Genetic Susceptibility to Renal Cell Carcinoma: The Role of DNA Double Strand Break Repair Pathway

Abstract

Alterations in DNA repair gene have been shown to cause a reduction in host DNA repair capacity and may influence host susceptibility to carcinogenesis. The double strand break (DSB) repair is a major DNA-repair pathway. This study tested the hypothesis that common sequence variants of the DSB pathway genes predispose susceptible individuals to increased risk of renal cell carcinoma (RCC). Towards this end, we evaluated the associations of 13 single nucleotide polymorphisms (SNP) in 10 candidate genes involved in the DSB pathway with RCC risk in a population-based case-control study that included 326 Caucasian RCC patients and 335 controls. Using the homozygous wild-type as the reference group, we observed a significantly increased RCC risk associated with the homozygous variant genotype of NBS1 (rs1805794; OR 2.13; 95% confidence interval (95% CI), 1.17-3.86). Carrying of at least one copy of the variant XRCC4 allele was also associated with a significantly increased risk (rs1805377; OR, 1.56; 95% CI, 1.08 - 2.26). Importantly, in pathway analysis, compared with the reference group (≤1 adverse alleles), individuals with 2 (OR: 1.26; 95% CI: 0.83-1.91), 3 (OR: 1.00; 95% CI: 0.64-1.56) and more than 3 adverse alleles (OR: 1.75; 95% CI: 1.03-2.98) were at increased risk of RCC with significant association in subjects carrying more than 3 adverse alleles. Results from this study provide evidence that individuals with a higher number of genetic variations in the DBS repair pathway are at an increased risk for RCC. These findings require further validation in independent populations.

Introduction

Renal cell carcinoma (RCC) represents the third leading cause of death among genitourinary malignancies and the twelfth leading cause of cancer death overall. It is estimated that in United States, approximately 51,000 people were diagnosed with RCC in 2007, and roughly one third of these patients will ultimately die from this disease (1).

Epidemiologic studies have suggested that gender, obesity, smoking, analgesic, diuretic abuse, and environmental factors are associated with RCC(2). Cigarette smoking, for example, doubles the risk of RCC and contributes to as many as one-third of all cases, yet only a fraction of smokers and a low number of non-smokers develop RCC, which implies influence of host factors on individual susceptibility(3). These inter-individual differences in susceptibility to RCC may be attributed to genetic polymorphisms in critical genes, including those involved in DNA repair(4).

DNA repair systems play a critical role in protecting the human genome from damage caused by carcinogens present in the environment (5). The double-strand break (DSB) pathway is responsible for repairing double-strand breaks caused by a variety of exposures, including ionizing radiation, free radicals, and telomere dysfunction. There are two distinct and complementary pathways for DSB repair—namely, homologous recombination (HR) and non-homologous end joining (NHEJ)(5). Many genes that encode enzymes involved in DNA repair carry single nucleotide polymorphisms (SNP) with potential to modulate gene function, and associations between RCC risk and variant alleles in different DNA repair genes have been reported(4).

The conventional single-gene–based approach to study the role of genetic variants in carcinogenesis has been fraught with inconsistencies and, sometimes, conflicting data, likely due to the fact that that carcinogenesis is a multigenic process. Consequently, a pathway-based approach, which evaluates the combined effects of a panel of SNPs in the same pathway, may amplify the effects of individual polymorphisms and provide enhanced risk assessment. Such pathway-based multigenic approaches have recently been shown to provide robust risk prediction in several solid organ cancers (6-8). In this context, we investigated thirteen potential functional polymorphisms in ten major DSB repair genes (XRCC3, XRCC2, NBS1, BRCA2, RAG1, KU70, KU80, LIG4, XRCC4 and ATM), and utilized a variety of analytical approaches to identify high-order gene-gene and gene-environment interactions in modulating risk of RCC.

Materials and Methods

Study Population

Beginning in 2002, incident RCC cases were recruited from The University of Texas M. D. Anderson Cancer Center in Houston, Texas. All cases were individuals with newly diagnosed, histologically confirmed RCC. There was no age, gender, ethnicity or cancer stage restrictions on recruitment. M. D. Anderson Cancer Center staff interviewers identified RCC cases through a daily review of computerized appointment schedules for the Departments of Urology and Genitourinary Medical Oncology. Healthy control subjects without a history of cancer, except non-melanoma skin cancer, were identified and recruited using the random digit dialing (RDD) methods. In RDD, randomly selected phone numbers from the household were used to contact potential control volunteers in the same residency of cases accordingly to the telephone directory listings. Controls must have lived in the same county or socio-economically matched surrounding counties that the case resides in for at least one year and have no prior history of cancer. The controls were frequency matched to the cases by age (±5 years), sex, ethnicity and county of residence.

Epidemiologic Data

After informed consent was obtained, all study participants completed a 45-min in-person interview that was administered by M. D. Anderson Cancer Center staff interviewers. The interview elicited information on demographics, smoking history, family history of cancer, occupational history and exposures and medical history. At the conclusion of the interview, a 40-mL blood sample was drawn into coded heparinized tubes and delivered to laboratory for molecular analysis. The study was approved by the Institutional Review Boards of M.D. Anderson Cancer Center. An individual who had smoked at least 100 cigarettes in his or her lifetime was defined as an ever smoker. Ever smokers include former smokers (those who had quit smoking for at least one year), current smokers, and recent quitters (those who had quit within the previous year).

Genotyping

Based on published association studies, we selected 13 potentially functional SNPs from ten DSB genes including 8 SNPs from the HR pathway and 5 SNPs from the NHEJ pathway. The selected SNPs include all published common non-synonymous SNPs (minor allele frequency > 5%) in these genes and a few other potential functional SNPs (5′ UTR, splice site, and one synonymous SNP) that have been investigated in previous cancer association studies.(9, 10) Genomic DNA was isolated from peripheral blood lymphocytes by proteinase K digestion, followed by isopropanol extraction and ethanol precipitations. DNA samples were stored at -80°C. Genotyping was performed using the Taqman real-time polymerase chain reaction method using 7900 HT sequence detector system (Applied Biosystems, Foster City, CA). The primer and probe sequences for each SNP are available on request. In probes, fluorescent dye FAM and VIC were labeled on the 5′ end and a quencher was labeled on the 3′ end. Typical amplification mixes (5 μl) consist of DNA sample (5 ng), 1x TaqMan Buffer A, 200 μm dNTPs, 5 mmol MgCl₂, 0.65 units of AmpliTaq Gold, 900 nmol/l primer each and 200 nmol/L probe each. The thermal cycling conditions included 1 cycle for 10 min at 95°C, and 40 cycles for 15 s at 95°C and 1 min at 60°C. The end point fluorescence was analyzed by SDS version 2.1 software (Applied Biosystems). Water control, internal controls and previously genotyped samples were added into each plate as the calibrator to ensure the accuracy and consistency of the genotyping. Positive and negative controls were used in each genotyping assay, and 5% of the samples were randomly selected and run in duplicates with 100% concordance.

Statistical analysis

The Pearson's χ² test and the Wilcoxon rank sum test were used to test the differences in characteristics between cases and controls. The Hardy-Weinberg equilibrium was tested by the goodness-of- fit χ² test. To evaluate the main effect of individual SNPs, we performed multivariate unconditional logistic regression analysis to estimate odds ratios (ORs) and the 95% confidence intervals (CIs), adjusting for age, sex, smoking status (never or ever smoker) and history of hypertension (yes or no). The SNPs were tested in association with RCC risk in additive, dominant and recessive models. The combined effects of minor alleles were analyzed as a categorical variable by grouping the subjects according to the number of minor alleles in each path way. We treated the minor allele at each locus as the “adverse” allele and tallied the total number of adverse alleles for each individual. For genes with multiple SNPs assayed, only one SNP was included in this pathway analysis. A trend test was performed to test for a linear trend in the ORs. All statistical analyses were two sided. To account for multiple comparisons, we used the false discovery rate (FDR) function based on the Benjamini-Hochberg method (11). We calculated the FDR-adjusted P values at 5%, 10% and 15% levels to assess whether the resulting P values were still significant after adjusting for multiple comparisons. All analyses were performed using the Intercooled Stata 8.0 statistical software package (Stata Co., College Station, TX). To assess RCC risk in the context of multiple genes, we applied a recursive partitioning technique (12). The recursive partitioning was derived from the methodology of Classification and Regression Tree (CART). In CART, a tree based model is created by recursive partitioning the data and enables identify effect modifications between variables that are less visible by traditional logistic regression. The algorithm splits the study sample into a number of homogenous subgroups based on risk factors. The final model is a tree structure with terminal nodes defining a range of risk subgroups. CART analysis was performed using the RPART package in the R software (version 2.5). We calculated ORs for each terminal node adjusting for age, sex, smoking status and history of hypertension.

Results

Subject characteristics

Due to the small number of the minority cases, we restricted our analysis to Caucasians only in this study. There were 326 RCC and 335 matched cancer-free controls available for this analysis (Table 1). Cases and controls were well matched on age (cases versus controls: 59.4 years versus 59.7 years, P = 0.66), gender (cases versus controls: males, 66.5% versus 61.2%; females, 33.5% versus 38.8%, P = 0.16). There were no significant differences in smoking status (cases: never smoker, 48.7%; ever smoker, 46.0%; controls: never smoker, 44.6%, ever smoker, 55.4%, P = 0.09). Among ever smokers, there were no differences in smoking duration, number of cigarettes smoked per day or pack-years of smoking (P=0.87, P=0.51 and P=0.72, respectively). Cases reported a significantly higher prevalence of hypertension than controls (cases versus controls: 57.4% versus 42.8%, P = 0.003).

Table 1.

Characteristics of study subjects^*

	Cases	Controls	P-value
	N=326	N=335
Gender, n (%)
Male	216 (66.5)	205 (61.2)	0.16
Female	109 (33.5)	130 (38.8)
Age, median (SD)	59.4 (10.8)	59.7 (10.8)	0.66
Smoking status, n (%)
Never	158 (48.5)	149 (44.5)	0.09
Ever	150 (46.0)	185 (55.2)
Unknown	18 (5.5)	1 (0.3)
Years of smoking, median (range)	25 (1-58)	24 (1-62)	0.87
Number of cigarettes/day, median (range)	20 (1-80)	20 (1-80)	0.51
Pack-year of smoking, median (range)	22 (0.25-150)	19 (0.25-133)	0.72
Hypertension
Yes	150 (54.74)	143 (42.81)
No	124 (45.26)	191 (57.19)	0.003

^*Total number of each variable may not sum up to 326 cases or 335 controls due to missing information.

Effects of individual polymorphisms on risk of RCC

All tested SNPs were in Hardy-Weinberg equilibrium in controls (data not shown). The overall RCC risks associated with the individual polymorphisms are listed in Table 2. When compared with the homozygous wild-type reference group, a significantly increased RCC risk was observed for the homozygous variant genotype of NBS1 E185Q (OR, 2.13; 95% CI, 1.17-3.86) and for heterozygous variant genotype of XRCC4 rs1805377 (OR, 1.7; 95% CI, 1.15-2.52). In addition, the variant allele of NBS1 E185Q also exhibited a significantly increased risk under a recessive model (homozygous variant genotype compared with wild-type containing genotype: OR, 2.18, 95% CI, 1.23-3.85) and the variant allele of XRCC4 rs1805377 was associated with an increased risk under a dominant model (variant-containing genotype compared with homozygous wild-type genotype: OR, 1.56, 95% CI 1.08-2.26). After adjusting for multiple comparisons at 15% level, the associations with NBS1 E185Q and the XRCC4 rs1805377 heterozygous variant genotype remained significant (Table 2).

Table 2.

Associations of individual polymorphisms and risk of RCC

	Control	Case	OR†	95% CI	P-value		Control	Case	OR†	95% CI	P-value
HR Pathway						NHEJ Pathway
XRCC3 T241M (rs861539)						KU70 G593G (rs132788)
CC	153	139	1			GG	139	144	1
CT	145	150	1.12	0.79 to 1.57	0.52	GT	156	140	0.82	0.58 to 1.16	0.27
TT	30	32	1.15	0.64 to 2.04	0.64	TT	29	31	1.01	0.56 to 1.81	0.97
CC VS. CT & TT			1.12	0.81 to 1.55	0.49	GG VS. GT & TT			0.85	0.61 to 1.18	0.34
CC & CT VS. TT			1.08	0.63 to 1.88	0.77	TT & GT VS. TT			1.11	0.64 to 1.95	0.70
	P for trend‡				0.51		P for trend‡				0.58
XRCC3 5′UTR (rs1799794)						KU80 3′ UTR (rs1051685)
AA	207	197	1			AA	257	246	1
AG	105	99	0.99	0.69 to 1.41	0.96	AG	70	76	1.15	0.78 to 1.70	0.47
GG	15	15	1.19	0.55 to 2.55	0.66	GG	7	2	0.38	0.08 to 1.87	0.23
AA VS. AG & GG			1.01	0.72 to 1.43	0.93	AA VS. AG & GG			1.09	0.74 to 1.58	0.67
AA & AG VS. GG			1.19	0.56 to 2.53	0.65	AA & AG VS. GG			0.37	0.08 to 1.80	0.22
	P for trend‡				0.81		P for trend‡				0.96
XRCC3 A17893G (rs1799796)						LIG4 T9I (rs1805388)
AA	71	44	1			CC	235	214	1
AG	79	33	0.72	0.38 to 1.36	0.31	CT	91	91	1.11	0.77 to 1.59	0.58
GG	19	9	0.47	0.15 to 1.43	0.18	TT	8	15	2.11	0.83 to 5.33	0.12
AA VS. AG & GG			0.67	0.36 to 1.22	0.19	CC VS. CT & TT			1.19	0.84 to 1.68	0.33
AA & AG VS. GG			0.55	0.19 to 1.61	0.27	CC & CT VS. TT			2.05	0.81 to 5.16	0.13
	P for trend‡				0.13		P for trend‡				0.18
XRCC2 R188H (rs3218536)						XRCC4 Splice Site l (rs1805377)
GG	287	275	1			GG	262	229	1
GA	38	32	0.94	0.56 to 1.57	0.80	GA	58	82	1.7	1.15 to 2.52	0.01*
AA	1	2	2.73	0.24 to 30.73	0.42	AA	13	12	0.95	0.41 to 2.23	0.91
GG VS. GA & AA			0.98	0.59 to 1.62	0.93	GG VS. GA & AA			1.56	1.08 to 2.26	0.02
GG & GA VS. AA			2.75	0.24 to 30.93	0.41	GG & GA VS. AA			0.84	0.36 to 1.96	0.69
	P for trend‡				0.92		P for trend‡				0.08
XRCC2 3′UTR C/T at Nucleotide 41657						ATM D1853N (rs1801516)
CC	301	292	1			GG	249	254	1
CT	30	33	1.1	0.63 to 1.94	0.73	GA	81	64	0.74	0.50 to 1.09	0.13
TT	2	0				AA	5	5	0.49	0.09 to 2.55	0.39
CC VS. CT & TT			1.03	0.59 to 1.79	0.93	GG VS. GA & AA			0.72	0.49 to 1.07	0.10
						GG & GA VS. AA			0.72	0.49 to 1.07	0.10
							P for trend‡				0.08
NBS1 E185Q (rs1805794)
GG	152	137	1
GC	160	142	0.96	0.68 to 1.34	0.80
CC	21	43	2.13	1.17 to 3.86	0.01*
GG VS. GC & CC			1.09	0.79 to 1.51	0.60
GG & GC VS. CC			2.18	1.23 to 3.85	0.01*
	P for trend‡				0.10
BRCA2 N372H (rs144814)
GG	169	157	1
GC	141	139	0.98	0.70 to 1.38	0.91
CC	19	25	1.47	0.77 to 2.83	0.25
GG VS. GC & CC			1.04	0.75 to 1.44	0.81
GG & GC VS. CC			1.48	0.79 to 2.80	0.22
	P for trend‡				0.49
RAG1 K820R (rs2227973)
AA	257	246	1
AG	68	70	1.13	0.76 to 1.67	0.55
GG	9	7	0.82	0.28 to 2.35	0.71
AA VS. AG & GG			1.09	0.75 to 1.59	0.65
AA & AG VS. GG			0.8	0.28 to 2.29	0.67
	P for trend‡				0.80

^*remained significant at 15% level after FDR adjustment for multiple comparisons.

^‡P-for trend was calculated for the additive model

^†ORs were adjusted for age, sex, smoking status and history of hypertension

When stratified by smoking status, in ever smokers, we observed a significant 1.76-fold increased risk for variant-containing genotype of XRCC4 rs1805377 in the dominant model (OR, 1.76, 95% CI, 1.00-3.11). The homozygous variant genotype of NBS1 E185Q was associated with 3.45-fold increased risk (95% CI, 1.47-8.11). None of the associations were observed in never smokers (data not shown).

Combined effects of multiple SNPs

To test the hypothesis that multiple SNPs in the same pathway may have an additive effect on RCC risk, we estimated the combined effect of these SNPs, and stratified the analyses by host characteristics (Table 3). The same combination of genes/SNPs was used in the overall analysis and in the stratified analysis. For those genes with multiple SNPs assayed, only one SNP was included in this combined analysis and others were excluded because of linkage disequilibrium.

Table 3.

HR, NHEJ and all DSB pathways and RCC risk

HR Pathway*			NHEJ Pathway‡			DSB Pathway
	Case/control	OR (95% CI)		Case/control	OR (95% CI)		Case/control	OR (95% CI)
Overall
0 and 1	86/106	Ref.	0 and 1	128/141	Ref.	≤3	116/126	Ref.
2	101/95	1.26 (0.83-1.91)	2	95/103	1.03 (0.70-1.51)	4	60/70	0.85 (0.54-1.34)
3	75/82	1.00 (0.64-1.56)	3	52/55	1.00 (0.62-1.60)	5	59/61	0.96 (0.61-1.53)
>3	53/36	1.75 (1.03-2.98)	>3	33/22	1.60 (0.86-2.97)	>5	64/49	1.39 (0.87-2.22)
P-trend		0.15	P-trend		0.3	P-trend		0.25
Smoking
Ever Smoker			Ever Smoker			Ever Smoker
0 and 1	37/57	Ref.	0 and 1	58/82	Ref.	≤3	48/80	Ref.
2	45/61	1.09 (0.62-1.94)	2	45/59	1.06 (0.63-1.79)	4	31/33	1.57 (0.85-2.88)
3	35/39	1.38 (0.73-2.59)	3	25/24	1.48 (0.76-2.89)	5	29/27	1.81 (0.96-3.43)
>3	28/19	2.31 (1.12-4.76)	>3	15/11	1.88 (0.80-4.44)	>5	31/28	1.88 (1.00-3.54)
P-trend		0.02	P-trend		0.1	P-trend		0.03
Never Smoker			Never Smoker			Never Smoker
0 and 1	38/49	Ref.	0 and 1	63/59	Ref.	≤3	58/46	Ref.
2	53/34	1.99 (1.08-3.64)	2	44/43	0.93 (0.53-1.61)	4	27/37	0.58 (0.31-1.08)
3	38/42	1.14 (0.62-2.11)	3	26/31	0.75 (0.40-1.43)	5	29/33	0.70 (0.37-1.31)
>3	24/17	1.79 (0.84-3.81)	>3	18/11	1.55 (0.67-3.57)	>5	32/21	1.18 (0.60-2.32)
P-trend		0.34	P-trend		0.79	P-trend		0.89
Gender
Male			Male			Male
0 and 1	61/61	Ref.	0 and 1	88/82	Ref.	≤3	85/71	Ref.
2	69/60	1.08 (0.65-1.82)	2	68/65	0.97 (0.60-1.55)	4	40/47	0.64 (0.37-1.13)
3	50/53	0.88 (0.51-1.52)	3	34/34	0.89 (0.49-1.60)	5	37/37	0.77 (0.43-1.38)
>3	30/23	1.18 (0.60-2.32)	>3	18/16	0.96 (0.44-2.11)	>5	40/35	0.93 (0.52-1.66)
P-trend		0.95	P-trend		0.77	P-trend		0.69
Female			Female			Female
0 and 1	25/45	Ref.	0 and 1	40/59	Ref.	≤3	31/55	Ref.
2	32/35	1.67 (0.81-3.43)	2	27/38	1.17 (0.59-2.30)	4	20/23	1.49 (0.67-3.32)
3	25/29	1.31 (0.60-2.89)	3	17/21	1.28 (0.57-2.89)	5	22/24	1.46 (0.67-3.20)
>3	23/13	3.48 (1.44-8.40)	>3	15/6	3.69 (1.27-10.43)	>5	24/14	3.22 (1.40-7.42)
P-trend		0.02	P-trend		0.04	P-trend		0.01
Age
<59			<59			<59
0 and 1	43/41	Ref.	0 and 1	68/52	Ref.	≤3	63/45	Ref.
2	49/47	0.89 (0.49-1.63)	2	43/55	0.59 (0.34-1.02)	4	26/38	0.49 (0.25-0.94)
3	29/40	0.57 (0.29-1.12)	3	22/25	0.65 (0.32-1.32)	5	21/31	0.41 (0.20-0.84)
>3	29/16	1.64 (0.77-3.53)	>3	13/10	0.93 (0.36-2.38)	>5	33/24	0.93 (0.48-1.83)
P-trend		0.74	P-trend		0.35	P-trend		0.42
≥59			≥59			≥59
0 and 1	43/65	Ref.	0 and 1	60/89	Ref.	≤3	53/81	Ref.
2	52/48	1.73 (0.96-3.10)	2	52/48	1.75 (1.01-3.03)	4	34/32	1.42 (0.74-2.72)
3	46/42	1.66 (0.90-3.05)	3	30/30	1.58 (0.82-3.05)	5	38/30	1.97 (1.05-3.68)
>3	24/20	1.76 (0.83-3.76)	>3	20/12	2.60 (1.13-5.99)	>5	31/25	2.02 (1.04-3.92)
P-trend		0.09	P-trend		0.02	P-trend		0.01
HR Pathway			NHEJ Pathway			DSB Pathway
	Case/control	OR† (95% CI)		Case/control	OR† (95% CI)		Case/control	OR† (95% CI)
History of Hypertension
Yes			Yes			Yes
0 and 1	32/53	Ref.	0 and 1	57/62	Ref.	≤3	49/62	Ref.
2	54/36	2.71 (1.43-5.13)	2	44/45	1.01 (0.57-1.79)	4	26/33	0.91 (0.47-1.78)
3	35/35	1.87 (0.96-3.65)	3	22/24	0.92 (0.46-1.87)	5	30/22	1.72 (0.87-3.41)
>3	23/13	2.96 (1.27-6.88)	>3	18/8	2.39 (0.94-6.04)	>5	31/16	2.31 (1.11-4.78)
P-trend		0.02	P-trend		0.22	P-trend		0.01
No			No			No
0 and 1	38/53	Ref.	0 and 1	49/79	Ref.	≤3	45/64	Ref.
2	34/59	0.78 (0.42-1.44)	2	35/57	1.02 (0.59-1.79)	4	25/37	0.91 (0.47-1.74)
3	25/46	0.70 90.36-1.36)	3	25/31	1.14 (0.58-2.21)	5	21/38	0.69 (0.35-1.37)
>3	23/23	1.42 (0.69-2.93)	>3	11/14	1.22 (0.49-3.00)	>5	25/33	1.08 (0.56-2.09)
P-trend		0.65	P-trend		0.61	P-trend		0.9

^*HR pathway genes include XRCC3, XRCC2, NBS1, BRCA2, and RAG1

^‡NHEJ pathway genes include KU70, KU80, LIG4, XRCC4 and ATM

^†ORs were adjusted for age, sex, smoking status and history of hypertension

In the HR pathway, compared with the reference group (≤1 adverse alleles), individuals with 2 (OR: 1.26; 95% CI: 0.83-1.91), 3 (OR: 1.00; 95% CI: 0.64-1.56) and more than 3 adverse alleles (OR: 1.75; 95% CI: 1.03-2.98) were at increased risk of RCC with significant association in subjects carrying more than 3 adverse alleles (Table 3). Stratified analysis showed that a significant gene-dosage trend was evident in ever smokers, hypertensive and female subjects. Compared with the reference group (0 and 1 adverse alleles), ever smokers with 2, 3 and >3 adverse alleles had ORs that increased to 1.09 (95% CI 0.62-1.94), 1.38 (95% CI 0.73-2.59), and 2.31 (95% CI 1.12-4.76), respectively (P for trend = 0.02) (Table 3). Female individuals, with 2, 3 and >3 adverse HR alleles, had ORs of 1.67 (95% CI 0.81-3.43), 1.31 (95% CI 0.60-2.89), and 3.48 (95% CI 1.44-8.40), respectively (P for trend = 0.02), compared to the reference group. Similarly, subjects with history of hypertension had ORs of 2.71 (95% CI 1.43-5.13), 1.87 (95% CI 0.96-3.65), and 2.96 (95% CI 1.27-6.88) when 2, 3 and >3 adverse HR alleles were respectively present (P for trend = 0.02) (Table 3).

For the NHEJ pathway, a statistically significant cumulative gene-dosage trend was observed in female subjects and individuals older than 59 (median age of the control subjects) years of age. Females, with 2, 3 and >3 adverse NHEJ alleles, had ORs of 1.17 (95% CI 0.59-2.30), 1.28 (95% CI 0.57-2.89), and 3.69 (95% CI 1.27-10.43), respectively (P for trend = 0.04). Likewise, subjects older than 59 years with 2, 3 and >3 adverse NHEJ pathway alleles had ORs that increased to 1.75 (95% CI 1.01-3.03), 1.58 (95% CI 0.82-3.05), and 2.60 (95% CI 1.13-5.99), respectively (P for trend = 0.02) (Table 3).

Finally, when all the DSB pathway genes were combined, compared with the reference group of ≤3 adverse alleles, hypertensive, ever-smokers, older and female subjects demonstrated significant gene-dosage trends in RCC risk (Table 3).

CART Analysis

Figure 1 depicts the tree structure generated using the CART analysis, which included all investigated genetic variants of the DSB pathway. The final tree structure contained 7 terminal nodes as defined by SNPs of the DBS pathway. The terminal nodes representing a range of low vs. high risk subgroups. The NBS1 genotype was singled out in the first splitting node, separating individuals with the wild-type-containing genotypes (low risk) from subjects with the homozygous variant genotype (high risk). Individuals with the variant genotypes of both ATM and BRCA2 exhibited the lowest RCC risk with a 32% case rate. Using this terminal node as the reference, the ORs of the other 5 terminal nodes ranged from 1.16 to 3.59. Subjects with the variant genotype of NBS1 exhibited a 3.59-fold (95% CI 1.67-7.72) increased risk of RCC (Figure 1).

Discussion

In this study, we used a polygenic approach to investigate the combined effects of 13 common potentially functional variants in ten DSB repair genes on RCC risk. Multivariate logistic regression revealed that the NBS1 E185Q and XRCC4 splicing variant allele exhibited a statistically significant association with the risk of RCC. More importantly, the combined analysis of multiple SNPs showed an increasing risk of RCC with an increasing number of adverse alleles. The results of the current study are in concordance with our previous report, where utilizing a comet assay, a significantly higher level of DNA damage at baseline and after mutagen induction was seen in RCC patients compared with control subjects, suggesting a functional correlation between DNA damage repair capacity and risk of RCC (13). To the best of our knowledge, this is the first study to examine associations between a panel of DSB repair SNPs and RCC risk.

Main effects on the risk of RCC were observed for the NBS1 E185Q SNP in subjects with the homozygous variant genotype, which exhibited a 2.17-fold increased risk of RCC. Homozygous germline mutations of the NBS1 gene lead to the Nijmegen breakage syndrome, a rare autosomal recessive disease characterized by microcephaly, growth and mental retardation, radiosensitivity, immunodeficiency, high incidence of malignancies at an early age, and elevated rates of chromosomal abnormalities (14). Nibrin, the protein encoded by NBS1, is part of a nuclear multiprotein complex that also contains the DNA repair proteins Mre11 and Rad50. Upon irradiation, this complex redistributes within the nucleus, forming foci that have been implicated as sites of DNA repair. Distinct domains of nibrin are required for focus formation, nuclear localization, and functional interactions with other DNA repair proteins (15). In accordance, the variant allele of the NBS1 E185Q SNP was also associated with altered RCC risk under a recessive model, consistent with the role as a tumor suppressor gene in an autosomal recessive genetic disease. This SNP has also been previously shown to be associated with increased risk of skin and lung cancer; however, the exact molecular mechanism remains to be determined (16, 17). The XRCC4 gene is necessary for DNA ligation in NHEJ, and XRCC4 variant allele in intron 7 (rs1805377) may have functional significance since the nucleotide change from G to A potentially abolishes an acceptor splice site at exon 8 (18, 19). In a recent study of 696 bladder cancer cases and 629 controls, XRCC4 variant allele in intron 7 was associated with an increased risk for bladder cancer (20). Likewise, we observed a significant increase in risk for RCC among carriers of the variant allele compared to common homozygotes (OR, 1.6; 95% CI, 1.08 - 2.26). However, these findings need to be replicated in additional study populations, since homozygote variants were rare in the control populations, and did not show significant associations with risk. Further, after adjustment for multiple comparisons, the associations were only significant at 15% FDR level.

Interestingly, we also noticed that consistent with the main effect logistic regression analysis, the NBS1 E185Q and XRCC4 splicing polymorphisms were also identified in the CART analysis as important genetic variants modulating RCC risk. NBS1 E185Q was the most prominent SNP to discriminate between cases and controls. Moreover, subsets of individuals with higher cancer risks were identified through CART modeling, based on differential combinations of DSB genotypes. The CART analysis used in this study demonstrates proof-of-principle ability for rapid identification of potential gene-gene interactions when dealing with a large number of variables in a complex disease process, such as cancer.

Since we observed only modest effects of individual variant DSB genotypes, they are expected to have limited practical value in predicting RCC risk, further supporting the role of a multigenic approach over the single candidate-gene approach. In concordance with this hypothesis, we found a significant trend of increased RCC risk with increasing numbers of adverse alleles in the HR pathway, NHEJ pathway and the entire DSB pathway. The gene-dosage effect of all the DSB pathway genes, observed in this study, was restricted to specific sub-populations, namely subjects with smoking history, females, older individuals and subjects with a history of hypertension.

We observed that carrying more than 5 adverse alleles of the entire DSB pathway was associated with significant increased risk of RCC in ever smokers, but not in never smokers. Oxidative stress due to cigarette smoking can induce oxidative DNA damage and DSB, and our data indicates that smokers with less efficient DSB capacity are more likely to develop RCC (21, 22). This finding is well supported by several other studies, which have reported similar interactions between DNA repair capacity and dose or duration of smoking (20, 23). Although our data indicates that smoking may modify the effects of all the DSB pathway genes examined, we observed no statistically significant interactions between smoking and individual SNPs. The moderate sample size in this study may not allow sufficient statistical power to detect gene-environment interactions.

We also observed that carrying more than 5 adverse alleles of DSB pathway was at significantly increased risk of RCC in older subjects (59 years or older) but not in younger subjects (less than 59 years old). The observation of significant DSB gene-dosage effect in older individuals is well supported by a large body of evidence linking DNA damage accumulation with age in mammals (24). The precise mechanism of cancer susceptibility in the elderly is not well understood, but immune function and DNA repair efficiency have been shown to decrease with age, which reduces protection against environmental carcinogens (22). Thus, there is emerging belief that the intrinsic fidelity of DNA repair mechanisms, such as DSB, may influence age-associated cancer risk.

Also, women carrying more than 5 adverse alleles in the DSB pathway were at increased risk, while the association was not observed in men. Our finding of a significantly increased risk for RCC in women carrying increased numbers of the variant DBS alleles is in line with evidence suggesting that women are inherently more susceptible to certain carcinogens than men (25). A number of epidemiologic studies, for example, indicate that women smokers are 1.7 to 3 times as likely as male smokers to develop lung cancer, given the same amount of exposure (25). Possible mechanisms that may underlie the enhanced susceptibility of women are greater activity of CYP P450 enzymes, enhanced formation of DNA adducts and P53 mutations, and hormonal effects on tumor promotion (26).

Finally, significant DBS gene-dosage effect observed in hypertensive individuals is particularly intriguing, but biologically plausible. Although, most epidemiologic studies have not been able to distinguish the effects of hypertension from those of diuretics or other antihypertensive drugs on the risk of RCC, a variety of angiogenic and other growth factors, the levels of which are increased in persons with hypertensive disease, may be involved in renal carcinogenesis (27). Conversely, only a small fraction of hypertensive patients develop RCC, and our data suggest that decreased efficacy of DNA repair machinery may place these individuals at an incrementally increased risk of RCC.

Several potential limitations of this study merit discussion. First, given the small sample size in some of the terminal nodes and in stratified analysis, the results should be interpreted with caution. Second, the inclusion of SNPs was based on potential functional SNPs in genes with higher possibilities of being related to cancer risk and a more comprehensive approach including tagging SNPs would provide more convincing evidence for the associations.

In summary, our study is one of the first to use a polygenic strategy to evaluate the involvement of DSB polymorphisms in RCC. We found that, in the context of specific host characteristics, variations in genes responsible for double strand DNA break repair may influence susceptibility to renal cancer.