Sage Journals: Discover world-class research

Abstract

In the past few decades, it has become increasingly evident that sensitivity to ionising radiation is variable. This is true for tissue reactions (deterministic effects) after high doses of radiation, for stochastic effects following moderate and possibly low doses, and conceivably also for non-cancer effects such as cardiovascular disease, the causal pathway(s) of which are not yet fully understood. A high sensitivity to deterministic effects is not necessarily correlated with a high sensitivity to stochastic effects. The concept of individual sensitivity to high and low doses of radiation has long been supported by data from patients with certain rare hereditary conditions. However, these syndromes only affect a small proportion of the general population. More relevant to the majority of the population is the notion that some part of the genetic contribution defining radiation sensitivity may follow a polygenic model, which predicts elevated risk resulting from the inheritance of many low-penetrance risk-modulating alleles. Can the different forms of individual radiation sensitivities be inferred from the reaction of cells exposed ex vivo to ionising radiation? Can they be inferred from analyses of individual genotypes? This paper reviews current evidence from studies of late adverse tissue reactions after radiotherapy in potentially sensitive groups, including data from functional assays, candidate gene approaches, and genome-wide association studies. It focuses on studies published in 2013 or later because a comprehensive review of earlier studies was published previously in a report by the UK Advisory Group on Ionising Radiation.

Keywords

Individual radiosensitivity Tissue effects Stochastic effects Predictive tests

1. CHARACTERISATION AND MECHANISMS OF RADIATION-INDUCED HEALTH EFFECTS

Exposure to ionising radiation in humans occurs in various settings, including fractionated high-dose exposures, such as those from cancer radiotherapy; a range of acute exposures, such as those experienced by the Japanese atomic bomb survivors; chronic and low-level exposures, such as those received by radiation workers; and fractionated low-dose exposures from diagnostic medical examinations. Radiation-induced health effects can be divided into stochastic effects, such as cancer, which are understood to be induced even after low-dose exposure; and deterministic tissue reactions, such as necrosis, which are only observed after relatively high-dose exposure, and cardiovascular disease and lens opacities, for which much less is known regarding the dose–response relationships. The mechanisms of these effects are diverse. As such, a radiosensitive phenotype towards one type of effect does not necessarily indicate radiosensitivity towards another type (Foray et al., 2016).

Late adverse tissue reactions to high doses of radiation have a complex aetiology. For a long time, they were believed to be solely related to the death of parenchymal cells in tissues (Thames and Hendry, 1987). Today, it is understood that they result both from cell death and interaction among surviving cells from multiple lineages, as well as from the action of various biologically active extracellular molecules (Denham and Hauer-Jensen, 2002). The reaction of normal tissues to high doses of radiation, especially when given in a fractionated manner, involves two complex and partially interacting components: (1) the healing of traumatic wounds which is perturbed by the fractionated radiation exposure; and (2) a combination of specific injuries, not only to the parenchymal cells, but also to connective tissue and, above all, the vascular system. The second component involves chronic inflammation and is believed to be responsible for the progression of injury over a period of many years, resulting in a steadily increasing risk of manifestation as adverse tissue reaction (Jung et al., 2001).

Longer-term effects of radiation exposure include elevated risk of cancer and, at higher levels of exposure, cardiovascular disease. The development of cancer is characterised by the acquisition of distinct biological properties over a long period of time. These hallmarks of cancer are driven by genomic instability (which generates genetic diversity leading to the acquisition of hallmark features) and inflammation (Hanahan and Weinberg, 2011). The carcinogenic effects of ionising radiation are generally understood to result from various types of damage to DNA, including damage to nucleotide bases, single-strand breaks, double-strand breaks (DSBs), and DNA crosslinks (BEIR VII, 2006), with DSBs playing a prominent role (Bernstein et al., 2013). Under normal circumstances, cells undergo cell cycle arrest in response to DNA damage, and the majority of the damage is repaired during this time. When the damage cannot be repaired, the cell dies within a few cell divisions by means of one of several pathways, including necrosis or apoptosis. Incorrectly repaired DNA damage, on the other hand, can cause genomic instability, whereby replication of damaged sites can lead to mutations contributing to cancer development following a lag period of many years.

The range of radiation-associated cardiovascular effects that have been described likely have complex aetiologies which remain poorly understood (Hamada et al., 2014). An important element is endothelial dysfunction, which is associated with increased expression of chemokines, increased endothelial permeability, and adhesion molecules which can result in a pro-fibrotic and pro-inflammatory environment (Stewart et al., 2010; Lowe and Raj, 2014; Kabacik and Raj, 2017). Further factors which are involved are: (1) transforming growth factor beta (Dobaczewski et al., 2011); (2) the renin–angiotensin system (Fyhrquist and Saijonmaa, 2008); (3) cytokine release from mast cells (Engels et al., 1995); and (4) the cardiac sensory nervous system (Sridharan et al., 2014). These various mechanisms interact, leading to different circulatory system pathologies such as myocardial infarction, stroke, and congestive heart failure (Stewart et al., 2010).

2. DETERMINANTS OF INDIVIDUAL RESPONSE TO IONISING RADIATION

Humans respond to ionising radiation in an individually variable manner. This is true for stochastic effects (ICRP, 2007) and tissue reactions (Joiner and van der Kogel, 2009). Individual responses at the tissue level and the stochastic level have been shown (with varying levels of evidence) to be influenced by various factors including age, smoking, diabetes, collagen vascular disease, and genotype (AGIR, 2013). Indeed, it is well established that individuals with certain rare hereditary cancer-prone disorders (e.g. ataxia telangiectasia and Nijmegen breakage syndrome) are particularly radiosensitive (Taylor et al., 1975; Weemaes et al., 1981). However, these cancer susceptibility syndromes only affect a small proportion of the general population (less than one in 10,000). More relevant for the majority of the population is the theory that some part of the genetic contribution defining radiation sensitivity is likely to follow a polygenic model, whereby inheritance of several low-penetrance risk alleles can lead to a larger increase in cancer risk following exposure to a certain radiation dose compared with subjects lacking those risk alleles. This theory (the ‘common disease–common variant’ model) is also plausible for late adverse tissue reactions following radiotherapy, supported by findings that multiple genetic pathways have been implicated, including DNA damage repair, radiation fibrogenesis, oxidative stress, and endothelial cell damage (Barnett et al., 2009).

3. PROSPECTS FOR PREDICTING INDIVIDUAL RESPONSE TO IONISING RADIATION

The fact that individual radiation sensitivity to deterministic and stochastic effects is partly determined genetically raises the question whether it can be inferred from the reaction of cells exposed ex vivo to ionising radiation. Similarly, can individual response be inferred from analyses of the genotype? Ideally, a proposed biomarker would have natural and meaningful variation in the human population and, in the sense of a binary classification test, would identify the most radiosensitive patients (high sensitivity) and few non-radiosensitive patients (high specificity), and be highly reproducible among different laboratories (Bentzen, 1997). For use in human populations, the biomarker would preferably be easily obtained with minimal discomfort or risk to the patient (Hall et al., 2017). Given that levels of a biomarker can change over time and can differ between cells or tissues, the source and appropriate timing of sample collection is also important. Finally, for large-scale population testing, a rapid readout of results is preferable.

This paper reviews the current evidence on assays proposed to predict individual radiation sensitivity, with particular focus on adverse tissue reactions after radiotherapy. It focuses on studies published in 2013 or later because a comprehensive review of earlier studies was published previously in a report by the UK Advisory Group on Ionising Radiation (AGIR, 2013).

3.1. Prediction of individual radiation sensitivity to radiotherapy-related toxicity

Before 2013, the majority of investigations aiming at finding a correlation between the in-vitro sensitivity of cells and the reaction of normal tissue to radiotherapy focused on functional assays, such as the chromosomal aberration test and the clonogenic cell survival test (AGIR, 2013). Thereafter, the effort shifted towards molecular assays, one being the gamma-H2AX assay which is able to measure the formation and repair of DNA DSBs in individual cells (Sedelnikova et al., 2003). Eleven studies have been published since 2013 (Table 1). The largest study involved patients suffering from cancer at various sites (Granzotto et al., 2016). The five patients with grade 5 skin reactions showed a high level of gamma-H2AX foci and a low level of ATM foci in fibroblasts 10 min after an in-vitro exposure, which was interpreted as evidence for impaired DNA repair. In a follow-up study with the same patient cohort, the low level of ATM 10 min after an in-vitro exposure was found to identify grade 3–4 patients with 80% sensitivity (Pereira et al., 2017). Other authors focused on the kinetics of gamma-H2AX focus formation for 24 h post exposure and often observed an impaired decay of foci in peripheral blood lymphocytes of over-responders. However, the sensitivity was very variable, with several publications showing no correlation at all (Mumbrekar et al., 2014; Markova et al., 2015; Pinkawa et al., 2016; Vandevoorde et al., 2016).

Table 1.

Studies using the gamma-H2AX/53BP1/ATM focus assay to investigate whether cellular ex-vivo radiosensitivity predicts clinical radiosensitivity.

Source	Cancer type	Number of patients	Cell type	Toxicity	Measured endpoint	Outcome
Granzotto et al. (2016)	Total 15 sites, 41% BC and 15% PC	100 (5 ORP)	Fibroblasts	Early and late	High residual gH2AX foci and low pATM foci after 10 min	G5 identified with 100% sensitivity
Pereira et al. (2017)	Same patient cohort as in Granzotto et al., 2016	30 (21 ORP)	Fibroblasts	Early and late	Low pATM foci after 10 min	G3–4 identified with 80% sensitivity
Lobachevsky et al. (2015)	12 HNC, 3 GC, 2 BC, 1 RC	28 (16 ORP)	PBL	Late	Combination of residual damage and rate of focus decay	G3–4 identified with 100% sensitivity
van Oorschot et al. (2014)	PC	24 (13 ORP)	PBL	Late (2 years)	Ratio of initial foci to residual foci	G3–4 identified with 69% sensitivity
van Oorschot et al. (2016)	PC	10 (5 ORP)	PBL	Late	Ratio of initial foci to residual foci	G3–4 identified with 100% sensitivity
Chua et al. (2014)	BC	16 (8 ORP)	PBL	Late	Level of residual foci	G3–4 identified with 63% sensitivity
Li et al. (2013)	HNC	25 (10 ORP)	PBL	Early	Level of residual foci	G3–4 identified with 20% sensitivity
Markova et al. (2015)	BC	38 (5 ORP)	PBL	Early and late	Level of residual foci	G2–3 identified with 0% sensitivity
Mumbrekar et al. (2014)	BC	80 (5 ORP)	PBL	Early	Level of residual damage	G3 identified with 0% sensitivity
Pinkawa et al. (2016)	PC	50 (25 ORP)	PBL	Early and late	Level of foci at various time points	QoL scale 0–100 identified with 0% sensitivity
Vandevoorde et al. (2016)	BC	24 (12 ORP)	PBL	Late	Level of residual foci	No correlation between in vitro results and LENT-SOMA > 2

BC, breast cancer; G, Radiation Therapy Oncology Group grade; GC, gynaecological cancer; HNC, head and neck cancer; LENT-SOMA, scale for side effects in all anatomic sites; ORP, over-responding patients; pATM, phosphorylated ATM protein; PBL, peripheral blood lymphocytes; PC, prostate cancer; QoL, quality-of-life grade; RC, rectal cancer. Sensitivity indicates the percentage of patients identified by an assay relative to patients showing a given grade of adverse effect (true positives).

Apart from analysing DSB repair, one study focused on finding a correlation between ex-vivo-induced apoptosis (8 Gy) in peripheral blood lymphocytes and breast fibrosis (Azria et al., 2015; Table 2). Promising results were observed, with a sensitivity of 80% to identify patients with grade 2 reactions. Four studies evaluated the correlation between the level of cytogenetic damage and apoptosis in ex-vivo-irradiated peripheral blood lymphocytes and adverse effects in healthy tissue following radiotherapy (Table 2). No correlation was found. Two studies assessed the predictive value of the capacity of peripheral blood lymphocytes to cope with radiation-induced oxidative stress by measuring the level of 8-oxo-2'-deoxyguanosine (8-oxo-dG) in serum of ex-vivo-irradiated blood of patients (Skiold et al., 2013; Danielsson et al., 2016) (Table 2). Over-responding patients were characterised by a low level of 8-oxo-dG which was interpreted as evidence for impaired ability to cope with oxidative stress. Although the sensitivity was high in the patients with breast cancer (Skiold et al., 2013), it was very low in the patients with head and neck cancer (Danielsson et al., 2016).

Table 2.

Studies investigating whether chromosomal aberrations, micronuclei, apoptosis, and oxidative stress markers predict clinical radiosensitivity.

Source	Cancer type	Number of patients	Cell type	Toxicity	Measured endpoint	Outcome
Azria et al. (2015)	BC	456	PBL	Early and late	Apoptosis	No correlation with early effects. For late effects, G2 and higher identified with 80% sensitivity
Danielsson et al. (2016)	HN	74 (37 ORP)	Blood serum	Late	Level of 8-oxo-dG	G3–4 identified with 8% sensitivity after 2 Gy in vitro and with 0% sensitivity after 5 mGy
Pinkawa et al. (2016)	PC	44 (16 ORP)	PBL	Early and late	G₂ chromosomal aberrations by Giemsa	Spontaneous aberrations higher in ORP group. No difference in radiation-induced aberrations
Schmitz et al. (2013)	PC	20 (10 ORP)	PBL	Early and late	G₀ chromosomal aberrations by FISH	QoL scale 0–100 identified with 0% sensitivity
Skiold et al. (2013)	BC	26 (14 ORP)	Blood serum	Early	Level of 8-oxo-dG	G3–4 identified with 100% sensitivity after 5 mGy in vitro and 26% sensitivity after 2 Gy
Vandevoorde et al. (2016)	BC	24 (12 ORP)	PBL	Late	G₀ and G₂/caffeine MN assay, apoptosis	No correlation between in-vitro results and LENT-SOMA > 2
Yeoh et al. (2016)	PC	106	PBL	Early and late	G₀ MN assay	No correlation between MN and GI symptom score

FISH, fluorescence in-situ hybridisation; G, Radiation Therapy Oncology Group grade; GI, gastrointestinal; MN, micronuclei; ORP, over-responding patients; PBL, peripheral blood lymphocytes; QoL, quality-of-life grade; LENT-SOMA, scale for side effects in all anatomic sites; 8-oxo-dG, 8-oxo-2'-deoxyguanosine. Sensitivity indicates the percentage of patients identified by an assay relative to patients showing a given grade of adverse effect (true positives).

Recent developments in molecular biology have made it possible to analyse single nucleotide polymorphisms (SNPs) in peripheral blood lymphocytes or saliva samples of patients in an attempt to assess the association of genotype with the probability of developing adverse effects to radiotherapy. Two experimental approaches have been applied: (1) quantitative polymerase chain reaction to identify a limited number of known SNPs for which there is some prior evidence; and (2) DNA sequencing to evaluate a larger number of genotyping markers in an agnostic manner. While earlier genetic studies focused on a handful of candidate genes, the development of technologies that can rapidly analyse thousands of genetic markers at relatively low cost, along with the mapping of linkage disequilibrium between common SNPs across the genome (Altshuler et al., 2010) and the definition of functional elements critical for regulation and genomic stability (Dunham and Kundaje, 2012), have allowed the comprehensive examination of the approximately 25,000 coding genes and associated functional elements in humans. The latter technique, known as the ‘genome-wide association study’ approach, uses markers across the genome and is a hypothesis-free approach.

Twelve studies of tissue endpoints have been published since 2013 using the candidate gene approach (Table 3). In addition, one case study was published which is not considered here, as it only included a single over-reacting patient (Lazzari et al., 2017). The magnitude of association between an SNP and an adverse tissue reaction is calculated as the odds ratio (OR). For simplicity, the OR levels are not emphasised, although accurate prediction requires high ORs. Rather, those SNPs which were found to be significantly associated with adverse reactions are listed. The 12 studies were based on patient populations with various types of cancers, and on SNPs in a number of genes related to DNA damage repair, inflammation, and oxidative stress (Table 3). The choice of analysed genes and SNPs was generally based on mechanistic considerations of radiation-induced normal tissue effects rather than on prior empirical evidence. Although there was some overlap in the SNPs across studies, with the exception of ATM, no common SNPs were identified: XRCC1 was analysed by Seibold et al. (2015) and Alsbeih et al. (2014), who found an association between late effects and the following SNPs: rs2682585, G28152A, and rs25487. GSTP1 was analysed by Danielsson et al. (2016) who found an association with the SNPs rs1695 and A313G. Only the SNP rs1801516 in ATM was found to be associated with adverse skin reactions by Alsbeih et al. (2014), Zhang et al. (2016), and Andreassen et al. (2016). The studies of Zhang et al. (2016) and Andreassen et al. (2016) are meta-analyses with strong statistical power.

Table 3.

Studies investigating whether single nucleotide polymorphisms (SNPs) in selected genes predict clinical radiosensitivity.

Source	Cancer type	Number of patients	Toxicity	Genotypes	Outcome
Alsbeih et al. (2014)	HNC	155 (48 ORP)	Late skin (fibrosis)	45 SNPs in 11 genes related to DNA repair	Three SNPs were associated with risk of skin reactions: rs1801516 in ATM gene, rs2279744 in HDM2, rs25487 in XRCC1
Andreassen et al. (2016)	BC and PC	Meta-analysis of studies	STAT, acute and late	rs1801516 in ATM: meta-analysis of published and unpublished studies with a total of 5456 patients	Variant rs1801516 in ATM gene correlates with both early and late STAT
Danielsson et al. (2016)	HNC	74 (37 ORP)	Late osteoradio-necrosis	Eight SNPs in eight genes related to oxidative stress	Variant rs1695 in GSTP1 gene was associated with increased risk of osteoradionecrosis
Edvardsen et al. (2013)	BC	92 (30 ORP)	Late skin effects	528 SNP in 203 genes	Variant rs1139793 in TXNRD2 was associated with grade of subcutaneous fibrosis, and validated in 283 independent patients
Li et al. (2013)	NSCLC	393 (58 ORP)	Pneumonitis	Three SNPs in the CBLB gene related to cell signalling and protein ubiquitination	Variant rs2305035 in CBLB was associated with increased risk of pneumonitis
Li et al. (2016)	NSCLC	167 (63 pneumonitis and 119 oesophagitis)	Pneumonitis or oesophagitis	161 SNPs within predicted microRNA binding sites in major cancer-related genes and 72 SNPs in seven miRNA processing genes	Increased risk of pneumonitis associated with variants rs720014, rs3757, and rs1633445 in DGCR8 gene. Decreased risk of oesophagitis associated with variant rs10274 in RPS6KB2 gene, and variants rs1061280 and rs1061285 in SMO gene
Mumbrekar et al. (2017)	BC	126 (44 ORP)	Early skin (grade ≥2)	22 SNPs in 18 genes related to DDR	Variant rs8193 in CD44 gene was weakly significantly associated with increased risk of skin toxicity (no multiple testing correction applied)
Pu et al. (2014)	NSCLC	201 (70 pneumonitis and 90 oesophagitis)	Pneumonitis or oesophagitis	12,000 SNPs in 904 genes related to inflammation	10 SNPs were associated with risk of oesophagitis (most significant: rs1239344 variant in OSMR gene), nine SNPs were associated with pneumonitis (most significant rs10711 variant in CDK1 gene)
Seibold et al. (2015)	BC	2636 (290 ORP)	Late skin, fibrosis, STAT	305 SNPs in 59 genes	One SNP (rs2682585 in XRCC1 gene) was significantly associated with skin toxicities and STAT but not fibrosis in combined discovery and validation set
Venkatesh et al. (2014)	HNC	183 (54 ORP mucositis and 39 ORP skin)	Early (mucositis) and late skin	22 SNPs in 17 genes related to DNA repair, inflammation, and oxidative stress	Variant rs1805794 in NBN gene and G-C haplotype in NBN gene were associated with increased risk of oral mucositis
Wen et al. (2014)	NSCLC	362 (56 ORP)	Pneumonitis	Eight SNPs in the LIN28 gene related to mRNA regulation	Variants rs314280 and rs314276 in LIN28B gene were associated with pneumonitis
Zhang et al. (2016)	All cancers	Meta-analysis of nine studies with a total of 2000 patients	Late skin effects	rs1801516 in ATM	Variant rs1801516 in ATM gene was significantly associated with late subcutaneous fibrosis, although there was high between-study heterogeneity

BC, breast cancer; DDR, DNA damage response; HNC, head and neck cancer; NSCLC, non-small cell lung cancer; ORP, over-responding patients (defined as having grade 3 or higher side effects); PC, prostate cancer; STAT, standardised total average toxicity.

In contrast to candidate SNP studies, measured SNPs in genome-wide association studies characterise the entire genome with more or less dense coverage. In addition, further unmeasured SNPs located between measured SNPs can be imputed using sequencing data from the 1000 genomes reference panel. However, the number of comparisons is generally much larger than the number of patients, and the signal-to-noise ratio is much lower than for candidate SNP studies. Also, the risk of false-positive findings increases substantially.

Four such studies have been identified among patients with breast and prostate cancer (Table 4). These studies have assembled relatively large numbers of patients (>1000) and have tightly controlled the type I error (patients falsely identified as radiosensitive) by requiring genome-wide statistical significance. Two studies included independent patient series for replication of findings. Among patients with prostate cancer, Fachal et al. (2014) observed strong associations (approximately six-fold OR) between several SNPs in the TANC1 gene and overall late toxicity (≥2 years), which were replicated in two independent patient series. However, these SNPs were apparently not found in the larger genome-wide association study by Barnett et al. (2014) evaluating similar outcomes. In that study, none of the initially associated SNPs replicated in an independent patient series. The three SNPs found by Kerns et al. (2016) and Dorling et al. (2016) have not been independently replicated to date.

Table 4.

Genome-wide association studies to identify single nucleotide polymorphisms (SNPs) predicting clinical radiosensitivity.

Source	Cancer type	Number of patients	Toxicity	Outcome
Barnett et al. (2014)	PC	2011	2 years, seven specific endpoints for PC and three for BC plus overall STAT	23 SNPs significantly associated with overall STAT, 63 with breast toxicity and 91 with prostate toxicity, but none replicated in independent patient series
Barnett et al. (2014)	BC	1572
Dorling et al. (2016)	BC	1160	Late (2 and 5 years), 11 toxicity outcomes plus STAT	Variant rs6964587 in AKAP9 gene significantly associated with breast oedema after 5 years, polygenic breast cancer risk score unrelated to toxicity outcomes. No independent validation
Fachal et al. (2014)	PC	1742	Acute and late (≥2 years), overall STAT	Significant association (approximately six-fold OR) between late toxicity and SNPs at TANC1 locus (rs10497203, rs7582141, rs6432512, rs264663, rs26451, rs264588, rs264631) with replication in two independent patient groups. TANC1 gene encodes BTF3L4P2 and GSTM3P2
Kerns et al. (2016)	PC	Meta-analysis with individual patient data from four GWAS including 1564 patients	2 years, urinary frequency, decreased urinary stream, rectal bleeding, STAT	Significant associations for two SNPs in non-coding regions: rs17599026 (for urinary frequency) and rs7720298 (for urine stream). No independent validation

BC, breast cancer; PC, prostate cancer; STAT, standardised total average toxicity; GWAS, genome-wide association study; OR, odds ratio. Number of patients represents the sum of patients from discovery and validation steps.

Despite the large sample sizes, all studies suffer from a substantial lack of power. For example, Dorling et al. (2016) evaluated the association between over 15 million SNPs and 13 late toxicity outcomes (telangiectasia, oedema, shrinkage, induration) and overall toxicity (standardised total average toxicity) at 2 and 5 years plus pigmentation, pain, and sensitivity at 2 years] using data from 1160 patients. This corresponds to almost 200 million comparisons, or 170,000 comparisons per patient. In contrast, the latest genome-wide association study investigating breast cancer risk SNPs included approximately 150,000 breast cancer cases and 120,000 controls, and investigated only one outcome (Michailidou et al., 2017). In addition, all four studies described in Table 4 include at least some patients from the RAPPER (Radiogenomics: Assessment of Polymorphisms for Predicting the Effects of Radiotherapy) study (Burnet et al., 2013). Overlap between studies might lead to false-positive conclusions.

3.2. Prediction of individual sensitivity towards radiation-induced cancer

With respect to radiosensitivity towards cancer induction, a number of population-based epidemiological studies conducted prior to 2013 have examined radiation-related risk of cancer with respect to genes in the DNA repair and other relevant biological pathways. However, results from functional SNPs have not been convincingly replicated to date (AGIR, 2013). The genome-wide association study approach described previously has successfully identified hundreds of risk loci in germline DNA for various cancer forms (Chung and Chanock, 2011). However, the assessment of gene–environment interaction for most environmental carcinogens, including radiation, has remained elusive. A genome-wide association study of 100 Hodgkin lymphoma survivors treated with radiotherapy identified two loci in the PRDM1 gene potentially associated with increased risk of secondary malignancy (Best et al., 2011), but these loci have yet to be independently confirmed. More recently, a genome-wide association study in US childhood cancer survivors reported some evidence that a locus on 1q41 was associated with subsequent breast cancer risk among survivors who received ≥10 Gy of breast radiation exposure (Morton et al., 2017). Again, these data remain to be replicated.

4. CONCLUSIONS

The aim of this paper was to review the current state of knowledge about individual radiosensitivity and the possibility of its prediction. It focused on reports describing the outcome of functional assays, candidate gene approaches, and genome-wide association studies which have been published since 2013, because a comprehensive review of earlier studies is given in a report by the UK Advisory Group on Ionising Radiation (AGIR, 2013).

The conclusions are similar to those reached by the authors of the earlier report. Although theoretical and empirical considerations suggest that individuals differ in their response to radiation exposure, no strong and consistently validated biomarkers of either tissue or stochastic effects have been identified to date. Studies of functional assays and candidate SNPs have been largely inconclusive. As pointed out by the authors of the AGIR report (AGIR, 2013), there may be several reasons for this:

functional assays are not standardised and there has been little attempt to ensure transferability across laboratories. The studies involve different radiation doses, dose rates, parameters, and assay conditions;

replication and validation studies are rarely carried out;

patient cohorts are heterogeneous, and different scales are often used to quantify adverse tissue effects; and

study designs vary considerably, and few involve power calculation and multi-variate analysis.

Developments in high-throughput molecular biology techniques make it possible to apply whole-genome sequencing to rapidly analyse thousands of genetic markers at relatively low cost, along with the mapping of linkage disequilibrium between common SNPs across the genome. The genome-wide association study approach has identified some promising leads, such as the TANC1 gene as a potential determinant of late toxicity after prostate cancer treatment, with some biological plausibility. While this gene could be an interesting potential target for a toxicity biomarker, further validation is needed. Similarly, loci identified in studies of second cancer will need additional validation in larger sample sizes before they can be further assessed as potential biomarkers.

Although these studies are faced with several challenges, such as the need for large sample sizes, high-quality exposure assessment, and meaningful replication and validation, initial results indicate that the integrated assessment of radiation exposure and genetic and epigenetic alterations may lead to a more nuanced characterisation of radiation-related detriment. In tandem with studying the role of biomarkers of sensitivity, better characterisation of the role of non-radiation risk factors, such as smoking and body mass index, may offer an opportunity to mitigate risk following radiation exposure.

References

AGIR, 2013. Human Radiosensitivity. Report of the Independent Advisory Group on Ionising Radiation. Doc HPA, RCE-21. Advisory Group on Ionising Radiation, Chilton.

Alsbeih

El-Sebaie

Al-Rajhi

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.

Altshuler

D.M.

Gibbs

R.A.

Peltonen

et al. , 2010. Integrating common and rare genetic variation in diverse human populations. Nature 467, 52–58.

Andreassen

C.N.

Rosenstein

B.S.

Kerns

S.L.

et al. , 2016. Individual patient data meta-analysis shows a significant association between the ATM rs1801516 SNP and toxicity after radiotherapy in 5456 breast and prostate cancer patients. Radiother. Oncol. 121, 431–439.

Azria

Riou

Castan

et al. , 2015. Radiation-induced CD8 T-lymphocyte apoptosis as a predictor of breast fibrosis after radiotherapy: results of the prospective multicentre French trial. EBioMed. 2, 1965–1973.

Barnett

G.C.

West

C.M.

Dunning

A.M.

et al. , 2009. Normal tissue reactions to radiotherapy: towards tailoring treatment dose by genotype. Nat. Rev. Cancer 9, 134–142.

Barnett

G.C.

Thompson

Fachal

et al. , 2014. A genome wide association study (GWAS) providing evidence of an association between common genetic variants and late radiotherapy toxicity. Radiother. Oncol. 111, 178–185.

BEIR VII, 2006. Health Risks from Exposure to Low Levels of Ionizing Radiation. Biological Effects of Ionizing Radiation (BEIR) VII Report. National Academy Press, Washington, DC.

Bentzen

S.M.

, 1997. Potential clinical impact of normal-tissue intrinsic radiosensitivity testing. Radiother. Oncol. 43, 121–131.

10.

Bernstein

Prasad

A.R.

Nfonsam

et al. , 2013. DNA damage, DNA repair and cancer. In: Chen

, (ed.) New Research Directions in DNA Repair, Intech, London: 413–465.

11.

Best

Skol

A.D.

et al. , 2011. Variants at 6q21 implicate PRDM1 in the etiology of therapy-induced second malignancies after Hodgkin's lymphoma. Nat. Med. 17, 941–943.

12.

Burnet

N.G.

Barnett

G.C.

Elliott

R.M.

et al. , 2013. RAPPER: the radiogenomics of radiation toxicity. Clin. Oncol. (R. Coll. Radiol.) 25, 431–434.

13.

Chua

M.L.

Horn

Somaiah

et al. , 2014. DNA double-strand break repair and induction of apoptosis in ex vivo irradiated blood lymphocytes in relation to late normal tissue reactions following breast radiotherapy. Radiat. Environ. Biophys. 53, 355–364.

14.

Chung

C.C.

Chanock

S.J.

, 2011. Current status of genome-wide association studies in cancer. Hum. Genet. 130, 59–78.

15.

Danielsson

Brehwens

Halle

et al. , 2016. Influence of genetic background and oxidative stress response on risk of mandibular osteoradionecrosis after radiotherapy of head and neck cancer. Head Neck 38, 387–389.

16.

Denham

J.W.

Hauer-Jensen

, 2002. The radiotherapeutic injury – a complex ‘wound'. Radiother. Oncol. 63, 129–145.

17.

Dobaczewski

Chen

Frangogiannis

N.G.

, 2011. Transforming growth factor (TGF)-beta signaling in cardiac remodeling. J. Mol. Cell. Cardiol. 51, 600–606.

18.

Dorling

Barnett

G.C.

Michailidou

et al. , 2016. Patients with a high polygenic risk of breast cancer do not have an increased risk of radiotherapy toxicity. Clin. Cancer Res. 22, 1413–1420.

19.

Dunham

Kundaje

, ENCODE Project Consortium, 2012. An integrated encyclopedia of DNA elements in the human genome. Nature 489, 57–74.

20.

Edvardsen

Landmark-Hoyvik

Reinertsen

K.V.

et al. , 2013. SNP in TXNRD2 associated with radiation-induced fibrosis: a study of genetic variation in reactive oxygen species metabolism and signaling. Int. J. Radiat. Oncol. Biol. Phys. 86, 791–799.

21.

Engels

Reiters

P.H.

Daemen

M.J.

et al. , 1995. Transmural changes in mast cell density in rat heart after infarct induction in vivo. J. Pathol. 177, 423–429.

22.

Fachal

Gomez-Caamano

Barnett

G.C.

et al. , 2014. A three-stage genome-wide association study identifies a susceptibility locus for late radiotherapy toxicity at 2q24.1. Nat. Genet. 46, 891–894.

23.

Foray

Bourguignon

Hamada

, 2016. Individual response to ionizing radiation. Mutat. Res. 770, 369–386.

24.

Fyhrquist

Saijonmaa

, 2008. Renin-angiotensin system revisited. J. Intern. Med. 264, 224–236.

25.

Granzotto

Benadjaoud

M.A.

Vogin

et al. , 2016. Influence of nucleoshuttling of the ATM protein in the healthy tissues response to radiation therapy: toward a molecular classification of human radiosensitivity. Int. J. Radiat. Oncol. Biol. Phys. 94, 450–460.

26.

Hall

Jeggo

P.A.

West

et al. , 2017. Ionizing radiation biomarkers in epidemiological studies – an update. Mutat. Res. 771, 59–84.

27.

Hamada

Fujimichi

Iwasaki

et al. , 2014. Emerging issues in radiogenic cataracts and cardiovascular disease. J. Radiat. Res. 55, 831–846.

28.

Hanahan

Weinberg

R.A.

, 2011. Hallmarks of cancer: the next generation. Cell 144, 646–674.

29.

ICRP, 2007. The 2007 Recommendations of the International Commission on Radiological Protection. ICRP Publication 103. Ann. ICRP 37(2–4).

30.

Joiner, M.C., van der Kogel, A.J., 2009. Basic Clinical Radiobiology, fourth ed. Hodder Arnold, London.

31.

Jung

Beck-Bornholdt

H.P.

Svoboda

et al. , 2001. Quantification of late complications after radiation therapy. Radiother. Oncol. 61, 233–246.

32.

Kabacik

Raj

, 2017. Ionising radiation increases permeability of endothelium through ADAM10-mediated cleavage of VE-cadherin. Oncotarget 8, 82049–82063.

33.

Kerns

S.L.

Dorling

Fachal

et al. , 2016. Meta-analysis of genome wide association studies identifies genetic markers of late toxicity following radiotherapy for prostate cancer. EBioMed. 10, 150–163.

34.

Lazzari

Natalicchio

M.I.

Terlizzi

et al. , 2017. Single nucleotide polymorphisms and unacceptable late toxicity in breast cancer adjuvant radiotherapy: a case report. Breast Cancer 9, 401–406.

35.

C.R.

W.C.

et al. , 2013. Correlation of dynamic changes in gamma-H2AX expression in peripheral blood lymphocytes from head and neck cancer patients with radiation-induced oral mucositis. Radiat. Oncol. 8, 155–165.

36.

Wang

Liu

et al. , 2016. Single nucleotide polymorphisms in CBLB, a regulator of T-cell response, predict radiation pneumonitis and outcomes after definitive radiotherapy for non-small-cell lung cancer. Clin. Lung Cancer 17, 253–262.

37.

Lobachevsky

Woodbine

Hsiao

K.C.

et al. , 2015. Evaluation of severe combined immunodeficiency and combined immunodeficiency pediatric patients on the basis of cellular radiosensitivity. J. Mol. Diagn. 17, 560–575.

38.

Lowe

Raj

, 2014. Premature aging induced by radiation exhibits pro-atherosclerotic effects mediated by epigenetic activation of CD44 expression. Aging Cell 13, 900–910.

39.

Markova

Somsedikova

Vasilyev

et al. , 2015. DNA repair foci and late apoptosis/necrosis in peripheral blood lymphocytes of breast cancer patients undergoing radiotherapy. Int. J. Radiat. Biol. 91, 934–945.

40.

Michailidou

Lindstrom

Dennis

et al. , 2017. Association analysis identifies 65 new breast cancer risk loci. Nature 551, 92–94.

41.

Morton

L.M.

Sampson

J.N.

Armstrong

G.T.

et al. , 2017. Genome-wide association study to identify susceptibility loci that modify radiation-related risk for breast cancer after childhood cancer. J. Natl Cancer Inst. 11, 109.

42.

Mumbrekar

K.D.

Fernandes

D.J.

Goutham

H.V.

et al. , 2014. Influence of double-strand break repair on radiation therapy-induced acute skin reactions in breast cancer patients. Int. J. Radiat. Oncol. Biol. Phys. 88, 671–676.

43.

Mumbrekar

K.D.

Bola Sadashiva

S.R.

Kabekkodu

S.P.

et al. , 2017. Genetic variants in CD44 and MAT1A confer susceptibility to acute skin reaction in breast cancer patients undergoing radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 97, 118–127.

44.

Pereira

Bodgi

Duclos

et al. , 2017. Fast and binary assay for predicting radiosensitivity based on the nucleoshuttling of ATM protein: development, validation and performances. Int. J. Radiat. Oncol. Biol. Phys. 100, 353–360.

45.

Pinkawa

Brzozowska

Kriehuber

et al. , 2016. Prediction of radiation-induced toxicity by in vitro radiosensitivity of lymphocytes in prostate cancer patients. Future Oncol. 12, 617–624.

46.

Wang

Chang

J.Y.

et al. , 2014. Inflammation-related genetic variants predict toxicity following definitive radiotherapy for lung cancer. Clin. Pharmacol. Ther. 96, 609–615.

47.

Schmitz

Brzozowska

Pinkawa

et al. , 2013. Chromosomal radiosensitivity analyzed by FISH in lymphocytes of prostate cancer patients and healthy donors. Radiat. Res. 180, 465–473.

48.

Sedelnikova

O.A.

Pilch

D.R.

Redon

et al. , 2003. Histone H2AX in DNA damage and repair. Cancer Biol. Ther. 2, 233–235.

49.

Seibold

Behrens

Schmezer

et al. , 2015. XRCC1 polymorphism associated with late toxicity after radiation therapy in breast cancer patients. Int. J. Radiat. Oncol. Biol. Phys. 92, 1084–1092.

50.

Skiöld

Näslund

Brehwens

et al. , 2013. Radiation-induced stress response in peripheral blood of breast cancer patients differs between patients with severe acute skin reactions and patients with no side effects to radiotherapy. Mutat. Res. 756, 152–157.

51.

Sridharan

Tripathi

Sharma

et al. , 2014. Roles of sensory nerves in the regulation of radiation-induced structural and functional changes in the heart. Int. J. Radiat. Oncol. Biol. Phys. 88, 167–174.

52.

Stewart

F.A.

Hoving

Russell

N.S.

, 2010. Vascular damage as an underlying mechanism of cardiac and cerebral toxicity in irradiated cancer patients. Radiat. Res. 174, 865–869.

53.

Taylor

A.M.

Harnden

D.G.

Arlett

C.F.

et al. , 1975. Ataxia telangiectasia: a human mutation with abnormal radiation sensitivity. Nature 258, 427–429.

54.

Thames

H.D.

Hendry

J.H.

, 1987. Fractionation in Radiotherapy, London: Taylor & Francis.

55.

van Oorschot

Hovingh

S.E.

Moerland

P.D.

et al. , 2014. Reduced activity of double-strand break repair genes in prostate cancer patients with late normal tissue radiation toxicity. Int. J. Radiat. Oncol. Biol. Phys. 88, 664–670.

56.

van Oorschot

Hovingh

Dekker

et al. , 2016. Predicting radiosensitivity with gamma-H2AX foci assay after single high-dose-rate and pulsed dose-rate ionizing irradiation. Radiat. Res. 185, 190–198.

57.

Vandevoorde

Depuydt

Veldeman

et al. , 2016. In vitro cellular radiosensitivity in relationship to late normal tissue reactions in breast cancer patients: a multi-endpoint case–control study. Int. J. Radiat. Biol. 92, 823–836.

58.

Venkatesh

G.H.

Manjunath

V.B.

Mumbrekar

K.D.

et al. , 2014. Polymorphisms in radio-responsive genes and its association with acute toxicity among head and neck cancer patients. PLoS One 9, e89079.

59.

Weemaes

C.M.

Hustinx

T.W.

Scheres

J.M.

et al. , 1981. A new chromosomal instability disorder: the Nijmegen breakage syndrome. Acta Paediatr. Scand. 70, 557–564.

60.

Wen

Liu

Wang

et al. , 2014. Genetic variants of the LIN28B gene predict severe radiation pneumonitis in patients with non-small cell lung cancer treated with definitive radiation therapy. Eur. J. Cancer 50, 1706–1716.

61.

Yeoh

E.K.

Krol

Dhillon

V.S.

et al. , 2016. Predictors of radiation-induced gastrointestinal morbidity: a prospective, longitudinal study following radiotherapy for carcinoma of the prostate. Acta Oncol. 55, 604–610.

62.

Zhang

Liu

Wang

et al. , 2016. Single nucleotide polymorphism rs1801516 in ataxia telangiectasia-mutated gene predicts late fibrosis in cancer patients after radiotherapy: a PRISMA-compliant systematic review and meta-analysis. Medicine (Baltimore) 95, e3267.

Human individual radiation sensitivity and prospects for prediction

Abstract

Keywords

1. CHARACTERISATION AND MECHANISMS OF RADIATION-INDUCED HEALTH EFFECTS

2. DETERMINANTS OF INDIVIDUAL RESPONSE TO IONISING RADIATION

3. PROSPECTS FOR PREDICTING INDIVIDUAL RESPONSE TO IONISING RADIATION

3.1. Prediction of individual radiation sensitivity to radiotherapy-related toxicity

3.2. Prediction of individual sensitivity towards radiation-induced cancer

4. CONCLUSIONS

References