predictive value of xrcc1 gene polymorphisms on platinum-based chemotherapy in advanced non-small...
TRANSCRIPT
2012;18:3972-3981. Published OnlineFirst June 15, 2012.Clin Cancer Res Junjie Wu, Jie Liu, Yuhao Zhou, et al. Cancer Patients: A Systematic Review and Meta-analysis
Small Cell Lung−Platinum-Based Chemotherapy in Advanced Non Gene Polymorphisms onXRCC1Predictive Value of
Updated Version 10.1158/1078-0432.CCR-11-1531doi:
Access the most recent version of this article at:
MaterialSupplementary
http://clincancerres.aacrjournals.org/content/suppl/2012/05/18/1078-0432.CCR-11-1531.DC1.htmlAccess the most recent supplemental material at:
Cited Articles http://clincancerres.aacrjournals.org/content/18/14/3972.full.html#ref-list-1
This article cites 43 articles, 10 of which you can access for free at:
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
SubscriptionsReprints and
[email protected] atTo order reprints of this article or to subscribe to the journal, contact the AACR Publications
To request permission to re-use all or part of this article, contact the AACR Publications Department at
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
Predictive Biomarkers and Personalized Medicine
Predictive Value of XRCC1 Gene Polymorphisms onPlatinum-Based Chemotherapy in Advanced Non–SmallCell Lung Cancer Patients: A Systematic Reviewand Meta-analysis
Junjie Wu1,3, Jie Liu1, Yuhao Zhou5, Jun Ying7, Houdong Zou6, Shicheng Guo1, Lei Wang4,Naiqing Zhao2, Jianjun Hu3, Daru Lu1, Li Jin1, Qiang Li3, and Jiu-Cun Wang1
AbstractPurpose: Published data have shown conflicting results about the relationship between X-ray repair
cross-complementing group 1 (XRCC1) gene polymorphisms (Arg399Gln and Arg194Trp) and clinical
outcome of platinum-based chemotherapy in patients with advanced non–small cell lung cancer (NSCLC).
A meta-analysis is needed to provide a systematic review of the published findings.
Experimental Design: We conducted a systematic review and meta-analysis to evaluate the predictive
value of XRCC1 gene polymorphisms on clinical outcome up to October 1, 2010. The quality of each study
was scored on the basis of predefined criteria.
Results:A total of 13 eligible follow-up studiesmet all the inclusion criteria. TheXRCC1194Trp allelewas
found to be significantly associated with a favorable response rate relative to 194Arg [Trp vs. Arg: OR, 1.88;
95% confidence interval (CI), 1.48–2.38]. XRCC1399Gln was less favorably associated with both response
rate (Gln vs. Arg: OR, 0.67; 95% CI, 0.52–0.87) and overall survival (Gln vs. Arg: HR, 1.30; 95% CI, 1.04–
1.63) than 399Arg in analyses using all available studies; but these associations became insignificant when
only high-quality studies were used.
Conclusion: These findings suggest a predictive role forXRCC1 genepolymorphisms in clinical outcome.
However, the role of 399Gln could be considered controversial because its impact on clinical outcome was
insignificant in high-quality studies. These findings show the importance of establishing suitable criteria,
including genetic epidemiologic, phenotypic, and clinical criteria, to improvequality control of studydesign
andmethods in pharmacogenomic studies related to XRCC1 gene polymorphism. Clin Cancer Res; 18(14);
3972–81. �2012 AACR.
IntroductionCurrently, themain conventional treatment for advanced
non–small cell lung cancer (NSCLC) is platinum-based
chemotherapy (Pt-chemo), which involves co-administra-tion of a platinum-containing drug and another cytotoxicagent (1). Despite improvements made to this treatmentover the past 2 decades, the response rates of chemotherapyregimens remains only 30% to 50% (2) and the 5-yearsurvival rate for NSCLC is still less than 15% (3).
X-ray repair cross-complementing group 1 (XRCC1) pro-tein plays a key role in base excision repair (BER). It serves asa scaffold protein in both single-strand break repair andbase excision repair activities (4). The amount of XRCC1transcription has shown a significant correlation with cis-platin resistance amongNSCLC cell lines (5). Another studyrevealed that XRCC1 protein could bind to platinum-con-taining DNA duplexes (6). These studies imply that XRCC1contributes to the repair of platinum-induced DNA dam-age. Arg399Gln and Arg194Trp are 2 common polymorph-isms in XRCC1. In erythrocytes, human placental aflatoxinB1 (AFB1-DNA) adducts have been shown to respond toenvironmental insults, and somatic glycophorin A (GPA)variants have been shown to respond to smoking. It hasbeen reported that the 399Gln allele is significantly
Authors' Affiliations: 1National Ministry of Education Key Laboratory ofContemporary Anthropology and State Key Laboratory of Genetic Engi-neering, School of Life Sciences, 2School of Public Health, Fudan Univer-sity; Departments of 3Pneumology and 4Cardiothoracic Surgery, ChanghaiHospital of Shanghai, 5Department of Health Statistics, 6Institute ofMilitaryHealth Service Management, Second Military Medical University, Shang-hai, China; and 7Department of Environmental Health, University of Cin-cinnati College of Medicine, Cincinnati, Ohio
Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).
Corresponding Authors: Qiang Li, Department of Pneumology, Chan-ghai Hospital, the Second Military Medical University, Shanghai200433, China. Phone: 86-21-81873231; Fax: 86-21-51190920; E-mail:[email protected]; and Jiu-Cun Wang, National Ministry of Edu-cation Key Laboratory of Contemporary Anthropology, School of LifeSciences, Fudan University, Shanghai 200433, China. Phone: 86-21-55665499; Fax: 86-21-556648845; E-mail: [email protected]
doi: 10.1158/1078-0432.CCR-11-1531
�2012 American Association for Cancer Research.
ClinicalCancer
Research
Clin Cancer Res; 18(14) July 15, 20123972
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
associated with higher levels of this type of genotoxicdamage (AFB1-DNA adducts and GPA variants), suggestingthat the Arg399Gln amino acid variant may alter the phe-notype of the XRCC1 protein, causing deficient DNA repair(7). XRCC1 codon 194 was shown to be a significantpredictor of progression-free survival. In a cohort of 229patients with NSCLCs who received radiotherapy, thepatients with haplotype pairs other than the homozygousTGG haplotype (194Trp-280Arg-399Arg) pairs survivedsignificantly longer than those with the homozygous TGGhaplotype pairs (8).The relationship between XRCC1 gene polymorphisms
(Arg399Gln and Arg194Trp) and clinical outcome(response rate and overall survival) in Pt-chemo foradvanced NSCLCs has been investigated extensively. How-ever to date, the evidence has been conflicting. In this study,we conducted a meta-analysis using all eligible studies toevaluate the association between the XRCC1 gene poly-morphisms and outcome in Pt-chemo for advancedNSCLCs.
Materials and MethodsData sources, search strategy, and selection of studiesEfforts were made to collect all published studies
related to the effects of XRCC1 gene polymorphisms onPt-chemo for advanced NSCLCs from various sources.Published articles were searched using databases(PubMed, Embase, and CNKI) up to October 1, 2010.Keywords such as "lung" or "pulmonary" and "cancer" or"carcinoma" and "XRCC1" or "X-ray cross-complement-ing group 1" or "base excision repair" or "BER" and"pharmacogenomics" or "polymorphism" or "variation"
were used in the searching process. Collected studies wereprescreened, and a study was excluded under either of thefollowing circumstances: (i) the study did not report anyclinical outcome; (ii) the clinical outcome reported in thestudy was either not specific to polymorphism or couldnot be attributed to a specific polymorphism; and (iii) theprincipal investigator declined or was unable to providerelevant information upon request. If the same researchgroup published multiple articles on the topic, we select-ed the article that used the most samples, used the mostrecent polymorphism data, or provided the most detailedinformation on each gene polymorphism (9, 10).
Quality assessmentThequality of each studywas also independently assessed
by 2 authors (Y. Zhou and H. Zou) using a predefined scale(Table 1). Our quality scoring criteria were followed fromother studies (11–13). The quality score of a given study(QSS) was determined using pharmacogenetic considera-tions and the following 6 factors: genotyping methods,platinum combinations, evaluation criteria, cancer stages,and sample size. With respect to genotyping methods, both
Translational RelevanceIn this report, we found that XRCC1 polymorphisms
in Arg194Trp could predict the clinical outcome ofplatinum-based chemotherapy for advanced non–smallcell lung cancer (NSCLC). The results, however, werebased upon studies using solely Chinese populations.The role of Arg399Gln was controversial, and itsrelationship to the response rate was found to be insig-nificant in high-quality studies, especially those usinghigh-quality genotyping methods. The importance ofconducting high-quality trials was confirmed in a studyof the effects of XRCC1 gene polymorphisms on clinicaloutcome in platinum-based chemotherapy for NSCLCs.Our study suggests that more studies using high-qualitygenotyping methods may be needed to confirm thepredictive roles of XRCC1 polymorphisms in differentpopulations. Our findings also show the importance ofestablishing suitable criteria, including genetic epidemi-ologic, phenotypic, and clinical criteria. These criteriamay help standardize the design of pharmacogenomicstudies and so aid in cancer research.
Table 1. Scale for quality assessment
Criteria Item Score
Evaluation criteriaa
WHO/RECIST 3Not detailed 0
Platinum combinationsa,b
One kind of platinum combinations 3TAX/TXT, DOC, GEM, or NVB 2Not detailed or other regimens 1
Stagea,b
Detailed 3Not detailed 0
Survivala
Original data 3Estimation of the log HR and variancefrom the Kaplan–Meier curves
1
Genotypinga,b
3D DNA microarray 3TaqMan 3PCR-RFLPc 2
Total sample sizea,b
�150 3>100 but <150 2�100 1
Abbreviations: 3D, 3-dimensional; RECIST, Response Eval-uation Criteria in Solid Tumors; WHO, World HealthOrganization.aCriteria for response rate.bCriteria for overall survival.cPCR-RFLP as genotyping method was categorized intosequencing or no sequencing.
XRCC1 Gene Polymorphisms and Platinum-Based Chemotherapy
www.aacrjournals.org Clin Cancer Res; 18(14) July 15, 2012 3973
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
DNA microarray and TaqMan were considered to be ofhigher quality than PCR-restriction fragment length poly-morphism (RFLP; refs. 14–17). PCR-RFLP was further cat-egorized intomethods that had been verified by sequencingand those that had not. For platinum combinations, weconsidered chemotherapy regimens, such as vinorelbine,gemzar, paclitaxel, and docetaxel combinedwith platinums(cisplatin or carboplatin) to be clinically equivalent (18–21). Newer agents such as vinorelbine (Navelbine) andpaclitaxel (Taxol) with cisplatin were considered as a moreeffective treatment of advanced disease than older regimensconsisting of cisplatin and a vinca alkaloid or a podophyl-lotoxin (22). The use of radiotherapy was also evaluatedcarefully during the rating of QSS (23). Total score rangedfrom 0 (worst) to 15 (best). A final QSS score was assignedto each study after consensus was reached between
reviewers. A study was considered low (or high) quality ifQSS < 10 (or �10).
Statistical analysisA total of 6 genetic models, with 3 main models (M1,
allele comparison, A vs. a; M2, recessive model, AA vs. Aaþaa; orM3, dominantmodel, AAþAa vs. aa) and 3models ofmultiple pairwise comparisons (M4, AA vs. aa;M5, Aa vs. aa;or M6, AA vs. Aa) were considered in this meta-analysis.Models M1 to M3 were considered primary genetic modelsof interest (24, 25). The ORs with 95% confidence intervals(CI) were estimated for response rate. The odds of responserate were defined as the ratio of complete or partial responseagainst stable or progressive disease. HRs and 95% CIswere estimated for 5-year survival, directly from the rawdata (26, 27), or indirectly from the Kaplan–Meier curve of
1,215 Potentially relevant articles
1,175 Excluded
129 Review
81 Other tumors
965 Other reasons
40 Evaluated articles in detail
17 Excluded
6 Abstract
4 Duplicate studies
7
23 Full text article analysis
13 Articles included in
meta-analysis
10 Excluded
4 SCLC
or including a non–platinum-based chemotherapy
6
Inclusion of
Insufficient information
Lung cancer susceptibility or radiotherapy
Figure 1. Literature search andselection of included studies.
Wu et al.
Clin Cancer Res; 18(14) July 15, 2012 Clinical Cancer Research3974
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
Tab
le2.
Cha
racteristic
sof
eligible
stud
iesco
nsidered
inthereport
QSS
Arg19
4Trp
Arg39
9Gln
Autho
r(Yea
r)Ethnicity
Sub
jects
Agemed
ian
(min–max
)Eva
luation
criterion
Metho
ds
IIIA,n
(%)IIIB,n
(%)IV,n
(%)
Others,
n(%
)Res
pons
erate
Ove
rall
survival
TrpTrp
TrpArg
ArgArg
Alle
licfreq
uenc
y%
(Trp)
GlnGln
GlnArg
ArgArg
Alle
licfreq
uenc
y%
(Gln)
Ref.
Gurub
haga
vatula
(200
4)
Cau
casian
s10
358
(32–
77)
—PCR-R
FLP
26(25.0)
30(29.0)
47(46.0)
——
9—
——
—10
b42
b51
b30
.10b
(23)
Wan
g(200
4)Chine
se10
556
(30–
74)
WHO
PCR-R
FLP
—17
(16.2)
41(39.0)
47(44.8)
9—
——
——
2/8a
9/33
a22
/31a
29.52a
(39)
DeLa
sPe~ n
as
(200
6)
Cau
casian
s13
562
(31–
81)
—Ta
qMan
—23
(17.0)
112(83.0)
——
14—
——
—18
b63
b49
b38
.08b
(27)
Yua
n(200
6)Chine
se20
056
(30–
74)
WHO
PCR-R
FLP
—49
(24.5)
151(75.5)
—13
—10
/13a
38/46a
24/69a
32.5
a—
——
—(38)
Gao
(200
6)Chine
se57
59(38–
77)
WHO
PCR-R
FLP
22(38.6)
34(59.6)
—12
—2/2a
12/11a
5/25
a27
.19a
0/3a
8/15
a11
/20a
25.44a
(36)
Son
g(200
7)Chine
se97
56(30–
68)
WHO
PCR-R
FLP
—38
(39.2)
59(60.8)
—11
——
——
—1/4a
11/29a
18/34a
25.77a
(10)
Son
g(200
7)Chine
se16
656
(30–
68)
WHO
PCR-R
FLP
—68
(41)
98(59.0)
—13
—4/12
a34
/32a
14/70a
29.52a
——
——
(9)
Kalikak
i(20
09)
Cau
casian
s11
961
(39–
85)
RECIST
PCR-R
FLP
6(5.0)
34(28.6)
79(66.4)
—11
9—
——
—10
b76
b11
/21a,3
3b40
.34b
(28)
Sun
(200
9)Chine
se87
59(34–
79)
WHO
3DDNA
microarray
——
87(100
.0)—
12—
5/6a
18/19a
8/31
a33
.91a
1/3a
8/22
a14
/39a
21.84a
(35)
Yao
(200
9)Chine
se10
861
(39–
79)
WHO
PCR-R
FLP
—37
(34.3)
71(65.7)
—12
12—
——
—9/48
a,6
0b12
/28a,
43b
1/4a,5
b75
.49a,
75.46b
(26)
Hon
g(200
9)Chine
se16
461
(27–
84)
WHO
PCR-R
FLP
100(61.3)
63(38.7)
14—
7/11
a31
/42a
19/54a
33.23a
3/10
a28
/53a
26/44a
32.62a
(34)
Yua
n(201
0)Chine
se19
956
(29–
74)
—PCR-R
FLP
43(21.6)
—15
6(78.4)
——
11—
——
—20
b74
b10
5b28
.64b
(29)
Qian(201
0)Chine
se10
761
(�55
),
46(<55
)
—PCR-R
FLP
—45
(42.1)
62(57.9)
—8
——
——
—2/6a
14/26a
32/27a
26.17a
(37)
Abbreviations
:3D,3
-dim
ension
al;R
ECIST,
Res
pon
seEva
luationCriteria
inSolid
Tumors;
WHO,W
orld
Hea
lthOrgan
ization.
aNum
berof
patientsforresp
onse
rate;infron
tof
obliq
uelineis
good
resp
onder
andbeh
indob
lique
lineis
poo
rresp
onder.
bNum
ber
ofpatientsforov
eralls
urviva
l.
XRCC1 Gene Polymorphisms and Platinum-Based Chemotherapy
www.aacrjournals.org Clin Cancer Res; 18(14) July 15, 2012 3975
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
Table 3. Analysis of the association between XRCC1Arg194Trp and response rate in main models
M1: Trp vs. Arg M2: TrpTrp vs. TrpArg þ ArgArg M3: TrpTrp þ TrpArg vs. ArgArg
Study groupsNo. studies(refs.)a
Random-effect;OR (95% CI) P I2 PQ
bRandom-effect;OR (95% CI) P I2 PQ
bRandom-effect;OR (95% CI) P I2 PQ
b
Overall 5 (9, 34–36, 38) 1.88 (1.48–2.38) <0.0001 0% 0.683 1.26 (0.75–2.09) 0.380 0% 0.848 2.91 (2.07–4.08) <0.0001 0% 0.419
Genotyping method
3D DNA microarray 1 (35) 2.15 (1.12–4.12) 0.020 — — 1.60 (0.45–1.75) 0.470 — — 3.56 (1.36–9.33) 0.010 — —
PCR-RFLP 4 (9, 34, 36, 38) 1.84 (1.42–2.37) <0.0001 0% 0.55 1.20 (0.69–2.09) 0.580 0% 0.75 2.87 (1.91–4.33) <0.0001 19% 0.300
PCR-RFLP
Sequencing 1 (34) 1.53 (0.95–2.46) 0.08 — — 1.22 (0.45–3.35) 0.7 — — 2.04 (1.04–3.98) 0.04 — —
No sequencing 3 (9, 36, 38) 1.98 (1.46–2.68) <0.0001 0% 0.53 1.19 (0.61–2.32) 0.61 0% 0.55 3.31 (2.03–5.39) <0.001 0.17 0.3
Tumor stage
Main of stage III 1 (34) 1.53 (0.95–2.46) 0.081 — — 1.22 (0.45–3.35) 0.697 — — 2.04 (1.04–3.98) 0.037 — —
Main of stage IV 4 (9, 35, 36, 38) 2.01 (1.53–2.65) <0.0001 0% 0.721 1.27 (0.70–2.29) 0.430 0% 0.712 3.29 (2.22–4.87) <0.0001 0% 0.486
Abbreviation: 3D, 3-dimensional.aThe detailed references are given in Table 2.bP value of heterogeneity.
Gao (2006) 2.96 (1.26–6.97) 7.7
Hong (2009) 1.53 (0.95–2.46) 25.0
Song (2007) 2.08 (1.27–3.41) 23.2
Sun (2009) 2.15 (1.12–4.12) 13.4
Yuan (2006) 1.72 (1.12–2.65) 30.6
Overall Overall
1.88 (1.48–2.38) 100.0
OR
0.3 1 5 15
Study OR (95% CI) Weight (%)
Study OR (95% CI) Weight (%)
Heterogeneity: χ2 = 2.29, df = 4 (P = 0.683); I2 = 0.0%
Test for overall effect: Z = 5.18 (P < 0.0001)
A
B
favoring 194Arg favoring 194Trp
favoring 399Arg favoring 399Gln
HighHigh quality uality
Gao (2006) 0.70 ( 0.28–1.77) 7.7
Song (2007) 0.73 ( 0.35–1.49) 12.7
Sun (2009) 0.99 ( 0.44–2.24) 10.0
Yao (2009) 0.62 ( 0.30–1.30) 12.3
Subtotal 0.77 ( 0.57–1.05) 70.3
LowLow quality uality
Qian (2010) 0.49 ( 0.26–0.92) 16.1
Wang (2004) 0.48 ( 0.24–0.96) 13.6
Subtotal 0.48 ( 0.30–0.77) 29.7
Overall
OR
0.3 1 5 15
Hong (2009) 0.82 ( 0.50–1.34) 27.6
0.67 ( 0.52–0.87) 100.0
Heterogeneity: χ2 = 0.83, df = 4 (P = 0.935); I 2 = 0.0%
Test for overall effect: Z = 1.65 (P = 0.098)
Heterogeneity: χ2 = 3.53, df = 6 (P = 0.740); I2 = 0.0%
Test for overall effect: Z = 3.04 (P = 0.002)
Heterogeneity: χ2 = 0.00, df = 1 (P = 0.965); I2 = 0.0%
Test for overall effect: Z = 3.04 (P = 0.002)
Figure 2. Forest plots forallele contrasts of XRCC1polymorphisms and clinicaloutcome in chemotherapystratified by study quality levels. A,ORs (and its 95% CI) of responserate between 194Trp and 194Arg.An OR > 1 (or <1) indicates that the194Trp is more (or less) likely toshow response than 194Arg. All 5studies were considered highquality in the analyses. B, ORs (andits 95% CI) of response ratebetween 399Gln and 194Arg. AnOR > 1 (or <1) indicates that the399Gln is more (or less) likely toshow response than 194Arg.
Wu et al.
Clin Cancer Res; 18(14) July 15, 2012 Clinical Cancer Research3976
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
an article (23, 28–30). Survival probabilities were tran-scribed using Grafula 3, version 2.10. For each trail (i) ineach time interval (t), the effective number of patients at risk[i.e., Ri(t)] and of deaths [i.e., Di(t)] for every arm (AAþ Aaand aAþ aa, respectively)were calculated using Ri(t) andDi(t) for different genotypes (AA, Aa, and aa).Data analyses were conducted as follows. First, Hardy–
Weinberg equilibrium (HWE) was assessed using a good-ness-of-fit test (c2 or the Fisher exact tests). Second,pooled allele frequencies for each ethnicity were thenestimated using a random-effects model with the inversevariance method based on the overall population datafrom each study (13). Third, the clinical outcome wasestimated, specifically the OR for response rate and theHR for overall survival using random-effect models withMantel–Haenszel statistics (31, 32). The heterogeneitybetween studies was investigated visually by scatter plotand estimated by a I2-statistic following the c2 test (33).The same statistical methods were applied in subanalysesusing stratified patient populations. All analyses wereconducted using STATA (version 10.0) except for thepooling of allele frequency, for which MetaAnalyst Ver-sion Beta 2.2 (Tufts Medical Center, Boston, MA) wasused. P values less than 0.05 were considered statisticallysignificant.
ResultsOverall 1,215 studies were selected during the first step
of systematic literature review, and a further review ofthe searched trials excluded 1,175 studies, including 129review articles, 81 studies on other tumors, and 965studies for other reasons. The remaining 23 studieswere identified through detailed assessment. In the end,
13 follow-up studies were considered to meet all ininclusion criteria (Fig. 1). These were included in finalanalyses (Fig. 1; refs. 9, 10, 23, 26–29, 34–39). Theyincluded 1,647 individuals. The baseline characteristicsof the included studies are given in Table 2. Three of thesestudies were conducted on Caucasian patients, and 10were conducted on Chinese patients. Six were publishedin English-language journals (23, 26–29, 35). Seven werepublished in Chinese-language journals (9, 10, 34, 36–39). The sample size of each report ranged from 57 to 200individuals. The quality score for studies of genetic con-trasts between Arg194Trp and Arg399Gln and thepatient’s response rate ranged from 12 to 14 and 8 to14, respectively, with 100% (5 of 5) and 75% (6 of 8) ofthe trials classified as high-quality. Studies of the contrastsbetween Arg399Gln and overall survival ranged from 9to 14 in quality score, with 60% (3 of 5) classified as high-quality. A total of 11 studies were reported using PCR-RFLP genotyping methods. Genotypes were verified bysequencing in all samples in one study (34), partiallyverified in 20% of samples in one study (28), and notverified in the rest of the studies. All of these studies usedsamples of peripheral blood.
Allele frequenciesTable 2 shows the distribution of XRCC1 genotypes with
respect to response rate and overall survival rate. It alsoshows the distribution of the XRCC1 allele frequencies.Using the frequencies of XRCC1 genotypes, all populationswere found to be inHWE except the one studied by Kalikaki(P ¼ 0.00028; ref. 28). This HWE-deviant population eval-uated by Kalikaki was not excluded from the study becauseno genotyping error was detected by PCR-RFLP combinedwith sequencing (28).
Figure 2. (Continued ) C, HRs(and its 95% CI) of 5-year survivalbetween 399Gln and 194Arg. An HR> 1 (or <1) indicates that the 399Gln isless (or more) likely to show survivalthan 194Arg.
C
favoring 399Arg favoring 399Gln0.5 1 2 5
High quality High quality
0.90 ( 0.66–1.23) 20.1
Yao (2009) 1.42 ( 1.06–1.89) 21.1
Yuan (2010) 1.73 ( 1.19–2.52) 17.0
Subtotal 1.29 ( 0.89–1.87) 58.2
Low quality Low quality
1.62 ( 1.15–2.28) 18.4
1.13 ( 0.88–1.45) 23.4
Subtotal 1.32 ( 0.93–1.88) 41.8
Overall 1.30 ( 1.04–1.63) 100.0
Study HR (95% CI) Weight (%)
Gurubhagavatula (2004)
De Las Peñas (2006)
Kalikaki (2009)
Heterogeneity: χ2 = 7.80, df = 2 (P = 0.020); I2 = 74.4%
Test for overall effect: Z = 1.36 (P = 0.175)
Heterogeneity: χ2 = 2.75, df = 1 (P = 0.097); I2 = 63.7%
Test for overall effect: Z = 1.57 (P = 0.116)
Heterogeneity: χ2 = 10.56, df = 4 (P = 0.032); I2 = 62.1%
Test for overall effect: Z = 2.30 (P = 0.022)
HR
XRCC1 Gene Polymorphisms and Platinum-Based Chemotherapy
www.aacrjournals.org Clin Cancer Res; 18(14) July 15, 2012 3977
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
The pooled frequency for 194Trp with respect toresponse rate was 31.72% (29.28%–34.25%) and thatfor 399Gln was 34.05% (22.72%–47.55%) in Chinesepatients. Heterogeneity was only observed in the pooledfrequency for 399Gln with respect to response rate instudies of Chinese patients (P < 0.0001). For Caucasianpatients, only the frequencies for 399Gln allele wereavailable for analysis, and the frequency in response ratewas 40.68% (34.59%–47.06%). The frequency in overallsurvival was 36.28% (30.62%–42.34%), which droppedto 34.19% after excluding the Kalikaki study amongCaucasian patients. No heterogeneity was found in thepooled frequencies in Caucasians.
Gene effectsData concerning the predictive value of XRCC1
Arg194Trp with respect to the sensitivity of lung cancerto platinum-based treatment were available in 5 trials(9, 34–36, 38). These covered 674 individuals. Table 3 andSupplementary Table S1 show that the 194Trp allele("increasing" allele) wasmore closely associatedwith betterresponse rates than the 194Arg allele. This indicates that the194Trp allele may be indicative of better response rates toplatinum-based treatment than the 194Arg allele (Fig. 2A).No significant heterogeneity was detected among the pre-dictive values from the 5 Chinese studies (Table 3 andSupplementary Table S1).
Data concerning the predictive value of the 399Gln allelewith respect to the resistance of lung cancer to platinum-based treatment were available from 8 studies covering atotal of 837 individuals (10, 26, 28, 34–37, 39). Table 4 andSupplementary Table S2 show an association between the399Gln allele ("increasing" allele) and response rate relativeto the 399Arg allele. Results suggest that the 399Gln alleleis associated with a poorer response rate to platinum-basedtreatment than the 399Arg allele (Fig. 2B). There was noevidence of heterogeneity with respect to predictive value(Table 4 and Supplementary Table S2).
Five studies covering a total of 659 patients for HRs wererecorded (23, 26–29). There was a statistically significantdifference between the association between the 399Glnallele and lower rates of 5-year survival and that betweenthe 399Arg allele and higher rates of survival. This suggeststhat the 399Gln allele is more closely associated withshorter survival time than the 399Arg allele (Fig. 2C).Significant heterogeneity was detected when these 5 studieswere combined. It was resolved by stratification analysiswith respect to the degree of PCR-RFLP or QSS. Similarresults were also obtained in anM1model when the studieswere stratified with respect to whether or not the PCR-RFLPmethod was used (Table 4 and Supplementary Table S3).
Subgroup analyses were conducted with respect toresponse rate (Tables 3 and 4 and Supplementary TablesS1 and S2). It was noted that the relationship between399Gln and response rate disappeared after the low-qualitystudies were excluded (Fig. 2B). This effect was especiallynoticeable in studies with low-quality genotyping methods(i.e., PCR-RFLP) andunsequenced subgroups. Similar results
were observed in patients from populations where stage IIItumors were common. The pooled ORs for 399Gln weresignificant in Chinese but not in Caucasian individuals. ThepooledHR for399Glnwas also found tobe insignificantwithrespect to overall survival, after studies that had used low-quality genotyping methods were excluded (Fig. 2C). Nosignificant differences were identified with respect to anyassociation between clinical outcome and other subgroups(Tables 3 and 4 and Supplementary Tables S1–S3).
DiscussionOur results showed that XRCC1 194Trp allele was pos-
itively associated with the response rate relative to 194Arg,and XRCC1 399Gln allele was negatively associated withboth the response rate and overall survival relative to399Arg. A previous study showed XRCC1 Arg399Gln to beassociated with the clinical outcome of chemotherapy inpatients with lung cancer (40). These findings, however,could be confounded by including both patients withNSCLCs and small cell lung cancer (SCLC) in the analysis,indicating that the study was biased and the results over-estimated. NSCLCs and SCLCs are different in terms ofdoubling time, metastasis, sensitivity to initial chemother-apy, and survival rate (1, 41). Our study using patients withNSCLCs only showed that the overall survival was unaf-fected by XRCC1 Arg399Gln with a relative risk of death(95% CI) of 0.85 (0.26–2.73).
Heterogeneity was detected in the analysis of XRCC1Arg399Gln to overall survival. It was also noted in the alleleanalysis. One study showed a Gln frequency of 75.46%,much higher than the range of 21.84% to 40.34% seen inother studies (26). In addition, the population evaluated inone study did not show HWE (28). The existence of het-erogeneity indicated variability, which may have beencaused by different characteristics, such as ethnicity, cancerstage, or method of genotyping used among patient popu-lations. Hence, stratified analyses of subpopulations areneeded to reduce such variability.
It is not uncommon that quality of studies (or trials) mayvary in meta-analyses of genetic association studies ingenetic epidemiology (11, 42, 43). In this article, we eval-uated each study using a QSS and provided subanalyses onhigh-quality studies. Both response to chemotherapy andoverall survival showed insignificant associations toXRCC1Arg399Gln, after excluding a few lower quality trials. Ourfindings suggest that the role of XRCC1 Arg399Gln inclinical outcomes might need to be investigated morecarefully in future studies incorporating more criteria inthe design and experimentation to ensure a more accurateand robust conclusion.
The overall survival was associated with the polymorph-isms in stage III but not in stage IV (23, 27). Such phenom-ena could be interpreted using double-edged sword theory(44, 45). However, overactive nucleotide excision repairand DNA replication systems may achieve clinically rele-vant outcomes, causing both favorable prognosis and drugresistance. Some researchers have found this to be
Wu et al.
Clin Cancer Res; 18(14) July 15, 2012 Clinical Cancer Research3978
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
Tab
le4.
Ana
lysisof
theas
sociationbetwee
nXRCC1Arg39
9Gln
andresp
onse
rate
andov
eralls
urviva
linmainmod
els
M1:G
lnvs
.Arg
M2:G
lnGln
vs.G
lnArg
þArgArg
M3:G
lnGln
þGlnArg
vs.A
rgArg
Studygroup
sNo.s
tudies(refs.)a
Ran
dom-effec
t(95%
CI)
PI2
PQb
Ran
dom-effec
t(95%
CI)
PI2
PQb
Ran
dom-effec
t(95%
CI)
PI2
PQb
Res
pon
serate
(OR)
Ove
rall
7(10,
26,34
–37
,39
)c0.67
(0.52–
0.87
)0.00
20%
0.74
0.49
(0.27–
0.88
)0.01
70%
0.99
70.67
(0.49–
0.92
)0.01
30%
0.70
4Pop
ulation
Cau
casian
s0c
——
——
——
——
0.83
(0.35–
1.96
)0.66
7—
—
Chine
se7(10,
26,34
–37
,39
)0.67
(0.52–
0.87
)0.00
20%
0.74
0.49
(0.27–
0.88
)0.01
70%
0.99
70.65
(0.46–
0.91
)0.01
20%
0.62
7QSS �10
5(10,
26,34
–36
)c0.77
(0.57–
1.05
)0.09
80%
0.94
0.50
(0.25–
1.00
)0.04
90%
0.97
40.83
(0.57–
1.21
)0.32
70%
0.99
6<1
02(37,
39)
0.48
(0.30–
0.77
)0.00
20%
0.97
0.45
(0.14–
1.41
)0.17
0%0.80
10.40
(0.22–
0.72
)0.00
20%
0.85
4Gen
otyp
ingmetho
d3D
DNAmicroarray
1(35)
0.99
(0.44–
2.24
)0.98
——
0.92
(0.09–
9.36
)0.95
——
1.00
(0.38–
2.66
)1
——
PCR-R
FLP
6(10,
26,34
,36–
37,3
9)c
0.64
(0.49–
0.84
)0.00
10%
0.77
0.47
(0.25–
0.86
)0.01
0%1
0.64
(0.46–
0.89
)0.00
80%
0.69
PCR-R
FLP
Seq
uenc
ing
1(34)
c0.82
(0.50–
1.34
)0.43
——
0.54
(0.14–
2.04
)0.36
——
0.83
(0.49–
1.39
)0.48
0%0.99
Nose
que
ncing
5(10,
26,36
–37
,39
)0.58
(0.42–
0.80
)9E
-04
0%0.88
0.45
(0.23–
0.89
)0.02
0%0.99
0.53
(0.34–
0.82
)0.00
40%
0.7
Tumor
stag
ed6(10,
26,34
–37
)c0.71
(0.54–
0.93
)0.01
50%
0.78
0.48
(0.26–
0.91
)0.02
40%
0.98
90.73
(0.52–
1.03
)0.07
10%
0.84
7Mainof
stag
eIII
1(34)
0.82
(0.50–
1.34
)0.43
1—
—0.54
(0.14–
2.04
)0.36
3—
—0.83
(0.44–
1.59
)0.58
——
Mainof
stag
eIV
5(10,
26,35
–37
)c0.66
(0.47–
0.92
)0.01
50%
0.75
0.47
(0.23–
0.96
)0.03
80%
0.96
80.70
(0.47–
1.04
)0.07
50%
0.78
1Ove
rallsu
rvival
(HR)
Ove
rall
5(23,
26–29
)1.30
(1.04–
1.63
)0.02
262
%0.03
2.68
(1.23–
5.82
)0.01
392
%<0
.000
11.09
(0.86–
1.37
)0.48
214
%0.32
6Pop
ulation
Cau
casian
s3(23,
27,28
)1.17
(0.86–
1.58
)0.30
967
%0.05
3.65
(1.36–
9.80
)0.01
89%
<0.000
11.10
(0.73–
1.65
)0.64
553
%0.12
Chine
se2(26,
29)
1.52
(1.21–
1.92
)<0
.000
10%
0.41
1.67
(1.20–
2.33
)0.00
30%
0.85
61.08
(0.78–
1.48
)0.65
70%
0.52
9QSS �10
3(26,
27,29
)1.29
(0.89–
1.87
)0.17
574
%0.02
1.60
(1.18–
2.16
)0.00
20%
0.82
20.95
(0.73–
1.24
)0.71
23%
0.35
8<1
02(23,
28)
1.32
(0.93–
1.88
)0.11
664
%0.1
6.49
(5.07–
8.32
)<0
.000
10%
0.51
1.36
(0.96–
1.94
)0.08
70%
0.87
2Gen
otyp
ingmetho
dTa
qMan
1(27)
0.90
(0.66–
1.23
)0.51
4—
—1.32
(0.65–
2.66
)0.44
3—
—0.75
(0.48–
1.17
)0.20
6—
—
PCR-R
FLP
4(23,
26,28
–29
)1.41
(1.16–
1.71
)0.00
137
%0.19
3.16
(1.35–
7.44
)0.00
893
%<0
.000
11.20
(0.94–
1.52
)0.13
90%
0.71
5PCR-R
FLP
Seq
uenc
ing
1(28)
1.13
(0.88–
1.45
)0.33
3—
—6.68
(5.14–
8.68
)<0
.000
1—
—1.32
(0.79–
2.20
)0.28
3—
—
Nose
que
ncing
3(23,
26,29
)1.55
(1.28–
1.88
)<0
.000
10%
0.69
2.31
(1.22–
4.37
)0.01
70%
0.03
51.16
(0.89–
1.52
)0.26
90%
0.55
7Stage Mainof
stag
eIII
1(23)
1.62
(1.15–
2.28
)0.00
6—
—5.08
(2.35–
10.99)
<0.000
1—
—1.40
(0.86–
2.29
)0.18
——
Mainof
stag
eIV
4(26–
29)
1.24
(0.96–
1.59
)0.09
564
%0.04
2.31
(0.92–
5.78
)0.07
494
%<0
.000
11.02
(0.80–
1.31
)0.86
410
%0.34
4
Abbreviation:
3D,3
-dim
ension
al.
aTh
edetailedreferenc
esaregive
nin
table
2.bPva
lueof
heteroge
neity
.cM
3includ
edan
addition
alstud
y[th
eworkof
Kalikak
iand
colleag
ues(re
f.28
)]in
analyses
.dTh
ereis
nodetailedinform
ationof
tumor
stag
ein
onestud
y[th
eworkof
Wan
gan
dco
lleag
ues(re
f.39
)]an
dthestud
ywas
exclud
ed.
XRCC1 Gene Polymorphisms and Platinum-Based Chemotherapy
www.aacrjournals.org Clin Cancer Res; 18(14) July 15, 2012 3979
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
reminiscent of the Roman god of entrances and exits, Janus,who is usually depicted with 2 faces pointing in oppositedirections (3). In our study, we found 399Gln to be neg-atively related to response to chemotherapy in patients whoweremostly in stage IV, but not in patients whoweremostlyin stage III. This suggests that 399Gln might have a distinctassociation with drug resistance in patients at differentstages of the disease. The association between 399Gln andoverall survival was found to be insignificant even afterstratification of cancer stages. This was due to a lack oforiginal staging data in each paper recruited into this meta-analysis. For this reason, the conclusions drawn in thismeta-analysis about the tumor stage subgroup should beweighed with caution. We also noted different outcomeswith respect to 399Gln, response rate, and ethnicity.
The study has some limitations. In some of the subana-lyses on stratified groups of patients, therewas only one trialavailable and hence the variability across trials could not beassessed. Some of the findings in subgroups may have beenundervalued because of the smaller sample size available foranalyses. Among all 13 trials used in meta-analysis, only 3were conducted upon Caucasian populations. None of theCaucasian trials were available for estimation of theresponse rate using the allelemodel (M1). Due to the natureof meta-analysis, the accuracy of inference and statisticalpower were usually limited because analyses could only beconducted on secondary data, other than the original datacollected directly from individual patients. In addition, thequality of trials could not be controlled directly by research-ers conducting the meta-analysis.
SummaryThese findings show a predictive role for XRCC1 gene
polymorphisms in clinical outcome. However, the pre-dictive role of 399Gln with respect to clinical outcomewas here found to differ by study quality, suggesting thatconsistent study quality is important and can be assessedusing the criteria established in our meta-analysis. Thesecriteria include genetic epidemiologic, phenotypic, and
clinical variables. To our knowledge, this quality assess-ment system was designed for the first time in pharma-cogenomic studies. Its feasibility was confirmed in ourstudy. These criteria will help standardize study designand may affect future pharmacogenomic studies in thefield of cancer research. More larger studies on patients ofdifferent ethnicities, especially studies stratified for inter-actions between tumor stage, genotyping method, andclinical outcome, should be conducted to confirm thepredictive roles of XRCC1 Arg399Gln and Arg194Trp.Functional studies may help validate the effects of thesepolymorphisms in Pt-chemo.
Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.
AcknowledgmentsThe authors thank Dr. Rafael Rosell from the Institut Catal�a Oncologia,
Badalona, Barcelona, Spain; Mar��a S�anchez Ronco from Departamento deMedicina Preventivay Salud P�ublica, Universidad Aut�onoma de Madrid,Madrid, Spain; andDr. Bo Shen andDr. ChengyunYao from theDepartmentof Chemotherapy, Jiangsu Cancer Hospital and Research Institute, NanjingMedical University, Nanjing, Jiangsu, China, for their kindness in providingus with original data. They also thank Jia He and Yingyi Qin from theDepartment of Health Statistics, Second Military Medical University, Shang-hai, China, for their helpful statistical advice and the processing of digitalgraphics; Wengsheng Guo from Division of Biostatistics, University ofPennsylvania School of Medicine, Philadelphia, PA; Lili Yan from the StateKey Laboratory of Genetic Engineering, Institute of Genetics, School of LifeScience, Fudan University, Shanghai, China; and Dr. Bobing Chen from theDepartment of Hematology, Huashan Hospital, Fudan University, for theirhelpful statistical advice.
Grant SupportThis studywas supported by grants from theNational Science Foundation
of China, grant numbers 30971594 and 30890034; the Science and Tech-nology Committee of Shanghai Municipality, grant number 09XD1400200;and the Major National Science and Technology Program of China, grantnumber 2008ZX10002-002.
The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.
Received June 17, 2011; revised May 9, 2012; accepted May 14, 2012;published OnlineFirst June 15, 2012.
References1. National Comprehensive Cancer Network, Inc. 2010 [homepage on the
Internet]. Fort Washington, PA. Available from: http://www.nccn.org/.2. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Murray T, et al. Cancer
statistics, 2008. CA Cancer J Clin 2008;58:71–96.3. Gazdar AF. DNA repair and survival in lung cancer-the two faces of
Janus. N Engl J Med 2007;356:771–3.4. Lindahl T, Wood RD. Quality control by DNA repair. Science
1999;286:1897–905.5. Weaver DA, Crawford EL, Warner KA, Elkhairi F, Khuder SA, Willey JC.
ABCC5, ERCC2, XPA and XRCC1 transcript abundance levels corre-late with cisplatin chemoresistance in non-small cell lung cancer celllines. Mol Cancer 2005;4:18.
6. Zhu G, Lippard SJ. Photoaffinity labeling reveals nuclear proteins thatuniquely recognize cisplatin-DNA interstrand cross-links. Biochemis-try 2009;48:4916–25.
7. Lunn RM, Langlois RG, Hsieh LL, Thompson CL, Bell DA. XRCC1polymorphisms: effects on aflatoxin B1-DNA adducts and glycophorinA variant frequency. Cancer Res 1999;59:2557–61.
8. Yoon SM, Hong YC, Park HJ, Lee JE, Kim SY, Kim JH, et al. Thepolymorphism and haplotypes of XRCC1 and survival of non-small-cell lung cancer after radiotherapy. Int J Radiat Oncol Biol Phys2005;63:885–91.
9. SongD, Liu J,WangZ, SongB, Li C. [Single nucleotide polymorphismsin XRCC1 and XPD and clinical response to platin-based chemother-apy in advanced non-small cell lung cancer]. Chin J Gerontol2007;27:1684–6.
10. SongD, Liu J,Wang Z, SongB, Li C. [Single nucleotide polymorphismsin XRCC1 and clinical response to platin-based chemotherapy inadvanced non-small cell lung cancer]. Cancer Res Prev Treat2007;34:845–7.
11. Attia J, Thakkinstian A, D'Este C. Meta-analyses of molecular asso-ciation studies: methodologic lessons for genetic epidemiology. J ClinEpidemiol 2003;56:297–303.
12. Jiang DK, Ren WH, Yao L, Wang WZ, Peng B, Yu L. Meta-analysis ofassociation between TP53 Arg72Pro polymorphism and bladder can-cer risk. Urology 2010;76:761–5.
Wu et al.
Clin Cancer Res; 18(14) July 15, 2012 Clinical Cancer Research3980
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531
13. Thakkinstian A, McEvoy M, Minelli C, Gibson P, Hancox B, Duffy D,et al. Systematic review andmeta-analysis of the association betweenb2-adrenoceptor polymorphisms and asthma: a HuGE review. Am JEpidemiol 2005;162:201–11.
14. Wang Y, Zhang D, Zheng W, Luo J, Bai Y, Lu Z. Multiple genemethylation of nonsmall cell lung cancers evaluated with 3-dimen-sional microarray. Cancer 2008;112:1325–36.
15. Xiao P, Huang H, Zhou G, Lu Z. Gel immobilization of acrylamide-modified single-stranded DNA template for pyrosequencing. Electro-phoresis 2007;28:1903–12.
16. WanY,WangY, Luo J, Lu Z. Bisulfitemodification of immobilizedDNAsfor methylation detection. Biosens Bioelectron 2007;22:2415–21.
17. Xiao PF, Cheng L, Wan Y, Sun BL, Chen ZZ, Zhang SY, et al. Animproved gel-based DNA microarray method for detecting singlenucleotide mismatch. Electrophoresis 2006;27:3904–15.
18. Scagliotti GV, De Marinis F, Rinaldi M, Crino L, Gridelli C, Ricci S,et al. Phase III randomized trial comparing three platinum-baseddoublets in advanced non-small-cell lung cancer. J Clin Oncol2002;20:4285–91.
19. Zatloukal P, Petruzelka L, ZemanovaM,KolekV, Skrickova J, PesekM,et al. Gemcitabine plus cisplatin vs. gemcitabine plus carboplatin instage IIIb and IV non-small cell lung cancer: a phase III randomized trial.Lung Cancer 2003;41:321–31.
20. Schiller JH, Harrington D, Belani CP, Langer C, Sandler A, Krook J,et al. Comparison of four chemotherapy regimens for advancednon-small-cell lung cancer. N Engl J Med 2002;346:92–8.
21. Kelly K, Crowley J, Bunn PJ, Presant CA, Grevstad PK, Moinpour CM,et al. Randomized phase III trial of paclitaxel plus carboplatin versusvinorelbine plus cisplatin in the treatment of patients with advancednon-small-cell lung cancer: a Southwest Oncology Group trial. J ClinOncol 2001;19:3210–8.
22. Johnson DH, Chang AY, Ettinger DS, Kim KM, Bonomi P. Recentadvances with chemotherapy for NSCLC: the ECOG experience.Eastern Cooperative Oncology Group. Oncology (Williston Park)1998;12:67–70.
23. Gurubhagavatula S, Liu G, Park S, Zhou W, Su L, Wain JC, et al. XPDand XRCC1 genetic polymorphisms are prognostic factors inadvanced non-small-cell lung cancer patients treated with platinumchemotherapy. J Clin Oncol 2004;22:2594–601.
24. Zintzaras E, Lau J. Synthesis of genetic association studies for per-tinent gene-disease associations requires appropriatemethodologicaland statistical approaches. J Clin Epidemiol 2008;61:634–45.
25. Minelli C, Thompson JR, Abrams KR, Thakkinstian A, Attia J. Thechoice of a genetic model in the meta-analysis of molecular associ-ation studies. Int J Epidemiol 2005;34:1319–28.
26. Yao CY, Huang XE, Li C, Shen HB, Shi MQ, Feng JF, et al. Lack ofinfluence of XRCC1 and XPD gene polymorphisms on outcome ofplatinum-based chemotherapy for advanced non small cell lung can-cers. Asian Pac J Cancer Prev 2009;10:859–64.
27. de Las Pe~nas R, Sanchez-Ronco M, Alberola V, Taron M, Camps C,Garcia-Carbonero R, et al. Polymorphisms in DNA repair genes mod-ulate survival in cisplatin/gemcitabine-treated non-small-cell lungcancer patients. Ann Oncol 2006;17:668–75.
28. Kalikaki A, Kanaki M, Vassalou H, Souglakos J, Voutsina A, Georgou-lias V, et al. DNA repair gene polymorphisms predict favorable clinical
outcome in advanced non-small-cell lung cancer. Clin Lung Cancer2009;10:118–23.
29. Yuan P, Liu L, Wu C, Zhong R, Yu D, Wu J, et al. No associationbetween XRCC1 polymorphisms and survival in non-small-cell lungcancer patients treated with platinum-based chemotherapy. CancerBiol Ther 2010;10:854–9.
30. Parmar MK, Torri V, Stewart L. Extracting summary statistics toperform meta-analyses of the published literature for survival end-points. Stat Med 1998;17:2815–34.
31. Ades AE, LuG, Higgins JP. The interpretation of random-effectsmeta-analysis in decision models. Med Decis Making 2005;25:646–54.
32. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control ClinTrials 1986;7:177–88.
33. Deeks J, Higgins J, Altman D. Analyzing data and undertaking meta-analyses. In: Higgins J, Green S, editors. Cochrane handbook forsystematic reviews of interventions 5.0.1. Oxford, UK: The CochraneCollaboration; 2008. p. 9.1–9.43.
34. HongC, XuQ, Yue Z, Zhang Y, Yuan Y. [Correlation of the sensitivity ofNP chemotherapy in non-small lung cancer with DNA repair geneXRCC1 polymorphism]. Ai Zheng 2009;28:1291–7.
35. SunX, Li F, SunN,ShukuiQ, BaoanC, Jifeng F, et al. Polymorphisms inXRCC1 and XPG and response to platinum-based chemotherapy inadvanced non-small cell lung cancer patients. Lung Cancer2009;65:230–6.
36. Gao C, Shi M, Wu J, Cao H, Feng J, Xu L. [Polymorphisms in XRCC1gene and sensitivity to gemcitabine/cisplatin chemotherapy in nonsmall cell lung cancer]. Pract J Cancer 2006;21:351–3.
37. Qian X, Qiu L, Yang Y, JiangM, Zhang Y, Yu L, et al. [Predictive value ofbase-excision repair gene polymorphisms in advanced non-small celllung cancer patients receiving platinum-based chemotherapy]. JMod-ern Oncol 2010;18:1303–5.
38. YuanP,Miao X, Zhang X,Wang Z, TanW, SunY, et al. [XRCCl andXPDgenetic polymorphisms predict clinical responses to platinum-basedchemotherapy in advanced non-small cell lung cancer]. ZhonghuaZhong Liu Za Zhi 2006;28:196–9.
39. Wang Z, Miao X, Tan W, Zhang X, Xu B, Lin D. [Single nucleotidepolymorphisms in XRCC1 and clinical response to platin-based che-motherapy in advanced non-small cell lung cancer]. Chin J Cancer2004;23:865–8.
40. Horgan AM, Yang B, Azad AK, Amir E, John T, Cescon DW, et al.Pharmacogenetic and germline prognostic markers of lung cancer. JThorac Oncol 2011;6:296–304.
41. SimonM,Argiris A,Murren JR.Progress in the therapyof small cell lungcancer. Crit Rev Oncol Hematol 2004;49:119–33.
42. Funke S, Brenner H, Chang-Claude J. Pharmacogenetics in colo-rectal cancer: a systematic review. Pharmacogenomics 2008;9:1079–99.
43. Wei SZ, Zhan P, ShiMQ, Shi Y, QianQ, Yu LK, SongY. Predictive valueof ERCC1andXPDpolymorphism inpatientswith advancednon-smallcell lungcancer receivingplatinum-based chemotherapy: a systematicreview and meta-analysis. Med Oncol 2011;28:315–21.
44. Rosell R, Lord RV, Taron M, Reguart N. DNA repair and cisplatinresistance in non-small-cell lung cancer. Lung Cancer 2002;38:217–27.
45. Wei Q, Frazier ML, Levin B. DNA repair: a double-edged sword. J NatlCancer Inst 2000;92:440–1.
XRCC1 Gene Polymorphisms and Platinum-Based Chemotherapy
www.aacrjournals.org Clin Cancer Res; 18(14) July 15, 2012 3981
American Association for Cancer Research Copyright © 2012 on January 31, 2013clincancerres.aacrjournals.orgDownloaded from
Published OnlineFirst June 15, 2012; DOI:10.1158/1078-0432.CCR-11-1531