia 1; a BCL-2 related, apoptosis-modulating cell-surface marker
M-CSF Macrophage colony-stimulating factor
M-CFU Macrophage colony-forming units
Mdm2 Mouse double minutes 2
MDS Myelodysplastic syndrome
MF Mutation frequency
MFH Malignant fibrous histiocytoma
MGMT O6-methylguanine-DNA methyltransferase
MI Mitotic index
Min Multiple intestinal neoplasia
miRNA MicroRNA (microribonucleic acid)
Mir-34 MicroRNA 34
MLL-AF9 Mixed-lineage leukaemia-acute lymphoblastic leukaemia 1 fused gene from chromosome 9 protein
MMTV Mouse mammary tumour virus
MNU Methylnitrosourea
MPP Multipotent progenitor
MR Marrow repopulating; identifies a self-renewing, proliferative haematopoietic cell type
MRA Marrow-repopulating activity; a feature of primitive, renewing haematopoietic stem cells
Mre11 Meiotic recombination 11
MSC Mesenchymal (stromal)/stem cell
Msi-1 Musashi 1
mTert Mouse telomerase reverse transcriptase
MUC1 Muchin 1
Myb Myeloblastoma proto-oncogene/transcription factor encoding gene; haematopoiesis regulating gene
Myc Proto-oncogene/ regulatory gene encoding for ‘master’ transcription factor
NBCC Nevoid basal cell carcinoma
NBN Nibrin
NCRP National Council on Radiation Protection and Measurements
NHEJ Non-homologous end joining
NIS Sodium/iodide symporter
NK-lymphocyte Natural killer cell-type lymphocyte
NOD-scid Non-obese diabetic severe combined immunodeficient mouse
Notch Mammalian gene homologue of Drosophila phenotype; conserved signalling pathway
NSC Neural stem cell
NSCLC Non-small cell lung cancer
NTRK1 Neurotrophic tyrosine kinase receptor type 1
NeuroD1 Neuronal differentiation 1
Oct3/4 Octamer-binding transcription factors 3/4
OPG Osteoprotegerin
OSCC Oxford Survey of Childhood Cancers
Osx Osterix
P4 Fourth position from the crypt base
p16 Cyclin-dependent kinase-inhibitor (number 2A)
p21 Cyclin-dependent protein kinase inhibitor (number 21); cell-cycle regulatory gene
p210 BCR/ABL fusion gene product
p53 Tumour-protein/tumour-suppressor-protein (number 53); cell-cycle regulatory protein/gene
PARP-1 Poly (ADP-ribose) polymerase 1
Pax5/8 Paired-box 5/8
PAX8/PPARG Paired box 8/peroxisome proliferator-activated receptor gamma fusion oncogene rearrangement
PCNA Proliferating cell nuclear antigen
PCR Polymerase chain reaction
PD Population doubling
PDGF Platelet-derived growth factor
PIG-A Phosphatidylinositol N-acetylglucosaminyltransferase subunit A enzyme
PI3K Phosphotidylitositol 3-kinase
PI3KCA Phosphotidylitositol 3-kinase catalytic subunit p110α
PLDR Potentially lethal damage repair
POU Pit-Oct-Unc
PR Progesterone
Pre-CFU Early immature ‘colony-forming unit’ type cell in marrow
Pre-GM-CFC Pregranulocyte/macrophage-colony-forming cell
Ptch1 Patched 1
PTEN Phosphatase and tensin homologue
PY-Sv Person year-sievert; a radiation exposure-based disease incidence rate
RAD51 Radiation 51 repair gene (number 51)
RANK Receptor activator of nuclear factor-κB
RANKL Receptor activator of nuclear factor-κB ligand
RARB Retinoic acid receptor β
RASSF1A Ras association domain family member 1A
Rb Retinoblastoma
RBE Relative biological effectiveness
RC Repair capacity
RET/PTC Rearranged during transfection/papillary thyroid carcinoma
Rho Phodamine; a fluorescent dye used in cell sorting
RNA Ribonucleic acid
ROS Reactive oxygen species
RR Relative risk
RTK Membrane receptor tyrosine kinase
Runx2 Runt-related transcription factor 2
SC Spontaneous cancer
Sca-1 Stem cell antigen 1
SCC Squamous cell carcinoma
SCF Stem cell factor; c-kit tyrosine kinase receptor ligand; haematopoietic progenitor growth factor
SCGBa1a Secretoglobin a1a
SCID Severe combined immunodeficient
SCLC Small cell lung carcinoma
SCN Solid cell nest
SE Standard error
SEER Surveillance, Epidemiology and End Results Program
SF2 Surviving fraction at 2 Gy
SHH Sonic hedgehog; mammalian homologue of Drosophila gene/phenotype
SI Small intestine
SKY Spectral karyotyping
SLAM Signalling lymphocytic activation molecule; a family of conserved haematopoietic stem cell surface markers
SLC Small light cell
SLDR Sublethal damage repair
SMO Smoothened
SNP Single nucleotide polymorphism
Sox2 Sex-determining region Y-box 2
SP Side population
SSB DNA single-strand break (within the DNA helix)
STAT Signal transducers and activators of transcription
St-HSC Short-term haematopoietic stem cell
SUFU Suppressor of fused
t(8,21) Translocation of chromosomes 8 and 21
T4 Thyroxine
TA Transit amplifying
TAR Total absolute incidence rate
TBI Total body irradiation
TEL/AML1 Translocation ETS-like leukaemia/acute myeloid leukaemia 1
Tg Thyroglobulin
TGFβ Transforming growth factor β
TGFBR2 Transforming growth factor β receptor 2
ThOX1/2 Thyroid oxidases 1/2
Thy-1 Thymus cell antigen 1
Tie2/Ang-1 Tie2, endothelia-specific tyrosine kinase receptor for angiopoietin 1
TIMP3 Tissue inhibitor of metalloproteinase 3
TITF-1 Thyroid transcription factor 1
T-lymphocyte Thymus-associated, cellular immune mediating lymphocyte
TPA 12-O-tetradecanoylphorbor-13-acetate
TPO Thyroperoxidase
TRAIL Tumour necrosis factor-related apoptosis-inducing ligand
TS DNA synthesis time
TSC Thyroid stem cell
TSH Thyroid stimulating hormone
TSHR Thyroid stimulating hormone receptor
TTF1 Thyroid transcription factor 1
UBB Ultimobranchial bodies
ULLC Undifferentiated large light cell
UV Ultraviolet
VCAM-1 Vascular cell adhesion molecule 1
VEGFR2 Vascular endothelial growth factor receptor 2
WAP Whey acidic protein
WHO World Health Organization
WL Working level
WLM Working-level month
Wnt Wingless-type mouse mammary tumour virus integration site family member; mammalian homologue of Drosophila wingless gene/phenotype
XP Xeroderma pigmentosum
XRCC3 X-ray repair cross-complementing group 3
XTT 2,3-bis-(2-methoxy-4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide
ZO-1 Zonula occludens 1
1. INTRODUCTION
1.1. Purpose of the report
(1) The risks of radiation-induced cancer have continued to dominate
the reasons behind recommendations on restricting radiation
exposures of workers and the public for many decades. Carcinogenesis
from radiation is considered to be a stochastic event, originating
in a single transformed target cell. Generally, the target cells are
considered to be the stem cells, and possibly some of their daughter
progenitor cells in each tissue. These target cells have
tissue-specific characteristics, and they reside in a
microenvironmental ‘niche’ that regulates their proliferation and
differentiation (see Fig. 2.1). Knowledge about
the stem cells and their regulation may help underpin extrapolated
risk estimates for different tissue and organ systems, and may also
help understand risk projections in different exposure
scenarios. (2) The location of target cells in different tissues is vitally
important regarding the risks of carcinogenesis from short-range
radionuclides and lightly penetrating radiation beams. In its
publications, the Commission has made various judgements and
assumptions about the location of these cells. In the skin, the
target cells are considered to be in the hair follicles as well as
in the basal interfollicular epidermis (ICRP, 1991). In the
respiratory tract, the target cells are considered to be in the
basal layer of the mucosa and in the alveoli of the lung (ICRP,
1995). In the intestine, the target cells are considered to
be near the bottom of the intestinal crypts, but there are
uncertainties regarding the carcinogenic potential of cells further
up the crypts (ICRP, 2007). In the skeleton, the target cells
responsible for radiation-induced osteosarcomas are considered to be
the osteoblasts in bone cavities, as well as mesenchymal stem cells
(MSCs) in the bone marrow (ICRP, 1995). The
information in the present report is likely to contribute further to
clarification of the locations of target cells in the tissues of
interest. (3) This report comprises a review of advances in knowledge of the
biology and radiation response of stem cells and progenitor cells in
the context of the tissue microenvironment, in relation to
mechanisms of radiation carcinogenesis. Recent progress in stem cell
biology and radiobiology is described, including tissue
architecture, the dynamic nature of tissue maintenance, stem cell
radiosensitivity and renewal, dose-rate effects, and age dependence.
This information is evaluated for deducing the implications of the
role of stem cells in the mechanisms of carcinogenic risk as a
function of acute and chronic radiation dose, including projections
of risk for short-range radiations. (4) In order to form a basis of scientific knowledge regarding
stochastic radiation effects in different tissues and organs, a
series of organ systems were selected as examples with known
radiation-induced risks. As such reviews were not available, a
detailed series of annexes were produced for these individual
examples. Not all organ systems were covered because of the enormity
of such a task. Selection was made on the basis of importance for
radiological protection purposes, and the extent of available
radiobiological knowledge and interest. Firstly, those with the
highest ICRP tissue weighting factor of 0.12 were selected, which
include bone marrow, breast, digestive tract, and lung. Among these,
haematopoietic tissue is best studied for the stem cell aspect
(Annex A), and there is good knowledge of stem cell locations and
other hierarchical aspects of tissue turnover in the digestive tract
(Annex D). The presence of stem cells is well proven for the mammary
gland (Annex B), but that for the lung has yet to be fully
characterised (Annex E). Thyroid was selected for its strong age
dependence of susceptibility to radiation carcinogenesis (Annex C).
Bone was selected because it is the best model system for cancer
induction by internal emitters (Annex G). Skin was selected for its
simple hierarchical tissue structure that facilitates understanding
regarding tissue cell types, cancer subtypes and susceptibility to
radiation (Annex F), and the high, but rather uncertain, incidence
rate for skin cancer (ICRP, 1991, 1992,
2007; Annex F). The format of the annexes was designed to
present similar information for each tissue in order to facilitate
comparisons.
1.2. Radiation carcinogenesis models and stem cell biology
(5) Current understanding of mechanisms for radiation-induced cancer
relies on the multistep model of Armitage and Doll (1954),
and its extension to the molecular changes postulated by Vogelstein
et al. (1988). However, these mechanisms have limited
impacts as they were not evaluated thoroughly in the context of
radiation carcinogenesis. Improvements in this situation can be
attained by positioning the linear-no-threshold (LNT) and relative
risk (RR) models in the context of the multistep model. Improvements
in this situation may arise from the identification of target cells
within tissues, understanding of radiation responses of these cells,
and the kinetics of turnover and cell-to-cell interactions that
control the potential of these target cells to progress in the
carcinogenic process. These issues are important for external
exposures, particularly for low chronic exposures, but they take on
additional importance when considering risks from internal emitters,
particularly those with short ranges in tissues. The newly emerging
stem cell biology provides a real opportunity to clarify these
issues and provide a foundation to better understand the dose and
dose-rate effects, and to better define the target cell locations
for dose effects from internal emitters. It may also contribute to
reasons supporting the way to transport risks across different
populations. (6) The number, sensitivity, location, and renewal characteristics of
target cells are important biological parameters with respect to
carcinogenesis from radiation. Many people consider that the target
cells for carcinogenesis are the tissue stem cells and some of their
daughter cells. The number of stem cells is often ill defined and
estimated using various assays in different tissues in experimental
animal systems. Stem cell radiosensitivity varies both within and
among tissues. For example, there are both resistant slow-cycling
and sensitive fast-cycling stem cells in spermatogenic epithelium,
sensitive apoptosis-susceptible cells and more resistant clonogenic
cells in intestinal mucosa, and more sensitive stem cells in
haematopoietic than in epithelial tissues. The microenvironment
within tissues differs in a variety of parameters including
cell-to-cell and cytokine-mediated signalling. (7) The location of target cells in different tissues is vitally
important regarding the risks of carcinogenesis following radiation
doses from short-range radionuclides and lightly penetrating
radiation beams. Stem cells are often found residing in a specific
microenvironment known as a stem cell ‘niche’. The target cells for
leukaemia are considered to be haematopoietic stem cells (HSCs) and
possibly some of their daughter cells, and their niches are situated
within small bone cavities as well as near sinusoids. The density of
some stem cell types near the central blood vessel is lower compared
with density closer to the bone surface, although there is some
evidence that the reverse is true regarding their renewal ability.
However, the general conclusion is that the primary stem cells are
protected in hypoxic niches near the bone surface (Parmar et al.,
2007). These uncertainties in target-cell type and
position have hampered attempts to provide accurate projections of
radiation risks pertaining to short-range radiations, based on risks
for homogeneous external irradiations. Moreover, the stem cell niche
and its morphological features need to be better defined for many
tissue types such as mammary gland and thyroid. (8) For the purposes of radiological protection programmes, ICRP
applied the LNT model in combination with the RR model to the risk
assessment of acute and chronic exposures. For the assessment of
chronic exposures, a dose and dose-rate effectiveness (DDREF) factor
is applied additionally. The numerical value of DDREF remains
controversial, and various proposals have been made (BEIR VII,
2006). Evaluation of stem cells and tissue turnover is
hoped to help clarify the appropriateness of using LNT for low-dose
extrapolation and a DDREF value for adjustment of risk for chronic
exposures. The latter is dependent, to some extent, on the uncertain
repair capacity and the renewal/retention rate of the target cells
in tissues. Although DNA repair has been discussed regarding its
role in the DDREF, the cell renewal rate is considered to be more
important for cancer types where the target cells are not only the
stem cells, but also their daughter progenitor cells, which are
subject to more vigorous renewal. In addition, recent studies have
shown that, in some cases, the damaged tissue stem cells are subject
to constant removal in competition with undamaged stem cells. This
would be expected to result in a lower risk from chronic exposures
than that predicted simply by the cumulative dose to the initial
complement of target cells, which are being slowly replaced by new
cells. (9) The sections below discuss the following questions and topics in
the light of recent progress in stem cell biology. Which are the
target cells for carcinogenesis, and where are they located? Does
the LNT model fit with considerations of stem-cell-based radiation
carcinogenesis, and if so, how? Is the current DDREF value supported
by information concerning stem-cell-based radiation carcinogenesis?
What are the mechanisms related to stem cell response, and do those
mechanisms help explain the tissue differences in sensitivity to
radiation carcinogenesis? What could be an underlying mechanism
related to stem cells for the age-dependent sensitivity to radiation
carcinogenesis, and hence risk?
2. GENERAL FEATURES OF TISSUE STEM CELLS
2.1. Cell division and differentiation in adult tissues
(10) Adult tissues fulfil various bodily needs with specialised sets
of functional cells. Functional cells are terminally differentiated
cells with a limited (often non-existent) capacity to proliferate,
and they have to be replaced when needed. Adult tissues can be
divided into two types: flexible (complex) and hierarchical. In
flexible tissues, such as liver, the functional cells have the
potential to divide and can increase their number in special
circumstances, especially when injured, but they are thought to be
supplied by stem cells under normal conditions. In contrast, the
turnover rate of the hierarchical tissues is high, where functional
cells are lost rapidly from the body. For the supply of a large
number of various functional cell types, the hierarchical tissues
have a discrete lineage consisting of stem cells, progenitor cells,
and differentiated cells. The hierarchical tissues are the main
target for radiation carcinogenesis, and they are the focus of this
report. (11) Stem cells differ between early embryogenesis, fetal
development, and after establishment of adult tissues. The stem
cells of the embryonic stage are embryonic stem (ES) cells that are
totipotent and have the capacity to differentiate into all tissue
types of later organogenesis. In the fetal stage, stem cells are
lineage-committed to a certain extent in order to contribute to
specific tissues of the adult stages. In these two stages of early
life, stem cells mainly undergo symmetric division to produce two
equal daughter stem cells, associated with an increase in the size
of the embryo and the fetus (Section 2.3.5 and Fig. 2.5). In contrast, adult
tissue stem cells are mainly fully committed with restricted
differentiation capabilities, and they divide in an asymmetric
fashion, although they undergo symmetric division to various
extents, especially when repairing tissue injuries. (12) The study of adult tissue stem cells started as early as the
1960s. Haematopoietic progenitor cells (HPCs, originally considered
as stem cells) were first identified in bone marrow as those cells
capable of forming splenic colonies after intravenous injection into
lethally irradiated mice (Till and McCulloch,
1961). Early studies focused on the regenerative capacity of
tissue progenitor cells, but later in-vivo studies revealed many
important characteristics of tissue stem and progenitor cells
including asymmetric division (Potten and Loeffler,
1990). (13) Asymmetric division of adult tissue stem cells produces both a
stem cell and a progenitor cell. Progenitor cells are also called
‘transit amplifying (TA) cells’, especially for epithelial tissues.
Although the terms ‘progenitor cells’ and ‘TA cells’ are used
somewhat differently in the context of the haematopoietic system,
these two terms are used interchangeably in the present report. The
progenitor cells divide further to increase in number, and
differentiate into functional cells that are eventually lost by
senescence from the tissue after serving their required functions.
In this general scheme of tissue turnover, the stem cells are
frequently quiescent while progenitors divide more rapidly with
limited proliferative capacity. This differential role sharing
between tissue stem cells and progenitor cells is the strategy for
life-long preservation of stem cells, by minimising
replication-mediated mutations while supplying a large number of
cells to the functional compartment of a tissue by vigorous division
of progenitor cells. The progression from stem cells to
differentiated cells is usually unidirectional, but can be
reversible under certain conditions such as when stem cells are lost
for some reason where the vacant stem cell niche becomes occupied by
a neighbouring stem cell or by a dedifferentiated progenitor cell.
The latter scenario was demonstrated for germ cells of fruit flies
and mice (Cheng
et al., 2008; Barrocal, 2009). The
balance between the production and loss of cells maintains the
bodily tissue mass. (14) Hierarchical tissues contain three cellular compartments: the
stem cell compartment, the progenitor cell compartment, and the
functional cell compartment. The cells in the former two
compartments have a capacity to divide, while those in the last
compartment generally do not. The steps from stem cells to
differentiated cells vary from tissue to tissue. This somewhat
oversimplified scheme is depicted in Fig. 2.1. Importantly, while
each cell resides in a defined compartment, the population as a
whole consists of a generally unidirectional gradient of cells
between compartments. (15) The number of cell stages in a lineage in a tissue varies
greatly. In some tissues, stem cells supply the relatively limited
number of lineages, as in the case of the epidermis, while those in
other tissues supply a variety of lineages as exemplified by HSCs in
bone marrow. In addition, the number of divisions from the stem
cells to functional cells is variable among tissues as schematised
in Fig. 2.2
(Potten and
Wilson, 2007). As mentioned earlier, differentiated cells
are capable of cell division in flexible tissues such as liver,
thyroid, and lung when injured. (16) The rate of turnover varies markedly between tissues. The
14C dating technique was used to directly estimate
the rate of cell turnover in various tissues, and demonstrated that
the majority of cells in the brain remain for the entire life span
of individuals without being replaced, while those in the blood and
intestine turn over rapidly (Spalding et al., 2005).
As the technique measures the level of 14C isotope of all
the cells in a tissue, it is not sufficiently sensitive to exclude
the possible existence of a small number of cells that turn over. In
fact, stem cells are known to be present in adult brain (Quinones-Hinojosa
et al., 2007), but their contribution to the total
turnover of brain tissue is too small to be detected by this
technique. (17) The same technique demonstrated the annual turnover rate of fat
cells in humans to be 10% (Spalding et al., 2008).
The turnover rate varied with age, and in cardiomyocytes, it
decreased from 1% annually at the age of 25 years to 0.45% at the
age of 75 years (Bergmann et al., 2009). The turnover rate and the number
and location of various cell types in a tissue are believed to be
important determinants of tissue-specific risk of radiation
carcinogenesis by external exposures as well as by internal
exposures. In adult tissues, hierarchical tissues with higher
renewal rates are generally more sensitive to radiation
carcinogenesis (small intestine is an important exception) than
flexible/complex tissues with lower renewal rates. Also, the
turnover rate varies by the developmental stage and age of
individuals. The turnover rate is one of the major determinants of
the tissue difference and age dependence of radiation carcinogenesis
(Sections 2.5.4 and 3.6). Stem cell division and the maintenance of tissue dynamics. A niche is
the location for stem cells that divide to replace themselves or to
produce progenitor cells. The latter divide further, producing
functional mature cells. The functional cells have a limited life
span before they die, requiring replacement by more divisions in the
cell lineage. A diagrammatic representation of a stem-cell-derived cell lineage.
The number of cell generations in the lineage varies considerably
between tissues, with stratified and glandular epithelia having the
fewest amplifying divisions of dividing transit cells. For example,
in the large intestinal crypt, there are five to nine amplifying
cell divisions. Seven divisions would produce 128 postmitotic
maturing cells from 64 cells that had multiplied (proliferative
cells) from one stem cell. Hence, the percentage of stem cells in
the proliferative cells in this example would be 1/65 × 100 = 1.5%.
In breast or tongue epithelium, the lineage is short with likely
only one or two amplifying cell divisions. Two divisions would
produce four maturing cells from two proliferative cells arising
from one stem cell. Hence, the stem cells form 1/3 × 100 = 33% of
the proliferative cells. Source: Potten and Wilson (2007).
Reproduced with permission from Elsevier.
2.2. Functional identification and isolation of tissue stem cells
(18) Tissue stem cells are defined by their ability to self-renew and
to produce progenitor cells of particular lineages, which in turn
give rise to all the cells of the tissue. The serial transplantation
of marrow into irradiated mice demonstrates that the cells in the
colonies have a capacity to self-renew. Also, histological
examination demonstrates the presence of several cell types in a
colony, indicating that the transplanted colony-forming cells have
the capacity to differentiate into various haematopoietic cell
lineages. In addition, colony survival analyses enable the
estimation of radiosensitivity of the bone marrow stem/progenitor
cells. Studies of total bone marrow transplantation in combination
with limited dilution of the cell preparation estimated the fraction
of the stem cells to be 1/10,000 to 1/100,000 (Szilvassy et al., 1990;
Harrison
et al., 1993). (19) The functional identification of stem/progenitor cells by
in-vivo colony formation was applied successfully for other tissues
including the skin epidermis and intestinal epithelium (Withers,
1967; Withers and Elkind,
1969). Transplantation of monodispersed cells and tissue
fragments was also used to demonstrate the presence of stem cells in
mouse and rat mammary glands, and in rat thyroid (Daniel et al.,
1971; Clifton et al., 1986). The
colony assay was also used for quantifying the radiation induction
of carcinogenic events in rat mammary and thyroid clonogenic cells
(Watanabe
et al., 1988; Kamiya et al., 1995).
Although these in-vivo studies identified stem cells by their
functions, lack of isolated stem cells hampered further studies on
the nature of the cells.
2.2.1. Isolation of embryonic and adult tissue stem cells
(20) ES cells were established by explant culture of mouse
blastocysts in vitro (Evans and Kaufman,
1981; Martin, 1981). There
are 20–40 ES cells in the inner cell mass of the blastocysts,
and they are responsible for further development of the entire
embryo proper. ES cells can be identified by their
characteristic colony morphology in vitro of small and tightly
packed cells. These cells are immortal and express telomerase to
protect their chromosome ends (Carpenter et al.,
2003). Also, they are totipotent as ES cells contribute
to almost all tissues, except for the trophectoderm. This
totipotency was shared by embryonic carcinoma cells and
early-stage embryos, and ES cells readily form teratocarcinomas
when transplanted into allogenic sites (Rossant and Papaioannou,
1984). ES cells have been isolated from a variety of
mammalian species including humans (Thomson et al.,
1998). ES cells serve as the target of gene ablation
techniques and play a pivotal role in creating ‘knockout’ mice,
thus contributing to the study of gene functions in the tissue
and whole-body contexts. A variety of repair gene knockout mice
have been created which offers great opportunities for the
functional analyses of these genes in terms of radiosensitivity,
DNA repair, mutagenesis, and carcinogenesis in mice (Griffin
et al., 2005; Zha et al., 2007). In
subsequent sections of this report, however, ES cells are only
discussed when necessary because they have no direct relevance
to radiation carcinogenesis. (21) Isolation and in-vitro cultivation of cells facilitate
qualitative and quantitative analyses of tissue stem cells.
Long-term in-vitro cultivation of tissue stem cells was
accomplished for mouse haematopoietic cells in 1976, but only to
produce granulocytic cells (Allen and Dexter,
1976). Numerous attempts have been made since then which
resulted in identification of various cytokines for the growth
and differentiation of HSCs. HSCs and their progenitors can now
be maintained in a defined culture medium in the presence of
cytokines (Miller and Eaves, 1997). However, the degree of the
expansion of HSCs under in-vitro cultivation is still modest,
while such expansion through serial transplantation in
irradiated mice has been shown to be more than 8000-fold (Iscove and
Nawa, 1997; Sauvageau et al.,
2004). (22) Isolation and in-vitro cultivation of tissue stem cells are
now greatly facilitated by identification of various
stem-cell-specific marker proteins (http://stemcells.nih.gov/info/scireport/appendixe).
Among these, cell-surface marker proteins are particularly
useful to isolate tissue stem cells by fluorescence-activated
cell sorting (FACS) (Gundry et al., 2008).
FACS sorting of HSCs relies on specific cell-surface markers,
such as the cellular homologue of the feline sarcoma viral
oncogene v-kit (c-kit), tyrosine kinase receptor ligand [stem
cell factor (SCF)], stem cell antigen 1 (Sca-1), and CD34 (Shizuru
et al., 2005). Cell-surface markers (Table
3.2) are described in the annexes for tissue-specific
stem cells. (23) In addition to the cell-surface markers, a unique cellular
property of the side population (SP) phenotype is shared by many
tissue stem cells. This was exploited to enrich and isolate stem
cells using flow cytometry. When bone marrow cells are stained
with fluorescence dyes of Rhodamine 123 and Hoechst 33342, the
most weakly stained fraction is found to contain long-term HSCs.
This weak staining is associated with the low metabolic and
mitotic activities of quiescent HSCs (Bertoncello and Williams,
2004). High expression of the adenosine triphosphate
(ATP)-binding cassette (ABC) transporter, breakpoint cluster
region pseudogene 1 (BCRP1), and the resulting efficient efflux
of the dye are responsible for the SP phenotype of the quiescent
HSCs (Zhou
et al., 2001). (24) Isolation and cultivation of stem cells were also
accomplished by exploiting another unique feature. Neural stem
cells (NSCs) and mammary gland stem cells (MaSCs) exhibit the SP
phenotype as in HSCs. In addition, they form spheroids when
cultured in vitro (Annex B). A majority of cells died when a
single-cell suspension of the periventricular region of the
adult mouse brain was cultured in a medium supplemented with
epidermal growth factor (EGF). However, a small population of
cells (approximately 1%) grew and formed spheroids (Reynolds and
Weiss, 1992). Neurospheres, as they were called, were
enriched with NSCs and their progenitor cells. Neurosphere
formation is a valuable tool for quantitative assessment of
radiation effects on the neural cells, and such assay has been
conducted on rat spinal cord stem cells (Lu and Wong, 2005).
Spheroid formation was also noted for human MaSCs cultivated in
the presence of EGF and/or basic fibroblast growth factor (bFGF)
(Dontu
et al., 2003). As is the case for neurospheres,
mammospheres can also be serially passaged. In addition, a
single cell from a mammosphere can regenerate an entire mammary
gland when transplanted into a mammary fat pad (Shackleton
et al., 2006). In-vitro cultivation and mammosphere
formation offer a great opportunity in analysing radioresponse
and radiosensitivity of MaSCs, considering that mammary gland is
one of the highly susceptible tissues for radiation
carcinogenesis. (25) HSCs originate from endodermal tissues, and NSCs and MaSCs
from ectodermal tissues (Annex A). In addition, MSCs can be
propagated successfully in vitro (Chamberlain et al.,
2007). Thus, stem cells are expected to be isolatable
from almost all of the tissue types. One problem of the current
systems of FACS-mediated isolation and in-vitro cultivation of
stem cells is that the propagated population, whilst enriched
for stem cells, still contains their descendants. Isolation of a
pure stem cell population is yet to be accomplished. (26) Recent advances in stem cell research have demonstrated a
hierarchy of stem cells, especially for tissues with a rapid
turnover rate. For example, stem cells of the haematopoietic
system can be classified into at least long-term HSCs and
short-term HSCs; the former being more primitive than the latter
(Annex A). As for the small intestine, three types of stem cells
have been identified within the various cell populations present
in murine crypts (Annex D). There are two distinct stem cell
populations located at the fourth position from the crypt base
(P4), which are highly apoptosis-sensitive P4 stem cells and
highly radioresistant stem cells with mouse telomerase reverse
transcriptase (mTert) expression. The third type of stem cells
is rapidly cycling columnar cells at the crypt base, which are
positive with leucine-rich repeat-containing G protein-coupled
receptor 5 (Lgr5). In the lung, there are region-specific stem
cells (Annex E): bronchioalveolar stem cells (BASCs), Clara
cells, and Clara variant cells, the hierarchical interplay of
which needs further clarification. Human skin seems to have a
clearer structure of epidermal stem cells (EpiSCs), early
progenitors, and late progenitors, giving rise to basal cell
carcinoma (BCC), squamous cell carcinoma (SCC), and papilloma,
respectively (Annex F).
2.3. Radiosensitivity and DNA damage response of tissue stem cells
2.3.1. Basic strategies of tissue stem cells for the maintenance of genomic integrity
(27) Current knowledge on DNA damage response pathways was
summarised in Publication 99 (ICRP,
2005). DNA damage response is particularly important
for maintaining the genomic integrity of stem cells, as they
have to supplement all the functional cells in a tissue
throughout life. This can be achieved by minimising DNA damage,
cell divisions, and DNA replication, while maximising DNA damage
repair and eliminating damaged/mutated cells. Indeed, failures
of DNA damage response and DNA repair have been implicated
repeatedly in human premature ageing syndromes, many of which
can be viewed as the consequence of premature exhaustion of
tissue stem cells and human cancer-prone syndromes (Friedberg
et al., 2006). Abundance of antioxidants in stem
cells and the provision of a hypoxic micro-environment by the
stem cell niche contribute to minimising DNA damage for some
tissue stem cells. Quiescence is a feature shared by stem cells
of a variety of tissues that facilitates DNA repair and
minimises accumulation of replication-mediated mutation.
Alternatively, damaged stem cells can be eliminated by apoptosis
and progression to more differentiated compartments of a tissue,
such as progenitor and functional compartments. Competition of
stem cells for occupancy of the tissue stem cell niche is likely
to help in eliminating damaged stem cells.
2.3.2. Radiosensitivity of tissue stem cells
(28) The DNA repair capacity of tissue stem cells is reflected in
their radiosensitivity. Radiosensitivity of stem cells can be
assessed by clonogenic assays in vitro or in vivo using
transplantation or in-situ techniques. The results of such
analyses are shown for epithelial colony-forming cells (Fig. 2.3).
However, these assays cannot generally distinguish
radiosensitivity of stem cells and progenitor cells of
epithelial tissues, as both types can be clonogenic whenever
necessary. (29) A moderate capacity of potentially lethal damage repair
(PLDR) was demonstrated for clonogenic cells of rat mammary
gland, thyroid, liver, and bone marrow (Mulcahy et al., 1980;
Gould
et al., 1984; Jirtle and Michaelopoulos,
1985; Kamiya et al., 1991).
A general feature in the epithelial systems was the increase in
survival when the cells were left in situ for 24 h before
transplantation in vivo to measure colony formation. This had
the effect of shifting the survival curve to higher doses (Fig. 2.4).
The doses used in those PLDR experiments were generally
>6 Gy, so any changes in the ‘α’ component at lower doses
were not measured directly. However, the survival curve,
assessed after a 24-h recovery period, did have more curvature
(i.e. the α/β ratio was lower, indicating that the decrease in α
was more than any decrease in β). Examination of the effect of
further delay times before assay was performed using quiescent
liver in vivo, with hepatocytes transplanted into fat pads for
colony growth (Fisher et al., 1987).
For these particular cells, the change at 24 h was a decrease in
survival-curve slope, which continued decreasing to the maximum
11-month delay examined. Cell proliferation was insufficient to
explain the long-term reduction in radiosensitivity in terms of
a dose-dependent replacement of damaged cells. Although there
was a reduction in the frequency of cells with micronuclei
postirradiation, the magnitude of this decrease was relatively
small. Thus, the long-term increase in clonogenicity could only
be explained partially in terms of long-term repair of
chromosome injury, assessed by the production of micronuclei. In
addition, dose-fractionation experiments were conducted where
the hepatocytes were assayed for survival either early or late
after irradiation in situ (Fisher and Hendry,
1987). When the assay was delayed 10 months, the
value of α showed a tendency to decrease only slightly. The β
component showed the greatest decrease with time, and the α/β
ratio (1–1.6 Gy at 24 h) remained low but increased slightly to
1.9–2.1 Gy at 10 months. It should be noted that these tissue
systems (thyroid, mammary gland, and liver) have different
kinetic, lineage, and structural properties; as such,
postirradiation temporal changes may be tissue specific. These
clonogenic assays also showed that bone-marrow-derived
progenitor cells were more radiosensitive than epithelial
clonogenic cells (Fig. 2.4) (Hendry,
1985). The exact molecular mechanism of PLDR has not
been elucidated, but ataxia telangiectasia mutated (ATM), the
master gene of radiation damage response, has been suggested to
be involved in the process; however, a number of uncertainties
are yet to be resolved (Lobrich and Jeggo,
2005). (30) Recent technological advances enabled in-vitro propagation
of a relatively pure population of tissue stem cells that
permits direct analysis of tissue stem cells. Human
bone-marrow-derived clonally expanded MSCs were analysed
directly for their radiosensitivity. It was found that they were
more radioresistant than human lung and breast cancer cell
lines, and this was found to be due to a better antioxidant
capacity of the cells (Chen et al., 2006). As
discussed in Annex F, when tested in vitro, skin stem cells were
more radioresistant than progenitor cells (Harfouche et al.,
2010). (31) A summary of radiosensitivities in relation to cellular
stage in various hierarchical lineages is given in Table
2.1. Of note is the high radiosensitivity observed
for cells that predominantly undergo an apoptotic form of cell
killing [some intestinal stem cells (ISCs), and types A,
intermediate, and B spermatogonia]. Deduced survival curves for epithelial clonogenic cells in
various tissues. J1, J2, cells regenerating jejunal crypts; G1,
G2, cells regenerating gastric crypts; C, cells regenerating
colonic crypts; E1, E2, E3, cells generating macroscopic
epidermal clones; S1, cells regenerating spermatogenic tubules;
M, mammary tissue-forming units; T, thyroid follicle-forming
units. Cells transplanted at 24 h after irradiation in situ.
Source: Potten and Hendry (1983). Reproduced with permission
from Elsevier. Survival curves assessed in vivo by transplantation. Dashed lines
indcate cells transplanted immediately after irradiation, and
solid lines indicate cells transplanted at 24 h after
irradiation [or 3–4 h in the case of marrow
pregranulocyte/macrophage-colony-forming cells (Pre-GM-CFC)].
The separation between the dashed and solid lines for each
tissue type indicates potentially lethal damage repair; note the
positive effect in the case of mammary and thyroid epithelial
cells, and liver (hepatocytes), and the negative effect for
marrow spleen colony-forming units assayed in vivo (CFU-S)
considered due to a radiation-induced differentiation effect.
Source: Hendry (1985). Reproduced with permission from
Informa. Radiosensitivity in relation to hierarchical status (updated from
Potten
and Hendry, 1983). CFU-F, fibroblastoid colony-forming units; SI, small
intestine; CBCC, crypt base columnar cells.
2.3.3. Quiescence/dormancy of stem cells and DNA damage response
(32) Mouse embryos before the compaction stage were shown to
possess a unique p53-dependent S-phase checkpoint with no
activation of p21 (Shimura et al., 2002;
Adiga
et al., 2007). ES cells derived from the blastocyst
stage were shown to lack p53-dependent p21 activation after x
irradiation (Aladjem et al., 1998; Malashicheva et al.,
2000). The cells of these stages are dividing
rapidly, yet their DNA damage response differs considerably from
that studied in cells in vitro. (33) The DNA damage responses of adult tissue stem cells also
vary, especially when analysed in the context of tissue
microenvironments. For example, the P4 stem cells of mouse small
intestine are known for their high sensitivity to
radiation-induced apoptosis (Potten, 1977, 2004b;
Potten
et al., 2002) (Annex D). This type of cell death was
thought to be a mechanism to eliminate damaged cells and
therefore to maintain genomic integrity. Interestingly, some of
the P4 cells were the first to undergo DNA replication after
irradiation and pass through the p53/p21 repair pathway (Potten
et al., 2009). Apoptosis of the P4 stem cells occurs
in two phases: early p53-dependent apoptosis at 4.5 h after
irradiation was induced at doses below 1 Gy, and delayed
p53-independent apoptosis at 24 h after irradiation was induced
after higher doses such as 8 Gy (Dove et al., 1998).
Absence of the early p53-dependent apoptosis of crypt stem cells
in p53-null mice was restored when combined with the homozygous
loss of the DNA-dependent protein kinase catalytic subunit
(DNA-PKcs) gene (Gurley et al.,
2009). (34) In addition to P4 stem cells, Lgr5+ crypt base
columnar cells (CBCCs) were claimed to be the primary stem
cells. These cells divide rapidly at a cycling time of 24 h and
are less sensitive to apoptosis (Barker et al., 2007).
Extensive evaluation of past and more recent publications,
however, led to a conclusion that the P4 cells are the likely
stem cells of small intestine, and that the CBCCs are the
intermediate progenitors, possibly differentiating into a
variety of cell types including Paneth cells (Potten
et al., 2009). Furthermore, a very small
subpopulation of cells was identified with the expression of
mTert among the P4 stem cells of mouse intestinal crypts. These
mTert+ cells were quiescent and did not exhibit
apoptotic cell death even after 10 Gy (Montgomery et al.,
2011). They gave rise to all the cell types in the
small intestine, including Lgr5+ cells. Although
further experiments need to be conducted, mTert+ P4
cells are probably the most primitive stem cells of the small
intestinal mucosa. They are rare and quiescent, and are
insensitive to radiation-induced apoptosis. These features
suggest that sensitivity to altruistic cell death may not be a
universal radiation response of tissue stem cells. (35) Dormancy or quiescence is a general feature of many tissue
stem cells. For the quiescence of stem cells, the damage sensor
ATM serves an essential role for HSCs (Ito et al., 2004).
ATM−/− mice displayed premature depletion of HSCs
in the bone marrow. The level of reactive oxygen species (ROS)
was high in ATM−/− mice, and the high ROS level
activated p38 mitogen-activated protein kinase (MAPK) to force
the quiescent HSCs into cell cycling, which then resulted in the
exhaustion of HSCs (Ito et al., 2006;
Liu and
Finkel, 2006). Thus, cellular senescence eliminates
overly replicated stem cells, and quiescence is a mechanism to
maintain stem cell potential. Further studies demonstrated that
p53 and p21 are also involved in the quiescent state of HSCs
(Cheng
et al., 2000; Liu et al., 2010). (36) The DNA damage response of stem cells includes loss of
stemness (characteristics that underlie self-renewal and the
ability to generate differentiated progeny) which results in
differentiation. A recent study has demonstrated that melanocyte
stem cells undergo terminal differentiation in the niche when
exposed to radiation. ATM was found to be involved in this
terminal differentiation of melanocyte stem cells as the lack of
its function sensitises mouse skin to radiation induction of
hair greying (Inomata et al., 2009).
As in the case of quiescence, p53 is also involved in regulating
cellular senescence, in addition to ATM kinase (Vigneron and
Vousden, 2010).
2.3.4. DNA repair in stem cells
(37) DNA repair in tissue stem cells may well differ from cells
of other types as it is known that the DNA repair pathway is
dependent on their stage of differentiation. For example, the
nucleotide excision repair pathway was reported to be attenuated
in terminally differentiated cells (Rasko et al., 1993;
Nouspikel and Hanawalt, 2002; Hsu et al., 2007).
Mouse models defective in excision repair pathways exhibited
tissue-specific differences in mutagenesis and carcinogenesis,
demonstrating that the repair system may differ between stem
cells of different origins (Wijnhoven et al.,
2007). It is tempting to speculate that the DNA
repair pathways are responding to the demands of specific cell
types. Mice have been created with specific defects in damage
response and repair, and they provide excellent models to study
the role of tissue stem cells. As for the ageing of mice, the
effect of repair defects usually manifests more severely in
tissues of rapid turnover such as haematopoietic tissue (Park and
Gerson, 2005). HSCs from ageing mice were shown to
have upregulated expression of a series of stress-responsive
genes, suggesting a strong correlation of ageing, DNA damage,
and stress responses (Chambers et al.,
2007). These studies demonstrate that DNA repair and
damage responses play important roles for tissue stem cells to
stay quiescent and to preserve genomic stability. (38) Ionising radiation induces DNA double-strand breaks (DSBs)
that are repaired either by homologous recombination (HR) or
non-homologous end joining (NHEJ). HR is potentially error free
as it takes place in S- and G2-phase cells to repair
the damaged region of DNA by copying the intact counterpart of
the sister DNA strand. Sublethal damage repair (SLDR) was shown
to be dependent on Rad54, and therefore represents the repair
activity of HR (Rao et al., 2007).
NHEJ takes place in non-cycling cells and in all cell cycle
phases to a varying degree, and is dependent on the repair
proteins Ku70, Ku80, and DNA-PKcs. PLDR is the repair of
non-cycling cells, and thus represents the repair activity of
NHEJ. NHEJ consists of at least two repair systems: a more
accurate NHEJ, and a highly error-prone NHEJ pathway termed
‘alternative NHEJ’ (Symington and Gautier,
2011; Deriano and Roth,
2013). The alternative NHEJ pathway catalyses many genome
rearrangements, some leading to oncogenic transformation. A more
accurate NHEJ may be the default pathway in non-cycling
mammalian cells. (39) ES cells with either Rad54−/− or
Ku70−/− were found to be equally sensitive to
ionising radiation, showing the importance of both repair
pathways (Gu et al., 1997). In contrast, although adult
Ku80−/− mice and DNA-PKcs−/− mice are
sensitive to radiation, adult Rad54−/− mice only
exhibit hypersensitivity when combined with the DNA-dependent
protein kinase (DNA-PK) deficiency (Essers et al., 2000).
A series of mice defective in DSB repair have been generated and
characterised (Brugmans et al.,
2007). Among these mice, DNA ligase IV defective mice
exhibit premature ageing of HSCs (Nijnik, 2007). This
indicates that NHEJ is likely to be the major pathway of
radiation damage repair in adult tissue stem cells. The
dormancy/quiescence of tissue stem cells is essential,
especially for the tissues producing vast numbers of cells such
as haematopoietic and gastrointestinal (GI) tissues. In the
non-cycling quiescent tissue stem cells, the NHEJ pathway is the
only way to repair DNA damage. Thus, the NHEJ pathway, likely
the more accurate NHEJ pathway, is associated with PLDR, which
is defined operationally as the repair occurring in the
stationary phase non-cycling cells. Hence, it is reasonable that
tissue stem cells exhibit a large capacity of PLDR as shown by
in-vivo and in-situ clonogenic assays (Fig. 2.4) (Hendry,
1985). (40) Quiescence of stem cells poses two problems to the strategy
of tissue stem cells in maintaining the integrity of the genome.
Firstly, quiescent stem cells rely on NHEJ, but this repair
pathway is considered to consist of both a more accurate and an
error-prone pathway. In addition, DNA damage accumulates in the
quiescent stem cells, as demonstrated in HSCs by the occurrence
of spontaneous γH2AX foci in ageing mice (Rossi et al., 2007).
Radiosensitivity was tested for haematopoietic stem and
progenitor cells (HSPCs: Sca-1+, CD34−),
common myeloid progenitors (CMPs: Sca-1−,
CD34+), and granulocyte/macrophage progenitors
(GMPs: Sca-1−, CD34+). Their
radiosensitivity, as tested directly by the in-vitro clonogenic
assay, demonstrated that quiescent HSPCs were more
radioresistant than CMPs and GMPs, as expected (Mohrin
et al., 2010). However, the frequency of chromosome
aberrations after 2-Gy irradiation, as measured by SKY analyses,
was more than two-fold higher for the quiescent CD34−
HSPCs than the two CD34+ cell types. A
stem-cell-enriched haematopoietic cell population was reported
to exhibit similar chromosome sensitivity to that of peripheral
lymphocytes, but the cell population in that study was
CD34+. Therefore, a conclusive judgement cannot
be made at present from the study on the chromosome sensitivity
of CD34+ HSPCs (Becker et al.,
2009). (41) These results may suggest that PLDR executed by the NHEJ
pathways confers better survival of quiescent tissue stem cells,
but that it may also cause more chromosome mutations. Consistent
with this notion, the classic study of PLDR using V79 cells also
indicated that the irradiated cells kept in the stationary phase
exhibited better survival, but that the hypoxanthine-guanine
phosphoribosyltransferase (HPRT) mutation frequency stayed the
same irrespective of the holding time (Thacker and Stretch,
1983). However, a recent study indicated that
irradiated human diploid fibroblasts have less chromosome
aberrations when they are in the non-cycling G0 phase
than in the cycling G1 phase (Liu et al., 2010).
Further analyses need to be made to clarify the role of DNA
repair pathways in relation to colony survival and mutagenesis
of quiescent tissue stem cells.
2.3.5. Cairns’ hypothesis: a way to suppress replication-mediated mutation in tissue stem cells
(42) As high-dose radiation is not common in natural conditions,
coping with radiation damage is not of importance to the normal
maintenance and function of tissue stem cells. Rather, avoiding
naturally occurring mutagenic events is more important. Among
such events, DNA replication is an unavoidable source of
mutation. Indeed, a recent analysis of human cancer registry
data indicated that variation in lifetime risk of natural
(spontaneous) cancer could be explained by the total number of
stem cell divisions, and hence the total number of stem cell DNA
replications (Tomasetti and Vogelstein,
2015). An interesting hypothesis, the so-called
‘immortal strand’ hypothesis, was proposed by Cairns
(1975) in which asymmetric segregation of DNA strands
minimised the replication error in the tissue stem cells. The
stem cell retains the template DNA strand after a round of DNA
synthesis, while the progenitor cells inherit the daughter
strand. As the template strand is never replaced, the
replication error is minimised for tissue stem cells, while the
strands with possible errors are passed to the progenitors that
are eventually lost by differentiation/maturation into
functional cells. The evidence supporting the hypothesis was
presented in mice injected with 3H-thymidine at
infancy that carried the long-term label-retaining cells (LRCs)
in the stem cell region of rapidly proliferating intestinal
crypts (Potten et al., 2002). In order for the template
strand to stay on one side of the sister chromatids, there
should be no recombination in the tissue stem cells. This
requirement is likely to make stem cells ‘recombination minus’,
which then leaves NHEJ as the only legitimate repair system in
stem cells. Also, the template-strand hypothesis requires that
all sister chromatids with the template strands have to
segregate to the stem cell side of the spindle poles at mitosis,
with the one-to-one connection of old centrosomes and old
centromeres. However, such requirements still have to be shown
on a molecular basis. (43) Proponents of the immortal strand hypothesis rely mainly on
histological data indicating the presence of LRCs, and P4 stem
cells in mouse intestinal crypts were shown to retain
3H-thymidine or bromodeoxyuridine (BrdU) for a
long time (Potten et al., 2009). Asymmetric chromosome
segregation is currently not generalised for all tissue stem
cells, and the immortal strand hypothesis remains under critical
debate (Lansdorp, 2007; Rando, 2007). It was
demonstrated that asymmetric segregation of DNA strands does not
take place at least in HSCs and hair follicle stem cells (Kiel et al.,
2007; Waghmare et al.,
2008). However, strand-specific segregation was reported
on chromosome 7 in mouse neuronal cells (Armakolas and Klar,
2006). Furthermore, the chromosome
orientation-fluorescence in-situ hybridisation (CO-FISH)
technique was used to address the problem. The results indicated
that whereas the segregation of sister chromatids was random in
mouse fibroblasts and ES cells, the segregation in mouse colon
cells was not random (Falconer et al.,
2010). The asymmetric segregation of the immortal sister
chromatid requires suppression of the HR pathway, and this is
consistent with the reliance of tissue stem cells on the NHEJ
repair pathway rather than the HR pathway. Recently, a genetic
sequencing approach has been used to estimate the mutation
accumulation rate in healthy stem cells of human colon, blood,
and head and neck tissues (Tomasetti and Bozic,
2015). They observed that mutations accumulated in
those tissues at rates strikingly similar to those expected
without any protection from an immortal strand mechanism; this
serves as evidence against non-random segregation of DNA during
replication, suggesting that DNA is passed randomly to daughter
cells of stem cells. The immortal strand hypothesis is likely to
remain controversial until further evidence becomes
available.
2.4. Ageing and exhaustion of tissue stem cells
2.4.1. Mortal nature of tissue stem cells
(44) Tissue stem cells divide and replenish the tissue for the
entire life of an individual. Tissue stem cells were once
thought to be immortal and immune to cellular senescence. One of
the determinants of cellular senescence is the loss of telomeres
through DNA replication. Rapidly replicating cells therefore
usually have telomerase activity to avoid telomere shortening
(Blasco,
2007). Indeed, tissue stem cells were reported to
possess telomerase activity, and this has been confirmed in a
variety of tissue stem cells (Harrington, 2004).
However, mouse and human HSCs lose telomeric DNA with serial
passages of cells in vitro and in-vivo ageing of animals (Vaziri
et al., 1994; Allsopp et al., 2001).
Thus, adult tissue stem cells are unlikely to be immortal,
although their division potential is enormous (e.g. as shown in
bone marrow and intestine). Telomerase-deficient mice were
reported to exhibit pronounced ageing with atrophy in rapidly
proliferating tissues such as bone marrow, intestine, and testis
(Lee
et al., 1998; Rudolph et al., 1999).
There is now ample evidence to demonstrate that telomere
shortening takes place in stem cells of an ageing body (Flores
et al., 2006).
2.4.2. Telomere length of stem cells
(45) As discussed, tissue stem cells have limited telomerase
activity, so erosion of telomere ends through rounds of DNA
replication is inevitable. Therefore, tissue stem cells have to
possess mechanisms to prevent the loss of telomeres. One way is
to have efficient repair of DNA damage, and failure to do this
results in the loss of cells that necessitates compensatory
replication of stem cells in a tissue. Therefore, damage
checkpoints and DNA repair are important for tissue stem cells.
Also, quiescence in the well-protected microenvironment of the
tissue stem cell niche acts to promote the genomic integrity of
tissue stem cells. Indeed, tissue stem cells in their niche seem
to have the longest telomeres, as quantitative FISH has revealed
for mouse hair follicles, small intestine, testis, cornea, and
brain (Flores et al., 2008). It is interesting to note that
in this particular study, fluorescent signals for telomeres of
the Lgr5+ CBCCs were less than those of the P4 cells,
implying that the former is higher in the hierarchy of ISCs.
Thus, although tissue stem cells have telomerase activity and
the longest telomeres, they nevertheless shorten their length in
ageing mice. Shortening of telomeres can also be accelerated
when HSCs are forced to replicate by serial bone marrow
transplantation (Allsopp et al.,
2001).
2.4.3. Telomere shortening and carcinogenesis
(46) When telomeres become critically short, the chromosome ends
lose protection. The exposed chromosome ends are recognised as
DSBs. DSBs are frequently rejoined incorrectly to create
dicentric chromosomes, which either block cell division or break
apart to create new double-strand ends at mitosis. This
breakage–fusion–bridge cycle can induce genomic instability in
the cells, as human embryonic kidney cells undergoing
replicative shortening of telomeres were shown to exhibit
chromosome instability (Counter et al., 1992).
Telomere shortening and chromosome instability lead to other
consequences by activating DNA damage responses including focus
formation of phosphorylated histone H2AX (γH2AX), p53-binding
protein 1 (53BP1), meiotic recombination 11 (Mre11) complexes,
and phosphorylated ATM (Takai et al., 2003).
Resulting DNA damage responses culminate in activation of a
series of p53-mediated responses including apoptotic cell death
and/or cellular senescence (Karlseder et al.,
1999; d'Adda di Fagagna et al.,
2003). Senescence acts as a powerful block for
carcinogenesis, as telomerase-deficient mice were found to
resist chemical induction of skin cancer (Gonzalez-Suarez et al.,
2000). Cellular senescence and apoptosis led to loss
of stem cells, forcing them to undergo further replication in
order to maintain tissue homeostasis. This creates a vicious
cycle of the additional loss of stem cells in tissues. Erosion
of telomeres due to natural or forced replications thus leads to
stem cell exhaustion, which is a hallmark of ageing tissues. (47) Cellular senescence brought about by short telomeres may act
as a block to carcinogenesis. However, short telomeres and the
resulting genomic instability are associated with cancer
induction in mice and humans (Murnane, 2012).
Although mice lacking telomerase exhibit premature ageing
phenotypes, the same mice on the p53 null genetic background are
highly prone to developing epithelial cancers (Artandi and
DePinho, 2010). Interestingly, the p53-null allele
does not have to be homozygous, and heterozygosity was enough
for high rates of cancers in these mice. Also, a
p53+/− genetic background restituted the
premature ageing and stem cell exhaustion phenotypes of
telomerase− mice (Flores and Blasco,
2009). Thus, the status of p53 is the major
determinant of the two opposing outcomes of telomere erosion in
tissue stem cells; namely, loss of cellularity in ageing tissue,
and unregulated cell proliferation in carcinogenesis. With
respect to radiation carcinogenesis in humans, it is interesting
that secondary cancers occurred after radiotherapy in those
patients with malignant Hodgkin’s lymphoma who showed a
shortening of telomeres with a resulting increase in chromosome
instability (M’Kacher et al.,
2007).
2.5. Tissue stem cell niche
2.5.1. Stem cell niche
(48) In homeostatic conditions, a tissue stem cell in adults is
considered to divide asymmetrically to produce a stem cell and a
progenitor cell. This asymmetry of cell division requires
spatial asymmetry of the stem cell microenvironment, and the
stem cell ‘niche’ (Schofield, 1978)
provides such a cue for the asymmetry (Watt and Hogan,
2000). In the stem cell niche, stem cells attach proximal
to specific stromas, while the daughter cell positions are
distal to these stromas. Asymmetric expression of adhesion
molecules on one side of a cell assures the specific
interaction, which in turn results in signalling by short-acting
factors, mainly tissue-specific cytokines, to maintain the
stemness of the cells. In addition, a study on the germline stem
cells in the Drosophila testis demonstrated that an alignment of
centrosomes plays a critical role in the division of stem cells
perpendicular to the stroma (Yamashita et al.,
2007). In the Drosophila testis, the stem cells are
positioned proximal to the stromal cells in the germline niche,
and green fluorescent protein (GFP)-tagged centrosomes align
with the proximal to distal direction. This structural feature
of stem cells and their centrosomes is shared by mammalian
tissue (Fuchs et al., 2004). (49) Positional information of the stem cell niche is therefore
pivotal to the hierarchical structuring of cells in a tissue by
activating expression of a series of genes. This means that the
gene expression patterns and the fates of stem cells,
progenitors, and terminally differentiated cells could be
modified reversibly by changing the positional information, or
by directly modifying the expression of fate-determining genes.
In the Drosophila testis, progenitor cells were demonstrated to
become stem cells by taking over the vacant stem cell niche
(Cheng
et al., 2008). This dedifferentiation is associated
with a specific pattern of gene expression. Direct manipulation
of genes led to dedifferentiation and transdifferentiation of
cells as demonstrated by a case of paired-box 5 (Pax5)-deficient
mature B cells that dedifferentiate and then transdifferentiate
into T cells (Cobaleda et al.,
2007). An extreme case of dedifferentiation and
transdifferentiation is the conversion of mouse fibroblasts to
induced pluripotent stem (iPS) cells by ectopic expression of
four stem-specific genes: octamer-binding transcription factors
3/4 (Oct3/4), sex determining region Y-box 2 (Sox2), c-Myc, and
Kruppel-like factor 4 (Klf4) (Takahashi and Yamanaka,
2006). All of the recent studies thus indicate that
the fate of cells is reprogrammable.
2.5.2. Stem cell niche as a shelter
(50) The stem cell niche was shown to promote rapid establishment
of the stem cell pool after genotoxic insults such as radiation
exposure. Indeed, transplantation of MSCs rescues lethally
irradiated mice through their HSC niche modulating activity
(Lange
et al., 2011). In the steady state, the stem cell
niche was shown to provide a shelter from various genotoxic
stresses. Tissue stem cells have to sustain themselves for the
entire life span of an individual, and they have a number of
strategies to circumvent genotoxic stresses. One of the
strategies to escape from ROS is to rely on the intracellular
antioxidants, as in the case of MSCs (Chen et al., 2006).
Another strategy is to stay quiescent in a low oxygen
environment, and the stem cell niche of HSCs is such an example
(Suda,
2007). (51) There are three types of niche for HSCs in the bone marrow:
an osteoblastic niche, a vascular niche, and a medullary niche
(Annex A) (Shiozawa and Taichman, 2012). HSCs in the
osteoblastic niche are in close association with osteoblastic
cells; thus, two of the most important targets of radiation
carcinogenesis are sharing the same microenvironment in a tissue
(Calvi
et al., 2003; Zhang et al., 2003).
HSCs are quiescent, and this state of cells is dependent on
their residence in the niche. The Tie2/angiopoietin 1 (Ang-1)
signalling between the two cell types regulates quiescence of
HSCs in the osteoblastic niche (Arai et al., 2004).
Quiescence is a property shared by stem cells in other tissues,
including EpiSCs (Nishikawa and Ozawa,
2007). Also, many types of tissue stem cells, such as
those of neural, mammary, mesenchymal, and adipose tissues, as
well as ES cells, favour a hypoxic condition for their stable
persistence in vitro and in vivo (Ivanovic et al., 2000;
Danet
et al., 2003; Ezashi et al., 2005;
Zhu
et al., 2005; Lin et al., 2006;
Grayson
et al., 2007). The osteoblastic niche of HSCs is
poorly vascularised, and some HSCs were shown to be stained
strongly by hypoxia-sensitive pimonidazole, suggesting that the
oxygen concentration of the niche is less than 2% (Parmar
et al., 2007). HSCs residing in the osteoblastic
niche are those of primitive and less committed types, and
hypoxia is favoured to protect cells from endogenous ROS. A more
recent view of the vital ‘haematopoietic niches’ is that they
consist of perivascular spaces formed by both mesenchymal
stromal and endothelial cells which predominantly are situated
near trabecular bone surfaces (Morrison and Scadden,
2014). The role of protection given by the stem cell
niche has to be considered when assessing the effect of
radiation on tissue stem cells. (52) Although hypoxia in the niche microenvironment is important,
it is interesting to note that an appropriate level of ROS is
also required for the maintenance of genomic stability in ES
cells in culture (Li and Marban, 2010).
The frequency of chromosome aberrations analysed in ES cells
decreased with decreasing levels of oxygen, and further
reduction of ROS by antioxidant treatment increased chromosome
aberrations in the cells. This increase was associated with the
depletion of expression of repair-related genes when ROS were
completely absent.
2.5.3. Stem cell competition for residence in the niche
(53) In contrast to the classic concept of asymmetric cell
division of stem cells, a number of studies revealed that stem
cells often divide symmetrically to produce two stem cells, or
to produce two committed stem cells/progenitor cells (Fig. 2.5).
In the former case, the surplus of stem cells can compete for
residence in the niche, and inferior stem cells are eliminated.
In the latter case, a vacant stem cell niche is created with two
committed cells leaving the niche. These processes lead to
turnover of stem cells, which were once thought to reside in a
tissue for a lifelong period. (54) The first evidence of such competition and turnover comes
from the analyses of mutant stem cells of intestinal crypts
(Potten
et al., 2009). For example, a mutant stem cell takes
over the entire crypt in 5–7 weeks in colon, and 12 weeks in
small intestine after mutagen treatment of mice (Loeffler
et al., 1993). Population expansion from a single
stem cell and the resulting monoclonality of the crypt have also
been shown by the lineage tagging of Lgr5+ stem cells
in intestinal crypts. From the pattern of the clone size
distribution, it was concluded that ISCs are equipotent in
replacing the neighbour, or being replaced by the neighbour, in
a neutral drift fashion without directional pressures. Thus,
ISCs are sometimes lost, or take over the entire crypt (Lopez-Garcia
et al., 2010). The lineage tagging method was applied
to a variety of tissues, and similar competition and resulting
turnover of stem cells were demonstrated in testicular germline
stem cells (Klein et al., 2010; Lopez-Garcia et al.,
2010) and skin stem cells (Clayton et al.,
2007). Three division patterns of tissue stem cells.
2.5.4. Competition of stem cells during establishment of the adult stem cell niche
(55) Although dependent on the tissue type, the adult stem cell
niche is usually established around the perinatal to postnatal
period. For example, the mouse intestinal tract is formed as a
simple tube of epithelial cells with high proliferative
activity. The first differentiation of these equipotent
epithelial cells, or fetal stem cells, is the formation of the
villi at embryonic day 15, which are necessary for absorption of
nutrition after birth. The formation of crypts that provide an
adult-type stem cell niche for the maintenance of tissue
proliferation appears on postnatal day 7 (Crosnier et al.,
2006). It is important to note that, whereas there are
few niches in the newly formed adult-type intestine, the number
of fetal stem cells in the fetal intestine is substantial. This
suggests strong competition of stem cells for the occupancy of
the niche. It is also interesting to note that the neonatal
crypt is occupied by a polyclonal population of stem cells, and
that monoclonality is established at approximately 2 weeks after
birth (Schmidt et al., 1988). This monoclonal conversion
demonstrates the occurrence of competition among fetal stem
cells for occupancy of the niche. (56) Haematopoietic development is first detected in the yolk sac
region of embryos, which then shifts to an
aorta/gonad/mesonephros site. A major site of haematopoiesis in
the fetal stage is the liver, while the site in the adult is the
red bone marrow. HSCs in the fetal liver migrate and settle in
the bone marrow niche (Orkin and Zon, 2008).
Although the migration of HSCs from liver to bone marrow is
generally preserved among mammalian species, there appear to be
distinct differences between the temporal patterns and major
sites of haematopoiesis in the developing fetus of mice and
humans; for example, in the developing human fetus,
haematopoietic activity increases steadily in bone marrow in the
second half of gestation. In contrast, in the mouse, the liver
is the dominant haematopoietic site from mid-gestation to birth,
and only very late in gestation and into the early postnatal
period is there active migration of HSCs from the liver to
skeletal sites. The process of ‘adult niche development’ appears
to take place at different times postnatally in the two
species. (57) Fetal liver HSCs differ in their characteristics from adult
bone marrow HSCs, in that the former are rapidly cycling while
the latter are in a state of quiescence. The gradual shift in
characteristics of HSCs takes place in the mouse 3 weeks after
birth. Redistribution to bone marrow is selective rather than
random, at least in the mouse, so that HSCs are deficient in
engraftment when they are transiting S/G2/M, while
those in G1 settle successfully in the bone marrow
niche (Bowie et al., 2006). Thus, homing into the adult
stem cell niche during the neonatal stage of development
functions as a selective process in the case of mouse HSCs,
where stem cells compete for residence in the niche with
retention of favourable cells and elimination of unwanted cells
(Fig.
2.6). Stem cell competition in the neonate. The competition is
particularly strong when the adult tissue stem cell niche is
established during the neonatal stage.
2.5.5. Effects of radiation on stem cell competition
(58) Irradiation affects the competition of HSCs for their
residence in the bone marrow niche. When two marked bone marrow
cell populations were mixed and transplanted to lethally
irradiated mice, both populations contributed equally to the
reconstituted HSCs. However, when one of the two populations was
first exposed to 1 Gy and subsequently mixed with the other
non-irradiated population prior to transplantation, the latter
population predominated in the reconstituted marrow HSCs (Bondar and
Medzhitov, 2010). In this competition, p53 played a
crucial role in sensing the stress of irradiation as the HSCs of
a p53+/− or p53−/− genotype exposed to
radiation were not outcompeted by the unirradiated wild-type
population. The competitiveness was solely dependent on the p53
level as bone marrow cells of mouse double minutes 2
(Mdm2)+/− mice with higher levels of p53 protein
were outcompeted by those from the wild-type mice, even without
irradiation. Somewhat similar observations were made using bone
marrow cells with the retrovirally transduced p53 with
functionally blocked dimerisation domain, in which the cells
were dominant over the wild-type cells after 2.5-Gy x-ray
irradiation (Marusyk et al., 2010).
Overall, stem cell competition for residence in the tissue stem
cell niche is sensitive to radiation stress which is sensed by
p53.
3. ROLE OF TISSUE STEM CELLS IN RADIATION CARCINOGENESIS
3.1. Role of stem cells in radiation carcinogenesis
(59) This chapter aims to depict subjects of importance for basic
processes of radiological health risk assessments, and the
development of guidelines deemed essential to the ICRP system of
radiological protection. Sections 1 and 2 deal with the mechanisms
of carcinogenesis in tissues with a special emphasis on the role of
tissue stem cells and other possible target cells with regard to the
LNT model and the RR model. The discussion then moves into the
radiation biology of stem cells, and the dynamic nature of stem cell
competition in the tissue stem cell niche. This tissue-level
dynamism is likely to play a role when considering the dose-rate
effect, which in the past has only been considered from the
viewpoint of cellular repair. Stem cell competition is a new
concept, but it can explain some features of radiation risk, and,
more importantly, the age dependence of radiation carcinogenesis.
Overall, this chapter is devoted to bridge the advancement of
knowledge of stem cell biology and issues of importance to radiation
risk assessment.
3.1.1. Multistage carcinogenesis
(60) Cancer in adulthood is envisaged to arise as a result of
accumulation of oncogenic mutations occurring mainly after
birth, whereas some childhood cancers are characterised by
mutations acquired during fetal development or inherited from
the parents. The incidence of cancer in adults, especially solid
cancer, exhibits a steady increase by age. Armitage and Doll
(1954) noted that this increase follows approximately
the fifth power of age, and proposed the multistage
carcinogenesis model. This was later supported by the molecular
analysis of human colon cancer, in which the conversion of
normal epithelial cells to adenoma and its progression to
carcinoma were shown to be associated with a stepwise
acquisition of mutations of oncogenes and tumour suppressor
genes (Fig. D.8 of Annex D; Vogelstein et al.,
1988). Acquisition of multiple mutations by
spontaneous processes takes a long time, which explains why
adulthood cancers arise late in life. (61) Cancers in childhood form a unique group of neoplasias that
occur before puberty, between birth and 15 years of age. In
contrast to adulthood cancer, childhood cancer such as
retinoblastoma was only found to require two steps (Knudson,
1971). This two-step carcinogenesis process can
explain why childhood cancer occurs with relatively short
latency in early life before puberty. The reason for the
difference in the number of steps (or mutations) for cancer of
adulthood and childhood onset is not fully understood. However,
it is likely that they differ in target-cell types, with some of
the former being fetal-stage primitive cells (the resulting
cancer often carries the suffix of ‘blastoma’) and the latter
being adult tissue stem cells and progenitor cells. (62) Animal experiments have shown that the process of stepwise
carcinogenesis can be categorised into four steps: the
initiation step with the irreversible change of a normal cell
into a preneoplastic state; the promotion step with the
proliferation and clonal expansion of initiated cells; the
malignant conversion step with an acquisition of neoplastic
characteristics of cells; and the progression step with further
accumulation of changes in cells with invasion to normal tissue
territories. Each step is associated with functional changes of
genes regulating cell proliferation, quiescence,
differentiation, senescence, and apoptosis (Perez-Losada
and Balmain, 2003). These changes in gene function
are often brought about by the genetic mechanism of mutation
induction and by the epigenetic mechanism of transcription
factors, chromatin modifications, DNA methylation, and
regulatory microRNA (miRNA) (Sharma et al., 2007).
However, the present report attempts to restrict the discussion
so as to focus on the multistage carcinogenesis model where
mutations are the determinant. This restriction is made in order
to simplify the discussion on the numerical aspects of radiation
carcinogenesis. For the same reason, mechanistic models
involving epigenetics and promotion/progression aspects are not
considered.
3.1.2. Target cells for carcinogenesis
(64) The target cells for carcinogenesis are considered to be the
tissue stem cells and their proximal progenitors (Reya et al.,
2001). This assumption is considered reasonable
because many characteristics of tissue stem cells resemble those
of cancer cells, and the resemblance became particularly strong
after the discovery of cancer stem cells in human leukaemia
(Lapidot et al., 1994). Cancer stem cells were also
found in a variety of solid tumour types arising in breast
(Dick,
2003), brain (Hemmati et al., 2003;
Singh
et al., 2003), prostate (Lawson and Witte,
2007), liver (Roskams, 2006), and
many other tissues. Cancer stem cells resemble normal tissue
stem cells in many respects. They have capacities to self-renew
and can initiate tumour formation when transplanted into
appropriate hosts. Cancer stem cells exhibit the SP phenotype
and are therefore resistant to chemotherapeutic agents, as are
normal tissue stem cells. Also, cancer stem cells in vivo are
often quiescent, in contrast to rapidly proliferating non-stem
cancer cells, because of constitutive upregulation of damage
checkpoints (Bao et al., 2006). (65) These similarities suggest that cancer stem cells may arise
from normal tissue stem cells and those of their proximal
progenitors, which are able to regain stemness. A stem cell
origin of cancer is suggested because, in some cases, the pool
size of tissue stem cells correlates with the risk of
carcinogenesis, and stem cells are the only cell type that has a
sufficiently long residence time in the body to accumulate
multiple mutations and gain malignant phenotypes (committed
progenitor cells do not have sufficient time) (ICRP,
2005). Also, regarding cancer risk in natural
(unirradiated) conditions, a strong correlation has been
reported recently between lifetime tissue-specific cancer risk
in (American) populations and the estimated lifetime total
number of stem cell divisions in each of a wide variety of
tissues (Fig.
3.1) (i.e. the product of estimates of the stem cell
number and the lifetime number of cell divisions per stem cell)
(Tomasetti and Vogelstein, 2015). This study has been
discussed variously regarding data selection, risk differences
between populations, and method of analysis (Sills,
2015). Nonetheless, apart from the increasing
knowledge of specific gene mutations associated with increased
susceptibility to particular cancers, the number of lifetime
stem cell divisions is the only other biological parameter to
date claimed to be associated with cancer risk in a normal human
population. Hence, this study highlights the potential
importance of replication-mediated somatic mutation in the
aetiology of spontaneous cancer (SC) rather than just the
traditionally entertained environmental and hereditary
components. This correlation has also been tested for radiation
risk, using either excess absolute risk (EAR) or excess relative
risk (ERR) per Gy. However, for the 10 body sites where
radiation risk values are available, no significant correlation
was found with estimated total stem cell divisions per lifetime
(Little et al., 2015). Hence, it is suggested from
current risk evidence that stem cell replications per se are not
a dominant mechanism of radiation risk among tissues. There is
as yet insufficient evidence among tissues concerning any effect
of other purported stem cell parameters on the above correlation
(e.g. DNA template retention, genomic instability, and stem cell
competition). (66) There are examples suggesting that lower ranking stem cells
and progenitor cells can also be the target for carcinogenesis.
The Cre-recombinase-mediated loss of the adenomatous polyposis
coli (APC) gene in Lgr5+ CBCCs, lower ranking to the
P4 stem cells, was found to result in the formation of full
adenomas (Barker et al., 2009). Also, in the human brain,
precancerous lesions are known to arise initially in the
differentiated compartment, and poorly differentiated
glioblastoma then evolve from well-differentiated astrocytoma
with a latency of 5–10 years (Klihues and Cavenee,
2000). At the molecular level, inactivation of
inhibitor of kinase 4a/alternative reading frame (Ink4a/Arf) was
shown to trigger dedifferentiation of astrocytes, and further
introduction of constitutionally active EGF receptor (EGFR)
conferred the malignant glioma phenotype to the cells (Bachoo et al.,
2002). (67) In haematopoietic malignancy, progenitor cells retain a
large proliferative capacity. Indeed, it was shown in mice that
long-lived progenitor cells may be in part responsible for
steady-state haematopoiesis during adulthood (Sun et al.,
2014). They are thought to be quite possibly a common
target for carcinogenesis, as many leukaemias and lymphomas
carry committed phenotypes of particular lineages. In addition,
progenitor cells have been shown to become leukaemia stem cells
by direct introduction of a fusion oncogene, mixed-lineage
leukaemia-acute lymphoblastic leukaemia 1 fused gene from
chromosome 9 protein (MLL-AF9) (Krivtsov et al.,
2006). The classic mouse model of radiation-induced
thymic lymphomas was shown to arise from the
CD4−/CD8− progenitors in the thymic
environment (Kominami and Niwa,
2006). It is interesting to note that proliferation
of haematopoietic malignancy is dependent on the niche;
proliferation of leukaemia cells requires a bone marrow
microenvironment, whereas proliferation of lymphoma cells needs
a lymph node environment. These two haematopoietic malignancies
carry phenotypes of stem cells and committed progenitor cells,
respectively. (68) Skin is another tissue where progenitor cells in addition to
stem cells can form cancer. As discussed in Annex F, three
cancer types are known in the skin; BCC, SCC, and papilloma
(Section F.5, Fig. F.5). EpiSCs were proposed to give rise to BCC
commonly found in people of European ancestry, early progenitor
cells were proposed to give rise to more malignant SCC, and late
progenitor cells were proposed to form benign papillomas.
Although this is an interesting and informative model, it is not
known how the early progenitor cells can resist the polarised
flow of cells without being discarded. (69) Other considerations are that the number of MaSCs is
determined by the level of hormones in utero, and this suggests
influences on subsequent risk of breast cancer after birth
(Trichopoulos, 1990). Similarly, body size of newborn
babies is known to correlate with risk of leukaemia (Caughey and
Michels, 2009). Involvement of insulin-like growth
factor 1 (IGF1) was recently implicated in this correlation,
suggesting that either a higher number of target cells or a
higher proliferation rate of HSCs is related to leukaemogenesis
(Chokkalingam et al., 2012). Finally, it is
interesting to note that the small intestine with high
proliferation is refractory to carcinogenesis in humans, while
the adjacent large intestine is prone to radiation
carcinogenesis, demonstrating that not all mutated stem cells
give rise to cancer. A high sensitivity of small intestinal P4
stem cells to apoptosis was proposed as a reason for a markedly
lower incidence of cancer in the small intestine compared with
the large intestine (Li et al., 1992; Potten
et al., 1992). Therefore, in addition to the number
of target cells, their behaviour also plays an important role in
carcinogenesis. Relationship between the estimated number of stem cell divisions
in the lifetime of a given human tissue, and the lifetime risk
of cancer in that tissue derived from the Surveillance,
Epidemiology and End Results Program (SEER) of the United States
National Cancer Institute. The dashed line is a visual
representation of the highly significant association between the
number of lifetime stem cell divisions and lifetime cancer risk
reported for different tissues in normal human (American)
populations by Tomasetti and Vogelstein
(2015). This was based on an analysis of data for 31
different tissues/cancer types and body sites including
osteosarcoma, medulloblastoma, ovarian germ cell, thyroid
medullary, pancreatic islet, duodenum, small intestine,
glioblastoma, gall bladder, oesophageal, testicular, lung
(smokers and non-smokers), acute myeloid leukaemia, chronic
lymphocytic leukaemia, hepatocellular, thyroid follicular, head
and neck, pancreatic ductal, melanoma, familial adenomatous
polyposis duodenum, colorectal, human papilloma virus head and
neck, hepatitis C virus hepatocellular, basal cell, Lynch
colorectal, and familial adenomatous polyposis colorectal
cancer. Note that in the present context of radiological
protection and using the RR model for the carcinogenic effects
of radiation exposure, a given dose of radiation could be
expected to increase (arrows and dotted line) the average
lifetime cancer risk among different tissues by a constant
factor on the log/log plot. Note also that risk transfer among
populations is accomplished using a combination of excess
relative risk and excess absolute risk models (see section
3.1.4).
3.1.3. Role of radiation in carcinogenesis
(70) The current model of radiation carcinogenesis assumes that
radiation acts as a mutagen, and gives possibly one or two
carcinogenic mutations to a target cell (ICRP, 2007). Radiation
is known to induce DSBs, especially those with clustered DNA
damage, which are prone to inducing large mutations such as
deletions and translocations. Induction of translocations
follows a linear-quadratic (LQ) dose response, while induction
of small deletions is likely to follow a linear dose response.
Deletion mutations inactivate tumour suppressor genes, while
translocations activate proto-oncogenes by juxtaposing them to
strong transcription promoter elements or making fusion genes
with oncogenic functions. These targeted actions of radiation
are the theoretical foundation for the LNT model, which is used
for projection of health risk at low dose and low dose rate. (71) The direct involvement of radiation in inducing the
oncogenic mutation has been tested experimentally by examining
the irradiated cells in culture for the presence of the
transcripts of rearranged genes responsible for leukaemias and
thyroid cancer. However, the dose required for such
rearrangements was found to be 50–100 Gy, which is extremely
high compared with the real situation where a few Gy induced
those cancers (Ito et al., 1993a,b).
In addition, the characteristic rearrangement of rearranged
during transfection/papillary thyroid carcinoma (RET/PTC)
elements in childhood thyroid cancer showed a strong age
dependence of occurrence, necessitating further studies to
address whether such translocation is attributable to radiation
exposures (Annex C). Therefore, a signature of radiation in
radiation-induced cancer has yet to be identified. (72) Recently, an interesting mechanism was discovered that
suggests a direct involvement of radiation in induction of
multiple carcinogenic mutations. In contrast to the stepwise
acquisition of mutations, genomic analyses of human cancer have
demonstrated that 2–3% of all cancers and approximately 25% of
bone cancers acquire multiple mutations by a single event (Stephens
et al., 2011). Chromothripsis, as it is called, is
multiple genomic rearrangements with sharply circumscribed
regions of one or a few chromosomes, crisscrossing back and
forth across involved regions. Involvement of a micronucleus in
the generation of chromothripsis was demonstrated recently
(Crasta
et al., 2012). Micronuclei are easily induced by
radiation through a single hit process, yet their role has been
implicated in cell death and not carcinogenesis. Thus, the role
of chromothripsis in radiation carcinogenesis needs to be
further investigated. (73) In addition to these targeted actions, radiation is known to
act in a non-targeted fashion that includes bystander effects
and the induction of genomic instability (ICRP, 2003, 2007;
UNSCEAR, 2006). Radiation has long been known to
induce transient changes in gene expression, but some of these
changes develop and persist for a long time after irradiation.
Indeed, the epigenetic mechanism of DNA methylation was shown to
underlie the non-targeted effect of radiation (Goetz et al.,
2011). It has been said that these effects may
increase risk if they cause non-lethal abnormalities in target
cells, or decrease risk if they are lethal (delayed reproductive
cell death). There is much evidence for the presence of such
effects after acute doses above approximately 0.5 Gy (UNSCEAR,
2006). However, the effects are, in most cases,
non-linear with dose, thus making extrapolations problematic,
and there are few studies at the dose levels relevant for
radiological protection. A pertinent example is the use of
C57BL/6 and CBA/Ca mice, resistant or susceptible, respectively,
to both radiation-induced myeloid leukaemia and chromosomal
instability in bone marrow cells, as well as exhibiting
differences in bystander signal generation after higher
radiation doses (Zyuzikov et al.,
2011). Over an acute and broad dose range of 1.7 mGy to
3 Gy, bystander effects in the bone marrow (assessed by p53
pathway signalling at 3 h after irradiation) were only observed
after doses above 100 mGy, and chromosomal instability at 30
days was found only after doses above 1 Gy. Although these
results were based on total marrow cells rather than stem cells
specifically, no evidence was found to support the potential
presence of these effects in the low dose range. In addition,
FISH analyses of clonally expanded T-cell populations with
characteristic translocations isolated from 50 atomic bomb
(A-bomb) survivors exposed to doses above 1 Gy did not show
excess levels of chromosome instability (Kodama et al., 2005).
These results are consistent with the views of UNSCEAR
(2010, 2012) that concluded:
There are now significantly more data available
on the biological consequences of low-dose radiation
exposure and non-targeted effects such as bystander
phenomena and transmissible genomic instability.
While mechanistic understanding of non-targeted
effects is improving, many studies remain primarily
observational. There are also reports of
differential gene and protein expression responses
at high and low radiation doses and dose rates.
These reports remain mixed in outcome and there is
little of the coherence required of robust data that
can be used confidently for risk assessment.
Similarly there is as yet no indication of a causal
association of non-targeted phenomena with
radiation-related disease and indeed, some may not
operate at low doses in vivo. A systems-level
framework should provide a useful guide for future
integration of mechanistic data into risk estimation
methods.
(74) Radiation is also known to change cell-to-cell and
cell-to-tissue interactions in a tissue’s microenvironment. In
the case of the mammary gland, radiation has been shown to
induce transforming growth factor β (TGFβ) from the matrix,
which plays a significant role in mammary carcinogenesis (Nguyen
et al., 2011). Irradiation modifies
stem-cell-to-niche interaction by giving selective advantage to
the Notch1 and p53-null stem cells for residence in the bone
marrow niche, contributing to the further development of
leukaemia (Marusyk, 2009; Marusyk et al., 2010).
These studies suggest a variety of roles played by radiation in
the induction of cancer. Nevertheless, the LNT model based on
the targeted mechanism of radiation is still used widely to
assess the risk at low-dose and low-dose-rate exposures. The
Commission considered that the LNT model was the best practical
approach and a prudent basis to managing radiation risk
commensurate with a precautionary principle (ICRP,
2005, 2007). Hence, LNT is
used for the purpose of establishing an appropriate radiological
protection programme. The model is generally consistent with the
epidemiological data of induction of cancer in radiation-exposed
human populations, in particular the A-bomb survivor study which
forms the ‘gold standard’ of human evidence, although there are
a few clear tissue-specific exceptions to the general rule and
other models can be equally applied in some cases.
3.1.4. Models and the risk of radiation carcinogenesis
(75) The radiation dose–incidence relationship regarding solid
cancers is fairly linear for A-bomb survivors, although there is
evidence for shallow curvilinearity in the dose range <2 Gy
when mortality was analysed (Preston et al., 2007;
Ozasa
et al., 2012). A majority of solid tumours among the
survivors occurred at a so-called ‘cancer-prone’ age, with a
similar age-incidence trend as in the non-exposed population,
but with the incidence being higher among the former in a
dose-dependent manner. These effects are consistent with the
assumption that radiation contributes one mutation out of
multiple mutations necessary for full malignancy (ICRP,
2007). Between three and seven mutations have been
suggested in the literature (Tomasetti et al.,
2015). Assuming that five mutations are necessary for
cancer development, SC develops from cells with four mutations
when any of them acquire one additional mutation by spontaneous
mutational processes, such as errors in DNA replication and
chromosome segregation augmented by internal/external mutagens.
Such spontaneous mutations were postulated to have been induced
through cell-division-associated DNA replication in stem cells.
On that basis, the lifetime risk of natural (spontaneous) cancer
of a particular tissue type was shown recently to be strongly
linearly correlated with the total number of stem cell divisions
when plotted in a log/log chart (Fig. 3.1; Tomasetti
and Vogelstein, 2015). Radiation-induced cancer is
also expected to arise from cells with the four spontaneous
mutations when one additional mutation is caused by radiation.
Assuming an RR value per unit dose of the same order for all
cancer types, one can predict that the lifetime risk of cancer
for a radiation-exposed population would be expected to shift
the linear correlation upwards in terms of risk, parallel to the
baseline linear correlation in the log/log plot (Fig.
3.1). (76) The overall incidence of cancer in the irradiated population
(OI) is the combination of the incidences of radiation-induced
cancer (RC) and SC. Therefore, the RR for a human population
exposed to radiation can be expressed by the equation
RR = OI/SC = RC/SC + 1. Altogether, the RR model, also called
the ‘multiplicative model’, is adequate for the risk assessment
of cancer of adulthood onset as radiation is expected to
increase the proportion of cells with five mutations. Thus, the
net effect is that radiation increases the risk of cancer
linearly with increasing dose, and proportionally above the
background incidence. However, these considerations indicate
that the role of radiation is relatively small compared with
that of other mutagenic factors. This raises the question of
whether or not it is reasonable to call a cancer
‘radiation-induced’, as radiation is not a major contributor to
carcinogenesis. One may call such a cancer
‘radiation-associated’ or ‘radiation-related’. However, the term
‘radiation-induced’ has been kept in the present report for
historical reasons and for simplicity of the discussion. In
contrast to the RR model, the absolute risk (AR) model
(sometimes called the ‘additive model’) assumes that the
radiation risk is independent of the background risk and depends
linearly on the dose. Today, both RR and AR models appear to be
a good fit for the risk of solid cancers in the A-bomb
survivors’ cohort, with coherent results about the modifying
effects of age at exposure and attained age (Ozasa et al.,
2012). (77) Apart from its role as a descriptive model, the RR model
implies an interesting practical application in the prospective
management of carcinogenic risk of people after exposure to
radiation. Postexposure risk management is thought to be
impossible because of the nature of radiation as an absolute
mutagen. However, as mentioned above, the RR model predicts that
the extent of the risk from a dose of radiation is proportional
to the background incidence. Hence, any action taken which
reduces the background incidence may also result in a reduction
in the radiation-related increase of EAR. An example of this can
be seen in the case of lung cancer from residential radon among
smokers and non-smokers. The ERR for radon is similar at
approximately 0.16 per 100 Bq m−3 for both smokers
and non-smokers, yet the background incidence at 75 years of age
differed approximately 25-fold between smokers and non-smokers
(Darby
et al., 2006). In mice, caloric restriction was shown
to reduce the spontaneous occurrence of myelocytic leukaemia.
Caloric restriction even after the radiation exposure was shown
to reduce the incidence in irradiated mice (Yoshida
et al., 1997). Application of caloric restriction and
other measures were reviewed recently for their various effects
on the incidence of radiation-induced cancer in experimental
animals (Oliai and Yang, 2014). These experimental findings
imply that human health-promoting actions such as stopping
smoking and improving dietary habits may lower not only the
background incidence but also the radiation-related increase of
some types of cancer. The benefit of such measures may be
expected, in theory, for the cancer types where the RR model
applies. Epidemiological studies could be designed to test
whether such measures effectively reduce the future occurrence
of cancer. (78) In contrast to solid tumours, the dose response for
leukaemia among A-bomb survivors, especially acute myeloid
leukaemia (AML), has a strong quadratic as well as a linear
component (Hsu et al., 2013). In addition, AML has a relatively
short latency time after radiation exposure. These suggest the
involvement of fewer mutations for this neoplasm. If the number
of mutations is small, radiation can act as an absolute
carcinogen to supply all such mutations. Similarly, childhood
cancer is known to be induced by a smaller number of mutations
with shorter latency than adult-onset cancers, as discussed in
Section 3.1.1. The AR model is likely to fit with these cancer
types, leukaemia and childhood cancer, as radiation is
sufficient to induce cancer regardless of the background
incidence in a particular target population. (79) From the above observations on current results, the
following paradigm can be derived concerning the role of
radiation, shape of the dose–response curve, latency, and the
radiation carcinogenesis model. Radiation induces mutations with
increasing dose in either a linear or an LQ fashion. The role of
radiation is to contribute only a few mutations (i.e. one or
two) to the carcinogenic process. Adult-onset solid tumours
require a relatively large number of mutations, and one mutation
by radiation has to be supplemented with additional mutations
from other processes before acquisition of full malignancy. This
requires a long latency after the radiation exposure. Such
cancers follow a linear dose response, fitting well with the LNT
model, and the risk is predicted by the RR model. In contrast,
leukaemia and childhood cancer require much fewer mutations
(e.g. two), which can be supplied by radiation alone. Such
cancers have an LQ dose response with relatively short latency,
and the associated risk can be best assessed by the AR model.
Such characterisation for the dose response, latency, and
radiation carcinogenesis model as summarised above is
oversimplified, but it offers a foundation for future studies
and for a better understanding of the mechanism. (80) An exception can be found for some cancer types, such as
radiation-induced childhood thyroid cancer. This appeared with a
short minimum latency of 4 years after the Chernobyl accident in
the presence of screening programmes. However, the linear dose
responses of childhood thyroid cancer were reported (Ron et al.,
2012). With the above rule in mind, it is tempting to
speculate that childhood thyroid cancer requires two mutations,
and a linear dose response with the short latency occurred for
those carrying the pre-existing RET/PTC rearrangement, with
radiation responsible for inducing the second hit necessary for
conversion of the cells to full malignancy. Indeed, the
rearrangement does exist frequently in benign nodules in the
thyroid, suggesting the possibility of such a mechanism (Marotta
et al., 2011). (81) The choice of risk model for radiological protection has
been made empirically when transferring the risk between two
populations differing in background risk (ICRP, 2007). A
comprehensive review is available on this important subject
(Wakeford, 2012). Currently, ICRP uses a number of
risk transfer models: a mixture of 50% RR model and 50% AR model
for all types of cancer, except for thyroid and skin (100% RR
model), breast and leukaemia (100% AR model), and lung cancer
(mixture of 30% RR model and 70% AR model). Among these cancer
types, breast cancer with the AR model is clearly an outlier of
the above simplified description. Its dose response for women
exposed under 40 years of age is linear with a slight upwards
curvature instead of expected LQ, the latency is relatively long
instead of just a few years, and the differences in background
rates between Japan and Europe or the USA are so large that an
AR model was considered more prudent and a better fit in
age-specific comparisons (Preston et al., 2007).
Thus, the working hypothesis as projected here is too simple to
predict the trend of cancer occurrence after radiation
exposures. However, the present consideration is useful to gain
a possible mechanistic insight into radiation carcinogenesis and
its implications for the choice of a risk model for the risk
transfer. (82) The risk and sensitivity of radiation carcinogenesis are
often evaluated by EAR and ERR. EAR quantifies the increment of
the cancer incidence rate due to radiation, while ERR describes
the relative increase over the background (control) incidence
rate due to radiation exposure. Both ERR and EAR not only differ
between age groups at the time of exposure but also change with
time since exposure. ERR is usually higher following exposure at
young ages, but the estimates decline with increasing years
since exposure or with increasing attained age. In contrast, EAR
at early years since exposure (i.e. when ERR is highest) are
usually smaller and increase along with an increase in years
since exposure, because the background rate increases sharply at
older ages. When attained age is the same, a younger age at
exposure in A-bomb survivors may give rise to a higher EAR for
some solid cancer types but not for others. (83) If it is assumed that cancer occurs as a result of
accumulation of a certain number of somatic mutations, the
occurrence of excess cancers as a result of inducing one
cancer-related mutation may be expressed as a function of three
factors: (a) the sensitivity of stem cells (and of proximal
progenitor or even differentiated cells for certain cancer
types) to radiation-induced mutation; (b) the retention of stem
cells that had undergone any mutation in any gene related to
cancer development; and (c) the population size of stem cells at
the time of exposure, which may, in the future, accumulate a
critical number of mutations (say four mutations) so that one
mutation added by irradiation may result in the elevated rate of
cancer. The number of stem cells with sufficient predisposing
mutations is expected to be proportional to the number with full
(say 5) mutations, which results in the background cancer
incidence rate. EAR is obtained by subtracting the background
absolute incidence rate (BAR) from the total absolute incidence
rate of cancer (TAR) in an exposed population (EAR = TAR – BAR).
In contrast, the excess relative increase in radiosensitivity is
quantified by ERR which is obtained by dividing EAR by BAR as in
the equation: ERR = EAR/BAR. EAR depends on BAR, especially for
cancer types where the RR model provides a better fit when
making risk transfer between different populations. A good
example of where EAR is strongly affected by BAR can be found in
radon-induced lung cancer in the smoking and non-smoking
populations (Darby et al., 2005,
2006). EAR was much larger for the smoking
population, but ERR was insensitive to the smoking
characteristic. A larger EAR for smokers may reflect a larger
BAR of that population, which has an increased number of
predisposed target stem cells. ERR was the same for both
populations. It is tempting to speculate that the
smoking-induced conditions did not markedly affect the
sensitivity of target stem cells to radiation induction of
carcinogenic mutations from radon. Overall, EAR and ERR
represent different mechanistic properties of stem cell
behaviour during radiation carcinogenesis (Section 3.6.2). (84) Representative values of EAR and ERR for some of the tissues
considered in this report are taken from ICRP (2007) and quoted
in Table
3.1. These values are standardised to the risk at 70
years of age following exposure to 1 Gy at 30 years of age. The
uncertainties in the values quoted are highest for breast and
lowest for thyroid, with stomach, colon, and lung in the middle.
The decline in EAR per decade after exposure at 30 years of age
is greatest for breast, lower and similar for thyroid, stomach,
and colon, with no change for lung. The values for ERR are of
the same order for all tissues quoted, with declines per decade
after exposure at 30 years of age being highest for thyroid,
lower for stomach and colon, zero for breast, and even a
positive increase for lung. The values for bone marrow, skin,
and bone surface are not quoted in Table 3.1 because the
values are not directly comparable in the same way. For bone
marrow, the preferred dose–response model is LQ (Hsu et al.,
2013), and hence, the coefficients cannot be compared
directly with those in Table 3.1 which are
based solely on a linear model. EAR dose coefficients for
leukaemia at 1 Gy (at 70 years of age after exposure at 30 years
of age) were 0.70 (linear term) and 0.71 (quadratic term) per
104 person-years for women, and 1.06 and 1.09 per
104 person-years for men, respectively. The
corresponding ERR values for men and women were 0.79 (linear
term) and 0.95 (quadratic term). Of note is the 17% increase in
ERR for lung per decade increase in age at exposure, compared
with decreases for the other tissues in Table 3.1. Skin is a
special case, as explained in detail in Section F.2.1. Risk
coefficients were calculated in a different way from those in
Table
3.1, and the values are very uncertain. For the
incidence of bone cancer in the Life Span Study (LSS) of A-bomb
survivors and assuming a linear dose response, EAR of 0.39 [95%
confidence interval (CI) 0.08–1.04] per 104
person-years Gy−1 for individuals at 70 years of age
exposed at 30 years of age, and ERR of 0.48 (95% CI 0.07–1.4)
were calculated (Preston et al., 2007).
A quadratic model has been suggested (UNSCEAR, 2006), and
in addition, a more recent analysis indicated that a linear ERR
model with a threshold at approximately 0.85 Gy appeared to be
the most plausible model from statistical and biological points
of view (Samartzis et al., 2013). A further EAR and LNT model
for induction of bone cancer has been developed based on
224Ra data (EPA, 2011), although
the LNT model is inconsistent with the 226Ra data
where a sigmoid response provides the best fit (Rowland
et al., 1978). In view of the above differences and
the complexities in making direct comparisons, in particular for
skin and bone vs the other tissues in Table 3.1, a more
detailed discussion of EAR and ERR values with reference to
stem-cell-based mechanisms would be very speculative (Section
3.6.2). (85) Values of a tissue weighting factor
wT are also quoted in Table
3.1 for completeness, because these formed part of
the rationale for the initial choice of tissues for this report
(Chapter 1). wT is one of the basic
elements of the ICRP risk model. It represents the relative
contribution of a tissue or organ to the total health risk,
mainly due to cancer, and used to weight the equivalent dose of
such organs (ICRP, 1991). The
weighted dose thus derived is the effective dose from which the
nominal risk for a dose can be calculated. The
wT values were revised in 2007
(ICRP,
2007), and again, it was confirmed that ERR of cancer
at 70 years of age for people acutely exposed to a unit dose at
30 years of age varies between tissues. ICRP recommended a
wT value of 0.12 for bone
marrow, colon, lung, stomach, and breast; 0.04 for bladder,
oesophagus, liver, and thyroid; and 0.01 for bone surface,
brain, salivary glands, and skin. The weighting factors refer to
total human health risk (i.e. mortality, which is a combination
of incidence and the probability that a particular cancer type
will be lethal, adjusted for quality of life and years of life
lost) (Box A.1, ICRP, 2007). The
detriment-adjusted nominal risk coefficient for cancer is based
upon lethality/life-impairment-weighted data on cancer incidence
with adjustment for relative life lost. Coefficients in the cancer-incidence-based excess absolute risk
(EAR) and excess relative risk (ERR) models, and tissue
weighting factors (wT) for some of
the different tissues considered in this report. M, male; F, female. Source: Tables A.4.7, A.4.6, and A.4.3 in Publication
103 (ICRP, 2007).
3.2. Stem cells and stem cell niche in radiation carcinogenesis
3.2.1. Stem cell radiation biology and radiation carcinogenesis
(86) Two factors are essential in carcinogenesis, especially for
cancers occurring in adults. One is the acquisition of oncogenic
mutations by the target cells, and the other is the retention of
such predisposed cells in the body to allow further accumulation
of mutations to gain full malignancy. Cellular radiosensitivity
and mutagenesis determine the former process, whilst the
dynamics of stem cells in the tissue micro-environment determine
the latter. (87) Radiobiological analyses of pure tissue stem cell
populations are still limited. Nevertheless, past studies and
emerging evidence suggest that the radiosensitivity of tissue
stem cells varies considerably between tissues and within a
tissue. How these differences relate to radiosensitivity for
carcinogenesis is not clear at present. In the case of the small
intestine, P4 stem cells are highly sensitive to apoptosis
induced by radiation at doses as low as 100 mGy, while the
telomerase-positive P4 stem cells, putatively the most primitive
stem cells of the intestine, survive even 10 Gy of radiation
(Para. 37). This high radioresistance is difficult to understand
because after such a dose, the cells are unlikely to remain
error free even with extremely efficient DNA repair. This raises
the possibility that the telomerase-positive P4 cells serve as
the reserve for emergency purposes, rather than serving as the
housekeeping supplier of lower ranking stem cells. HSCs are
among the most radiosensitive tissues, but the long-term HSCs
are more radioresistant than short-term HSCs (Para. A75). In the
skin, stem cells are more radioresistant than progenitor cells
(Section F4.1). In general, primitive tissue stem cells are more
radioresistant than their committed counterparts, but the
relevance of this to their sensitivity to radiation
carcinogenesis is not known. (88) Quiescence plays an important role in the radioresistance of
stem cells, as discussed in the previous section. In addition,
staying quiescent is the best way for tissue stem cells to avoid
replication-mediated mutations and exhaustion of tissue stem
cells. At the same time, quiescence results in accumulation of
spontaneous DNA damage as shown in mouse HSCs (Rossi et al.,
2007). In addition, quiescent cells have to be
dependent on NHEJ repair for coping with DNA damage, which
comprises both a more accurate and a highly error-prone NHEJ
pathway. Thus, the benefit of quiescence is dependent on the
trade-off of avoiding replication-mediated mutation vs taking a
chance on damage accumulation and resulting mutations. Another
way to avoid replication-mediated mutation is the unique
mechanism of immortal DNA strand retention and
stem-cell-specific chromosome segregation, which are likely to
operate in stem cells of the intestine and mammary gland. It
seems that stem cells have evolved multiple strategies to avoid
replication-mediated mutations. Radiobiological characteristics
of quiescent tissue stem cells in vivo need to be analysed
further in order to understand the cellular processes of
radiation carcinogenesis.
3.2.2. Tissue radiation biology and radiation carcinogenesis
(89) The steady-state maintenance of a tissue involves three
tissue compartments: the stem cell compartment, the progenitor
compartment, and the functional cell compartment. Once cells
move out of the first compartment, they will be discarded
eventually from the body, except for some progenitors of long
residential time in a tissue. Thus, stem cells are the major
target with a sufficiently long period of residence in the body
to accumulate mutations and thus acquire a malignant phenotype.
However, as discussed in Section 3.1.2, even stem cells are not
immune to being replaced by other stem cells as they are under
constant competition for residence in the stem cell niche. This
competition is due to occasional symmetric division of tissue
stem cells in which extra stem cells compete for residence in
the niche. (90) Analyses of Drosophila germ cells revealed E-cadherin as the
major gene involved in the competitiveness of the germline stem
cells in the niche (Zhao and Xi, 2010).
In this Drosophila system, rapid turnover of stem cells
functions as an efficient mechanism for the removal of aberrant
stem cells in the gonad. This is likely to be so for mammalian
stem cell systems where stem cells are in contact with various
niche cells. HSCs interact with osteoblasts through N-cadherin
and integrin, and receive Tie2/Ang-1 signalling to regulate the
quiescence (Suda, 2007). As for
Lgr5+ stem cells in the small intestine, adhesion
to Paneth cells was found to be essential for keeping the
stemness of the cells through essential signals received for
their maintenance (Sato et al., 2011).
Change in the expression of these genes, brought about by any
stress including radiation and/or by mutations in relevant
genes, is likely to affect the competitiveness of stem cells and
their subsequent occupancy in the niche. The effect of low-dose
radiation on the niche interaction of stem cells has not been
studied. Expression of E-cadherin was studied at >2 Gy with
differing results; downregulation in mammary epithelial cells
and upregulation in rat liver (Andarawewa et al.,
2007; Moriconi et al.,
2009). The consequence of such changes is expected to
result in the less-fit stem cells being competed out of the
niche. Indeed, it was reported that irradiation of bone marrow
with >0.5 Gy decreased the competitiveness of HSCs in the
recipient hosts (Marusyk et al., 2010).
This tissue microenvironment-based selection process is likely
to function as a tissue-based quality control, independent of
the molecular and cellular-based quality controls such as DNA
repair and apoptosis.
3.3. Dose-rate effect for radiation carcinogenesis
3.3.1. Cell-based considerations
(91) For radiological protection, the risk of radiation has to be
estimated, especially for low-dose and low-dose-rate exposures,
where the risk to be taken into consideration is that for
stochastic effects (ICRP, 2007). ICRP
estimates such risk by use of the LNT model and DDREF to adjust
for the conditions of low-dose and low-dose-rate exposures. ICRP
made a choice of 2.0 for the value of DDREF and is continuing to
use it (ICRP, 1991, 2007). The dose
response of cancer for acute exposures generally follows an
upwards concave curve, which can be described adequately by a
number of models. Two of the cell-based biophysical models have
been well received: the sublesion model of Kellerer and Rossi
(1972), and the repair saturation model of Goodhead
(1985). In particular, the sublesion model is widely
used with an LQ equation as below (UNSCEAR, 2006):
(92) The linear term of Eq. (3.1)
represents single-track events in cells that are supposed to be
dose rate independent. The quadratic term represents two-track
events that are subject to cellular repair, and therefore this
term becomes negligible at low doses and low dose rates. As for
the definition of low dose, any doses below 200 mSv were
proposed by UNSCEAR (1993) as, in
this dose range, the linear term dominates the magnitude of the
dose response. In addition, under the assumption of Eq. (3.1), the risk of radiation carcinogenesis
becomes identical for low dose and low dose rate. This is the
reason why low dose and low dose rate can be handled by one
factor of DDREF. Then, DDREF can be described by the following
equation: (93) The US Biological Effects of Ionizing Radiation (BEIR) VII
Committee (BEIR VII, 2006) applied Eq. (3.2) to the epidemiological data of the A-bomb
survivors and animal data using a Bayesian approach, and
obtained a DDREF value of 1.5. Furthermore, studies of radiation
workers exposed at low dose rates reported similar risk
coefficients for solid cancer as those observed for acute
exposures in the LSS, suggesting a DDREF value of 1.0 (Jacob et al.,
2009). UNSCEAR abandoned the use of DDREF and applied
the LQ model to the LSS data to directly obtain the
low-dose/low-dose-rate risk coefficient (UNSCEAR, 2006). The
risk coefficient thus obtained was consistent with the ICRP risk
values based on the LSS data adjusted by the DDREF value of
2.0. (94) The calculation-based derivation of the DDREF by BEIR VII
(2006) and UNSCEAR (2006) relies
on the validity of the LQ equation, and especially on the dose
rate independence of the linear term. At the cellular level,
there is ample evidence that the linear term of the mutation
induction rate by radiation is independent of dose rate.
However, at the tissue level, there are cases in which the slope
of the linear dose response of mutation and cancer decreases by
lowering the dose rate. Such a decrease can be expected when the
tissue-level elimination of aberrant cells by stem cell
competition is in operation.
3.3.2. Tissue-based considerations
(95) As discussed in Chapter 2, HSCs of mice irradiated with
≥1 Gy are effectively competed out by those of unirradiated mice
as demonstrated by cotransplantation into lethally irradiated
mice (Bondar and Medzhitov, 2010). This suggests that any
deviation from total stemness by irradiation leads to
elimination of such cells from the tissue. Stem cell competition
is regarded as a quality control at the tissue level to
eliminate phenotypically unfit cells. Thus, there are three
levels of quality control systems in a body: the molecular-level
quality control of DNA repair; the cellular-level quality
control of apoptosis; and the tissue-level quality control of
stem cell competition. (96) Stem cell elimination is expected to affect the linear term
of the LQ Eq. (3.1). One
example of such case is seen in the radiation induction of
germline mutations in mice. In the case of the male mouse, the
induction is a linear function of radiation dose, yet the
induction rate was lowered when the dose rate was reduced (Russell and
Kelly, 1982). In the case of the female mouse, the
dose-rate effect is extreme in that the linear dose response
became completely flat when the dose rate was decreased (Searle,
1974). Molecular and cellular quality controls cannot
explain such data. As discussed in the previous section, the
low-dose-rate sparing of mutation by DNA repair should only
affect the quadratic term of the LQ equation, and the linear
term should not be affected by the dose rate. In addition,
cellular quality control of apoptosis functions is less
efficient when the dose rate becomes low. Thus, loss of the
irradiated cells by whatever mechanism is likely to contribute
not to the dose effectiveness factor (DEF), but to the dose-rate
effectiveness factor (DREF). (97) Key issues in this topic are whether low-dose exposure
influences stemness, and if it does, how low can the dose that
affects stemness be? If the dose is as low as an elemental dose
of radiation, that is, the lowest dose given by a single track
of radiation to a nucleus of a cell, an interesting possibility
emerges. An elemental dose of 60Co γ rays is
approximately 1 mGy for the typical mammalian cellular nucleus
of 8 µm in diameter (Feinendegen, 1985).
Chronic exposures at a dose rate of a few mGy per year mean that
every cell in the body is hit by a track of radiation every few
months. This then makes a hit stem cell, at any time, compete
against surrounding non-hit stem cells within a niche. Thus, if
the elemental dose affects the stemness, the hit cell will be
preferentially lost by competition from the tissue stem cell
niche. This elimination theory lowers the linear term. Hence,
stem cell competition at the tissue level leaves an ample
possibility for a DREF value larger than unity, as in the case
of the current DDREF value used by ICRP.
3.4. Experimental animal studies to supplement human data
(98) There have been many studies of radiation-induced tumours in
animal systems that have contributed knowledge of dose–incidence
relationships and dose rate/fractionation effects. These have
focused mainly on AML and solid tumours of the Harderian gland,
pituitary, ovary, lung, breast, skin, and bone (NCRP,
2005; ICRP, 2006). Overall, it was considered by the
Commission that the animal tumour data tended to support the
hypothesis of a linear relationship of incidence vs dose at low
doses and at low dose rates, with no threshold dose (ICRP,
2005). However, there are various caveats to this conclusion,
notably that some of the tumour types studied in experimental
animals are not the most relevant types in humans, and there are
many strain and species differences in sensitivity. In addition,
over a broad range of doses, all neoplasms are not increased in
frequency within a given strain or species, and certain neoplasms
are decreased in frequency by irradiation. It has been argued that
‘From the diversity of observed dose–incidence relationships, it is
clear that no one mathematical model for relating incidence to dose
is universally applicable’ (Upton, 1985).
Nonetheless, the study of radiation-induced tumours in one or more
defined animal strains can help elucidate the molecular, cellular,
and tissue mechanisms that are fundamental to any animal species
including humans. (99) Concerning radiation-induced stochastic events in the
haematopoietic system (Annex A), AML is the type of leukaemia in
mice that has been studied in most detail. AML comprises less than
5% of all childhood leukaemias, but is one of three
radiation-inducible leukaemia subtypes: AML; chronic myeloid
leukaemia (CML); and acute lymphoblastic leukaemia (ALL). Regarding
the target cells for AML, recent evidence in mice suggests that the
initial radiation-induced AML stem cell may originate not only from
irradiated HSCs but also from multipotent progenitor cells and CMPs
(Hirouchi
et al., 2011; Section A.5). Also, hemizygous deletion of
dual specificity protein phosphatase 2 (Dusp2) in chromosome 2 may
contribute to the self-renewal potential of radiation-induced AML
stem cells. A detailed study of the ‘immortal strand hypothesis’ in
highly purified HSCs revealed that BrdU had poor specificity and
poor sensitivity as an HSC marker (Kiel et al., 2007). All
HSCs segregated their chromosomes randomly, and division of
individual HSCs in culture revealed no asymmetric segregation of the
label. Hence, HSCs did not retain older DNA strands during division.
The multistage theory of carcinogenesis and the importance of the
microenvironment in promotional events continue to suggest that the
target cells are likely to be the more slowly renewing cells in the
lineage, and those cells are generally the more primitive stem cells
in the population. Indeed, the correlation between chronic
radionuclide doses/location and the incidence of AML was closest for
target cells in the central marrow sinusoidal region (Lord et al.,
2001), which is a principal site of such primitive cells.
These are identified in marker studies as
CD34−CD48−CD150hi lineage
negative (Lin−)/Sca-1+/c-Kit+
(LSK). (100) As noted in Section A.1.1, the incidence of AML at very low
levels of exposure (from background levels to approximately 1 Gy) is
fairly linear with increasing dose. At doses from 1 to 2–3 Gy, the
dose–incidence curve tends to exhibit upwards curvature, and at
>3 Gy, the curve tends to bend over downwards. Also, DDREF values
estimated from such data vary considererably, but generally fall in
the range of 2–5 over the dose range to 3 Gy (UNSCEAR, 1993). Values at
doses <1 Gy are expected to be lower, mainly reflecting the
single-hit (α component) of the dose–response curve. In often-quoted
mouse experiments delivering continuous irradiation for 28 days at
0.04–0.11 mGy min−1 with a total dose of 1.5 Gy, the
incidence of AML of 5% (Mole et al., 1983) was
also produced by an acute dose of approximately 0.5 Gy (Mole and Major,
1983), which gives a DREF value of ≤3 at ≤1.5 Gy. In
addition, after higher doses of 3.0 and 4.5 Gy delivered at the same
low dose rate, the incidence of AML was inexplicably also 5%. This
led to the postulate of some ‘biological factors’ related to
stem/clonogenic cells influencing the response to dose protraction
(Mole and
Major, 1983), as discussed further in Section A.2.2. (101) A number of studies have been published on the effects of dose
and dose rate on mammary tumour induction in rodents (Annex B; UNSCEAR,
1993). Studies with Sprague-Dawley rats found
approximately linear incidence with increasing dose. The DDREF
values ranged from <2 to approximately 4 for dose rates varying
by a factor of ≥150 and for doses (at high dose rate) of
approximately 2–3 Gy (Shellabarger et al., 1966;
Gragtmans
et al., 1984). With BALB/c mice, the dose–incidence curve
at high dose rates was LQ up to approximately 0.25 Gy, and the
linear term was similar to that obtained after low-dose-rate
(0.07 mGy min−1) exposures (Ullrich, 1983).
Dose-fractionation studies showed a significant contribution from
the quadratic component at doses as low as 0.1 Gy
fraction−1, and acute daily fractions of 0.01 Gy gave
a tumour incidence similar to that observed after low-dose-rate
exposure to a total dose of 0.25 Gy in both cases (Ullrich et al.,
1987). The DDREF decreased with decreasing dose from 11.7
at 0.25 Gy, and would be predicted to be near unity at a dose of
between 0.1 and 0.01 Gy. The mouse studies showed that dose
fractionation/protraction to 25 days (i.e. 0.25 Gy delivered in
0.01 Gy day−1 fractions) produced the incidence of
mammary cancer predicted from the LQ analysis of the acute exposure
response. (102) Regarding age-at-exposure effects, the relatively restricted
window of carcinogen susceptibility that is evident during or around
puberty in both rodents and humans has been postulated to either
contain the greatest number of target cells or be a critical period
of stem cell regulation. There is a clear hierarchical lineage in
mammary epithelium and the many factors that control it. For
example, the CD24+ CD29high population in the
mouse mammary epithelium is highly enriched for cells with
multilineage and self-renewal potentials; the two properties that
define an MaSC. Also, there is evidence that epithelial LRCs in
mouse mammary gland divide asymmetrically and retain their template
DNA strands (Smith, 2005). However, the target cell origin of
radiation-induced breast cancers in terms of stem and progenitor
cells has not yet been elucidated. (103) Concerning the thyroid (Annex C), radiation induces both
papillary and follicular carcinomas, but the former type
predominates in humans whereas the latter type predominates in the
common rat model. Dose–incidence relationships for carcinomas
(mostly follicular) in 3000 female Long-Evans rats showed increasing
incidence with increasing x-ray dose from 0.8 Gy, flattening off at
the higher doses to 10.6 Gy (Lee et al., 1982).
However, adenomas were in the majority, and in contrast, these
showed a continuously rising dose–incidence curve. Hence, the curves
for the two tumour types appeared to be significantly different in
shape. In addition, concurrent studies with 131I and
detailed dosimetry (Lee et al., 1979) showed a
similar response to the high-dose-rate x-ray results for the
carcinomas, but there was a tendency towards a lower incidence of
adenomas at the higher doses of 131I compared with x
rays. If the adenoma yields are interpreted on an LQ basis, it can
be estimated that the solely linear component (not modified by dose
rate) may be at doses up to approximately the first dose point of
0.8 Gy, and a DDREF of approximately 2 may apply at approximately
2 Gy. However, for the aggregated yields of both tumour types, the
solely linear component could be higher and the DDREF lower, albeit
with large uncertainties. The similarity of tumour yields in the
rats at low doses of acute x rays and low-dose-rate 131I
is compatible with the human data (i.e. ERR for external radiation
exposure was compatible with ERR for internal radiation exposure
following the Chernobyl accident) (Section C.2). Although there is
evidence for the presence of a stem-cell-type lineage in the thyroid
epithelium (Section C.3), there is no knowledge of whether the
different tumour types originate from the same or different target
cells in the lineage. (104) Cancers of the stomach and colon (Annex D) in rodents are only
induced by high radiation doses (e.g. ≥8 Gy) (Boice and Fry, 1995), and
are very rarely found in the small intestine. A possible reason
invoked for the latter is the radiation-induced apoptosis of mutated
stem cells in the small intestine, which is prevented in the large
bowel by expression of the survival (anti-apoptotic) gene B-cell
lymphoma 2 (bcl-2) (Merritt et al., 1995).
Nonetheless, the stomach and colon are fairly resistant to cancer
induction. The multiple intestinal neoplasia (Min) mouse provides a
sensitive model for the study of tumourigenesis in irradiated mice.
Min mice are genetically heterozygous for a germline truncating
mutation of the Apc gene (i.e. ApcMin/+), and develop
multiple intestinal tumours and sporadic colon tumours in their
intestinal tracts within several weeks of birth. The following
yields of tumours were observed in CHB6 ApcMin/+ mice
exposed to 2-Gy x rays in utero on day 7 (30 tumours per mouse, not
significantly higher than numbers in unirradiated controls) or day
14 postconception (44 tumours per mouse), as 2-day-old neonates (85
tumours per mouse) or 10-day-old neonates (130 tumours per mouse),
and as 35-day-old young adults (70 tumours per mouse) (Ellender et al.,
2006). Hence, neonates were more sensitive to tumour
induction than young adults. The x-ray dose–incidence curve for
adenomas was LQ over the range 0–5 Gy, and strikingly, there were
more tumours in the small intestine than in the caecum and
colorectum (Ellender et al., 2011). Tumour incidence was elevated in
the caecum after ≥2 Gy, and in the colorectum after ≥1 Gy. In
general, adenomas in the small intestine were sessile, while the
smaller numbers of adenomas in the large intestine were
pedunculated. There was also an incidence of microadenomas in the
small intestine, which was greater after the higher doses in the
range used, but none were found in the large intestine. (105) The target cells for colonic tumours are still considered to be
crypt stem cells, and the potential inclusion of progenitor cells as
target cells is not yet resolved. An interesting development is the
finding of rare (one per 150 crypts), slowly cycling, long-lived,
and radioresistant mTert+ stem cells in both small and
large intestine, giving rise to all differentiated intestinal cell
types. These are probably the best candidate target cells in the
colon in terms of the multistage model for carcinogenesis. Germline
mutation of the APC gene predisposes both humans and mice to
intestinal carcinogenesis. In humans, inheritance of mutant APC is
associated with the cancer-predisposing disorder, familial
adenomatous polyposis (FAP), and mutation of APC is an early somatic
event in sporadic colon cancer. Individuals carrying germline
mutations in the APC gene develop hundreds to thousands of
colorectal adenomatous polyps, some of which will progress to
carcinomas if left untreated. (106) From studies of A-bomb survivors and uranium miners,
radiation-induced lung cancers (Annex E) appeared to be more likely
to be small cell lung carcinomas (SCLCs), and less likely to be
adenocarcinomas (ADCs) (Land et al., 1993). In
mice, SCLCs do not occur, and ADCs are the most common type of lung
cancer. The induction of ADCs in female BALB/c mice following acute
irradiation was shown to be consistent with an LQ model in which the
linear term was independent of dose rate by comparing responses
using 0.4 Gy min−1 and 0.06 mGy min−1 (Ullrich,
1983). Also, the DDREF was approximately 4.2 at 3 Gy, and
approximately 3.2 at 2 Gy, and from dose-fractionation studies, the
DDREF was predicted to be approximately 1.1 at 0.1 Gy (Ullrich et al.,
1987). As for mammary tumours induced in the same strain
of mouse, the induced frequencies of lung tumours from acute and
protracted fractionated or low-dose-rate exposures were consistent
with each other in an LQ analysis. (107) In the respiratory tract, the target cells for
radiation-associated carcinogenesis are considered to be in the
basal cells in the trachea and larger bronchi of the central lung,
and in the Clara variant and type II alveolar cells of the
peripheral lung (Annex E). An epithelial stem cell niche has been
identified in the zone where airways terminate and form alveoli. The
putative mouse BASCs in the bronchiolar alveolar junction co-express
secretoglobin a1a (SCGBa1a), the type II cell marker surfactant
protein C, Sca-1, and are negative for CD45 and CD31. Molecular
analysis showed that despite their distinct histopathological
phenotypes in human ADC and SCC, genomic profiles showed a nearly
complete overlap, with only one clear SCC-specific amplicon (Tonon et al.,
2005). Hence, the common or different cellular origin of
lung cancer types may become better understood. In addition, there
may be influences from the irradiated microenvironment. For example,
migration of MSCs into irradiated and stressed regions has been
invoked as a potential alternative or contributory mechanism in
carcinogenesis. (108) Radiation-induced skin cancers (Annex F) in humans are
predominantly BCC. The traditional view was that a threshold dose
exists for radiation-induced skin cancer in the range of 8–10 Gy,
but the A-bomb survivor data indicated that BCC can be induced by
acute exposure at moderate doses even as low as about 1 Gy. In mice,
radiation readily produces SCC but no BCC, whereas in rats,
approximately 20% of induced skin tumours are BCC. The dose–response
curve for total tumours in rats was compatible with an LQ model (and
linear for high-LET radiation), albeit with a tendency for adnexal
(hair follicle and sebaceous) tumours to be more common and later
after low doses compared with epidermoid tumours being more common
and earlier after high doses. With repeated weekly doses of 0.75 or
1.5 Gy over a lifetime, more tumours were produced than expected
from single-exposure dose responses, suggesting either that the
number of events increased per unit dose (i.e. induced
sensitisation) or that clonal growth expanded the number of early
transformed cells (Burns and Albert,
1986a,b). The radiation had to penetrate at least
approximately 180 µm to induce tumours, and 300 µm was approximately
optimum irrespective of follicle growth phase and size, thus
demonstrating that the main target cells were in the stem cell zone
of the hair follicle. Also, there was a marked effect of age on the
incidence of radiation-induced cancer. In the rat, the dose to
induce two tumours per rat (the mid-range induced) by 70 weeks
postirradiation increased from 10 Gy (1 day old) to 15 Gy (28 days),
17.5 Gy (58 days), and 30 Gy (99 days). If this increasing
resistance is expressed as an iso-effect per unit area of skin
irradiated, the factor of three in iso-effective dose per rat
between 1 and 99 days of age would be greater by approximately two
orders of magnitude (i.e. 3 × ∼100) when allowing for the smaller
skin area of the irradiated newborn compared with the much larger
adult rat. (109) A model for human skin cancer proposed that stem cells were
likely target cells for BCC, early progenitor cells for SCC, and
late progenitor cells for papillomas (Sell, 2004). Molecular
characterisation of these different cell populations is continuing,
and the most potent quiescent stem cells appear to be marked as high
levels of the adhesion molecule integrin α6 and low levels of the
transferrin receptor CD71 (α6briCD71dim).
There are quantitative data of mouse follicle-bulge cell divisions
marked with BrdU, which support the long-standing infrequent
SC-division model (Waghmare et al., 2008).
However, it was shown that hair follicle stem cells do not retain
the older DNA strands or sort their chromosomes. To date, there are
no distinct markers for the target cells of the different types of
skin cancer. (110) Radiation-induced bone sarcoma has been associated with high
doses of ionising radiation from therapeutic or occupation-related
exposures (Annex G). However, the development of bone sarcoma
following lower doses remains speculative. Analysis of 80,000
individuals in the LSS cohort to assess the development of bone
sarcoma (most commonly osteosarcoma) showed a preferred fit with a
dose threshold at approximately 0.85 Gy (95% CI 0.12–1.85 Gy) and a
linear dose–response association above this threshold (Samartzis et al.,
2011). Chadwick et al. (1995)
fitted the radium dial painter data using a two-mutation
carcinogenic model with clonal expansion. The analysis showed that
an LQ dose–effect relationship can be applied and, because of the
very low natural (spontaneous) incidence of bone sarcoma, is
consistent with very low AR at low doses and dose rates. Much of the
experimental work on radiation-induced bone cancers has been
performed using dogs. For the low-LET β emitter, 90Sr,
the dose response was non-linear with no tumours occurring at doses
below 18 Gy cumulative average bone dose (Annex G). This much higher
threshold dose may reflect differences in dosimetry, the protraction
of the radionuclide dose, and the much shorter life span of dogs
compared with humans. Osteogenic MSCs for the osteoblast lineage
reside in the bone marrow. CD34− stem cells as well as
mesenchymal precursors are possible target cells for
radiation-induced bone cancer. In mice, these target cells for
bone-associated α emitters lie within a few tens of μm from the
endosteal surface, which is the range of the α particles.
3.5. Location of target cells in a tissue
(111) The location of target cells in different tissues is an
important consideration in the calculation of doses received from
short-range particulate emissions from radionuclides retained in
body tissues, including α particles and low-energy electrons. Thus,
the Commission has made some judgements and assumptions about the
location of target cells in the skin, the respiratory and alimentary
tracts, and the skeleton (ICRP, 1991, 1995,
1996, 2007). Based on the data in the present report and
recent publications, the location and characteristics of target
cells are assembled in Table 3.2. (112) For tissues where the stem cell location is understood and well
defined, it may be reasonable to estimate doses specifically to this
location as stem cells are the primary target for accumulation of
mutations in the initiation and development of carcinogenesis.
However, the extent to which the immediate progeny of stem cells may
also be targets for the development of particular cancer types
varies between tissues, and is not well established in many cases.
For some tissues, this possibility may not have implications for the
definition of targets for dosimetric purposes because stem cells and
their immediate progeny occupy the same microenvironment. For
example, in the epidermis, the stem cells and immediate daughter
cells are found in the basal layer. The nominal depth of target
cells in human epidermis is taken to be 70 µm for dosimetric
protection calculations, although there is significant variation in
depth of the skin undulations. For the example of the colon, the
distance from the mucosal surface is taken to be the crypt base at
280–300-µm depth, and widening this band to include progeny cells
makes little difference to dosimetric calculations when considering
irradiation predominantly from radionuclides in the intestinal
lumen. For the respiratory tract, however, assumptions regarding the
type and location of target cells within the airways can be a major
determinant of estimated doses for some radionuclides, including the
α-particle-emitting progeny of 222Rn. For bone cancer,
until recently, the target has been taken to be a 10-µm layer
adjacent to bone surfaces, but now it is recognised that a 40–50-µm
layer would be more appropriate. For leukaemia, while it is known
that stem cells are located in endosteal and vascular niches,
refinement from the calculation of average red bone marrow dose has
not proven feasible, although it is recognised that risks of
leukaemia from bone-seeking α-particle emitters (e.g.
239Pu and 226Ra) are substantially
overestimated by such calculations. For other tissues such as
mammary gland and thyroid, regional distributions of potential stem
cells are not considered, and radiation doses are calculated as
tissue averages. Locations and characteristics of target cells for radiation-induced
cancers in different tissues. HSCs, haematopoietic stem cells.
3.6. Cell-based and tissue-based considerations
(113) Regarding the relationship between the cancer incidence
parameters (Table 3.1) and the three considered mechanistic factors
of: (a) the number and sensitivity of stem cells to
radiation-induced mutation; (b) the retention of mutated stem cells
in a tissue; and (c) the population size of stem cells with a
sufficient number of predisposing mutations, there is a lack of
definitive evidence for the contribution of one or more factors to
the values of radiation risk for the different tissues. One example
is that on current estimates, differences in the stem cell number
among tissues are unlikely to be an important factor, albeit with
the caveat of large uncertainties in the numbers. Some estimates of
total stem cells per human in bone marrow are approximately
108 (Annex A), approximately 4 × 107 in
colon (Annex D), and approximately 2 × 107 in skin (Annex
F; one functional stem cell per 35,000 epidermal cells, and
8 × 1010 total epidermal cells per human). These
numbers span more than three orders of magnitude, and their ranking
clearly does not show a similar ranking of sensitivities regarding
cancer incidence expressed by either EAR or ERR (Table
3.1). Also, the various total stem cell numbers in the 31
different tissues/locations considered by Tomasetti and Vogelstein
(2015) do show a trend towards an association between
stem cell number and the risk of natural (spontaneous) cancer, but
the trend is not as strong as when the number of stem cells was
multiplied by the number of lifetime divisions per stem cell. In any
case, a direct correlation between radiation risk and the total
number of stem cell divisions alone is unlikely to explain the
relative lack of radiation-induced cancer in small intestine.
Biological reasons for this include the p53-mediated apoptosis in
the stem cell zone in the small intestine, its prevention by Bcl-2
expression in the colon, and evidence for the retention of the
parental DNA template strand (Potten, 2004). Also, there
may be differences in the true target cell population in the two
sites, depending on whether it includes extra ‘potential’ stem cells
and hence is larger, or whether it consists of a minority of
radioresistant mTert+ stem cells that differs between
sites and hence forms a smaller population. (114) Mechanisms behind any tissue differences in the sensitivity to
radiation carcinogenesis are likely to be multiple, including
tissue-specific mechanisms of stem cell number and turnover, and
cell-type-specific mechanisms of DNA replication, DNA repair,
cell-cycle control, and apoptosis. Knowledge in these areas is still
lacking, especially for tissue stem cells. One general trend is that
the risk of radiation carcinogenesis is higher for some tissues with
higher rates of renewal such as skin, colon, and stomach, and lower
for those with lower rates of renewal such as oesophagus, liver,
thyroid, and bone surfaces, although there are exceptions. It has
been considered that a higher rate of stem cell proliferation could
contribute to a more rapid rate of mutation accumulation, but a
higher turnover rate of progenitor cells may not allow time for the
cells to accumulate enough mutations to acquire full malignancy.
Adult tissues without proliferation are almost refractory to
radiation carcinogenesis; brain can be such an example, although a
small number of NSCs are slowly proliferating even in adult primate
brain (Gould
et al., 1999). In contrast, some highly sensitive tissues
such as breast are not particularly active in proliferation (Annex
B). Similarly, bladder is known to be sensitive to radiation
carcinogenesis, yet this tissue is rather quiescent. Thus, simple
proliferative activity of tissues does not appear to predict tissue
sensitivity to radiation carcinogenesis. (115) Some tissue stem cells may be characterised by specific
mechanisms such as asymmetric segregation of template strands when
replicating DNA, use of a specific DNA repair system such as NHEJ,
altruistic apoptotic cell death, and cellular differentiation. These
cellular features are tightly regulated by the tissue
micro-environment of the stem cell niche, which is essential in
maintaining the ‘stemness’ of the stem cells. Thus, any perturbation
in these features is likely to lead to accumulation of mutations. In
the case of mammary gland, the niche was shown to be the target of
radiation carcinogenesis (Section B.4.2). For some tissues, stem
cells are under constant competition for the occupancy of the niche,
which may lead to elimination of some of its stem cells. This
elimination could be a contributing factor for the tissue-specific
sensitivity to radiation. It is concluded that a number of
mechanisms are known or speculated upon in this report that may
contribute in some way to cancer susceptibility among tissues, but
their applicability and importance are unknown to date.
3.7. Age dependence of radiation carcinogenesis
3.7.1. Age-dependent occurrence of spontaneous cancers
(116) Childhood cancer is defined as cancer occurring after birth
until puberty, and thought to arise in children carrying
predisposing mutations inherited from their parents. The
mutation may also be acquired during fetal development or
childhood growth. Epidemiological studies of childhood cancer
and some aetiological factors including ionising radiation were
discussed in detail in Publication 90 (ICRP,
2003). The incidence of childhood cancer is
approximately 1 × 10−4 per live birth. The type of
childhood cancer is restricted, and each cancer type has a
specific age window of occurrence (Ries et al., 1999).
For example, retinoblastoma of the hereditary form develops
within 1 year of birth. In addition, these carriers develop bone
sarcoma around the time of adolescence (Knudson, 1971; Abramson
et al., 1984; Friend et al., 1986).
This bimodal pattern of occurrence is shared by the
non-hereditary form of retinoblastoma and bone sarcoma (Ries et al.,
1999). (117) Adult cancers are thought to occur as the result of
mutations acquired somatically, and exhibit a steady increase in
incidence with age (Armitage and Doll,
1954). This age-dependent increase makes cancer the
leading cause of death in developed countries with a long life
expectancy. The incidence of cancer differs by sex. It is
slightly higher for males up to adolescence, and almost twice as
high for males than females after 70 years of age (Bleyer et al.,
2006). The incidence of adulthood cancer is twice as
high in females as in males in their 40s due to the
female-specific cancers in this age group.
3.7.2. Risks from prenatal radiation exposures
(118) Russell and Russell (1954) were the first to
thoroughly investigate the developmental-stage dependence of
radiation effects in mice. They exposed pregnant female mice to
radiation, and found that the pre-implantation stage was highly
sensitive to embryonic death. However, no malformation was
identified, and this is still the general rule today. However,
Pampfer
and Streffer (1988) observed the induction of
gastroschisis (a birth defect with infant's intestines outside
the body) after exposure to x rays and neutrons during the
pre-implantation period in a mouse strain with a specific
genetic predisposition for such a malformation. Similar effects
have been reported in other mouse strains after exposure to
ionising radiation and genotoxic agents. However, such cases
were only observed with regard to specific genetic
predispositions (ICRP, 2003; Jacquet,
2012). Malformations were induced mainly by exposures
at the later stage of embryonic development when organogenesis
takes place. It is interesting to note the classic work by Howlett
et al. (1988), who demonstrated that chick embryos
lacked tumour formation even when inoculated with highly
oncogenic Rous sarcoma virus, suggesting strong suppression of
transformed phenotype in embryos. Interestingly, chicks
inoculated with the virus after hatching became viraemic, and
developed tumours at any site of injury. Based on these
observations, it is tempting to speculate that cancer is a
disease taking place in the context of adult-type tissue
architecture with the tissue stem cells and their niche, as
shown by radiation-induced preneoplastic tissue microdysplasias
in mammary gland (Annex B), thyroid (Annex C), and colon (Annex
D). (119) One of the largest human studies on the effect of
fetal-stage exposures was the Oxford Survey of Childhood Cancers
(OSCC); a case–control study of mortality from childhood cancer
in Great Britain. OSCC found an association with intra-uterine
x-ray examination, and indicated that fetal stages were highly
sensitive to radiation with ERR per Gy of 50 for childhood
leukaemia and other childhood cancers alike (Wakeford and
Little, 2003). The epidemiological study of the
incidence of cancer among the in-utero-exposed A-bomb survivors
suggested a high ERR for childhood cancers other than leukaemia
of 22 per Gy (although based upon just two incident cases and
thus not statistically meaningful), but with no increased risk
of childhood leukaemia (Wakeford and Little,
2003). The study of Ohtaki et al. (2004)
of chromosome translocations in peripheral blood lymphocytes
sampled from A-bomb survivors exposed in utero found no dose
response above a dose of approximately 100 mGy, in contrast to
the dose response found in some of the mothers and other adults.
The authors suggested that the lack of chromosome aberrations is
due to high sensitivity of the fetal-stage haematopoietic cells
to killing by radiation at moderate doses (i.e. doses much lower
than those normally associated with cell killing affecting
cancer risks). Competition-mediated elimination of stem cells
from newly established bone marrow niches may also explain the
lack of chromosome aberrations in lymphocytes of
in-utero-exposed A-bomb survivors. A moderate ERR of 1.0 per Gy
was found for mainly adult-onset solid tumours, with an overall
risk lower than that for childhood exposures with ERR of 1.7
(Preston et al., 2008). It is interesting to note
that a combined analysis of in-utero-exposed and
childhood-exposed individuals indicated a dose response of
upwards curvature, suggesting a quadratic component in the
induction of cancer for these cohorts. The Commission made an
extensive review, but at that time was unable to reach a
clear-cut conclusion on the risks of fetal-stage exposures
(ICRP,
2003). Based on the data available and uncertainty on
the development of solid cancer for in-utero-exposed
individuals, ICRP (2007) made a
judgement that the lifetime cancer risk following in-utero
exposure is similar to that following irradiation in early
childhood. Based on more recent follow-up, this assumption
appears to overestimate the lifetime risk of in-utero exposure
(Preston et al., 2008). (120) Ideally, the unresolved issue of the epidemiological
studies has to be augmented with the help of experimental
studies. However, one problem of such an experimental approach
is the lack of an appropriate animal model of human childhood
cancer. For example, human childhood cancer is relatively rare
with a cumulative incidence of approximately 10−4
from birth to 15 years of age. Experimental studies usually use
a group of less than 100 animals, which is too small to detect
cancers arising at a frequency of 10−4. Therefore,
experimental studies are conducted to analyse the effects of
in-utero exposures on the lifetime occurrence of cancer, which
mimics adult-onset cancer in humans. Such studies using
laboratory mice and rats demonstrated that in-utero exposures
are less effective than neonatal exposures in inducing leukaemia
and various solid tumours (Upton et al., 1960;
Sasaki,
1991; Inano et al., 1996;
Di Majo
et al., 2003). Detailed studies on the age-dependent
sensitivity of the ApcMin/+ mice to radiation
indicated that the sensitivity was highest for 10-day-old
neonates and decreased in the order of 2-day-old neonates,
35-day-old young adults, 14-day-old fetuses, and 7-day-old
embryos (Ellender et al., 2006). Mouse studies thus indicate
that the fetal stage in general is less sensitive than the
neonatal stage, and that the earlier embryonic stage is much
less sensitive to radiation induction of leukaemia and solid
tumours. However, the fetal stage is shorter in mouse than in
man, and this could be one of the reasons for the different
results between these two species. A range of parameters
affecting the sensitivity to radiation carcinogenesis during the
prenatal period of development was discussed in detail in
Publication 90 (ICRP, 2003). OSCC
reported that the first trimester was most sensitive to
induction of childhood cancer (Bithell and Stiller,
1988), and this contradicts the mouse data where the
early embryogenesis stages are generally insensitive to
radiation carcinogenesis. One possible reason is that the doses
in the first trimester could have been greater than those
received during the second and third trimesters (Mole,
1990), and the apparent greater sensitivity could
just be related to the dose received during an examination
(Doll
and Wakeford, 1997). It is important to note that
none of the case–control studies of prenatal exposure and
childhood cancer risk performed individual dose reconstructions
on the cases and the controls involved. Dose estimates were
based on national surveys, and there are uncertainties with
regard to machine parameters, repeat examinations, and
undocumented procedures. (121) When considering cellular characteristics, it is difficult
to discern the difference between stem cells of fetal stages and
after birth. Therefore, the difference in the sensitivity to
radiation carcinogenesis has to be sought at the tissue level.
One distinction of fetal-stage tissues is the lack of a clear
niche-like micro-architecture, while in the adult, tissue stem
cells are thought to reside in a distinct microenvironment of a
stem cell niche. The adult-type niche is established after birth
for many tissues. As discussed in Section 2.5.4, for example, a
major site of HSC proliferation is the liver during fetal
development. HSCs then migrate and colonise the bone marrow
niche. Numerous HSCs migrate to the newly established bone
marrow niche housing a limited number, and it has been shown
that the only HSCs that can settle in the niche are those in the
G0 phase (Bowie et al., 2006).
This selective settlement functions as an efficient mechanism to
remove aberrant HSCs. (122) Ohtaki
et al. (2004) found that in-utero-exposed A-bomb
survivors did not generally express chromosome aberrations in
their lymphocytes (genomic instability), although they observed
a small increase in the aberration yield at doses <50 mGy.
This could be the underlying mechanism for the lack of leukaemia
among in-utero-exposed A-bomb survivors. A similar lack of
chromosome aberrations was observed for in-utero-exposed mice
(Abramsson-Zetterberg et al., 2000; Nakano
et al., 2007). Another study showed that although
unstable chromosome aberrations induced (0.5–1.5 Gy) in the
fetal haematopoietic short-term repopulating stem cells were
eliminated by subsequent cell division, they persisted in the
long-term repopulating stem cells (Devi and Satyamitra,
2005). A more recent study has indicated that the
removal of in-utero-exposed HSCs is somewhat ‘leaky’, and clonal
expansion of surviving HSCs can be detected (Nakano
et al., 2012). These studies can be explained by two
assumptions. Firstly, fetal HSCs do not have to be more
sensitive to cell killing by radiation. Secondly, fetal HSCs
with chromosome mutations are preferentially removed during
fetal to neonatal stages, possibly by competition for residence
in the bone marrow niche. Although further studies are
definitely needed, the tissue-level competition is likely to
serve as an effective filter to remove aberrant stem cells in
haematopoietic tissue. However, in contrast to the findings in
haematopoietic tissue, in rats exposed in utero, the aberrant
cells with chromosome translocations in mammary epithelial cells
were found to persist (Nakano et al., 2014).
Also, the risk of mammary cancer was not elevated following
radiation exposure of fetal rats (Imaoka et al., 2013).
Thus, the fate of aberrant stem cells in the fetally exposed
animal may well differ among tissues. A somewhat similar
explanation can be made for the low sensitivity of fetal
exposures to radiation induction of intestinal tumours in
ApcMin/+ model mice (Ellender et al.,
2006), as discussed in the previous section. During fetal
development, the intestine is formed as a simple tube with a
layer of stem cells (Crosnier et al.,
2006). In the case of mice, the first differentiation of
the villus formation takes place in 15-day-old fetuses. However,
crypt formation only begins on day 7 after birth. This suggests
that the stem cells settling into the crypt stem cell niche are
few compared with the large number of fetal-stage ISCs. This
leads to strong competition among fetal stem cells that is
likely to remove such aberrant cells for contributing to the
maintenance of the adult intestinal tissue. (123) The competition-mediated elimination of aberrant cells
during neonatal stages is likely to operate not only for
radiation-damaged stem cells, but also for spontaneously
aberrant cells (Nakamura, 2005). ALL
is a major childhood cancer with characteristic translocations
specific to certain types of leukaemia. With the examination of
cord blood samples by polymerase chain reaction (PCR),
approximately 1% of newborns were found to carry the
translocation ETS-like leukaemia/acute myeloid leukaemia 1
(TEL/AML1) fusion gene; one of the major translocations specific
to childhood ALL. This translocation was shown to be generated
during normal fetal development (Mori et al., 2002). In
fact, the number of translocation carriers is higher by two
orders of magnitude than the incidence of childhood leukaemia of
this type, suggesting a possible elimination of the cells with
predisposing mutations during childhood growth. The elimination
of predisposing cells after birth was suggested by the fact that
the background incidence of ALL is highest at approximately 3
years of age, then declines rapidly with age up to 20 years, and
increases again in the elderly (Smith, 2005). It is
tempting to speculate that the stable lodging of HSCs into the
bone marrow niche during postnatal stages of development may
serve as a barrier to select out unfavourable cells with any
predisposing mutations. (124) Such elimination of cells with cancer-predisposing
mutations may operate for many other childhood cancers of
sporadic types (Ries et al., 1999).
Neuroblastoma, the tumour of neural crest origin, is one of the
most common malignancies among children, and its incidence
before 15 years of age is approximately eight in 10,000.
However, necropsy samples of babies dying within 3 months of
birth demonstrated that precancerous lesions are rather common,
and found at a rate of approximately one in 200 (Beckwith and
Perrin, 1963). Thus, the occurrence of the lesion in
newborn babies and the incidence of neuroblastoma differ
considerably, suggesting that most precancerous lesions are
eliminated by some mechanisms. In accordance with this, the
incidence of tumours is highest during the first few months
after birth, and declines rapidly to almost zero at 15 years of
age (Goodman et al., 1999). Similar patterns of rapid
decline in incidence after birth are noted for other childhood
cancers such as retinoblastoma, Wilms’ tumour, and
hepatoblastoma. For those tumours, terminal differentiation of
precancerous cells may be involved as a mechanism of the
decline, in addition to real elimination of
aberrant/premalignant stem cells as in the case of ALL. (125) Mouse data on fetal exposures tend to demonstrate that the
risk is minimal at early stages of gestation. However, exposure
at 17 days of gestation was shown to induce cancers of lung and
pituitary gland (Sasaki, 1991). It is
interesting to note that these two organs are well developed by
day 17 of mouse gestation (Sheng and Westphal,
1999; Yu et al., 2004).
Thus, even fetal exposure is capable of inducing cancer during
the perinatal stage, where certain tissues are well developed.
In contrast to mice, humans have a much longer period of fetal
development; therefore, it is likely that certain tissues at
perinatal stages are as sensitive as after birth. Wakeford
(2008) noted that 90% of the pelvimetric diagnoses in
the OSCC were performed during the last month of gestation.
Therefore, such late stages of fetal development may exhibit
similar sensitivity to radiation carcinogenesis as that of the
neonatal stage after birth. On the other hand, the remarkable
similarity in the RRs of all childhood cancer types in the study
raises concern about the causal relationship with prenatal x-ray
examinations (Boice and Miller, 1999;
ICRP,
2003). It is noteworthy that the lifetime RRs of
childhood exposures are fairly variable between cancer types
(UNSCEAR, 2013).
3.7.3. High sensitivity of children to cancer induction from radiation
(126) It is well recognised that children are highly sensitive to
radiation induction of leukaemia and some solid tumours (ICRP,
2003). The sensitivity of children to
radiation-induced cancer has been reviewed extensively in a
recent publication by UNSCEAR, and the high sensitivity is
strongly dependent on cancer type (UNSCEAR, 2013).
Children are more sensitive for approximately 25% of 23 cancer
types analysed compared with adults. These types include
leukaemia, thyroid, skin, and brain cancer. Children have the
same sensitivity as adults for approximately 15% of cancer types
including bladder, and lower sensitivity for approximately 10%
of cancer types including lung. For 20% of cancer types, the
data are insufficient for concluding a difference in
sensitivity, and for 30% of cancer types, no increase in the
risk is observed after radiation exposure. ERR of radiogenic
cancer, especially for early onset, is inversely related to the
age at exposure for many cancer types (i.e. high for the young
with an attained-age-dependent decline in risk) (Preston
et al., 2007). ERR per Gy for ALL is more than 15 for
children aged <10 years, and the risk declines rapidly with
increasing age (Hsu et al., 2013). It
is noteworthy that even with the high ERR for early-onset ALL,
EAR for such cancer is rather low because the background
incidence of cancer is considerably lower in children than in
adults. ERR per Gy for adult solid cancer following exposure in
childhood is as high as 3–5, which declines sharply as the age
at exposure increases. In addition, the latency of cancer
development is relatively short for some cancers after childhood
exposures; for example, the minimum latency interval for thyroid
cancer occurred within 4 years in children after the Chernobyl
accident (Section C.1). As a consequence, the lifetime ERR per
unit dose for solid cancer after childhood exposure is
approximately 1.0 (Pierce et al.,
1996). (127) The higher sensitivities of children for radiation
induction of leukaemia and some adult solid cancers are
considered to be due to the high proliferation rate of stem
cells and progenitor cells in children. However, these cells
also proliferate rapidly in embryos and fetuses, and the
sensitivity to radiation carcinogenesis for these stages does
not appear to be as sensitive as in childhood, as discussed at
length in the previous sections. Therefore, a cellular feature
such as proliferation alone is unlikely to explain the high
sensitivity to radiation-induced cancer after exposure in
childhood. However, there is one feature of a tissue that
differs considerably between children and adults, which could
contribute to the high sensitivity of children to radiation
carcinogenesis. As discussed, the adult stem cell niche is
established around perinatal stages, although the time of
establishment is likely to vary between tissues and between
species. During childhood growth, the stem cell niche together
with its stem cells increases in number as a unit to match the
demand of body growth (Fig. 3.2). (128) In the case of the intestine, this process of stem
cell/niche expansion is accomplished by fission of crypts (Fujimitsu
et al., 1996). The fission starts with an increase in
crypt size, which then splits from the bottom to form two
crypts. The increased size of a crypt means increased
availability of crypt niches that eases the competition of stem
cells within a niche. In addition, competition between niches is
also eased. The end result of these is a better chance for an
aberrant stem cell to remain to accumulate more mutations in the
intestine. Thus, the process of expansion of the number of
stem/niche units could potentially contribute to a higher
sensitivity of children to the development of cancer after
radiation exposure. Expansion of tissue stem cell niche multiplicity during childhood
growth.
3.7.4. Risk of carcinogenesis from exposures in adulthood
(129) The risk of radiation carcinogenesis in general decreases
inversely with age at exposure. The risk also decreases by
attained age after radiation exposure. Adulthood exposures give
moderate risk compared with childhood exposures, and this can
partly be explained by a change in the cellular features of stem
cells. For example, childhood stem cells divide frequently in a
symmetrical fashion to cope with demands for bodily growth,
while adult stem cells do so less frequently. This makes the
former more prone to mutate than the latter. Also, the stem
cells during childhood growth are expected to experience less
competition when the stem cell niche increases in number in
response to the demand of bodily growth as discussed above.
Indeed, adult stem cells are under stronger competition than
stem cells of childhood growth, which is likely to keep the risk
low, as discussed in Section 3.3.2. (130) Elimination of aberrant stem cells has been demonstrated in
animal models. In the case of mouse mammary carcinogenesis,
radiation exposures induce many more initiated cells than those
progressing to full malignancy (Adams et al., 1987;
Kamiya
et al., 1995). Indeed, it was shown in a rat mammary
carcinogenesis model that the frequency was as high as one in 13
irradiated mammary clonogenic cells, which could not be
accounted for by a specific mutation induced by irradiation
(Para. B55). Even with highly frequent initiated cells,
neoplasias arise much less frequently, indicating either
efficient elimination of such cells or suppression of their
aberrant phenotypes. (131) As for radiation exposure in adulthood, ERR at some
specified time after exposure is generally smaller than that for
childhood exposures. The risk rises within a few years for
leukaemia and after 10 years or more for solid cancer, and the
elevated ERR eventually starts to decline with increasing
attained age of those exposed. This pattern of elevation and
decline of cancer risk has been observed repeatedly, but the
most reliable data come from epidemiological studies (Boice et al.,
1985), and notably from studies of A-bomb survivors
(Preston et al., 2007; Richardson et al.,
2009; Hsu et al., 2013).
Based on the multistage carcinogenesis model by Armitage and
Doll (1954), the RR of a population given one hit of
acute irradiation was predicted to decline over the attained age
by a rate of 1/age (Pierce and Mendelsohn, 1999). The rates of
decline of RR were estimated in the recent compilation of A-bomb
survivor data. Although the estimates varied among cancer types,
the rates were, in general, in the range of approximately
1/age2 (Preston et al., 2007).
In addition, a study of radon-exposed uranium miners indicated
that the RR of lung cancer declined approximately 50% for each
10 years after they stopped working in mines, suggesting that
the decline is proportional to approximately 1/age3
(Tomasek et al., 2008). These rates of the decline in
RR are higher than 1/age and may suggest the loss of
initiated/precancerous cells in a tissue over time, as discussed
earlier. (132) Overall, the above-described age-dependent sensitivities to
radiation carcinogenesis can be summarised as follows: embryo
and fetal stages, low to moderate; children, high; and adults,
low. However, a mechanistic insight for this pattern of
radiosensitivity is still lacking. In the past, it was naively
assumed that the high sensitivity to radiation carcinogenesis
for children could be attributed to high rates of proliferation
of somatic cells at this stage of human life. However, this
simple assumption makes it difficult to explain why the fetal
stages with even higher rates of proliferation are not
associated, in general, with extremely high sensitivity to
radiation carcinogenesis, although the latter remains somewhat
controversial.
3.8. Recommendations for future research
Knowledge of stem cell/niche systems and their sensitivity to
radiation injury resulting in carcinogenesis in bone marrow, large
intestine, and skin is more developed than for other tissues at
risk, such as mammary gland, thyroid, lung, and bone. Hence, there
is a need to further investigate the stem cell systems in these
latter (more slowly renewing) tissues, their controlling factors,
and their mutational/mechanistic responses to acute and protracted
irradiation. Research programmes on stem cells and their
low-dose-radiation responses are helping to support research in this
highly important field; for example, the Euratom multicentre RISK-IR
(Risk, Stem Cells and Tissue Kinetics - Ionising Radiation) project
(http://www.risk-ir.eu). Genomic instability, non-targeted effects, and adaptation have been
researched extensively and discussed at length in the literature.
However, there remains uncertainty in their contribution to
carcinogenic effects after low chronic doses, and how to manage any
consensus within the framework of radiological protection
principles. Hence, this is a challenge for the future. The immortal DNA-strand hypothesis, whereby the parental DNA template
is retained during asymmetric stem cell divisions so that the
mutational burden in the stem cell population is kept low, may act
to protect against radiation carcinogenesis. There is evidence to
support this mechanism in some tissues considered in the present
report (e.g. small intestinal crypts, mammary epithelium, and
epidermis), but the mechanism was not found to apply in HSCs. Hence,
it would be informative to study further tissue types, and attempt
to measure purportedly relevant carcinogenic mutations in stem cells
following acute or protracted irradiations in tissues that show this
phenomenon. The concept of competition between normal and radiation-injured stem
cells for residence in the stem cell niche, which may act to give
less carcinogenic events than expected after irradiation, is
supported by studies in haematopoietic tissue after acute doses of
≥1 Gy. It would be useful to study this effect after lower doses
and, in particular, after chronic low radiation exposures, as well
as in other tissues where possible. The numerical value of the DDREF is controversial, and appears to be
lower in some human populations than in others and in experimental
animal systems. The reasons for this are currently unclear. Also,
regarding the effect of age at exposure, there are inconsistencies
between the evidence in human populations and in experimental animal
systems regarding carcinogenesis at the fetal and neonatal stages of
development. Reconciling these differences at the biological level
should help underpin the consistency and robustness of the
radiological protection system.