Down-Regulated Xanthine Oxidoreductase Is a Feature of ...clincancerres.aacrjournals.org/content/clincanres/11/12/4372.full.pdfDown-Regulated Xanthine Oxidoreductase Is a Feature of

Down-Regulated Xanthine Oxidoreductase Is a Feature ofAggressive Breast CancerNina Linder,1Johan Lundin,2 Jorma Isola,3 Mikael Lundin,2 Kari O. Raivio,1and HeikkiJoensuu2

Abstract Purpose: Xanthine oxidoreductase (XOR) is a key enzyme in the degradation of DNA, RNA, andhigh-energy phosphates and also plays a role inmilk lipid globule secretion. Given the strong andregulated expression of XOR in normal breast epithelium, and the previously shown alterations ofits expression in experimental tumorigenesis, we hypothesized that XOR may be differentiallyexpressed in breast cancer.Experimental Design:XORexpressionwas analyzedby immunohistochemistry in tissuemicro-array specimens of1,262 breast cancer patients with a median follow-up of 9.5 years.Results: Expression of XOR was moderately decreased in 50% and undetectable in another7% of the tumors. Decreased XOR expression was associated with poor histologic grade ofdifferentiation, ductal and lobular histologic types, large tumor size, high number of positive ax-illary lymph nodes, and high cyclooxygenase-2 expression, but not with estrogen or progester-one receptor status, Ki-67, p53, or ERBB2 amplification. Absence of XOR expression wasassociated with unfavorable outcome, and patients with no XOR expression had more thantwice the risk of distant recurrence as compared with those with a moderately decreased ornormal expression (hazard ratio, 2.21; P < 0.0001). Thiswas also true inpatientswithnode-neg-ative disease (hazard ratio, 2.75;P < 0.0001) aswell as inpatientswith small (V1cm) tumors (haz-ard ratio, 3.09; P = 0.027). In a multivariate survival analysis, negative XOR emerged as anindependent prognostic factor both in the entire series (P = 0.01) and amongpatients withnode-negative disease (P =0.0009).Conclusion: Loss of XOR identifies breast cancer patients with unfavorable prognosis.

Metabolically active and proliferating tissues require largeamounts of purine nucleotides for transmission of metabolicenergy and synthesis of nucleic acids. Xanthine oxidoreductase(XOR) catalyzes the final reactions of the purine catabolicpathway, oxidizing hypoxanthine to xanthine, and xanthine touric acid (Fig. 1). XOR is coded for by a single gene, located onhuman chromosome 2p22 (1), and the protein is mainlyexpressed in the cytoplasm of hepatocytes, intestinal epithelialcells, vascular endothelial cells, and breast acinar and ductalepithelium where it is strongly induced during lactogenesis(2, 3). In the mouse mammary gland, prolactin and dexa-methasone increase XOR de novo synthesis (4, 5), and miceheterozygous for a loss-of-function mutation in the XOR geneare unable to maintain lactation (3). Hypoxia activates XOR

both at the transcriptional and posttranscriptional levels (6, 7)and proinflammatory cytokines induce XOR transcription incell culture (8, 9).

Progressive decrease of XOR activity has been shown in themouse breast during carcinogenesis (10), and XOR activity (11)and protein (12) are decreased in rat hepatomas as well as inhuman renal carcinoma (13) as compared with thecorresponding normal tissues. In a recent study using cDNAmicroarray analysis, the XOR gene was down-regulated inchemically induced rat mammary gland carcinomas (14).

Despite the evidence of a role for XOR in cancer, the enzymehas been previously studied in only a very limited number ofhuman tumor specimens. We hypothesized that XOR may bedifferentially expressed in human cancers, breast cancer inparticular, because XOR is strongly expressed in the normalresting mammary epithelium. To address this question, weexamined the expression of XOR in an unselected nationwidepatient series, and analyzed whether the expression of XORprotein is associated with clinicopathologic variables andclinical outcome in breast cancer.

Materials and Methods

Patients and preparation of tumor tissue microarrays. Using the filesof the nationwide Finnish Cancer Registry, all women diagnosed withbreast cancer in 1991 and 1992 were identified. Five well-definedgeographic regions, comprising more than 50% of the Finnishpopulation, were selected for the study. The database includes

www.aacrjournals.orgClin Cancer Res 2005;11(12) June15, 2005 4372

Imaging, Diagnosis, Prognosis

Authors’Affiliations: 1Research Program for Developmental and ReproductiveBiology and Hospital for Children and Adolescents, Biomedicum Helsinki,University of Helsinki; 2Department of Oncology, Helsinki University CentralHospital, Helsinki, Finland; and 3Institute of Medical Technology, University ofTampere,Tampere, FinlandReceived11/8/04; revised 2/17/05; accepted 2/23/05.The costs of publication of this article were defrayed in part by the payment of pagecharges.This article must therefore be hereby marked advertisement in accordancewith18 U.S.C. Section1734 solely to indicate this fact.Requests for reprints: Nina Linder, Research Program for Developmental andReproductive Biology, Room B524b, Biomedicum Helsinki, University of Helsinki,P.O. Box 63 (Haartmaninkatu 8), FIN-00014 Helsinki, Finland. Phone: 358-9-4717-1976; Fax: 358-9-4717-1947; E-mail: [email protected].

F2005 American Association for Cancer Research.

Cancer Research. on January 28, 2020. © 2005 American Association forclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

information on clinical and pathologic characteristics extracted fromthe hospital records and, in addition, data on a series of tumormarkers determined from the primary tumor specimens (15, 16). Atotal of 2,846 patients (93% of all breast cancer patients within theselected regions) with sufficient clinical data available were enteredinto the database (the FinProg Breast Cancer Database, accessible athttp://www.finprog.org). The median follow-up of patients alive at theend of follow-up is 9.5 years (range, 0.2-10.8 years). Routinely fixedparaffin-embedded tumor samples were extracted from the files ofpathology laboratories, and histopathologically representative tumorregions were used for preparation of tumor tissue array blocks (17).From the 1,931 tumor samples available, 19 tissue microarray blockswere prepared, each containing 50 to 144 tumor samples. In addition,20 whole-slide specimens were prepared for evaluation of tumorheterogeneity.

XOR protein expressions in the normal mammary gland (n = 14)and the lactating mammary gland (n = 3) were evaluated in specimensobtained at breast surgery for reduction mammoplasty or suspectedcancer.

Immunohistochemistry. The antigen was enhanced in Target Re-trieval Solution (pH 6.0; DAKO, Carpentaria, CA) at 95jC to 97jC for30 minutes on routinely processed paraffin sections. The sections werethen treated with 3% hydrogen peroxide and XOR protein was detectedusing a well-characterized rabbit polyclonal anti-XOR antibody (2, 18).The antibody was diluted 1:50 in Blocking Solution (Powervision,Immunovision, Inc., Daly City, CA), and incubated with the samples(overnight at + 4jC). An antimouse-peroxidase polymer (30 minutes atroom temperature) and diaminobenzidine as a chromogen (Power-vision) were used for visualization. Specificity of the XOR localizationwas confirmed by staining slides with preimmune serum and withoutthe primary antibodies.

Of the 1,931 tissue core biopsies stained for XOR, 314 eitherhad detached (n = 229; 8%) or did not contain identifiable tumorcells (n = 85; 3%). Cases with in situ carcinoma, distant metastasis atthe time of diagnosis, synchronous or metachronous bilateral breastcancer, cases with a history of malignancy other than breast cancerexcept for basal cell carcinoma or cervical in situ carcinoma, andwomen who did not undergo breast surgery were excluded. Followingthese exclusions, 1,262 unilateral, invasive breast carcinomas wereeligible for analysis and had an interpretable tissue XOR stainingresult. Eight hundred seventy-eight patients (70%) were treated withmastectomy and 376 (30%) with breast conserving surgery. Postop-erative radiotherapy was given to 731 (59%) patients. Only 474(37%) of the patients had received systemic adjuvant therapy, andthese consisted of 290 (61%) patients who received tamoxifen, 177

(37%) women who were treated with cyclophosphamide, methotrex-ate, and 5-fluorouracil, and 7 (2%) who were given a combination ofchemotherapy and tamoxifen. Adjuvant systemic therapy was given to9% of the patients with node-negative disease and to 92% of thepatients with node-positive disease. Twenty of the excluded samples ofductal carcinoma in situ were separately analyzed to evaluate XORprotein expression in this entity.

Immunostainings for the estrogen receptor, the progesteronereceptor, Ki-67 antigen, and p53 protein, as well as chromogenic insitu hybridization for ERBB2 (HER-2) gene amplification, were carriedout using established procedures (16), and cyclooxygenase-2 (COX-2)protein was visualized as described elsewhere (19).

Scoring of xanthine oxidoreductase immunostaining. Expression ofXOR was evaluated by two of the investigators (N.L. and J.L.). Bothinvestigators were blinded to the clinicopathologic data at the time ofXOR expression evaluation. Cytoplasmic and nuclear XOR staining ofthe tumor cells were scored separately. Cytoplasmic XOR stainingintensity was scored as follows: strong = staining comparable to that ofthe normal epithelial cells; moderate = clearly decreased staining;negative = no staining for the XOR protein in more than 90% of thecancer cells.

Nuclear XOR staining intensity was scored as follows: strong staining;moderate staining; and negative staining. Both strong and moderatestainings of the nuclei represent clearly increased XOR expression ascompared with the normal breast.

Cell lines. The established breast cancer cell lines SkBr-3 andMCF-7 were obtained from the American Type Culture Collection(Rockville, MD). The MPE-600 cell line was a gift from GeraldineBrush Cancer Research Institute (San Francisco, CA). The cells werecultured using the recommended culture conditions and resuspendedin 50 mmol/L potassium phosphate buffer (pH 7.8).

Xanthine oxidoreductase activity measurements. XOR activity wascalculated by using [14C]xanthine as substrate and separating theproduct uric acid by high-performance liquid chromatography asdescribed (20).

Statistical analysis. The association between XOR expression andother clinicopathologic factors was analyzed using the m2 test. The levelof agreement between the observers in scoring of XOR expression wasestimated by percent agreement and j statistics.

Life-tables were calculated according to the Kaplan-Meier method.Distant disease-free survival (DDFS) was calculated from the date of thediagnosis to the first occurrence of metastases outside the locoregionalarea, or death from breast cancer in cases with missing data on the dateof distant recurrence (n = 31). Deaths due to intercurrent causes werecensored. Survival curves were compared with the log-rank test or thelog-rank test for trend in case of three or more ordered categories.Multivariate survival analyses were done with the Cox proportionalhazards model using a backward stepwise selection of variables, and aP value of 0.05 was adopted as the limit for inclusion of a covariate. Theassumption of proportional hazards was ascertained with complemen-tary log plots. All statistical tests are two sided.

Results

Xanthine oxidoreductase expression in normal and lactatingbreast. XOR was strongly expressed in the cytoplasm of allacinar cells in the terminal ducts as well as in the epithelialcells of the larger ducts in all specimens of nonlactatingnormal breast. Cytoplasmic XOR staining was also seen inendothelial cells of capillaries and arterioles. The cytoplasm ofthe stromal cells showed weak XOR expression. In thelactating mammary gland, the acinar and ductal epithelialcells were intensely stained for XOR, whereas the endothelialand the stromal cells showed similar staining as in thenonlactating mammary gland (Fig. 2).

Fig.1. A simplified overviewof the pathways leading to the formation of substratesfor XOR. Reactions catalyzed by XORand the purine salvage pathway. HPRT,hypoxanthine phosphoribosyltransferase; cGMP, cyclic GMP.

www.aacrjournals.org Clin Cancer Res 2005;11(12) June15, 20054373

XORExpression in Breast Cancer



XOR was expressed in a small fraction of the nuclei (f5%)of the acinar and ductal epithelial cells in 3 of the 14 normalbreasts studied. In the lactating breast, the nuclei of theepithelial cells were negative, whereas the nuclei of thestromal cells were weakly stained for XOR in all the breasttissue samples analyzed. Control sections stained with thepreimmune serum (Fig. 2) and slides processed withoutprimary antiserum showed no immunoreactivity (data notshown).

Xanthine oxidoreductase activity in human breast cancer celllines. None of the breast cancer cell lines studied (MCF-7,SkBr-3, and MPE-600) showed XOR activity.

Xanthine oxidoreductase expression in breast cancer. Cyto-plasmic XOR was scored into three categories in 1,262 invasivebreast carcinomas. The percent agreement between the twoindependent investigators in allocation of the tumors into thethree staining categories was 82%. The corresponding j valuewas 0.71, which can be interpreted as a good level ofagreement. All specimens with discordant scores were reeval-uated by the two investigators, and the consensus score wasused for further analyses.

Forty-three percent (n = 534) of the tumors showed strongstaining for cytoplasmic XOR similar to the XOR expression inthe normal breast, whereas 50% (n = 634) showed moderatestaining, corresponding to a clearly decreased XOR expression,and 7% (n = 93) had no cytoplasmic XOR expression (Fig. 2).In the 20 ductal carcinomas in situ samples analyzed, 50%

(n = 10) showed strong XOR expression, 40% (n = 8) weremoderate, and 10% (n = 2) were negative. The cytoplasms ofthe stromal cells were either negative or weakly positive for theXOR protein.

Although most nuclei of the normal breasts did not expressXOR, 30% (n = 334) of the breast tumors showed moderateand 21% (n = 227) showed strong nuclear expression (Fig. 2),and only 49% (n = 535) were negative for nuclear XOR.

The heterogeneity of XOR expression was minimal in the20 whole tumor section slides studied. Therefore, the stainingresults obtained from the tissue microarray cores wereconsidered as representative for the entire tumor.

Association of xanthine oxidoreductase expression with clinico-pathologic variables. Decreased cytoplasmic XOR expressionwas significantly associated with poor histologic grade ofdifferentiation, ductal and lobular histologic type, high COX-2 expression, high number of metastatic axillary lymph nodes,and large tumor size, whereas no statistically significantassociation was found between cytoplasmic XOR and age atdiagnosis, estrogen receptor or progesterone receptor status,Ki-67 or p53 protein expression, or ERBB2 amplification(Table 1).

Presence of nuclear XOR expression (moderate or strongstaining) was significantly associated with small primary tumorsize (P = 0.033), absence of ERBB2 amplification (P = 0.025),and low COX-2 (P < 0.0001), but not with any of the otherclinicopathologic characteristics analyzed.

Fig. 2. XOR antibody stained sections of normal (A and B), and lactating (C) human breast tissues. Ductal carcinoma in situ ; normal epithelium (N) and carcinoma cells(Ca ; E), and tumor microarray cores with primary invasive breast cancer tissue showing strong (F), moderate (G), and negative (H), and nuclear (I) staining of XOR.Representative control staining of normal breast processed with the preimmune rabbit serum (D). Bar, 25 Am.





Association of xanthine oxidoreductase expression with distantdisease-free survival. Decreased cytoplasmic XOR expressionwas significantly associated with decreased DDFS among the1,262 breast cancer patients (log-rank for trend, P < 0.0001;Fig. 3). The difference in DDFS between patients with normal(strong staining) versus decreased (either moderate or nostaining) cytoplasmic XOR expression was highly significant,but the greatest decrease in survival was seen in breast cancerpatients who had no detectable cytoplasmic XOR expression

(Table 2). Therefore, the groups with strong and moderate XORstainings were combined for further survival analyses (Table 2;Fig. 4). Patients with strong or moderate XOR expression hadan 8-year DDFS of 76%, whereas those with no XOR expressionhad a DDFS of only 52% (Table 2).

Absence of cytoplasmic XOR expression was associatedwith poor outcome in almost all subgroups analyzed. Significantdifferences were observed in patients with breast tumors ofductal type, in those with axillary node–negative disease as well

Table1. Association of cytoplasmic XORexpressionwith clinicopathologic characteristics

Parameter N XOR C2 P

Strong (%) Moderate (%) Negative (%)

Age at diagnosis (y)<50 406 169 (42) 202 (50) 35 (9) 1.38 0.50z50 856 366 (43) 432 (50) 58 (7)

Tumor size (cm)0.1-1 224 107 (48) 105 (47) 12 (5) 12.98 0.0431.1-2 503 223 (44) 245 (49) 35 (7)2.1-5 430 164 (38) 232 (54) 34 (8)>5 48 14 (29) 27 (56) 7 (15)

Axillary nodal status0 771 344 (45) 376 (49) 51 (7) 14.79 0.0221-3 285 120 (42) 147 (52) 18 (6)4-9 99 31 (31) 57 (58) 11 (11)z10 26 7 (27) 14 (54) 5 (19)

Histologic grade1 216 114 (53) 93 (43) 9 (4) 22.33 0.00022 447 179 (40) 236 (53) 32 (7)3 251 82 (33) 143 (57) 26 (10)

Histologic typeDuctal 932 395 (42) 469 (50) 68 (7) 20.31 0.0004Lobular 202 66 (33) 118 (58) 18 (9)Special 128 74 (58) 47 (37) 7 (5)

Estrogen receptor statusPositive 652 270 (41) 336 (52) 46 (7) 0.17 0.92Negative 306 131 (43) 154 (50) 21 (7)

Progesterone receptor statusPositive 571 241(42) 289 (51) 41 (7) 0.19 0.91Negative 387 160 (41) 201 (52) 26 (7)

Ki-67Negative 67 28 (42) 32 (48) 7 (10) 2.21 0.70Low 586 250 (43) 298 (51) 38 (6)High 385 169 (44) 194 (50) 22 (6)

p53Negative 509 202 (40) 262 (51) 45 (9) 6.67 0.15Low 320 148 (46) 155 (48) 17 (5)High 198 76 (38) 107 (54) 15 (8)

ERBB2 amplificationNegative 908 387 (43) 458 (50) 63 (7) 2.40 0.30Positive 208 80 (38) 108 (52) 20 (10)

COX-2Negative-low 731 329 (45) 347 (47) 55 (8) 8.39 0.015Moderate-high 473 175 (37) 264 (56) 34 (7)

Method of detectionScreen detected 241 112 (46) 112 (46) 17 (7) 2.29 0.32Not screen detected 992 408 (41) 510 (51) 74 (7)





as node-positive disease, in patients with small tumors 0.1 to

1 cm, and in those with tumors larger than 2 cm (Table 2; Fig. 4).Survival analysis within subgroups according to histologic gradewas restricted to tumors of ductal type, and significant survivaldifferences according to XOR expression were found both withingrade 1 and grade 2 to 3, even though the number of patientswith XOR-negative grade 1 tumors was small. XOR showedsignificant prognostic value among patients with estrogenreceptor–positive and those with estrogen receptor–negativetumors (Fig. 4), as well as among subgroups of progesteronereceptor–positive and progesterone receptor–negative tumors.Furthermore, within subgroups defined by low and high Ki-67expression, low and high p53 expression, negative and positiveERBB2 amplification, as well as in tumors with low andhigh COX-2 expression, XOR expression status further dividedthe patients into groups with significantly different DDFS(Table 2).

Fig. 3. DDFS of1,262 patients with breast cancer according to XOR proteinexpression. ��, strong (n = 535); - - -, moderate (n = 634); D, negative (n = 93).P < 0.0001 (log-rank test for trend).

Table 2. Eight-year DDFS according to the cytoplasmic XORexpression level

Clinicopathologic variable XOR score n DDFS (95% CI) P* Risk ratio (95% CI)

XORexpression Strong 535 79 (75-82) Refc 1Moderate 634 74 (71-78) 0.084 1.23 (0.97-1.56)Negative 93 52 (41-63) <0.0001 2.50 (1.75-3.56)

Strong 535 79 (75-82) 0.0067 1.37 (1.09-1.72)Moderate-negative 727 71 (68-75)

Strong-moderate 1,169 76 (74-79) <0.0001 2.21 (1.60-3.08)Negative 93 52 (41-63)

Age at diagnosis (y)V50 Strong-moderate 371 77 (73-82) 0.0079 2.05 (1.19-3.54)

Negative 35 57 (41-74)

>50 Strong-moderate 798 75 (72-79) <0.0001 2.35 (1.54-3.56)Negative 58 48 (33-63)

Tumor size (cm)0.1-1 Strong-moderate 212 91 (87-95) 0.027 3.09 (1.08-8.91)

Negative 12 67 (40-93)

1.1-2 Strong-moderate 468 79 (75-83) 0.47 1.29 (0.65-2.55)Negative 35 71 (55-88)

>2 Strong-moderate 437 67 (62-72) <0.0001 2.65 (1.76-4.03)Negative 41 31 (16-47)

Axillary node statusNegative Strong-moderate 720 86 (84-89) <0.0001 2.75 (1.65-4.59)

Negative 51 63 (49-78)

Positive Strong-moderate 397 59 (54-64) 0.0043 1.87 (1.21-2.91)Negative 38 37 (20-53)

Histologic gradeb

1 Strong-moderate 140 91 (86-96) 0.019 7.75 (1.12-56.5)Negative 5 50

2-3 Strong-moderate 568 69 (64-73) 0.0027 1.83 (1.23-2.73)Negative 53 45 (31-59)

Histologic typeDuctal Strong-moderate 864 73 (70-77) <0.0001 2.45 (1.73-3.50)

Negative 68 43 (31-56)

(Continued on the following page)





In contrast to cytoplasmic XOR expression, nuclear XORexpression showed no statistically significant association withsurvival. Patients with strong or moderate nuclear XORexpression had an 8-year DDFS of 75% [n = 561; 95%confidence interval (95% CI), 71-79] and those with no nuclearXOR expression had a DDFS of 72% (n = 535; 95% CI, 68-76);(risk ratio, 1.09; 95% CI, 0.86-1.37; P = 0.48).

The DDFS of the patients whose tumor XOR expressioncould not be analyzed (tissue array cores were detached orlacking tumor cells; n = 314) did not differ from that of thepatients who had an interpretable XOR staining result (n =1,262; P = 0.24).

Multivariate analysis. Absence of cytoplasmic XOR expres-sion was an independent prognostic factor in a Cox

Clinicopathologic variable XOR score n DDFS (95% CI) P* Risk ratio (95% CI)

Lobular Strong-moderate 184 79 (73-85) 0.35 1.56 (0.61-3.96)Negative 18 71 (44-97)

Special Strong-moderate 121 91 (86-97) 0.40 0.05 (0-2657)Negative 7 100 (100-100)

Estrogen receptorPositive Strong-moderate 606 76 (73-80) 0.0014 2.10 (1.32-3.36)

Negative 46 54 (39-70)

Negative Strong-moderate 285 70 (64-75) 0.019 2.09 (1.12-3.92)Negative 21 45 (22-68)

Progesterone receptorPositive Strong-moderate 530 80 (76-83) 0.04 1.77 (1.02-3.10)

Negative 41 64 (48-80)

Negative Strong-moderate 361 67 (62-72) 0.0001 2.62 (1.58-4.37)Negative 26 32 (14-50)

Ki-67Neg-low Strong-moderate 608 80 (77-84) <0.0001 2.99 (1.86-4.81)

Negative 45 50 (34-66)

High Strong-moderate 363 66 (61-71) 0.0001 2.75 (1.62-4.73)Negative 22 27 (7-47)

p53Neg-low Strong-moderate 767 78 (75-81) <0.0001 2.82 (1.91-4.21)

Negative 62 47 (33-61)

High Strong-moderate 183 66 (59-74) 0.0065 2.45 (1.25-4.76)Negative 15 33 (10-57)

ERBB2 amplificationNegative Strong-moderate 845 79 (76-82) <0.0001 2.43 (1.62-3.65)

Negative 63 53 (40-67)

Positive Strong-moderate 188 60 (52-67) 0.0197 2.03 (1.10-3.74)Negative 20 36 (11-61)

COX-2Neg-low Strong-moderate 676 80 (77-83) <0.0001 2.62 (1.69-4.07)

Negative 55 54 (40-69)

Mod-high Strong-moderate 439 70 (65-74) 0.0031 2.07 (1.27-3.40)Negative 34 41 (23-60)

Method of detectionScreen Strong-moderate 224 90 (86-94) 0.92 1.06 (0.25-4.48)

Negative 17 88 (73-100)

Not screen Strong-moderate 918 73 (70-76) <0.0001 2.35 (1.67-3.33)Negative 74 44 (31-57)

*All statistical tests were two sided.cRef, reference category.bDuctal type only.

Table 2. Eight-year DDFS according to the cytoplasmic XORexpression level (Cont’d)





multivariate analysis along with the number of positiveaxillary lymph nodes, primary tumor size, progesteronereceptor status, and histologic grade, whereas age, histologictype, and estrogen receptor and ERBB2 status were not

retained in the model (Table 3). In patients with node-negative breast cancer who did not receive adjuvant therapy,absence of cytoplasmic XOR expression retained its inde-pendent prognostic value in addition to size and grade,

Fig. 4. A , DDFS of 771patients with node-negative breast cancer according to XOR protein expression. D, moderate to strong* (n = 720); D, negative (n = 51). P < 0.0001(log-rank test). B, DDFS of 224 patients with breast cancer with a tumor less than1cm in diameter according to XOR protein expression. D, moderate to strong* (n = 212);D, negative (n = 12). P = 0.02 (log-rank test). C, DDFS of 775 patients with estrogen receptor ^ positive breast cancer according to XOR protein expression. D, moderate tostrong* (n = 722); D, negative (n = 53). P = 0.0015 (log-rank test). D, DDFS of 352 patients with estrogen receptor ^ negative breast cancer according to XOR proteinexpression.D,moderate to strong* (n = 323);D, negative (n = 29). P < 0.0001 (log-rank test). *, therewas little difference in patient outcome between strong andmoderateXOR staining and, therefore, these groups were combined and compared with the patients with absent XOR in the survival analysis.

Table 3. Multivariate survival analysis in 741breast cancer patients

Clinicopathologic variable b-coef P C2 Hazard ratio 95% CI

No. positive lymphnodes 0.140 <0.0001 69.8 1.15 1.11-1.19Tumor size (cm) 0.179 <0.0001 16.9 1.19 1.10-1.30Progesterone receptor (pos vs neg) 0.529 0.0004 12.4 1.70 1.26-2.28Grade (grade1vs 2-3) 0.798 0.0022 9.2 2.22 1.33-3.72XOR

Strong 1.00Moderate 0.023 0.89 0.02 1.02 0.75-1.40Negative 0.524 0.027 4.86 1.69 1.06-2.69

Estrogen receptor (pos vs neg) NSHistologic type (nonductal vs ductal) NSERBB2 (pos vs neg) NSAge at diagnosis (V50 vs >50 y) NS

NOTE: Cox proportional hazards regression models for DDFS in 741patients with breast cancer. The total patient data set was reduced in the Cox regression analysesbecause of the need for all patients to have data available on all the covariates. b-coef, regression coefficient; NS, not significant. Number of positive lymph nodes andtumor sizewere entered as continuous variables; XORexpression as a categorical variable; and grade, progesterone receptor, estrogen receptor, histologic type, andage asbinary variables.





whereas XOR expression did not have independent prog-nostic value among patients with node-positive breast cancer(Tables 4 and 5).

Discussion

XOR seems to play an important role in the physiologyof the mammary gland because it is expressed in normalbreast epithelium and induced on lactation (2, 5), andheterozygosity for a loss-of-function mutation of XOR causesa failure of lactation (3). As compared with the cytoplasmof normal resting breast epithelium and lactating breastepithelial cells, expression of the XOR protein was decreasedin 50% of breast carcinomas and undetectable in another 7%

of the cancers. Decreased cytoplasmic expression of XORwas associated with several adverse prognostic features, suchas poor grade of differentiation, large primary tumor size,high number of positive axillary lymph nodes, and COX-2expression. Yet, these associations were relatively weak, andno significant association was found between expression ofcytoplasmic XOR and that of estrogen receptor, progesteronereceptor, Ki-67 antigen, and p53 protein, or the presence ofERBB2 amplification.

Patients whose tumor tissue did not contain immunoreactiveXOR had about twice as high risk for distant recurrence aswomen whose cancer expressed at least some cytoplasmic XORprotein. Absence of immunoreactive XOR was relativelyuncommon (7% of all tumors), but had independentprognostic significance both in the entire patient series and in

Table 4. Multivariate survival analysis in a subgroup of 468 node-negative breast cancer patients


Tumor size (cm) 0.377 0.0001 14.85 1.46 1.20-1.77XOR

Strong 1.00Moderate �0.053 0.84 0.04 0.95 0.57-1.58Negative 1.025 0.003 8.74 2.79 1.41-5.50

Grade (grade1vs 2-3) 1.047 0.004 8.26 2.85 1.40-5.82Progesterone receptor (pos vs neg) NSEstrogen receptor (pos vs neg) NSHistologic type (nonductal vs ductal) NSERBB2 (pos vs neg) NSAge at diagnosis (V50 vs >50 y) NS

NOTE: Cox proportional hazards regressionmodels for DDFS in a subgroup of 468 patients with node-negative breast cancer not treated with adjuvant systemic therapy.The total patient data set was reduced in the Cox regression analyses because of the need for all patients to have data available on all the covariates. Number of positivelymph nodes and tumor size were entered as continuous variables; XOR expression as a categorical variable; and grade, progesterone receptor, estrogen receptor, histo-logic type, and age as binary variables.

Table 5. Multivariate survival analysis in a subgroup of 365 node-positive breast cancer patients


No. positive lymphnodes 0.139 <0.0001 63.93 1.15 1.11-1.19Tumor size (cm) 0.111 0.0096 6.72 1.12 1.03-1.22Progesterone receptor (pos vs neg) 0.611 0.0004 12.77 1.84 1.32-2.58XOR

Strong 1.00Moderate 0.075 0.68 0.17 1.08 0.75-1.54Negative 0.328 0.25 1.34 1.39 0.80-2.42

Grade (grade1vs 2-3) NSEstrogen receptor (pos vs neg) NSHistologic type (nonductal vs ductal) NSERBB2 (pos vs neg) NSAge at diagnosis (V50 vs >50 y) NS

NOTE: Cox proportional hazards regression models for DDFS in a subgroup of 365 patients with node-positive breast cancer of whom 337 (92%) were treated withadjuvant systemic therapy. The total patient data set was reduced in the Cox regression analyses because of the need for all patients to have data available on all thecovariates. Number of positive lymph nodes and tumor size were entered as continuous variables; XOR expression as a categorical variable; and grade, progesteronereceptor, estrogen receptor, histologic type, and age as binary variables.





the subgroup of axillary node–negative disease. XOR expres-sion was not associated with outcome in patients with node-positive disease, of whom a majority received adjuvant systemictherapy, unlike the women with node-negative disease. Furtherresearch needs to be carried out to study whether XORexpression of the tumor may be related to the response toadjuvant therapy.

Little is known about the expression of XOR in differenttypes of human cancer, but the sparse data available suggestthat XOR activity may decrease in malignant tumors becauseXOR expression was low or absent in six hepatocellularcarcinomas (21) and in four invasive breast carcinomasstudied (22). We found that XOR activity is absent in malig-nant cell lines originating from human breast carcinoma,including MPE-600 cells which have a relatively well-preservedoriginal karyotype (23). Although a statistically significantassociation between decreased XOR expression and poor histo-logic grade was observed in our study, 33% of the poorlydifferentiated tumors showed strong XOR expression, indica-ting that dedifferentiation does not invariably lead to loss ofXOR. The mechanisms of down-regulation of XOR in cancerremain unknown and might include loss of heterozygosityat chromosome 2p where XOR is located (24), decreasedgene promoter activity, increased XOR mRNA degradation,or posttranslational changes.

Purine nucleotides are made available for cells via tworoutes, either by de novo synthesis from low molecular weightprecursors or by reutilization of nucleotides catabolized topurine bases, mainly hypoxanthine. XOR and hypoxanthinephosphoribosyltransferase compete for the substrate hypo-xanthine, resulting either in irreversible purine loss, whenXOR oxidizes hypoxanthine to xanthine and further to uricacid, or in salvage of the purine ring, when hypoxanthinephosphoribosyltransferase converts hypoxanthine to IMP andfurther to adenine and guanine nucleotides (Fig. 1). Salvage

of the purine ring is six times more efficient in terms of ATPequivalents than de novo purine synthesis, and thus cancercells with effective shunting of purine bases to the salvagepathway might gain a growth advantage. Some evidence fora shift in the purine anabolic-catabolic balance can be foundin studies on rodent tumors in which XOR activity hasbeen found to be decreased (12, 21, 25–27) as comparedwith the corresponding normal tissues, whereas the acti-vities of hypoxanthine phosphoribosyltransferase (28) andthe enzymes involved in purine biosynthesis are increased(25, 29).

Interestingly, no association between XOR expression andthe hormone receptor status was found. This is unlike mostother cancer biological factors we have investigated in the samebreast cancer series, such as p53, Ki-67 antigen, and ERBB2expression, and suggests that down-regulation of XOR mayoccur independently of the estrogen and progesterone receptorpathways.

In conclusion, many breast cancer samples stain positivelyfor XOR, but its expression in the cytoplasm is usually lessthan that found in normal breast acinar and ductal cells.Breast cancers that lack cytoplasmic XOR expression areassociated with about a 2.5-fold greater risk for distantmetastases than cancers showing strong XOR expression.The reasons why loss of XOR is associated with pooroutcome in breast cancer remain unknown. Further studieswill elucidate whether tumors lacking XOR respond differ-ently to adjuvant therapy compared with those showingXOR expression.

Acknowledgments

We thankTiina LehtimJki, Kaija Holli, Liisa Elomaa, Liisa Pylkkanen,Vesa Kataja,andTainaTurpeenniemi-Hujanen for collecting the clinical and follow-up data.


References1. Rytkonen EM,Halila R, LaanM, et al.Thehumangenefor xanthine dehydrogenase (XDH) is localized onchromosome band 2q22. Cytogenet Cell Genet1995;68:61^3.

2. Linder N, Rapola J, Raivio KO. Cellular expressionof human xanthine oxidoreductase protein in normalhuman tissues. Lab Invest1999;79:967^74.

3.Vorbach C, Scriven A, CapecchiMR.Thehousekeep-ing gene xanthine oxidoreductase is necessary formilkfat droplet enveloping and secretion: gene sharing inthe lactating mammary gland. Genes Dev 2002;16:3223^35.

4. McManamanJL, Hanson L, Neville MC,Wright RM.Lactogenic hormones regulate xanthine oxidoreduc-tase and h-casein levels inmammary epithelial cells bydistinct mechanisms. Arch Biochem Biophys 2000;373:318^27.

5. Kurosaki M, Zanotta S, Li CM, Garattini E, TeraoM. Expression of xanthine oxidoreductase inmouse mammary epithelium during pregnancy andlactation: regulation of gene expression by gluco-corticoids and prolactin. Biochem J 1996;319:801^10.

6. LinderN,Martelin E, Lapatto R, Raivio KO.Posttrans-lational inactivationof humanxanthine oxidoreductaseby oxygen under standard cell culture conditions. AmJPhysiol Cell Physiol 2003;285:C48^55.

7. Terada LS, Guidot DM, Leff JA, et al. Hypoxia injuresendothelial cells by increasing endogenous xanthine

oxidase activity. Proc Natl Acad Sci U S A 1992;89:3362^6.

8. Pfeffer KD, Huecksteadt TP, Hoidal JR. Xanthine de-hydrogenase and xanthine oxidase activity and geneexpression in renal epithelial cells. Cytokine and ste-roid regulation. J Immunol1994;153:1789^97.

9. Dupont GP, HuecksteadtTP, Marshall BC, Ryan US,Michael JR, Hoidal JR. Regulation of xanthine dehy-drogenase and xanthine oxidase activity and gene ex-pression in cultured rat pulmonary endothelial cells.JClin Invest1992;89:197^202.

10. Lewin I, LewinR, Bray RC. Xanthine oxidase activityduring mammary carcinogenesis in mice. Nature1957;180:763^4.

11. Prajda N,Weber G. Malignant transformation-linkedimbalance: decreased xanthine oxidase activity in he-patomas. FEBSLett1975;59:245^9.

12. Ikegami T, NatsumedaY,Weber G. Decreased con-centration of xanthine dehydrogenase (EC 1.1.1.204)in rat hepatomas. Cancer Res1986;46:3838^41.

13. Durak I, BedukY, Kavutcu M, et al. Activity of theenzymes participating in purine metabolism of cancer-ous and noncancerous human kidney tissues. CancerInvest1997;15:212^6.

14. Shan L, HeM,YuM, et al. cDNAmicroarray profilingofratmammaryglandcarcinomasinducedby2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine and 7,12-dimethylbenz[a]anthracene. Carcinogenesis 2002;23:1561^8.

15. LundinJ, LundinM, Holli K, et al. Omission of histo-logic grading from clinical decision making may resultin overuse of adjuvant therapies in breast cancer:results from a nationwide study. J Clin Oncol 2001;19:28^36.

16. Joensuu H, Isola J, Lundin M, et al. Amplificationof erbB2 and erbB2 Expression Are Superior to Es-trogen Receptor Status As Risk Factors for DistantRecurrence in pT1N0M0 Breast Cancer: A NationwidePopulation-based Study. Clin Cancer Res 2003;9:923^30.

17. KononenJ, Bubendorf L, Kallioniemi A, et al. Tissuemicroarrays for high-throughput molecular profiling oftumor specimens. NatMed1998;4:844^7.

18. SarnestoA, Linder N, Raivio KO. Organ distributionand molecular forms of human xanthine dehydroge-nase/xanthine oxidase protein. Lab Invest 1996;74:48^56.

19. Ristimaki A, Sivula A, LundinJ, et al. Prognostic sig-nificance of elevated cyclooxygenase-2 expression inbreast cancer. Cancer Res 2002;62:632^5.

20. Saksela M, Lapatto R, Raivio KO. Xanthine oxido-reductase gene expression and enzyme activity indeveloping human tissues. Biol Neonate 1998;74:274^80.

21. Stirpe F, Ravaioli M, Battelli MG, Musiani S,Grazi GL. Xanthine oxidoreductase activity in hu-man liver disease. Am J Gastroenterol 2002;97:2079^85.





22. Cook WS, Chu R, Saksela M, Raivio KO,YeldandiAV. Differential immunohistochemical locali-zation of xanthine oxidase in normal and neo-plastic human breast epithelium. Int J Oncol 1997;11:1013^7.

23. Rummukainen J, Kytola S, Karhu R, Farnebo F,Larsson C, Isola JJ. Aberrations of chromosome 8 in16 breast cancer cell lines by comparative genomichybridization, fluorescence in situ hybridization, andspectral karyotyping. Cancer Genet Cytogenet 2001;126:1^7.

24. O’Connell P, Pekkel V, Fuqua SA, Osborne CK,Clark GM, Allred DC. Analysis of loss ofheterozygosity in 399 premalignant breast lesionsat 15 genetic loci. J Natl Cancer Inst 1998;90:697^703.

25. Prajda N, Morris HP,Weber G. Imbalance of purinemetabolism in hepatomas of different growth rates asexpressed inbehaviorof xanthine oxidase (EC1.2.3.2).Cancer Res1976;36:4639^46.

26.Weber G, Hager JC, Lui MS, et al. Biochemicalprograms for slowly and rapidly growing human co-

lon carcinoma xenografts. Cancer Res 1981;41:854^9.

27. Prajda N, Donohue JP,Weber G. Increased amido-phosphoribosyltransferase and decreased xanthineoxidase activity in human and rat renal cell carcinoma.Life Sci1981;29:853^60.

28. MurrayAW. Purine Phosporibosyltransferase acti-vities in rat and mouse tissues and in Ehrlich ascites-tumor cells. BiochemJ1966;100:664^70.

29.Weber G. Enzymes in purine metabolism in cancer.Clin Biochem1983;16:57^63.




2005;11:4372-4381. Clin Cancer Res Nina Linder, Johan Lundin, Jorma Isola, et al. Aggressive Breast CancerDown-Regulated Xanthine Oxidoreductase Is a Feature of

Updated version

http://clincancerres.aacrjournals.org/content/11/12/4372

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/11/12/4372.full#ref-list-1

This article cites 27 articles, 10 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/11/12/4372.full#related-urls

This article has been cited by 3 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] at

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

Permissions

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

.http://clincancerres.aacrjournals.org/content/11/12/4372To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/11/12/4372.full#ref-list-1

http://clincancerres.aacrjournals.org/content/11/12/4372.full#related-urls

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

mailto:[email protected]



Documents

Down-Regulated Xanthine Oxidoreductase Is a Feature of ...clincancerres.aacrjournals.org/content/clincanres/11/12/4372.full.pdfDown-Regulated Xanthine Oxidoreductase Is a Feature of