KEAP1 Genetic Polymorphisms Associate with Breast Cancer ...clincancerres.aacrjournals.org/content/clincanres/21/7/1591.full.pdf · Breast Cancer Risk and Survival Outcomes ... 4Laboratory

Personalized Medicine and Imaging

KEAP1 Genetic Polymorphisms Associate withBreast Cancer Risk and Survival OutcomesJaana M. Hartikainen1, Maria Tengstr€om2,3, Robert Winqvist4,5, Arja Jukkola-Vuorinen6,Katri Pylk€as4,5, Veli-Matti Kosma1,7, Ylermi Soini1,7, and Arto Mannermaa1,7

Abstract

Purpose: Defective oxidative stress response may increasecancer susceptibility. In tumors, these rescue mechanisms maycause chemo- and radioresistance impacting patient outcome.We previously showed that genetic variation in the nuclearfactor erythroid 2–related factor 2 (NFE2L2) is associated withbreast cancer risk and prognosis. Here we further studied thispathway by investigating Kelch-like ECH-associated protein 1(KEAP1).

Experimental Design: Five tagging SNPs in the KEAP1 genewere genotyped in 996 breast cancer cases and 880 controls fromtwo Finnish case–control sets. KEAP1 protein expression wasstudied in 373 invasive breast cancer tumors.

Results: rs34197572 genotype TT was associated withincreased risk of breast cancer in the KBCP samples [P ¼1.8�10�4; OR, 7.314; confidence interval (CI), 2.185–24.478].rs11085735 allele A was associated with lower KEAP1 proteinexpression (P ¼ 0.040; OR,¼ 3.545) and high nuclear NRF2expression (P ¼ 0.009; OR, 2.445) and worse survival in all

invasive cases (P¼ 0.023; HR, 1.634). When including treatmentdata, rs11085735 was associated with recurrence-free survival(RFS; P ¼ 0.020; HR, 1.545) and breast cancer–specific survival(P¼ 0.016;HR, 1.683) and rs34197572with overall survival (P¼0.045; HR, 1.304). rs11085735 associated with RFS also amongtamoxifen-treated cases (P ¼ 0.003; HR, 3.517). Among radio-therapy-treated cases, overall survival was associated withrs34197572 (P ¼ 0.018; HR, 1.486) and rs8113472 (P ¼0.025; HR, 1.455). RFS was associated with rs9676881 (P ¼0.024; HR, 1.452) and rs1048290 (P¼ 0.020; HR, 1.468) amongall invasive cases and among estrogen receptor (ER)-positivetamoxifen-treated cases (P ¼ 0.018; HR, 2.407 and P ¼ 0.015;HR, 2.476, respectively).

Conclusions: The present findings suggest that the investigatedSNPs have effects related to oxidative stress induced by cancertreatment, supporting involvement of the NRF2/KEAP1 pathwayin breast cancer susceptibility and patient outcome. Clin Cancer Res;21(7); 1591–601. �2015 AACR.

IntroductionRadiation, inflammation, and various chemicals can lead to

oxidative stress, which in turn causes DNA changes and maylead to cancer development (1, 2). Antioxidative enzymes—including superoxide dismutases (SOD), thioredoxins andperoxiredoxins—and nonenzymatic antioxidants contribute to

combating oxidative stress (1, 2). The main mediator ofcellular adaptation to oxidative and xenobiotic stress is theNuclear factor erythroid 2–related factor 2 protein (NRF2,encoded by NFE2L2). NRF2 normally resides in the cytoplasmand forms a complex with Kelch-like ECH-associated protein 1(KEAP1), which directs NRF2 through ubiquitination to pro-teasomal degradation (3). In the event of oxidative stress,NRF2 is released from the complex and moves to the nucleus.In the nucleus, NRF2 heterodimerizes with small Maf or c-Junand binds antioxidant response elements (ARE) in down-stream genes, thus inducing the expression of several cytopro-tective genes—including detoxifying enzymes, drug transpor-ters, antiapoptotic proteins, and proteasomes (3). Once redoxbalance (basal state) is returned to the cell, KEAP1 moves tothe nucleus, binds to NRF2 by its ETGE domain, and takes itback to the cytoplasm for degradation (3, 4). Several factorsreportedly interrupt the NRF2/KEAP1 interaction, inducing theNRF2 accumulation, and thus interfering with reactive oxygenspecies (ROS) scavenging (5–11). Recent findings suggest thatenhanced ROS detoxification through additional NRF2 func-tions may be also protumorigenic (12).

In cancer cells, large amounts of unbound NRF2 can result inchemoresistance (13). Tumor cells may contain mutations inthe NFE2L2 and KEAP1 genes, underlining the importance ofNRF2 and KEAP1 in tumorigenesis. Somatic mutations inKEAP1 predominate in lung cancer (14, 15), whereas NFE2L2mutations occur with a 6% to 13% frequency in esophageal,skin, larynx, and lung cancer, with the highest prevalence found

1School of Medicine, Institute of Clinical Medicine, Pathology andForensic Medicine, and Cancer Center of Eastern Finland, Universityof Eastern Finland, Kuopio, Finland. 2School of Medicine, Institute ofClinical Medicine, Oncology, University of Eastern Finland, Kuopio,Finland. 3Cancer Center, Kuopio University Hospital, Kuopio, Finland.4Laboratory of Cancer Genetics and Tumor Biology, Department ofClinical Chemistry and Biocenter Oulu, University of Oulu, Oulu, Fin-land. 5Laboratory of Cancer Genetics and Tumor Biology, NorthernFinland Laboratory Centre NordLab, Oulu, Finland. 6Department ofOncology, University of Oulu, Oulu University Hospital, University ofOulu, Oulu, Finland. 7Imaging Center, Clinical Pathology, Kuopio Uni-versity Hospital, Kuopio, Finland.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

Y. Soini and A. Mannermaa contributed equally to this article.

Corresponding Author: Jaana M. Hartikainen, School of Medicine, Institute ofClinical Medicine, Pathology and Forensic Medicine, University of Eastern Fin-land, Yliopistonranta 1 C, FI-70210 Kuopio, Finland. Phone: 358-403552754; Fax:358-17162753; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-14-1887

�2015 American Association for Cancer Research.

ClinicalCancerResearch

www.aacrjournals.org 1591

on May 5, 2018. © 2015 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst January 14, 2015; DOI: 10.1158/1078-0432.CCR-14-1887

http://clincancerres.aacrjournals.org/

in squamous cell carcinomas (16). The NRF2/KEAP1 pathwayhas been reported to be somatically significantly altered also inclear cell renal cell carcinoma (17). In breast carcinomas,KEAP1 mutations and putative copy number alterations haverarely been detected (18, 19) and NFE2L2 mutations are evenless common (19).

Breast carcinoma—the most common cancer type amongwomen in Western countries—has been associated with manyrisk factors, some of them with a hereditary basis (20). The linkbetween breast carcinoma development and oxidative damageis supported by the expression of 8-hydroxydeoxyguanosine(8-OHdG) and 4-hydroxy-2-nonenal (NHE) in breast cancertissue and the fact that estrogen metabolites induce ROSformation and thus are able to cause oxidative stress (21,22). We previously reported that germline polymorphisms inNFE2L2 and SRXN are associated with breast cancer risk andoutcome (23). Another study demonstrated that a combina-tion of risk alleles in iron-related oxidative stress pathwaygenes (NFE2L2, NQO1, NOS3, and HO-1) are associatedwith increased risk of postmenopausal breast cancer (24).KEAP1 polymorphisms have not previously been shown tobe associated with breast cancer risk or survival. However, theKEAP1 polymorphism rs1048290 in tumor tissue associatedwith progression-free survival in endometrioid endometrialadenocarcinoma (25). In addition, in non–small cell lungcancer, low KEAP1 protein expression reportedly associatedwith worse overall survival (OS; ref. 26), and loss of KEAP1function leads to increased NRF2 concentration and chemore-sistance (27).

Here, we studied KEAP1 protein expression in 373 cases ofinvasive breast carcinoma and continued to investigate theNRF2/KEAP1 pathway. We analyzed the influence of 5 taggingSNPs—rs34197572, rs9676881, rs1048290, rs11085735, andrs8113472—in KEAP1 on KEAP1 protein expression, on breastcancer risk, and on patient prognosis. KEAP1 protein expres-sion was also evaluated as a predictive factor and in relation tothe immunohistochemical expression of the NRF2 protein. Thisis the first study to observe the role of KEAP1 polymorphisms asgenetic risk factors in breast cancer.

Materials and MethodsDNA samples

This study included 996 patients with breast cancer and 880healthy control individuals, all of whom were Finnish Caucasianfemales. Data and DNA samples for genotyping were obtainedfrom 2 population-based sample sets from Finland: the KuopioBreast Cancer Project (KBCP) from Eastern Finland and theNorthern Finnish Oulu Breast Cancer Study (OBCS).

From the KBCP sample set, DNA for genotyping was availablefrom 452 patients with invasive breast cancer (age, 23–91 years)and 370 control subjects (age, 26–77 years; Table 1). The KBCPsample set includes a total of 497 prospective breast cancer casesfrom the province ofNorthern Savo in Eastern Finland, diagnosedat the Kuopio University Hospital between 1990 and 1995. It alsoincludes 458 control individuals matched for age and long-termarea of residencewhowere selected from theNational PopulationRegister during the same time period. The patient follow-up dataextend up to 20 years to February 2011.More details regarding theKBCP have been previously published (e.g., ref. 28). The studyparticipants gave their informed consent and the KBCP has beenapproved by the ethical committee of the University of EasternFinland and Kuopio University Hospital.

From the OCBS sample set, DNA for genotyping was availablefrom 544 patients with breast cancer (age, 28–92) and 510healthy control subjects (age, 18–66 years; ref. 29). The OBCSinvasive breast cancer cases were consecutively operated, and dataand blood samples collected at the Oulu University Hospitalbetween 2000 and 2007 (Table 1). Patient follow-up data extendup to 12.7 years. The Northern Finnish control subjects werecollected from cancer-free Finnish Red Cross blood donorsrecruited in 2002 and originating from the same geographicalregion as the studied cancer patient cohort. Control subjects wereanonymous. Their genders, ages, and places of birth were known.All control individuals were cancer-free at the time of donation ofthe blood sample. Therewas no follow-up ondonor health status.Informed consent for participation in the study has been obtainedfromeachpatient, and the studyhas been approvedby the FinnishMinistry of Social Affairs andHealth, and the ethical committee ofOulu University Hospital.

Genomic DNA was extracted from peripheral blood lympho-cytes of both cases and controls using standard procedures (30).

Tumor samplesTumormaterial was available from 373 cases of invasive breast

carcinomas included in the KBCP. Table 1 shows the clinicalcharacteristics of these cases. Tumor tissue was obtained from theprimary tumor during breast cancer surgery. Tissue was paraffinembedded and used to construct a tissue microarray (TMA) aspreviously described (28).

ImmunohistochemistryImmunohistochemical staining of KEAP1 protein was per-

formed as previously described (23). The primary antibody goatpolyclonal anti-human KEAP1 (E-20, sc-15246, Santa Cruz Bio-technology, Inc.) was diluted with 1.5% normal rabbit serum inPBS tomake a 1:200working solution. The antibody dilutionwasoptimized using 3 confirmed KEAP1-positive lung tissue speci-mens under the pathologist's supervision.Oneof these lung tissuesamples was used as a positive control when staining the breastcancer TMAs. Negative controls included lung tissue and breast

Translational Relevance

Oxidative stress is linked to cancer development. Oxidativestress rescue mechanisms also play a role, as they potentiallycause chemo- and radioresistance and affect patient outcome.Kelch-like ECH-associated protein 1 (KEAP1) is a regulator ofthe main mediator of the redox stress response, Nuclear factorerythroid 2–related factor 2 (NRF2). Although oxidative dam-age is considered amechanism for cancer development, its rolein breast cancer risk and outcome has not been extensivelystudied. The present study demonstrated that genetic poly-morphisms inKEAP1 also affect breast cancer risk and survival.In particular, the investigated polymorphisms modified theeffects of radiotherapy and tamoxifen treatment on patientsurvival. These results contribute to the present knowledgeregarding the involvement of NRF2 pathway in breast cancerand in the future may be translated into clinical use and beused in treatment planning and assessing prognosis of patientswith breast cancer.

Hartikainen et al.

Clin Cancer Res; 21(7) April 1, 2015 Clinical Cancer Research1592




cancer tissue samples prepared using the same reagents andprocedure except that the primary antibody was replaced by thebuffer used for dilution. No staining was observed in the negativecontrols.

The immunohistochemical staining in the microarray for eachtumor was done in triplicate. The immunostaining was semi-quantitatively categorized into 4 groups: 0, 0%–5% positive cells;1, 5%–50% positive cells; 2, 50%–75% positive cells; and 3, over75% positive cells. The intensity was assessed as: 0, no staining; 1,

weak staining; 2, moderate staining; and 3, strong staining. Thequantity and intensity values were summed to create an immu-nostaining index value: 0, negative (0); 1–3, weak (1); and 4–6,strong (2). The final immunostaining value for each tumor wasthe arithmetic mean (from values 0–2) of the array triplicates. Forstatistical analyses, thesemean values were divided into 2 groups:�1.33 (representing negative/weak staining) and �1.67 (repre-senting strong/moderate KEAP1 staining). To analyze the associ-ation of KEAP1 protein expression and genotypes with NRF2

Table 1. Clinicopathologic characteristics of the patients

KBCP samples in genotyping OBCS samples in genotypingKBCP cases in TMACases Controls Cases Controls

Characteristic n (%) n (%) n (%) n (%) n (%)

Number of patients 452 (100) 370 537 (100) 510 373 (100)Tumor grade — —

1 116 (25.7) 91 (16.9) 89 (23.9)2 210 (46.5) 229 (42.6) 173 (46.4)3 122 (26.9) 198 (36.9) 109 (29.2)NA 4 (0.9) 19 (3.5) 2 (0.5)

Histological type — —

Ductal 294(65.0) 413 (76.9) 238 (63.8)Lobular 80 (17.7) 92 (17.1) 73 (19.6)Medullary 8 (1.8) 2 (0.4) 8 (2.1)Other 70 (15.5) 30 (5.6) 54 (14.5)

Tumor size — —

T1 236 (52.2) 313 (58.5) 177 (47.5)T2 174 (38.6) 191 (35.7) 160 (42.9)T3 25 (5.5) 17 (3.2) 21 (5.6)T4 17 (3.7) 14 (2.6) 15 (4.0)

Nodal status — —

Negative 262 (58.0) 299 (55.7) 203 (54.4)Positive 180 (39.8) 238 (44.3) 165 (44.2)Unknown 10 (2.2) 5 (1.3)

ER status — —

Negative 99 (21.9) 106 (19.7) 85 (22.8)Positive 335 (74.1) 431 (80.3) 282 (75.6)NA 18 (4.0) 6 (1.6)

Stage — —

I 177 (39.2) 209 (38.9) 129 (34.6)II 213 (47.1) 270 (50.3) 191 (51.2)III 39 (8.6) 34 (6.3) 37 (9.9)IV 13 (2.9) 24 (4.5) 11 (2.9)Unknown 10 (2.2) 5 (1.3)

HER2 status — —

Negative 350 (77.4) 464 (86.4) 300 (80.4)Positive 52 (11.5) 73 (13.6) 46 (12.3)NA 50 (11.1) 27 (7.2)

Cause of death — —

Alive 230 (50.9) 429 (80.8) 182 (48.7)Breast cancer 117 (25.9) 63 (11.9) 105 (28.2)Other 105 (23.2) 39 (7.3) 86 (23.1)

Age, y<50 141 (31.2) 148 (40%) 155 (28.9) 316 (62%) 118 (31.6)�50 311 (68.8) 222 (60%) 382 (71.2) 194 (38%) 255 (68.4)

Menopausal statusPremenopausal 131 (29.0) 151 162 (30.2) 108 (29.0)Postmenopausal 299 (66.2) 213 346 (64.4) 248 (66.5)NA 22 (4.8) 29 (5.4) 510 17 (4.5)

Radiation therapy — —

Yes 258 (57.1) 445 (83.8) 211 (56.6)No 194 (42.9) 86 (16.2) 162 (43.4)

Chemotherapy — —

Yes 86 (19.0) 225 (42.3) 74 (19.8)No 366 (81.0) 307 (57.7) 299 (80.2)

Endocrine therapy — —

Yes 116 (25.7) 256 (48.3) 106 (28.4)No 336 (74.3) 274 (51.7) 267 (71.6)

Abbreviations: NA, data not available; —, data not applicable.

KEAP1 Genetic Polymorphism Predicts Outcome in Breast Cancer

www.aacrjournals.org Clin Cancer Res; 21(7) April 1, 2015 1593




protein expression, we used our previously published data onNRF2 protein expression (23).

SNP selectionTagging SNPs (tagSNPs) for KEAP1 were selected using the

HapMap Genome Browser release 2 (Phase 3, NCBI build 36,dbSNP b126) as of February 24, 2010 (31). TagSNPs for regionchr19:10454325–10478524 were determined for the CEU pop-ulation using the Tagger multimarker algorithm with an r2 cutoffof 0.8 and a minor allele frequency (MAF) cutoff of 0.05.

GenotypingFour KEAP1 tagSNPs were genotyped using MassARRAY

(Sequenom Inc.) and iPLEX Gold (Sequenom Inc.) with a 384-well plate format as previously described (23). For quality control,duplicate analysis was performed for 6.7% of the KBCP samplesand 6.5%of theOBCS samples. All primer sequences and reactionconditions are available upon request. The KEAP1 tagSNPrs34197572 was genotyped by 50 nuclease assay (TaqMan) usingthe Mx3000P Real-Time PCR System (Stratagene) following themanufacturer's instructions. Primers and probes for rs34197572were supplied from Applied Biosystems as Custom TaqManGenotyping Assay. Reactions were carried out in a 10-mL volumein a 96-well format as previously described (28). For qualitycontrol, genotypingwas performed in duplicate for 4.2%of KBCPsamples and 9.1% of OBCS samples. If the duplicate genotyperesults were discordant, the sample was discarded.

SequencingWe detected great deviation from the Hardy–Weinberg equi-

librium. Therefore, to confirm the results of the TaqMan analysis,we sequenced all samples with the rs34197572 minor homozy-gous genotype (29 samples in KBCP and 30 samples inOBCS), 10samples representing the common homozygous and 5 represent-ing the heterozygous genotypes. The sequencing primers were asfollows: forward primer: 50-ataaaaggtggcaccacagg-30; reverse prim-er: 50-tccactgctcagggttgtaa-30. PCR conditions are available onrequest. Sequencing was performed using the BigDye Terminatorv1.1 Cycle Sequencing Kit (Applied Biosystems), an ABI PRISM3100Genetic Analyzer (Applied Biosystems) andData Collectionversion 2.0 software (Applied Biosystems) following the manu-facturer's instructions. Sequences were analyzed using SequenceScanner v1.0 software (Applied Biosystems). All obtainedsequences were in agreement with the TaqMan results.

Statistical analysesSignificance levels for overall comparisons of SNP genotype

frequencies between cases and controls were computed usingFisher exact test implemented in IBM SPSS Statistics 19 software(IBM). The Armitage trend test was used to estimate the linearadditive trend effect (increasing number of alleles) of the riskallele (32). Logistic regression analysis in IBM SPSS Statistics 19was performed to calculate genotype-specific ORs and to test theassociation between genotypes and protein expressions. Thedeviation from Hardy–Weinberg equilibrium was tested usinga standardc2 test. Survival datawere analyzed using the univariateKaplan–Meiermethodwith the log-rank test andmultivariate Coxregression analysis in IBMSPSS Statistics 19.OS andbreast cancer-specific survival (BCSS) were calculated as the time from the dateof diagnosis to the date of last follow-up or death. Cause of death

was coded as either breast cancer or not caused by breast cancer.Relapse-free survival (RFS) was calculated as the time of diagnosisto the time of first relapse (local relapse, contralateral breastcancer, or metastatic disease). A P value of 0.05 (2-sided) or lesswas considered statistically significant. The initial P values fromthe breast cancer risk association analyses were corrected formultiple testing using the Benjamini–Hochberg false discoveryrate method (33).

ResultsPositive KEAP1 protein expression in breast tumors

Cytoplasmic KEAP1 protein expression was negative or weak(0–0.67) in 16.6% of the tumors, moderate (1–1.33) in 48.9%,and high (1.67–2) in 34.6% (Supplementary Fig. S1). KEAP1staining did not significantly differ between different histologictypes of breast carcinoma. Lower KEAP1 protein expressionwas associated with estrogen receptor (ER)-positive tumors[P ¼ 0.040; OR, 1.711; 95% confidence interval (CI), 1.025–2.856], progesterone receptor (PR)-positive tumors (P ¼ 0.025;OR, 1.683; 95% CI, 1.069–2.652) and HER2-negative tumors(P ¼ 0.034; OR, 1.984; 95% CI, 1.052–3.741; SupplementaryTable S1).

KEAP1 protein expression associated with NRF2High KEAP1 protein expression (�1.67) was associated with

high cytoplasmic NRF2 protein expression (P¼ 0.003;OR, 2.154;95% CI, 1.288–3.603; Poverall ¼ 0.003, logistic regression; Sup-plementary Table S2). Low cytoplasmic NRF2 expression wasmore closely associated with lobular tumors than with ductaltumors (with medullary tumors and other histologic types as areference group; P ¼ 0.007; OR, 2.915; 95% CI, 1.332–6.378, inlogistic regression analysis; data not shown). With the analysislimited to tumors with high KEAP1 expression (�1.67), weobserved that tumor histology was associated with cytoplasmicNRF2 expression (P ¼ 0.020; logistic regression). In particular,lobular histology was associated with negative/low cytoplasmicNRF2 (P ¼ 0.034; OR, 6.250; 95% CI, 1.145–34.123; logisticregression; data not shown).With the analysis restricted to tumorswith high cytoplasmic NRF2 protein expression, KEAP1 proteinexpression not significantly associated with tumor histology.

KEAP1 rs34197572 associated with breast cancer riskWe analyzed 5 tagging SNPs in the KEAP1 gene region:

rs34197572, rs9676881, rs1048290, rs11085735, and rs8113472(Supplementary Fig, S2). The controls were in Hardy–Weinbergequilibrium with all studied SNPs (Supplementary Table S3).Within the KBCP sample set, an increased risk of breast cancer wasassociated with the rs34197572 minor allele T (P ¼ 0.015; OR,1.74 from the Armitage trend test) and the homozygous genotypeTT (P ¼ 0.0006; OR, 6.41 compared with the CC genotype). Onthe other hand, the major allele C was associated with decreasedbreast cancer risk (P ¼ 0.015; OR, 0.66 from the Armitage trendtest; and P¼ 0.0006; OR, 0.157 for allele positivity; Table 2). Theassociations were not significant within the OBCS set. However,they were found to be significant in the combined KBCP andOBCS sample set. In the KBCPþOBCS sample set, increased riskof breast cancer was significantly associated with the rs34197572allele T (P¼ 0.044;OR, 1.30 from theArmitage trend test) and thehomozygous genotype TT (P ¼ 0.003; OR, 2.39 compared withthe CC genotype), whereas the common allele C was associated

Hartikainen et al.





with decreased breast cancer risk (P ¼ 0.044; OR, 0.793 from theArmitage trend test; and P ¼ 0.003; OR, 0.421 for allelepositivity; Table 2). In addition, in the KBCP þ OBCS sampleset, the rs11085735 minor allele A was associated with PR-negative tumors (P ¼ 0.017; OR, 1.612; 95% CI, 1.089–2.397;log reg; data not shown).

KEAP1 genotypes associated with KEAP1 and NRF2 proteinexpressions

Within the KBCP sample set, higher KEAP1 protein expressionwas significantly associated with the KEAP1 rs9676881 minoralleleA (P¼ 0.027;OR, 2.507; 95%CI, 1.109–5.667 for rs9676881AA) and rs1048290minor allelesG (P¼ 0.024;OR, 2.545; 95%CI,1.128–5.746 for rs1048290 GG; Supplementary Table S4). Asrs9676881 and rs1048290 were in tight linkage disequilibrium(LD) with an r2 value of 1, they most likely represent the sameassociation. We also found that the KEAP1 rs11085735 minorallele A was significantly associated with lower KEAP1 proteinexpression (P ¼ 0.040; OR, 3.545; 95% CI, 1.061–11.837; Sup-plementary Table S4), and with high NRF2 expression (P¼ 0.009;OR, 2.445; 95% CI, 1.232–4.853; Supplementary Table S5).

rs34197572 and rs8113472 associated with OSTo study the effect of the SNPs on the patients' response to breast

cancer treatment, we included the treatment data in the survivalanalyses of the combined KBCP and OBCS cases. Among all

invasive cases, theKEAP1 rs34197572minor alleleTwas associatedwith poorerOS (P¼ 0.045; HR, 1.304, Cox regression analysis; Fig.1).When the analysiswas limited to all those patientswho receivedradiotherapy (RT), theminoralleleTwasalsoassociatedwithOS(P¼ 0.018; HR, 1.486, Cox regression analysis; Fig. 2A), and thedifference between genotypes was even more pronounced amongthe patients who received only RT (P ¼ 0.037; HR, 1.948, Coxregression analysis; Fig. 2B). The rs8113472 minor allele A wasassociated with poorer OS among the patients who received RT (P¼ 0.025;HR, 1.455, Cox regression analysis; Fig. 2C), although thisassociation did not reach statistically significance among all inva-sive cases (P ¼ 0.063, data not shown). Again, the differencebetween the genotypes (major homozygous vs. minor allele car-riers) was more pronounced among the patients receiving only RT(P ¼ 0.044; HR, 1.820, Cox regression analysis; Fig. 2D).

Among patients who did not receive adjuvant therapy (n ¼187), the rs34197572 and rs8113472 genotypes were not asso-ciated with survival. Within both SNPs, OS significantly differedbetween the treatment-defined strata (RT vs. no adjuvant therapy:P¼ 0.024 for rs34197572 and P¼ 0.028 for rs8113472; log-rankKaplan–Meier analysis, data not shown). AmongRT-treated cases,poorer OS was associated with the rs34197572 allele T (P ¼0.009) and rs8113472 alleleA (P¼ 0.015, log-rank Kaplan–Meieranalysis, data not shown). No such associations were foundamong patients who did not receive adjuvant therapy (P ¼0.818 for rs34197572 and P ¼ 0.692 for rs8113472, log-rank

Table 2. Association between KEAP1 SNP genotypes and risk of breast cancer in KBCP, OBCS, and combined samples

Genotype counts Homozygouse Allele positivityf

SNP Cases Controls Pa PFDRb PTrend

c PTrend-FDRd P OR (95% CI) P OR (95% CI)

KBCPrs34197572 0.001 0.005 0.006 0.030CC 327 287 0.00018 0.137 (0.041–0.458) 0.0002 0.139 (0.042–0.464)CT 84 68TT 25 3 0.00018 7.314 (2.185–24.478) 0.084 1.347 (0.961–1.890)

rs9676881 0.342 0.428 0.143 0.179 nsrs1048290 0.286 0.428 0.117 0.179 nsrs11085735 0.952 0.952 0.954 0.954 nsrs8113472 0.151 0.378 0.087 0.179 ns

OBCSrs34197572 0.874 0.943 0.801 0.801 nsCC 417 389CT 110 104TT 17 13


KBCP and OBCSrs34197572 0.011 0.055 0.044 0.218CC 744 676 0.0028 0.419 (0.234–0.753) 0.0028 0.421 (0.235–0.755)CT 194 172TT 42 16 0.0028 2.385 (1.329–4.282) 0.237 1.141 (0.917–1.419)


NOTE: Values shown in bold are statistically significant.Abbreviation: ns, no significant association observed.aP from the c2 test for overall association of the SNP genotypes with breast cancer risk.bP from the c2 test for overall association of the SNP genotypes with breast cancer risk corrected with the Benjamini–Hochberg FDR method.cP from the Armitage trend test for the linear additive trend effect (increasing number of alleles) of the risk allele.dP from the Armitage trend test for the linear additive trend effect of the risk allele corrected with the Benjamini–Hochberg FDR method.eP, OR, and 95% CI for the homozygous risk allele carriers versus the other homozygotes.fP, OR, and 95% CI for the risk allele carriers, that is, the homozygous and heterozygous allele carriers versus the other homozygotes. For example the values on row"TT" are for the T allele positivity, that is TT and CT versus CC.






Kaplan–Meier analysis, data not shown). Multivariate analysisshowed that the rs34197572 and rs8113472 genotypes were notassociated with survival when analyzed separately in the KBCPand OBCS samples (Supplementary Table S6).

KEAP1 rs1048290 and rs9676881 associated with RFSWhen the treatment data were included in the analysis of the

combined KBCP and OBCS invasive cases, shorter RFS wassignificantly associatedwith theKEAP1minor homozygous geno-type rs1048290 GG (P¼ 0.020; HR, 1.468) and rs9676881 AA (P¼ 0.024; HR, 1.452, Cox regression analysis; Fig. 3A and B). Thesegenotypes were also significantly associated with shorter RFSamong the ER-positive patients treated with tamoxifen (P ¼0.015; HR, 2.476 and P ¼ 0.018; HR, 2.407, respectively, Coxregression analysis; Fig. 3C and D) and in the OBCS sample setalone (P ¼ 0.011; HR, 2.210 and P ¼ 0.011; HR, 2.210, respec-tively, Cox regression analysis; Supplementary Table S6). In theKBCP sample set alone, the associations of rs1048290 andrs9676881 with RFS among tamoxifen-treated cases were closeto significant (P¼ 0.053; HR, 2.139 for rs1048290 and P¼ 0.063for rs9676881, Cox regression analysis; Supplementary Table S6).Among patients who did not receive adjuvant therapy (n¼ 186),the rs1048290 and rs9676881 genotypeswere not associatedwithsurvival. For both SNPs, RFS significantly differed between thetreatment-defined strata (ER-positive tamoxifen-treated vs. caseswith no adjuvant therapy): P¼ 0.029 for rs1048290 and 0.037 forrs9676881 (log-rank Kaplan–Meier analysis, data not shown).The genotypes rs1048290 GG and rs9676881 AA were signifi-cantly associated with shorter RFS among tamoxifen-treatedpatients (P ¼ 0.011 and 0.015, respectively, log-rank Kaplan–Meier analysis, data not shown) but not among patients who didnot receive adjuvant therapy (P ¼ 0.522 and 0.569, respectively,log-rank Kaplan–Meier analysis, data not shown).

rs34197572 OS

CT and TTn = 206HR = 1.304(95% CI, 1.006–1.692)

CCn = 615

P = 0.045

Survival time in years

Cu

m. s

urv

ival

0

0.0

0.2

0.4

0.6

0.8

1.0

5 10 15 20

Figure 1.Association of KEAP1 rs34197572 with OS among all invasive KBCP andOBCScases. Multivariate analysis stratified by age at diagnosis, stage, ER status, PRstatus, adjuvant chemotherapy, radiotherapy, and hormone therapy. HRs ofdeath with 95% CI in Cox regression survival analysis.

rs8113472 OS, RT only

CA and AAn = 59HR = 1.820(95% CI, 1.016−3.262)

CCn = 136

P = 0.044

D

rs34197572 OS, RT only

CT and TTn = 57HR = 1.948(95% CI, 1.041−3.647) P = 0.037

B

rs8113472 OS, RT

CA and AAn = 154HR = 1.455(95% CI, 1.048−2.021)

CC

CC

n = 438

P = 0.025

C

rs34197572 OS, RT

CT and TTn = 152HR = 1.486(95% CI, 1.069−2.065)

CC

n = 440 n = 136

P = 0.018

A


Cu

m. s

urv

ival

0

0.0

0.2

0.4

0.6

0.8

1.0

Cu

m. s

urv

ival

0.0

0.2

0.4

0.6

0.8

1.0

5 10 15 20


0 5 10 15 20

Cu

m. s

urv

ival

0.0

0.2

0.4

0.6

0.8

1.0


0 5 10 15 20


Cu

m. s

urv

ival

0

0.0

0.2

0.4

0.6

0.8

1.0

5 10 15 20

y

CA CC and AAn = 59nHR = 1.820(95% CI, 1.016−3.262)

CCn = 136 n

P = 0.044

CA CC and AAn = 154nHR = 1.455(95% CI, 1.048−2.021)

CCn = 438 n

P = 0.025

Figure 2.Association of KEAP1 rs34197572 andrs8113472 with OS among KBCP andOBCS patients who receivedradiotherapy. OS with rs34197572genotypes (A) among patients whoreceived radiotherapy and (B) amongpatients who received onlyradiotherapy. OS with rs8113472genotypes (C) among patients whoreceived radiotherapy and (D) amongpatients who received onlyradiotherapy. A and C, HRs of deathwith 95% CI in Cox regression survivalanalyses stratified by age at diagnosis,stage, ER status, PR status, adjuvantchemotherapy, and hormone therapy.B and D, HRs of death with 95% CI inCox regression survival analysesstratified by age at diagnosis, stage,ER status, and PR status.

Hartikainen et al.





KEAP1 rs11085735 associated with BCSSWhen treatment data were included, Cox multivariate survival

analysis revealed that the KEAP1 rs11085735 minor allele A wasassociated with poorer RFS and BCSS among all invasive cases (P¼ 0.020; HR, 1.545 and P ¼ 0.016; HR, 1.683, respectively) andwith shorter RFS among tamoxifen-treated cases (P ¼ 0.003; HR,3.517; 95% CI, 1.556–7.951; Fig. 4). Within only the KBCPsample set, the rs11085737 allele A was associated with shorterRFS among all invasive cases (P ¼ 0.038; HR, 1.574; 95% CI,1.026–2.404) andamong tamoxifen-treated cases (P¼0.024;HR,2.868; 95% CI, 1.152–7.145) and was associated with BCSSamong the tamoxifen-treated cases (P ¼ 0.009; HR, 3.900;95% CI, 1.395–10.905; Supplementary Table S6). Within onlytheOBCS sample set, the rs11085735 alleleAwas associated withBCSS among all invasive cases (P ¼ 0.035; HR, 2.408; 95% CI,1.064–5.450) and among RT-treated cases (P¼ 0.041; HR, 2.387;95% CI, 1.037–5.491; Supplementary Table S6). Within theOBCS sample set, the genotype association with RFS amongtamoxifen-treated cases was close to significant (P ¼ 0.070, datanot shown).

Of the 5 studied SNPs, KEAP1 rs11085735 was the only SNPassociated with patient survival when treatment data were notincluded in the survival analyses. In the combined KBCP andOBCS cases, the minor allele A was associated with poorer breastcancer survival in univariate Kaplan–Meier analysis (P ¼ 0.039;Supplementary Fig. S3A) and in Cox multivariate analysis (P ¼

0.023; HR, 1.634; Supplementary Fig. S3B and Table S7). Amongthe ER-positive combined KBCP andOBCS cases, the rs11085735minor allele A was also associated with poorer breast cancersurvival in the univariate analysis (P ¼ 0.007, Kaplan–Meieranalysis; Supplementary Fig. S4A and Table S8) but not in theCoxmultivariate analysis. Among theER-positive cases only in theKBCP sample set, this association was also significant in themultivariate analysis (P ¼ 0.017; HR, 2.106, Cox regressionanalysis; Supplementary Tables S8 and S9 and Fig. S4B). Amongthe KBCP cases with lower KEAP1 protein expression (�1.33), thers11085735 minor allele A was associated with poorer breastcancer survival in univariate Kaplan–Meier analysis (P ¼ 3.6 �10�4; Supplementary Table S8 and Fig. S5A) and in multivariateanalysis (P ¼ 0.001; HR, 2.733, Cox regression analysis; Supple-mentary Table S10 and Fig. S5B). KEAP1 protein expression itselfwas not significantly associated with survival in Kaplan–Meieranalysis, although a trend toward worse survival with lowerKEAP1 protein expression was observed (Supplementary Fig. S6).

DiscussionOxidative stress is linked with cancer development, and defec-

tive oxidative stress rescue mechanisms can also play a role inincreasing cancer susceptibility, potentially causing chemo- andradioresistance and affecting treatment outcome. KEAP1 is aregulator of NRF2, which is the main mediator of the redox stress

rs9676881 RFS

AAn = 166HR = 1.452(95% CI, 1.051−2.006)

GG and GAn = 654

P = 0.024

Brs1048290 RFS

GGn = 166HR = 1.468(95% CI, 1.063−2.027)

CC and CGn = 659

P = 0.020

A

rs1048290 RFS, TAM+

GGn = 20HR = 2.476(95% CI, 1.197−5.123)

CC and CGn = 84

P = 0.015

Crs9676881 RFS, TAM+

AAn = 20HR = 2.407(95% CI, 1.163−4.979)

GG and GAn = 82

P = 0.018

D

RFS in years

Cu

m. s

urv

ival

Cu

m. s

urv

ival

0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

5 10 15 20

RFS in years

0 5 10 15 20

Cu

m. s

urv

ival

0.0

0.2

0.4

0.6

0.8

1.0

Cu

m. s

urv

ival

0.0

0.2

0.4

0.6

0.8

1.0

RFS in years

0 53 108 13

RFS in years

0 53 108 13

Figure 3.Association of KEAP1 rs1048290 andrs9676881withRFS inKBCPandOBCScases. A, rs1048290 among all invasivecases. B, rs9676881 among all invasivecases. C, rs1048290 among ER-positive tamoxifen-treated patientswith breast cancer. D, rs9676881among ER-positive tamoxifen-treatedpatients with breast cancer. A and B,HRs of breast cancer recurrence with95% CI in Cox regression survivalanalysis stratified by age at diagnosis,stage, ER status, PR status, adjuvantchemotherapy, radiotherapy, andhormone therapy. C and D, HRs ofbreast cancer recurrence with 95% CIin Cox regression survival analysesstratified by age at diagnosis, stage,PR status, and radiotherapy.






response. We previously reported that polymorphisms in theoxidative stress response pathway affect breast cancer risk andthe survival outcome of patients with breast cancer (23). Thepresent results show that genetic polymorphisms in KEAP1 alsoseem to affect the breast cancer risk and survival, further support-ing the involvement of the NRF2 pathway in breast cancer. Inparticular, the investigated polymorphisms seemed tomodify theeffects of radiotherapy and tamoxifen treatment on patient sur-vival. Implications of the modifying role of lower KEAP1 proteinexpression were also observed.

In 2 populations, we examined 5 tagging SNPs covering theKEAP1 gene region, along with KEAP1 protein expression byimmunohistochemistry. We found that the minor alleles of 2SNPs in KEAP1—rs9676881 and rs1048290—were associatedwith higher KEAP1 protein expression and shorter RFS. Theassociation with RFS was observed among all invasive cases, andamong ER-positive tamoxifen-treated cases. The SNP rs9676881is located only 16 bp downstream from the 30-untranslated region(UTR) of KEAP1 gene in a possible transcription factor–bindingand enhancer region (34, 35). SNP rs1048290 is a synonymousvariant in the KEAP1 coding region, which is located in thesequence encoding the beta strand B of blade IV in the Kelch-1domain, residing 33 bp (11 codons) from the codon for theamino acid whose side chain contacts the NRF2-derived peptidein the crystal structure (36).

The rs1048290 and rs9676881 polymorphisms are in tight LDwith each other (r2 ¼ 1, Ensembl, 1000Genomes data, Finnishpopulation; refs. 35, 37) and could also be in LD with anotherfunctional variant that affects KEAP1 protein expression or func-

tion, such as amiRNA-binding site in the 30UTR ofKEAP1. Loss ofmiR-200a is reportedly correlated with KEAP1 upregulation andNRF2 downregulation in breast cancer cell lines (38). The KEAP130UTR contains a binding site for miR-200a (TargetScan; ref. 39)that is 410 bp upstream from rs9676881, and rs9676881 andrs1048290 could be in LD with a variant that affects miR-200abinding, resulting in a higher KEAP1 level. Five other miRNA-binding sites are predicted in the KEAP1 30UTR (microRNA.org;ref. 40). These SNPs appear to have the potential to affect KEAP1protein level, although the effects on patient survival are probablyvery complex.

Antiestrogen agents have been reported to stimulate NAD(P)H:dehydrogenase, quinone 1 (NQO1) transcription and toprotect against estrogen-induced DNA damage in estrogen-dependent breast cancer cells (41, 42). Tamoxifen acts as aselective estrogen receptor modulator that binds the NQO1promoter and activates NQO1 gene transcription in breastcancer cells (41, 42). Thus, tamoxifen has been proposed asa possible chemopreventive agent for use against breast cancer.However, after tumor development and the use of tamoxifen,cancer cell survival could actually be promoted, as tamoxifeninduces NQO1, which would protect the cancer cells againstestrogen-induced DNA damage. Tamoxifen can also induceoxidative stress (43–45), which would then activate the NRF2pathway. Estrogen also reportedly acts on the ERa ligand-binding domain and inhibits NQO1 promoter activity; there-fore, inhibition of estrogen signaling leads to activation ofNRF2-dependent NQO1 transcription (46). Breast cancer cellstransfected with ERa shRNA exhibit reduced nuclear NRF2

A

CA + AAn = 109HR = 1.545 (95% CI, 1.070−2.231)

CC n = 714

rs11085735, RFS

P = 0.020

Crs11085735, RFS, TAM

CA + AAn = 16HR = 3.517 (95% CI, 1.556−7.951)

CC n = 86

P = 0.003

B

CA + AAn = 109HR = 1.683 (95% CI, 1.104−2.566)

CC n = 714

rs11085735, BCSS

P = 0.016

RFS in years Survival time in years

Cu

m. s

urv

ival

0

0.0

0.2

0.4

0.6

0.8

1.0

Cu

m. s

urv

ival

0.0

0.2

0.4

0.6

0.8

1.0

5 10 15 20

RFS in years

Cu

m. s

urv

ival

0

0.0

0.2

0.4

0.6

0.8

1.0

5 83 10 13

0 5 10 15 20

CA CC + AAn = 16nHR = 3.517 (95% CI, 1.556−7.951)

CCn = 86n

P = 0.003

CA CC + AAn = 109nHR = 1.545 (95% CI, 1.070−2.231)

CCn = 714n

P = 0.020

CACC + AAn = 109nHR = 1.683 (95% CI, 1.104−2.566)

CCn = 714n

P = 0.016

Figure 4.Association of KEAP1 rs11085735 withRFS andBCSS amongKBCP andOBCScases. A, RFS in all invasive cases. B,BCSS in all invasive cases. C, RFS intamoxifen-treated cases. A and B, HRsof breast cancer recurrence or deathwith 95% CI in Cox regression survivalanalyses stratified by age at diagnosis,stage, ER status, PR status, radiationtherapy, adjuvant chemotherapy, andhormone therapy. C, HRs of breastcancer recurrence with 95% CI in Coxregression survival analyses stratifiedby age at diagnosis, stage, ER status,PR status, and radiotherapy.

Hartikainen et al.





localization as well as cytoplasmic KEAP1 accumulation, whichwould bind any remaining NRF2, preventing antioxidantresponse and allowing ROS generation (47). Among patientswith endometrioid endometrial adenocarcinoma who receivedadjuvant (CT and/or RT) treatment, heterozygous rs1048290minor allele carriers (analyzed from tumor tissue) have previ-ously been found to show shorter progression-free survivalcompared with CC homozygotes (25). Furthermore, in ovariancancer, KEAP1 downregulation by miR-141 reportedly inducescisplatin resistance, whereas KEAP1 overexpression significant-ly enhances cisplatin sensitivity (48). However, this did notlead to activation of NRF2 target genes but rather to activationof the NF-kB pathway.

The polymorphism rs34197572 is downstream of the KEAP130UTR and was also associated with breast cancer risk. SNPrs34197572 was not associated with KEAP1 protein level butcan be connected to breast cancer susceptibility via the ROSscavenging/NRF2 pathway, as impaired ROS response increasescancer risk. SNP rs34197572 was also associated with OSamong all cases of invasive breast cancer, as well as amongthose treated with RT and those treated with only RT. Review ofthe previously published data revealed no direct functionalconsequences for this variation. Approximately 700 bpupstream of rs34197572, there is a possible transcriptionfactor–binding site (Encode; ref. 49). It is possible thatrs34197572 is in LD with another functional variant, whichcould explain a potential connection with breast cancer risk andsurvival. This SNP could also impact survival via anothermechanism, independent of oxidative stress.

The SNP rs8113472 was associated with OS among allinvasive cases, including those treated with RT and thosetreated with only RT. This SNP was not associated with KEAP1protein expression. The SNP rs8113472 is located in intron 1 ofKEAP1. It has not been reported to impact any regulatoryregions or having functional consequences. Therefore, thefunctional variant explaining this association with survivalremains to be elucidated.

For the final studied SNP rs11085735, the minor allele A wasassociated with lower KEAP1 protein expression and increasednuclear NRF2 expression. rs11085735 resides in intron 2, whichintervenes with the exon sequence encoding the Kelch-1 domain(between the B and C beta strands of blade III). As the SNP islocated near a coding region of a functional element of theprotein, it may be in LD with a variant that affects the Kelch-1domain structure and thus affects the interaction between KEAP1and NRF2. Impairment of the affinity of KEAP1 for NRF2 couldpotentially enhanceNRF2 accumulation in the nucleus. Increasednuclear NRF2 expression implies that NRF2 is activating genes foran ROS response and therefore facilitating the cancer cell survivaland resulting in poorer patient survival. Indeed, the rs11085735minor allele was associated with poorer BCSS and RFS among allinvasive cases, poorer OS among RT-treated patients, and shorterRFS among tamoxifen-treated patients. Cases with lower KEAP1protein expression and the rs11085735 A allele also showedpoorer breast cancer survival. One intriguing possibility is thatimpaired interaction of KEAP1 and NRF2 in patients with thers11085735 minor allele A shows an enhanced effect of ROSscavenging. This would result in lower KEAP1 protein expressionand higher nuclear NFR2, which were both observed to beassociated with this allele, and would lead to poorer patientsurvival.

Overall, here we report that in certain treatment groups, thesepresently studied SNPs seem to worsen the prognosis of patientswith breast cancer, thus supporting the possible involvement ofthe NRF2/KEAP1 pathway. RT, estrogen, and tamoxifen eachincrease oxidative stress, which activates NRF2 function(13, 22, 43–45, 50). Enhanced ROS scavenging in tumors couldlead to cancer recurrence or poor survival. None of the investi-gated SNPs showed significant effects within the group of patientswho did not receive any adjuvant therapy, implying that theseSNPs specifically modified the effects of radiotherapy and tamox-ifen treatment on patient survival.

In conclusion, here we report that genetic variants in the KEAP1gene were associated with the outcome of patients with breastcancer. One of these variants, rs34197572, was also associatedwith risk of breast cancer. These SNPs may cause defects in theoxidative stress rescue mechanisms. Impaired response to oxida-tive stress causes prolonged exposure of the tissues to damagingagents, thus increasing cancer susceptibility. On the other hand, ifthe stress response system is functioning in the cancer tissue inresponse to damage caused by therapy, it could also help thecancer cells survive. The present findings support the hypothesisthat the KEAP1/NRF2 pathway plays a role in breast cancer.Further studies are needed to identify the functional variantsaffecting KEAP1 protein expression and interaction with NRF2.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: J.M. Hartikainen, V.-M. Kosma, Y. Soini, A.MannermaaDevelopment of methodology: J.M. Hartikainen, Y. Soini, A. MannermaaAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): R. Winqvist, K. Pylk€as, Y. Soini, A. MannermaaAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): J.M. Hartikainen, M. Tengstr€om, Y. Soini,A. MannermaaWriting, review, and/or revision of the manuscript: J.M. Hartikainen,M. Tengstr€om, R. Winqvist, A. Jukkola-Vuorinen, V.-M. Kosma, Y. Soini,A. MannermaaAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): R. Winqvist, A. Jukkola-VuorinenStudy supervision: V.-M. Kosma, Y. Soini, A. Mannermaa

AcknowledgmentsKBCP thanks Eija My€oh€anen, Helena Kemil€ainen, and Aija Parkkinen for

their skillful technical assistance. OBCS thanks Mervi Grip, Kari Mononen, andMeeri Otsukka for their help with the OBCS sample and data collection.

Grant SupportThis study was funded by grants from the Finnish Cancer Society, the

Academy of Finland (grants 127220, 250083, 251314, and 122715), thespecial Governmental EVO Research Fund of Kuopio University Hospital(grants 53125, 53127, and 53132) and of Oulu University Hospital (grantK36717); the Sigrid Juselius Foundation, the University of Oulu andBiocenter Oulu, and the strategic funding of the University of EasternFinland.

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received July 30, 2014; revisedDecember 16, 2014; accepted January 7, 2015;published OnlineFirst January 14, 2015.






References1. Loft S, Poulsen HE. Cancer risk and oxidative DNA damage in man. J Mol

Med (Berl) 1996;74:297–312. Erratum in: J Mol Med 1997;75:67–8.2. Reuter S, Gupta SC, Chaturvedi MM, Aggarwal BB. Oxidative stress,

inflammation, and cancer: how are they linked? Free Radic Biol Med2010;49:1603–16.

3. Niture SK, Khatri R, Jaiswal AK. Regulation of Nrf2-an update. Free RadicBiol Med 2014;66:36–44.

4. Sun Z, Zhang S, Chan JY, Zhang DD. Keap1 controls postinductionrepression of the Nrf2-mediated antioxidant response by escorting nuclearexport of Nrf2. Mol Cell Biol 2007;27:6334–49.

5. Adam J, Hatipoglu E, O'Flaherty L, Ternette N, Sahgal N, LockstoneH, et al.Renal cyst formation in Fh1-deficient mice is independent of the Hif/Phdpathway: roles for fumarate in KEAP1 succination and Nrf2 signaling.Cancer Cell 2011;20:524–37.

6. ChenW, Sun Z, Wang XJ, Jiang T, Huang Z, Fang D, et al. Direct interactionbetween Nrf2 and p21(Cip/WAF1) upregulates the Nrf2-mediated anti-oxidant response. Mol Cell 2009;34:663–73.

7. Lau A,Wang XJ, Zhao F, Villeneuve NF,Wu T, Jiang T, et al. A noncanonicalmechanism of Nrf2 activation by autophagy deficiency: direct interactionbetween Keap 1 and p62. Mol Cell Biol 2010;30:3275–85.

8. Komatsu M, Kurokawa H, Waguri S, Taguchi K, Kobayashi A, Ichimura Y,et al. The selective autophagy substrate p62 activates the stress responsivetranscription factor Nrf2 through inactivation of Keap1. Nat Cell Biol2010;12:213–23.

9. Camp ND, James RG, Dawson DW, Yan F, Davison JM, Houck SA, et al.Wilms tumor gene on the X chromosome (WTX) inhibits the degradationof NRF2 through competitive binding to KEAP1. J Biol Chem 2012;287:6539–50.

10. Kim WD, Kim YW, Cho IJ, Lee CH, Kim SG. E-cadherin inhibits nuclearaccumulation of Nrf2: implications for chemoresistance of cancer cells.J Cell Sci 2012;125:1284–95.

11. Ma J, Cai H, Wu T, Sobhian B, Huo Y, Alcivar A, et al. PALB2 interacts withKEAP1 to promoteNRF2nuclear accumulation and function.MolCell Biol2012;32:1506–17.

12. Hayes JD, McMahon M. NRF2 and KEAP1 mutations: permanent activa-tion of an adaptive response in cancer. Trends Biochem Sci 2009;34:176–88.

13. Hu L,MiaoW, LoignonM, KandouzM, Batist G. Putative chemopreventivemolecules can increase Nrf2-regulated cell defence in some human cancercell lines, resulting in resistance to common cytotoxic therapies. CancerChemoter Pharmacol 2010;66:467–74.

14. The Cancer Genome Atlas Research Network. Comprehensive genomiccharacterization of squamous cell lung cancers. Nature 2012;489:519–25.

15. Kandoth C, McLellan MD, Vandin F, Ye K, Niu B, Lu C, et al. Mutationallandscape and significance across 12 major cancer types. Nature2013;502:333–9.

16. Kim YR, Oh JE, KimMS, KangMR, Park SW, Han JY, et al. Oncogenic NRF2mutations in squamous cell carcinomas of oesophagus and skin. J Pathol2010;220:446–51.

17. Sato Y, Yoshizato T, Shiraishi Y, Maekawa S, Okuno Y, Kamura T, et al.Integrated molecular analysis of clear-cell renal cell carcinoma. Nat Genet2013;45:860–7.

18. YooNJ, KimHR, Kim YR, An CH, Lee SH. Somatic mutations of the KEAP1gene in common solid cancers. Histopathology 2012;60:943–52.

19. The Cancer Genome Atlas Network. Comprehensivemolecular portraits ofhuman breast cancer. Nature 2012;490:61–70.

20. Tavassoli FA, Devilee P. Tumours of the breast and female genital organs.World Health Organization classification of tumours. Lyon, France: IARCPress; 2003.

21. Karihtala P, Kauppila S, Puistola U, Jukkola-Vuorinen A. Divergent behav-iour of oxidative stress markers 8-hydroxydeoxyguanosine (8-OHdG) and4-hydroxy-2-nonenal (HNE) in breast carcinogenesis. Histopathology2011;58:854–62.

22. Okoh V, Deoraj A, Roy D. Estrogen-induced reactive oxygen species-mediated signaling contribute to breast cancer. Biochim Biophys Acta2011;1815:115–33.

23. Hartikainen JM, Tengstr€om M, Kosma V-M, Kinnula VL, Mannermaa A,Soini Y. Genetic polymorphisms and protein expression of NRF2 and

sulfiredoxin predict survival outcomes in breast cancer. Cancer Res2012;72:5537–46.

24. Hong CC, Ambrosone CB, Ahn J, Choi JY, McCullough ML, Stevens VL,et al. Genetic variability in iron-related oxidative stress pathways (Nrf2,NQO1, NOS3, and HO-1), iron intake, and risk of postmenopausal breastcancer. Cancer Epidemiol Biomarkers Prev 2007;16:1784–94.

25. Wong TF, Yoshinaga K, Monma Y, Ito K, Niikura H, Nagase S, et al.Association of Keap1 and Nrf2 genetic mutations and polymorphismswith endometrioid endometrial adenocarcinoma survival. Int J GynecolCancer 2011;21:1428–35.

26. Solis LM, Behrens C, Dong W, Suraokar M, Ozburn NC, Moran CA, et al..Nrf2 and Keap1 abnormalities in non-small cell lung carcinoma andassociation with clinicopathologic features. Clin Cancer Res 2010;16:3743–53.

27. Singh A, Ling G, Shasini AN, Zhang P, Yamamoto M, Navas-Acien A, et al.Nrf2-dependent sulfiredoxin-1 expression protects against cigarettesmoke-induced oxidative stress in lungs. Free Radic Biol Med 2009;46:376–86.

28. Kauppinen JM, Kosma V-M, Soini Y, Sironen R, Nissinen M, Nykopp TK,et al. ST14 gene variant anddecreasedmatriptase protein expressionpredictpoor breast cancer survival. Cancer Epidemiology Biomarkers Prev2010;19:2133–42.

29. Solyom S, Aressy B, Pylk€as K, Patterson-Fortin J, Hartikainen JM, Kallio-niemi A, et al. Breast cancer-associated Abraxas mutation disrupts nuclearlocalization and DNA damage response functions. Sci Transl Med2012;4:122rs23.

30. Vandenplas S, Wiid J, Grobler-Rabie A, Brebner K, Ricketts M, Wa�llis G,

et al. Blot hybridization of genomic DNA. J Med Genet 1984;21:164–72.31. International HapMap Project. [cited 2010 Feb 24]. Available from: http://

hapmap.ncbi.nlm.nih.gov/cgi-perl/gbrowse/hapmap3r2_B36/.32. Institute for Human Genetics. [cited 2014 Nov 1]. Available from: http://

ihg.gsf.de/cgi-bin/hw/hwa2.pl.33. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical

and powerful approach to multiple testing. J R Statist Soc Ser B (Method-ological) 1995;57:289–300. doi:10.2307/2346101.

34. UCSC Genome Bioinformatics. [Human Feb. 2009 genome GRCh37/hg19 assembly, cited 2014 May 9]. Available from: http://genome.ucsc.edu/.

35. Ensembl. [cited 2014 May 9]. Available from: www.ensembl.org.36. Lo S-C, Li X, Henzl MT, Beamer LJ, Hannik M. Structure of the Keap1:Nrf2

interface provides mechanistic insight into Nrf2 signaling. EMBO J2006;25:3605–17.

37. 1000 Genomes. [cited 2014 May 9]. Available from: www.1000genomes.org.

38. Eades G, Yang M, Yao Y, Zhang Y, Zhou Q. miR-200a regulates Nrf2activation by targeting Keap1 mRNA in breast cancer cells. J Biol Chem2011;286:40725–33.

39. TargetScan. [cited 2014 May 9]. Available from: www.targetscan.org.40. microRNA.org. [cited 2014 May 9]. Available from: www.microrna.org.41. Montano MM, Chaplin LJ, Deng H, Mesia-Vela S, Gaigwad N, Zahid M,

et al. Protective roles of quinone reductase and tamoxifen againstestrogen-induced mammary tumorigenesis. Oncogene 2007;26:3587–90.

42. Sripathy SP, Chaplin LJ, GaikwadNW, Rogan EG,MontanoMM. hPMC2 isrequired for recruiting an ERb coactivator complex to mediate transcrip-tional upregulation of NGO1 and protection against oxidative DNAdamage by tamoxifen. Oncogene 2008;27:6376–84.

43. Ferlini C, Scambia G, Marone M, Distefano M, Gaggini C, FerrandinaG, et al. Tamoxifen induces oxidative stress and apoptosis in oestro-gen receptor-negative human cancer cell lines. Br J Cancer 1999;79:257–63.

44. Nazarewicz RR, Zenebe WJ, Parihar A, Larson SK, Alidema E, Choi J,et al. Tamoxifen induces oxidative stress and mitochondrial apoptosisvia stimulating mitochondrial nitric oxide synthase. Cancer Res2007;67:1282–90.

45. Wozniak K, Kolacinska A, Blasinska-Morawiec M, Morawiec-Bajda A,Morawiec Z, Zadrozny M , et al. The DNA-damaging potential oftamoxifen in breast cancer and normal cells. Arch Toxicol 2007;81:519–27.


Hartikainen et al.




46. Yao Y, Brodie AMH, Davidson NE, Kensler TW, Zhou Q. Inhibition ofestrogen signaling activates the NRF2 pathway in breast cancer. BreastCancer Res Treat 2010;124:585–91.

47. Cook KL, Clarke PAG, Parmar J, Hu R, Schwarz-Roberts JL, Abu-Asab M,et al. Knockdown of estrogen receptor-a induces autophagy and inhibitsantiestrogen-mediated unfolded protein response activation, promotingROS-induced breast cancer cell death. FASEB J. 2014 May 23. —[Epubahead of print].

48. van Jaarsveld MTM, Helleman J, Boersma AWM, van Kuijk PF, van IjckenWF, Despierre E, et al. miR-141 regulates KEAP1 and modulates cisplatinsensitivity in ovarian cancer cells. Oncogene 2013;32:4284–93.

49. Wang J, Zhuang J, Iyer S, Lin X, Whitfield TW, Greven MC, et al. Sequencefeatures and chromatin structure around the genomic regions bound by119 human transcription factors. Genome Res. 2012;22:1798–812.

50. Riley PA. Free radicals in biology: oxidative stress and the effects of ionizingradiation. Int J Radiat Biol 1994;65:27–33.






2015;21:1591-1601. Published OnlineFirst January 14, 2015.Clin Cancer Res Jaana M. Hartikainen, Maria Tengström, Robert Winqvist, et al. and Survival Outcomes

Genetic Polymorphisms Associate with Breast Cancer RiskKEAP1

Updated version

10.1158/1078-0432.CCR-14-1887doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2015/01/15/1078-0432.CCR-14-1887.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/21/7/1591.full#ref-list-1

This article cites 41 articles, 13 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/21/7/1591.full#related-urls

This article has been cited by 1 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/21/7/1591To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-14-1887

http://clincancerres.aacrjournals.org/content/suppl/2015/01/15/1078-0432.CCR-14-1887.DC1

http://clincancerres.aacrjournals.org/content/21/7/1591.full#ref-list-1

http://clincancerres.aacrjournals.org/content/21/7/1591.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/21/7/1591


Documents

KEAP1 Genetic Polymorphisms Associate with Breast Cancer ...clincancerres.aacrjournals.org/content/clincanres/21/7/1591.full.pdf · Breast Cancer Risk and Survival Outcomes ... 4Laboratory