Carcinogenesis, 2015, Vol. 36, No.

3, 307–317

doi:10.1093/carcin/bgv007
Advance Access publication January 20, 2015
Review

review
Radiogenomics helps to achieve personalized therapy
by evaluating patient responses to radiation treatment

Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019

Zhen Guo1, Yan Shu2, Honghao Zhou1, Wei Zhang1,* and Hui Wang3,*
1
Department of Clinical Pharmacology, Xiangya Hospital, Central South University and Institute of Clinical Pharmacology,
Central South University; Hunan Key Laboratory of Pharmacogenetics, Changsha 410008, P.R. China; 2Department of
Pharmaceutical Sciences, School of Pharmacy, University of Maryland, Baltimore, MD 21201, USA and, 3Department of
Radiation Oncology, Hunan Provincial Tumor Hospital & Affiliated Tumor Hospital of Xiangya Medical School, Central South
University, Changsha 410013, P.R. China
*To whom correspondence should be addressed. Tel: +86 13973168774; Fax: +86 731 8235 4476; Email: yjsd2003@163.com. Correspondence may also be
addressed to Hui Wang. Tel: +86 13973135460; Fax: +86 731 8865 1999; Email: wanghui710327@163.com

Abstract
Radiogenomics is the whole genome application of radiogenetics, which focuses on uncovering the underlying genetic causes
of individual variation in sensitivity to radiation. There is a growing consensus that radiosensitivity is a complex, inherited
polygenic trait, dependent on the interaction of many genes involved in multiple cell processes. An understanding of the
genes involved in processes such as DNA damage response and oxidative stress response, has evolved toward examination of
how genetic variants, most often, single nucleotide polymorphisms (SNPs), may influence interindividual radioresponse. Many
experimental approaches, such as candidate SNP association studies, genome-wide association studies and massively parallel
sequencing are being proposed to address these questions. We present a review focusing on recent advances in association
studies of SNPs to radiotherapy response and discuss challenges and opportunities for further studies. We also highlight the
clinical perspective of radiogenomics in the future of personalized treatment in radiation oncology.

Introduction
Radiation therapy is a key modality in the treatment of cancer.
Moreover, when radiotherapy is combined with chemotherapy,
surgery or other targeted therapies, treatment efficiency is
improved and recurrence and cancer death rates are reduced
(1). The main aim of radiotherapy is to maximize local control
of the tumor while minimizing the damage to patient’s normal
tissues. There is a spectrum of side effects (toxicities) in the
surrounding normal tissues associated with radiotherapy (2).
Acute/early toxicities, usually occur within 90 days after treat-
ment, are transient and healing. However, late toxicities are
more persistent and troublesome, which may include chronic
inflammatory processes, vascular changes, fibrosis and atrophy
(3). Tumor radiation response is the determining factor of the
radiotherapeutic effect. How to elevate radiation response in
tumor cells while decrease radiosensitivity in adjacent normal
tissues has become a core issue in the tumor radiotherapeutic
field (4). Previous interest has primarily focused on tumor radio-
curability, but a shift toward cancer survivorship in recent years
has seen growing interest in understanding radiation-induced
complications in normal tissues (5).
A large patient-to-patient variability exists in radiotherapy
response despite uniform treatment protocols (6). Although
some of this may be ascribed to comorbidities, body habitus and
stochastic factors (7–9), it has become increasingly believed that
genetic background takes a leading role in the interindividual
variation in radiosensitivity (10–13). Heritability analysis sug-
gested that 58–78% of the variance can be attributed to genetic
factors (14,15). The initial evidence for the genetic basis of radio-
sensitivity came from the observation that certain individuals
with specific genetic disorders, such as ataxia telangiectasia,
Nijmegen breakage syndrome, Fanconi anemia and DNA ligase
IV deficiency, were hypersensitive to radiation (16). Around the

Received: September 15, 2014; Revised: December 21, 2014; Accepted: January 15, 2015

© The Author 2015. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com.

307
 308 | Carcinogenesis, 2015, Vol. 36, No. 3
Abbreviations thus contributing to radioresistance (29,30). Multiple DNA repair
BC breast cancer
GWAS genome-wide association study
HNC head and neck cancer
mRNA messenger RNA
NPC nasopharyngeal carcinoma
OS overall survival
RP radiation pneumonitis
SNP single nucleotide polymorphism
turn of the century, there was a growing consensus that the
genetics behind individual radiosensitivity is more compli-
cated and must be considered as a so-called polygenic or com-
plex trait dependent on the combined influence of risk alleles
with different frequencies and different relative risks (10). It is
of importance to increase our understanding of the molecular
pathogenesis of radiotherapy toxicity, find ways of predicting
those patients likely to suffer with long­term side effects and
develop new approaches for their amelioration.
With the evolution of DNA sequencing and bioinformatics,
radiogenomics has emerged as a new research field which stud-
ies the influence of genetic variations on radiation response
(17–20). One aim of radiogenomics studies is to develop a risk
model including genetic factors and clinical covariants capable
of predicting individual radiosensitivity and likelihood of devel-
oping side effects after radiotherapy. Another aim is to improve
the radiotherapeutic effect with individualized radiotherapy,
through prevention or intervention of radiation-induced mor-
bidities, by increasing our understanding of the biological mech-
anisms behind them (4,21–24). Many cancer patients achieve
survivorship at the cost of treatment complications occurring
in normal tissues. The ability to predict a predisposition for
severe radiotherapy-induced adverse effects in normal tissues
could potentially aid treatment decision making and improve
therapy effects. Avoiding or reducing radiation in these patients
could lessen the likelihood of radiotoxicity-related morbidities
and may also potentially reduce the burden of healthcare costs
incurred for the supportive care required for these conditions
(25). There has been intense interest in the phenomenon of inter-
individual variation in normal tissues and tumor responses to
radiotherapy in the postgenomics era. Experimental approaches,
such as candidate single nucleotide polymorphism (SNP) asso-
ciation studies, genome-wide association studies (GWASs) and
massively parallel sequencing are being proposed to address
these questions. Evidence of genetic polymorphisms underlying
interindividual differences in radiotherapy responses is rapidly
increasing. In this review, we will present a systematic summary
of the progress in radiogenomics studies.
Biological pathways associated with
radiosensitivity
Radiosensitivity can be defined as the relative susceptibility of
cells, tissues or organs to the effects of radiation (4). To preserve
genomic stability, cells have developed elaborate response path-
ways to handle DNA lesions caused by ionizing radiation. The
DNA damage response is a core determining factor of tumor
radioresistance or radiosensitivity which decides the cell’s fate
either to repair DNA damage or to undergo apoptosis if there is
too much damage by manipulating cell cycle checkpoints, DNA
damage repair, apoptosis and other responses via multiple sig-
nal transduction pathways (26–28). DNA damage­induced activa-
tion of cell cycle checkpoints at either G1/S or G2/M give cells
time to employ DNA repair machineries to resolve DNA lesions,
pathways, including single-strand break repair, nonhomologous
end joining repair, homologous recombination repair, base exci-
sion repair, nucleotide excision repair, have evolved to deal with
various DNA lesions (31,32). Blocking these repair pathways will
likely result in increased radiosensitivity. DNA damage levels
that exceed the repair capacity of cells lead to the activation
of apoptosis, which is controlled by a number of proapoptotic
factors and antiapoptotic factors. Details of the DNA dam-
age response network are illustrated in Figure  1. As radiation
results in free radical formation and inflammatory responses
in exposed tissues, genes involved in oxidative stress reactions
and inflammation processes also exert a considerable influence
on the response to ionizing radiation (33,34).
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
Candidate gene approach
Genetic association studies have been employed to identify
causal functional SNPs in normal tissue radiotoxicities. Most
association studies have based on a candidate gene approach
and investigated SNPs in genes involved in the complex radi-
oresponse pathways as mentioned above (Figure  2). Through
a number of case–control studies, in which SNPs at candidate
loci were genotyped across patients with or without radiother-
apy side effects, many genetic markers have been investigated.
Studies into the most frequently investigated candidate genes
within each pathway are highlighted below (Table 1).
DNA damage repair genes
XRCC1, XRCC2, XRCC3, XRCC4, XRCC5
Among the DNA damage repair genes, XRCC1 is the most fre-
quently researched candidate. Contrasting results have been
reported on the impact of XRCC1 on radiation-induced normal
tissue toxicities. Evidence from lung cancer patients showed
that XRCC1 rs25487 (G > A) may influence radiation sensitiv-
ity (35) and contribute to the development of grade ≥2 radia-
tion pneumonitis (RP) (36). As for nasopharyngeal carcinoma
(NPC) patients, rs25487 A  allele was conferred a reduced risk
of late normal tissue complications after radiotherapy (37–39).
Nevertheless, another independent study demonstrated that
head and neck cancer (HNC) patients with the rs25487 AA/
AG genotypes have a higher probability of radiation-induced
mucositis (40). When it comes to breast cancer (BC) patients, no
positive association has been found between rs25487 and acute
toxicity (41).
Previous findings demonstrated that XRCC2 rs3218536 (G > A)
was correlated with overall survival (OS) in non-small-cell lung
cancer (NSCLC) patients receiving radiotherapy (42). In addition,
results from classification and regression tree analysis displayed
male patients with TT genotype of XRCC4 rs6869366 (G > T) and
female patients with AG/AA genotypes of XRCC5 rs3835 (G > A)
were also at increased risk of severe RP (43). Alsbeih et al. (37,38)
have initially indicated the role of XRCC5 rs1051677 (T > C) C allele
in radiation toxicity among NPC patients. More recently, another
study in Chinese NPC patients demonstrated a trend of possible
association between rs861539 (C > T) of XRCC3 and radiation-
induced fibrosis, but not rs25487 of XRCC1 (44). Additionally, BC
patients with the variant allele of XRCC3 rs861539 were more
likely to experience skin toxicity (45).
RAD51
The RAD51 gene family consists of several proteins that show
DNA-stimulated ATPase activity and play a central role in the
homologous recombination repair activation. A functional SNP
 Guo et al. | 309

Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019

Figure 1.  Highly simplified schematic of the radiation-triggered DNA damage response (DDR). The ionizing radiation-triggered DDR is a signaling network initiated by
lesion recognition and amplified by multiple signaling transducers, which eventually activate downstream effectors to modulate cell fate by manipulating DNA damage
repair, cell cycle progression and apoptosis. Four classical signal transduction pathways, including PI3K/AKT, MAPK/ERK, NF-κB, TGF-β, participate in the regulation of
tumor radiation response. The PI3K/AKT and MAPK/ERK pathways can be activated either through ionizing radiation or through the receptor tyrosine kinases (RTK) of
EGFR and IGFR, leading to the activation of DNA-PKcs and apoptotic-associated proteins, thus regulating the NHEJ process and apoptosis. TGF-β signaling is initiated
by the binding of TGF-β to its receptors, TGFBR1 and TGFBR2, and RTK activation which then phosphorylates SMAD2/3. Activated SMAD2/3 together with SMAD4 trans-
locate into the nucleus to regulate transcription factors and gene expression. TGF-β pathway is necessary for the full activation of ATM in response to DNA damage,
affecting DNA repair by NHEJ and HR. Besides, TGF-β promotes activation of NF-κB pathway via TAK1, a TGF-β activated kinase. Ionizing radiation also activates the
NF­κB pathway by ATM. With the help of NEMO and RIP1, ATM transfer the signal to the cytoplasm and mediate the activation of TAK1 and IκB, which finally promote
NF-κB activation. All these four pathways will finally affect the expression of crucial genes in nucleus which take part in the processes of DNA damage repair, cell cycle
progression and apoptosis. In terms of DNA damage repair, γH2AX, MDC1, 53BP1 and MRE11-RAD50-NBS1 complex act as the core sensors and could directly bind to
the damaged ends of DNA fragments. ATM and ATR are primary transducers in the center of the entire DDR, which can detect various forms of damaged DNA and
recruit multiple downstream effectors, including DNA-PKcs, Ku70/Ku80, LIG4, XRCC4, BRCA1, BRCA2, RAD51 and RAD52, in order to complete two types of DNA damage
repair (NHEJ and HR) processes. Meanwhile, ATM and ATR can separately interplay with CHK2 and CHK1 and activate other cell cycle regulators like CDC25, P53, P21,
so as to disturb the function of cell cycle checkpoints and interfere with the G1/S and G2/M phase progression. 53BP1, tumor protein p53 binding protein 1; ATM, ataxia
telangiectasia mutated kinase; ATR, ATR serine/threonine kinase; BRCA1, breast cancer 1; BRCA2, breast cancer 2; CDC25, cell division cycle 25C; CHK1, checkpoint
kinase 1; CHK2, checkpoint kinase 2; DNA-PKcs, catalytic subunit of DNA-dependent protein kinase; EGFR, epidermal growth factor receptor; H2AX, H2A histone family,
member X; HDM2, HDM2 proto-oncogene; HR, homologous recombination; IGFR, insulin-like growth factor receptor; IκB, inhibitor of κB; Ku70/Ku80, X-ray repair com-
plementing defective repair in Chinese hamster cells 6/5 (XRCC6/5); LIG4, ligase IV; MAPK/ERK, mitogen-activated protein kinase/extracellular signal-regulated kinase;
MDC1, mediator of DNA damage checkpoint 1; MRE11, DNA double-strand break repair protein Mre11; NBS1, nibrin; NEMO, NF-κB essential modulator; NF-κB, nuclear
factor-kappaB; NHEJ, nonhomologous end joining repair; P21, cyclin-dependent kinase inhibitor 1A (CDKN1A); P53, tumor protein p53; PI3K/AKT, phosphatidylinositol
3-kinase/protein kinase B; RAD50, RAD50 homolog; RAD51, RAD51 recombinase; RAD52, RAD52 homolog; RIP1, receptor-interacting protein 1; SMAD2/3, SMAD family
member 2/3; TAK1, TGF-beta-activated kinase 1; TF, transcription factors; TGF-β, transforming growth factor-β; TGFBR1, transforming growth factor, beta receptor 1;
TGFBR2, transforming growth factor, beta receptor 2; XRCC4, X-ray repair complementing defective repair in Chinese hamster cells 4.

rs1801320 (G > C), located in the promoter region of RAD51, result-
ing in upregulated gene expression level through an increased
promoter activity by substituting G for C allele, displayed a pre-
dictive value for RP development in NSCLC patients and dyspha-
gia among HNC patients after radiotherapy (40), which is also
an independent prognostic factor for OS in NSCLC patients (42).
Moreover, a recent study firstly confirmed that RAD51 rs1801321
(G > T) T allele carriers may have a better OS as for cervical can-
cer patients upon radiotherapy (46).
ATM
ATM protein presents kinase activity induced by ionizing radia-
tion, and it is involved in DNA damage detection and cell cycle
progression (47). A number of studies have evaluated the role of
ATM polymorphisms in RP developing for NSCLC patients upon
radiotherapy. Two independent studies have consistently con-
firmed the rs189037 (G > A) A allele as a risk allele for RP (48–50).
Both in vitro and in vivo studies demonstrated that rs189037
may affect ATM expression by reducing transcriptional activity
and interfering nuclear protein binding (49). Another two SNPs
of ATM [rs373759 (G > A) and rs228590 (C > T)] were also found
to be correlated with RP (49,50). In addition, haplotype analysis
showed that the rs189037/rs228590/rs1801516 (G-T-G) haplotype
have a significantly lower risk of severe RP (hazard ratio = 0.52,
95% confidence interval: 0.35–0.79, P  =  0.002) (50). The variant
A allele of ATM rs1801516 (G > A) was correlated with increased
 310 | Carcinogenesis, 2015, Vol. 36, No. 3

Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019

Figure 2.  Strategies for candidate gene selection. DNA repair pathway, cell cycle progression and apoptosis mediated via four radiation-induced signaling pathways
construct a DNA damage response (DDR) signaling network, which can influence cell radiation response by determining cell’s fate either to repair DNA damage or to
undergo apoptosis. Besides DDR, inflammatory responses and oxidative stress reactions may have a considerable effect on radiosensitivity by regulating the tumor
microenvironment. Genes involved in these pathways may serve as good candidates for radiogenomics study. BIM, BCL-2-interacting mediator of cell death; EGF,
endothelial growth factor; FAS, Fas cell surface death receptor; GST, glutathionine S-transferase; HIF-1, hypoxia-inducible factor-1; IL, interleukin; MCL1, myeloid cell
leukemia sequence 1; NFE2L2, nuclear factor, erythroid 2-like 2; SOD2, superoxide dismutase 2; TNF-α, tumor necrosis factor- α.

risk of severe fibrosis in NPC patients (37,38). For prostate cancer
patients (51) and BC patients (52), however, there seems to be no
association between rs1801516 and radiation toxicity.
NBN
NBN is a component of MRE complex (MRE11-RAD50-NBN)
which is involved in damage sensing, signaling and responding
to DNA double-strand breaks (53). HNC patients experiencing
severe oral mucositis after radiotherapy were correlated to NBN
rs1805794 (G > C) with an odds radio of 4.72 (54). However, nega-
tive results have also been found between rs1805794 and radia-
tion toxicity in BC patients (55) and NSCLC patients (42).
Oxidative stress response genes
Radiation treatment exert its antineoplastic effects either directly
by attacking cellular macromolecules, including DNA, or indi-
rectly by generating reactive oxygen species and their by-prod-
ucts. TXNRD2 is a key player in the defense against oxidative
damage, by reducing the oxidoreductase enzyme thioredoxin
involved in reactive oxygen species scavenging (56). rs1139793
(C > T) was found significantly associated with messenger RNA
(mRNA) expression level of TXNRD2 in blood. A recent study with
a relatively long median follow-up time of >17 years has reported
the association between radiation-induced subcutaneous fibrosis
and rs1139793 in a cohort of 92 BC survivors, which was then con-
firmed in an independent cohort of 283 BC survivors (57). Subjects
carrying polymorphic variant rs1695 (A > G) of GSTP1, an antioxi-
dant enzyme, have shown a higher significant risk of developing
fibrosis or fat necrosis, but not acute skin toxicity after radiation
treatment in BC patients (58,59). The endothelial nitric oxide syn-
thase participates in the oxidative stress response by catalyz-
ing the production of the free radical nitric oxide. An increased
risk of acute skin toxicity has been observed in BC patients with
endothelial nitric oxide synthase rs1799983 (G > T) (41).
Apoptosis-related genes
Apoptosis occurs through two main pathways, the intrinsic
pathway that is triggered by various intracellular stimuli, includ-
ing DNA damage and oxidative stress, and the extrinsic pathway
that is stimulated by binding of death ligands of the tumor
necrosis factor superfamily to their cell surface death receptors.
FASL, a member of the tumor necrosis factor superfamily, by
interacting with its receptor FAS, can initiate the extrinsic apop-
totic pathway. The expression of FAS and FASL is increased after
irradiation and has been associated with radiation-induced cell
death (60). FASL  rs763110 (C > T) is located in the binding motif
for the transcription factor CAAT/enhancer-binding protein β
and has been shown to affect the biological activity of FASL pro-
moter and FASL expression. Thurner et al. (61) have provided the
first evidence that the variant allele of rs763110 may have a pro-
tective effect against the development of high-grade late rectal
and/or urinary side effects after prostate cancer radiotherapy.
Polymorphisms of TP53 gene have been investigated as pos-
sible causes of normal tissue radiation sensitivity in numer-
ous cancers. rs1042522 (G > C), the most frequently studied
SNP of TP53 to date, is associated with the ability to induce cell
apoptosis (62). Mountainous evidence have demonstrated that
rs1042522 has a potential in predicting radiation responses, such
as radiation-induced telangiectasia for BC patients (63), RP for
lung cancer patients (48) and disease progression (local recur-
rence and distant metastases) for NPC patients (64). A  study
with a relatively small sample size of 48 prostate cancer patients
noted a new SNP of TP53, rs35117667 (C > T), which was reported
for the first time and has a predictive value for developing acute
skin adverse effects (51). Evidence also showed that the intronic
polymorphism at TP53 rs17883323 (C > A) was linked to high-
grade urinary toxicity among prostate cancer patients (51).
Additionally, TP73 rs3765701 (A > G) was associated with survival
in stage III–IV NSCLC patients receiving chemoradiation therapy
(65).
HDM2 is an essential negative regulator in the p53 path-
way that directly binds to the p53 transactivation domain and
inhibits its transcriptional activity. Initial results from Alsbeih
et  al. (37,38) have indicated that the variant allele of HDM2
rs2279744 (T > G) and rs1196333 (T > A) were associated with
reduced risk to develop late normal tissues complications
among NPC patients. The functional polymorphic variant in
the promoter at rs2279744 has been suggested to increase the
 Guo et al. | 311

Table 1.  Studies addressing the associations between candidate genes with radiation toxicities

Gene investigated SNP site Endpoint (tumor type, N) OR/HR References

XRCC1 rs25487 (G > A) Fibrosis (NPC, 155) OR = 0.41 (35–41)

Radiation pneumonitis (NSCLC, 165) HR = 0.48
Mucositis (HNC, 101) OR = 4.02
Acute skin toxicity (BC, 286) No associa-
tion
XRCC2 rs3218536 (G > A) OS (NSCLC, 228) HR = 1.70 (42)
XRCC3 rs861539 (C > T) Fibrosis (NPC, 120) OR = 3.88 (44,45)
Acute skin toxicity (BC, 87) HR: unquan-
tifiably
high
XRCC4 rs6869366 (G > T) Radiation pneumonitis (NSCLC, 261) HR = 8.67 (43)
XRCC5 rs3835 (G > A) Radiation pneumonitis (NSCLC, 261) HR = 4.69 (43)

Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019

rs1051677 (T > C) Fibrosis (NPC, 155) OR = 0.39 (37)
RAD51 rs1801320 (G > C) Dysphagia (HNC, 101) HR = 4.00 (40,42)
Radiation pneumonitis (NSCLC, 228) HR = 0.52
rs1801321 (G > T) OS (cervical cancer, 193) HR = 0.373 (46)
ATM rs189037 (G > A) Radiation pneumonitis (lung cancer, 253) HR = 2.49 (48,50)
Radiation pneumonitis (NSCLC, 362) HR = 0.49
rs1801516 (G > A) Fibrosis (NPC, 155) OR = 2.86 (37,38,51,52)
Acute and chronic radiation toxicities (prostate cancer, 48) No associa-
Erythema (BC, 83) tion
No associa-
tion
NBN rs1805794 (G > C) Mucositis (HNC, 183) OR = 4.728 (42,54,55)
OS (NSCLC, 228) No associa-
Acute skin reactions (BC, 446) tion
No associa-
tion
TXNRD2 rs1139793 (C > T) Fibrosis (BC, 375) OR = 2.38 (57)
GSTP1 rs1695 (A > G) Necrosis (BC, 57) OR = 2.90 (58,59)
Fibrosis (BC, 257) OR = 2.756
Endothelial nitric rs1799983 (G > T) Acute skin toxicity (BC, 286) OR = 2.473 (41)
oxide synthase
FAS rs763110 (C > T) Late rectal or urinary side effects (prostate cancer, 607) HR = 0.585 (61)
TP53 rs1042522 (G > C) Local recurrence and distant metastases (NPC, 75) HR = 0.30 (48,63,64)
Telangiectasia (BC, 409) OR = 1.66
Radiation pneumonitis (lung cancer, 253) HR = 4.53
rs35117667 (C > T) Acute skin toxicity (prostate cancer, 48) OR = 0.007 (51)
rs17883323 (C > A) Chronic toxicity of urinary tract (prostate cancer, 48) OR = 0.071 (51)
TP73 rs3765701 (A > G) OS (NSCLC, 598) HR = 1.87 (65)
HDM2 rs2279744 (T > G) Fibrosis (NPC, 155) OR = 0.49 (38)
rs1196333 (T > A) Fibrosis (NPC, 155) OR = 0.13 (38)
TGFB1 rs1800469 (C > T) Erythema (BC, 83) OR = 3.10 (38,52,68,69,74,75,79–82)
Erectile dysfunction (prostate cancer, 141) OR = 4.00
Esophageal toxicity (lung cancer, 213) HR = 3.86
Fibrosis (NPC, 155) HR = 0.57
OS (NSCLC, 109) HR = 0.46
Esophagitis (NSCLC, 198) HR = 2.53
Late adverse effects (BC, 1716) No associa-
Overall toxicity (BC, 2782) tion
Late radiation toxicity (BC, 778) No associa-
Gastrointestinal and genitourinary toxicities (prostate cancer, tion
413) No associa-
tion
No associa-
tion
rs1982073 (T > C) Fibrosis (BC, 257) OR = 0.295 (59,70)
Radiation pneumonitis (NSCLC, 164) HR = 0.39

HR, hazard ratio; N, number of sample size; OR, odds ratio.

transcriptional activator SP1 binding, resulting in higher lev- (51) noted that rs2279744 did not play a role in the incidence
els of HDM2 mRNA and protein and the subsequent attenua- of radiotherapy-induced acute and chronic effects in prostate
tion of the p53 pathway (66). On the other hand, Cintra et  al. cancer patients.
 312 | Carcinogenesis, 2015, Vol. 36, No. 3
Fibroblast proliferation gene
TGFB1 is a versatile cytokine involved in inflammation, cell pro-
liferation and fibrosis, which has been convincingly linked to the
development of radiation-induced fibrosis (67). Multiple studies
have observed an association among TGFB1 SNPs and radiation
toxicities, including erythema (52), fibrosis (59), erectile dys-
function (68), esophageal toxicity (69), radiation pneumonitis
(70) and miscellaneous severe complications (38). Among them,
the two most commonly studied SNPs are rs1800469 (C > T) and
rs1982073 (T > C), which have been reported to be correlated
with serum level of TGFB1 (71,72). Unlike the typical studies that
use the clinical radiation toxicities as research endpoint, Kelsey
et al. (73) applied an objective radiologic endpoint, the slope of
the dose–response curve, and confirmed that rs1800469 may
affect radiation sensitivity. In addition, the rs1800469 might act
as an independent predictor for radiotherapy outcomes as the
CC genotype carriers showed better OS, whereas CT/TT geno-
types carriers displayed a higher probability of severe radiation
esophagitis in lung cancer patients (74,75). Meanwhile, certain
attention should be given to ethnic differences as the evidence
noted that rs1982073, associated with RP risk in whites, was not
correlated with RP in Chinese patients (76,77).
Although a large number of exploratory studies and a small
number of confirmatory studies have claimed the positive asso-
ciations between TGFB1 SNPs and the probability of develop-
ing late adverse effects of radiotherapy, it is remarkable that
few studies with relatively large sample size failed to find any
association. Strikingly, results from 124 BC patients’ lympho-
cytes and 15 human fibroblast strains clearly demonstrated that
rs1800469 genotype has no effect on cellular radiosensitivity,
TGFB1 protein secretion or TGFB1 expression (78). This is of par-
ticular importance, because it implies that none of the SNPs or
SNP networks that affect TGFB1 expression seem to be involved
in the regulation of cellular radiosensitivity. Consistently, the
absence of a link with radiation toxicity has just been confirmed
for rs1800469 in BC patients and prostate cancer patients from
different research groups with quite large sample size (79–82).
So, even for TGFB1 polymorphisms which are among the most
extensively examined SNPs in normal tissue radiobiology, it is
difficult to draw any definite conclusion.
Genome-wide approach
The recent development of GWASs has provided a radical alter-
native to the candidate gene approach. GWASs offer the oppor-
tunity to conduct a hypothesis-free survey for SNP associations
without any need for a prior understanding of the biology
underlying the phenotype of interest. For at least two reasons
GWASs represent an important breakthrough in the attempts to
unravel the genetics of complex traits. Above all, GWASs have
dramatically increased the number of compelling SNP associa-
tions reported in the study of various phenotypes (83). Secondly,
they shed important new light on the overall properties of the
allelic architecture presumably underlying most complex phe-
notypes (84,85).
The first GWAS in radiogenomics aimed at identifying SNPs
associated with erectile dysfunction following radiotherapy for
prostate cancer in a specific African-American ancestry with a
modest sample size and noted a significant SNP rs2268363 (86).
With a two-stage design (discovery and replication), the same
research group expanded the sample size and further con-
ducted the GWAS in a cohort composed predominantly of men
of European ancestry. Twelve SNPs, which lie in or near genes
involved in controlling erectile function or cell adhesion and
signaling, have been addressed to be correlated with erectile
dysfunction (87). Meanwhile, they demonstrated that a region
on chromosome 11q14.3 was associated with late rectal bleed-
ing following radiotherapy for prostate cancer patients (88).
Cell line-based GWAS to identify genes associated with radi-
ation sensitivity also yields certain novel biomarkers. Through
the initial screening in 227 human lymphoblastoid cell lines and
further functional validation using small interfering RNA knock-
down in multiple tumor cell lines, it has been demonstrated that
C13orf34, MAD2L1, PLK4, TPD52, DEPDC1B, were all significantly
linked with radiosensitivity (89). More recently, Barnett et  al.
(90) have performed a phase-designed GWAS (1850 discovery +
1733 replication) with a replication phase in three independent
cohorts who had radiotherapy for prostate or BC. Interestingly,
none of the SNPs with potential associations to radiotherapy
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
toxicity in this study have previously considered to be candi-
dates. It is noteworthy to point that the study provided strong
evidence that the identified SNPs were correlated with tumor
site-specific toxicity (90).
As mentioned above, a striking finding in most GWASs is
that the identified SNPs are not involved in any of the genes
related to the major pathways assumed to be of importance
for the investigated phenotype. Many of the identified SNPs
are located in noncoding regions without any known function
(87,90). This experience seriously questions the value of the can-
didate gene approach and broadens our understanding of the
mechanisms underlying radiation-induced normal tissue com-
plications at the same time. The ‘intronic’ genetics and biology
have renewed enthusiasm for exploring functional implications
as it may lead to alternative splicing or microRNAs regulation
(91,92). Noncoding mutations with roles in radiosensitivity will
likely open new doors to understand genome biology and gene
regulation in radiation-induced side effects.
Even though GWASs have certainly provided several interest-
ing insights and proven their worth, they do at the same time
display a number of challenges. First of all, the SNPs detected
through GWASs are mostly limited to ‘index SNPs’, which are a
subset of common variants derived from haplotype blocks, and
that will typically be inefficient to capture the impact of rare var-
iants with minor allele frequency below 5% (93). In other words,
it is unable to uncover the genetics of a complex polygenic trait
once and for all. This is an issue of major importance since the
genetic background of normal tissue radiosensitivity is made
up by a spectrum ranging from rare highly penetrant alleles to
common low-risk alleles (93). Deeper sequencing-based charac-
terization of genomic variation, fine mapping, imputation and
denser SNP array are needed as they can extend the reach of
GWAS to ever lower ranges of minor allele frequency (94,95).
Although GWAS arrays are extraordinarily efficient at genotyp-
ing SNPs, they are less effective at capturing structural varia-
tions, such as insertions, deletions, inversions and copy number
variants, which may also have a role in interindividual varia-
tion in radiation response (96,97). In addition, GWAS is a double-
edged sword as it facilitates the initial identification of a region
but makes it difficult to distinguish causal mutation(s), of which
the challenge will be to separate true associations from the bliz-
zard of false positives (91). Moreover, in order to keep the risk of
false positives at a reasonable level and still obtain acceptable
statistical power, most GWASs have been based on very large
patient cohorts (98). For example, achieving 90% power to detect
an allele with 5% frequency and a factor of 1.2 effect at a sta-
tistical significance of 10−8 requires 24 994 samples. For further
discovery studies, larger GWAS will be necessary to include rare
sequence variants through next generation sequencing. Besides,

rigorous re-testing in an independent validation set is also of
utmost importance.
Challenge and opportunity
Though many positive results have been reported, the findings
have been always inconsistent and lack the ability to replicate.
Thus, none of the associations reported so far can by any means
be regarded as unambiguously proven. This can presumably be
attributed to certain methodological shortcomings. Based on
theoretical considerations and experiences from other scientific
fields, we address a number of critical points in order to suc-
cessfully unravel the genetics of normal tissue radiosensitivity.
Clearly defined endpoint
The search for clinical and biologic biomarkers associated with
late radiotherapy toxicity is hindered by the use of multiple and
different endpoints from a variety of scoring systems, which
make it tanglesome for comparisons across studies and pool-
ing of data. In addition, although simple and useful in clini-
cal, subjective scoring systems (e.g. mild, moderate, severe)
for mixed endpoints (like breast appearance or urinary quality
of life) cannot sufficiently distinguish the underlying biologi-
cal mechanisms, which probably are determined by different
genetic components. Thus, more quantifiable, biologically rel-
evant phenotypes must be clearly and specifically defined, both
to pinpoint the variable genetic components and to compare
data across institutions. Barnett et  al. (99) have proposed such
a novel metric, the Standardized Total Average Toxicity score, to
try to overcome these difficulties.
Confounding factors
Another major issue relating to the phenotypes is confounding
factors. Briefly, not all studies pay equal attention to the potential
confounding effects of differences in tumor site, target volume,
radiotherapy technique, radiation dose, juxtaposed skin surface,
overall treatment time, concomitant chemotherapy, body habi-
tus and comorbidities (8,9). For example, Barnett et al. (100) have
noted that a larger breast volume was the greatest nongenetic
risk factor for the development of late toxicity in a cohort of
1014 BC patients. A  decision tree considered clinicopathologi-
cal variables and two SNPs demonstrated that mean lung dose
was the strongest risk factor for RP among NSCLC patients (36).
For late morbidities, the length of follow-up is a critical issue
because the frequency and severity of late reactions tend to
progress over time (101). Additionally, the differences between
clinical specimen sources (e.g. blood, tissue, plasma), the time
of specimen collection (e.g. before, during, after treatment) and
the assays used for analysis (e.g. genomics, transcriptomics,
proteomics) may also have subtle impact on outcomes (22). It is
essential to account for all dosimetric-, patient- and treatment-
related factors that may predispose to increased late effects in
order to estimate the residual or unexplained toxicity likely due
to genetics. As individualized radiotherapy is increasingly used,
many centers have made great efforts to develop standardized
procedures for storing and evaluating treatment plans and clini-
cal outcomes.
The need for replication study
A very important aspect of any association study, whether it
is derived from a candidate gene approach or whole genome
analysis, is the need for replication studies validating the initial
findings. With a large (N = 1613), well-powered prospective study,
Barnett et  al. (102) have failed to replicate previously reported
Guo et al. | 313
individual radiation toxicity associations with candidate gene
SNPs, implying that the published associations were either false
positives or true weak positives. It is worth to highlight that a
three-staged internal replication study, involving 1938 patients
from three independent cohorts, has positively addressed that
polymorphisms near tumor necrosis factor-α were associated
with overall radiotherapy toxicities in BC patients (103).
Many instances have arisen in which initial findings have
not been reproduced in follow-up studies (104). Considering that
small sample size can provide imprecise or incorrect estimate
of the magnitude of the observed effect, sufficient sample size
is necessary in replication study to convincingly distinguish the
proposed effect from no effect (105). Heterogeneity in clinical
confounding factors and endpoint phenotypes between ini-
tial and replication studies can undermine the opportunity to
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
compare among them. Because many initial studies have been
reported in populations of European descent, the challenge
remains to extend the studies to other populations (106,107).
Data ‘dredging’ is also a major problem, especially when criteria
for defining phenotypes are altered to achieve statistical signifi-
cance worthy of publication (108). Efforts are underway to pool
homogenous patient cohorts with well-documented radiotox-
icity endpoints to provide sufficient statistical power to detect
what will probably be modest functional effects. The National
Cancer Institute-National Human Genome Research Institute
working group on replication in association studies has pub-
lished a comprehensive set of guidelines, providing a number
of essential criteria for establishing positive replication studies
(108).
Predictive models construction
Development of user-friendly tools for the prediction of radia-
tion-induced complications support the patient and the phy-
sician in selecting the best therapeutic approach and avoid
unnecessary worsening of quality of life. In the last years,
numerous efforts have been performed to construct well-cali-
brated, integrated predictive models by combining genotyping
profiles, clinical data and treatment parameters, which are suit-
able in clinical practice to identify patients at risk for developing
radiotherapy-induced toxicities.
Multimetric normal tissue complication probability models,
such as Lyman-Kutcher-Burman model (109), nearest-neighbor
risk prediction model (110), relative seriality model (111) and
Logit-equivalent uniform dose model (111), have been proposed
to evaluate radiation-induced toxicities based on clinical and
dosimetric parameters. By incorporating SNP information into
the Lyman-Kutcher-Burman model based on mean lung dose,
the predictive ability for severe radiation pneumonitis among
patients with NSCLC has been significantly improved (109).
It is worth to note that a robust statistical method, the
EMLasso technique, has been increasingly used for model
selection in predictive radiogenetics. The EMLasso is a penal-
ized logistic regression method with a stochastic expectation–
maximization algorithm handling missing data and 10-fold
cross-validation avoiding overfit (112). In light of this method, a
multicomponent prediction model involving clinical, treatment
and genetic parameters for grade >2 acute esophagitis postra-
diotherapy toxicity in lung cancer patients with a sensitivity
of 84% and a specificity of 75.3% has been constructed by De
Ruyck et al. (113). Clinical usefulness assessment of the model
demonstrated that 69.4% patients predicted to suffer from acute
esophagitis will actually develop this toxicity and will benefit
from therapy modification. Following this, a number of predic-
tion models achieved by EMLasso have been developed recently,
 314 | Carcinogenesis, 2015, Vol. 36, No. 3
for example, late genitourinary complaints predictive model for
prostate cancer (114), and dysphagia predictive model for HNC
(115).
The development of EMLasso is appealing in constructing
predictive models as it selects models with the smallest number
of parameters while preserving predictive value, which really
benefits clinical practice in consideration of time and cost effi-
ciency. Cross-validation is another fortunate property of the
EMlasso that avoids over/under-fitting. Besides, EMLasso is very
suited for datasets with missing values which are common in
radiogenetics studies due to the use of high-throughput geno-
typing techniques lacking a 100% call rate (115). Nevertheless,
validation and assessment of clinical usefulness are necessary
before implementing these models in the clinic for routine use.
Clinical usefulness can be quantified by decision analytic meth-
ods, such as net benefit and decision curve analysis (116,117).
The biological mechanism of genotype–phenotype
associations
The strong association between a genotype and a phenotype
does not necessarily imply that the genetic variation is itself
the cause. The true variants affecting a given phenotype can be
located within genes, or they can be located in regions that have
not yet been assigned any functional elements. Thus, under-
standing the biological mechanism is of utmost importance in
genotype–phenotype association studies.
Since expression levels can be regulated by genetic variants
in regulatory elements, examining the gene expression patterns
after irradiation in vitro or in vivo with microarray to map the
complex traits may shed new light on the pathways activated
by ionizing radiation and help to identify novel candidate genes
(118,119). For example, by comparing the expression profiles of
radiation-resistant and radiation-sensitive NPC patient biopsy
specimens with DNA microarray, Yang et  al. (118) have dem-
onstrated that 26 pathways (such as cell ion homeostasis, cell
proliferation, receptor protein signaling, membrane system,
humoral immune response, as well as cytokines and inflamma-
tion) were probably linked to radiation resistance with biosta-
tistics and bioinformatics analysis. Although most studies have
focused on mRNA profiles, proteomic studies will also be impor-
tant because mRNA expression is not always correlated with
protein expression or posttranslational modification (120,121).
Experiments based on animal models, gene knockout and small
interfering RNAs techniques could provide interesting insights
into functional candidates (122,123). Besides SNPs, structural
variations, such as insertions, deletions, inversions and copy
number variants, may also have important implications on
radiosensitivity (92). Furthermore, epigenetic changes such as
DNA methylation, microRNA and histone modification, may
represent an additional layer of complexity to understanding
cancer cell biology and response to radiation treatment by post-
translational modifications (124–126). A recent study analyzing
DNA methylation changes at >450 000 loci after ionizing radia-
tion has identified that the most differentially methylated genes
were enriched in gene ontology categories relating to cell cycle,
DNA repair and apoptosis pathways, with a radiation dose and
postirradiation time-dependent manner (127). Evaluating the
link among genetic variants, epigenetic modifications and dis-
ease predispositions is currently a very active area in research.
Large cohort, multicentered collaboration
Recent advances in technology and emerging insights into genetic
variations provide unprecedented opportunities. Nonetheless,
to fully exploit these new possibilities, international coopera-
tion is essential. Several national and international cooperation
genetic association studies have been initiated, such as Gene-
PARE (128), RadGenomics (129), RAPPER (130) and REQUITE pro-
ject (131). The International Radiogenomics Consortium has
been established in 2009 in order to foster large-scale collabora-
tive research projects (132). The future of the field is to create
large patient cohorts for multiple cancer types, to validate the
genetic loci and build reliable predictive models.
Conclusion
The identified predictors of radiosensitivity are aimed to ulti-
mately contribute to an algorithm for guiding oncology treatment
decisions. Patients at high risk of developing radiation-induced
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
toxicities may be offered an alternative treatment approach, or,
for patients who have received radiotherapy, advanced planning
corrections can be introduced to better individualized radia-
tion treatment. In addition to predictive and prognostic testing,
the products of the identified genes could become targets for
innovative therapies in susceptible individuals. Although we
are still at the foot of the mountain, we have faith to believe
that research on radiogenomics will help us achieve individual-
ized radiotherapy protocols and broadly personalized medicine,
leading to safer and more effective outcomes.
Funding
National Natural Science Fund (81273595); National High
Technology Research and Development Program (2012AA02A518,
2012AA02A517).
Acknowledgement
We thank F.He, M.Liu and Z.Luo for helpful comments on the
manuscript.
Conflict of Interest Statement: None declared.
References
1. De Ruysscher, D. et al. (2012) Individualised isotoxic accelerated radio-
therapy and chemotherapy are associated with improved long-term
survival of patients with stage III NSCLC: a prospective population-
based study. Radiother. Oncol., 102, 228–233.
2. Ghazali, N. et al. (2012) Genomic determinants of normal tissue toxicity
after radiotherapy for head and neck malignancy: a systematic review.
Oral Oncol., 48, 1090–1100.
3. Parliament, M.B. et al. (2010) Single nucleotide polymorphisms of DNA
repair genes as predictors of radioresponse. Semin. Radiat. Oncol., 20,
232–240.
4. West, C.M. et al. (2011) Genetics and genomics of radiotherapy toxicity:
towards prediction. Genome Med., 3, 52.
5. Aziz, N.M. et  al. (2003) Trends and advances in cancer survivorship
research: challenge and opportunity. Semin. Radiat. Oncol., 13, 248–
266.
6. Bentzen, S.M. et al. (1994) Patient-to-patient variability in the expres-
sion of radiation-induced normal tissue injury. Semin. Radiat. Oncol.,
4, 68–80.
7. Barnett, G.C. et  al. (2011) The impact of clinical factors on the devel-
opment of late radiation toxicity: results from the Medical Research
Council RT01 trial (ISRCTN47772397). Clin. Oncol. (R. Coll. Radiol)., 23,
613–624.
8. Vogelius, I.R. et  al. (2012) A literature-based meta-analysis of clinical
risk factors for development of radiation induced pneumonitis. Acta
Oncol., 51, 975–983.
9. Qin, Q. et al. (2013) Clinical risk factors for late intestinal toxicity after
radiotherapy: a systematic review protocol. Syst. Rev., 2, 39.

10. Andreassen, C.N. et  al. (2002) Does variability in normal tissue reac-
tions after radiotherapy have a genetic basis–where and how to look
for it? Radiother. Oncol., 64, 131–140.
11. Andreassen, C.N. (2005) Can risk of radiotherapy-induced normal tis-
sue complications be predicted from genetic profiles? Acta Oncol., 44,
801–815.
12. Barnett, G.C. et  al. (2009) Normal tissue reactions to radiotherapy:
towards tailoring treatment dose by genotype. Nat. Rev. Cancer, 9, 134–
142.
13. Becze, E. (2014) Radiogenomics offers promise of personalized radia-
tion cancer therapy. ONS Connect, 29, 38–39.
14. Finnon, P. et al. (2008) Evidence for significant heritability of apoptotic
and cell cycle responses to ionising radiation. Hum. Genet., 123, 485–
493.
15. Roberts, S.A. et  al. (1999) Heritability of cellular radiosensitivity: a
marker of low-penetrance predisposition genes in breast cancer? Am.
J. Hum. Genet., 65, 784–794.
16. Gatti, R.A. (2001) The inherited basis of human radiosensitivity. Acta
Oncol., 40, 702–711.
17. Proud, C. (2014) Radiogenomics: the promise of personalized treatment
in radiation oncology? Clin. J. Oncol. Nurs., 18, 185–189.
18. Andreassen, C.N. (2014) The future has begun in radiogenomics! Radio-
ther. Oncol., 111, 165–167.
19. Rosenstein, B.S. et  al. (2014) Radiogenomics: radiobiology enters the
era of big data and team science. Int. J.  Radiat. Oncol. Biol. Phys., 89,
709–713.
20. Parliament, M.B. (2012) Radiogenomics: associations in all the wrong
places? Lancet. Oncol., 13, 7–8.
21. Ivanov, S.D. (2012) [Biochemical markers predicting response to radia-
tion- and radiochemo-therapy in cancer patients]. Biomed. Khim., 58,
635–650.
22. West, C.M. et al. (2007) The genomics revolution and radiotherapy. Clin.
Oncol. (R. Coll. Radiol)., 19, 470–480.
23. Kerns, S.L. et al. (2014) Radiogenomics: using genetics to identify can-
cer patients at risk for development of adverse effects following radio-
therapy. Cancer Discov., 4, 155–165.
24. Norman, A. et al. (1988) The importance of genetics for the optimiza-
tion of radiation therapy. A hypothesis. Am. J. Clin. Oncol., 11, 84–88.
25. Bekiroglu, F. et  al. (2011) Adjuvant radiotherapy and health-related
quality of life of patients at intermediate risk of recurrence follow-
ing primary surgery for oral squamous cell carcinoma. Oral Oncol., 47,
967–973.
26. Lord, C.J. et al. (2012) The DNA damage response and cancer therapy.
Nature, 481, 287–294.
27. Hosoya, N. et al. (2014) Targeting DNA damage response in cancer ther-
apy. Cancer Sci., 105, 370–388.
28. Zhao, L. et  al. (2013) MicroRNA and signal transduction pathways in
tumor radiation response. Cell. Signal., 25, 1625–1634.
29. Smith, J. et al. (2010) The ATM-Chk2 and ATR-Chk1 pathways in DNA
damage signaling and cancer. Adv. Cancer Res., 108, 73–112.
30. Garrett, M.D. et  al. (2011) Anticancer therapy with checkpoint inhibi-
tors: what, where and when? Trends Pharmacol. Sci., 32, 308–316.
31. Furgason, J.M. et al. (2013) Targeting DNA repair mechanisms in cancer.
Pharmacol. Ther., 137, 298–308.
32. Mladenov, E. et  al. (2011) Induction and repair of DNA double strand
breaks: the increasing spectrum of non-homologous end joining path-
ways. Mutat. Res., 711, 61–72.
33. Karar, J. et  al. (2009) Modulating the tumor microenvironment to
increase radiation responsiveness. Cancer Biol. Ther., 8, 1994–2001.
34. Chen, B.P. et al. (2012) New insights into the roles of ATM and DNA-PKcs
in the cellular response to oxidative stress. Cancer Lett., 327, 103–110.
35. Chris, R. et al. (2013) Analysis of single nucleotide polymorphisms and
radiation sensitivity of the lung assessed with an objective radiologic
endpoint. Clin. Lung Cancer, 14, 267–274.
36. Yin, M. et al. (2011) Functional polymorphisms of base excision repair
genes XRCC1 and APEX1 predict risk of radiation pneumonitis in
patients with non-small cell lung cancer treated with definitive radia-
tion therapy. Int. J. Radiat. Oncol. Biol. Phys., 81, e67–e73.
37. Alsbeih, G. et al. (2013) SNPs in genes implicated in radiation response
are associated with radiotoxicity and evoke roles as predictive and
prognostic biomarkers. Radiat. Oncol., 8, 125.
Guo et al. | 315
38. Alsbeih, G. et al. (2014) Among 45 variants in 11 genes, HDM2 promoter
polymorphisms emerge as new candidate biomarker associated with
radiation toxicity. 3 Biotech., 4, 137–148.
39. Alsbeih, G. et al. (2010) Association between normal tissue complica-
tions after radiotherapy and polymorphic variations in TGFB1 and
XRCC1 genes. Radiat. Res., 173, 505–511.
40. Pratesi, N. et al. (2011) Association between single nucleotide polymor-
phisms in the XRCC1 and RAD51 genes and clinical radiosensitivity in
head and neck cancer. Radiother. Oncol., 99, 356–361.
41. Terrazzino, S. et al. (2012) Common variants of eNOS and XRCC1 genes
may predict acute skin toxicity in breast cancer patients receiving
radiotherapy after breast conserving surgery. Radiother. Oncol., 103,
199–205.
42. Yin, M. et  al. (2011) Polymorphisms of homologous recombination
genes and clinical outcomes of non-small cell lung cancer patients
treated with definitive radiotherapy. PLoS One, 6, e20055.
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
43. Yin, M. et al. (2012) Genetic variants of the nonhomologous end joining
gene LIG4 and severe radiation pneumonitis in nonsmall cell lung can-
cer patients treated with definitive radiotherapy. Cancer, 118, 528–535.
44. Cheuk, I.W. et al. (2014) Association of XRCC1 and XRCC3 gene haplo-
types with the development of radiation-induced fibrosis in patients
with nasopharyngeal carcinoma. Mol. Clin. Oncol., 2, 553–558.
45. Mangoni, M. et al. (2011) Association between genetic polymorphisms
in the XRCC1, XRCC3, XPD, GSTM1, GSTT1, MSH2, MLH1, MSH3, and
MGMT genes and radiosensitivity in breast cancer patients. Int.
J. Radiat. Oncol. Biol. Phys., 81, 52–58.
46. Nogueira, A. et  al. (2012) Role of the RAD51 G172T polymorphism in
the clinical outcome of cervical cancer patients under concomitant
chemoradiotherapy. Gene, 504, 279–283.
47. Canman, C.E. et  al. (2012) Activation of the ATM kinase by ionizing
radiation and phosphorylation of p53. Science, 281, 1677–1679.
48. Yang, M. et al. (2011) Association of P53 and ATM polymorphisms with
risk of radiation-induced pneumonitis in lung cancer patients treated
with radiotherapy. Int. J. Radiat. Oncol. Biol. Phys., 79, 1402–1407.
49. Zhang, L. et  al. (2010) ATM polymorphisms are associated with risk
of radiation-induced pneumonitis. Int. J. Radiat. Oncol. Biol. Phys., 77,
1360–1368.
50. Xiong, H. et  al. (2013) ATM polymorphisms predict severe radiation
pneumonitis in patients with non-small cell lung cancer treated with
definitive radiation therapy. Int. J. Radiat. Oncol. Biol. Phys., 85, 1066–
1073.
51. Cintra, H.S. et al. (2013) Investigation of genetic polymorphisms related
to the outcome of radiotherapy for prostate cancer patients. Dis. Mark-
ers, 35, 701–710.
52. Raabe, A. et al. (2012) Association of single nucleotide polymorphisms
in the genes ATM, GSTP1, SOD2, TGFB1, XPD and XRCC1 with risk of
severe erythema after breast conserving radiotherapy. Radiat. Oncol.,
7, 65.
53. Silva, J. et  al. (2012) DNA repair system and prostate cancer progres-
sion: the role of NBS1 polymorphism (rs1805794). DNA Cell Biol., 31,
1182–1186.
54. Venkatesh, G.H. et al. (2014) Polymorphisms in radio-responsive genes
and its association with acute toxicity among head and neck cancer
patients. PLoS One, 9, e89079.
55. Popanda, O. et  al. (2006) Genetic polymorphisms in the DNA double-
strand break repair genes XRCC3, XRCC2, and NBS1 are not associated
with acute side effects of radiotherapy in breast cancer patients. Can-
cer Epidemiol. Biomarkers Prev., 15, 1048–1050.
56. Sugiyama, T. et al. (2010) Thiol-metabolizing proteins and endothelial
redox state: differential modulation of eNOS and biopterin pathways.
Am. J. Physiol. Heart Circ. Physiol., 298, H194–H201.
57. Edvardsen, H. et al. (2013) SNP in TXNRD2 associated with radiation-
induced fibrosis: a study of genetic variation in reactive oxygen spe-
cies metabolism and signaling. Int. J.  Radiat. Oncol. Biol. Phys., 86,
791–799.
58. Falvo, E. et al. (2012) SNPs in DNA repair or oxidative stress genes and
late subcutaneous fibrosis in patients following single shot partial
breast irradiation. J. Exp. Clin. Cancer Res., 31, 7.
59. Terrazzino, S. et  al. (2012) Common variants of GSTP1, GSTA1, and
TGFβ1 are associated with the risk of radiation-induced fibrosis in
breast cancer patients. Int. J. Radiat. Oncol. Biol. Phys., 83, 504–511.
 316 | Carcinogenesis, 2015, Vol. 36, No. 3
60. Nishioka, A. et al. (1999) An augmentation of Fas (CD95/APO-1) antigen
induced by radiation: flow cytometry analysis of lymphoma and leuke-
mia cell lines. Int. J. Mol. Med., 3, 275–278.
61. Thurner, E.M. et al. (2014) Association of genetic variants in apoptosis
genes FAS and FASL with radiation-induced late toxicity after prostate
cancer radiotherapy. Strahlenther. Onkol., 190, 304–309.
62. Dumont, P. et al. (2003) The codon 72 polymorphic variants of p53 have
markedly different apoptotic potential. Nat. Genet., 33, 357–365.
63. Chang-Claude, J. et  al. (2009) Genetic polymorphisms in DNA repair
and damage response genes and late normal tissue complications of
radiotherapy for breast cancer. Br. J. Cancer, 100, 1680–1686.
64. Xie, X. et al. (2013) Expression of pAkt affects p53 codon 72 polymor-
phism-based prediction of response to radiotherapy in nasopharyn-
geal carcinoma. Radiat. Oncol., 8, 117.
65. Yin, J. et al. (2011) Common genetic variants in cell cycle pathway are
associated with survival in stage III-IV non-small-cell lung cancer. Car-
cinogenesis, 32, 1867–1871.
66. Bond, G.L. et al. (2004) A single nucleotide polymorphism in the MDM2
promoter attenuates the p53 tumor suppressor pathway and acceler-
ates tumor formation in humans. Cell, 119, 591–602.
67. Yarnold, J. et al. (2010) Pathogenetic mechanisms in radiation fibrosis.
Radiother. Oncol., 97, 149–161.
68. Peters, C.A. et  al. (2008) TGFB1 single nucleotide polymorphisms are
associated with adverse quality of life in prostate cancer patients
treated with radiotherapy. Int. J. Radiat. Oncol. Biol. Phys., 70, 752–759.
69. Zhang, L. et al. (2010) Association of TGF-β1 and XPD polymorphisms
with severe acute radiation-induced esophageal toxicity in locally
advanced lung cancer patients treated with radiotherapy. Radiother.
Oncol., 97, 19–25.
70.
Yuan, X. et  al. (2009) Single nucleotide polymorphism at
rs1982073:T869C of the TGFbeta 1 gene is associated with the risk of
radiation pneumonitis in patients with non-small-cell lung cancer
treated with definitive radiotherapy. J. Clin. Oncol., 27, 3370–3378.
71. Grainger, D.J. et al. (1999) Genetic control of the circulating concentra-
tion of transforming growth factor type beta1. Hum. Mol. Genet., 8,
93–97.
72. Shah, R. et al. (2006) A molecular mechanism for the differential reg-
ulation of TGF-beta1 expression due to the common SNP -509C-T (c.
-1347C > T). Hum. Genet., 120, 461–469.
73. Kelsey, C.R. et al. (2012) A polymorphism within the promoter of the
TGFβ1 gene is associated with radiation sensitivity using an objec-
tive radiologic endpoint. Int. J.  Radiat. Oncol. Biol. Phys., 82, e247–
e255.
74. Xue, S.L. et al. (2013) Association between single nucleotide polymor-
phisms of the transforming growth factor-β1 gene and overall survival
in unresectable locally advanced non-small-cell lung cancer patients
treated with radio(chemo)therapy in a Chinese population. Med.
Oncol., 30, 512.
75. Guerra, J.L. et  al. (2012) Association between single nucleotide poly-
morphisms of the transforming growth factor β1 gene and the risk of
severe radiation esophagitis in patients with lung cancer. Radiother.
Oncol., 105, 299–304.
76. Niu, X. et al. (2012) A study of ethnic differences in TGFβ1 gene poly-
morphisms and effects on the risk of radiation pneumonitis in non-
small-cell lung cancer. J. Thorac. Oncol., 7, 1668–1675.
77. Wang, L. et al. (2010) TGF-beta1 gene polymorphisms for anticipating
radiation-induced pneumonitis in non-small-cell lung cancer: differ-
ent ethnic association. J. Clin. Oncol., 28, e621–e622.
78. Reuther, S. et al. (2013) No effect of the transforming growth factor β1
promoter polymorphism C-509T on TGFB1 gene expression, protein
secretion, or cellular radiosensitivity. Int. J.  Radiat. Oncol. Biol. Phys.,
85, 460–465.
79. Martin, S. et al. (2010) Test of association between variant tgβ1 alleles
and late adverse effects of breast radiotherapy. Radiother. Oncol., 97,
15–18.
80. Barnett, G.C. et al. (2012) Individual patient data meta-analysis shows
no association between the SNP rs1800469 in TGFB and late radiother-
apy toxicity. Radiother. Oncol., 105, 289–295.
81. Barnett, G.C. et al. (2010) No association between SNPs regulating TGF-
β1 secretion and late radiotherapy toxicity to the breast: results from
the RAPPER study. Radiother. Oncol., 97, 9–14.
82. Fachal, L. et al. (2012) TGFβ1 SNPs and radio-induced toxicity in pros-
tate cancer patients. Radiother. Oncol., 103, 206–209.
83. Altshuler, D. et  al. (2008) Genetic mapping in human disease. Sci-
ence, 322, 881–888.
84. Manolio, T.A. et al. (2009) Finding the missing heritability of complex
diseases. Nature, 461, 747–753.
85. Cirulli, E.T. et al. (2010) Uncovering the roles of rare variants in com-
mon disease through whole-genome sequencing. Nat. Rev. Genet.,
11, 415–425.
86. Kerns, S.L. et  al. (2010) Genome-wide association study to identify
single nucleotide polymorphisms (SNPs) associated with the devel-
opment of erectile dysfunction in African-American men after
radiotherapy for prostate cancer. Int. J. Radiat. Oncol. Biol. Phys., 78,
1292–1300.
87. Kerns, S.L. et al. (2013) A 2-stage genome-wide association study to
identify single nucleotide polymorphisms associated with develop-
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
ment of erectile dysfunction following radiation therapy for prostate
cancer. Int. J. Radiat. Oncol. Biol. Phys., 85, e21–e28.
88. Kerns, S.L. et al. (2013) Genome-wide association study identifies a
region on chromosome 11q14.3 associated with late rectal bleeding
following radiation therapy for prostate cancer. Radiother. Oncol.,
107, 372–376.
89. Niu, N. et  al. (2010) Radiation pharmacogenomics: a genome-wide
association approach to identify radiation response biomarkers
using human lymphoblastoid cell lines. Genome Res., 20, 1482–1492.
90. Barnett, G.C. et al. (2014) A genome wide association study (GWAS)
providing evidence of an association between common genetic vari-
ants and late radiotherapy toxicity. Radiother. Oncol., 111, 178–185.
91. Edwards, S.L. et al. (2013) Beyond GWASs: illuminating the dark road
from association to function. Am. J. Hum. Genet., 93, 779–797.
92. Abbaszadeh, F. et al. (2010) A novel splice variant of the DNA-PKcs
gene is associated with clinical and cellular radiosensitivity in a
patient with xeroderma pigmentosum. J. Med. Genet., 47, 176–181.
93. Andreassen, C.N. et al. (2009) Genetic variants and normal tissue tox-
icity after radiotherapy: a systematic review. Radiother. Oncol., 92,
299–309.
94. 1000 Genomes Project Consortium. et  al. (2010) A map of human
genome variation from population-scale sequencing. Nature, 467,
1061–1073.
95. Tennessen, J.A. et al. (2012) Evolution and functional impact of rare
coding variation from deep sequencing of human exomes. Science,
337, 64–69.
96. Tuzun, E. et  al. (2005) Fine-scale structural variation of the human
genome. Nat. Genet., 37, 727–732.
97. Manolio, T.A. (2013) Bringing genome-wide association findings into
clinical use. Nat. Rev. Genet., 14, 549–558.
98. Chung, C.C. et al. (2011) Current status of genome-wide association
studies in cancer. Hum. Genet., 130, 59–78.
99. Barnett, G.C. et al. (2012) Standardized Total Average Toxicity score:
a scale- and grade-independent measure of late radiotherapy toxic-
ity to facilitate pooling of data from different studies. Int. J. Radiat.
Oncol. Biol. Phys., 82, 1065–1074.
100. Barnett, G.C. et al. (2011) The Cambridge Breast Intensity-modulated
Radiotherapy Trial: patient- and treatment-related factors that influ-
ence late toxicity. Clin. Oncol. (R. Coll. Radiol)., 23, 662–673.
101. Bentzen, S.M. et al. (1989) Latent-time estimation for late cutaneous
and subcutaneous radiation reactions in a single-follow-up clinical
study. Radiother. Oncol., 15, 267–274.
102. Barnett, G.C. et al. (2012) Independent validation of genes and poly-
morphisms reported to be associated with radiation toxicity: a pro-
spective analysis study. Lancet. Oncol., 13, 65–77.
103. Talbot, C.J. et  al. (2012) A replicated association between polymor-
phisms near TNFα and risk for adverse reactions to radiotherapy. Br.
J. Cancer, 107, 748–753.
104. Kerns, S.L. et al. (2014) STROGAR - STrengthening the Reporting Of
Genetic Association studies in Radiogenomics. Radiother. Oncol.,
110, 182–188.
105. Neale, B.M. et al. (2004) The future of association studies: gene-based
analysis and replication. Am. J. Hum. Genet., 75, 353–362.
106. Ioannidis, J.P. et al. (2001) Replication validity of genetic association
studies. Nat. Genet., 29, 306–309.

107. Vieland, V.J. (2001) The replication requirement. Nat. Genet., 29, 244–
245.
108. Chanock, S.J. et  al. (2007) Replicating genotype-phenotype associa-
tions. Nature, 447, 655–660.
109. Tucker, S.L. et  al. (2013) Incorporating single-nucleotide polymor-
phisms into the Lyman model to improve prediction of radiation
pneumonitis. Int. J. Radiat. Oncol. Biol. Phys., 85, 251–257.
110. Saligan, L.N. et  al. (2014) Supervised classification by filter methods
and recursive feature elimination predicts risk of radiotherapy-related
fatigue in patients with prostate cancer. Cancer Inform., 13, 141–152.
111. Bakhshandeh, M. et al. (2013) Normal tissue complication probability
modeling of radiation-induced hypothyroidism after head-and-neck
radiation therapy. Int. J. Radiat. Oncol. Biol. Phys., 85, 514–521.
112. Sabbe, N. et al. (2013) EMLasso: logistic lasso with missing data. Stat.
Med., 32, 3143–3157.
113. De Ruyck, K. et al. (2011) Development of a multicomponent predic-
tion model for acute esophagitis in lung cancer patients receiving
chemoradiotherapy. Int. J. Radiat. Oncol. Biol. Phys., 81, 537–544.
114. De Langhe, S. et  al. (2014) Integrated models for the prediction of
late genitourinary complaints after high-dose intensity modulated
radiotherapy for prostate cancer: making informed decisions. Radio-
ther. Oncol., 112, 95–99.
115. De Ruyck, K. et al. (2013) A predictive model for dysphagia following
IMRT for head and neck cancer: introduction of the EMLasso tech-
nique. Radiother. Oncol., 107, 295–299.
116. Van Calster, B. et  al. (2013) Evaluation of markers and risk predic-
tion models: overview of relationships between NRI and decision-
analytic measures. Med. Decis. Making, 33, 490–501.
117. Steyerberg, E.W. et al. (2010) Assessing the performance of prediction
models: a framework for traditional and novel measures. Epidemiol-
ogy, 21, 128–138.
118. Yang, S. et  al. (2012) Identification of prognostic biomarkers for
response to radiotherapy by DNA microarray in nasopharyngeal car-
cinoma patients. Int. J. Oncol., 40, 1590–1600.
119. Li, L. et  al. (2013) Identification of radioresistance-associated pro-
teins in human nasopharyngeal carcinoma cell lines by proteomic
analysis. Cancer Biother. Radiopharm., 28, 380–384.
Guo et al. | 317
120. Feng, X.P. et  al. (2010) Identification of biomarkers for predicting
nasopharyngeal carcinoma response to radiotherapy by proteomics.
Cancer Res., 70, 3450–3462.
121. Guo, Y. et  al. (2012) Identification of genes involved in radioresist-
ance of nasopharyngeal carcinoma by integrating gene ontology and
protein-protein interaction networks. Int. J. Oncol., 40, 85–92.
122. Fabre, K.M. et al. (2011) Murine Prkdc polymorphisms impact DNA-
PKcs function. Radiat. Res., 175, 493–500.
123. Shang, Z.F. et al. (2010) Inactivation of DNA-dependent protein kinase
leads to spindle disruption and mitotic catastrophe with attenuated
checkpoint protein 2 phosphorylation in response to DNA damage.
Cancer Res., 70, 3657–3666.
1 24. Metheetrairut, C. et  al. (2013) MicroRNAs in the ionizing radia-
tion response and in radiotherapy. Curr. Opin. Genet. Dev., 23,
12–19.
125. Yang, L.H. et al. (2013) Axin gene methylation status correlates with
Downloaded from https://academic.oup.com/carcin/article-abstract/36/3/307/315471 by guest on 27 May 2019
radiosensitivity of lung cancer cells. BMC Cancer, 13, 368.
126. Liang, L. et  al. (2014) MicroRNA-223 enhances radiation sensitivity
of U87MG cells in vitro and in vivo by targeting ataxia telangiectasia
mutated. Int. J. Radiat. Oncol. Biol. Phys., 88, 955–960.
127. Antwih, D.A. et al. (2013) Radiation-induced epigenetic DNA meth-
ylation modification of radiation-response pathways. Epigenetics, 8,
839–848.
128. Ho, A.Y. et  al. (2006) Genetic predictors of adverse radiotherapy
effects: the Gene-PARE project. Int. J.  Radiat. Oncol. Biol. Phys., 65,
646–655.
129. Iwakawa, M. et  al. (2002) [RadGenomics project]. Nihon Igaku
Hoshasen. Gakkai Zasshi., 62, 484–489.
130. Burnet, N.G. et al.; RAPPER Investigators. (2013) RAPPER: the radiog-
enomics of radiation toxicity. Clin. Oncol. (R. Coll. Radiol)., 25, 431–
434.
131. West, C. et al. (2014) The REQUITE Project: validating predictive mod-
els and biomarkers of radiotherapy toxicity to reduce side-effects
and improve quality of life in cancer survivors. Clin. Oncol. (R. Coll.
Radiol)., 26, 739–742.
132. West, C. et al. (2010) Establishment of a radiogenomics consortium.
Radiother. Oncol., 94, 117–118.