Breast Cancer Research and Treatment

, Volume 115, Issue 2, pp 307–313

No evidence that CDKN1B (p27) polymorphisms modify breast cancer risk in BRCA1 and BRCA2 mutation carriers

Authors

    • Queensland Institute of Medical Research
    • Genetics and Population Health DivisionQueensland Institute of Medical Research
  • Andrew J. Deans
    • Peter MacCallum Cancer Centre
  • David Duffy
    • Queensland Institute of Medical Research
  • David E. Goldgar
    • Department of DermatologyUniversity of Utah
  • Xiaoqing Chen
    • Queensland Institute of Medical Research
  • Jonathan Beesley
    • Queensland Institute of Medical Research
  • kConFaB
    • The Kathleen Cunningham Foundation Consortium for Research into Familial Breast CancerPeter MacCallum Cancer Centre
  • Douglas F. Easton
    • Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary CareUniversity of Cambridge
  • Antonis C. Antoniou
    • Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary CareUniversity of Cambridge
  • Susan Peock
    • Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary CareUniversity of Cambridge
  • Margaret Cook
    • Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary CareUniversity of Cambridge
  • EMBRACE Study Collaborators
    • Cancer Research UK Genetic Epidemiology Unit, Department of Public Health and Primary CareUniversity of Cambridge
  • Katherine L. Nathanson
    • Department of Medicine and Abramson Cancer CenterUniversity of Pennsylvania School of Medicine
  • Susan M. Domchek
    • Department of Medicine and Abramson Cancer CenterUniversity of Pennsylvania School of Medicine
  • Grant A. MacArthur
    • Peter MacCallum Cancer Centre
    • Department of MedicineSt Vincents Hopsital
  • Georgia Chenevix-Trench
    • Queensland Institute of Medical Research
Preclinical Study

DOI: 10.1007/s10549-008-0083-5

Cite this article as:
Spurdle, A.B., Deans, A.J., Duffy, D. et al. Breast Cancer Res Treat (2009) 115: 307. doi:10.1007/s10549-008-0083-5

Abstract

The p27kip1 protein functions as an inhibitor of cyclin dependent kinase-2, and shows loss of expression in a large percentage of BRCA1 and BRCA2 breast cancer cases. We investigated the association between CDKN1B gene variants and breast cancer risk in 2359 female BRCA1 and BRCA2 mutation carriers from Australia, the UK, and the USA. Samples were genotyped for five single nucleotide polymorphisms, including coding variant rs2066827 (V109G). Cox regression provided no convincing evidence that any of the polymorphisms modified disease risk for BRCA1 or BRCA2 carriers, either alone or as a haplotype. Borderline associations were observed for homozygote carriers of the rs3759216 rare allele, but were opposite in effect for BRCA1 and BRCA2 carriers (adjusted hazard ratio (HR) 0.72 (95% CI = 0.53–0.99; P = 0.04 for BRCA1, HR 1.47 (95% CI = 0.99–2.18; P = 0.06 for BRCA2). The 95% confidence intervals for per allele risk estimates excluded a twofold risk, indicating that common CDKN1B polymorphisms do not markedly modify breast cancer risk among BRCA1 or BRCA2 carriers.

Keywords

P27BRCA1BRCA2Modifier

Introduction

Germline mutations in BRCA1 or BRCA2 genes confer a high risk of predisposition to breast cancer. However estimated risk has been shown to be higher in studies based on multiple case families than in population-based studies [1], and genetic modifiers may explain some of this difference. Of particular interest in this regard are those genes (and their protein products) which are shown to be dysregulated in the development of BRCA-cancers, and contribute to the unique properties of familial breast cancer (reviewed in [2]).

One such example is p27Kip1, which shows loss of expression in a large percentage of BRCA1/2 breast cancer cases [35]. Immunodetection of p27 has been used as a prognostic factor in a variety of cancer types, with low levels being correlated with reduced median survival time (reviewed by [6]). Furthermore, characterization of p27-deficient breast cancer cell lines which promoted progression in mouse tumour migration experiments have provided evidence that p27 plays an essential role in the restriction of breast cancer progression [7]. In one study of cancers in BRCA1 and BRCA2 carriers, p27 was found to be low in 82% of cases, and was an independent prognostic marker for survival after treatment [3]. In contrast, p27 protein accumulates in normal mammary tissue in BRCA1-loss of function mice, and crossing such mice onto a p27-deficient background leads to accelerated proliferation and increased ductal content [8]. These findings led to the hypothesis that the development of BRCA-associated cancers is repressed by p27 and that loss of p27 expression may be selected for as an early event in cancer development.

The CDKN1B gene product p27 functions as an inhibitor of cyclin dependent kinases, particularly CDK2, and regulates cell cycle progression, apoptosis and differentiation in the breast [9]. Consistent with its role as a tumor suppressor, P27-null mice develop multi-organ hyperplasia and increased susceptibility to sporadic and carcinogen induced cancers [10, 11]. In humans, several single nucleotide polymorphisms (SNPs) have been identified in the CDKN1B gene, which have variously been associated with cancer risk or outcome. The −79C > T variant has been associated with familial prostate cancer risk [12]. The −838C > A variant was reported to be associated with a higher risk of myocardial infarction [13], and was demonstrated to have lower promoter activity, potentially resulting in reduced p27 protein levels. A silent G > A variant at codon 55 was found at increased frequency in hepatocellular cases compared to controls, although the difference was only marginally significant [14]. There have been numerous studies of the +326G (V109G) variant. The common T allele was non-significantly overtransmitted in a family-based associated test in prostate cancer families [12], while the TT genotype has been associated with an increased risk of advanced prostate cancer [15], and also with an increased risk of esophageal squamous cell carcinoma and gastric cardiac adenocarcinoma [16]. In contrast, the GG genotype at the same locus was associated with increased risk of squamous cell carcinoma of the head and neck among alcohol users, and both increased risk and worse outcome in oral squamous cell carcinoma [17]. The G allele was associated with high grade breast tumours [18], lymph node metastasis in breast cancer [19], and greater nodal involvement and shorter disease-free survival in node-negative breast cancer [20]. Recently, a large study of 1,115 breast cancer cases and 710 controls showed that the G allele was associated with a modest protective effect in adjusted analyses [21]. Given the reported associations between various CDKN1B SNPs and cancer risk and/or outcome, the demonstrated functional consequences for at least one of these SNPs, and the evidence that decreased p27 promotes proliferation of BRCA1-deficient cells, we sought to investigate the contribution of CDKN1B SNPs to breast cancer risk in BRCA1 and BRCA2 mutation carriers.

Subjects

The characteristics of the study samples are shown in Table 1. A total of 2359 living female carriers of pathogenic BRCA1 or BRCA2 mutations were identified from three sources, as part of clinic-based research studies described in detail elsewhere [22, 23]. Briefly, the Epidemiological Study of BRCA1 and BRCA2 Mutation Carriers (EMBRACE) recruits participants from among women and men referred for genetic testing at clinical genetics centres in the UK and Eire. The Kathleen Cuningham Consortium for Research into Familial Breast Cancer (kConFaB) recruits participants from families attending high-risk family cancer clinics across Australia. The UPENN sample set was recruited from high risk breast cancer clinics within the University of Pennsylvaian Health System. Mutation classification was as described previously [23]. Missense alterations were included only if considered pathogenic by BIC, and/or classified according to multifactorial likelihood modelling approaches [24, 25]. Ethical approvals for recruitment and genotyping were obtained from the institutional review boards or ethics committees at all sites. Written informed consent was obtained from each participant.
Table 1

Characteristics of study subjects

 

BRCA1

BRCA2

BRCA1&2

n

% of total

n

% of total

n

Group total

EMBRACE

663

(47.8)

471

(48.7)

3

1,137

KConFaB

438

(31.6)

363

(37.5)

0

801

UPENN

287

(20.7)

133

(13.8)

1

421

Total

1,388

 

967

 

4

2,359

Affected with breast cancera

749

(54.0)

524

(54.2)

2

1,275

Affected with ovarian cancera

119

(8.6)

39

(4.0)

0

158

Unaffectedb

520

(37.5)

404

(41.8)

2

926

aRefers to first cancer for individuals reporting both breast and ovarian cancer. Two individuals with breast cancer were censored at prior mastectomy. Individuals with ovarian cancer were censored as unaffected at age of diagnosis for analysis, unless otherwise specified

bAbout 32 individuals were censored for prior bilateral mastectomy (avg. 4.5 years prior to interview, range 1.4–19.0 years)

Molecular methods

Sequenom massarray genotyping was performed using methodology as described previously [26], initially for six CDKN1B SNPs. Criteria for selection were as follows: validated 5′ upstream SNPs with minor allele frequency > 0.05 (rs375216 and rs375217); a promoter SNP −838C > A reported to be associated with increased risk of myocardial infarction, and have functional consequences [13]; and three haplotype tagging SNPs tagging all known HapMap SNPs with minor allele frequency >0.05, at R2 > 0.8 by pairwise tagging—located within the 5′UTR (rs34430), exon 1 (the V109G coding SNP rs2066827), and in intron 2 (rs34329) (http://www.ncbi.nlm.nih.gov/projects/SNP/). Details of PCR primer sequences, extension primers, and assay conditions are available on request. Initial haplotype analysis on a subset of 746 carriers indicated that the promoter SNP −838C > A was in complete linkage disequilibrium with rs375216, and so was not genotyped in additional samples.

After applying standard quality control measures [23], genotypes were missing for a maximum of 28 carriers (1%) for any individual SNP. For haplotype risk analysis, genotypes were inferred for individuals missing data at one or two loci only using the EM approach implemented in the R haplo.stats package (see section ‘Statistical methods’), so that a total of 21 individuals were excluded from the analysis due to missing genotype information.

Statistical methods

Individuals with a first primary invasive breast cancer diagnosis were considered to be affected (1,275 carriers), while individuals with no reported breast or ovarian cancer were censored at age of interview, or age of prior prophylactic mastectomy (926 carriers). Individuals with a first primary ovarian cancer diagnosis were censored as unaffected at age at onset of ovarian cancer (158 carriers).

Analyses of association between genotype and disease risk were performed using Cox regression with time to breast cancer onset as the endpoint. Potential confounders considered in the analysis included year of birth (categorised into subgroups 1910–1939, 1940–1949, 1950–1959, and 1960+), ethnicity (Caucasian, Ashkenazi Jewish, other), bilateral prophylactic oophorectomy, parity (categorized as 0, ≥1 live births). Reported ethnicity was Caucasian for 85% of subjects, Ashkenazi Jewish for 7% and “other” for 8%. Hazard ratios (HR) and 95% confidence intervals (CIs) were estimated separately for BRCA1 and BRCA2 carriers, with year of birth, ethnicity, and study group entered as covariates in the survival analysis. Confidence limits for the rate ratio were calculated using a robust variance approach to allow for the dependence among individuals in the same family [27]. To address the problem of non-random sampling of mutation carriers with respect to the disease phenotype, analyses used the weighted Cox regression approach [28], where individuals were weighted such that observed breast cancer incidence rates in the study sample are consistent with established breast cancer risk estimates for BRCA1 and BRCA2 mutation carriers [1]. Additional analyses further adjusted for oophorectomy and parity as time-dependent covariates (never/ever). Oophorectomy prior to age at interview or diagnosis of breast cancer was reported by 219 carriers—135 BRCA1 carriers (31 with primary breast cancer) and 84 BRCA2 carriers (31 with primary breast cancer).

For haplotype analysis, Haploview V 3.2 was used to assess correlation between SNPs on a subset of data from 746 kConFab subjects. Four SNPs (rs3759216, rs3759217, the promoter −838C > A SNP, and rs34330) formed a linkage disequilibrium block, but high correlation (R2 = 0.93) was found only between the −838C > A and rs3759216 SNPs. Further genotyping thus excluded the −838C > A SNP. Haplotypes were imputed using the EM algorithm implemented in the R haplo.stats package [29], and the highest probability haplotype pair was selected for each individual, resulting in 18 distinct imputed haplotypes. Haplotypes with a frequency of less than 5% were pooled, and per-haplotype hazard ratios were estimated for each haplotype relative to the most common haplotype.

R version 2.2.1 was used for all statistical analyses.

Results

The estimated hazard ratios associated with the CDKN1B SNPs are shown in Table 2. There was no convincing evidence for an association between any of the CDKN1B SNPs assessed and breast cancer risk. The rs3759216 rare allele was associated with decreased risk in BRCA1 mutation carriers for homozygote carriers only (homozygote RR 0.72, 95% CI 0.53–0.97, P = 0.04), with marginal evidence that risk decreased with each additional variant allele (Ptrend = 0.05). In contrast, the same SNP was associated with a borderline significant increased risk in BRCA2 carriers with two copies of the rare allele (homozygote RR 1.47, 95% CI 0.99–2.18, P = 0.06). In addition, the rs3759217 SNP was associated with a borderline significant increased risk in BRCA2 carriers only, for individuals who were homozygote for the rare variant (homozygote RR 2.21, 95% CI 0.94–5.21, P = 0.07). The only other result of similar marginal significance was for rs34329 in BRCA2 carriers, and is very likely due to chance, with an estimate of decreased risk in heterozygotes and increased risk in homozygotes. There was no evidence for heterogeneity in the hazard ratios between studies for any SNP (P ≥ 0.1). The overall findings were not markedly different when analysis was carried out including all genotypes in a multivariate model. The only risk estimates that approached significance in a multivariate model were for rs3759216 among BRCA1 and BRCA2 carriers, and were little different to those presented in Table 2 (BRCA1 RR for homozygotes 0.68, CI 0.45–1.03, P = 0.07, BRCA2 RR for homozygotes 1.60, CI 0.94–2.73, P = 0.08).
Table 2

Association between CDKN1B SNPs and breast cancer risk in BRCA1 and BRCA2 carriers

CDKN1B SNPa

Genotype

Freq

BRCA1

BRCA2

Adjusted group, ethnicity, year of birth

Adjusted group, ethnicity, year of birth

RR

(95% CI)

P

RR

(95% CI)

P

rs3759216

het

0.47

0.93

(0.76–1.15)

0.5

0.92

(0.69–1.24)

0.6

5′ upstream G > A

hz

0.15

0.72

(0.53–0.99)

0.04

1.47

(0.99–2.18)

0.06

 

per allele

 

0.86

(0.75–1.00)

0.05

1.17

(0.96–1.44)

0.1

rs3759217

het

0.21

1.03

(0.82–1.30)

0.8

0.82

(0.60–1.12)

0.2

5′ upstream C > T

hz

0.01

1.04

(0.55–1.96)

0.9

2.21

(0.94–5.21)

0.07

 

per allele

 

1.03

(0.85–1.25)

0.8

0.93

(0.70–1.24)

0.6

rs34330

het

0.40

0.99

(0.81–1.21)

0.9

0.96

(0.72–1.29)

0.8

5′UTR C > T tag SNP

hz

0.07

1.36

(0.93–2.00)

0.1

0.81

(0.47–1.39)

0.5

 

per allele

 

1.08

(0.92–1.27)

0.3

0.93

(0.75–1.14)

0.5

rs2066827

het

0.36

0.99

(0.80–1.22)

0.9

0.95

(0.73–1.25)

0.7

+326 exon 1 T > G V109G

hz

0.06

0.82

(0.53–1.27)

0.4

1.20

(0.68–2.13)

0.5

(also tag SNP)

per allele

 

0.95

(0.81–1.12)

0.5

1.01

(0.81–1.26)

0.9

rs34329

het

0.47

1.08

(0.88–1.34)

0.5

0.78

(0.59–1.03)

0.08

Intron 2 C > G tag SNP

hz

0.09

1.14

(0.81–1.59)

0.5

1.32

(0.87–1.99)

0.2

 

per allele

 

1.07

(0.92–1.25)

0.4

1.00

(0.81–1.23)

1.0

aAllele substitutions are presented as the alteration from common allele > rare allele

There was virtually no difference between the estimates adjusted for only study group, ethnicity and year of birth (Table 1), and those adjusted also for oophorectomy and parity. For example, among BRCA1 carriers the risk estimates for the rs3759216 SNP were 0.95 (0.77–1.17) for heterozygote carriers, and 0.73 (0.54–1.00) for homozygote carriers of the rare allele. Among BRCA2 carriers, the risk estimates for the rs3759217 SNP were 0.81 (0.60–1.11) for heterozygotes and 2.17 (0.92–5.21) for homozygotes.

A total of 18 distinct haplotypes with non-zero frequencies were identified with only 8 of these occurring at frequency greater than 5% (See Table 3). The most common haplotype Ht3 (frequency 28%) was selected as reference for risk analyses. There was no evidence for significantly altered risk for any single haplotype among BRCA1 or BRCA2 carriers. There was only borderline evidence (P = 0.06) for an association between Ht12 and increased risk among BRCA1 carriers, and since analyses did not incorporate error in haplotype inference, the true P values will be even greater.
Table 3

Association between CDKN1B haplotypes and breast cancer risk in BRCA1 and BRCA2 carriers

Haplotype

rs3758216

rs3759217

rs34330

rs2066827

rs35329

Haplotype frequency

BRCA1 mutation carriers

BRCA2 mutation carriers

P

RRa

(95% CI)

P

RRa

(95% CI)

3

A

C

C

T

C

0.28

Ref

1.00

 

Ref

1.00

 

6

G

C

C

G

C

0.05

0.2

1.13

(0.93–1.37)

1.0

1.00

(0.79–1.26)

8

G

C

C

T

C

0.09

0.6

1.06

(0.87–1.30)

0.3

0.84

(0.63–1.14)

9

G

C

C

T

G

0.05

0.2

1.15

(0.91–1.46)

0.9

1.02

(0.78–1.34)

12

G

C

T

T

C

0.15

0.06

1.17

(0.99–1.37)

0.1

0.85

(0.69–1.05)

13

G

C

T

T

G

0.07

0.9

0.99

(0.78–1.25)

0.3

0.87

(0.70–1.10)

17

G

T

C

T

G

0.11

0.2

1.11

(0.94–1.32)

0.3

0.89

(0.73–1.09)

20

A

C

C

G

C

0.09

0.7

0.96

(0.77–1.21)

0.5

0.93

(0.72–1.19)

Rare Htsb

0.12

0.3

1.11

(0.93–1.32)

0.6

0.95

(0.78–1.16)

aAdjusted study group, ethnicity, year of birth

bTen haplotypes of frequency <0.05 (range 0.0004–0.03) pooled for analysis

Discussion

Our data provide no convincing evidence for association between CDKN1B SNPs and breast cancer risk among BRCA1 and/or BRCA2 carriers, individually or as a haplotype. While there was marginal evidence for altered breast cancer risk associated with rs3759216, the risk estimates were opposite directions for BRCA1 and BRCA2 carriers, with a marginally significant decreased risk in BRCA1 carriers, and a borderline increased risk in BRCA2 carriers. While it is certainly possible that a SNP in the 5-upstream region could influence P27 expression, or be linked to another SNP regulating P27 expression, and might therefore influence breast cancer risk, it is seems improbably that the direction of such an effect would differ between BRCA1 and BRCA2 carriers. It seems likely, therefore, that these observation simply represent chance associations. The role of p27 in BRCA2-related cancers is expected to be similar to that of BRCA1-related cancers, in that low protein expression was observed in 7/7 BRCA2 carriers and 16/20 BRCA1 carriers, and was associated with shorter distant disease-free survival in the pooled sample of carriers [30]. We cannot, however, exclude the possibility that the effect of a SNP may be modulated by the cellular milieu of the tumour, which is likely to differ between BRCA2 carriers and BRCA1 carriers, in that the latter commonly display a basal-like tumour phenotype (reviewed in [2]).

There are some suggestions in the literature to support associations between the rs2066827 V109G amino acid substitution variant and cancer risk. The rare G allele has been associated with a decreased risk of familial prostate cancer [12], a decreased risk of advanced prostate cancer [15], and a decreased risk of esophageal squamous cell carcinoma and gastric cardiac adenocarcinoma [16]. Conversely, the G allele has been associated with an increased risk of squamous cell carcinoma [17], and breast cancer [18] and prognosis in breast cancer [20, 31]. Insert new ref. However, there is no evidence from the current study to support an association of this particular SNP with breast cancer risk, either alone or as a haplotype. It is possible that post-translational regulation of p27 may explain in part the differences in functional levels of p27 [32], including high p27 expression in improved prognosis breast cancer [31].

Conclusion

We found no convincing evidence for association between any of the SNPs in CDKN1B and breast cancer risk in either BRCA1 or BRCA2 carriers. Given the number of tests carried out in this study, the marginally significant associations observed would not be robust to Bonferroni correction. The 95% confidence intervals for the per allele risk estimate exclude a twofold risk associated with any SNP. Equally, haplotype analysis provided no evidence for an association that might suggest an unmeasured risk variant. We conclude that common CDKN1B SNPs do not modify to any marked extent the breast cancer risk among BRCA1 or BRCA2 carriers.

Acknowledgments

kConFab—The Kathleen Cuningham Consortium for Research into Familial Breast Cancer: We wish to thank Heather Thorne, Eveline Niedermayr, all the kConFab research nurses and staff, the heads and staff of the Family Cancer Clinics, and the Clinical Follow Up Study (funded by NHMRC grants 145684, 288704 and 454508) for their contributions to this resource, and the many families who contribute to kConFab. kConFab is supported by grants from the National Breast Cancer Foundation, the National Health and Medical Research Council (NHMRC) and by the Queensland Cancer Fund, the Cancer Councils of New South Wales, Victoria, Tasmania and South Australia, and the Cancer Foundation of Western Australia. EMBRACE—M.C., S·P., and EMBRACE are funded by Cancer Research-UK. The following are EMBRACE collaborating centers. Coordinating Centre, Cambridge: Susan Peock, Margaret Cook, Alexandra Bignell and Debra Frost. North of Scotland Regional Genetics Service, Aberdeen: Neva Haites, Helen Gregory. Northern Ireland Regional Genetics Service, Belfast: Patrick Morrison. West Midlands Regional Clinical Genetics Service, Birmingham: Trevor Cole and Carole McKeown. South West Regional Genetics Service, Bristol: Alan Donaldson. East Anglian Regional Genetics Service, Cambridge: Joan Paterson. Medical Genetics Services for Wales, Cardiff: Alexandra Murray, Mark Rogers and Emma McCann. St James’s Hospital, Dublin and National Centre for Medical Genetics, Dublin: Peter Daly and David Barton. South East of Scotland Regional Genetics Service, Edinburgh: Mary Porteous and Michael Steel. Peninsula Clinical Genetics Service. Exeter: Carole Brewer and Julia Rankin.West of Scotland Regional Genetics Service, Glasgow: Rosemarie Davidson and Victoria Murday. South East Thames Regional Genetics Service, Guys Hospital London: Louise Izatt and Gabriella Pichert. North West Thames Regional Genetics Service, Harrow: Huw Dorkins. Leicestershire Clinical Genetics Service, Leicester: Richard Trembath.Yorkshire Regional Genetics Service, Leeds: Tim Bishop and Carol Chu. Merseyside and Cheshire Clinical Genetics Service, Liverpool: Ian Ellis. Manchester Regional Genetics Service, Manchester: D. Gareth Evans and Fiona Lalloo. North East Thames Regional Genetics Service, NE Thames: Alison Male, James Mackay, and Anne Robinson. Nottingham Centre for Medical Genetics, Nottingham: Carol Gardiner. Northern Clinical Genetics Service, Newcastle: Fiona Douglas and John Burn. Oxford Regional Genetics Service, Oxford: Lucy Side, Lisa Walker and Sarah Durell. Institute of Cancer Research and Royal Marsden NHS Foundation Trust: Rosalind Eeles. North Trent Clinical Genetics Service, Sheffield: Jackie Cook and Oliver Quarrell. South West Thames Regional Genetics Service, London: Shirley Hodgson. Wessex Clinical Genetics Service. Southampton: Diana Eccles and Anneke Lucassen. UPENN—Breast Cancer Research Foundation (KLN), QVC Network and the Fashion Footwear Association of New York, Marjorie B. Cohen Foundation, National Cancer Institute Cancer Genetics Network [HHSN216200744000C] (SMD). The genotyping was supported by an NHMRC Programme grant to GCT. ABS was funded by an NHMRC Career Development Award, AJD is a recipient of a Cancer Council of Victoria Postdoctoral fellowship, DD is an NHMRC Senior Research Fellow. ACA is funded by Cancer Research UK, DFE is a Principal Research Fellow of Cancer Research UK, GAM is the Weary Dunlop Fellow of the Cancer Council of Victoria, and GC-T is an NHMRC Senior Principal Research Fellow.

Copyright information

© Springer Science+Business Media, LLC. 2008