To the editor:

A recent Blood paper by Hill et al1  suggested a link between the risk of developing non-Hodgkin lymphoma (NHL) and polymorphic variations in several genes involved in DNA strand break induction and DNA repair pathways. We examined 2 of these polymorphisms where Hill et al reported statistically significant associations, in the Recombination-activating 1 (RAG1) gene (K820R; odds ratio [OR] = 2.7 [AA vs GG], 95% confidence interval [CI], 1.4-5.0) and the Breast cancer 2 (BRCA2) gene (N372H; OR = 1.5 [AA vs CC], 95% CI, 1.0-2.1), in a United Kingdom case-control study of lymphoma.

Local ethical approval was obtained for this research. Informed consent was obtained in accordance with the Declaration of Helsinki.

A total of 741 white patients with NHL and 806 white control participants randomly selected from population registers were recruited as part of a lymphoma case-control study conducted in parts of north and southwest England.2  Our study participants consented to give a blood sample, and represented 67% of patients with incident NHL aged 16 to 69 years in the study area and 66% of the control participants who were contacted successfully. We genotyped the participants' DNA while blinded to case-control status using TaqMan SNP genotyping assays (Applied Biosystems, Warrington, United Kingdom) and ABI PRISM 7000 software (Applied Biosystems). To assess the reliability of the Taqman assays, 20 samples were randomly selected for direct DNA sequencing: concordance with TaqMan was 100% for both polymorphisms. Among our control population, RAG1 K820R and BRCA2 N372H were in Hardy-Weinberg equilibrium, and the distributions of variants were consistent with those reported by Hill et al and the cancer genome anatomy project SNP500 cancer database.3  Unmatched statistical analyses to estimate ORs and 95% CIs were conducted using unconditional logistic regression adjusting for age and sex,4  and the likelihood ratio test was used to test for interactions.

We observed a modest but nonsignificant decrease in risk of NHL with BRCA2 N372H AC and CC variants compared with the AA variant, and no evidence of any association between NHL and the RAG1 K820R polymorphism (Table 1). Risks for B-cell lymphoma, and the 2 major diagnostic subgroups of diffuse large B-cell lymphoma and follicular lymphoma (data not shown), were similar to those for total NHL for both variants, while risk estimates were elevated for T-cell lymphoma, but were not significant (Table 1). Risks were generally similar among men and women, and varied little by age (data not shown). However, for T-cell lymphoma, men who carried at least 1 copy of the BRCA2 N372H C allele had a significantly increased risk (OR = 2.30, 95% CI, 1.08-4.87 based on 23 patients and 194 control participants with the C allele compared with 11 patients and 208 control participants), while women were at decreased risk (OR = 0.43, 95% CI, 0.17-1.06 based on 7 patients and 176 control participants with the C allele compared with 17 patients and 179 control participants) (test for interaction: χ2 = 8.42, P = .004).

Table 1

Number of patients and control participants, adjusted ORs, and 95% CIs for BRCA2 N372H and RAG1 K820R by immunophenotype

Genes and variantsControls, no. (%)NHL
B-cell lymphoma
T-cell lymphoma
Patients, no. (%)OR (95% CI)Patients, no. (%)OR (95% CI)Patients, no. (%)OR (95% CI)
Total 806 (100) 741 (100) NA 665 (100) NA 59 (100) NA 
BRCA2 N372H, A > C*        
    AA 387 (51.1) 375 (55.5) 1 (reference) 340 (56.4) 1 (reference) 28 (48.3) 1 (reference) 
    AC 307 (40.6) 253 (37.4) 0.85 (0.68-1.05) 224 (37.1) 0.83 (0.66-1.03) 22 (37.9) 1.03 (0.58-1.85) 
    CC 63 (8.3) 48 (7.1) 0.79 (0.53-1.18) 39 (6.5) 0.71 (0.46-1.09) 8 (13.8) 1.61 (0.70-3.75) 
    AC + CC 370 (48.9) 301 (44.5) 0.84 (0.68-1.03) 263 (43.6) 0.81 (0.65-1.00) 30 (51.7) 1.14 (0.67-1.96) 
RAG1 K820R, A > G       
    AA 622 (81.4) 557 (81.2) 1 (reference) 496 (81.3) 1 (reference) 46 (78.0) 1 (reference) 
    AG 136 (17.8) 124 (18.1) 1.01 (0.77-1.33) 109 (17.9) 1.00 (0.76-1.32) 13 (22.0) 1.38 (0.72-2.65) 
    GG 6 (0.8) 5 (0.7) 0.94 (0.29-3.10) 5 (0.8) 1.07 (0.32-3.52) 0 (0.0) 0 (0-8.84) 
    AG + GG 142 (18.6) 129 (18.8) 1.01 (0.78-1.33) 114 (18.7) 1.00 (0.76-1.32) 13 (22.0) 1.31 (0.69-2.51) 
Genes and variantsControls, no. (%)NHL
B-cell lymphoma
T-cell lymphoma
Patients, no. (%)OR (95% CI)Patients, no. (%)OR (95% CI)Patients, no. (%)OR (95% CI)
Total 806 (100) 741 (100) NA 665 (100) NA 59 (100) NA 
BRCA2 N372H, A > C*        
    AA 387 (51.1) 375 (55.5) 1 (reference) 340 (56.4) 1 (reference) 28 (48.3) 1 (reference) 
    AC 307 (40.6) 253 (37.4) 0.85 (0.68-1.05) 224 (37.1) 0.83 (0.66-1.03) 22 (37.9) 1.03 (0.58-1.85) 
    CC 63 (8.3) 48 (7.1) 0.79 (0.53-1.18) 39 (6.5) 0.71 (0.46-1.09) 8 (13.8) 1.61 (0.70-3.75) 
    AC + CC 370 (48.9) 301 (44.5) 0.84 (0.68-1.03) 263 (43.6) 0.81 (0.65-1.00) 30 (51.7) 1.14 (0.67-1.96) 
RAG1 K820R, A > G       
    AA 622 (81.4) 557 (81.2) 1 (reference) 496 (81.3) 1 (reference) 46 (78.0) 1 (reference) 
    AG 136 (17.8) 124 (18.1) 1.01 (0.77-1.33) 109 (17.9) 1.00 (0.76-1.32) 13 (22.0) 1.38 (0.72-2.65) 
    GG 6 (0.8) 5 (0.7) 0.94 (0.29-3.10) 5 (0.8) 1.07 (0.32-3.52) 0 (0.0) 0 (0-8.84) 
    AG + GG 142 (18.6) 129 (18.8) 1.01 (0.78-1.33) 114 (18.7) 1.00 (0.76-1.32) 13 (22.0) 1.31 (0.69-2.51) 

B-cell lymphoma includes International Classification of Diseases for Oncology version 3 codes 9673/3, 9679/3, 9680/3, 9684/3, 9689/3, 9690/3, 9691/3, 9695/3, 9698/3, and 9699/3; T-cell lymphoma includes 9700/3, 9701/3, 9702/3, 9705/3, 9708/3, 9709/3, 9714/3, 9716/3, 9717/3, 9718/3, 9719/3, and 9827/3.

Immunophenotype was not known for 17 patients. ORs and 95% CIs were calculated using unconditional logistic regression adjusting for age and sex.

NA indicates not applicable.

*

BRCA2 N372H and RAG1 K820R were in Hardy-Weinberg equilibrium among controls (χ2 = 0.04, P = 0.85; χ2 = 0.23, P = 0.63, respectively). Samples did not amplify for 65 patients and 49 control participants for BRCA2 N372H, and 55 patients and 42 control participants for RAG1 K820R.

OR and 95% CI were calculated using exact methods.

Our data suggest that there is little association between NHL and either of these 2 polymorphisms. However, we cannot exclude the possibility of an association, and further investigation is required using larger datasets to elucidate the role of these polymorphisms in determining the risk of developing NHL.

Correspondence: Kathryn Scott, Department of Biology, University of York, York. United Kingdom YO10 5YW; e-mail: KS27@YORK.AC.UK

Conflict-of-interest disclosure: The authors declare no competing financial interests.

1
Hill
 
DA
Wang
 
SS
Cerhan
 
JR
et al. 
Risk of Non-Hodgkin lymphoma (NHL) in relation to germline variation in DNA repair and related genes.
Blood
2006
, vol. 
108
 (pg. 
3161
-
3167
)
2
Willett
 
EV
Smith
 
AG
Dovey
 
GJ
et al. 
Tobacco and alcohol consumption and the risk of non-Hodgkin lymphoma.
Cancer Causes Control
2004
, vol. 
15
 (pg. 
771
-
780
)
3
National Cancer Institute
Cancer Genome Anatomy Project SNP500 Cancer Database.
Accessed June 2006. 
4
Breslow
 
NE
Day
 
NE
Classical methods of analysis of matched data. Statistical methods in cancer research: the analysis of case-control studies volume 1.
1980
Lyon
International Agency for Research in Cancer
(pg. 
162
-
189
)
Sign in via your Institution