Molecular Biology Reports

, Volume 39, Issue 12, pp 10873–10879

HSP90, HSPA8, HIF-1 alpha and HSP70-2 polymorphisms in breast cancer: a case–control study

Authors

  • Flora Zagouri
    • Department of Clinical Therapeutics, Alexandra HospitalUniversity of Athens
    • 1st Propaedeutic Surgical Department, Hippocrateio HospitalUniversity of Athens
  • Maria Gazouli
    • Department of BiologyUniversity of Athens
  • Alexandra Tsigginou
    • Department of Obstetrics and Gynaecology Alexandra HospitalUniversity of Athens
  • Constantine Dimitrakakis
    • Department of Obstetrics and Gynaecology Alexandra HospitalUniversity of Athens
  • Irene Papaspyrou
    • Department of Clinical Therapeutics, Alexandra HospitalUniversity of Athens
  • Evaggelos Eleutherakis-Papaiakovou
    • Department of Clinical Therapeutics, Alexandra HospitalUniversity of Athens
  • Dimosthenis Chrysikos
    • 1st Propaedeutic Surgical Department, Hippocrateio HospitalUniversity of Athens
  • George Theodoropoulos
    • 1st Propaedeutic Surgical Department, Hippocrateio HospitalUniversity of Athens
  • George C. Zografos
    • 1st Propaedeutic Surgical Department, Hippocrateio HospitalUniversity of Athens
  • Aris Antsaklis
    • Department of Obstetrics and Gynaecology Alexandra HospitalUniversity of Athens
  • Athanassios-Meletios Dimopoulos
    • Department of Clinical Therapeutics, Alexandra HospitalUniversity of Athens
  • Christos A. Papadimitriou
    • Department of Clinical Therapeutics, Alexandra HospitalUniversity of Athens
Article

DOI: 10.1007/s11033-012-1984-2

Cite this article as:
Zagouri, F., Sergentanis, T.N., Gazouli, M. et al. Mol Biol Rep (2012) 39: 10873. doi:10.1007/s11033-012-1984-2
  • 568 Views

Abstract

This case control study aims to investigate the role of HSP90 Gln488His (C > G), HSP70-2 P1/P2, HIF-1 alpha C1772T and HSPA8 intronic 1541–1542delGT polymorphisms as potential risk factors and/or prognostic markers for breast cancer. 113 consecutive incident cases of histologically confirmed ductal breast cancer and 124 healthy cases were recruited. The above mentioned polymorphisms were genotyped; multivariate logistic regression was performed. HSP90 GG (His/His) genotype was associated with elevated breast cancer risk. Similarly, the allele dose–response model pointed to increase in breast cancer risk per G allele. HSP70-2 P1/P2, HSPA8 intronic 1541–1542delGT and HIF-1 alpha polymorphisms were not associated with breast cancer risk, as evidenced by the dose–response allele models. The positive association between HSP90 G allele and breast cancer risk seemed to pertain to both premenopausal and postmenopausal women. With respect to survival analysis, none of the aforementioned polymorphisms was associated with either disease-free survival or overall survival. HSP90α Gln488His polymorphism seems to be a risk factor for breast cancer. On the other hand, our study did not point to excess risk conferred by HSPA8 1541–1542delGT, Hsp70-2 P1/P2 and HIF-1α C1772T.

Keywords

PolymorphismBreast cancerHSP90Case–control

Introduction

Single nucleotide polymorphisms may be associated not only with inter-individual predisposition to breast cancer, but also with phenotypic traits, treatment outcomes with anticancer agents and disease prognosis. Numerous recent meta-analyses have emphasized on the role of polymorphisms in various genes regarding breast cancer, such as such as PON1 [1], AIB1 [2], COMT [3], MDM2 [4], LEPR [5], CXCL12 [6] and any many others.

Hsp90 (heat shock protein 90, also known as HSPC1 according to the most recent nomenclature assigned by the HUGO Gene Nomenclature Committee [7]) is a chaperone protein essential for preserving and regulating the function of various cellular proteins. Hsp90 mediates its actions through the formation of discrete subcomplexes, each containing proteins-cochaperones that assist protein folding and refolding during stress, protein transport and degradation [8, 9]. Hsp90 interacts with a number of protein-cornerstones in breast carcinogenesis; including estrogen receptors (ER), tumor suppressor p53 protein, angiogenesis transcription factor HIF-1alpha, antiapoptotic kinase Akt, Raf-1 MAP kinase and a variety of receptor tyrosine kinases of the erbB family (reviewed in [9]). Hsp90α Gln488His polymorphism has been recently recognized as a molecular variation of the aforementioned gene in Caucasians [10]. The functional implications of Hsp90α Gln488His have been also highlighted, as the substitution may abrogate the Hsp90 function due to reduced dimerization [11, 12]. To our knowledge, the relevance of this in vitro observation has not been examined in the context of breast cancer.

Hsp70-2 is another member of the chaperone family, protecting cell proteins from various stressful stimuli through its binding to denatured or inappropriately folded proteins [13]. HSP70-2 P1/P2 polymorphism status has been associated with risk for nasopharyngeal [14], gastric [15, 16]. No associations were observed for multiple myeloma [17]. Regarding breast cancer, only one study has appeared, pointing to excess risk conferred by the HSP 70-2 P2 allele [18]. Nevertheless, it should be stressed that HSP70-2 P1/P2 polymorphism represents a silent mutation [14, 18].

HSPA8 (alias HSC70) is a member of the HSP70 family located on chromosome 11q23.3, a region deleted in 40 % of sporadic breast and other cancers [19]. It is constitutively expressed under non-stressful conditions and also participated in protein folding as well as differentiation procedures [2022]. HSPA8 intronic 1541–1542delGT polymorphism status has been associated with lung cancer risk [23]; indeed functional relevance of this intronic polymorphism has been supported, as it is located in intron two close to a splice acceptor consensus sequence and may well interfere with HSPA8 protein expression [23]. The 1541–1542delGT polymorphism has been identified within breast cancer tissue [19]; however, to our knowledge, no case–control studies have evaluated this polymorphism in the context of breast cancer.

Hypoxia-inducible factor-1 (HIF-1) is a heterodimeric transcription factor that consists of HIF-1α and HIF-1β subunits [24]. HIF-1α is upregulated by hypoxic stress and, as mentioned above, functionally interacts with Hsp90 protein; specifically, Hsp90 binds to the PAS-domain of HIF-1α, protecting the latter from misfolding and degradation [25]. As a result, various Hsp90 endogenous competitors or exogenous inhibitors lead to HIF-1α degradation [26]. HIF-1α C1772T (Pro582Ser) polymorphism has been demonstrated as a functionally meaningful substitution, leading to increased transcriptional activity of HIF-1α [27]. HIF-1α C1772T status has been associated with susceptibility to various cancer types; nevertheless, regarding breast cancer, a recent meta-analysis based on only three studies pointed to a null association, although the need for further data accumulation was declared [28].

Based on a well designed case–control study of 113 patients with sporadic ductal invasive breast cancer and 124 controls, this study aims to investigate the role of HSPA8, HIF-1 alpha C1772T, HSP70-2 and HSP90 polymorphisms as potential risk factors and/or prognostic markers for breast cancer.

Methods

Subjects

During the period 1997–2006 consecutive incident cases of histologically confirmed ductal breast cancer were recruited; patients were operated either in the Gynecology Department, “Alexandra” Hospital, Medical School, University of Athens, Greece, or in the 1st Department of Propaedeutic Surgery, Hippokratio Hospital, University of Athens, Greece and received chemotherapy in the Oncology Department, “Alexandra” Hospital, Medical School, University of Athens, Greece. Regarding cases, exclusion criteria were: histology other than ductal carcinoma, in situ lesions, family history of breast cancer (so as to obtain a clear picture of sporadic cancer) ethnicity other than Greek, inadequate tissue and not signed informed consent.

Controls were recruited among women with normal results on routine screening mammogram; all controls were unrelated to cases and were derived from the same hospital as cases. Both cases and controls had to be free of any prior diagnosis of other cancer. Cases and controls were frequency matched on age (±2 years); all subjects were Caucasian and resided in the same geographic region in Greece (the greater metropolitan area of Athens, Attica).

All women responded to a pre-coded questionnaire covering sociodemographic parameters, lifestyle habits and parameters related to their reproductive history. Ever smoking was defined as smoking of 100 or more cigarettes during lifetime and alcohol consumption was subclassified as <1 and ≥1 glasses/week. Somatometrics were also measured by the interviewers i.e., height with participants wearing no shoes and body weight, with participants wearing no clothes. BMI (kg/m2) was then appropriately calculated. Regarding cases, additional information on tumor size, nodal status, grade, overall survival and disease-free survival were retrieved from the medical records of patients. For immunohistochemistry regarding Estrogen receptors and Progesterone receptors, the following antibodies were used: 636, Dako for PR, and ID5, Dako for ER.

Written informed consent was obtained by all subjects participating in the study. The study is in accordance with the Helsinki Declaration and has been approved by the local Institutional Review Board.

Genotyping of HSP90 Gln488His, HSP70-2 P1/P2, HSPA8 intronic 1541–1542delGT and HIF-1α C1772T

Regarding breast cancer patients, DNA from paraffin embedded breast normal tissues was isolated with the Nucleopsin Tissue kit (Macherey Nigel, Germany). Concerning controls, DNA from blood was isolated with the same kit. For HSP90 Gln488His, we used allele specific PCR. The primers were: 5′-CCTGTGATATAATAGATATGTTTC-3′, 5′-CCTGTGATATAATAGATATGTTTG-3′ and the common primer 5′-TGGATAACTGTGAGGAGCTAA-3′. The PCR cycling parameters were 94 °C for 4 min, followed by 35 cycles of 94 °C for 30 s, 56 °C for 45 min, 72 °C for 45 min and final cycle at 72 °C for 10 min.

P1/P2 Polymorphism within HSP70-2 gene has been characterized by Milner et al. [29], who identified a polymorphic Pst I site at position 1,267 of the HSP70-2 gene. The position 1,267 of the HSP70-2 gene lies in the coding region. The coding sequence of the HSP70-2 gene was amplified from genomic DNA using sequence specific oligonucleotide primers: the primer: 5′-TCCGAAGGACTGAGCTCTTG-3′ was used in combination with the 30-primer: 5′-CAGCAAAGTCCTTGAGTCCC-3′. Amplification was accomplished by initial incubation at 94C for 5 min followed by 30 cycles of incubation at 94 and 60 °C for 1 min each; 72 °C for 3 min and a final incubation at 72 °C for 10 min.

To assess the polymorphism of the HSP70-2 at position 1,267, the corresponding PCR products were digested with Pst I. The presence of a Pst I site was indicated by the cleavage of the 2,075 bp amplified product to yield fragments of 1,139 and 936 bp. The two allelic forms of HSP70-2, corresponding to the presence or the absence of the Pst I site, are referred to as HSP70-2 P1 and HSP70-2 P2, respectively.

The fragment of HSPA8 intron two harboring the 1541–1542delGT polymorphism was amplified using the following primers: (sense) CAG GCT GAA ATC TGG ATA ACG, (antisense) AAGACA CCT CCT CTG GAT AGA AGC TTT TCG TCT. The antisense primer creates the control XmnI restriction site by G to C substitution. The temperature profile of amplification was: initial denaturation and activation of enzyme 95 °C for 10 min, then 35 cycles of 30 s at 94 °C, 30 s at 58 °C, and 30 s at 72 °C, followed by the final extension in 72 °C for 7 min. The PCR product was subsequently digested, with two U of XmnI restriction enzyme (New England Biolabs, Beverly, MA). The 1541–1542delGT polymorphism (minor allele) created the XmnI restriction site, which allowed to cut the PCR product into three fragments: 24 bp (control site), 143 bp, and 155 bp. The PCR product of wild-type allele (without GT deletion) was cut at the control site only and two digestion products were formed (24 and 300 bp).

HIF-1α C1772T polymorphism was analyzed using polymerase chain reaction-restriction fragment length polymorphism (PCR–RFLP) method and the primer sequences were as described by Ollerenshaw et al. [30]. The primer sequences used to amplify the polymorphic site of C1772T (346 bp) were as follows: forward 5′-AAGGTGTGGCCATTGTAAAAACTC-3′, reverse 5′-GCACTAGTAGTTTCTTTATGTATG-3′. The PCR cycling parameters for C1772T were 94 °C for 4 min followed by 35 cycles of 94 °C for 30 s, 57 °C for 1.5 min, 72 °C for 1.5 min, and final cycle at 72 °C for 7 min. For C1772T amplified PCR products were digested with HphI (New England BioLabs, Beverly, MA, USA) at 37 °C. The digested products were separated on 2.5 % agarose gel. The restriction digest for C1772T reveals 228 and 118 bp fragments in the presence of C allele, and the product is uncut in the presence of T allele (346 bp).

Statistical analysis

Three separate models were constructed to assess the association between the examined polymorphisms and breast cancer risk: heterozygous vs. wild type (wild type was referred to as the most frequent homozygous genotype), homozygous versus wild type and dose–response allele model (0: wild type, 1: heterozygous, 2: homozygous carriers). Univariate (unadjusted, crude) and multivariate odds ratios (ORs) with 95 % CI were derived from unconditional logistic regression. The multivariate ORs were adjusted for age, smoking, alcohol, BMI, menopausal status age at menarche and education. In addition, subanalyses for premenopausal and postmenopausal women were conducted.

Moreover, allele frequencies in controls were examined so as to detect any significant deviation from the Hardy–Weinberg equilibrium (HWE), given that the deviation may denote bias [31]. For the assessment of the deviation from HWE, the appropriate goodness-of-fit Chi-square test was performed [31, 32].

Regarding the association of polymorphisms with overall survival and disease-free survival, univariate Cox regression was performed for all three aforementioned models. Kaplan–Meier survival curves were also estimated for the graphical representation of results. 30 November 2011 was the censoring date.

Statistical analysis was performed using STATA 11.1 statistical software (Stata Corporation, College Station, TX, USA).

Results

Table 1 presents descriptive statistics regarding demographic features, lifestyle habits, anthropometric and reproductive parameters, in cases and controls. Cases presented with younger age at menarche (p = 0.047), consumed greater amounts of alcohol (p = 0.041) and tended to be of higher education (p = 0.078).
Table 1

Distribution of the 113 ductal breast cancer cases and the 124 age-matched controls by demographic, lifestyle habits, anthropometric and reproductive variables

Variable

Cases

Controls

p value

Continuous variables

Mean (SD)

Mean (SD)

Age (years)

55.1 (11.1)

55.7 (10.7)

Matched variable

BMI (kg/m2)

26.8 (4.0)

26.4 (3.9)

0.456MWW

Age at menarche (years)

13.1 (1.7)

13.5 (1.6)

0.047MWW

Categorical and ordinal variables

N (%)

N (%)

 

Education

 Uneducated/primary

10 (8.8)

17 (13.7)

0.078CT

 Secondary

17 (15.0)

26 (21.0)

 High School

54 (47.8)

53 (42.7)

 College/university

32 (28.3)

28 (22.6)

Menopausal status

 Premenopausal

37 (32.7)

32 (25.8)

0.240C

 Postmenopausal

76 (67.3)

92 (74.2)

Ever smoking

 Yes

34 (30.1)

32 (25.8)

0.463C

 No

79 (69.9)

92 (74.2)

Alcohol consumption

 <1 glasses/week

75 (66.4)

97 (78.2)

0.041C

 ≥1 glasses/week

38 (33.6)

27 (21.8)

Tumor size

 T1

44 (38.9)

  

 T2

57 (50.4)

 

 T3

10 (8.9)

 

 T4

2 (1.8)

 

Nodal status

 N0

16 (14.1)

  

 N1

69 (61.1)

 

 N2

16 (14.1)

 

 N3

12 (10.6)

 

Grade

 G1

4 (3.5)

  

 G2

50 (44.3)

 

 G3

59 (52.2)

 

Estrogen receptor status

 Positive

79 (69.9)

  

 Negative

34 (30.1)

 

Progesterone receptor status

 Positive

61 (54.0)

  

 Negative

52 (46.0)

 

MWW Mann–Whitney–Wilcoxon test for independent samples, CT Chi-square for trend, C Chi-square

Table 2 presents genotype frequencies, unadjusted and adjusted and ORs for the examined polymorphisms. HSP90 GG (His/His) genotype was associated with elevated breast cancer risk (OR = 4.78, 95 % CI: 2.32–9.81 at the adjusted model). Similarly, the allele dose–response model pointed to increase in breast cancer risk per G allele (OR = 2.09, 95 % CI: 1.49–2.94). HSP70-2 P1/P2 and HIF-1 alpha polymorphisms were not associated with breast cancer risk. An inverse association noted in HSPA8 intronic 1541–1542delGT heterozygous carriers was not reproduced upon homozygous carriers; as a result, the allele dose–response model did not yield statistically significant results.
Table 2

Genotype frequencies and odds ratios regarding the examined polymorphisms

Genotype

Cases

Controls

OR (95 % CI)a

OR (95 % CI)b

N (%)

N (%)

HSP90 Gln488His (C > G)

 CC

54 (47.8)

90 (72.6)

1.00

1.00

 CG

20 (17.7)

21 (16.9)

1.59 (0.79–3.19)

1.62 (0.80–3.30)

 GG

39 (34.5)

13 (10.5)

5.00 (2.45–10.20)

4.78 (2.32–9.81)

 Allele dose–response

  

2.14 (1.53–2.99)

2.09 (1.49–2.94)

HSP70-2 P1/P2

 P1/P1

24 (21.2)

32 (25.8)

1.00

1.00

 P1/P2

82 (72.6)

76 (61.3)

1.44 (0.78–2.66)

1.52 (0.81–2.83)

 P2/P2

7 (6.2)

16 (12.9)

0.58 (0.21–1.64)

0.59 (0.20–1.75)

 Allele dose–response

  

0.93 (0.59–1.47)

0.96 (0.61–1.53)

HIF-1 alpha C1772T

 CC

98 (86.7)

107 (86.3)

1.00

1.00

 CT

15 (13.3)

17 (13.7)

0.96 (0.46–2.03)

0.94 (0.44–2.01)

 TT

0 (0.0)

0 (0.0)

Not estimable

Not estimable

 Allele dose–response

  

0.96 (0.46–2.03)

0.94 (0.44–2.01)

HSPA8 intronic 1541–1542delGT

 wt

83 (73.5)

71 (57.3)

1.00

1.00

 het

25 (22.1)

51 (41.1)

0.42 (0.24–0.74)

0.42 (0.24–0.76)

 del/del

5 (4.4)

2 (1.6)

2.14 (0.40–11.36)

2.36 (0.44–12.75)

 Allele dose–response

  

0.63 (0.39–1.02)

0.64 (0.39–1.05)

aUnadjusted OR; b OR adjusted for age, smoking, alcohol, BMI, menopausal status age at menarche and education

Regarding deviation from HWE, significant deviation was noted regarding HSP90 Gln488His (Pearson’s chi2(1) = 24.97, p < 0.001), HSP70-2 P1/P2 (Pearson’s chi2(1) = 7.54, p = 0.006) and HSPA8 (Pearson’s chi2(1) = 4.55, p = 0.033). On the other hand, regarding HIF-1 alpha C1772T the deviation was not statistically significant (Pearson’s chi2(1) = 0.67, p = 0.413).

Table 3 presents genotype frequencies and odds ratios regarding the examined polymorphisms. The positive association between HSP90 G allele and breast cancer risk seemed to pertain to both premenopausal (OR = 2.21, 95 %CI: 1.18–4.15 per allele) and postmenopausal women (OR = 2.04, 95 % CI: 1.36–3.07). The null associations regarding HSP70-2 P1/P2, HSPA8 intronic 1541–1542delGT and HIF-1α C1772T seemed reproducible upon both premenopausal and postmenopausal women.
Table 3

Genotype frequencies and odds ratios regarding the examined polymorphisms

Genotype

Premenopausal women

Postmenopausal women

Cases N (%)

Controls N (%)

OR (95 % CI)

Cases N (%)

Controls N (%)

OR (95 % CI)

HSP90 Gln488His (C > G)

 CC

15 (40.5)

24 (75.0)

1.00

39 (51.3)

66 (71.7)

1.00

 CG

9 (24.3)

3 (9.4)

4.92 (1.10–22.05)

11 (14.5)

18 (19.6)

1.07 (0.45–2.52)

 GG

13 (35.1)

5 (15.6)

4.01 (1.17–13.78)

26 (34.2)

8 (8.7)

5.12 (2.09–12.56)

 Allele dose–response

  

2.21 (1.18–4.15)

  

2.04 (1.36–3.07)

HSP70-2 P1/P2

 P1/P1

10 (27.0)

8 (25.0)

1.00

14 (18.4)

24 (26.1)

1.00

 P1/P2

24 (64.9)

21 (65.6)

1.02 (0.33–3.21)

58 (76.3)

55 (59.8)

1.96 (0.90–4.24)

 P2/P2

3 (8.1)

3 (9.4)

0.54 (0.07–4.30)

4 (5.3)

13 (14.1)

0.59 (0.15–2.25)

 Allele dose–response

  

0.90 (0.38–2.15)

  

1.01 (0.58–1.76)

HIF-1 alpha C1772T

 CC

32 (86.5)

28 (87.5)

1.00

66 (86.8)

79 (85.9)

1.00

 CT

5 (13.5)

4 (12.5)

1.01 (0.24–4.26)

10 (13.2)

13 (14.1)

0.90 (0.36–2.20)

 TT

0 (0.0)

0 (0.0)

Not estimable

0 (0.0)

0 (0.0)

Not estimable

 Allele dose–response

  

1.01 (0.24–4.26)

  

0.90 (0.36–2.20)

HSPA8 intronic 1541–1542delGT

 wt

28 (75.7)

17 (53.1)

1.00

55 (72.4)

54 (58.7)

1.00

 het

8 (21.6)

14 (43.8)

0.35 (0.12–1.01)

17 (22.4)

37 (40.2)

0.47 (0.23–0.93)

 del/del

1 (2.7)

1 (3.1)

0.53 (0.03–9.51)

4 (5.3)

1 (1.1)

4.42 (0.47–41.56)

 Allele dose–response

  

0.43 (0.17–1.08)

  

0.75 (0.42–1.34)

Odds ratios were derived from unconditional logistic regression adjusted for age, smoking, alcohol, BMI, age at menarche and education

With respect to survival analysis, none of the aforementioned polymorphisms was associated with either disease-free survival or overall survival (Table 4). With respect to survival analysis, cases were followed-up for a total of 661.4 person-years (mean follow-up: 70.2 months). None of the examined polymorphisms was associated with either disease-free survival or overall survival (Table 4). Figure 1 presents Kaplan–Meier overall survival curves for HSP90 genotype subgroups.
Table 4

Results of univariate cox regression analysis regarding the examined polymorphisms

Genotype

Cases

Disease-free survival

Overall survival

N (%)

HR (95 % CI)

HR (95 % CI)

HSP90 Gln488His (C > G)

 CC

54 (47.8)

1.00

1.00

 CG

20 (17.7)

1.01 (0.40–2.58)

1.20 (0.40–3.63)

 GG

39 (34.5)

1.33 (0.64–2.74)

1.46 (0.62–3.46)

 Allele dose–response

 

1.16 (0.81–1.68)

1.24 (0.80–1.90)

HSP70-2 P1/P2

 P1/P1

24 (21.2)

1.00

1.00

 P1/P2

82 (72.6)

1.32 (0.55–3.19)

1.09 (0.41–2.91)

 P2/P2

7 (6.2)

1.28 (0.32–5.17)

0.79 (0.15–4.18)

 Allele dose–response

 

1.16 (0.62–2.17)

0.99 (0.48–2.07)

HIF-1 alpha C1772T

 CC

98 (86.7)

1.00

1.00

 CT

15 (13.3)

0.49 (0.17–1.42)

0.68 (0.23–2.02)

 TT

0 (0.0)

Not estimable

Not estimable

 Allele dose–response

 

0.49 (0.17–1.42)

0.68 (0.23–2.02)

HSPA8 intronic 1541–1542delGT

 wt

83 (73.5)

1.00

1.00

 het

25 (22.1)

1.47 (0.65–3.33)

1.23 (0.45–3.36)

 del/del

5 (4.4)

0.61 (0.08–4.51)

0.83 (0.11–6.21)

 Allele dose–response

 

1.06 (0.58–1.93)

1.04 (0.51–2.13)

https://static-content.springer.com/image/art%3A10.1007%2Fs11033-012-1984-2/MediaObjects/11033_2012_1984_Fig1_HTML.gif
Fig. 1

Kalpan–Meier survival estimates

Discussion

This is the first case–control study examining the HSP90α Gln488His polymorphism in breast cancer; of note, a positive association between HSP90 Gln488His G allele emerged and seemed to pertain to both premenopausal and postmenopausal women. The direction of the association seemed to be in accordance with the published functional data, according to which this G(His) allele may be directly linked to reduced Hsp90 function due to reduced dimerization [11, 12]. Indeed, Hsp90 seems to be a crucial factor-chaperone for the optimal function of the cell, orchestrating the regulation of various key proteins, such as ER, p53, HIF-1alpha, Akt, Raf-1 MAP kinase and tyrosine kinases of the erbB family (reviewed in [9]).

Nevertheless, regarding survival, no statistically significant associations were documented as far as HSP90 Gln488His status is concerned. Despite a trend (HR = 1.46) towards shorter overall survival in G carriers, that finding did not reach statistical significance. Indeed, further larger studies with longer follow-up periods seem of special importance to disentangle the effects of Gln488His polymorphism. Our findings may also be of clinical significance, as currently various Hsp90 inhibitors that bind to Hsp90 at its ATP-binding site are tested in clinical trials; 17-allylamino, 17-demethoxygeldanamycin (17-AAG) is the first Hsp90 inhibitor to be clinically investigated and has yielded promising results in patients with HER2-overexpressing metastatic breast cancer [33]. The interaction of Gln488His polymorphism with outcomes in the ongoing trials seem an undiscovered field of particular significance.

The present study is the first to examine HSPA8 1541–1542delGT in the context of breast cancer epidemiology. This polymorphism has been identified within breast cancer tissue [19] and may modify HSPA8 protein expression [23]. Nevertheless, our data did not support a sizeable effect mediated by this intronic polymorphism either in the context of relative risk or regarding survival.

Regarding Hsp70-2 P1/P2 polymorphism, our null finding is not in accordance with the study by Mestiri et al. [18], according to which the HSP 70-2 P2 allele may be associated with increased breast cancer risk. However, our study seems closed to the functional data, as HSP70-2 P1/P2 polymorphism represents a silent mutation and any effects have been attributed to linkage disequilibrium with nearby, unknown genes in chromosome six [14, 18]. The accumulation of data and their quantitative synthesis will eventually result in a clearer picture of the underlying phenomenon.

Our null results on HIF-1α C1772T polymorphism are in accordance with the recent meta-analysis, which synthesized three studies and suggested a null association with breast cancer risk [28].

Despite the originality and the elaborate multivariate modeling of the present case–control study, certain limitations should be acknowledged and discussed. First, the documentation of significant deviation from the Hardy–Weinberg Equilibrium should be declared. To our knowledge the reasons for this deviation in our sample cannot be explained with certainty; population stratification [31] i.e., genotype frequencies varying among the Greek ancestry regions of controls may potentially explain this observation. Indeed, Greece is a mountainous country with numerous islands, where geographical segregation prevails and may have profound impact even upon the population synthesis of the capital Athens due to the major internal migration and rapid urbanization. At any case, Hardy–Weinberg disequilibrium may not necessarily invalidate the results of an association study [31]. Moreover, a larger sample size with longer follow-up seems desirable to further examine the non-signficant trends in terms of survival. Furthermore, the fact that polymorphism status in cases was evaluated based on paraffin embedded breast normal tissues rather than peripheral blood should be acknowledged; future studies on peripheral blood are anticipated to extrapolate and validate the present findings.

In conclusion, HSP90α Gln488His polymorphism seems to be a risk factor for breast cancer. On the other hand, our study did not point to excess risk conferred by HSPA8 1541–1542delGT, Hsp70-2 P1/P2 and HIF-1α C1772T. Further, larger studies on other countries and subjects coming from other races seem warranted.

Acknowledgments

A research grant has been received by HeSMO. FZ is receiving a research grant from HeSMO.

Conflict of interest

The authors have declared no conflicts of interest.

Copyright information

© Springer Science+Business Media Dordrecht 2012