Genetic Markers for Prediction of Normal Tissue Toxicity After Radiotherapy

Genetic Markers for Prediction of Normal Tissue Toxicity After Radiotherapy

Genetic Markers for Prediction of Normal Tissue Toxicity After Radiotherapy Jan Alsner, PhD, Christian Nicolaj Andreassen, MD, PhD, and Jens Overgaard...

316KB Sizes 0 Downloads 46 Views

Genetic Markers for Prediction of Normal Tissue Toxicity After Radiotherapy Jan Alsner, PhD, Christian Nicolaj Andreassen, MD, PhD, and Jens Overgaard, MD, DMSc, FRCR, FACR During the last decade, a number of studies have supported the hypothesis that there is an important genetic component to the observed interpatient variability in normal tissue toxicity after radiotherapy. This review summarizes the candidate gene association studies published so far on the risk of radiation-induced morbidity and highlights some recent successful whole-genome association studies showing feasibility in other research areas. Future genetic association studies are discussed in relation to methodological problems such as the characterization of clinical and biological phenotypes, genetic haplotypes, and handling of confounding factors. Finally, candidate gene studies elucidating the genetic component of radiation-induced morbidity and the functional consequences of single nucleotide polymorphisms by studying intermediate phenotypes will be discussed. Semin Radiat Oncol 18:126-135 © 2008 Elsevier Inc. All rights reserved.

C

Single Nucleotide Polymorphisms and GWA Studies

Department of Experimental Clinical Oncology, Aarhus University Hospital, Aarhus, Denmark. Address reprint requests to Jan Alsner, Department of Experimental Clinical Oncology, Aarhus University Hospital, Noerrebrogade 44, Bldg 5, 8000 Århus C, Denmark. E-mail: [email protected]

Definitions and Numbers Single nucleotide polymorphisms (SNPs) account for most of the known genetic variation between individuals and are usually defined as polymorphisms in which the minor variant (allele) is present in at least 1% of a given population. SNPs can affect protein function by altering the amino acid composition or by affecting various aspects of transcriptional and translational control. However, most SNPs are located in regions without any apparent genes or yet-identified functional elements. About 11.8 million SNPs are now included in the National Center for Biotechnology Information public database dbSNP Build 127 (www.ncbi.nlm.nih.gov/projects/ SNP). Nearly half of these SNPs have been validated experimentally, whereas the rest are yet-unconfirmed variations mainly identified through computational analysis. Although not uniformly distributed among the 3 billion base pair (bp) human genome, SNPs thus occur on average every ⬃300 bp. With an average gene length of ⬃27,000 bp,5,6 about 100 SNPs are present in a typical gene. Most SNPs in dbSNP (⬃80%) are single nucleotide variations, also known as “true” SNPs. The remaining SNPs are mainly small insertions/ deletions including 1 or a few nucleotides. Through recent technological developments, arrays have been made available that allow the simultaneous analysis (genotyping) of more than 500,000 SNPs in each sample. Using information from the Human Genome Project and the

ancer patients receiving radiotherapy alone or in combination with chemotherapy display a large patient-topatient variability in their risk of developing normal tissue reactions. Although part of this variability can be ascribed to differences in treatment and patient characteristics, it has become increasingly clear that there might be an important genetic component.1-4 The term “radiogenomics” has been applied to the study of these genetic variants that are associated with the observed interpatient variability.1 One aim of radiogenomics studies is to develop tools for biologically individualized radiotherapy, taking into consideration the individual risk of developing radiation-induced morbidity. Another aim is to improve the ability to reduce the morbidity, through prevention or intervention, by increasing our understanding of the biological mechanisms behind them. In this review, the progress in radiogenomics studies on radiation-induced morbidity will be summarized. Recently, large successful genome-wide associations (GWA) studies have been reported in other research areas. Based on the lessons learned from these studies, suggestions will be made for the design of future studies on genetic variants associated with interpatient variability in radiation-induced morbidity.

126

1053-4296/08/$-see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.semradonc.2007.10.004

Genetic markers HapMap project, it has been possible to design SNP arrays that span the entire genome. The HapMap project describes the statistical relatedness of SNPs by providing a catalog of how SNPs are organized on chromosomes and distributed among different populations (www.hapmap.org). Adjacent SNPs are often linked together, and regions of linked variants that are inherited together are known as haplotypes. Typically, only approximately 5 different common haplotypes (with frequencies ⬎5%) are found for most parts of our genome, and the median haplotype length is ⬃50,000 bp.7 Within a given haplotype, it is therefore possible to identify particular variants that can predict or “tag” the presence of particular variants at other sites. These SNPs that can be used to uniquely identify haplotypes are known as “tag” SNPs. Because we are a relatively young species originating out of Africa,8,9 the haplotypes in non-African populations tend to be subsets of the haplotypes in African populations, and the haplotype blocks tend to be longer in non-African populations. Through random chance and natural selection, the frequency of haplotypes has come to vary from region to region as modern humans have expanded throughout the world. Thus, a given haplotype can occur at quite different frequencies in different populations. Quantitative Complex Traits SNPs provide a powerful tool for GWA studies in which dense sets of SNPs are genotyped to identified genetic variations associated with a certain phenotype. The phenotype (or “trait”) can either be represented by distinct categories (such as disease occurrence or number of tumors) or can display continuous variation (such as measurements of height and weight). Sometimes a threshold must be crossed for the phenotype to be expressed. A quantitative complex trait is a biological trait that has such measurable phenotypic variation.10 Most common diseases and also the risk of radiationinduced morbidity1,2 are regarded as quantitative complex traits. The genetic basis of these traits often involves the effect of several genes. Some genes might affect the phenotype in an almost qualitative “all-or-none” way (like genes that cause dwarfism), but usually each causal gene only makes a small contribution to overall susceptibility making it very difficult to identify the relevant genes.10 Furthermore, quantitative complex traits are often under environmental influences. Because of the complexity of quantitative traits, GWA studies have mainly addressed binary phenotypes (like disease occurrence “yes or no”), but methods are currently being developed for mapping quantitative traits displaying continuous variations.11 Whole Genome Strategies Different strategies can be applied to GWA studies. One strategy is the stage design,12 elegantly exemplified in a recent study on the identification of novel breast cancer susceptibility loci13 (in genetic terms, a locus refers to a position on a chromosome and may refer to a marker, a gene, or any other landmark that can be described). In the first stage, more than 225,000 SNPs were analyzed in about 400 cases and 400 controls. In the second stage, more than 10,000 SNPs were selected based on the significance of the difference in genotype frequency between cases and controls. These were ana-

127 lyzed in about 4,000 cases and 4,000 controls. In the final step, 30 of the most significant SNPs were analyzed in about 22,000 cases and 22,000 controls.13 This study showed that common susceptibility loci (with a minor allele frequency ⬎10%) can be detected by a stage-design approach. Because the study only included a low number of SNPs with minor allele frequency ⬍10%, the power of the stage-design approach to detect low-frequency risk alleles could not be determined. Another concern is the possibility of missing alleles that might be common but confer a low risk. One such SNP that did not pass the statistical requirements for going from stage 1 to stage 2 is a caspase 8 variant previously validated in about 17,000 of the cases and 16,000 of the controls from stage 3.14 Another strategy is to go directly for the large number of SNPs in large cohorts. The first impressive example of such a strategy has been presented by the Welcome Trust Case Control Consortium (WTCCC) studying more than 500,000 SNPs in 14,000 cases (2,000 cases for each of seven common diseases) and 3,000 shared controls.15 Validation studies have been performed and have confirmed the importance for almost all the newly identified susceptibility loci.16-20 The Need for Validation A very important aspect of any association study, whether it is based on a candidate gene approach or whole genome analysis, is the need for replication studies validating the initial findings. The National Cancer Institute-National Human Genome Research Institute (NCI-NHGRI) Working Group on Replication in Association Studies has published a comprehensive set of guidelines, providing a number of important points to consider both when reporting (and evaluating) initial genotype-phenotype reports and also suggesting a number of essential criteria for establishing positive replication studies.21 Understanding the Biological Function of Genotype-Phenotype Associations Another important point is that strong associations between a genotype and a phenotype does not necessarily imply that the genetic variation is itself the cause and SNPs identified through WGA studies often serve as marker SNPs for particular genomic regions. 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, understanding the biological function and importance of genotype-phenotype associations identified through WGA studies is challenging. One way to address this issue is to increase the number of SNPs investigated dramatically. Using data on “tag” SNPs from projects like HapMap7 to infer the genotype of unidentified SNPs, much larger datasets of SNPs can now be generated “in silico” (ie, performed by computer simulation), and these can be tested for associations in exactly the same way as truly genotyped SNPs.22 Although very powerful, this approach has some obvious limitations when the population structure in the cohort used in the association study differs from the cohort used to define linkage patterns. Another promising development is the progress made by the ENCyclopedia Of DNA Elements consortium aiming at identifying every sequence element with functional properties in the human genome.23 The recent

128 report analyzing 1% of the genome shows a surprisingly large number of long RNA transcripts from genomic regions previously not thought to be transcribed.24 Perhaps more relevant to the studies on SNPs are the studies by ENCyclopedia Of DNA Elements on small evolutionary conserved regions that are not part of known genes.24 It will be interesting to see if such studies can provide clues to the functional consequences of SNPs located in these regions.

Genetic Variations and Radiation-Induced Morbidity The Candidate Gene Approach So far, association studies on genetic variants and radiationinduced morbidity (Table 1) have all used the candidate gene approach. Based on mechanistic understanding of the radiation pathogenesis of early and late morbidity, a number of genes have been selected for further analysis. These analyses have either been restricted to known polymorphisms (most often SNPs but in a few cases also microsatellites) or used various resequencing approaches to identify rare variants. Although the focus of this review is on SNPs, both types of studies have been included in Table 1 because the latter approach also identifies known polymorphisms. With about 40 studies published so far, it is tempting to perform a meta-analysis on the most frequently studied genes. However, it is unlikely that any firm conclusions, being either negative or positive, can be drawn from such an analysis. First of all, from a genetic point of view, radiationinduced morbidity most likely does not represent a single phenotype.1 Although patients affected with radiosensitive syndromes do seem to have a general increased risk of both early and late morbidity,25 there are clear indications that differences exist between the genetic component of various types of radiation-induced morbidity in unselected patients. One example is from studies on breast cancer patients treated with more than one radiation field in which no correlation was observed between the risk of two different endpoints in the same patient.26,27 Strong correlations only existed between the observed risk of a particular endpoint in one field and the risk of the same endpoint in a different field in the same patient,27 indicating an element of tissue specificity in any genetic determinants. We have previously proposed a model viewing radiation-induced morbidity as a quantitative trait with some genes affecting overall radiosensitivity and others displaying a high degree of tissue specificity.1,3 If this is correct, great care should be taken when trying to compare results from different studies. Even if the same overall endpoint is evaluated, clinically defined phenotypes might represent a different underlying molecular pathology. As an example, alterations in breast appearance after irradiation28 might not reflect exactly the same biological mechanisms as when late radiation morbidity is assessed by palpation of subcutaneous induration.29 Also, some studies have chosen to look at selected overreactors; other studies include cohorts of consecutive patients, and some include reactors and nonreactors in a case-control like manner. Any genetic associa-

J. Alsner, C.N. Andreassen, and J. Overgaard tions identified in one of these types of studies may not be the same as in patients with significantly less or more severe phenotypes. Particularly for late effects, the length of follow-up is also very critical. Confounding Factors Possible differences in confounding factors is another reason that studies published so far are difficult to compare.3 Briefly, not all studies pay equal attention to the potential confounding effects of differences in radiation dose and type, target volume, target dose specification (especially when at a variable depth like tumor location), overall treatment time, fractionation, concomitant chemotherapy, juxtaposed skin surfaces, immobilizing and dose-modifying devices, and comorbidity. An example of the latter is connective tissue diseases, which are associated with an increased risk of late radiation morbidity30 and presumably have their own distinct genetic components. One way to evaluate heterogeneity in treatment characteristics in SNP association studies is to generate dose-response curves and calculate median effective dose (ED)50 values for patients with different genotypes, as shown in Figure 1. In this study on breast cancer patients treated with postmastectomy radiotherapy using a 3-field technique,31,32 most of the patients received hypofractionated radiotherapy and there was a large variance in the dose received at skin level between patients and between different fields in the same patient.33 A simple scoring based on the absence and presence of fibrosis would therefore only reflect differences in treatment characteristics27 and would not have provided any meaningful information regarding underlying genetic factors associated with the risk of fibrosis. Another approach is to generate dose-volume histograms for patients stratified by genotype, as described in a recent study on the incidence of rectal bleeding in patients treated with brachytherapy for prostate cancer.34 Tagging SNPs Versus Functional Variants The likely tissue-specificity of some genetic determinants, the differences in patient cohorts, the length of follow-up, and the differences in confounding factors are all well-known factors.1-4 A less recognized but very important problem has recently emerged, namely the issue of haplotypes and the difference between tagging genetic variants and functional variants actually affecting some biological mechanism. An illustrative example is the story of XRCC1, the gene encoding the X-ray repair cross-complementing 1 (XRCC1) protein required for DNA single-strand break repair in human cells.35 As listed in Table 1, several studies have evaluated individual SNPs in this gene, and the results have been fairly inconsistent. It now appears that the SNPs analyzed so far (Arg194Trp, Arg280His, and Arg399Gln) might merely serve as tagging SNPs for a potentially functionally important haplotype. Not even all 3 in combination can be used to fully identify this haplotype, which is characterized by a different SNP, -77C¡T, in the 5=-untranslated region (5=-UTR).35 Furthermore, there are large ethnic differences in the frequency of the 5=-UTR SNP alleles and the haplotypes, which can also explain the discrepancy between some studies.35 To complicate the story further, it is still not clear whether the 5=-UTR SNP itself has any functional consequences or

First Author, Year

Conclusion

CAT, SOD2, MPO, eNOS

446

Ambrosone, 200663

GSTA1, GSTP1, GSTM1, GSTT1

446

Andreassen, 200329

TGFB1, SOD2, XRCC1, XRCC3, APEX

41

Andreassen, 200528

TGFB1, SOD2, XRCC1, XRCC3, APEX, ATM

52*

Andreassen, 200664

ATM

41

Andreassen, 200633

TGFB1, SOD2, XRCC1, XRCC3, APEX, ATM

120

Angele, 200365

ATM

254

Appleby, 199766 Borgmann, 200252

ATM ATM, NBS, MRE11, RAD50, DNA ligase IV

Brem, 200635

XRCC1

Bremer, 200367

ATM

10

Cesaretti, 200568

ATM

37

Cesaretti, 200734

ATM

108

Chang-Claude, 200569

XRCC1, APEX, XPD

446

Clarke, 199870

ATM

Damaraju, 200671

83

De Ruyck, 200572

ATM, BCL2, BRCA1, BRCA2, CY3A5*3, CYP11B2, CYP17A1, CYP1A1, CYP2C13*3, CYP2C9*2, CYP2D6*4, CYP2D6*6, CYP2D6*8*14, DNA lig 4, ERCC2, XPF, ESR1, NR3C1, MLH1, MSH6, NBN, RAD51, RAD52, TGFB1, XRCC1, XRCC2, XRCC3 XRCC1, XRCC3, OGG1

No significant associations with acute skin toxicity. Association between obesity and skin toxicity possibly modified by MPO and eNOS SNPs GSTP1 codon 105 Val allele significantly associated with increased risk of acute skin toxicity Risk of subcutaneous fibrosis significantly associated with the TGFB1 position ⴚ509 T and codon 10 Pro alleles, SOD2 codon 16 Ala and XRCC1 codon 399 Arg alleles. XRCC3 codon 241 Thr allele associated with risk of subcutaneous fibrosis and telangiectasia Risk of altered breast appearance significantly associated with the position ⴚ509 T and codon 10 Pro alleles in 26 matched case-control pairs Risk of subcutaneous fibrosis significantly associated with the codon 1883 Asp/Asn and Asn/Asn genotypes No significant associations between the investigated SNPs and risk of radiation-induced subcutaneous fibrosis Significant association between the codon 1853 Asn/Asn genotype and risk of various acute and late adverse normal tissue reactions, intronic IVS22 to 77 CC genotype associated with reduced risk No ATM mutations detected in 23 patients with severe acute or late toxicity No mutations detected in five patients with severe late toxicity, possible DNA ligase IV polymorphism detected in one patient Haplotype consisting of majority alleles in 4 polymorphic sites associated with increased radiosensitivity (mixed endpoint) No indications of increased acute or late radio-sensitivity in 10 patients being heterozygous for pathogenic ATM mutations Possession of missense mutations significantly associated with radiation-induced rectal bleeding and erectile dysfunction Possession of genetic variants significantly associated with radiation-induced rectal bleeding XRCC1 codon 399 Gln allele in combination with APEX codon 148 Glu allele significantly associated with reduced risk of acute skin reactions in a sub group of 104 patients (8 cases and 96 controls) No ATM truncations detected in neither five patients with severe acute toxicity nor in four controls SNPs in DNA lig 4, ERCC2 and CYP2D6 identified as putative markers of rectal and bladder toxicity

De Ruyck, 200573

XRCC1, XRCC3, XRCC5

62

Ahn,

Gene(s) Investigated

23 5 247

9

62

XRCC3 IVS5 to 14 G allele significantly associated with increased risk of late gastro-intestinal damage, XRCC1 194 Trp allele with reduced risk Significant associations between microsatellite polymorphisms in XRCC3 and late toxicity

129

N

200662

Genetic markers

Table 1 Summary of Studies Associating Genetic Variants With Radiation-Induced Morbidity

130

Table 1 Continued First Author, Year

Gene(s) Investigated

Conclusion Non-significant association between possession of the ⴚ509 TT and codon 10 Pro/Pro genotypes and risk of late gastro-intestinal damage GSTP1 codon 105 Val allele significantly associated with increased risk of pleural thickening No evidence of excessive acute toxicity in 21 patients with BRCA1 or BRCA2 mutations TGFB1 codon ⴚ509 T allele significantly associated with increased risk of radiation induced fibrosis, XRCC1 codon 399 Gln allele associated with telangiectasia No association between the SOD2 codon 16 Val/Ala SNP and radiation-induced alteration of breast appearance in 41 cases and 39 matched controls Three patients with ‘significant’ ATM mutations in 17 severe late reactors, no mutations in four controls, difference not statistically significant Risk of rectal bleeding significantly associated with the codon 1853 Asn allele Significant association between missense mutations and severe subcutaneous late damage. Risk of need for long-term feeding tube after radiotherapy significantly reduced in patients with the T2505C (codon 835 Ser/Ser [silent]) SNP No BRCA1 or BRCA2 mutations detected in 22 patients with severe acute or late toxicity Codon 399 Gln allele in combination with the codon 194 Trp allele associated with increased risk of various acute and late adverse normal tissue reactions No ‘unequivocal’ ATM mutations in 20 patients with severe acute or late toxicity No exacerbation of acute or late toxicity in 71 patients with BRCA1 or BRCA2 mutations compared to 213 matched controls No association with acute skin toxicity Significant associations between microsatellite polymorphisms in XRCC3/XRCC5 and severe acute or late toxicity in 8 radiosensitive patients and 11 “normal reactors” Risk of subcutaneous fibrosis significantly associated with the position ⴚ509 T and codon 10 Pro allele in 15 breast cancer patients with severe subcutaneous fibrosis compared to 88 controls No ATM truncations detected in 15 patients with severe late toxicity Nonconservative G1441A transition found in 1/19 and a T1440C substitution found in 6/19 radiosensitive patients. No control group to compare with One ATM truncation in 41 patients with radiation induced breast shrinkage, none in 39 matched controls, difference not statistically significant Non-significant association between TP53 codon 72 SNP and acute skin toxicity but only in a subgroup No evidence of excessive acute or late toxicity in 13 ataxia-telangiectasia heterozygotes

TGFB1

Edvardsen, 200775

GSTM1, GSTP1, GSTM1

Gaffney, 199876

BRCA1, BRCA2

Giotopoulos, 200777 Green, 200278

TGFB1, XRCC1, APEX, DHFR, CX3CR1, Hyl-1, MS, MTHFR SOD2

Hall, 199879

ATM

17

Ho, 200780 Iannuzzi, 200281

ATM ATM

131 46

Kornguth, 200582

ERCC4

130

Leong, 200083

BRCA1, BRCA2

Moullan, 200384

XRCC1

254

Oppitz, 199985 Pierce, 200086

ATM BRCA1, BRCA2

20 284

Popanda, 200687 Price, 199788

XRCC3, XRCC2 XRCC1, XRCC3, XRCC5

446 18

Quarmby, 200389

TGFB1

103

Ramsay, 199890 Severin, 200191

ATM hHR21

15 19

Shayeghi, 199892

ATM

80

Tan, 200693

TP53, CDKN1A

Weissberg, 199894

ATM

253 21 167 80*

22

446 13

NOTE. *These studies were based on patient cohorts that were partially identical, and the results for the SOD2 codon 16 SNP cannot be considered as independent.

J. Alsner, C.N. Andreassen, and J. Overgaard

N 78

De Ruyck, 200674

Genetic markers TGFβ1 Codon 10 Pro/Pro Codon 10 Pro/Leu Codon 10 Leu/Leu

100

Severe fibrosis (%)

131

80 60 40

ED50 Enhancement ratio 1.21 (1.06 – 1.39)

20 0 30

40

50

60

70

80

Radiation dose (Gy) (equivalent dose of 2 Gy per fraction) Figure 1 Dose-response curves for subcutaneous fibrosis in 41 breast cancer patients treated with postmastectomy radiotherapy using a 3-field technique stratified by the their TGF-␤1 codon 10 genotype. The enhancement ratio (blue) between patients with Pro/Pro and Leu/Leu is the ratio between ED50 values (defined as the radiation dose, which, on average, is expected to cause moderate or severe fibrosis in 50% of the treatment fields) with 95% confidence intervals. (Adapted with permission.29)

whether it acts as a tagging SNP for another yet-unidentified variant. It is also possible that some of the commonly studied SNPs may affect XRCC1 function.35 TGFB1, the gene encoding transforming growth factor ␤1 (TGF-␤1), is another example of a frequently studied gene where various SNPs have been analyzed individually (Table 1). These SNPs are in linkage disequilibrium,36 meaning that they are linked in distinct haplotypes and that it is very difficult to compare individual SNPs from different cohorts if the haplotypes are not reported. As with XRCC1, it is still not clear which SNPs/haplotypes are actually affecting the production or secretion or activity of TGF-␤137-40 and the distribution of genotypes varies among ethnic populations.41 The Candidate Gene Approach: A Summary To summarize, about 40 studies, each including relatively few patients, have addressed the associations between various genetic variants and different kinds of radiation-induced morbidity. For a number of reasons, these studies cannot be easily compared, and no firm conclusions (negative or positive) can be made. On the other hand, many of the individual studies have been positive, and the most important question now does not seem to be whether there is a genetic component to the large patient-to-patient variability in their risk of developing normal tissue reactions but rather how to analyze it.

Future Directions: GWA Studies Genetic (SNP) association studies can be divided into different components. The first step is the identification of candidate association SNPs followed by validation of these SNPs. The next steps involve identification of the functional genetic

variants linked with these SNPs and elucidation of the possible interactions between the genetic variants and any environmental factors. As described earlier, successful GWA studies in other research areas have recently been published on the first 2 steps, identification and validation. One characteristic of these studies is that the phenotype investigated is disease occurrence, which, compared with radiation-induced morbidity, makes it less complicated to distinguish cases from controls. Another characteristic of these studies is that they are performed by very large consortia. The current speed of development in genotyping technology and bioinformatics makes it difficult to provide detailed suggestions for how to perform GWA studies in the field of SNPs and radiation-induced morbidity. Considering the complexity of the problem and the number of cases and controls needed, formation of a large research consortium could be a step in the right direction. One of the major issues that need to be addressed is the description of the phenotypes. Although clinically very useful and relevant, subjective scoring systems (eg, mild, moderate, or severe) for mixed endpoints (like breast appearance or urinary quality of life) do not allow for sufficient resolution of the underlying biological mechanisms, which probably are affected by different genetic components. More quantifiable, biologically relevant phenotypes need to be defined, both to pinpoint the variable genetic components and to compare data across institutions. Another major issue relating to the phenotypes is the problem of confounding factors and comorbidity. As described earlier, confounding factors have a profound impact on the risk of morbidity and can outweigh any influence of genetic factors. Improved guidelines are needed for the recording of treatment-related factors and handling of dosimetric data. As individualized radiotherapy is increasingly used, standardized methods for storing and evaluating treatment plans are urgently needed. For late morbidity, the length of follow-up is a critical issue because the frequency and severity of late reactions tend to increase with time.42,43 In genetic association studies, this has been referred to as misclassification bias (ie, when a proportion of the controls either have the phenotype or will develop it in the future). With a modest misclassification, the effect is fortunately limited. In the WTCCC study, it was estimated that the loss of power by a 5% misclassification would correspond to a reduction of the sample size by 10%.15 Several national and international genetic association studies, each recruiting thousands of patients, have been initiated including Genetic Predictors of Adverse Radiotherapy Effects (Gene-PARE),44 RadGenomics,45 and Radiogenomics: Assessment of Polymorphisms for Predicting the Effects of Radiotherapy.46 These studies are each approaching sample sizes needed for WGA studies like the WTCCC study.15 Alternatively, a 2-step WGA approach12 could be applied, somewhat like the first 2 stages of the breast cancer susceptibility study.13 In either case, large validation studies are still needed. For this, the GENEPI 2 project aiming at a database with biological material and clinical data from at least 15,000 patients47 seems ideally suited. However, the optimal design of any future WGA studies to a

J. Alsner, C.N. Andreassen, and J. Overgaard

132

number of other important points to consider, this recommendation can be found in the comprehensive guidelines by the NCI-NHGRI Working Group on Replication in Association Studies.21 Candidate gene studies can provide valuable information on the composition of quantitative complex traits and the functional consequences of different SNPs. One approach is to break down the clinical phenotype into several intermediate phenotypes, also known as endophenotypes, which correlate with the clinical phenotype but are associated more closely with the genetic variants and may be portrayed by gene/protein expression patterns and may be applicable to pathway analysis.48,49 Figure 2 presents examples of proposed intermediate phenotypes that either have shown or may be expected to show some correlations with clinical endpoints. These include clonogenic survival,50,51 chromosome aberrations,52,53 messenger RNA expression patterns,54-57 differentiation,58,59 microRNA expression patterns, DNA repair, reactive oxygen species (ROS) scavenging, extracellular matrix remodeling, vascular damage, cytokine secretion, and cell-cell interactions. Another approach to identify

Figure 2 Intermediate phenotypes between genotypes and the clinical phenotypes of normal tissue morbidity. CNVs, copy number variations.

All patients

80

Induced gene expression Patients

60

Genes

Moderate/severe fibrosis (%)

100

40 20

Pattern A

Pattern B

0 30

40

50

60

70

80

large extend depend on how the issues of phenotype description and confounding factors can be addressed. These are not easily resolved and can be expected to have a negative impact on the power of WGA studies. It is therefore important to continue studies based on the candidate gene approach to ensure that relevant genes and SNPs are included in any upcoming large validation studies and are not lost through lack of power in initial WGA studies. Also, current SNP arrays do not have the power to resolve all possible haplotypes and, as described earlier for XRCC1 and TGFB1, it is important not to miss the relevant functional variant.

Future Directions: Candidate Gene Studies A basic recommendation for future association studies on candidate genes is that data on the haplotypes of each gene (or different parts of the gene) should always be included along with data on the individual SNPs. Together with a

Moderate/severe fibrosis (%)

100 Patients with pattern A ( )

80

Patients with pattern B ( )

60 40

ED50 Enhancement ratio 1.25 (1.12-1.40)

20 0 30

40

50

60

70

80

Radiation dose (Gy) (equivalent dose of 2 Gy per fraction)

Figure 3 Dose-response curves for subcutaneous fibrosis in 26 breast cancer patients treated with postmastectomy radiotherapy using a 3-field technique. (Top) Dose-response curves for all patients (insert: induced gene expression patterns in fibroblast irradiated with 3 ⫻ 3.5 Gy/3 days). (Bottom) Dose-response curves for patients stratified by induced gene expression pattern. (Adapted with permission.57)

Genetic markers

133

Others

Others

Ionizing radiation

Cascade of cytokine activity

ROS (Reactive Oxygen Species)

e.g. TGFβ1 (SMADs)

CTGF family

Fibrosis

Others

Others

WISP2 Superoxide dismutases Metallothioneins

TGFβ1 TNFα

Adlican, Lumican, Fibroblast activating protein

Figure 4 A simplified overview of the radiation-induced fibrosis processes (black) with indications of some of the candidate genes (red) identified through differential gene expression analysis in irradiated fibroblasts. “Others” refers to other genes or pathways that are not necessarily involved in fibrotic processes.

ical points, including better definitions of clinical and biological phenotypes and improved methods for registration and evaluation of confounding factors, that need to be addressed before successful WGA studies on the risk of radiation-induced morbidity can be expected. Future association studies by the candidate gene approach should consider the guidelines by the NCI-NHGRI Working Group on Replication in Association Studies,21 which include reporting haplotype data along with data on individual SNPs. Finally, candidate gene studies elucidating the genetic component of radiationinduced morbidity and the functional consequences of SNPs are highly encouraged. One approach is to study intermediate phenotypes that correlate with the clinical phenotype but are associated more closely with the genetic variants.

References human candidate genes and to study complex traits of radiationinduced morbidity is the use of mouse genetics.60 One example of the intermediate phenotype approach is the studies on irradiated fibroblasts56,57 from breast cancer patients treated with postmastectomy radiotherapy using a 3-field technique.31,32 Briefly, differences in patterns of induced gene expression in fibroblasts after in vitro irradiation (3 ⫻ 3.5 Gy/3 days) correlated with the risk of radiationinduced fibrosis (Fig 3). Based on the fold-induction levels of gene expression between irradiated and nonirradiated cells, patients can be divided into 2 groups (Fig 3, insert), which are characterized by different dose-response curves for the risk of moderate or severe fibrosis with an ED50 enhancement ration of 1.25 (Fig 3, bottom). Many of the differentially expressed genes have functions that are relevant for the fibrotic processes. This is shown in Figure 4, which shows a simplified overview of the processes and genes involved. Marked in red are genes directly identified as being differentially expressed and/or identified through pathway analysis. Further studies on these genes (and the genes that regulate them) have the potential to identify new SNPs for association analysis and may also identify relevant targets for future intervention studies. In a recent review,61 the use of gene expression profiling for elucidating the biological mechanisms of radiation-induced morbidity and risk prediction has been discussed in details.

Conclusions Thus far, a number of studies have reported mainly positive associations between certain genetic variants and the risk of normal tissue toxicity after radiotherapy. Although the studies are not easily compared, it seems very likely that radiation-induced morbidity should be considered as a complex quantitative trait and that naturally occurring genetic variations can account for some part of the observed interpatient variability in normal tissue toxicity. Recent WGA studies have shown feasibility in other research areas and opened the possibility for similar studies on the risk of radiation-induced morbidity. Large biobanks and clinical databases are currently being established, but there are still a number of crit-

1. Andreassen CN, Alsner J, Overgaard J: Does variability in normal tissue reactions after radiotherapy have a genetic basis—Where and how to look for it? Radiother Oncol 64:131-140, 2002 2. Fernet M, Hall J: Genetic biomarkers of therapeutic radiation sensitivity. DNA Repair (Amst) 3:1237-1243, 2004 3. Andreassen CN: Can risk of radiotherapy-induced normal tissue complications be predicted from genetic profiles? Acta Oncol 44:801-815, 2005 4. West CM, Elliott RM, Burnet NG: The genomics revolution and radiotherapy. Clin Oncol (R Coll Radiol) 19:470-480, 2007 5. Lander ES, Linton LM, Birren B, et al: Initial sequencing and analysis of the human genome. Nature 409:860-921, 2001 6. Venter JC, Adams MD, Myers EW, et al: The sequence of the human genome. Science 291:1304-1351, 2001 7. International HapMap Consortium: A haplotype map of the human genome. Nature 437:1299-1320, 2005 8. Liu H, Prugnolle F, Manica A, et al: A geographically explicit genetic model of worldwide human-settlement history. Am J Hum Genet 79: 230-237, 2006 9. Manica A, Amos W, Balloux F, et al: The effect of ancient population bottlenecks on human phenotypic variation. Nature 448:346-348, 2007 10. Hirschhorn JN, Daly MJ: Genome-wide association studies for common diseases and complex traits. Nat Rev Genet 6:95-108, 2005 11. Li J, Zhou Y, Elston RC: Haplotype-based quantitative trait mapping using a clustering algorithm. BMC Bioinformatics 7:258, 2006 12. Satagopan JM, Verbel DA, Venkatraman ES, et al: Two-stage designs for gene-disease association studies. Biometrics 58:163-170, 2002 13. Easton DF, Pooley KA, Dunning AM, et al: Genome-wide association study identifies novel breast cancer susceptibility loci. Nature 447: 1087-1093, 2007 14. Cox A, Dunning AM, Garcia-Closas M, et al: A common coding variant in CASP8 is associated with breast cancer risk. Nat Genet 39:352-358, 2007 15. The Wellcome Trust Case Control Consortium: Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 447:661-678, 2007 16. Parkes M, Barrett JC, Prescott NJ, et al: Sequence variants in the autophagy gene IRGM and multiple other replicating loci contribute to Crohn’s disease susceptibility. Nat Genet 39:830-832, 2007 17. Todd JA, Walker NM, Cooper JD, et al: Robust associations of four new chromosome regions from genome-wide analyses of type 1 diabetes. Nat Genet 39:857-864, 2007 18. Zeggini E, Weedon MN, Lindgren CM, et al: Replication of genomewide association signals in UK samples reveals risk loci for type 2 diabetes. Science 316:1336-1341, 2007 19. Frayling TM, Timpson NJ, Weedon MN, et al: A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science 316:889-894, 2007 20. Saxena R, Voight BF, Lyssenko V, et al: Genome-wide association analysis identifies loci for type 2 diabetes and triglyceride levels. Science 316:1331-1336, 2007

134 21. Chanock SJ, Manolio T, Boehnke M, et al: Replicating genotype-phenotype associations. Nature 447:655-660, 2007 22. Marchini J, Howie B, Myers S, et al: A new multipoint method for genome-wide association studies by imputation of genotypes. Nat Genet 39:906-913, 2007 23. ENCODE Project Consortium.: The ENCODE (ENCyclopedia Of DNA Elements) Project. Science 306:636-640, 2004 24. ENCODE Project Consortium: Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature 447:799-816, 2007 25. Rogers PB, Plowman PN, Harris SJ, et al: Four radiation hypersensitivity cases and their implications for clinical radiotherapy. Radiother Oncol 57:143-154, 2000 26. Bentzen SM, Overgaard M: Relationship between early and late normaltissue injury after postmastectomy radiotherapy. Radiother Oncol 20: 159-165, 1991 27. Bentzen SM, Overgaard M, Overgaard J: Clinical correlations between late normal tissue endpoints after radiotherapy: implications for predictive assays of radiosensitivity. Eur J Cancer 29A:1373-1376, 1993 28. Andreassen CN, Alsner J, Overgaard J, et al: TGFB1 polymorphisms are associated with risk of late normal tissue complications in the breast after radiotherapy for early breast cancer. Radiother Oncol 75:18-21, 2005 29. Andreassen CN, Alsner J, Overgaard M, et al: Prediction of normal tissue radiosensitivity from polymorphisms in candidate genes. Radiother Oncol 69:127-135, 2003 30. Hölscher T, Bentzen SM, Baumann M: Influence of connective tissue diseases on the expression of radiation side effects: A systematic review. Radiother Oncol 78:123-130, 2006 31. Overgaard M, Bentzen SM, Christensen JJ, et al: The value of the NSD formula in equation of acute and late radiation complications in normal tissue following 2 and 5 fractions per week in breast cancer patients treated with postmastectomy irradiation. Radiother Oncol 9:1-11, 1987 32. Bentzen SM, Overgaard M, Thames HD, et al: Early and late normaltissue injury after postmastectomy radiotherapy alone or combined with chemotherapy. Int J Radiat Biol 56:711-715, 1989 33. Andreassen CN, Alsner J, Overgaard M, et al: Risk of radiation-induced subcutaneous fibrosis in relation to single nucleotide polymorphisms in TGFB1. SOD2, XRCC1, XRCC3, APEX and ATM– a study based on DNA from formalin fixed paraffin embedded tissue samples. Int J Radiat Biol 82:577-586, 2006 34. Cesaretti JA, Stock RG, Atencio DP, et al: A genetically determined dose-volume histogram predicts for rectal bleeding among patients treated with prostate brachytherapy. Int J Radiat Oncol Biol Phys 68: 1410-1406, 2007 35. Brem R, Cox DG, Chapot B, et al: The XRCC1-77T-⬎C variant: haplotypes, breast cancer risk, response to radiotherapy and the cellular response to DNA damage. Carcinogenesis 27:2469-2474, 2006 36. Syrris P, Carter ND, Metcalfe JC, et al: Transforming growth factorbeta1 gene polymorphisms and coronary artery disease. Clin Sci (Lond) 95:659-667, 1998 37. Awad MR, El-Gamel A, Hasleton P, et al: Genotypic variation in the transforming growth factor-beta1 gene: Association with transforming growth factor-beta1 production, fibrotic lung disease, and graft fibrosis after lung transplantation. Transplantation 66:1014-1020, 1998 38. Grainger DJ, Heathcote K, Chiano M, et al: Genetic control of the circulating concentration of transforming growth factor type beta1. Hum Mol Genet 8:93-97, 1999 39. Hinke V, Seck T, Clanget C, et al: Association of transforming growth factor-beta1 (TGFbeta1) T29 –⬎ C gene polymorphism with bone mineral density (BMD), changes in BMD, and serum concentrations of TGF-beta1 in a population-based sample of postmenopausal german women. Calcif Tissue Int 69:315-320, 2001 40. Grainger DJ, Metcalfe JC: Tamoxifen: teaching an old drug new tricks? Nat Med 2:381-385, 1996 41. Hoffmann SC, Stanley EM, Cox ED, et al: Ethnicity greatly influences cytokine gene polymorphism distribution. Am J Transplant 2:560-567, 2002

J. Alsner, C.N. Andreassen, and J. Overgaard 42. Bentzen SM, Thames HD, Overgaard M: Latent-time estimation for late cutaneous and subcutaneous radiation reactions in a single-follow-up clinical study. Radiother Oncol 15:267-274, 1989 43. Jung H, Beck-Bornholdt HP, Svoboda V, et al: Quantification of late complications after radiation therapy. Radiother Oncol 61:233-246, 2001 44. Ho AY, Atencio DP, Peters S, et al: Genetic predictors of adverse radiotherapy effects: The Gene-PARE project. Int J Radiat Oncol Biol Phys 65:646-655, 2006 45. Iwakawa M, Imai T, Harada Y, et al: RadGenomics project. Nippon Igaku Hoshasen Gakkai Zasshi 62:484-489, 2002 46. Burnet NG, Elliott RM, Dunning A, et al: Radiosensitivity, radiogenomics and RAPPER. Clin Oncol (R Coll Radiol) 18:525-528, 2006 47. GENEPI Homepage. Available at: www.genepi-estro.org. Accessed December 13, 2007 48. Strauch K, Fimmers R, Baur MP, et al: How to model a complex trait. 1. General considerations and suggestions. Hum Hered 55:202-210, 2003 49. Pan WH, Lynn KS, Chen CH, et al: Using endophenotypes for pathway clusters to map complex disease genes. Genet Epidemiol 30:143-154, 2006 50. Johansen J, Bentzen SM, Overgaard J, et al: Relationship between the in vitro radiosensitivity of skin fibroblasts and the expression of subcutaneous fibrosis, telangiectasia, and skin erythema after radiotherapy. Radiother Oncol 40:101-109, 1996 51. West CM, Davidson SE, Elyan SA, et al: Lymphocyte radiosensitivity is a significant prognostic factor for morbidity in carcinoma of the cervix. Int J Radiat Oncol Biol Phys 51:10-15, 2001 52. Borgmann K, Roper B, El-Awady R, et al: Indicators of late normal tissue response after radiotherapy for head and neck cancer: Fibroblasts, lymphocytes, genetics. DNA repair, and chromosome aberrations. Radiother Oncol 64:141-152, 2002 53. Hoeller U, Borgmann K, Bonacker M, et al: Individual radiosensitivity measured with lymphocytes may be used to predict the risk of fibrosis after radiotherapy for breast cancer. Radiother Oncol 69:137-144, 2003 54. Rieger KE, Hong WJ, Tusher VG, et al: Toxicity from radiation therapy associated with abnormal transcriptional responses to DNA damage. Proc Natl Acad Sci U S A 101:6635-6640, 2004 55. Svensson JP, Stalpers LJ, Esveldt-van Lange RE, et al: Analysis of gene expression using gene sets discriminates cancer patients with and without late radiation toxicity. PLoS Med 3:e422, 2006 56. Rødningen OK, Børresen-Dale AL, Alsner J, et al: Radiation-induced gene expression in human subcutaneous fibroblasts is predictive of radiation-induced fibrosis. Radiother Oncol (in press) 57. Alsner J, Rødningen OK, Overgaard J: Differential gene expression before and after ionizing radiation of subcutaneous fibroblasts identifies breast cancer patients resistant to radiation-induced fibrosis. Radiother Oncol 83:261-266, 2007 58. Herskind C, Bentzen SM, Overgaard J, et al: Differentiation state of skin fibroblast cultures versus risk of subcutaneous fibrosis after radiotherapy. Radiother Oncol 47:263-269, 1998 59. Herskind C, Johansen J, Bentzen SM, et al: Fibroblast differentiation in subcutaneous fibrosis after postmastectomy radiotherapy. Acta Oncol 39:383-388, 2000 60. Travis EL: Genetic susceptibility to late normal tissue injury. Semin Radiat Oncol 17:149-155, 2007 61. Kruse JJ, Stewart FA: Gene expression arrays as a tool to unravel mechanisms of normal tissue radiation injury and prediction of response. World J Gastroenterol 13:2669-2674, 2007 62. Ahn J, Ambrosone CB, Kanetsky PA, et al: Polymorphisms in genes related to oxidative stress (CAT. MnSOD, MPO, and eNOS) and acute toxicities from radiation therapy following lumpectomy for breast cancer. Clin Cancer Res 12:7063-7070, 2006 63. Ambrosone CB, Tian C, Ahn J, et al: Genetic predictors of acute toxicities related to radiation therapy following lumpectomy for breast cancer: A case-series study. Breast Cancer Res 8:R40, 2006 64. Andreassen CN, Overgaard J, Alsner J, et al: ATM sequence variants and risk of radiation-induced subcutaneous fibrosis after postmastectomy radiotherapy. Int J Radiat Oncol Biol Phys 64:776-783, 2006

Genetic markers 65. Angele S, Romestaing P, Moullan N, et al: ATM haplotypes and cellular response to DNA damage: association with breast cancer risk and clinical radiosensitivity. Cancer Res 63:8717-8725, 2003 66. Appleby JM, Barber JB, Levine E, et al: Absence of mutations in the ATM gene in breast cancer patients with severe responses to radiotherapy. Br J Cancer 76:1546-1549, 1997 67. Bremer M, Klopper K, Yamini P, et al: Clinical radiosensitivity in breast cancer patients carrying pathogenic ATM gene mutations: No observation of increased radiation-induced acute or late effects. Radiother Oncol 69:155-160, 2003 68. Cesaretti JA, Stock RG, Lehrer S, et al: ATM sequence variants are predictive of adverse radiotherapy response among patients treated for prostate cancer. Int J Radiat Oncol Biol Phys 61:196-202, 2005 69. Chang-Claude J, Popanda O, Tan XL, et al: Association between polymorphisms in the DNA repair genes. XRCC1, APE1, and XPD and acute side effects of radiotherapy in breast cancer patients. Clin Cancer Res 11:4802-4809, 2005 70. Clarke RA, Goozee GR, Birrell GW, et al: Absence of ATM truncations in patients with severe acute radiation reactions. Int J Radiat Oncol Biol Phys 41:1021-1027, 1998 71. Damaraju S, Murray D, Dufour J, et al: Association of DNA repair and steroid metabolism gene polymorphisms with clinical late toxicity in patients treated with conformal radiotherapy for prostate cancer. Clin Cancer Res 12:2545-2554, 2006 72. De Ruyck K, Van Eijkeren M, Claes K, et al: Radiation-induced damage to normal tissues after radiotherapy in patients treated for gynecologic tumors: Association with single nucleotide polymorphisms in XRCC1. XRCC3, and OGG1 genes and in vitro chromosomal radiosensitivity in lymphocytes. Int J Radiat Oncol Biol Phys 62:1140-1149, 2005 73. De Ruyck K, Wilding CS, Van Eijkeren M, et al: Microsatellite polymorphisms in DNA repair genes XRCC1. XRCC3 and XRCC5 in patients with gynecological tumors: Association with late clinical radiosensitivity and cancer incidence. Radiat Res 164:237-244, 2005 74. De Ruyck K, Van Eijkeren M, Claes K, et al: TGFbeta1 polymorphisms and late clinical radiosensitivity in patients treated for gynecologic tumors. Int J Radiat Oncol Biol Phys 65:1240-1248, 2006 75. Edvardsen H, Kristensen VN, Grenaker Alnaes GI, et al: Germline glutathione S-transferase variants in breast cancer: Relation to diagnosis and cutaneous long-term adverse effects after two fractionation patterns of radiotherapy. Int J Radiat Oncol Biol Phys 67:1163-1171, 2007 76. Gaffney DK, Brohet RM, Lewis CM, et al: Response to radiation therapy and prognosis in breast cancer patients with BRCA1 and BRCA2 mutations. Radiother Oncol 47:129-136, 1998 77. Giotopoulos G, Symonds RP, Foweraker K, et al: The late radiotherapy normal tissue injury phenotypes of telangiectasia, fibrosis and atrophy in breast cancer patients have distinct genotype-dependent causes. Br J Cancer 96:1001-1007, 2007 78. Green H, Ross G, Peacock J, et al: Variation in the manganese superoxide dismutase gene (SOD2) is not a major cause of radiotherapy complications in breast cancer patients. Radiother Oncol 63:213-216, 2002

135 79. Hall EJ, Schiff PB, Hanks GE, et al: A preliminary report: Frequency of A-T heterozygotes among prostate cancer patients with severe late responses to radiation therapy. Cancer J Sci Am 4:385-389, 1998 80. Ho AY, Fan G, Atencio DP, et al: Possession of ATM sequence variants as predictor for late normal tissue responses in breast cancer patients treated with radiotherapy. Int J Radiat Oncol Biol Phys 69:677-684, 2007 81. Iannuzzi CM, Atencio DP, Green S, et al: ATM mutations in female breast cancer patients predict for an increase in radiation-induced late effects. Int J Radiat Oncol Biol Phys 52:606-613, 2002 82. Kornguth DG, Garden AS, Zheng Y, et al: Gastrostomy in oropharyngeal cancer patients with ERCC4 (XPF) germline variants. Int J Radiat Oncol Biol Phys 62:665-671, 2005 83. Leong T, Whitty J, Keilar M, et al: Mutation analysis of BRCA1 and BRCA2 cancer predisposition genes in radiation hypersensitive cancer patients. Int J Radiat Oncol Biol Phys 48:959-965, 2000 84. Moullan N, Cox DG, Angele S, et al: Polymorphisms in the DNA repair gene XRCC1, breast cancer risk, and response to radiotherapy. Cancer Epidemiol Biomarkers Prev 12:1168-1174, 2003 85. Oppitz U, Bernthaler U, Schindler D, et al: Sequence analysis of the ATM gene in 20 patients with RTOG grade 3 or 4 acute and/or late tissue radiation side effects. Int J Radiat Oncol Biol Phys 44:981-988, 1999 86. Pierce LJ, Strawderman M, Narod SA, et al: Effect of radiotherapy after breast-conserving treatment in women with breast cancer and germline BRCA1/2 mutations. J Clin Oncol 18:3360-3369, 2000 87. Popanda O, Tan XL, Ambrosone CB, et al: 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. Cancer Epidemiol Biomarkers Prev 15:1048-1050, 2006 88. Price EA, Bourne SL, Radbourne R, et al: Rare microsatellite polymorphisms in the DNA repair genes XRCC1. XRCC3 and XRCC5 associated with cancer in patients of varying radiosensitivity. Somat Cell Mol Genet 23:237-247, 1997 89. Quarmby S, Fakhoury H, Levine E, et al: Association of transforming growth factor beta-1 single nucleotide polymorphisms with radiationinduced damage to normal tissues in breast cancer patients. Int J Radiat Biol 79:137-143, 2003 90. Ramsay J, Birrell G, Lavin M: Testing for mutations of the ataxia telangiectasia gene in radiosensitive breast cancer patients. Radiother Oncol 47:125-128, 1998 91. Severin DM, Leong T, Cassidy B, et al: Novel DNA sequence variants in the hHR21 DNA repair gene in radiosensitive cancer patients. Int J Radiat Oncol Biol Phys 50:1323-1331, 2001 92. Shayeghi M, Seal S, Regan J, et al: Heterozygosity for mutations in the ataxia telangiectasia gene is not a major cause of radiotherapy complications in breast cancer patients. Br J Cancer 78:922-927, 1998 93. Tan XL, Popanda O, Ambrosone CB, et al: Association between TP53 and p21 genetic polymorphisms and acute side effects of radiotherapy in breast cancer patients. Breast Cancer Res Treat 97:255-262, 2006 94. Weissberg JB, Huang DD, Swift M: Radiosensitivity of normal tissues in ataxia-telangiectasia heterozygotes. Int J Radiat Oncol Biol Phys 42: 1133-1136, 1998