Objective: To explore the possible relationship between single nucleotide polymorphisms (SNP) in candidate genes encoding DNA damage recognition/repair/response and steroid metabolism proteins with respect to clinical radiation toxicity in a retrospective cohort of patients previously treated with three-dimensional conformal radiotherapy (3-DCRT) for prostate cancer.

Experimental Design: One hundred twenty-four patients with prostate cancer underwent 3-DCRT at our institution between September 1996 and December 2000. Of these, 83 consented for follow-up of blood sampling and SNP analysis. Twenty-eight patients were documented as having experienced grade ≥2 late bladder or rectal toxicity (scoring system of Radiation Therapy Oncology Group) on at least one follow-up visit. We analyzed 49 SNPs in BRCA1, BRCA2, ESR1, XRCC1, XRCC2, XRCC3, NBN, RAD51, RAD52, LIG4, ATM, BCL2, TGFB1, MSH6, ERCC2, XPF, NR3C1, CYP1A1, CYP2C9, CYP2C19, CYP3A5, CYP2D6, CYP11B2, and CYP17A1 genes using the Pyrosequencing technique.

Results: Significant univariate associations with late rectal or bladder toxicity (grade ≥2) were found for XRCC3 (A>G 5′ untranslated region NT 4541), LIG4 (T>C Asp568Asp), MLH1 (C>T, Val219Ile), CYP2D6*4 (G>A splicing defect), mean rectal and bladder dose, dose to 30% of rectum or bladder, and age <60 years. On Cox multivariate analysis, significant associations with toxicity were found for LIG4 (T>C, Asp568Asp), ERCC2 (G>A, Asp711Asp), CYP2D6*4 (G>A, splicing defect), mean bladder dose >60 Gy, and dose to 30% of rectal volume >75 Gy.

Conclusions: In this study, we identified SNPs in LIG4, ERCC2, and CYP2D6 genes as putative markers to predict individuals at risk for complications arising from radiation therapy in prostate cancer.

External beam radiotherapy is an efficacious and widely used treatment modality for localized prostate cancer. Using image-guided techniques, such as three-dimensional conformal radiotherapy (3-DCRT) and intensity-modulated radiotherapy, dose escalation (13) has been suggested as a means to improve prostate cancer control while incurring acceptable rates of toxicity. A randomized trial of dose-escalated 3-DCRT (2) reported improved biochemical freedom from relapse after 78 versus 70 Gy for localized prostate cancer. However, the risk of radiation-related late rectal complications was 26% (78 Gy) versus 12% (70 Gy; P = 0.001), suggesting a need for caution in the routine application of dose-escalated 3-DCRT for prostate cancer.

Estimates of the risk of late bladder and rectal toxicity after 3-DCRT vary from 7% to 38%, depending upon the time interval reported (2, 4, 5). The best strategy to identify patients who are potentially at risk for late radiotherapy toxicity before undergoing 3-DCRT for prostate cancer is currently unknown. One phenomenological approach has been the development of biophysical models based upon observations of organ-at-risk tolerance to gradients of radiation dose delivered by 3-DCRT to predict which patients may display significant radiation toxicity (6, 7). This approach is limited by uncertainties in the actual delivery of the physical radiation dose and by nonstochastic factors, the most important of which is the individual patient's normal tissue sensitivity to ionizing radiation.

Despite the advances in radiotherapy delivery and modalities to reduce normal tissue irradiation, the inherent biological complexity among individuals due to variations in their genome has been a limiting factor in predicting tissue radiosensitivity. Cellular radiosensitivity assays using fibroblast and lymphoblastoid cell lines have had limited, if any, effect in predicting the in vivo response of individuals to ionizing radiation (8) for reasons that are not entirely clear. Assay precision or in vitro selection pressures may play a role, as well as differences in the way that cells respond to radiation in the complex normal tissue microenvironment. The observation that interindividual differences exist in the occurrence of normal tissue complications after radiotherapy has led past authors to explore the hypothesis that heritable, genetic factors (at least in part) influence the development of toxicity (913). Opinion is evolving from the supposition that this phenomenon is monogenic in nature (involving single but uncommon predisposing genes inherited in a Mendelian fashion) to a polygenic concept, characterized by a complex inheritance of risk influenced by multiple genomic loci (1416).

Ionizing radiation produces its biological effects mainly through the generation of short-lived but highly reactive DNA radicals that evolve into stable/long-lived DNA lesions, such as double-strand breaks (DSB; ref. 17) or through interactions with the plasma membrane (18), leading to cell death. Elucidation of the molecular effectors of the cellular response to such damage is currently of great interest in the radiation research community. Cellular processes in which the encoded factors are involved include DNA damage recognition, DNA repair, and recombination pathways, including nonhomologous end-joining of DSBs, homologous recombination repair of DSBs, and base excision repair of other types of single and simple clustered lesions, as well as proteins involved in the activation of cell cycle checkpoints, signal transduction, and stress responses, and apoptotic or premature senescence pathways.

The human genome project has helped to identify at least 130 DNA repair genes (19). In a recent single nucleotide polymorphism (SNP) discovery study, Mohrenweiser et al. described 133 variants from 37 DNA repair genes in groups of 36 to 164 unrelated apparently healthy individuals (20). In the present study, we have tested the hypothesis that there exists a relationship between SNPs in genes encoding DNA damage recognition/response proteins and tissue homeostasis factors with respect to clinical radiation toxicity in a cohort of patients who have previously undergone 3-DCRT for prostate cancer. Organ-specific exposure and the ensuing damage to exposed normal tissue upon radiation treatment show distinct phenotypes due to cell-specific characteristics of the exposed tissue. The use of gene polymorphisms to predict radiation-induced injury to normal tissues in prostate cancer patients undergoing radiotherapy has not been previously reported. In this study, we analyzed 49 SNPs in 12 genes from DNA repair pathways and 12 genes from cell signaling, apoptosis, tissue modeling, and steroid metabolism pathways for their association to radiation-induced toxicity. Ultimately, we seek to discover and validate predictive methods incorporating data from polymorphisms in genes for DNA repair and tissue repair to predict which individuals may be at risk for complications arising from radiation therapy.

Selection of cohort. One hundred twenty-four patients predominantly with histologically confirmed adenocarcinoma of the prostate were identified who had received external beam 3-DCRT between September 1996 and December 2000. All patients had clinically localized prostate cancer (T classification, T1c-T3b) with a negative staging bone scan and no evidence of pelvic lymphadenopathy on computed tomography scan. From this cohort, 71 Caucasian, six Caucasian-French Canadian, three Aboriginal, two Pacific Asian, and one Afro-Caribbean patients with adequate follow-up and documentation of clinical late toxicity were identified (Table 1). The number of patients with >18 months follow-up was 81. Informed consent was obtained before blood sampling, and the research protocol had received approval by the Research Ethics Committee of the Alberta Cancer Board. Twenty-eight patients were documented as having experienced Radiation Therapy Oncology Group grade ≥2 late bladder or rectal toxicity on at least one follow-up visit. We analyzed 49 SNPs in ATM, BRCA1, BRCA2, XRCC1, XRCC2, XRCC3, NBN, RAD51, RAD52, LIG4, BCL2, TGFB1, MLH1, MSH6, ERCC2, XPF, and the CYP family of genes (see Table 2 for abbreviations) using the Pyrosequencing technique. All SNPs were first validated on DNA samples from apparently healthy anonymous male controls from the same geographic region.

Table 1.

Patient clinical characteristics

Mean (range)
Age at diagnosis (y) 67 (45-78) 
PSA level at diagnosis (ng/mL) 11.8 (0.2-74.0) 
T classification  
    T1 18 
    T2 44 
    T3 20 
    Tx 
History of diabetes  
    Yes 
    No 78 
Previous abdominopelvic surgery  
    Yes 30 
    No 53 
History of hypertension  
    Yes 32 
    No 51 
Neoadjuvant hormone therapy  
    Yes 35 
    No 48 
Adjuvant hormone therapy  
    Yes 
    No 77 
Mean rectal dose (Gy) 55.4 (29.3-72.2) 
Dose to 30% of rectum (Gy) 66.4 (43.6-78.1) 
Mean bladder dose (Gy) 50.2 (25.8-74.1) 
Mean (range)
Age at diagnosis (y) 67 (45-78) 
PSA level at diagnosis (ng/mL) 11.8 (0.2-74.0) 
T classification  
    T1 18 
    T2 44 
    T3 20 
    Tx 
History of diabetes  
    Yes 
    No 78 
Previous abdominopelvic surgery  
    Yes 30 
    No 53 
History of hypertension  
    Yes 32 
    No 51 
Neoadjuvant hormone therapy  
    Yes 35 
    No 48 
Adjuvant hormone therapy  
    Yes 
    No 77 
Mean rectal dose (Gy) 55.4 (29.3-72.2) 
Dose to 30% of rectum (Gy) 66.4 (43.6-78.1) 
Mean bladder dose (Gy) 50.2 (25.8-74.1) 

Abbreviation: PSA, prostate-specific antigen.

Table 2.

SNP analysis from select candidate genes

No.GenedbSNP IDChangeObserved genotype
Sample sizeAllele frequency
AAAaaaAllele 1Allele 2
ATM rs1801516 A>G, Asp1853Asn 59 20 84 0.82 0.18 
ATM rs1801673 A>G, Asp1853lle 80 84 0.98 0.02 
BCL2 rs1801018 G>A, Thr7Thr 21 48 15 84 0.54 0.46 
BRCA1 rs16940 A>G, Leu771Leu 38 38 84 0.68 0.32 
BRCA1 rs1799967 G>A, lle1652Met 78 83 0.97 0.03 
BRCA1 rs1799949 C>T, Ser694Ser 37 38 83 0.67 0.33 
BRCA1 rs1060915 T>C, Ser1436Ser 38 37 83 0.68 0.32 
BRCA1 rs1799966 A>G, Gly1613Ser 37 38 83 0.67 0.33 
BRCA1 rs16941 A>G, Gly742Glu 37 37 82 0.68 0.32 
10 BRCA1 rs16942 A>G, Arg1183Lys 37 38 83 0.67 0.33 
11 BRCA1 rs1799950 A>G, Arg356Gln 73 83 0.93 0.07 
12 BRCA1 rs12516 C>T, 3′ UTR NT 79705 37 38 83 0.67 0.33 
13 BRCA2 rs1799955 A>G, Ser2414Ser 45 33 83 0.74 0.26 
14 BRCA2 rs1799944 A>G, Asp991Asn 76 83 0.95 0.05 
15 BRCA2 rs1801406 A>G, Lys1132Lys 33 36 14 83 0.61 0.39 
16 BRCA2 rs1801499 T>C, His743His 77 83 0.96 0.04 
17 BRCA2 rs1799943 G>A, 5′ UTR NT-26 34 38 11 83 0.64 0.36 
18 BRCA2 rs1801439 A>G, Ser455Ser 76 83 0.95 0.05 
19 BRCA2 rs766173 C>A, Asn289His 76 83 0.95 0.05 
20 CY3A5*3 rs776746 G>A, Intron/splice variant 75 84 0.95 0.05 
21 CYP11B2 rs4541 T>C, Val386Ala 68 14 84 0.89 0.11 
22 CYP17A1 rs6163 T>G, Ser65Ser 35 37 12 84 0.64 0.36 
23 CYP1A1 rs1048943 A>G, lle462Val 79 84 0.97 0.03 
24 CYP1A1 rs1799814 A>C, Asn461Thr 83 84 0.99 0.01 
25 CYP2C19*3 NA G>A, Stop codon 84 84 1.00 — 
26 CYP2C9*2 rs1799853 C>T, Cys144Arg 79 84 0.97 0.03 
27 CYP2D6*4 rs1800716 G>A, Splicing defect 60 21 83 0.85 0.15 
28 CYP2D6*6 rs5030655 T>del, Frameshift 83 83 1.00 — 
29 CYP2D6*8*14 rs5030865 G>T, Stop codon 83 83 1.00 — 
30 DNA Lig4 rs1805386 T>C, Asp568Asp 61 21 84 0.85 0.15 
31 ERCC2 rs1052555 G>A, Asp711Asp 35 40 84 0.65 0.35 
32 ERCC2 rs1052559 T>G, Gln751Lys 30 44 10 84 0.62 0.38 
33 XPF rs1799801 C>T, Ser835Ser 43 38 84 0.74 0.26 
34 ESR1 rs1801132 C>G, Pro325Pro 48 32 84 0.76 0.24 
35 NR3C1 rs6195 A>G, Ser363Asn 77 84 0.96 0.04 
36 MLH1 rs1799977 C>T, Val219lle 47 28 84 0.73 0.27 
37 MSH6 rs2020910 A>T, Thr1102Thr 83 84 0.99 0.01 
38 MSH6 rs1800935 T>C, Asp180Asp 45 36 84 0.75 0.25 
39 NBN rs1805794 G>C, Glu185Gln 31 39 14 84 0.60 0.40 
40 RAD51 rs1801320 G>C, 5′ UTR, NT 135 69 14 84 0.90 0.10 
41 RAD51 rs1801321 G>T 5′ UTR, NT 172 28 45 11 84 0.60 0.40 
42 RAD52 rs11226 C>T, 3′ UTR, NT 2259 24 39 21 84 0.52 0.48 
43 TGFB1 rs1800471 G>C Arg25Pro 71 13 84 0.92 0.08 
44 TGFB1 rs1800469 C>T, 5′ UTR NT 1347 45 32 84 0.73 0.27 
45 XRCC1 rs1799782 C>T, Arg194Trp 70 13 84 0.91 0.09 
46 XRCC1 rs25489 A>G, Arg280Hls 78 84 0.96 0.04 
47 XRCC2 rs3218536 G>A, Arg188His 70 13 84 0.91 0.09 
48 XRCC3 rs1799794 A>G, 5′ UTR NT 4541 52 29 84 0.79 0.21 
49 XRCC3 rs1799796 A>G, IVS5-14, NT 17893 40 39 84 0.71 0.29 
No.GenedbSNP IDChangeObserved genotype
Sample sizeAllele frequency
AAAaaaAllele 1Allele 2
ATM rs1801516 A>G, Asp1853Asn 59 20 84 0.82 0.18 
ATM rs1801673 A>G, Asp1853lle 80 84 0.98 0.02 
BCL2 rs1801018 G>A, Thr7Thr 21 48 15 84 0.54 0.46 
BRCA1 rs16940 A>G, Leu771Leu 38 38 84 0.68 0.32 
BRCA1 rs1799967 G>A, lle1652Met 78 83 0.97 0.03 
BRCA1 rs1799949 C>T, Ser694Ser 37 38 83 0.67 0.33 
BRCA1 rs1060915 T>C, Ser1436Ser 38 37 83 0.68 0.32 
BRCA1 rs1799966 A>G, Gly1613Ser 37 38 83 0.67 0.33 
BRCA1 rs16941 A>G, Gly742Glu 37 37 82 0.68 0.32 
10 BRCA1 rs16942 A>G, Arg1183Lys 37 38 83 0.67 0.33 
11 BRCA1 rs1799950 A>G, Arg356Gln 73 83 0.93 0.07 
12 BRCA1 rs12516 C>T, 3′ UTR NT 79705 37 38 83 0.67 0.33 
13 BRCA2 rs1799955 A>G, Ser2414Ser 45 33 83 0.74 0.26 
14 BRCA2 rs1799944 A>G, Asp991Asn 76 83 0.95 0.05 
15 BRCA2 rs1801406 A>G, Lys1132Lys 33 36 14 83 0.61 0.39 
16 BRCA2 rs1801499 T>C, His743His 77 83 0.96 0.04 
17 BRCA2 rs1799943 G>A, 5′ UTR NT-26 34 38 11 83 0.64 0.36 
18 BRCA2 rs1801439 A>G, Ser455Ser 76 83 0.95 0.05 
19 BRCA2 rs766173 C>A, Asn289His 76 83 0.95 0.05 
20 CY3A5*3 rs776746 G>A, Intron/splice variant 75 84 0.95 0.05 
21 CYP11B2 rs4541 T>C, Val386Ala 68 14 84 0.89 0.11 
22 CYP17A1 rs6163 T>G, Ser65Ser 35 37 12 84 0.64 0.36 
23 CYP1A1 rs1048943 A>G, lle462Val 79 84 0.97 0.03 
24 CYP1A1 rs1799814 A>C, Asn461Thr 83 84 0.99 0.01 
25 CYP2C19*3 NA G>A, Stop codon 84 84 1.00 — 
26 CYP2C9*2 rs1799853 C>T, Cys144Arg 79 84 0.97 0.03 
27 CYP2D6*4 rs1800716 G>A, Splicing defect 60 21 83 0.85 0.15 
28 CYP2D6*6 rs5030655 T>del, Frameshift 83 83 1.00 — 
29 CYP2D6*8*14 rs5030865 G>T, Stop codon 83 83 1.00 — 
30 DNA Lig4 rs1805386 T>C, Asp568Asp 61 21 84 0.85 0.15 
31 ERCC2 rs1052555 G>A, Asp711Asp 35 40 84 0.65 0.35 
32 ERCC2 rs1052559 T>G, Gln751Lys 30 44 10 84 0.62 0.38 
33 XPF rs1799801 C>T, Ser835Ser 43 38 84 0.74 0.26 
34 ESR1 rs1801132 C>G, Pro325Pro 48 32 84 0.76 0.24 
35 NR3C1 rs6195 A>G, Ser363Asn 77 84 0.96 0.04 
36 MLH1 rs1799977 C>T, Val219lle 47 28 84 0.73 0.27 
37 MSH6 rs2020910 A>T, Thr1102Thr 83 84 0.99 0.01 
38 MSH6 rs1800935 T>C, Asp180Asp 45 36 84 0.75 0.25 
39 NBN rs1805794 G>C, Glu185Gln 31 39 14 84 0.60 0.40 
40 RAD51 rs1801320 G>C, 5′ UTR, NT 135 69 14 84 0.90 0.10 
41 RAD51 rs1801321 G>T 5′ UTR, NT 172 28 45 11 84 0.60 0.40 
42 RAD52 rs11226 C>T, 3′ UTR, NT 2259 24 39 21 84 0.52 0.48 
43 TGFB1 rs1800471 G>C Arg25Pro 71 13 84 0.92 0.08 
44 TGFB1 rs1800469 C>T, 5′ UTR NT 1347 45 32 84 0.73 0.27 
45 XRCC1 rs1799782 C>T, Arg194Trp 70 13 84 0.91 0.09 
46 XRCC1 rs25489 A>G, Arg280Hls 78 84 0.96 0.04 
47 XRCC2 rs3218536 G>A, Arg188His 70 13 84 0.91 0.09 
48 XRCC3 rs1799794 A>G, 5′ UTR NT 4541 52 29 84 0.79 0.21 
49 XRCC3 rs1799796 A>G, IVS5-14, NT 17893 40 39 84 0.71 0.29 

NOTE:SNP analysis from candidate genes for possible association with radiation-induced toxicity in prostate cancer cases. SNPs from 24 genes were selected based on the criteria described in Materials and Methods. SNPs in genes, the National Center for Biotechnology Information SNP database (http://www.ncbi.nlm.nih.gov/SNP/) unique identifiers (rs numbers) and the changes in nucleotides are summarized. Altered nucleotides in the coding, intervening sequences, promoter, or intronic splice site regions are identified with nucleotide numbering based on the position of codons in protein sequence or contiguous nucleotide sequence number in the database (http://snpper.chip.org). CYP gene alleles are identified (*). The genotype scores observed for each SNP are indicated as AA (wild type, homozygous), Aa (heterozygous), and aa (homozygous variant). Sample size represents total number of cases genotyped. The calculated allele frequencies for major and minor alleles (1 and 2) are shown.

Abbreviations: ATM, ataxia telangiectasia mutated; BCL2, B-cell lymphoma protein 2 α isoform; BRCA 1/2, breast cancer 1/2, early onset; CYP, cytochrome P450; LIG4, ATP-dependent DNA ligase 4; ESR1, estrogen receptor; NR3C1, glucocorticoid receptor or nuclear receptor subfamily 3, group C, member 1; MLH1, Homo sapiens mutL homologue 1; MSH6, Homo sapiens mutS homologue 6; NBN, Nijmegen breakage syndrome 1; RAD, Rad homologue isoforms; TGFB1, transforming growth factor β 1; IVS, intervening sequences; NA, not applicable; NT, nucleotide; UTR, untranslated region.

Treatment. All patients underwent a pelvic computed tomography scan with contrast urethrography. Custom-expanded foam immobilization casts were used for the majority of patients; however, implanted prostatic fiducial markers were not employed at the time of this study. The clinical target volume (prostatic tissue containing biopsy-proven adenocarcinoma or its suspected microscopic extensions) was typically considered to be the entire prostate gland ± bilateral seminal vesicles. Axial clinical target volume images underwent segmentation with digital delineation of the clinical target volume, planning target volume (a 10-mm tangential expansion of the clinical target volume), rectum from anorectal junction to rectosigmoid junction (as per the guidelines of the Radiation Therapy Oncology Group 1994-2006 clinical trial; ref. 3), and entire bladder using the Helax TMS treatment planning system versions 5 to 6 (Nucletron Canada, Kanata, Ontario, Canada). The organs at risk (rectum, bladder, and proximal femoral bones) were delineated as “solid” organs (21, 22). 3-DCRT was planned using beam's eye view techniques with either custom cerrobend or multileaf collimator shaped fields using anterior, posterior, right, and left lateral 15 to 18 MV X-rays (Varian Medical Systems, Corona, CA). A few patients received treatment with right and left anterior oblique fields in addition to the four fields described above. The planning target volume received a mean dose of 77.1 Gy (range, 68.3-82.1 Gy) in five daily fractions per week. The number of fractions ranged from 35 to 44, and in all patients, the planning target volume received a minimum dose per fraction of 1.8 to 2 Gy. Pelvic lymph nodes were not treated by intention in any patient. Doses to “partial” volumes of the organs at risk and planning target volume were recorded from the dose-volume histograms at each “decile” of fractional volume (i.e., 10%, 20%, 30%, etc.). At the discretion of the treating radiation oncologist, short-course (2-4 months pre-3-DCRT) neoadjuvant hormonal therapy (leuprolide depot, goserelin acetate, or buserelin ± flutamide) was given to 35 patients (42%). Adjuvant post-radiation hormonal therapy (typically for 2 years after 3-DCRT) was given to six patients (7%). Clinical follow-up involved obtaining clinical history, physical examination, measurement of serum prostate-specific antigen, and documentation of clinical toxicity every 3 months for the first 2 years, every 4 months for the third year, and at 6-month intervals for the fourth and fifth years, and yearly thereafter. The Radiation Therapy Oncology Group late effects scale was used by two observers to grade toxicity retrospectively in an independent fashion, with scores assigned based upon written comments by the clinicians in the charts. Discrepancies were adjudicated by one of us (M.P.) to yield the final toxicity score for that time point.

Choice of SNPs. Candidate SNPs selected for the present study (Table 2) were based on the reported association of several genes identified in repair of DNA damage upon exposure of cells to ionizing radiation and other environmental agents (19, 23, 24). Several of these SNPs have also been associated with cancer predisposition in case-control studies (10, 14, 2528). In addition, we included SNPs from genes encoding integral membrane proteins, such as estrogen receptor α (ESR1) and glucocorticoid receptor (NR3C1), involved in cell signaling. Polymorphisms for xenobiotic and steroid metabolism genes from the cytochrome P450 (CYP) gene family were evaluated for possible association because many of these SNPs have been implicated in predisposition to various cancers (Genetic Association Database: http://geneticassociationdb.nih.gov/) and hence could also serve as surrogate markers in the identification of novel/causative genes in radiation-induced toxicity. A balance of anti-inflammatory actions of steroids and proinflammatory and anti-inflammatory cytokines produced in vivo in response to radiation damage of the tissue may determine the rate of tissue repair/wound-healing processes, and a number of CYP gene family members are known to mediate these processes (29). In the current study, we selected CYP gene SNPs whose gene expression levels have been documented in relevant tissues, such as prostate, bladder, colon, and small intestine (3035). CYP genes, such as CYP1A1, CYP1B1, CYP3A4, CYP3A5, CYP17A1, ESR1, and CYP19A1, are also involved in steroid hormone metabolism and response (36).

Many DNA repair proteins form complexes in vivo to repair damaged DNA (28). Therefore, it is reasonable to speculate that SNPs in multiple DNA repair genes might not only contribute to potential alteration in the protein structure and function but might also modulate protein-protein interactions among members of the DNA repair family of enzymes. We assessed synonymous and nonsynonymous SNPs in protein coding regions and SNPs in regulatory regions (5′ or 3′ regions) for their possible influence on regulating gene expression and for indications of linkage disequilibrium with disease causing genes.

Genotyping technology. Genotyping using Pyrosequencing is based on the principle of sequencing by synthesis, in which the complimentary DNA strand synthesis is initiated following annealing of primer a few bases away from the putative SNP site using standard DNA strand synthesis reagents with Klenow polymerase enzyme, except that the enzyme cascade system also uses luciferase, ATP sulfurylase, and apyrase. The target sequence containing the polymorphic site is amplified using standard PCR conditions in which one of the primers is biotinylated. A single-stranded template is generated by stripping off the nonbiotinylated strand on streptavidin affinity matrix in alkali. The bound DNA template on the affinity matrix is directly used to synthesize a short stretch of complimentary strand (10-15 bases) flanking the SNP site. Quantitative nucleotide incorporation (the pyrophosphate released at every base extension is converted to ATP by the above enzyme cascade system) is measured by coupling ATP synthesis to the oxidation of luciferin, resulting in the generation of light with an intensity proportional to the number of nucleotides incorporated that can be visualized in real-time. Pyrosequencing reagents were purchased and used as recommended by the manufacturer (http://www.pyrosequencing.com or http://www.biotagebio.com). All other reagents were of molecular biology grade obtained from commercial sources.

Blood collection and DNA preparation. Venous blood samples were collected into EDTA Vacutainer vials and separated buffy coat samples were snap frozen in liquid nitrogen and stored at -80°C until use. Genomic DNA was isolated from buffy coat cells using the QiaAmp kit (Qiagen, Mississauga, ON, Canada). The oligonucleotide primers for PCR were designed using Primer3 software (Whitehead Institute for Biomedical Research, Cambridge, MA), and the sequencing primers were designed using the software from Pyrosequencing, Inc. (Westborough, MA). Primers were custom synthesized at the Operon Biotechnologies Facility. Genotype data were tested for deviations from Hardy-Weinberg equilibrium criteria.

Statistical considerations. Genotype calls for individual SNPs (homozygous wild type, heterozygous, and homozygous mutant or variant) were grouped dichotomously into genotype classes to maximize the differences in time to toxicity. Cumulative risk of late toxicity curves were derived according to genotype class for each SNP using the Kaplan-Meier method, and log-rank tests were done to obtain univariate significance of the differences between the genotype classes (37). The dichotomized SNPs with significance level of <0.15 for between-genotype class differences were chosen for entry into the multivariate analysis. This analysis also included radiation dosimetric variables and age covariates that were treated as binary variables for entry into the multivariate analysis based upon previously published data (38). Exploratory multivariate analysis was done using the SAS score method (SAS v9.1, SAS Institute, Inc., Cary, NC) to determine the best overall model fit. Because of the small sample size (n = 83), we applied the Cox proportional hazards regression procedure (selection = score option) to select the best subset of models from all possible variables. This approach is believed to reduce the instability of the estimates produced. Thus, we obtained two best models for each subset of four and five variables providing a plausible number of 20 and 16 cases, respectively, for each covariate in the model.

Patient cohort. Median follow-up since the commencement of radiotherapy for this analysis was 28.4 months (range, 5-66 months). Table 1 shows the clinical characteristics of the patients analyzed. The risk of grade ≥2 late genitourinary or gastrointestinal toxicity is shown in Fig. 1. There were four cases of grade 3 toxicity (4.8%), but no cases of grade 4 or 5 toxicity.

Fig. 1.

A, cumulative risk of Radiation Therapy Oncology Group grade ≥2 (Gr. 2+) rectal or bladder toxicity in the cohort (n = 83). B to H, cumulative risk of Radiation Therapy Oncology Group grade ≥2 rectal or bladder toxicity with respect to SNPs from (B) ERCC2, (C) LIG4, (D) MSH6, (E) CYP2D6*4, (F) dose received by 30% of rectal volume, (G) age at diagnosis, and (H) mean bladder dose. RT, radiation therapy.

Fig. 1.

A, cumulative risk of Radiation Therapy Oncology Group grade ≥2 (Gr. 2+) rectal or bladder toxicity in the cohort (n = 83). B to H, cumulative risk of Radiation Therapy Oncology Group grade ≥2 rectal or bladder toxicity with respect to SNPs from (B) ERCC2, (C) LIG4, (D) MSH6, (E) CYP2D6*4, (F) dose received by 30% of rectal volume, (G) age at diagnosis, and (H) mean bladder dose. RT, radiation therapy.

Close modal

Analysis of gene polymorphisms. Forty-nine SNPs in 24 genes were genotyped in 83 patient samples (Table 2). Of these, 15 were synonymous and 28 were nonsynonymous SNPs, and the remaining in regulatory regions or splice variants. Sixteen SNPs from BRCA1 and BRCA2 and 10 SNPs from CYP450 genes were also analyzed in this study for their association with radiation-induced toxicity. The minor allele frequencies (MAF) for the 49 SNPs ranged from 1% to 48%. Thirty-seven of the SNPs showed MAF of >5%, nine SNPs showed a MAF of <5%, and three SNPs showed no polymorphism in our study population (Table 2). Only one synonymous SNP in BRCA2 (T>C, His743His; Table 2) showed deviation from the Hardy-Weinberg equilibrium in cases (n = 83) but not in the apparently healthy controls (n = 90; data not shown). CYP1A1 (A>C, Asn461Thr) and MLH6 (A>T, Thr1102Thr) showed only one heterozygote and no variants, and the CYP2C19*3, CYP2D6*6, and CYP2D6*8*14 SNPs were not polymorphic. We eliminated these six SNPs and analyzed the remaining 43 SNPs for their association with radiation-induced toxicity. Allele calls for SNPs in TGFB1 (Leu10Pro), XRCC1 (A>G, Arg399Gln), and XRCC3 (C>T, Thr241Met) were difficult to interpret using the Pyrosequencing technique and were not included for analysis in this study.

Association of candidate gene SNPs to radiation-induced toxicity. In the Radiation Therapy Oncology Group late rectal or bladder toxicity grading scheme, complications are scored as they arise ≥90 days after radiation treatment. The risk of late rectal or bladder toxicity (grade ≥2) was significantly associated (P < 0.03) with the following SNPs in the univariate analysis (Table 3): MLH1 (C>T, Val219Ile), XRCC3 (A>G, 5′ untranslated region NT 4541), LIG4 (T>C, Asp568Asp), and CYP2D6*4 (G>A, splicing defect). Significant univariate associations with toxicity were found for the following clinical and dosimetric variables (Table 3): mean rectal dose >60 Gy (P = 0.03), and a significance value of P < 0.01 for mean bladder dose >60 Gy, dose received by 30% of rectal volume >75 Gy, dose received by 30% of bladder volume >75 Gy, and age at diagnosis <60 years. ERCC2 (G>A, Asp711Asp), MSH6 (T>C, Asp180Asp), BRCA2 (A>G, Lys1132Lys), and BRCA1 (G>A, Ile1652Met) gene SNPs had marginal significance on univariate analysis (Table 3). Use of androgen deprivation therapy was not significantly associated with risk of toxicity (data not shown). On the Cox multivariate analysis, we found high statistical significance for LIG4 (T>C, Asp568Asp; P = 0.0004) and ERCC2 (G>A, Asp711Asp; P = 0.02) SNPs, mean bladder dose >60 Gy (P = 0.001), dose to 30% of rectal volume >75 Gy (P = 0.0037), and age <60 years (P = 0.01; Table 4A and B). Kaplan-Meier graphs for select variables from univariate analysis are presented (Fig. 1) as these variables exhibited the highest level of significance in the multivariate analysis. The genotype CC or CT in LIG4 had the highest incidence of toxicity (close to 55%) followed by the variant allele TT (close to 35%). ERCC2 showed the most toxicity for the AA variant genotype followed by the wild-type and heterozygous genotypes GG and GA. Similarly, MSH6 and CYP2D6*4 showed the highest toxicities for the genotypes GA + GG and AA + GA, respectively.

Table 3.

Univariate analysis of factors associated with Radiation Therapy Oncology Group grade ≥2 chronic toxicity

VariableTypeCasesCox PHazard ratio (95% confidence interval)
MLH1 C>T, Val219Ile CC 0.024 2.86 (1.15-7.11) 
ERCC2 G>A, Asp711Asp AA 0.078 2.41 (0.91-6.39) 
MSH6 T>C, Asp180Asp GA or GG 39 0.053 2.10 (0.99-4.46) 
XRCC3 A>G, 5′ UTR NT 4541 AA 52 0.004 4.83 (1.67-13.99) 
NBN G>C, Glu185Gln CC or CG 70 0.150 2.89 (0.69-12.19) 
TGFB1 G>C, Arg25Asp GC 13 0.170 1.89 (0.76-4.67) 
L1G 4 T>C, Asp568Asp CC or CT 23 0.014 2.64 (1.22-5.70) 
CYP2D6*4 G>A, splicing defect AA or AG 23 0.013 2.63 (1.23-5.63) 
BRCA1 G>A, Met1652Ile TC 0.085 2.88 (0.86-9.61) 
BRCA1 A>G, Arg356Gln CT 0.110 2.23 (0.84-5.87) 
BRCA2 A>G, Lys1132Lys AA or GG 47 0.061 2.20 (0.97-5.00) 
ATM A>G, Asp1853Asn GG 58 0.130 2.12 (0.81-5.59) 
Mean rectal dose >60 Gy 25 0.030 2.31 (1.09-4.93) 
Mean bladder dose >60 Gy 20 0.003 3.11 (1.45-6.68) 
Dose to 30% of rectal volume >75 Gy 17 0.001 3.52 (1.63-7.57) 
Dose to 30% of bladder volume >75 Gy 25 0.012 2.60 (1.23-5.46) 
Age at diagnosis ≥60 y 75 0.006 0.25 (0.09-0.66) 
Abdominopelvic surgery Yes 30 0.090 1.90 (0.91-3.99) 
VariableTypeCasesCox PHazard ratio (95% confidence interval)
MLH1 C>T, Val219Ile CC 0.024 2.86 (1.15-7.11) 
ERCC2 G>A, Asp711Asp AA 0.078 2.41 (0.91-6.39) 
MSH6 T>C, Asp180Asp GA or GG 39 0.053 2.10 (0.99-4.46) 
XRCC3 A>G, 5′ UTR NT 4541 AA 52 0.004 4.83 (1.67-13.99) 
NBN G>C, Glu185Gln CC or CG 70 0.150 2.89 (0.69-12.19) 
TGFB1 G>C, Arg25Asp GC 13 0.170 1.89 (0.76-4.67) 
L1G 4 T>C, Asp568Asp CC or CT 23 0.014 2.64 (1.22-5.70) 
CYP2D6*4 G>A, splicing defect AA or AG 23 0.013 2.63 (1.23-5.63) 
BRCA1 G>A, Met1652Ile TC 0.085 2.88 (0.86-9.61) 
BRCA1 A>G, Arg356Gln CT 0.110 2.23 (0.84-5.87) 
BRCA2 A>G, Lys1132Lys AA or GG 47 0.061 2.20 (0.97-5.00) 
ATM A>G, Asp1853Asn GG 58 0.130 2.12 (0.81-5.59) 
Mean rectal dose >60 Gy 25 0.030 2.31 (1.09-4.93) 
Mean bladder dose >60 Gy 20 0.003 3.11 (1.45-6.68) 
Dose to 30% of rectal volume >75 Gy 17 0.001 3.52 (1.63-7.57) 
Dose to 30% of bladder volume >75 Gy 25 0.012 2.60 (1.23-5.46) 
Age at diagnosis ≥60 y 75 0.006 0.25 (0.09-0.66) 
Abdominopelvic surgery Yes 30 0.090 1.90 (0.91-3.99) 

NOTE: Univariate analysis of factors prognostic for Radiation Therapy Oncology Group grade ≥2 rectal or bladder toxicity. Cox P values < 0.03 are indicated in bold.

Table 4.

Multivariate analysis of factors associated with Radiation Therapy Oncology Group grade ≥2 chronic toxicity

A.
Best subset of 4 predictors (Score χ2 = 36.6, P < 0.0001)
Second best subset of four predictors (score χ2 = 34.9, P < 0.0001)
VariablePHazard (95% confidence interval)PHazard (95% confidence interval)
LIG 4 T>C, Asp568Asp 0.0004 4.86 (2.04-11.56) 0.0034 3.56 (1.52-8.31) 
CYP2D6*4 G>A, splicing defect — — 0.0110 2.85 (1.27-6.38) 
Age at diagnosis 0.0108 0.24 (0.08-0.72) 0.0002 0.13 (0.04-0.38) 
Mean bladder dose 0.0111 3.02 (1.29-7.08) — — 
Dose to 30% of rectal volume 0.0037 3.27 (1.47-7.29) 0.0005 4.14 
     
B.
 
    
 Best subset of five predictors (score χ2 = 42.0, P < 0.0001)
 
 Second best subset of five predictors (score χ2 = 41.7, P < 0.0001)
 
 
Variable P Hazard (95% confidence interval) P Hazard (95% confidence interval) 
LIG 4 T>C, Asp568Asp 0.0002 5.37 (2.22-12.97) 0.0003 5.22 (2.14-12.73) 
ERCC2 G>A, Asp711Asp 0.0207 3.54 (1.21-10.35) — — 
MSH6 T>C, Asp180Asp — — 0.0074 3.08 (1.35-7.01) 
Age at diagnosis 0.0036 0.19 (0.06-0.58) 0.0057 0.22 (0.07-0.64) 
Mean bladder dose 0.0086 3.19 (1.34-7.59) 0.0029 3.90 (1.60-9.55) 
Dose to 30% of rectal volume 0.0105 2.89 (1.28-6.53) 0.0012 3.81 (1.70-8.55) 
A.
Best subset of 4 predictors (Score χ2 = 36.6, P < 0.0001)
Second best subset of four predictors (score χ2 = 34.9, P < 0.0001)
VariablePHazard (95% confidence interval)PHazard (95% confidence interval)
LIG 4 T>C, Asp568Asp 0.0004 4.86 (2.04-11.56) 0.0034 3.56 (1.52-8.31) 
CYP2D6*4 G>A, splicing defect — — 0.0110 2.85 (1.27-6.38) 
Age at diagnosis 0.0108 0.24 (0.08-0.72) 0.0002 0.13 (0.04-0.38) 
Mean bladder dose 0.0111 3.02 (1.29-7.08) — — 
Dose to 30% of rectal volume 0.0037 3.27 (1.47-7.29) 0.0005 4.14 
     
B.
 
    
 Best subset of five predictors (score χ2 = 42.0, P < 0.0001)
 
 Second best subset of five predictors (score χ2 = 41.7, P < 0.0001)
 
 
Variable P Hazard (95% confidence interval) P Hazard (95% confidence interval) 
LIG 4 T>C, Asp568Asp 0.0002 5.37 (2.22-12.97) 0.0003 5.22 (2.14-12.73) 
ERCC2 G>A, Asp711Asp 0.0207 3.54 (1.21-10.35) — — 
MSH6 T>C, Asp180Asp — — 0.0074 3.08 (1.35-7.01) 
Age at diagnosis 0.0036 0.19 (0.06-0.58) 0.0057 0.22 (0.07-0.64) 
Mean bladder dose 0.0086 3.19 (1.34-7.59) 0.0029 3.90 (1.60-9.55) 
Dose to 30% of rectal volume 0.0105 2.89 (1.28-6.53) 0.0012 3.81 (1.70-8.55) 

NOTE: Summary of the exploratory multivariate analysis of factors prognostic for Radiation Therapy Oncology Group grade ≥2 rectal or bladder toxicity.

Using multivariate analysis (SAS score method), a combination of five predictors, providing 16 cases per predictor (Table 4B), helped determine the best overall model fit compared with the combination of four predictors (Table 4A). The significance level of each predictor was obtained with its hazard ratio and 95% confidence interval for the first and second best combination of five predictors. The first set (overall model fit, χ2 = 42, P < 0.0001) contained ERCC2 (G>A, Asp711Asp; P = 0.021), LIG4 (T>C, Asp568Asp; P = 0.0002), mean bladder dose (P = 0.0086), dose to 30% of rectal volume (P = 0.0105), and age (P = 0.0036), whereas the second set (overall model fit, χ2 = 41.7, P < 0.0001), consisted of MSH6 (T>C, Asp180Asp; P = 0.073), LIG4 (T>C, Asp568Asp; P = 0.0042), CYP2D6*4 (G>A, splicing defect, P = 0.032), dose to 30% of rectal volume (P = 0.0003), and age (P = 0.0002).

Association of polymorphisms in the DNA repair pathways and radiation-induced late toxicity. Following damage to DNA, the DSB repair pathway is activated, with an early event being the activation of a multiprotein complex comprised of MRE11, RAD50, and NBN (MRN). Damaged DNA is repaired by one or a combination of pathways of which nonhomologous end joining and homologous recombination are the subject of intense investigation in DSB repair. In nonhomologous end joining pathway, additional proteins, such as Ku70/Ku80, DNA-PK, XRCC4, and Ligase 4, are recruited to the MRN complex. In the homologous recombination repair pathway, RAD51 and RAD52 are involved in a strand exchange reaction, and a multiprotein complex containing XRCC2 and XRCC3 is initiated to repair the DSBs in the damaged DNA. Rad51 has been shown to interact with XRCC3 and BRCA1 (28). ATM mediates both upstream (NBN) and downstream (BRCA1) phosphorylation events (28). Representative SNPs analyzed in this study were derived from all of these classes of DNA repair genes. In addition, we considered other DNA repair pathway gene polymorphisms, such as mismatch repair (MLH and MSH) and nucleotide excision repair (ERCC). The mismatch repair pathway contributes to DNA replication by correcting errors in proofreading and controls homologous recombination by preventing strand exchange between divergent DNA sequences. Thus, mismatch repair has also been implicated in the repair of DNA damage by ionizing radiation (39). The nonhomologous end joining pathway is considered not very stringent and is therefore error prone. The eukaryotic cell, therefore, evolved mechanisms to overcome these limitations by using several distinct yet convergent DNA repair pathways (40, 41).

We have genotyped a total of 49 SNPs in DNA repair, xenobiotic metabolism, and tissue repair genes, of which 37 showed a MAF of >5% and were analyzed in combination with the six SNPs with MAF of <5%. Although case-control association studies typically only include SNPs with MAF of >5% (representing more common variants in the population), we included the low-frequency alleles (<5% MAF) in the analysis because such alleles may confer disease susceptibility. We did not observe any associations with the low-frequency alleles. The significant SNPs from univariate analysis showed no selectivity for a particular DNA repair pathway.

The XRCC3 SNP in the 5′ untranslated region showed a statistically significant association with toxicity on univariate analysis (P = 0.004), whereas the RAD51 SNP in the 5′ untranslated region showed no significance. The MSH6 (P = 0.053) and MLH1 (P = 0.024) mismatch repair genes, which encode proteins with roles in damage recognition, as well as LIG4, showed significance (P = 0.014) in the univariate analysis (Fig. 1; Table 3).

The nonsynonymous SNP in MLH1 is predicted to yield a functionally conserved amino acid change from Val-to-Ile at position 219, whereas the MSH6, ERCC2, and LIG4 gene SNPs are synonymous changes. The MSH6 and LIG4 SNPs are associated with breast cancer risk, presumably through linkage disequilibrium with variant alleles in neighboring genes (28, 42). Two ERCC2 SNPs (C>T, Asp312Asn and G>T, Lys751Gln) were investigated in cancer predisposition case-control studies and for association with the acute skin reactions in breast cancer patients following radiotherapy (43, 44) and showed no significance. However, in prostate cancer patients receiving radiotherapy, a synonymous SNP in ERCC2 (G>A, Asp711Asp) showed marginal significance in both univariate (P = 0.078) and multivariate (P = 0.021) analyses.

Ligase 4 plays an important role in nonhomologous end joining and in V(D)J recombination and was shown to physically interact with XRCC4 (45). The LIG4 (T>C, Asp568Asp) SNP showed an association with decreased breast cancer risk in a study of >1,000 patients (28). It was suggested that linkage disequilibrium with other functional variants in the same or adjoining genes could have led to this association. Our results confirmed that the LIG4 (T>C, Asp568Asp) SNP is not only important in the assessment of cancer risk but is also an important marker associated with the susceptibility of individuals to clinical radiosensitivity. LIG4 syndrome patients, who have an Arg278His mutation, have also been characterized as radiosensitive (45). One LIG4 syndrome patient who only had the Arg278His mutation (SNP, by definition, has a MAF of >1% and mutations are those occurring below 1%) was diagnosed with immunodeficiency but showed no developmental delays and was clinically normal. However, more severely affected patients had two additional polymorphisms (Ala3Val and Thr9Ile) in the LIG4 gene. Detailed biochemical investigations revealed that the activity of Ligase 4 with the Arg278His mutation was only 10% of wild-type protein activity. Near complete loss of activity occurred when the Arg278His mutation and the neighboring polymorphisms coexisted on the same LIG4 gene. These studies indicated that polymorphisms that are otherwise silent, or those exerting marginal effects when assessed individually, could affect protein function when present simultaneously. Additive effects of such SNPs in the variant protein may have profound clinical manifestations. In a separate study, LIG4 variant protein with combined Ala3Val and Thr9Ile mutations showed ∼50% of wild-type activity (46, 47). It would be interesting to also directly evaluate the association of the Ala3Val and Thr9Ile variants to clinical radiosensitivity in prostate cancer patients in our cohort as well as resequence the LIG4 gene for possible Arg278His mutations.

Two SNPs in XRCC3 (C>T, Thr241Met and A>G, intervening sequence 5-14) were shown to be associated with increased risk and dominant protective effects, respectively, for breast cancer (28). The mutant (C>T, Thr241Met) XRCC3 protein was associated with increased risk for fibrosis in breast cancer patients receiving radiotherapy (28). Microsatellite polymorphisms and SNPs in XRCC3 were associated with radiosensitivity in melanoma and bladder cancer (10). In our study, XRCC3 (A>G, intervening sequence 5-14) did not show association with radiotherapy toxicity, whereas XRCC3 (A>G, 5′ untranslated region 4541) showed significance in univariate analysis. Future expanded studies will include the XRCC3 (C>T, Thr241Met) SNP.

The ATM gene has been extensively investigated in the context of radiation-induced normal tissue damage (27). Individuals with alterations, such as deletions or stop codons in the ATM, BRCA1, or BRCA2 genes, are highly sensitive to radiation (high-penetrance alleles). The SNPs considered in this study may confer a radiation sensitive phenotype in a finite but additive manner with other gene polymorphisms (polygenic) and are considered as low-penetrance alleles. ATM (A>G, Asp1853Asn) was correlated to risk of radiation-induced fibrosis in breast cancer patients treated with radiotherapy (48). ATM sequence variants in prostate cancer patients (n = 37) treated with radiation showed correlation to radiosensitivity for various exonic and intronic SNPs but not for ATM (A>G, Asp1853Asn; ref. 48). We also did not find a significant association between this SNP and the late toxicity of rectum or bladder (Tables 3 and 4). Similarly, the SNPs analyzed in XRCC1 and NBN did not show an association with radiation-induced toxicity in rectum or bladder (Table 3). BRCA1 (G>A, Met1652Ile) and BRCA2 (A>G, Lys1132Lys) showed marginal significance in univariate analysis (Table 3).

Association of CYP2D6 gene polymorphism and radiation-induced late toxicity. Surprisingly, the steroid/xenobiotic metabolism gene CYP2D6*4 showed high significance (P = 0.013) in univariate analysis. A G>A change in exon 4 creates a splicing defect in CYP2D6*4 because this SNP is located in the exon-intron boundary. CYP2D6*4 has been associated with the metabolism of naturally occurring estrogens and antiestrogen metabolite drugs used in the treatment of various cancers, particularly neoplasms of the breast and prostate (49). We speculate that there may be a role for the CYP2D6*4 protein as a modulator of radiosensitivity by an unknown mechanism, similar to the modulatory role of estrogens in mediating DSBs, which in turn are predicted to increase susceptibility to breast cancer (26). Radiation exposure can result in tissue inflammation and increased synthesis of anti-inflammatory steroids. Cytokine gene regulatory networks also play a crucial role in the response of irradiated tissues. For instance, increased circulatory levels of TGFB1 were documented in patients treated with radiation (11). This cytokine was associated with radiation-induced fibrosis (11). Cytokines, such as tumor necrosis factor-α, interleukin-1, interleukin-6, and interleukin-10, were also shown to mediate proinflammatory or profibrotic response in normal tissues, and the phenotypic characteristics may vary in a tissue-specific manner (10). Evidence for the down-regulation of CYP genes in response to increased cytokine production in inflammatory diseases, such as rheumatoid arthritis, has been investigated (29, 50).

Tissue-specific expression of CYP2D6 in extrahepatic tissues, such as bladder, as well as an association of CYP2D6 alleles in prostate cancer, has been reported (51, 52). Despite the marginal association in multivariate analysis between CYP2D6*4 and radiation toxicity described in this article, a putative role of CYP2D6 in tissue damage or repair has not been clearly established. Linkage of CYP2D6 to ankylosing spondylitis (a familial inflammatory rheumatic disorder) has been shown (50). Another potential explanation for the observed association of clinical radiation toxicity with CYP2D6*4, which maps to chromosome 22q13.1, could be envisaged if this SNP is in linkage disequilibrium with the putative functional/causative allele. Tissue inhibitor of metalloproteinase 3 (TIMP3), the gene for which maps to chromosome 22 (22q12.1-q13.2; 22q12.3), is a natural inhibitor of matrix metalloproteinases involved in degradation of extracellular matrix (53). Production of reactive oxygen species following radiation exposure induces expression of the TIMP3 gene through TGFB1 in inflamed tissues (54). TIMP3 expression is also correlated to benign prostatic hyperplasia, hormonal regulation of endometrial tissue remodeling, and inhibition of vascular endothelial factor–mediated angiogenesis (55, 56). The proximity of CYP2D6 and TIMP3 on chromosome 22 and the biological relevance of TIMP3 in tissue modeling would suggest that TIMP3 could be one of the genes in linkage disequilibrium with the CYP2D6*4 allele (53). Further investigations are needed to validate this premise.

A polymorphism in exon 10 of BRCA2 (A>C, Asn372His), a gene product involved in DSB repair, was shown to confer increased risk for breast cancer (57), and the nonconserved amino acid change was thought to affect binding of histone acetyl transferase to the BRCA2 protein. However, it was also postulated that linkage disequilibrium may exist with another SNP in BRCA2 or another gene on the same chromosome. Interestingly, the KL (klotho) gene was described within the 1.4 Mb of BRCA2. KL encodes a membrane protein related to β-glucosidase that plays a role in premature ageing, renal failure, and atherosclerosis; importantly, a SNP in KL was found to be associated with osteoporosis in aged postmenopausal women (57). Such associations are not uncommon with low-penetrance alleles in complete linkage disequilibrium with causal variants on the same chromosome.

Association of TGFB1 gene polymorphisms and radiation-induced late toxicity. SNPs in TGFB1 have been associated with radiation-induced normal tissue fibrosis (11). TGFB1 alleles (G>A at −800; C>T at −509; T>C, Leu10Pro; G>C, Arg25Pro) have been analyzed in radiation sensitivity studies of breast and lung cancers (11). We have included two TGFB1 SNPs (G>C, Arg25Pro and C>T at −509) in our analysis (Table 1). Both of these SNPs showed no correlation to radiation-induced toxicity in prostate cancer patients. TGFB1 (G>C, Arg25Pro) was shown to be associated with fibrosis in patients receiving radiation therapy for lung cancer (58). However, this SNP showed no association for fibrosis phenotype in breast cancer (11) or prostate cancer (this study) patients following radiotherapy.

It is recognized that in this historical cohort the risk of grade ≥2 toxicity was ∼40%, which although mostly indicating grade 2 toxicity, is higher than the 22% to 33% range of other cohorts in the literature, which increasingly have used more sophisticated radiotherapy planning, such as intensity-modulated radiotherapy and implanted radio-opaque intraprostatic fiducial markers with online setup correction (5961). The risk of serious, grade ≥3 toxicity in our cohort was no greater than in these studies. A caveat should, however, be noted regarding the relative use of a physician-scored retrospective review versus a prospective patient-scored toxicity profile instrument, as it relates to the toxicity data. Clearly, prospective scoring has potential advantages as a means of obtaining toxicity data because of the potential for reporting bias with retrospective collection. Prospective collection of patient-reported toxicity scores would be even more ideal, and we recommend that this be incorporated into future studies.

With regard to the observed association between age at diagnosis and toxicity, in larger cohorts previously described in the literature, this association has usually been absent (62). Caution should be exercised in generalizing this result because the cohort with age at diagnosis <60 years is quite small (eight patients). Further studies with much larger cohorts of younger men will be required before it is possible to draw definitive conclusions about the apparent association of radiotherapy toxicity and age.

These results presented here confirm the convergence of the multiple tissue and DNA repair pathways in response to radiation damage to DNA. Further studies are in progress for a more comprehensive analysis on a genome wide scale, including those shown in published reports using candidate SNPs in apoptosis and cell cycle genes in predicting the adverse radiotherapy toxicity response. If validated in future studies, these results could aid in the construction of predictive risk models for the occurrence of bothersome and/or severe late radiotherapy toxicity in individual patients with prostate cancer undergoing dose-escalated radiotherapy with curative intent. Compared with current radiobiological models that use population data to predict radiotherapy late effects, this could lead to a model based upon genomically and dosimetrically derived variables to predict late effects.

Grant support: Research Initiatives Program of the Alberta Cancer Board.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Kathryn Calder and Lillian Cook from the PolyomX Program for assistance in the banking of clinical samples.

1
Zelefsky MJ, Fuks Z, Hunt M, et al. High dose radiation delivered by intensity modulated conformal radiotherapy improves the outcome of localized prostate cancer.
J Urol
2001
;
166
:
876
–81.
2
Pollack A, Zagars GK, Starkschall G, et al. Prostate cancer radiation dose response: results of the M.D. Anderson phase III randomized trial.
Int J Radiat Oncol Biol Phys
2002
;
53
:
1097
–105.
3
Ryu JK, Winter K, Michalski JM, et al. Interim report of toxicity from 3D conformal radiation therapy (3D-CRT) for prostate cancer on 3DOG/RTOG 9406, level III (79.2 Gy).
Int J Radiat Oncol Biol Phys
2002
;
54
:
1036
–46.
4
Koper PC, Jansen P, van Putten W, et al. Gastro-intestinal and genito-urinary morbidity after 3D conformal radiotherapy of prostate cancer: observations of a randomized trial.
Radiother Oncol
2004
;
73
:
1
–9.
5
Chism DB, Horwitz EM, Hanlon AL, Pinover WH, Mitra RK, Hanks GE. Late morbidity profiles in prostate cancer patients treated to 79–84 Gy by a simple four-field coplanar beam arrangement.
Int J Radiat Oncol Biol Phys
2003
;
55
:
71
–7.
6
Tucker SL, Cheung R, Dong L, et al. Dose-volume response analyses of late rectal bleeding after radiotherapy for prostate cancer.
Int J Radiat Oncol Biol Phys
2004
;
59
:
353
–65.
7
Skwarchuk MW, Jackson A, Zelefsky MJ, et al. Late rectal toxicity after conformal radiotherapy of prostate cancer (I): multivariate analysis and dose-response.
Int J Radiat Oncol Biol Phys
2000
;
47
:
103
–13.
8
Russell NS, Grummels A, Hart AA, et al. Low predictive value of intrinsic fibroblast radiosensitivity for fibrosis development following radiotherapy for breast cancer.
Int J Radiat Biol
1998
;
73
:
661
–70.
9
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
2003
;
63
:
8717
–25.
10
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
2002
;
64
:
131
–40.
11
Quarmby S, Fakhoury H, Levine E, et al. Association of transforming growth factor beta-1 single nucleotide polymorphisms with radiation-induced damage to normal tissues in breast cancer patients.
Int J Radiat Biol
2003
;
79
:
137
–43.
12
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
2001
;
50
:
1323
–31.
13
Varghese S, Schmidt-Ullrich RK, Dritschilo A, Jung M. Enhanced radiation late effects and cellular radiation sensitivity in an ATM heterozygous breast cancer patient.
Radiat Oncol Investig
1999
;
7
:
231
–7.
14
Andreassen CN, Alsner J, Overgaard M, Overgaard J. Prediction of normal tissue radiosensitivity from polymorphisms in candidate genes.
Radiother Oncol
2003
;
69
:
127
–35.
15
Andreassen CN, Sorensen FB, Overgaard J, Alsner J. Optimisation and validation of methods to assess single nucleotide polymorphisms (SNPs) in archival histological material.
Radiother Oncol
2004
;
72
:
351
–6.
16
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
1997
;
23
:
237
–47.
17
Leadon SA. Repair of DNA damage produced by ionizing radiation: a minireview.
Semin Radiat Oncol
1996
;
6
:
295
–305.
18
Haimovitz-Friedman A. Radiation-induced signal transduction and stress response.
Radiat Res
1998
;
150
:
S102
–8.
19
Wood RD, Mitchell M, Sgouros J, Lindahl T. Human DNA repair genes.
Science
2001
;
291
:
1284
–9.
20
Mohrenweiser HW, Xi T, Vazquez-Matias J, Jones IM. Identification of 127 amino acid substitution variants in screening 37 DNA repair genes in humans.
Cancer Epidemiol Biomarkers Prev
2002
;
11
:
1054
–64.
21
Ting JY, Wu X, Fiedler JA, Yang C, Watzich ML, Markoe A. Dose-volume histograms for bladder and rectum.
Int J Radiat Oncol Biol Phys
1997
;
38
:
1105
–11.
22
Stavrev PV, Hristov DH, Seuntjens JP. On the consistent use of organ definitions and radiobiological models for the evaluation of complication probabilities of “tubular” organs.
Int J Radiat Oncol Biol Phys
2002
;
52
:
1150
–2.
23
Rosen EM, Fan S, Goldberg ID, Rockwell S. Biological basis of radiation sensitivity. Part 1: factors governing radiation tolerance.
Oncology (Williston Park)
2000
;
14
:
543
–50.
24
Rosen EM, Fan S, Rockwell S, Goldberg ID. The molecular and cellular basis of radiosensitivity: implications for understanding how normal tissues and tumors respond to therapeutic radiation.
Cancer Invest
1999
;
17
:
56
–72.
25
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
2005
;
11
:
4802
–9.
26
Fernet M, Hall J. Genetic biomarkers of therapeutic radiation sensitivity.
DNA Repair (Amst)
2004
;
3
:
1237
–43.
27
Stone HB, Coleman CN, Anscher MS, McBride WH. Effects of radiation on normal tissue: consequences and mechanisms.
Lancet Oncol
2003
;
4
:
529
–36.
28
Kuschel B, Auranen A, McBride S, et al. Variants in DNA double-strand break repair genes and breast cancer susceptibility.
Hum Mol Genet
2002
;
11
:
1399
–407.
29
Morgan ET. Regulation of cytochrome p450 by inflammatory mediators: why and how?
Drug Metab Dispos
2001
;
29
:
207
–12.
30
Chang BL, Zheng SL, Isaacs SD, et al. Polymorphisms in the CYP1A1 gene are associated with prostate cancer risk.
Int J Cancer
2003
;
106
:
375
–8.
31
Chang BL, Zheng SL, Isaacs SD, et al. Polymorphisms in the CYP1B1 gene are associated with increased risk of prostate cancer.
Br J Cancer
2003
;
89
:
1524
–9.
32
Finnstrom N, Bjelfman C, Soderstrom TG, et al. Detection of cytochrome P450 mRNA transcripts in prostate samples by RT-PCR.
Eur J Clin Invest
2001
;
31
:
880
–6.
33
Fukatsu T, Hirokawa Y, Araki T, et al. Genetic polymorphisms of hormone-related genes and prostate cancer risk in the Japanese population.
Anticancer Res
2004
;
24
:
2431
–7.
34
Tanaka Y, Sasaki M, Kaneuchi M, Shiina H, Igawa M, Dahiya R. Polymorphisms of estrogen receptor alpha in prostate cancer.
Mol Carcinog
2003
;
37
:
202
–8.
35
Wadelius M, Andersson AO, Johansson JE, Wadelius C, Rane E. Prostate cancer associated with CYP17 genotype.
Pharmacogenetics
1999
;
9
:
635
–9.
36
Johnson MM, Houck J, Chen C. Screening for deleterious nonsynonymous single-nucleotide polymorphisms in genes involved in steroid hormone metabolism and response.
Cancer Epidemiol Biomarkers Prev
2005
;
14
:
1326
–9.
37
Vittinghoff E, Glidden DV, Shiboski SC, McCulloch CE. Regression methods in biostatistics: New York (NY): Springer Science + Business Media Inc.; 2005.
38
Jackson A, Skwarchuk MW, Zelefsky MJ, et al. Late rectal bleeding after conformal radiotherapy of prostate cancer. II. Volume effects and dose-volume histograms.
Int J Radiat Oncol Biol Phys
2001
;
49
:
685
–98.
39
Cejka P, Stojic L, Marra G, Jiricny J. Is mismatch repair really required for ionizing radiation-induced DNA damage signaling?
Nat Genet
2004
;
36
:
432
–3; author reply 4.
40
Allen C, Halbrook J, Nickoloff JA. Interactive competition between homologous recombination and non-homologous end joining.
Mol Cancer Res
2003
;
1
:
913
–20.
41
Takata M, Sasaki MS, Sonoda E, et al. Homologous recombination and non-homologous end-joining pathways of DNA double-strand break repair have overlapping roles in the maintenance of chromosomal integrity in vertebrate cells.
EMBO J
1998
;
17
:
5497
–508.
42
Listgarten J, Damaraju S, Poulin B, et al. Predictive models for breast cancer susceptibility from multiple single nucleotide polymorphisms.
Clin Cancer Res
2004
;
10
:
2725
–37.
43
Qiao Y, Spitz MR, Shen H, et al. Modulation of repair of ultraviolet damage in the host-cell reactivation assay by polymorphic XPC and XPD/ERCC2 genotypes.
Carcinogenesis
2002
;
23
:
295
–9.
44
King CR, Yu J, Freimuth RR, McLeod HL, Marsh S. Interethnic variability of ERCC2 polymorphisms.
Pharmacogenomics J
2005
;
5
:
54
–9.
45
Girard PM, Kysela B, Harer CJ, Doherty AJ, Jeggo PA. Analysis of DNA ligase IV mutations found in LIG4 syndrome patients: the impact of two linked polymorphisms.
Hum Mol Genet
2004
;
13
:
2369
–76.
46
Riballo E, Doherty AJ, Dai Y, et al. Cellular and biochemical impact of a mutation in DNA ligase IV conferring clinical radiosensitivity.
J Biol Chem
2001
;
276
:
31124
–32.
47
Riballo E, Critchlow SE, Teo SH, et al. Identification of a defect in DNA ligase IV in a radiosensitive leukaemia patient.
Curr Biol
1999
;
9
:
699
–702.
48
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
2005
;
61
:
196
–202.
49
Wojnowski L, Hustert E, Klein K, et al. Re: modification of clinical presentation of prostate tumors by a novel genetic variant in CYP3A4.
J Natl Cancer Inst
2002
;
94
:
630
–1; author reply 1–2.
50
Brown MA, Edwards S, Hoyle E, et al. Polymorphisms of the CYP2D6 gene increase susceptibility to ankylosing spondylitis.
Hum Mol Genet
2000
;
9
:
1563
–6.
51
Romkes-Sparks M, Mnuskin A, Chern HD, et al. Correlation of polymorphic expression of CYP2D6 mRNA in bladder mucosa and tumor tissue to in vivo debrisoquine hydroxylase activity.
Carcinogenesis
1994
;
15
:
1955
–61.
52
Febbo PG, Kantoff PW, Giovannucci E, et al. Debrisoquine hydroxylase (CYP2D6) and prostate cancer.
Cancer Epidemiol Biomarkers Prev
1998
;
7
:
1075
–8.
53
Collins JE, Goward ME, Cole CG, et al. Reevaluating human gene annotation: a second-generation analysis of chromosome 22.
Genome Res
2003
;
13
:
27
–36.
54
Li WQ, Qureshi HY, Liacini A, Dehnade F, Zafarullah M. Transforming growth factor Beta1 induction of tissue inhibitor of metalloproteinases 3 in articular chondrocytes is mediated by reactive oxygen species.
Free Radic Biol Med
2004
;
37
:
196
–207.
55
Qi JH, Ebrahem Q, Moore N, et al. A novel function for tissue inhibitor of metalloproteinases-3 (TIMP3): inhibition of angiogenesis by blockage of VEGF binding to VEGF receptor-2.
Nat Med
2003
;
9
:
407
–15.
56
Tunuguntla R, Ripley D, Sang QX, Chegini N. Expression of matrix metalloproteinase-26 and tissue inhibitors of metalloproteinases TIMP-3 and -4 in benign endometrium and endometrial cancer.
Gynecol Oncol
2003
;
89
:
453
–9.
57
Healey CS, Dunning AM, Teare MD, et al. A common variant in BRCA2 is associated with both breast cancer risk and prenatal viability.
Nat Genet
2000
;
26
:
362
–4.
58
Awad MR, El-Gamel A, Hasleton P, Turner DM, Sinnott PJ, Hutchinson IV. 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
1998
;
66
:
1014
–20.
59
Kupelian PA, Thakkar VV, Khuntia D, Reddy CA, Klein EA, Mahadevan A. Hypofractionated intensity-modulated radiotherapy (70 Gy at 2.5 Gy per fraction) for localized prostate cancer: long-term outcomes.
Int J Radiat Oncol Biol Phys
2005
;
63
:
1463
–8.
60
Nichol A, Chung P, Lockwood G, et al. A phase II study of localized prostate cancer treated to 75.6 Gy with 3D conformal radiotherapy.
Radiother Oncol
2005
;
76
:
11
–7.
61
Michalski JM, Winter K, Purdy JA, et al. Toxicity after three-dimensional radiotherapy for prostate cancer on RTOG 9406 dose level V.
Int J Radiat Oncol Biol Phys
2005
;
62
:
706
–13.
62
Liu M, Pickles T, Agranovich A, et al. Impact of neoadjuvant androgen ablation and other factors on late toxicity after external beam prostate radiotherapy.
Int J Radiat Oncol Biol Phys
2004
;
58
:
59
–67.