Purpose: p21 is a main effector of growth arrest induced by p53. In addition, a second transcript from the same gene (p21B) has been linked to apoptosis. We previously analyzed p21 status in breast cancer and reported two novel polymorphisms of the p21 gene. In the present study, we present a larger study designed to explore a possible association between these novel polymorphisms and breast cancer.

Experimental Design: The p21/p21B polymorphisms were analyzed in 507 breast cancer patients and 1,017 healthy individuals using cDNA or genomic DNA from tumor and/or blood samples.

Results: We detected five polymorphisms of the p21 gene. Three of these polymorphisms are earlier reported by others, whereas two were reported for the first time in a recent study by us. The presence of the A allele of the p21G251A polymorphism was observed more frequently among patients with primary stage III breast cancer (4.5%) compared with stage I and II tumors (1.5%) and healthy female controls (1.4%; P = 0.007, comparing the three groups; P = 0.0049 and P = 0.0057, comparing locally advanced to stage I/II and healthy controls, or to healthy controls alone, respectively). The allele frequencies of the remaining four polymorphisms were evenly distributed among patients and healthy individuals.

Discussion: The finding of an association between locally advanced breast cancer and one particular polymorphism of the p21 gene suggests this polymorphism to be related to tumor behavior, including enhanced growth rate. If confirmed in other studies, this may add significant information to our understanding of the biology as well as of the clinical behaviour of locally advanced breast cancers.

The cyclin-dependent kinase inhibitor p21 (CIP1, WAF1, CDKN1A) is thought to be the main executor of p53-induced growth arrest (14). Transcriptionally induced by p53, the p21 protein binds and inhibits the cdk2-cyclin B/E complexes but increases stability and nuclear localization of cyclin D, stimulating assembly of the cdk4-cyclin D complex (3, 513). Interestingly, a second p53-induced transcript from the p21 gene, p21B (Fig. 1) was recently identified and found to encode a protein expressing proapoptotic activity (14), suggesting a role for the p21 gene in both cell cycle arrest and apoptosis.

Fig. 1.

p21 and p21B with observed polymorphisms. Organization of the coding exons of p21 and p21B with the observed polymorphisms. The functional domains of p21B are not known, but the p21 domains are shown together with the position of the polymorphisms. The CDK inhibitor domain (CDI) is the defining part of the CIP/KIP protein family. p21 contains two cyclin-binding domains (Cy1 and Cy2) and a CDK-binding domain, as well as a nuclear localization domain (NLS). None of the p21 polymorphisms directly affect any functional domains of p21.

Fig. 1.

p21 and p21B with observed polymorphisms. Organization of the coding exons of p21 and p21B with the observed polymorphisms. The functional domains of p21B are not known, but the p21 domains are shown together with the position of the polymorphisms. The CDK inhibitor domain (CDI) is the defining part of the CIP/KIP protein family. p21 contains two cyclin-binding domains (Cy1 and Cy2) and a CDK-binding domain, as well as a nuclear localization domain (NLS). None of the p21 polymorphisms directly affect any functional domains of p21.

Close modal

We recently described p21/p21B status in 73 locally advanced breast cancers (15), observing no promoter methylation or somatic mutations. We did, however, observe two novel polymorphisms, p21G251A (Arg84Gln) and p21BT35C (Leu12Pro), in addition to the previously known p21C93A (Ser31Arg). Although several groups have reported an increased frequency of the p21C93A polymorphism in cancer patients (1620), others reported a similar distribution of the polymorphism in both cancer and healthy individuals (2125). The frequencies of patients harboring the A allele of p21G251A or the C allele of p21BT35C were 8.2% and 16.4%, respectively. Although we are not aware of other analyses of the p21B coding sequence, the fact that the p21G251A polymorphism had not been reported previously made us hypothesize that it may occur at low frequency in non–breast cancer individuals. Analyzing a large set of healthy individuals and breast cancer patients, we here report an increased frequency of the p21G251A polymorphism among breast cancer patients. Most interestingly, this increased frequency was limited to the subgroup of breast cancer patients diagnosed with locally advanced tumors.

Patients. In this study, we analyzed samples from several cohorts of patients and healthy controls. Ninety-one and 35 patients all suffering from locally advanced breast cancer enrolled in prospective studies exploring resistance to treatment with doxorubicin (26) or mitomycin and 5-fluorouracil (27), respectively, were included. In addition, we analyzed an extra group of six patients with locally advanced breast cancer that were excluded from the mitomycin and 5-fluorouracil study for reasons other than tumor status and 60 patients from a prospective study regarding resistance to paclitaxel. Seventy-three of the locally advanced breast cancers were analyzed in our previous work on p21 (15).

In addition, we analyzed tumor samples from 201 patients suffering from metastatic breast cancer and blood DNA from 170 breast cancer patients (132 stage I-II and 38 stage III) collected at the Norwegian Radiumhospital during the period 1984 to 1990 under an institutional review board–approved protocol.

As controls, we analyzed blood DNA from a total of 1,017 healthy postmenopausal Norwegian females recruited from the national mammographic program into other studies described elsewhere (28, 29).

RNA purification. RNA was purified by Trizol (Life Technologies, Gaithersburg, MD) extraction from snap-frozen tissue samples as per the manufacturer's instructions. After extraction, the RNA was dissolved in 100 μL DEPC-treated double-distilled water. cDNA was synthesized by reverse transcription-PCR from 4 μL RNA solution using either Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA) or Transcriptor reverse transcriptase (Roche, Indianapolis, IN).

DNA purification. Genomic DNA was prepared from lymphocytes using chloroform/phenol extraction followed by ethanol precipitation (Nucleic Acid Extractor 340A; Applied Biosystems, Foster City, CA) according to standard procedure for the samples from the Norwegian Radiumhospital, or by QIAamp DNA Mini kit (Qiagen, Chatsworth, CA) according to the manufacturer's instructions. Before the latter procedure, lymphocytes were prepared by cultivating 0.4 mL blood at 37°C for 3 days in 10 mL Medium 199 (Life Technologies) containing 18% FCS (Life Technologies).

Genotyping of p21 and p21B. The p21 gene was analyzed on cDNA regarding patients from the doxorubicin, mitomycin and 5-fluorouracil, paclitaxel, and metastatic protocols. Results were verified on genomic DNA from blood when available. The remaining patients and all controls were analyzed on DNA from blood. Amplification of p21 and p21B, both from cDNA and genomic DNA, were carried out as described earlier (15).

One microliter of the PCR product was used as template for sequencing, and the sequencing reaction was done with the Big Dye terminator mix (Applied Biosystems), as described (15). Primers for the sequencing reactions were the same as those used for PCR amplification. Capillary gel electrophoresis, data collection, and sequence analysis were done on an automated DNA sequencer (ABI 3700).

Haplotype analysis. Haplotypes were first constructed using the program PHASE 2.1 (30, 31). PHASE 2.1 was also used to perform permutation testing to determine if there was a difference in the distribution of haplotypes between cases and controls. The permutation test examines the null hypothesis that the case and control haplotypes are a random sample from a single set of haplotype frequencies versus the alternative hypothesis that cases are more similar to other cases than to control haplotypes. The significance was calculated giving the P for a permutation test of the null hypothesis. The algorithm was run on 10 permuted data sets for 10 separate times to check for consistency. The final run lasted 10 times longer than the other runs to increase the number of interactions. The final output was the run with the best average value for the goodness of fit variables. The extent of linkage disequilibrium was expressed as |D′| (recombination rates) and r2 (association power) as calculated using DNAsp, and significance was calculated using Fisher's exact tests with Bonferroni correction for multiple testing (32). The result was visualized using the GOLD software (33).

The program Haploview (34) was used to perform an independent analysis of the haplotypes in cases versus control, and the result was compared with the results from PHASE analysis. The software was also used to ensure that the genotypes are in Hardy-Weinberg equilibrium, to exclude technical genotyping errors.

Statistical analysis. Each 2 × 2 test was done using the Fisher's exact test. When three or more groups were compared, analysis was done by the chi-quadrate test. All computations were done using the SAS System (SAS Institute, Inc., Cary, SC); Ps reported are two–sided.

In our previous study analyzing locally advanced breast cancers (15), we reported one known (p21C93A) and two novel (p21G251A and p21BT35C) polymorphisms of the p21 gene (Fig. 1). Preliminary screening of the p21 and p21B coding sequence in 100 healthy controls (29) indicated that the allele frequencies of the p21C93A and the p21BT35C polymorphisms were equally distributed in normal individuals and tumor samples. However, we found the A allele of the novel p21G251A polymorphism in only 1 out of 100 normal samples compared with 6 of 73 individuals suffering from breast cancer (P = 0.043, Fisher's exact test). Based on these preliminary findings, we increased the number of patients and controls to test whether this association would remain statistically significant.

In addition to the three polymorphisms reported in our original study (15), we observed two other p21 polymorphisms (Fig. 1). The polymorphism p21C515T is located 20 bp downstream of the p21 stop codon and rarely analyzed for but is always reported together with the p21C93A polymorphism (17). Similarly, all patients in our study carrying the p21C93A polymorphism that were analyzed for the region encompassing nucleotide 515 (n = 27) also carried the p21C515T polymorphism. The G allele of the previously reported p21T74G (Val25Gly) polymorphism (35) was recorded once, in a patient suffering from locally advanced breast cancer (Fig. 1).

Our observations on p21C93A, p21G251A, and p21BT35C are summarized in Tables 1, 2 and 3 respectively. Homozygosity for the rare allele is infrequent (Tables 1-3), and all further statistical analyses thus concern wild-type individuals versus the combined groups of heterozygous and homozygous for the rare allele. The p21C93A and p21BT35C polymorphisms occur at similar frequencies among healthy controls and cancer patients. (Frequency of presence of the rare allele was 9-11% for each polymorphism in each group; Tables 1 and 3). p21G251A is more infrequent, present in only in 1.9% of the samples (Table 2). Although presence of the rare allele is observed more frequently in patients (2.7%) than in controls (1.0 %), the difference is not statistically significant (P = 0.076, Fisher's exact test). Dividing our patients into two subgroups based on tumor stage, we found the p21G251A variant in 4.5% of our locally advanced breast cancers versus 1.5% among individuals diagnosed with stage I and II tumors (P = 0.0526, Fisher's exact test). Comparing the frequency of p21G251A among healthy controls, stage I/II patients, and locally advanced patients, a clear difference between the three groups was detected (P = 0.007, χ2). Similarly, comparing stage III cancer patients with either stage I/II patients and healthy controls together, or healthy controls alone, we found a statistically higher frequency of carriers of the A allele of the p21G251A polymorphism among stage III patients (P = 0.0049 and P = 0.0057, respectively, Fisher's exact test).

Table 1.

Observations of p21C93A

No. patients analyzedNo. observations
Frequency of individuals carrying the A allele
Homozygous (CC)Homozygous (AA)Heterozygous (CA)
Healthy controls 1,006 892 113 0.113 
Stage I/II breast cancers 326 296 28 0.092 
Locally advanced breast cancers 223 200 20 0.094 
No. patients analyzedNo. observations
Frequency of individuals carrying the A allele
Homozygous (CC)Homozygous (AA)Heterozygous (CA)
Healthy controls 1,006 892 113 0.113 
Stage I/II breast cancers 326 296 28 0.092 
Locally advanced breast cancers 223 200 20 0.094 
Table 2.

Observations of p21G251A

No. patients analyzedNo. observations
Frequency of individuals carrying the A allele
Homozygous (GG)Homozygous (AA)Heterozygous (GA)
Healthy controls 1,006 996 13 0.014 
Stage I/II breast cancers 326 321 0.015 
Locally advanced breast cancers 223 213 10 0.045 
No. patients analyzedNo. observations
Frequency of individuals carrying the A allele
Homozygous (GG)Homozygous (AA)Heterozygous (GA)
Healthy controls 1,006 996 13 0.014 
Stage I/II breast cancers 326 321 0.015 
Locally advanced breast cancers 223 213 10 0.045 
Table 3.

Observations of p21BT35C

No. patients analyzedNo. observations
Frequency of individuals carrying the C allele
Homozygous (TT)Homozygous (CC)Heterozygous (TC)
Healthy controls 1,017 917 97 0.098 
Stage I/II breast cancers 333 299 34 0.102 
Locally advanced breast cancers 219 191 28 0.128 
No. patients analyzedNo. observations
Frequency of individuals carrying the C allele
Homozygous (TT)Homozygous (CC)Heterozygous (TC)
Healthy controls 1,017 917 97 0.098 
Stage I/II breast cancers 333 299 34 0.102 
Locally advanced breast cancers 219 191 28 0.128 

The haplotype structure, considering all the five coding polymorphism in the p21, gene was determined. Overall, six different haplotypes were observed for the combined cases and controls data set. PHASE calculated six different haplotypes for the case group and four for the controls. Two infrequent haplotypes were found in the cases only, whereas no control-specific haplotypes were observed. A case-control permutation test did not show a significantly different distribution of haplotypes in cases compared with controls (P = 0.47).

In this article, we report the presence of previously identified as well as novel polymorphisms of p21/p21B among 474 Norwegian breast cancer patients and 1,017 healthy individuals.

The polymorphisms are rarely found together (with the exception of the linkage between p21C93A and p21C515T), and when two are observed together, they are always heterozygous. This indicates that no haplotype containing more than one of the polymorphism p21T74G, p21G251A, or p21BT35C; or both p21C93A and p21C515T are present among patients and controls analyzed. This constitutes five of the six PHASE reported haplotypes (including the haplotype wild type in all positions). The final haplotype predicted by PHASE is disregarded, as it results from estimation of status for one of the polymorphisms, and thus may not be counted among our observed haplotypes. We here refer to the haplotypes as wild type or by the polymorphism defining the haplotype, as no haplotype carries two or more of the p21T74G, p21G251A, p21BT35C, or p21C93A polymorphisms.

Expanding on our previous results (15), we observe the novel p21 polymorphism, p21BT35C, as well as the previously defined p21C93A, to occur with a similar frequency among breast cancer patients and non–breast cancer individuals. Although the p21G251A polymorphism was observed more frequently among breast cancer patients compared with controls, this difference did not reach statistical significance. In contrast, we found this polymorphism to be significantly associated with stage III (locally advanced) breast cancer. Although locally advanced breast cancer seems to present cDNA microarray profiles resembling the profile of breast tumors in general (36, 37), recent studies suggest a potential difference in biology between tumors detected by mammographic screening and so-called “interval cancers” (38). “Interval cancers” are thought to include subgroups of fast-growing tumors. Many of our patients revealing a large tumor at time of diagnosis had recognized symptoms for short time intervals only (26, 27), and the possibility exists that some locally advanced cancers may have fast-growing tumors due to particular genetic disturbances. Considering p21 to be critical to growth arrest, any disturbance in the function of this gene is expected to affect cellular growth rate, explaining an association with locally advanced tumor development. Ongoing studies in our laboratory are exploring the function of the different alleles of this polymorphism in the p21 protein's ability to inhibit cellular growth.

When observing a novel polymorphism of p21 the possibility exists that this is a geographically limited variant. However, linking p21G251A to the subset of locally advanced breast cancers presents another plausible explanation as to why the variant has not been reported previously. Ours was the first study addressing the full p21 status of locally advanced breast cancers, whereas many earlier screenings only evaluated status of p21 polymorphisms already known. As the frequencies of the polymorphism in stage I/II tumors and among healthy individuals are low, a geographic limitation of the variant is not necessary to explain that others did not observe the p21G251A polymorphism.

As the p21G251A polymorphism is rare, homozygotes are expected to be very low in number. We observe only one homozygote in all our samples, and we are thus not able to evaluate the significance of homozygosity versus heterozygosity in relation to the clinical implications of this polymorphism.

Although the p21BT35C polymorphism has previously been analyzed in only one small case control study and not investigated in larger cohort studies, conflicting evidence has linked the p21C93A polymorphism to cancer risk as well as prognosis (1625, 39). We found both polymorphisms to be evenly distributed among our groups of individuals, showing no association to any clinical aspect of our patients.

Currently, there is much focus on “low-risk” genes with respect to breast cancer risk (40). If the association of the p21G251A polymorphism to locally advanced, but not other forms of breast cancer, is confirmed, it suggests we may start looking not at elevated risk of breast cancer in general, but rather whether particular tumor forms may reveal a family association. Notably, whereas tumors in BRCA2 carriers in general reveal a biology resembling the biology of tumors in patients with no germ line mutation, the opposite is the case for tumors among BRCA1 carriers (4143).

In conclusion, we find the novel p21G251A polymorphism to be associated not to breast cancers in general but to a subgroup of the disease characterized by aggressive growth.

Grant support: Norwegian Cancer Society and the Haukeland Hospital Program for Gene Therapy.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1
Brugarolas J, Chandrasekaran C, Gordon JI, Beach D, Jacks T, Hannon GJ. Radiation-induced cell-cycle arrest compromised by p21 deficiency.
Nature
1995
;
377
:
552
–7.
2
Deng CX, Zhang PM, Harper JW, Elledge SJ, Leder P. Mice lacking p21(C/P1/Waf1) undergo normal development, but are defective in G1 checkpoint control.
Cell
1995
;
82
:
675
–84.
3
Dulic V, Kaufmann WK, Wilson SJ, et al. P53-dependent inhibition of cyclin-dependent kinase-activities in human fibroblasts during radiation-induced G1 arrest.
Cell
1994
;
76
:
1013
–23.
4
Luo Y, Hurwitz J, Massague J. Cell-cycle inhibition by independent Cdk and Pcna binding domains in P21(Cip1).
Nature
1995
;
375
:
159
–61.
5
Cheng MG, Olivier P, Diehl JA, et al. The p21(Cip1) and p27(Kip1) CDK ‘inhibitors’ are essential activators of cyclin D-dependent kinases in murine fibroblasts.
EMBO J
1999
;
18
:
1571
–83.
6
Dulic V, Stein GH, Far DF, Reed SI. Nuclear accumulation of p21(Cip1) the onset of mitosis: a role at the G(2)/M-phase transition.
Mol Cell Biol
1998
;
18
:
546
–57.
7
Gu Y, Turck CW, Morgan DO. Inhibition of Cdk2 activity in-vivo by an associated 20k regulatory subunit.
Nature
1993
;
366
:
707
–10.
8
Harper JW, Adami GR, Wei N, Keyomarsi K, Elledge SJ. The p21 Cdk-interacting protein Cip1 is a potent inhibitor of G1 cyclin-dependent kinases.
Cell
1993
;
75
:
805
–16.
9
Medema RH, Klompmaker R, Smits VAJ, Rijksen G. p21(waf1) can block cells at two points in the cell cycle, but does not interfere with processive DNA-replication or stress-activated kinases.
Oncogene
1998
;
16
:
431
–41.
10
Niculescu AB, Chen XB, Smeets M, Hengst L, Prives C, Reed SI. Effects of p21(Cip1/Waf1) at both the G(1)/S and the G(2)/M cell cycle transitions: pRb is a critical determinant in blocking DNA replication and in preventing endoreduplication.
Mol Cell Biol
1998
;
18
:
629
–43.
11
Sherr CJ, Roberts JM. Inhibitors of mammalian G(1) cyclin-dependent kinases.
Genes Dev
1995
;
9
:
1149
–63.
12
Sherr CJ, Roberts JM. CDK inhibitors: positive and negative regulators of G1-phase progression.
Genes Dev
1999
;
13
:
1501
–12.
13
Xiong Y, Hannon GJ, Zhang H, Casso D, Kobayashi R, Beach D. P21 is a universal inhibitor of cyclin kinases.
Nature
1993
;
366
:
701
–4.
14
Nozell S, Chen X. p21B, a variant of p21(Waf1/Cip1), is induced by the p53 family.
Oncogene
2002
;
21
:
1285
–94.
15
Staalesen V, Leirvaag B, Lillehaug JR, Lonning PE. Genetic and epigenetic changes in p21 and p21B do not correlate with resistance to doxorubicin or mitomycin and 5-fluorouracil in locally advanced breast cancer.
Clin Cancer Res
2004
;
10
:
3438
–43.
16
Chen WC, Wu HC, Hsu CD, Chen HY, Tsai FJ. p21 gene codon 31 polymorphism is associated with bladder cancer.
Urol Oncol
2002
;
7
:
63
–6.
17
Facher EA, Becich MJ, Deka A, Law JC. Association between human cancer and two polymorphisms occurring together in the p21Waf1/Cip1 cyclin-dependent kinase inhibitor gene.
Cancer
1997
;
79
:
2424
–9.
18
Harima Y, Sawada S, Nagata K, Sougawa M, Ostapenko V, Ohnishi T. Polymorphism of the WAF1 gene is related to susceptibility to cervical cancer in Japanese women.
Int J Mol Med
2001
;
7
:
261
–4.
19
Roh J, Kim M, Kim J, et al. Polymorphisms in codon 31 of p21 and cervical cancer susceptibility in Korean women.
Cancer Lett
2001
;
165
:
59
–62.
20
Sjalander A, Birgander R, Rannug A, Alexandrie AK, Tornling G, Beckman G. Association between the p21 codon 31 A1 (arg) allele and lung cancer.
Hum Hered
1996
;
46
:
221
–5.
21
Hsieh YY, Tsai FJ, Chang CC, et al. p21 gene codon 31 arginine/serine polymorphism: non-association with endometriosis.
J Clin Lab Anal
2001
;
15
:
184
–7.
22
Ressiniotis T, Griffiths PG, Keers SM, Chinnery PF, Birch M. A polymorphism at codon 31 of gene p21 is not associated with primary open angle glaucoma in Caucasians.
BMC Ophthalmol
2005
;
5
:
5
.
23
Shih CM, Lin PT, Wang HC, Huang WC, Wang YC. Lack of evidence of association of p21WAF1/CIP1 polymorphism with lung cancer susceptibility and prognosis in Taiwan.
Jpn J Cancer Res
2000
;
91
:
9
–15.
24
Su L, Liu G, Zhou W, et al. No association between the p21 codon 31 serine-arginine polymorphism and lung cancer risk.
Cancer Epidemiol Biomarkers Prev
2003
;
12
:
174
–5.
25
Tsai MH, Chen WC, Tsai FJ. Correlation of p21 gene codon 31 polymorphism and TNF-alpha gene polymorphism with nasopharyngeal carcinoma.
J Clin Lab Anal
2002
;
16
:
146
–50.
26
Geisler S, Lonning PE, Aas T, et al. Influence of TP53 gene alterations and c-erbB-2 expression on the response to treatment with doxorubicin in locally advanced breast cancer.
Cancer Res
2001
;
61
:
2505
–12.
27
Geisler S, Børresen-Dale A-L, Johnsen H, et al. TP53 gene mutations predict the response to neoadjuvant treatment with FUMI in locally advanced breast cancer.
Clin Cancer Res
2003
;
9
:
5582
–8.
28
Gram IT, Bremnes Y, Ursin G, Maskarinec G, Bjurstam N, Lund E. Percentage density, Wolfe's and Tabar's mammographic patterns: agreement and association with risk factors for breast cancer.
Breast Cancer Res
2005
;
7
:
R854
–61.
29
Helle SI, Ekse D, Holly JMP, Lonning PE. The IGF-system in healthy pre- and postmenopausal women: relations to demographic variables and sex-steroids.
J Steroid Biochem Mol Biol
2002
;
81
:
95
–102.
30
Stephens M, Donnelly P. A comparison of Bayesian methods for haplotype reconstruction from population genotype data.
Am J Hum Genet
2003
;
73
:
1162
–9.
31
Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data.
Am J Hum Genet
2001
;
68
:
978
–89.
32
Rozas J, Sanchez-DelBarrio JC, Messeguer X, Rozas R. DnaSP, DNA polymorphism analyses by the coalescent and other methods.
Bioinformatics
2003
;
19
:
2496
–7.
33
Abecasis GR, Cookson WO. GOLD: graphical overview of linkage disequilibrium.
Bioinformatics
2000
;
16
:
182
–3.
34
Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps.
Bioinformatics
2005
;
21
:
263
–5.
35
Koopmann J, Maintz D, Schild S, et al. Multiple polymorphisms, but no mutations, in the WAF1/CIP1 gene in human brain tumours.
Br J Cancer
1995
;
72
:
1230
–3.
36
Sorlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications.
Proc Natl Acad Sci U S A
2001
;
98
:
10869
–74.
37
Sorlie T, Tibshirani R, Parker J, et al. Repeated observation of breast tumor subtypes in independent gene expression data sets.
Proc Natl Acad Sci U S A
2003
;
100
:
8418
–23.
38
Collett K, Stefansson IM, Eide J, et al. A Basal epithelial phenotype is more frequent in interval breast cancers compared with screen detected tumors.
Cancer Epidemiol Biomarkers Prev
2005
;
14
:
1108
–12.
39
Powell BL, van Staveren IL, Roosken P, Grieu F, Berns EM, Iacopetta B. Associations between common polymorphisms in TP53 and p21WAF1/Cip1 and phenotypic features of breast cancer.
Carcinogenesis
2002
;
23
:
311
–5.
40
Antoniou AC, Pharoah PDP, McMullan G, Day NE, Ponder BAJ, Easton D. Evidence for further breast cancer susceptibility genes in addition to BRCA1 and BRCA2 in a population-based study.
Genet Epidemiol
2001
;
21
:
1
–18.
41
Goffin JR, Chappuis PO, Begin LR, et al. Impact of germline BRCA1 mutations and overexpression of p53 on prognosis and response to treatment following breast carcinoma: 10-year follow-up data.
Cancer
2003
;
97
:
527
–36.
42
Moller P, Borg A, Evans DG, et al. Survival in prospectively ascertained familial breast cancer: analysis of a series stratified by tumour characteristics, BRCA mutations and oophorectomy.
Int J Cancer
2002
;
101
:
555
–9.
43
Rebbeck TR, Lynch HT, Neuhausen SL, et al. Prophylactic oophorectomy in carriers of BRCA1 or BRCA2 mutations.
N Engl J Med
2002
;
346
:
1616
–22.