EPIDEMIOLOGY

Meta-analysis of two ERCC2 (XPD) polymorphisms, Asp312Asnand Lys751Gln, in breast cancer

Noel Pabalan • Ofelia Francisco-Pabalan •

Lillian Sung • Hamdi Jarjanazi • Hilmi Ozcelik

Received: 17 March 2010 / Accepted: 19 March 2010

� Springer Science+Business Media, LLC. 2010

Abstract The excision repair cross-complementing group

2 gene (ERCC2) plays a key role in DNA repair. Several

polymorphisms in the ERCC2 gene have been described,

including the commonly occurring Lys751Gln and

Asp312Asn polymorphisms. Studies investigating the asso-

ciation of these polymorphisms with breast cancer risk pro-

duced controversial results. To evaluate these associations

presented in diverse populations, we have conducted a meta-

analysis based on 40 studies from 33 publications in PubMed

which included analyses of Lys751Gln (14,545 cases,

15,352 controls) and Asp312Asn polymorphisms (16,254

cases, 14,006 controls). Overall findings of both polymor-

phisms have implicated null effects (OR = 1.01–1.03) when

the analyses were limited to the statistically powerful

(C80%) studies. Although modestly increased statistically

significant breast cancer risk was detected in the

underpowered studies (B80%), removal of outliers resulted

in null associations. Ethnic stratification showed non-sig-

nificant and relatively null associations for both polymor-

phisms with breast cancer risk for the overall Caucasians as

well as North American and the European sub-populations.

Although statistically increased and decreased risks were

observed for the homogenous populations of African-

Americans (Lys751Gln, OR 1.25, 95% CI 1.03–1.53,

P = 0.03) and Asians (Asp312Asn, ORs: 0.53–0.55, P val-

ues: 0.02–0.03), respectively, this may be the result of small

sample size. Analyses of the homogeneous adduct studies,

with relatively large sample size, exhibited increased risk for

Lys751Gln (OR 1.20, 95% CI (1.02–1.41), P = 0.03) and

Asp312Asn (OR 1.17 95% CI 1.02–1.34, P = 0.03) under

the dominant genetic model. In conclusion, our results sug-

gest null associations of both polymorphisms in the overall

and the Caucasian subgroups, although some effects can be

suggested for relatively smaller minority studies. Increased

risk effect was more visible when the adduct studies are

considered, suggesting the role of these polymorphisms in

the presence of exposure to DNA damaging agents.

Keywords Breast cancer � ERCC2 � XPD � Asp312Asn �Lys751Gln � Adducts

Introduction

The excision repair cross-complementing group 2 (ERCC2)

or the xeroderma pigmentosum complementary group D

(XPD) is a DNA repair gene, which encodes an ATP-

dependent DNA helicase. It is involved in separating the

double helix at lesion sites in the nucleotide excision repair

pathway (NER), and causes Xeroderma Pigmentosum when

mutated in the germ line [1]. ERCC2 is a component of the

N. Pabalan

College of Natural Sciences, Saint Louis University,

Baguio City 2600, Philippines

O. Francisco-Pabalan

Analytical Genetics Technology Centre, Princess Margaret

Hospital, University Health Network, Toronto, ON, Canada

L. Sung

Division of Hematology/Oncology, Hospital for Sick Children,

Toronto, ON, Canada

H. Jarjanazi

Environmental Monitoring and Reporting Branch, Ontario

Ministry of the Environment, Toronto, ON, Canada

H. Ozcelik (&)

Fred A. Litwin Centre for Cancer Genetics, Samuel Lunenfeld

Research Institute, Mount Sinai Hospital, 60 Murray St., Room

L6-304, Box 29, Toronto, ON M5T 3L9, Canada

e-mail: ozcelik@lunenfeld.ca

123

Breast Cancer Res Treat

DOI 10.1007/s10549-010-0863-6

transcription factor complex, TFIIH, which participates in

both NER and basal transcription [2]. Several polymor-

phisms have been identified in the coding region of ERCC2,

whereas the two commonly occurring, Asp312Asn and

Lys751Gln, have been studied most extensively [3]. The

XPD Asp312Asn polymorphism (rs1799793) is character-

ized by a G to A substitution resulting in an aspartic acid

(Asp [D] 3) to asparagine (Asn [N]) amino acid transition at

codon 312, whereas the Lys751Gln polymorphism

(rs13181) is characterized by an A to C substitution causing

a lysine (Lys [K] 3) to glutamine [Gln (Q)]) amino acid

exchange at codon 751.

Many studies have investigated the functional properties

of these two ERCC2 variants; however, the findings have

been controversial. The two polymorphisms have shown

little or no effect on the protein using predictive models [4]

or evolutionary analysis [5]. Structural evidence indicated

that these polymorphisms are located outside the main

catalytic sites [6] and far from regulatory [7] and interacting

domains [8] suggesting that they have no direct effect on the

ATPase activity of ERCC2. Additional studies detected no

measurable effect of the two variants on NER capacity and

basal transcription activation [2], and genotype-specific

differences in DNA repair rates was inconsistent [9]. On the

other hand, a study has shown that homozygous genotypes

of both variants, 312 Asn-Asn (AA) and 751 Gln-Gln (CC),

were associated with defective repair of ultraviolet light-

induced DNA damage [10]. Although, beset with sample

size and replication issues [5], other studies also showed

that the UV induced DNA repair capacity was also reduced

in subjects with homozygous 312 Asn–Asn (AA) or 751

Gln–Gln (CC) genotypes compared with the respective

homozygous wild-type genotypes [11]. Another study with

similar methodology issues have shown that homozygous

751 Lys–Lys (AA) genotype was associated with reduced

repair of X-ray induced DNA damage [12]. In another

study, apoptotic response to irradiation was observed in the

variant allele of 312 Asn (A) but not in any allelic combi-

nation of 751 [13]. Interestingly, Wolfe et al. [14] has

demonstrated that the two polymorphisms significantly

reduced constitutive ERCC2 mRNA levels (312N:

P \ 0.0004; 751Q: P \ 0.002) in lymphocytes of healthy

subjects and that this decrease was significantly greater in

smokers, exacerbated by smoking duration and intensity.

In addition to the controversies surrounding functional

studies, accumulating evidence from epidemiological

studies on the association of these polymorphisms and

breast cancer has also been conflicting. The increasing

number of such studies prompted us to examine all related

published literature. Adhering to an established framework

[15] we aim to clarify the effect of variation in the ERCC2

Asp312Asn and Lys751Gln polymorphisms on breast

cancer risk.

Materials and methods

Selection of studies

Using PubMed, we identified all published case–control

studies, written in English which investigated the association

of the two ERCC2 polymorphisms and breast cancer risk.

The terms and their combinations used for search included:

‘‘ERCC2’’, ‘‘XPD’’, ‘‘751C’’, ‘‘312A’’, ‘‘breast cancer’’ and

‘‘polymorphism’’, ‘‘Asp312Asn’’, ‘‘Lys751Gln’’, ‘‘D312N’’,

‘‘K751Q’’, ‘‘rs1799793’’, and ‘‘rs 13181’’. Additional studies

were manually searched in the reference list of all studies

identified. Eligible studies had genotypic data with a case–

control design and duplicates of previous studies were

excluded [16]. As a result, we have identified a total of 33

eligible articles, which investigated the association of one or

both polymorphisms, Asp312Asn and Lys751Gln, with

breast cancer risk where 29 studied a single ethnic popula-

tion. Three [17–19] and one articles [20] focused on two and

three ethnic populations, respectively (Table 1). Treated as

separate data were matched and unmatched case controls in

one article [21] and familial and sporadic data in another [22].

Thus, here we have presented the various population studies

of 33 publications in 40 separate studies or populations.

Of the 40 studies, 32 investigated Caucasian populations

including 17 European [17, 20, 21, 23–32] and 13 North

American subgroups [18, 19, 33–43]. To obtain geo-

graphical homogeneity of these ethnicities, the Brazilian

[44] and Australian [20] studies were excluded from these

subgroupings. Six studies were Asian [22, 45–48] and two

were African-American [18, 19].

Data extraction and power calculations

Two investigators independently extracted and revised

eligibility of data. For each study, we abstracted the first

author’s name, year of publication, country and ethnicity of

the study populations, genotype data as well as number of

cases and controls. Departures of genotypic frequencies

from the Hardy–Weinberg Equilibrium (HWE) in control

subjects were determined with the v2 test.

Assuming an odds ratio (OR) of 1.5 at a genotypic risk

level of a = 0.05 (two-sided), power was considered ade-

quate at C80%. In Lys751Gln, this level of power was

found in 16 (50.0%) of the 32 studies and in 14 (58.3%) of

the 24 studies in Asp312Asn (Table 1). Statistical powers

for the combined studies in overall analysis and all sub-

groups were above adequate (C97%).

Meta-analysis

We estimated OR of association with the variant CC

(Lys751Gln) and variant AA (Asp312Asn) genotypes

Breast Cancer Res Treat

123

Table 1 Characteristics of studies of two ERCC2 polymorphisms and their associations with breast cancer

First author (year) [Ref] Nationality/

ethnicity

ERCC2 (XPD) A751C ERCC2 (XPD) G312A

Case/

control (N)

Powerc (a = 0.05)

OR = 1.5

vafb in

controls

HWE Case/

control (N)

Powerc (a = 0.05)

OR = 1.5

vafb in

controls

HWE

Crew (2007) [34] American US – – – – 1031/1083 99.6 0.34 0.038

Forsti (2004) [17] Polish EU – – – – 170/181 46.3 0.61 0.0002

Jorgensen (2007) [36] American US – – – – 260/274 63.5 0.37 0.047

Kuschel AU (2005) [20] Australian – – – – 1453/793 99.5 0.34 0.16

Kuschel GE (2005) [20] German EU – – – – 2738/769 99.8 0.41 0.33

Lee (2005) [46] Korean AS – – – – 528/445 87.4 0.05 0.10

LSHTM (2006) [23] UK EU – – – – 579/591 92.8 0.13 0.33

Ribas (2006) [29] Spanish EU – – – – 833/806 98.1 0.69 0.62

Bernard-Gallon (2008)

[24]

French EU 908/995 99.1 0.66 0.84 904/994 99.1 0.33 0.13

Debniak (2006) [21] Polish EU 1830/511 97.9 0.39 0.93 1726/511 97.8 0.40 0.55

Debniak (2006) [21] Polish EU 1830/1141 99.9 0.38 0.60 1726/1262 99.9 0.37 0.70

Forsti (2004) [17] Finnish EU 222/314 62.4 0.41 0.12 223/310 62.3 0.39 0.37

Jakubowska (2010) [32] Polish EU 315/290 68.9 0.42 0.12 314/290 68.9 0.40 0.59

Justenhoven (2004) [26] German EU 586/643 93.8 0.36 0.66 567/610 92.9 0.34 0.10

Kuschel UK (2005) [20] UK EU 1676/1718 99.9 0.37 0.54 1605/1742 99.9 0.33 0.96

Mechanic (2006) [18] American US 1273/1133 99.8 0.37 0.53 1262/1133 99.8 0.34 0.64

Mechanic (2006) [18] African-

American US

761/679 96.6 0.24 0.85 760/675 96.6 0.13 0.45

Nexo (2003) [58] Danish EU 425/435 83.4 0.37 0.01 413/418 82.1 0.38 0.053

Shen (2006) [39] American US 154/153 41.6 0.33 0.051 156/153 41.8 0.41 0.10

Shi (2004) [40] American US 69/79 22.6 0.30 0.59 69/79 22.6 0.25 0.47

Smith (2008) [19] American US 314/399 75.4 0.39 0.40 304/391 74.3 0.35 0.24

Smith (2008) [19] African-

American US

52/72 19.3 0.20 0.13 49/74 19 0.12 0.92

Tang (2002) [42] American US 103/121 38.4 0.36 0.049 90/194 34.6 0.21 0.006

Zhang (2005) [48] Chinese AS 220/310 62 0.41 0.09 220/310 62 0.39 0.37

Brewster (2006) [33] American US 309/318 71.5 0.64 0.11 – – – –

Costa (2007) [25] Portuguese EU 282/660 80.2 0.70 0.10 – – – –

Dufloth (2005) [44] Brazilian 86/117 28.9 0.70 0.21 – – – –

Faraglia (2003) [35] American US 144/53 23.6 0.70 0.91 – – – –

Hsu (2010) [45] Chinese AS 401/533 85.6 0.08 0.20 – – – –

Kipikasova (2008) [27] Polish EU 114/113 32 0.37 0.58 – – – –

Li (2008) [47] Chinese AS 486/479 87.4 0.10 0.22 – – – –

Romanowicz-Makowska

(2007) [30]

Polish EU 92/110 29.1 0.47 0.59 – – – –

Metsola (2005) [28] Finnish EU 481/480 87.2 0.43 0.88 – – – –

Onay (2006) [37] Canadian NA 398/372 80.0 0.33 0.82 – – – –

Rajaraman (2008) [38] American US 839/1080 99.1 0.38 0.43 – – – –

Shore (2008) [41] American US 611/611 93.7 0.65 0.61 – – – –

Syamala F (2009) [22]a Indian AS 140/367 52 0.19 0.0006 – – – –

Syamala S (2009) [22]a Indian AS 219/367 64.7 0.19 0.0006 – – – –

Synowiec (2008) [31] Polish EU 43/48 15.6 0.40 0.42 – – – –

Terry (2004) [43] American US 1053/1102 99.6 0.36 0.45 – – – –

a Publication treated familial and sporadic separatelyb vaf: variant allele frequency in controls; Hardy–Weinberg Equilibrium (HWE) in controls (P \ 0.10)c Power was calculated with the G*Power program (http://www.psycho.uni-duesseldorf.de/aap/projects/gpower) as probability of detecting an association

between Asp312Asn and Lys751Gln and breast cancer assuming odds ratios (OR) of 1.5 (small effects size)

Breast Cancer Res Treat

123

compared with the wild-type AA (Lys751Gln) or GG

(Asp312Asn) genotypes, respectively. To evaluate impor-

tance of the heterozygous genotype, dominant and recessive

genetic models were also applied. For Lys751Gln, we

examined contrast of CC vs. AC ? AA genotypes as well as

the CC ? AC vs. AA genotypes. Likewise, for Asp312Asn,

we analyzed contrasts of AA vs. AG ? GG and AA ? GA

vs. GG genotypes. These contrasts correspond to recessive

and dominant effects of the C allele in Lys751Gln and A

allele in Asp312Asn. Raw data for genotype frequencies,

without adjustment, were used for calculating study-specific

estimates of the OR. Significance of the pooled OR (sum-

mary estimates) was determined by the Z-test. Pooled ORs

were obtained using either the fixed (Mantel–Haenszel) or

random (DerSimonian–Laird) effects models. The fixed-

effects model was used in the absence of heterogeneity [49]

while the random-effects model was used in its presence

[50]. Assuming genuine diversity in the results of various

studies, the random-effects model incorporates between

study variance. All analyses were done using Review

Manager (RevMan, v.4.2, Oxford, England), SigmaStat

(v.2.03), and SigmaPlot (v.9.01). All P values were two-

sided, significance of which was set at \0.05 throughout

except in heterogeneity estimation and publication bias.

Heterogeneity and outlier analysis

Heterogeneity between studies was estimated using the

v2-based Q test [51] significance of which was set at

P \ 0.10 [52]. Effect of heterogeneity was quantified with

the I2 statistic which measures the degree of inconsistency

among the studies [53]. Heterogeneity was explored using

subgroup analysis [51] with ethnicity and statistical power

as variables and whether a risk factor profile (aromatic

adducts) played a role in ERCC2 associations with breast

cancer. The Galbraith plot [54] was used to identify

potential outlier studies after which their influence on

pooled effects and/or heterogeneity was graphically

examined [55]. Outlier analysis was directed at studies in

the overall findings and in the power groups.

Of the seven significant pooled ORs in Lys751Gln, five

were heterogeneous and in Asp312Asn, all but one of the

five significant pooled ORs was homogeneous (Table 2).

Outliers were detected, three each for Lys751Gln [22, 30,

47] and Asp312Asn [17, 39, 48] (Fig. 1), their removal

followed by re-analysis resulted in loss of heterogeneity for

both polymorphisms (Fig. 2a, b).

Population admixture and publication bias

Admixtures of the USA Caucasian population were found

in five [35, 38, 39, 41, 42] (15.6%) of the 32 studies in

Lys751Gln and in two [34, 36] (8.3%) of the 24 studies of

Asp312Asn. Publication bias was statistically evaluated

with Egger’s regression asymmetry test, significance of

which was set at P \ 0.10 [56] which detects whether the

intercept deviates significantly from zero in a regression of

the standardized effect estimates against their precision

[57].

Results

Overall analyses

Here we have carried out meta-analyses of 40 separate

study populations, which provided genotype data from

30,799 cases and 29,358 controls for a total of 60,157

subjects. Seven [21, 30, 31, 40, 42, 44, 46] of the 40 studies

were hospital-based but a much larger proportion was

population-based (82.5%), thus representing the general

population. Of the 31 matched studies, 17 (54.8%) com-

prised the age criterion while six (19.4%) comprised the

geography criterion. The remaining eight (26.7%) was

matched using a combination of the above-mentioned cri-

teria plus race and gender. Egger’s test indicated presence

of overall publication bias in studies investigating

Lys751Gln but this was not detectable for Asp312Asn

studies. Assuming a genotypic risk of C1.5 (a = 0.05, two-

sided) when considered individually, half (50%) to over

half (58%) of the 32 Lys751Gln and 24 Asp312Asn studies

had adequate power (C80%), respectively. Allowing a

Type I error of 5%, the present meta-analysis has power

greater than 80% to detect an effect size of 1.5 for the

overall analysis and all subgroups in both polymorphisms.

Overall analyses of Lys751Gln have shown statistically

significant, moderately increased risk associated with

breast cancer under the homozygous (OR 1.14, P = 0.02)

and dominant models (OR 1.13, P = 0.02) (Table 2).

Stratifying the studies by statistical power (B80% vs.

C80%), has implicated that this effect comes from heter-

ogeneous (I2 range: 60–74%) and underpowered studies

(OR range: 1.31–1.43, P value range: 0.01–0.04). These

significant effects exhibited no publication bias (Table 3).

On the other hand, the observed risk effects were non-

significant and null (OR = 1.01–1.03) for the relatively

homogenous (I2 = 0–45%) studies with high statistical

power (Table 2). Furthermore, significant effects detected

for the underpowered studies were lost after removal of the

outlier populations [22, 30, 47] within this category

(Fig. 2a). For the 312 analyses, heterogeneity of under-

powered category was lower (I2 range: 12–40%) compared

to the statistically powered category (I2 range: 61–97%).

Analyses of the underpowered studies have shown pro-

tective effect for Asp312Asn under the homozygous (OR

0.80, 95% CI 0.66–0.98, P = 0.03) and recessive (OR

Breast Cancer Res Treat

123

Ta

ble

2S

um

mar

yo

fp

oo

led

od

ds

rati

os

for

var

iou

sco

ntr

asts

Gen

etic

mo

del

Ho

mo

zyg

ou

sR

eces

siv

eD

om

inan

t

Ly

s75

1G

lnN

o.

com

par

iso

ns

(sam

ple

size

s

case

/co

ntr

ol)

Ly

sLy

sv

s.G

lnG

lnG

lnG

lnv

s.G

lnL

ys

?L

ysL

ys

Gln

Gln

?G

lnL

ys

vs.

Ly

sLy

s

OR

(95

%C

I)P

val

ue

Phet

I2(%

)O

R(9

5%

CI)

Pv

alu

eP

het

I2(%

)O

R(9

5%

CI)

Pv

alu

eP

het

I2(%

)

All

32

(14

,54

5/1

5,3

52

)1

.14

(1.0

2–

1.2

8)

0.0

2R

0.0

02

47

.91

.09

(0.9

9–

1.2

0)

0.0

9R

0.0

03

45

.21

.13

(1.0

2–

1.2

4)

0.0

2R

\0

.00

01

69

.6

Sta

tist

ical

po

wer

[8

0%

16

(11

,94

9/1

2,4

88

)1

.03

(0.9

5–

1.1

2)

0.4

70

.48

0.0

1.0

1(0

.94–

1.0

8)

0.8

50

.74

0.0

1.0

2(0

.95–

1.1

0)

0.5

8R

0.0

34

4.7

\8

0%

16

(2,5

96

/2,8

64

)1

.43

(1.0

8–

1.9

0)

0.0

1R

0.0

01

59

.91

.31

(1.0

1–

1.7

1)

0.0

4R

0.0

00

36

3.7

1.3

6(1

.08

–1

.71

)0

.01

R\

0.0

00

17

4.1

Eth

nic

ity

All

Cau

casi

ans

25

(12

,26

6/1

2,9

12

)1

.05

(0.9

7–

1.1

3)

0.2

40

.43

2.2

1.0

4(0

.95–

1.1

3)

0.4

0R

0.0

53

4.4

1.0

3(0

.98–

1.0

9)

0.2

20

.39

5.2

Eu

rop

ean

s1

3(6

,97

4/7

,45

8)

1.0

4(0

.94–

1.1

5)

0.4

50

.20

24

.11

.08

(0.9

3–

1.2

5)

0.3

2R

0.0

05

57

.71

.02

(0.9

5–

1.0

9)

0.5

90

.34

10

.7

No

rth

Am

eric

ans

11

(5,2

06

/5,3

37

)1

.06

(0.9

4–

1.1

9)

0.3

60

.56

0.0

0.9

9(0

.90–

1.1

0)

0.8

70

.69

0.0

1.0

6(0

.97–

1.1

5)

0.2

00

.33

12

.4

Afr

ican

-Am

eric

ans

2(8

13

/75

1)

1.2

3(0

.81–

1.8

7)

0.3

30

.75

0.0

1.1

7(0

.78–

1.7

7)

0.4

40

.94

0.0

1.2

5(1

.03

–1

.53

)0

.03

0.1

35

6.6

Asi

ans

5(1

,46

6/1

,68

9)

1.6

0(0

.70–

3.6

6)

0.2

6R

0.0

00

28

2.3

1.4

4(0

.80–

2.6

1)

0.2

2R

0.0

16

8.1

1.3

5(0

.70–

2.5

7)

0.3

7R

\0

.00

01

93

.6

Aro

mat

icad

du

cts

4(1

,28

0/1

,21

8)

1.2

1(0

.94–

1.5

6)

0.1

30

.84

0.0

1.0

9(0

.87–

1.3

8)

0.4

50

.80

0.0

1.2

0(1

.02

–1

.41

)0

.03

0.9

30

.0

Asp

31

2A

snN

o.

com

par

iso

ns

(sam

ple

size

s

case

/co

ntr

ol)

Asn

Asn

vs.

Asp

Asp

Asn

Asn

vs.

Asp

Asn

?A

spA

spA

snA

sn?

Asp

Asn

vs.

Asp

Asp

OR

(95

%C

I)P

val

ue

Phet

I2(%

)O

R(9

5%

CI)

Pv

alu

eP

het

I2(%

)O

R(9

5%

CI)

Pv

alu

eP

het

I2(%

)

All

24

(16

,25

4/1

4,0

06

)0

.94

(0.8

2–

1.0

6)

0.3

1R

0.0

00

55

5.7

0.8

3(0

.65

–1

.05

)0

.13

R\

0.0

00

18

9.6

1.1

3(0

.92

–1

.38

)0

.26

R\

0.0

00

19

4.1

Sta

tist

ical

po

wer

[8

0%

14

(14

,39

9/1

1,8

32

)0

.99

(0.8

6–

1.1

4)

0.8

5R

0.0

02

60

.50

.85

(0.6

1–

1.1

7)

0.3

1R

\0

.00

01

93

.61

.14

(0.8

6–

1.5

1)

0.3

7R

\0

.00

01

96

.6

\8

0%

10

(1,8

55

/2,1

74

)0

.80

(0.6

6–

0.9

8)

0.0

30

.11

37

.10

.77

(0.6

0–1

.00

)0

.05

R0

.09

40

.01

.02

(0.9

0–

1.1

6)

0.7

90

.34

11

.6

Eth

nic

ity

All

Cau

casi

ans

20

(14

,69

7/1

2,5

02

)0

.95

(0.8

3–

1.0

8)

0.4

40

.00

05

58

.90

.83

(0.6

4–

1.0

7)

0.1

5R

\0

.00

01

91

.21

.13

(0.9

0–

1.4

2)

0.3

0R

\0

.00

01

95

.1

Eu

rop

ean

s1

2(1

0,0

72

/8,4

84

)0

.92

(0.7

7–

1.1

1)

0.3

7R

\0

.00

01

70

.70

.77

(0.5

2–

1.1

3)

0.1

9R

\0

.00

01

94

.71

.14

(0.8

0–

1.6

4)

0.4

7R

\0

.00

01

97

.1

No

rth

Am

eric

ans

7(3

,17

2/3

,22

5)

1.0

0(0

.85

–1

.18

)0

.95

0.2

03

0.3

0.9

6(0

.82

–1

.11

)0

.56

0.2

22

7.7

1.0

8(0

.98

–1

.19

)0

.12

0.1

53

7.0

Afr

ican

-Am

eric

ans

2(8

09

/74

9)

1.1

8(0

.58

–2

.41

)0

.65

0.3

60

.01

.14

(0.5

6–

2.3

2)

0.7

20

.39

0.0

1.1

7(0

.93

–1

.47

)0

.18

0.4

10

.0

Asi

ans

2(7

48

/75

5)

0.5

5(0

.32–

0.9

6)

0.0

30

.58

0.0

0.5

3(0

.32

–0

.90

)0

.02

0.5

60

.00

.96

(0.7

3–

1.2

5)

0.7

50

.71

0.0

Aro

mat

icad

du

cts

3(1

,60

2/1

,70

2)

1.1

7(0

.94

–1

.45

)0

.16

0.2

33

1.4

1.1

1(0

.91

–1

.36

)0

.31

0.7

90

.01

.17

(1.0

2–

1.3

4)

0.0

30

.31

15

.1

OR

od

ds

rati

o,

CI

con

fid

ence

inte

rval

,P

Pv

alu

efo

rsu

mm

ary

effe

cts,

Phet

Pv

alu

efo

rh

eter

og

enei

ty

Po

ole

dO

Rs

mar

ked

wit

hsu

per

scri

pt

Rw

ere

der

ived

fro

mra

nd

om

-eff

ects

mo

del

oth

erw

ise,

fix

ed-e

ffec

tsm

od

elw

asu

sed

;B

old

ind

icat

essi

gn

ifica

nt

resu

lts

Breast Cancer Res Treat

123

0.77, 95% CI 0.64–0.93, P = 0.006) models, however,

these associations were not statistically significant for the

overall analyses and the analyses of statistically powered

studies (Table 2). Further removal of outliers [17, 26, 39,

48] within the underpowered category also resulted in non-

significant associations (Fig. 2b).

Genotype distribution of the control group in four [22,

42, 58] (Ref. [22] consists of two studies) of the 32 studies

for Lys751Gln and four [17, 34, 36, 42] of the 24 studies

for Asp312Asn deviated from the HWE indicating poten-

tial biases in the selection of controls (Table 1). For both

polymorphisms, these HWE-deviating studies were omit-

ted followed by recalculation of the summary effects in the

overall and adduct analyses as well as the underpowered

subgroup. In Lys751Gln, where pooled ORs were signifi-

cant, this resulted in loss of heterogeneity in all analyses

and significance (overall analyses and B80% homozygous

and recessive subgroups). Significance in the dominant

model of the B80% subgroup and in the adduct analyses

were unaffected (not shown). In Asp312Asn, pooled ORs

and heterogeneity were unchanged in the overall analyses

but incurred loss of significance in the B80% subgroup. In

the adduct analyses (N = 3), removal of two studies [34,

42] that deviated from the HWE left study [40]-specific

ORs that ranged from 1.60 to 2.11 across the genetic

models (not shown).

Ethnic group analyses

Ethnic analyses of both Lys751Gln and Asp312Asn vari-

ants have shown non-significant and relatively null asso-

ciations with breast cancer risk for the overall Caucasians

as well as North American and European sub-populations

(Table 2). In Lys751Gln there was increased risk in Asians

(OR = 1.40–1.59) and African-Americans (1.17–1.25),

where the effect reached significance only for relatively

homogenous latter subgroup (I2 = 0.0–56%) under the

dominant model (OR 1.25, 95% CI 1.03–1.53, P = 0.03).

The only statistically significant association in Asp312Asn

were those in Asians where homogeneous effects

(I2 = 0.0%) were protective under the homozygous (OR

0.55, 95% CI 0.32–0.96, P = 0.03) and recessive (OR

0.53, 95% CI 0.32–0.90, P = 0.02) models (Table 2).

Fig. 1 Galbraith plot analysis to evaluate heterogeneity. Parenthesized numbers indicate reference numbers of outlier studies. H homozygous, Rrecessive, D dominant, (22S) Syamala et al. sporadic, (22F) Syamala et al. familial

Breast Cancer Res Treat

123

Aromatic adducts

All adduct data were from Caucasian-American women

with 6.2 and 7.3% population admixture [34, 35]. Analysis

of homogeneous studies (I2 range: 0.0–31.4%) have shown

slight increased risk which was only significant under the

dominant model for Lys751Gln (OR 1.20, 95% CI 1.02–

1.41, P = 0.03) and Asp312Asn (OR 1.17, 95% CI 1.02–

1.34, P = 0.03) (Table 2). However, only a single study in

Lys751Gln [43] and in Asp312Asn [34] with 99.6% sta-

tistical power, accounted for the bigger portion of the effect

detected in the pooled analyses for all adduct data. Allowing

type 1 error of 5%, our analyses of the adduct data had

power C80% in detecting an effect size range of 1.1–1.5 for

both polymorphisms. Combined statistical power at a

genotypic risk of 1.5 was[99%. Significant effects in both

polymorphisms exhibited no publication bias (Table 3).

Discussion

Independent meta-analysis of the two well-characterized

polymorphisms of ERCC2, Lys751Gln and Asp312Asn,

has shown no association with breast cancer risk. Although

significantly associated modest alterations in risk were

detected in the overall analyses, this effect was mainly

contributed by the underpowered studies. Removal of the

underpowered and outlier/heterogeneous studies consis-

tently confirmed the null effects associated with the two

ERCC2 polymorphisms. The ratio of cases for powered

studies was about five- to tenfold higher when compared to

the underpowered studies, further emphasizing the null

effects associated with these polymorphisms.

To the best of our knowledge, this is so far the largest

meta-analysis undertaken for ERCC2 Lys751Gln and

Asp312Asn polymorphisms in breast cancer. A recent

meta-analysis by Wang et al. [59] has also reported null

association of Lys751Gln with breast cancer risk in a

homogenous population of 15,664 subjects (about half the

size of our population of 29,897 subjects). The same study

has shown a protective effect for Asp312Asn with breast

cancer; however, these results came from a heterogeneous

population of 11,443 subjects (less than half the size of our

population of 30,260 subjects). Although, half of the studies

were underpowered, Wang et al. [59] did not investigate

subgroups stratified by statistical power. The findings of our

Fig. 2 Effect of removing outlier studies on heterogeneity and

summary odds ratios (OR). Genetic models: H homozygous, Rrecessive, D dominant. Parenthesized numbers left and right of the

y-axis: number of studies and omitted references (*), respectively. CIconfidence interval. Filled blocks in the forest plot indicate

significance. Analysis models: R random-effects, F fixed-effects, Phet

P value for heterogeneity. Effects on heterogeneity: LOH loss of

heterogeneity, NE no effect on heterogeneity, IH increase in homo-

geneity, Hom homogeneous, Het heterogeneous. a Lys751Gln, (22S)Syamala et al. sporadic; (22F) Syamala et al. familial; b Asp312Asn

Breast Cancer Res Treat

123

study support the importance of homogenous and statisti-

cally powerful studies in evaluating risk of commonly

occurring polymorphisms in the human population.

Ethnicity-specific differences were evident for both

polymorphisms with null effects among Caucasians as well

as European and North American sub-populations.

Although a significantly associated risk was observed for

African-Americans and Asians for Lys751Gln and

Asp312Asn, respectively, the sample size was quite small

and the confidence intervals were quite large. Associations

found in one population, but not in another, could poten-

tially be explained by variability in statistical power and

heterogeneity of the association studies. The linkage

between Lys751Gln and Asp312Asn was r2 = 0.56 in

populations of European ancestry and r2 = 0.11 in popu-

lations of African ancestry [19]. Therefore, linkage dis-

equilibrium differences among these populations may also

impact on the outcomes of these polymorphisms [60].

Among the carcinogenic compounds contained in

tobacco smoke, the polycyclic aromatic hydrocarbons and

aromatic amines have been regarded as significant etiologic

environmental factors in breast cancer [61–64]. With

exposure, reactive metabolites may bind to DNA, forming

adducts with mutagenic consequences, mainly transver-

sions and frameshift mutations [65, 66]. Our adduct anal-

yses consisted of a relatively large sample ([2,000

subjects) homogeneous and statistically powered studies of

Caucasian-American populations. Our findings of 1.2-fold

significant increased risk for breast cancer in both poly-

morphisms were found to be correlated with high levels of

DNA adducts (or lower DNA repair capacity) [67].

Reduced DNA repair capacity is in turn influenced by

polymorphic variations in genes that are responsible for

removing these adducts [68, 69]. Variations such as

Lys751Gln and Asp312Asn of the ERCC2 gene that par-

ticipate in culling these adducts have been found to be

predictive of DNA repair capacity [40]. Other studies

investigating correlations between the Gln allele of

Lys751Gln and higher DNA adduct levels or lower DNA

repair efficiency have reported positive [70] and null

effects [71]. Similarly, a cytogenetic study showed that the

variant 751Gln genotype was not associated with increase

in bulky or polyphenol DNA adducts [72]. Functional

studies have demonstrated a higher level of DNA adducts,

measured by 32P-postlabeling, in lymphocytes of subjects

with the ERCC2 751Gln/Gln genotype [73]. Higher levels

of DNA adducts were found in workers with one Gln allele

who were exposed to traffic pollution compared to those

with two alleles [74]. In another study, peripheral blood

lymphocytes of subjects with the 312Asn and 751Gln

alleles were found to have increased number of aromatic

DNA adducts [75].

The presence of heterogeneity and publication bias may

have impacted upon the overall significant findings in

Lys751Gln and limited the ability of the meta-analysis in

finding estimates of true associations. Absence of publi-

cation bias in Asp312Asn, on the other hand, contributes to

the overall strength of our meta-analysis. Central to its

strength, however, is the substantial number of cases and

controls pooled from different studies which translates to

sufficient statistical power in the combined analyses and in

majority of the individual studies. Lack of proper matching

of controls to breast cancer cases may indicate selection

bias. Although controls were selected mainly from healthy

populations, a substantial number did not mention physi-

ological condition of this group allowing for the possibility

of non-differential misclassification bias owing to the

inclusion of control groups with different risks of devel-

oping breast cancer. However, this bias is unlikely in our

meta-analysis, since 75% (30 of 40) of the studies were

matched for age, gender, geography and/or ethnicity.

To conclude, in light of studies with high statistical

power and relatively homogenous populations, our study

implicates null associations for both Lys751Gln and

Asp312Asn. Although African-Americans and Asians were

shown to be associated with altered breast cancer risk for

both Lys751Gln and Asp312Asn, the sample size was quite

small, also suggested by large confidence intervals. The

statistically significant 1.2-fold increased breast cancer risk

was observed for adduct analyses for both polymorphisms.

Considered individually, these two polymorphisms have

been shown to have little or no influence and would

Table 3 Results of Egger’s

regression asymmetry tests for

publication bias in the overall

analyses and subgroups (N C 3

studies) with significant effects

* Bold indicates presence of

publication bias

Homozygous Recessive Dominant

N Intercept P value Intercept P value Intercept P value

Lys751Gln

Overall 32 1.18 0.02* – – 1.46 0.05*

\80 Power 16 1.40 0.10 1.37 0.10 1.15 0.26

Aromatic adducts 4 – – – – 0.02 0.92

Asp312Asn

\80 Power 10 1.00 0.32 0.69 0.54 – –

Aromatic adducts 3 – – – – 0.58 0.55

Breast Cancer Res Treat

123

probably require haplotype analysis to discern combined

effects. Such analysis may shed light on the complexities

of the many pathways involved in DNA repair and breast

cancer development, providing hypotheses for future

functional studies.

Acknowledgments The Philippine Department of Science and

Technology (DOST) awarded Noel Pabalan with a Balik-Scientist

Status. The Canadian Breast Cancer Foundation (CBCF) grant sup-

ports Hilmi Ozcelik. Lillian Sung is supported by a New Investigator

Award from the Canadian Institutes of Health Research. We thank Dr.

Xiangdong Liu of Analytic Genetics Technology Centre, Princess

Margaret Hospital, Toronto, Canada; Hong Li of Ozcelik’s Labora-

tory, Mount Sinai Hospital, Toronto, Canada; Dorothy Joy Ireneo and

Darwin Casuga of the Library Services in Saint Louis University for

their support.

References

1. Coin F, Marinoni JC, Rodolfo C, Fribourg S, Pedrini AM, Egly

JM (1998) Mutations in the XPD helicase gene result in XP and

TTD phenotypes, preventing interaction between XPD and the

p44 subunit of TFIIH. Nat Genet 20(2):184–188

2. Laine JP, Mocquet V, Bonfanti M, Braun C, Egly JM, Brousset P

(2007) Common XPD (ERCC2) polymorphisms have no mea-

surable effect on nucleotide excision repair and basal transcrip-

tion. DNA Repair (Amst) 6(9):1264–1270

3. Shen MR, Jones IM, Mohrenweiser H (1998) Nonconservative

amino acid substitution variants exist at polymorphic frequency

in DNA repair genes in healthy humans. Cancer Res 58(4):604–

608

4. Xi T, Jones IM, Mohrenweiser HW (2004) Many amino acid

substitution variants identified in DNA repair genes during

human population screenings are predicted to impact protein

function. Genomics 83(6):970–979

5. Clarkson SG, Wood RD (2005) Polymorphisms in the human

XPD (ERCC2) gene, DNA repair capacity and cancer suscepti-

bility: an appraisal. DNA Repair (Amst) 4(10):1068–1074

6. Bienstock RJ, Skorvaga M, Mandavilli BS, Van Houten B (2003)

Structural and functional characterization of the human DNA

repair helicase XPD by comparative molecular modeling and

site-directed mutagenesis of the bacterial repair protein UvrB.

J Biol Chem 278(7):5309–5316

7. Dubaele S, Proietti De Santis L, Bienstock RJ, Keriel A, Stefanini

M, Van Houten B, Egly JM (2003) Basal transcription defect

discriminates between xeroderma pigmentosum and trichothio-

dystrophy in XPD patients. Mol Cell 11(6):1635–1646

8. Giglia-Mari G, Coin F, Ranish JA, Hoogstraten D, Theil A,

Wijgers N, Jaspers NG, Raams A, Argentini M, van der Spek PJ

et al (2004) A new, tenth subunit of TFIIH is responsible for the

DNA repair syndrome trichothiodystrophy group A. Nat Genet

36(7):714–719

9. Hemminki K, Xu G, Angelini S, Snellman E, Jansen CT, Lambert

B, Hou SM (2001) XPD exon 10 and 23 polymorphisms and

DNA repair in human skin in situ. Carcinogenesis 22(8):1185–

1188

10. Au WW, Salama SA, Sierra-Torres CH (2003) Functional char-

acterization of polymorphisms in DNA repair genes using

cytogenetic challenge assays. Environ Health Perspect 111(15):

1843–1850

11. Qiao Y, Spitz MR, Shen H, Guo Z, Shete S, Hedayati M,

Grossman L, Mohrenweiser H, Wei Q (2002) Modulation of

repair of ultraviolet damage in the host-cell reactivation assay by

polymorphic XPC and XPD/ERCC2 genotypes. Carcinogenesis

23(2):295–299

12. Lunn RM, Helzlsouer KJ, Parshad R, Umbach DM, Harris EL,

Sanford KK, Bell DA (2000) XPD polymorphisms: effects on

DNA repair proficiency. Carcinogenesis 21(4):551–555

13. Seker H, Butkiewicz D, Bowman ED, Rusin M, Hedayati M,

Grossman L, Harris CC (2001) Functional significance of XPD

polymorphic variants: attenuated apoptosis in human lympho-

blastoid cells with the XPD 312 Asp/Asp genotype. Cancer Res

61(20):7430–7434

14. Wolfe KJ, Wickliffe JK, Hill CE, Paolini M, Ammenheuser MM,

Abdel-Rahman SZ (2007) Single nucleotide polymorphisms of

the DNA repair gene XPD/ERCC2 alter mRNA expression.

Pharmacogenet Genomics 17(11):897–905

15. Thakkinstian A, McElduff P, D’Este C, Duffy D, Attia J (2005) A

method for meta-analysis of molecular association studies. Stat

Med 24(9):1291–1306

16. Dufloth RM, Arruda A, Heinrich JK, Schmitt F, Zeferino LC

(2008) The investigation of DNA repair polymorphisms with

histopathological characteristics and hormone receptors in a

group of Brazilian women with breast cancer. Genet Mol Res

7(3):574–582

17. Forsti A, Angelini S, Festa F, Sanyal S, Zhang Z, Grzybowska E,

Pamula J, Pekala W, Zientek H, Hemminki K et al (2004) Single

nucleotide polymorphisms in breast cancer. Oncol Rep

11(4):917–922

18. Mechanic LE, Millikan RC, Player J, de Cotret AR, Winkel S,

Worley K, Heard K, Heard K, Tse CK, Keku T (2006) Poly-

morphisms in nucleotide excision repair genes, smoking and

breast cancer in African Americans and whites: a population-

based case-control study. Carcinogenesis 27(7):1377–1385

19. Smith TR, Levine EA, Freimanis RI, Akman SA, Allen GO,

Hoang KN, Liu-Mares W, Hu JJ (2008) Polygenic model of DNA

repair genetic polymorphisms in human breast cancer risk. Car-

cinogenesis 29(11):2132–2138

20. Kuschel B, Chenevix-Trench G, Spurdle AB, Chen X, Hopper JL,

Giles GG, McCredie M, Chang-Claude J, Gregory CS, Day NE

et al (2005) Common polymorphisms in ERCC2 (Xeroderma

pigmentosum D) are not associated with breast cancer risk.

Cancer Epidemiol Biomarkers Prev 14(7):1828–1831

21. Debniak T, Scott RJ, Huzarski T, Byrski T, Masojc B, van de

Wetering T, Serrano-Fernandez P, Gorski B, Cybulski C, Gron-

wald J et al (2006) XPD common variants and their association

with melanoma and breast cancer risk. Breast Cancer Res Treat

98(2):209–215

22. Syamala VS, Syamala V, Sreedharan H, Raveendran PB, Kuttan

R, Ankathil R (2009) Contribution of XPD (Lys751Gln) and

XRCC1 (Arg399Gln) polymorphisms in familial and sporadic

breast cancer predisposition and survival: an Indian report. Pathol

Oncol Res 15(3):389–397

23. Breast Cancer Association Consortium (2006) Commonly studied

single-nucleotide polymorphisms and breast cancer: results from

the Breast Cancer Association Consortium. J Natl Cancer Inst

98(19):1382–1396

24. Bernard-Gallon D, Bosviel R, Delort L, Fontana L, Chamoux A,

Rabiau N, Kwiatkowski F, Chalabi N, Satih S, Bignon YJ (2008)

DNA repair gene ERCC2 polymorphisms and associations with

breast and ovarian cancer risk. Mol Cancer 7:36

25. Costa S, Pinto D, Pereira D, Rodrigues H, Cameselle-Teijeiro J,

Medeiros R, Schmitt F (2007) DNA repair polymorphisms might

contribute differentially on familial and sporadic breast cancer

susceptibility: a study on a Portuguese population. Breast Cancer

Res Treat 103(2):209–217

26. Justenhoven C, Hamann U, Pesch B, Harth V, Rabstein S, Baisch

C, Vollmert C, Illig T, Ko YD, Bruning T et al (2004) ERCC2

genotypes and a corresponding haplotype are linked with breast

Breast Cancer Res Treat

123

cancer risk in a German population. Cancer Epidemiol Bio-

markers Prev 13(12):2059–2064

27. Kipikasova L, Wolaschka T, Bohus P, Baumohlova H, Bober J,

Blazejova J, Mirossay L, Sarissky M, Mirossay A, Cizmarikova

M et al (2008) Polymorphisms of the XRCC1 and XPD genes and

breast cancer risk: a case-control study. Pathol Oncol Res

14(2):131–135

28. Metsola K, Kataja V, Sillanpaa P, Siivola P, Heikinheimo L,

Eskelinen M, Kosma VM, Uusitupa M, Hirvonen A (2005)

XRCC1 and XPD genetic polymorphisms, smoking and breast

cancer risk in a Finnish case-control study. Breast Cancer Res

7(6):R987–R997

29. Ribas G, Gonzalez-Neira A, Salas A, Milne RL, Vega A,

Carracedo B, Gonzalez E, Barroso E, Fernandez LP, Yankilevich

P et al (2006) Evaluating HapMap SNP data transferability in a

large-scale genotyping project involving 175 cancer-associated

genes. Hum Genet 118(6):669–679

30. Romanowicz-Makowska H, Sobczuk A, Smolarz B, Fiks T, Kulig

A (2007) XPD Lys751Gln polymorphism analysis in women with

sporadic breast cancer. Pol J Pathol 58(4):245–249

31. Synowiec E, Stefanska J, Morawiec Z, Blasiak J, Wozniak K

(2008) Association between DNA damage, DNA repair genes

variability and clinical characteristics in breast cancer patients.

Mutat Res 648(1–2):65–72

32. Jakubowska A, Gronwald J, Menkiszak J, Gorski B, Huzarski T,

Byrski T, Tołoczko-Grabarek A, Gilbert M, Edler L, Zapatka M,

Eils R, Lubinski J, Scott RJ, Hamann U (2010) BRCA1-associ-

ated breast and ovarian cancer risks in Poland: no association

with commonly studied polymorphisms. Breast Cancer Res Treat

119(1):201–211

33. Brewster AM, Jorgensen TJ, Ruczinski I, Huang HY, Hoffman S,

Thuita L, Newschaffer C, Lunn RM, Bell D, Helzlsouer KJ

(2006) Polymorphisms of the DNA repair genes XPD

(Lys751Gln) and XRCC1 (Arg399Gln and Arg194Trp): rela-

tionship to breast cancer risk and familial predisposition to breast

cancer. Breast Cancer Res Treat 95(1):73–80

34. Crew KD, Gammon MD, Terry MB, Zhang FF, Zablotska LB,

Agrawal M, Shen J, Long CM, Eng SM, Sagiv SK et al (2007)

Polymorphisms in nucleotide excision repair genes, polycyclic

aromatic hydrocarbon-DNA adducts, and breast cancer risk.

Cancer Epidemiol Biomarkers Prev 16(10):2033–2041

35. Faraglia B, Chen SY, Gammon MD, Zhang Y, Teitelbaum SL,

Neugut AI, Ahsan H, Garbowski GC, Hibshoosh H, Lin D et al

(2003) Evaluation of 4-aminobiphenyl-DNA adducts in human

breast cancer: the influence of tobacco smoke. Carcinogenesis

24(4):719–725

36. Jorgensen TJ, Visvanathan K, Ruczinski I, Thuita L, Hoffman S,

Helzlsouer KJ (2007) Breast cancer risk is not associated with

polymorphic forms of xeroderma pigmentosum genes in a cohort

of women from Washington County, Maryland. Breast Cancer

Res Treat 101(1):65–71

37. Onay VU, Briollais L, Knight JA, Shi E, Wang Y, Wells S, Li H,

Rajendram I, Andrulis IL, Ozcelik H (2006) SNP-SNP interac-

tions in breast cancer susceptibility. BMC Cancer 6:114

38. Rajaraman P, Bhatti P, Doody MM, Simon SL, Weinstock RM,

Linet MS, Rosenstein M, Stovall M, Alexander BH, Preston DL

et al (2008) Nucleotide excision repair polymorphisms may

modify ionizing radiation-related breast cancer risk in US

radiologic technologists. Int J Cancer 123(11):2713–2716

39. Shen J, Desai M, Agrawal M, Kennedy DO, Senie RT, Santella

RM, Terry MB (2006) Polymorphisms in nucleotide excision

repair genes and DNA repair capacity phenotype in sisters dis-

cordant for breast cancer. Cancer Epidemiol Biomarkers Prev

15(9):1614–1619

40. Shi Q, Wang LE, Bondy ML, Brewster A, Singletary SE, Wei Q

(2004) Reduced DNA repair of benzo[a]pyrene diol epoxide-

induced adducts and common XPD polymorphisms in breast

cancer patients. Carcinogenesis 25(9):1695–1700

41. Shore RE, Zeleniuch-Jacquotte A, Currie D, Mohrenweiser H,

Afanasyeva Y, Koenig KL, Arslan AA, Toniolo P, Wirgin I

(2008) Polymorphisms in XPC and ERCC2 genes, smoking and

breast cancer risk. Int J Cancer 122(9):2101–2105

42. Tang D, Cho S, Rundle A, Chen S, Phillips D, Zhou J, Hsu Y,

Schnabel F, Estabrook A, Perera FP (2002) Polymorphisms in the

DNA repair enzyme XPD are associated with increased levels of

PAH-DNA adducts in a case-control study of breast cancer.

Breast Cancer Res Treat 75(2):159–166

43. Terry MB, Gammon MD, Zhang FF, Eng SM, Sagiv SK, Paykin

AB, Wang Q, Hayes S, Teitelbaum SL, Neugut AI et al (2004)

Polymorphism in the DNA repair gene XPD, polycyclic aromatic

hydrocarbon-DNA adducts, cigarette smoking, and breast cancer

risk. Cancer Epidemiol Biomarkers Prev 13(12):2053–2058

44. Dufloth RM, Costa S, Schmitt F, Zeferino LC (2005) DNA repair

gene polymorphisms and susceptibility to familial breast cancer

in a group of patients from Campinas, Brazil. Genet Mol Res

4(4):771–782

45. Hsu MS, Yu JC, Wang HW, Chen ST, Hsiung CN, Ding SL, Wu

PE, Shen CY, Cheng CW (2010) Synergistic effects of poly-

morphisms in dna repair genes and endogenous estrogen expo-

sure on female breast cancer risk. Ann Surg Oncol 17(3):760–771

46. Lee SA, Lee KM, Park WY, Kim B, Nam J, Yoo KY, Noh DY,

Ahn SH, Hirvonen A, Kang D (2005) Obesity and genetic

polymorphism of ERCC2 and ERCC4 as modifiers of risk of

breast cancer. Exp Mol Med 37(2):86–90

47. Li J, Jin W, Chen Y, Di G, Wu J, Shao ZM (2008) Genetic

polymorphisms in the DNA repair enzyme ERCC2 and breast

tumour risk in a Chinese population. J Int Med Res 36(3):479–

488

48. Zhang L, Zhang Z, Yan W (2005) Single nucleotide polymor-

phisms for DNA repair genes in breast cancer patients. Clin Chim

Acta 359(1–2):150–155

49. Mantel N, Haenszel W (1959) Statistical aspects of the analysis

of data from retrospective studies of disease. J Natl Cancer Inst

22(4):719–748

50. DerSimonian R, Laird N (1986) Meta-analysis in clinical trials.

Control Clin Trials 7(3):177–188

51. Lau J, Ioannidis JP, Schmid CH (1997) Quantitative synthesis in

systematic reviews. Ann Intern Med 127(9):820–826

52. Berman NG, Parker RA (2002) Meta-analysis: neither quick nor

easy. BMC Med Res Methodol 2:10

53. Higgins JP, Thompson SG (2002) Quantifying heterogeneity in a

meta-analysis. Stat Med 21(11):1539–1558

54. Galbraith RF (1988) A note on graphical presentation of esti-

mated odds ratios from several clinical trials. Stat Med 7(8):889–

894

55. Pabalan N, Bapat B, Sung L, Jarjanazi H, Francisco-Pabalan O,

Ozcelik H (2008) Cyclin D1 Pro241Pro (CCND1–G870A)

polymorphism is associated with increased cancer risk in human

populations: a meta-analysis. Cancer Epidemiol Biomarkers Prev

17(10):2773–2781

56. Zafarmand MH, van der Schouw YT, Grobbee DE, de Leeuw

PW, Bots ML (2008) The M235T polymorphism in the AGT

gene and CHD risk: evidence of a Hardy-Weinberg equilibrium

violation and publication bias in a meta-analysis. PLoS ONE

3(6):e2533

57. Egger M, Davey Smith G, Schneider M, Minder C (1997) Bias in

meta-analysis detected by a simple, graphical test. BMJ Clinical

researched 315(7109):629–634

58. Nexo BA, Vogel U, Olsen A, Ketelsen T, Bukowy Z, Thomsen

BL, Wallin H, Overvad K, Tjonneland A (2003) A specific

haplotype of single nucleotide polymorphisms on chromosome

19q13.2–3 encompassing the gene RAI is indicative of

Breast Cancer Res Treat

123

post-menopausal breast cancer before age 55. Carcinogenesis

24(5):899–904

59. Wang F, Chang D, Hu FL, Sui H, Han B, Li DD, Zhao YS (2008)

DNA repair gene XPD polymorphisms and cancer risk: a meta-

analysis based on 56 case-control studies. Cancer Epidemiol

Biomarkers Prev 17(3):507–517

60. King CR, Yu J, Freimuth RR, McLeod HL, Marsh S (2005)

Interethnic variability of ERCC2 polymorphisms. Pharmacoge-

nomics J 5(1):54–59

61. Brody JG, Rudel RA (2003) Environmental pollutants and breast

cancer. Environ Health Perspect 111(8):1007–1019

62. Gorlewska-Roberts K, Green B, Fares M, Ambrosone CB,

Kadlubar FF (2002) Carcinogen-DNA adducts in human breast

epithelial cells. Environ Mol Mutagen 39(2–3):184–192

63. Lodovici M, Akpan V, Evangelisti C, Dolara P (2004) Sidestream

tobacco smoke as the main predictor of exposure to polycyclic

aromatic hydrocarbons. J Appl Toxicol 24(4):277–281

64. DeBruin LS, Josephy PD (2002) Perspectives on the chemical

etiology of breast cancer. Environ Health Perspect 110(Suppl

1):119–128

65. Kadlubar FF (1994) DNA adducts of carcinogenic aromatic

amines. IARC Sci Publ 125:199–216

66. Schut HA, Snyderwine EG (1999) DNA adducts of heterocyclic

amine food mutagens: implications for mutagenesis and carci-

nogenesis. Carcinogenesis 20(3):353–368

67. Zhao H, Wang LE, Li D, Chamberlain RM, Sturgis EM, Wei Q

(2008) Genotypes and haplotypes of ERCC1 and ERCC2/XPD

genes predict levels of benzo[a]pyrene diol epoxide-induced

DNA adducts in cultured primary lymphocytes from healthy

individuals: a genotype-phenotype correlation analysis. Carcino-

genesis 29(8):1560–1566

68. Berwick M, Vineis P (2000) Markers of DNA repair and sus-

ceptibility to cancer in humans: an epidemiologic review. J Natl

Cancer Inst 92(11):874–897

69. Neumann AS, Sturgis EM, Wei Q (2005) Nucleotide excision

repair as a marker for susceptibility to tobacco-related cancers: a

review of molecular epidemiological studies. Mol Carcinog

42(2):65–92

70. Kiyohara C, Yoshimasu K (2007) Genetic polymorphisms in the

nucleotide excision repair pathway and lung cancer risk: a meta-

analysis. Int J Med Sci 4(2):59–71

71. Duell EJ, Wiencke JK, Cheng TJ, Varkonyi A, Zuo ZF, Ashok

TD, Mark EJ, Wain JC, Christiani DC, Kelsey KT (2000) Poly-

morphisms in the DNA repair genes XRCC1 and ERCC2 and

biomarkers of DNA damage in human blood mononuclear cells.

Carcinogenesis 21(5):965–971

72. Matullo G, Palli D, Peluso M, Guarrera S, Carturan S, Celentano

E, Krogh V, Munnia A, Tumino R, Polidoro S et al (2001)

XRCC1, XRCC3, XPD gene polymorphisms, smoking and (32)P-

DNA adducts in a sample of healthy subjects. Carcinogenesis

22(9):1437–1445

73. Matullo G, Peluso M, Polidoro S, Guarrera S, Munnia A, Krogh

V, Masala G, Berrino F, Panico S, Tumino R et al (2003)

Combination of DNA repair gene single nucleotide polymor-

phisms and increased levels of DNA adducts in a population-

based study. Cancer Epidemiol Biomarkers Prev 12(7):674–677

74. Palli D, Russo A, Masala G, Saieva C, Guarrera S, Carturan S,

Munnia A, Matullo G, Peluso M (2001) DNA adduct levels and

DNA repair polymorphisms in traffic-exposed workers and a

general population sample. Int J Cancer 94(1):121–127

75. Hou SM, Falt S, Angelini S, Yang K, Nyberg F, Lambert B,

Hemminki K (2002) The XPD variant alleles are associated with

increased aromatic DNA adduct level and lung cancer risk.

Carcinogenesis 23(4):599–603

Breast Cancer Res Treat

123