GermLine Variation in Superoxide Dismutase-2 (SOD2) and Survival Outcomes After Radiation Therapy for Prostate Cancer: Results of a Test and Validation Set Analysis.

BACKGROUND: Genetic variants in antioxidant pathways might decrease the efficacy of radiation therapy (RT) by suppressing the generation of reactive oxygen species. We studied the association between single nucleotide polymorphisms (SNPs) in the antioxidant gene superoxide dismutase-2 (SOD2) and cancer-specific outcomes after RT. PATIENTS AND METHODS: Among 816 prostate cancer patients who received radiation as primary therapy from the Physicians' Health Study and the Health Professionals Follow-up Study, we evaluated the association of 7 tagging SNPs in SOD2 with lethal prostate cancer (death from prostate cancer or distant metastasis among living patients). We sought to validate findings in a separate cohort of 612 prostate cancer patients treated with RT with a greater proportion of intermediate and high-risk Gleason scores at the Dana-Farber Cancer Institute. Genetic effects were analyzed using a codominant model, using the genotype homozygous for the major allele as baseline.
RESULTS: Among patients who underwent RT in the test cohort, there was a significant association between 3 of the 7 SOD2 SNPs and lethal prostate cancer: rs6917589 (overall P = .006), rs2758331 (P = .04) and the functional valine to alanine polymorphism in rs4880 (P = .04). These SNPs were not associated with outcome among men who had undergone prostatectomy. The associations were not replicated in the validation cohort.
CONCLUSION: Germline genetic variation in the SOD2 gene might be a predictive biomarker of response to RT for prostate cancer but is not consistently associated with outcome after RT across prostate cancer cohorts with different clinical characteristics.

INTRODUCTION

Germline variation in antioxidant pathways may alter the effect of cancer therapies that rely on the generation of cytotoxic reactive oxygen species (ROS). Somatic alterations in the antioxidant environment are also postulated to result in enhanced cancer cell survival (1). There is growing interest in molecular-based strategies that target antioxidant pathways to promote cancer cell killing via oxidative stress (2–4). Radiation therapy generates ROS that mediate DNA damage and other downstream effects on cancer cells (5). Patient germline variability in endogenous antioxidant enzymes involved in neutralizing ROS may explain variability in cancer-specific outcomes after radiation therapy (RT). For example, patients with increased capacity for neutralizing ROS may receive less benefit from RT compared to patients with impaired ability to neutralize cytotoxic ROS.

Superoxide dismutase-2 (SOD2) is a mitochondrial antioxidant enzyme that is an important ROS scavenger. SOD2 reduces superoxide anion to hydrogen peroxide and oxygen, which is then converted to water by catalase (CAT) and glutathione peroxidase (GPX) (Figure 1). Overexpression of mitochondrial SOD was previously shown to protect cells from radiation-induced neoplastic transformation (6) and decreased levels of SOD increased the radiosensitivity of prostate cancer cells in vitro (7). A specific polymorphism in codon 16 of SOD2, rs4880, results in a valine to alanine amino acid change and is postulated to decrease mitochondrial ROS by causing more efficient transport of the enzyme into the mitochondria (8, 9). The polymorphism would be expected to decrease the effectiveness of cancer therapies such as radiation therapy, which rely on formation of ROS. Polymorphisms in SOD2 were previously shown to be associated with late toxicity after radiation therapy for prostate cancer (10), breast cancer (11) and head and neck cancer (12).

Figure 1

Simplified schema of the relationship between SOD2, reactive oxygen species (ROS), and other enzymes involved in free-radical scavenging in tissue.

There is conflicting data on the prognostic significance of SOD2 polymorphisms and survival after cancer therapy (13, 14). This study sought to validate the association between SOD2 polymorphisms and cancer outcomes after radiation therapy for prostate cancer. We hypothesized that germline genetic variation in SOD2 is associated with outcome after RT and that the functional rs4880 polymorphism is associated with adverse prostate cancer outcomes.

MATERIALS AND METHODS

Patients and Outcomes

The test cohort was comprised of 816 participants from two prospective cohort studies, the Physicians’ Health Study (PHS; 1982 – 2009 N=387) and the Health Professionals Follow-up Study (HPFS; 1993–2010 N=429). The PHS (15–17) was a 2x2 randomized double-blind, placebo-controlled trial that began in 1982 and enrolled 22,071 U.S. male physicians ages 40–84 years to take 325 mg aspirin and/or 50 mg beta-carotene every other day or placebo. Participants were free from diagnosed cancer at enrollment and were followed with yearly questionnaires and postcards at 6 month intervals to ascertain endpoints, including prostate cancer. At baseline, 14,916 (68%) participants provided blood before randomization and cancer diagnosis. The prospective HPFS enrolled 51,529 male medical professionals in 1986 to investigate the causes of cancer and heart disease. These cohort participants are subsequently followed with biennial questionnaires designed to collect information about medical diagnoses and lifestyle factors. Response rates to the follow-up surveys are high at approximately 96% and 18,018 participants provided a blood sample between 1993 and 1995.

When a participant reported a diagnosis of prostate cancer, hospital records and pathology reports were requested and study physicians verified diagnosis by reviewing medical records and pathology reports to determine the Gleason grade, stage, and prostate specific antigen (PSA) level at diagnosis. This study includes men in the PHS and HPFS blood cohorts who were diagnosed with prostate cancer between 1982 and 2010 and who underwent radiation therapy. Participants were excluded if radiation therapy was not their primary treatment or if their first treatment was radical prostatectomy.

For comparison, we also analyzed the association of polymorphisms in SOD2 in patients who underwent radical prostatectomy without radiation therapy, reasoning that the genetic variations would have no impact after surgical intervention. This separate cohort included 1094 patients from the PHS (N=555) and the HPFS (N=539). These studies were approved by the institutional review board at the Harvard School of Public Health and Partners Health Care.

The validation cohort consisted of patients from the Prostate Clinical Research Information System (CRIS; 1990–2008 N=612) at the Dana-Farber Cancer Institute. CRIS consists of a central secure data repository of patient data, including baseline clinical and disease characteristics and information about treatment and outcomes. All prostate cancer patients at Dana Farber Cancer Institute and Brigham and Women’s Hospital were offered enrollment and 647 patients were initially identified for the validation cohort. Selected patients had prostate cancer, were treated with external beam radiation or brachytherapy, consented to provide information and tissue, and donated blood for research purposes. Patients were excluded if they had lymph node or distant metastases prior to radiation therapy, or if the samples failed > 50% of the genotyping assays.

For the test cohort, the primary outcome was time to development of lethal prostate cancer, defined as the time from initiation of radiation therapy (RT) to prostate-cancer-specific death or distant metastasis among living participants. Outcomes, including cause of death, were verified via death certificates and medical record review. Since it was not routinely verified in the PHS, we did not use biochemical recurrence as an outcome in the test cohort. For the validation cohort, the primary analysis evaluated the association between SNP genotypes and time to distant metastasis, which was defined as the time from the initiation of RT to the time when metastases developed. Due to shorter follow-up in the validation cohort, prostate-cancer death was not used as the primary outcome. As a secondary analysis, we also evaluated the association with time to biochemical recurrence. Time to biochemical recurrence was defined as the time from the start of radiation therapy to the time when nadir+2ng/mL occurred or to time of salvage therapy. If the outcome of interest did not occur, follow-up was censored on the last PSA date.

Genotyping

We characterized one candidate SNP (rs4880) and 6 tagging SNPs from SOD2 that were selected to capture genetic variation across the SOD2 gene, including 5kb upstream and downstream, with an average r2 > 0.80 (Tagger, http://www.broadinstitute.org/mpg/tagger/, using HapMap Release 21, CEU analysis panel: Utah residents of Northern and Western European Ancestry). For the test cohort, genotyping was performed at the Harvard Medical School – Partners Healthcare Center for Genetics and Genomics after extraction of DNA from whole blood using Biotrove Open Genetics and Genomics with a standard QIAmp kit (QIAGEN Inc. Chatsworth, CA) protocol. All SNPs had greater than 90% completion and the concordance was greater than 99% for blinded quality control samples. All SNPs were in Hardy-Weinberg equilibrium.

For the validation cohort, all DNA samples were extracted from patients’ peripheral whole blood by QIAamp DNA Blood mini kit (QIAGEN Inc.) according to the manufacturer’s instruction. Genotyping was performed at the core facility of Boston Children’s Hospital using Sequenom iPLEX matrix–assisted laser desorption/ionization time-of-flight mass spectrometry technology. Approximately 5% of randomly selected duplicates were included as the quality control. All SNPs had greater than 99% genotype passing rates and no discrepancy between duplicates was observed in the genotyping data. Laboratory personnel were blinded to all case status information.

Statistical Methods

Patient clinical and disease characteristics at the time of diagnosis were summarized by median and inter-quartile range for continuous variables and by number and percentage for categorical variables. For both the test and validation cohorts, we analyzed the genetic effects of SOD2 SNPs using the co-dominant model, where the heterozygous and homozygous minor allele genotypes were treated as separate categories and compared to the homozygous major allele genotype. For minor alleles with less than 10% frequency in the cohorts, we combined the minor homozygous with the heterozygous genotypes. The co-dominant model was used as it makes fewer assumptions about the nature of the effect of the minor allele on outcome as compared to the additive model.

Cox proportional hazards models were used to assess the unadjusted and adjusted association between SNP and outcome and were used to calculate hazard ratios and associated 95% confidence intervals. The adjusted models included biopsy Gleason score, log-transformed PSA at diagnosis, clinical stage, and age at treatment. The median age of diagnosis and treatment were the same. For the test cohort, year of diagnosis and cohort (PHS or HPFS) were also used as adjustment covariates, and missing values for the clinical variables used in the adjusted models were imputed using Multiple Imputation for Chained Equations (MICE) in R. The use of hormonal therapy was included in the adjusted model for only the validation cohort.

All reported p-values are 2-sided, with Bonferroni-corrected p<0.007 considered statistically significant and p<0.05 considered nominally significant. SAS version 9.3 (SAS institute Inc, Cary, North Carolina) and R version 3.0.2 were used for all analyses.

RESULTS

Table 1 shows the patient characteristics from the test (N=816) and validation cohorts (N=612). Patients in the test cohort were older (median age 73 versus 64 years in the validation cohort) and had longer follow-up compared to the validation cohort (median 10.2 years versus 6.8 years). They were more likely to have low grade Gleason ≤6 tumors (60%) and to be treated in an earlier time period than the validation cohort where most patients had higher risk Gleason ≥7 (43%) or Gleason 8–10 (28%) tumors. As shown in Table 2, the minor allele frequencies for the 7 polymorphisms in SOD2 were similar among the two cohorts. Three of the SNP’s (rs4880, rs2758331, rs2758329) were in linkage disequilibrium with r2≥0.8.

Table 1

Patient characteristics at diagnosis for the test and validation cohorts

		Test cohort (n=816)	Validation cohort (n=612)
Follow-up (years), median		10.2	6.8
Age at treatment (years), median [IQR]		73 [68,76]	64 [59,70]
Gleason Score, no. (%)	2–6	486 (60)	148 (24)
	7	196 (24)	261 (43)
	8–10	85 (10)	173 (28)
	Unknown	49 (6)	30 (5)
Clinical stage, no. (%)	T1/T2	744 (91)	458 (75)
	T3/T4/N1	50 (6)	23 (4)
	Unknown	22 (3)	131 (21)
PSA at diagnosis, median [IQR]		7.3 ng/mL [5.4,11.0]	7.7 ng/mL [5.2,15]
Year of treatment, no. (%)	1982–1991	83 (10)	18 (3)
	1992–2001	535 (66)	285 (47)
	2002–2010	198 (24)	309 (51)

Abbreviations: No. – number; IQR – interquartile range; PSA – prostate-specific antigen

Table 2

Allelic variation of the 7 candidate single nucleotide polymorphisms (SNPs) in the SOD2 gene among study patients in the test (N=816) and validation (N=612) cohorts of men with prostate cancer treated with radiation

SNP	Major/Minor Allele	MAF (Test)	MAF (Validation)	Type	Annotation
rs6917589	(A/G)	24 %	24%	-	31 of SOD2
rs2758331†	(G/T)	48%	47%	Synonymous	Intron
rs4880†	(T/C)	49%	49%	Non- Synonymous (Valine/Alanine)	Exon2
rs2758329†	(A/G)	48%	48%	-	31 of SOD2
rs5746151	(G/A)	<10%	<10%	-	31 of SOD2
rs2842980	(A/T)	21%	21%	-	31 of SOD2
rs7855	(T/C)	<10%	<10%	-	31 UTR, exon

Abbreviations: SNP – Single nucleotide polymorphism; MAF – minor allele frequency; A – adenine; G – guanine; T – thymine; C – cytosine; UTR – untranslated region

SNPs are in linkage disequilibrium with each other

During follow-up in the PHS and HPFS cohorts, there were 77 lethal prostate cancer events, of which 52 were cancer deaths and 25 were distant metastases among living patients. Known prognostic factors, including biopsy Gleason score (p-value < 0.001), log PSA (p-value = 0.008), clinical TNM-stage (p-value < 0.001), and year of diagnosis (p-value < 0.001) were associated with lethal prostate cancer. Table 3 shows that three of the seven SNPs were statistically significantly associated with the composite enpoint of prostate-cancer death or metastases among living participants, at p<0.05. rs6917589 polymorphism was associated with risk of lethal prostate cancer (p=0.006). Carriage of the C allele in rs4880, which results in the valine to alanine isoform of the enzyme, was associated with a nominally statistically significant decrease in risk of lethal prostate cancer (HR 0.37 for homozygous C/C and HR 0.84 for T/C genotype, p-value = 0.04) as compared to the T/T genotype. This borderline association was also observed for the minor allele genotypes amongst the other two tagging SNPs in linkage disequilibrium with rs4880 (rs2758331 and rs2758329).

Table 3

Associations between SOD2 polymorphisms and lethal prostate cancer among prostate cancer patients undergoing radiation therapy or radical prostatectomy in the test cohort, adjusted for Gleason score, PSA, clinical stage, age at treatment, year of diagnosis and cohort (PHS or HPFS)

	Radiation Therapy Cohort (N=816)						Radical Prostatectomy Cohort (N=1094)
SNP	Total	Events	Unadjusted HR (95% CI)	P-value	Adjusted HR (95% CI)	P-value	Total	Events	Unadjusted HR (95% CI)	P-value	Adjusted HR (95% CI)	P-value
SOD2
rs6917589
GG	46	5	1.70 (0.66, 4.38)	0.06	2.62 (0.98, 7.02)	0.006	39	2	0.86 (0.21, 3.56)	0.64	0.78 (0.18, 3.25)	0.54
AG	230	32	1.80 (1.10, 2.94)		2.21 (1.31, 3.72)		334	25	1.26 (0.75, 2.10)		1.31 (0.76, 2.25)
AA	429	32	REF		REF		598	35	REF		REF
rs2758331†
TT	177	9	0.52 (0.24, 1.17)	0.05	0.37 (0.16, 0.83)	0.04	234	18	1.05 (0.55, 2.01)	0.66	1.10 (0.56, 2.13)	0.67
GT	338	46	1.28 (0.74, 2.20)		0.84 (0.47, 1.50)		498	31	0.82 (0.47, 1.46)		0.84 (0.47, 1.52)
GG	193	18	REF		REF		246	19	REF		REF
rs4880†
CC	182	9	0.52 (0.23, 1.17)	0.04	0.37 (0.16, 0.84)	0.04	254	19	1.02 (0.53, 1.94)	0.77	1.10 (0.57, 2.13)	0.67
TC	343	47	1.31 (0.75, 2.28)		0.90 (0.50, 1.61)		495	32	0.85 (0.48, 1.51)		0.85 (0.47, 1.53)
TT	182	17	REF		REF		232	18	REF		REF
rs2758329†
GG	170	9	0.53 (0.24, 1.18)	0.05	0.39 (0.17, 0.88)	0.06	228	18	0.99 (0.52, 1.87)	0.27	1.08 (0.56, 2.09)	0.37
AG	336	46	1.27 (0.74, 2.19)		0.86 (0.48, 1.53)		500	27	0.66 (0.37, 1.18)		0.72 (0.40, 1.31)
AA	188	18	REF		REF		236	20	REF		REF
rs5746151
GA/AA	86	7	0.75 (0.35, 1.65)	0.48	0.59 (0.25, 1.37)	0.22	121	6	0.71 (0.31, 1.64)	0.42	0.62 (0.27, 1.46)	0.27
GG	627	63			REF		858	61	REF		REF
rs2842980
TT	35	2	0.59 (0.14, 2.42)	0.69	0.86 (0.20, 3.62)	0.64	43	3	0.92 (0.29, 2.97)	0.94	0.78 (0.24, 2.53)	0.79
AT	224	25	1.09 (0.67, 1.78)		1.26 (0.76, 2.07)		327	22	0.91 (0.55, 1.53)		0.85 (0.50, 1.44)
AA	444	46	REF		REF		604	43	REF		REF
rs7855
TC/CC	75	7	0.90 (0.41, 1.97)	0.80	0.97 (0.44, 2.13)	0.94	110	6	0.74 (0.32, 1.71)	0.48	0.45 (0.19, 1.06)	0.07
TT	639	66	REF		REF		875	64	REF		REF

Abbreviations: CI – Confidence Interval; SNP – Single nucleotide polymorphism; REF – reference;

SNPs in linkage disequilibrium with each other

In the cohort of patients who underwent radical prostatectomy for prostate cancer (N=1094), the median age at prostatectomy was 65 years and the median PSA at diagnosis was 6.2 ng/mL [IQR 4.7, 9.7]. In this cohort, 65% of patients had Gleason 6 or less and 94% of patients had clinical T1/2 tumors. With a median follow-up of 12 years, there were 71 occurrences of lethal prostate cancer, of which 43 were from prostate cancer deaths and 28 were from distant metastases among living patients. There was no association between any of the seven SNPs in SOD2 and lethal prostate cancer outcome after adjustment for age at radical prostatectomy, clinical TNM stage, log PSA, biopsy Gleason score, year of diagnosis, and cohort (Table 3).

We further examined the association of the 7 SNPs in SOD2 with prostate cancer recurrence and with development of metastatic disease in a separate higher risk cohort of prostate cancer patients undergoing radiation from the Dana-Farber Cancer Institute (N=612). The median follow-up time was 6.8 years (range 2 months – 20 years) from the initiation of radiation therapy. There were 277 patients that experienced biochemical recurrence, with a median time to biochemical recurrence of 4.5 years (95% CI: 3.9–5.2 years). Distant metastasis was also assessed as an outcome of interest based on a total of 168 patients that developed distant metastases and had a median time to distant metastasis of 11 years (95% CI: 10.4–13.5 years). In adjusted and unadjusted analyses, there was no association between rs6917589, rs4880, or other SNPs in SOD2 and distant metastasis or biochemical recurrence (Table 4 and Supplementary Table 1).

Table 4

Associations between SOD2 polymorphisms and distant metastases in the validation cohort (N=612), Dana-Farber Cancer Institute prostate cancer patients who underwent radiation therapy

SNP	Total	Events	Unadjusted HR (95% CI)	P-value	Adjusted HR (95% CI)	P-value
SOD2
rs6917589				0.97		0.97
GG	42	13	1.04 (0.59, 1.86)		1.04 (0.58, 1.88)
AG	210	51	0.97 (0.69, 1.36)		1.05 (0.74, 1.48)
AA	354	100	REF		REF
rs2758331				0.99		0.85
TT	128	36	1.02 (0.66, 1.57)		0.99 (0.63, 1.55)
GT	296	82	1.03 (0.72, 1.48)		1.09 (0.76, 1.58)
GG	180	46	REF		REF
rs4880				0.94		0.87
CC	141	38	0.93 (0.60, 1.44)		0.93 (0.60, 1.46)
TC	298	82	0.98 (0.68, 1.42)		1.04 (0.71, 1.52)
TT	165	44	REF		REF
rs2758329				0.99		0.90
GG	136	37	0.97 (0.63, 1.50)		0.94 (0.60, 1.46)
AG	296	82	1.00 (0.70, 1.44)		1.03 (0.71, 1.50)
AA	173	45	REF		REF
rs5746151				0.18		0.14
GA/AA	78	25	1.34 (0.87, 2.06)		1.40 (0.89, 2.20)
GG	528	139	REF		REF
rs2842980				0.54		0.33
AT/TT	238	63	0.91 (0.66, 1.24)		0.85 (0.61, 1.18)
AA	367	101	REF		REF
rs7855				0.63		0.14
TC/CC	72	19	0.89 (0.55, 1.44)		0.69 (0.41, 1.14)
TT	534	145	REF		REF

Abbreviations: SNP – Single nucleotide polymorphism; HR – Hazard ratio; CI – Confidence Interval; REF – reference

DISCUSSION

In a cohort of patients with predominantly lower risk prostate cancer that was treated with definitive radiation therapy, the SOD2 rs6917589 was associated with risk of lethal prostate cancer. There were borderline statistically significant associations between rs2758331 and the functional SOD2 rs4880 polymorphism and lethal prostate cancer in the test cohort. Of note, these three SOD2 polymorphisms were not predictive of cancer-specific outcomes after radical prostatectomy.

The initial finding was not reproduced in a cohort of men with a higher proportion of intermediate to high-grade Gleason scores, where there was no association between any SOD2 polymorphism and risk of biochemical recurrence or distant metastasis. This study comes after attention has focused on the lack of reproducibility of candidate gene association studies (18, 19). The RAPPER study included 637 patients that received radical prostate radiotherapy and it rigorously assessed the association between toxicity outcomes and 92 SNPs in 46 genes which had been previously reported to be statistically significantly associated with radiation toxicity. The study failed to reproduce any of the findings, but did report borderline statistical significance for the SOD2 rs4880 (20). This current study benefits from having a total of 1,428 patients treated with radiation therapy and is the largest study to date to investigate the relationship between candidate gene polymorphisms and prostate cancer outcome after radiation therapy.

While it is possible that our initial observations of statistically significant associations for SOD2 SNPs and outcomes were due to chance, it is also possible that differences in the study population, follow-up times, available outcomes, and clinical variables may also account for the lack of consistent results in the validation cohort. For example, patients in the validation cohort tended to be younger and to have more intermediate-risk disease than the older, predominantly low-risk patients in the test cohort. Also, the test cohort had substantially longer follow-up than the validation cohort. Androgen deprivation therapy (ADT) was also commonly used in the validation cohort and is estimated to have been used much less often in the test cohort. The endpoints were also different. The test cohort utilized lethal prostate cancer as the outcome, with the majority of events being death from prostate cancer. Distant failure was not used as a separate endpoint in the first cohort due to the low number of verified self-reported events which was potentially due to less use and availability of post-treatment PSA monitoring or radiographic imaging to detect distant metastases compared with the more modern validation cohort. In the validation cohort, biochemical recurrence and distant metastasis were validated via medical records and were deemed the most appropriate as few deaths from cancer had occurred by end of follow-up. Data were not available for local recurrence after radiation therapy due to a lack of consistent screening and reporting of local recurrence in the test cohort. Lastly, many of the prostate biopsies from the test cohort were assigned a Gleason score during an earlier time period than the DFCI cohort. We previously reported that there is an upgrading in Gleason score after modern standardized review of the original biopsy specimens from these cohorts (21), making it challenging to compare the distribution of Gleason scores across the test and validation cohorts.

The SOD2 rs4880 T/C polymorphism has been well studied and postulated to result in increased ability to neutralize ROS due to more efficient uptake into the mitochondrial matrix (22). It has been associated with aggressive prostate cancer incidence among men with low antioxidant nutritional intake (23, 24). However, there are conflicting data regarding the association between rs4880 and toxicity after radiation therapy. Some studies identified an association of rs4880 with increased risk of subcutaneous fibrosis in breast cancer patients that underwent radiation therapy (11) and with grade ≥3 side effects in predominantly breast cancer and head and neck cancer populations (13). Another study by Green et al. refuted the association between SOD2 and radiotherapy complications in breast cancer patients (14). Our study did not find a reproducible association between 7 of the SOD2 SNPs and prostate cancer outcomes, but there was a suggestion of increased survival after radiation therapy for the rs4880 polymorphism and of decreased survival after radiation therapy for the rs6917589 polymorphism.

Since the interaction between SOD2 and the tumor microenvironment is more complex than a single enzymatic reaction, pathway analysis of SNPs may yet detect clinically significant associations by taking into account other key enzymes involved in regulating oxidative stress. For example, as shown in Figure 1, after SOD2 catalyzes the conversion of superoxide anion to hydrogen peroxide, the myeloperoxidase enzyme catalyzes the conversion of hydrogen peroxide to hydrochlorous acid, which is another oxidizing agent that may cause a net effect of increased ROS. Alternatively, catalase and glutathione peroxidase may catalyze conversion of hydrogen peroxide to neutral species. Therefore, the overall effect of SOD2 polymorphisms may be dependent on the activity of MPO, CAT, GPX and other factors that alter the local ROS concentration. The model may also need to take the nutritional status of the patient into account, as our collaborative group has previously reported an interaction between antioxidant status, such as plasma selenium, and a SOD polymorphism as related to the incidence of aggressive prostate cancer (23, 25).

This study benefits from a large sample size and two diverse cohorts to independently assess the association between SOD2 polymorphisms and prostate cancer outcome after radiation therapy. Moreover, in the test cohort, we were able to make comparisons with men who were undergoing radical prostatectomy. A limitation of the study is that we examined only germline polymorphisms and therefore cannot assess the genetic changes within the tumor that may affect tolerance to oxidative stress. We also were not able to directly measure the degree of ROS within the tumor or stroma. Lastly, a pathway analysis may improve the ability to determine the complex interaction between SOD2 polymorphisms and other genes involved in regulating antioxidant stress.

CONCLUSION

In summary, the study shows that most common germline polymorphisms in the SOD2 enzyme are unlikely to have a clinically significant impact on all patient outcome after radiation therapy when treated individually. Though not validated, genetic variants in SOD2 may have an effect that is specific to low-risk prostate cancer patients, and merits further study.