A comparative biomarker study of 514 matched cases of male and female breast cancer reveals gender-specific biological differences

Download A comparative biomarker study of 514 matched cases of male and female breast cancer reveals gender-specific biological differences

Post on 26-Aug-2016




0 download

Embed Size (px)


<ul><li><p>PRECLINICAL STUDY</p><p>A comparative biomarker study of 514 matched cases of maleand female breast cancer reveals gender-specific biologicaldifferences</p><p>Abeer M. Shaaban Graham R. Ball Rebecca A. Brannan Gabor Cserni </p><p>Anna Di Benedetto Jo Dent Laura Fulford Helen Honarpisheh </p><p>Lee Jordan J. Louise Jones Rani Kanthan Loaie Maraqa Maria Litwiniuk </p><p>Marcella Mottolese Steven Pollock Elena Provenzano Philip R. Quinlan </p><p>Georgina Reall Sami Shousha Mark Stephens Eldo T. Verghese </p><p>Rosemary A. Walker Andrew M. Hanby Valerie Speirs</p><p>Received: 3 October 2011 / Accepted: 25 October 2011 / Published online: 18 November 2011</p><p> Springer Science+Business Media, LLC. 2011</p><p>Abstract Male breast cancer remains understudied</p><p>despite evidence of rising incidence. Using a co-ordinated</p><p>multi-centre approach, we present the first large scale</p><p>biomarker study to define and compare hormone receptor</p><p>profiles and survival between male and female invasive</p><p>breast cancer. We defined and compared hormone receptor</p><p>profiles and survival between 251 male and 263 female</p><p>breast cancers matched for grade, age, and lymph node</p><p>status. Tissue microarrays were immunostained for ERa,ERb1, -2, -5, PR, PRA, PRB and AR, augmented by</p><p>HER2, CK5/6, 14, 18 and 19 to assist typing. Hierarchical</p><p>clustering determined differential nature of influences</p><p>between genders. Luminal A was the most common phe-</p><p>notype in both sexes. Luminal B and HER2 were not seen</p><p>in males. Basal phenotype was infrequent in both. No</p><p>differences in overall survival at 5 or 10 years were</p><p>observed between genders. Notably, AR-positive luminal</p><p>A male breast cancer had improved overall survival over</p><p>female breast cancer at 5 (P = 0.01, HR = 0.39, 95%</p><p>CI = 0.260.87) but not 10 years (P = 0.29, HR = 0.75,</p><p>95% CI = 0.461.26) and both 5 (P = 0.04, HR = 0.37,</p><p>95% CI = 0.070.97) and 10 years (P = 0.04, HR =</p><p>0.43, 95% CI = 0.120.97) in the unselected group.</p><p>Hierarchical clustering revealed common clusters between</p><p>genders including total PRPRAPRB and ERb1/2 clus-ters. A striking feature was the occurrence of ERa on</p><p>Presented in part at the 32nd Annual San Antonio Breast Cancer</p><p>Symposium, 9-13 December, 2009, San Antonio, TX, Breast Cancer</p><p>Research 2010, 18 May 2010, London UK and the Pathological</p><p>Society of Great Britain and Ireland 2010 Summer Meeting, 30 June-</p><p>3 July, St Andrews, UK.</p><p>A. M. Shaaban R. A. Brannan G. Reall E. T. Verghese A. M. HanbySt Jamess Institute of Oncology, St Jamess University Hospital,</p><p>Leeds, UK</p><p>G. R. Ball</p><p>Nottingham Trent University, Nottingham, UK</p><p>R. A. Brannan H. Honarpisheh L. Maraqa S. Pollock E. T. Verghese A. M. Hanby V. Speirs (&amp;)Leeds Institute of Molecular Medicine, Wellcome Trust Brenner</p><p>Building, University of Leeds, Leeds LS9 7TF, UK</p><p>e-mail: v.speirs@leeds.ac.uk</p><p>G. Cserni</p><p>Bacs-Kiskun County Teaching Hospital, Nyiri ut 38,</p><p>Kecskemet 6000, Hungary</p><p>A. D. Benedetto M. MottoleseRegina Elena Cancer Institute, Rome, Italy</p><p>J. Dent</p><p>Calderdale Hospital, Halifax, UK</p><p>L. Fulford</p><p>Surrey &amp; Sussex NHS Trust, Redhill, UK</p><p>L. Jordan P. R. QuinlanUniversity of Dundee/NHS Tayside, Dundee, UK</p><p>J. L. Jones</p><p>Barts Cancer Institute, Barts and The London School</p><p>of Medicine and Dentistry, London, UK</p><p>R. Kanthan</p><p>University of Saskatchewan, Saskatoon, Canada</p><p>M. Litwiniuk</p><p>Poznan University of Medical Sciences, Poznan, Poland</p><p>123</p><p>Breast Cancer Res Treat (2012) 133:949958</p><p>DOI 10.1007/s10549-011-1856-9</p></li><li><p>distinct clusters between genders. In female breast cancer,</p><p>ERa clustered with PR and its isoforms; in male breastcancer, ERa clustered with ERb isoforms and AR. Our datasupports the hypothesis that breast cancer is biologically</p><p>different in males and females suggesting implications for</p><p>clinical management. With the incidence of male breast</p><p>cancer increasing this provides impetus for further study.</p><p>Keywords Male breast cancer Hormone receptors Androgen receptor Hierarchical clustering</p><p>Introduction</p><p>According to figures from Cancer Research UK, there were</p><p>45,695 cases of female breast cancer (FBC) and 277 cases</p><p>of male breast cancer (MBC) diagnosed in the UK in 2007</p><p>[1]. In the US it was estimated that 1,970 men and 207,090</p><p>women would be diagnosed with breast cancer in 2010 [2].</p><p>Whilst MBC accounts for less than 1% of breast cancer</p><p>diagnoses worldwide, the overall improvements in survival</p><p>and mortality observed in FBC has not been seen to the</p><p>same extent in MBC, as demonstrated in a recent interro-</p><p>gation of the Surveillance Epidemiology and End Results</p><p>(SEER) database [3]. Moreover, the incidence rate of MBC</p><p>is rising steadily [47].</p><p>The etiology of MBC is poorly understood with most of</p><p>our current knowledge regarding its biology, natural his-</p><p>tory, and treatment extrapolated from FBC. Retrospective</p><p>studies are generally weakened by the small numbers of</p><p>cases available from any one centre with studies published</p><p>on as few as 15 cases [8], making it hard to draw biolog-</p><p>ically meaningful conclusions. It is, therefore, a challenge</p><p>to accrue sufficiently large numbers to allow comparative</p><p>analysis of possible prognostic or predictive biomarkers.</p><p>Many articles imply a general similarity of MBC to FBC</p><p>and this has resulted in MBC patients being treated in</p><p>exactly the same way as females in the clinic, which may</p><p>not be optimal. Survival rates for MBC are generally</p><p>assumed to be lower than FBC, probably as a result of later</p><p>diagnosis and the assumption that treatments which are</p><p>proven in FBC through clinical trials will have the same</p><p>impact in men [9].</p><p>A 40-year review of records of 759 cases from invasive</p><p>MBC from the US Armed Forces Institute of Pathology</p><p>database showed that the frequency of histological sub-</p><p>types in men was comparable to that of FBC, with the</p><p>exception of papillary carcinoma which was twice more</p><p>common in MBC [10]. To date, modern molecular sub-</p><p>typing has been reported in a single study of MBC where</p><p>only luminal A (35/42) and luminal B (7/42) subtypes were</p><p>observed [11].</p><p>Using a co-ordinated multi-centre approach, the aim of</p><p>this study was to conduct the first large scale study to</p><p>address and compare the expression profile of hormone</p><p>receptors and their effect on survival in FBC and MBC.</p><p>Methods</p><p>Patient cohorts</p><p>Following ethical approval from the Leeds (West) Research</p><p>Ethics Committee (06/Q1205/156), 514 formalin-fixed</p><p>paraffin-embedded blocks of male (251) and matched</p><p>female (263) breast cancers were obtained retrospectively.</p><p>The latter were all from Europe and the former from Europe</p><p>(n = 196) and Canada (n = 55). Informed consent was not</p><p>required as the anonymised material pre-dated September</p><p>2006, came from a Tissue Bank approved by the UK Human</p><p>Tissue Authority (or equivalent) or were from non-UK</p><p>patients. Patients had not received any therapy before sur-</p><p>gery. Details on adjuvant therapy were not extensively</p><p>available; where available this was predominantly endocrine</p><p>therapy (tamoxifen). Patient characteristics are presented in</p><p>Table 1. Cases were reviewed by specialised breast con-</p><p>sultant histopathologists (AMH, AMS, RAB) to confirm</p><p>histology and marked up for assembly into tissue micro-</p><p>arrays (TMAs) using 3 9 0.6 mm tissue cores per case</p><p>taken from formalin-fixed paraffin-embedded material as</p><p>previously described [10].</p><p>Immunohistochemistry (IHC)</p><p>Antibodies, dilution, and retrieval methods are listed in</p><p>Table 1. The antibody panel was focused on hormone</p><p>receptors oestrogen receptor (ER)a, ERb isoforms, pro-gesterone receptor (PR) isoforms and androgen receptor</p><p>(AR) and additional biomarkers selected to distinguish</p><p>molecular subtypes of breast cancer (CK5/6, 14, 18,</p><p>HER2). Each marker was run as a batch with appropriate</p><p>positive (tissue known to express the biomarker of interest)</p><p>and negative (no primary antibody) controls. Scoring was</p><p>overseen by AMH, AMS, and RAB. Following visualisa-</p><p>tion of the signal with 3-30diaminobenzidine chromogen,TMAs were digitised (Aperio Technologies), and hormone</p><p>E. Provenzano</p><p>Addenbrookes Hospital, Cambridge, UK</p><p>S. Shousha</p><p>Imperial College, London, UK</p><p>M. Stephens</p><p>University Hospital of North Staffordshire, Stoke-on Trent, UK</p><p>R. A. Walker</p><p>University of Leicester, Leicester, UK</p><p>950 Breast Cancer Res Treat (2012) 133:949958</p><p>123</p></li><li><p>receptor immunoreactivity was scored using the Allred</p><p>system with the following cut offs: ERa[ 2, ERb (andisoforms) [ 3, AR [ 2, PR (and isoforms) [ 2, as vali-dated in previous studies [1214].</p><p>Hierachical clustering and principal components</p><p>analysis (PCA)</p><p>For hierarchical cluster analysis, IHC measurements were</p><p>used as inputs for all cases in each of the male and female</p><p>cohorts. A Euclidian distance measure was employed with</p><p>complete linkage of clusters. Clustering was conducted for</p><p>the data structure of cases with the immunohistochemical</p><p>parameters and for the immunohistochemical parameters as</p><p>they were expressed through the population. Cluster</p><p>dendrograms were plotted for both of these analyses for</p><p>each gender and compared. PCA was applied to the same</p><p>dataset. Analysis was based on covariances between</p><p>parameters and cases in the data. Variances were computed</p><p>based on the sum of squares/n - 1. Plots of the influences</p><p>of variables in the factor plane of the first and second and</p><p>the second and third principal components for both the</p><p>male and female cases were plotted separately. As with the</p><p>hierarchical clustering, both the male and female cases</p><p>were combined into a single data set. The influence of</p><p>variables in the factor plane of the combined cases were</p><p>plotted for the first and second and the second and third</p><p>principal components. The distributions of cases within the</p><p>combined sets were also plotted.</p><p>Statistical analysis</p><p>Patient and disease characteristics were compared between</p><p>male and FBCs using the v2 test (GraphPad). Associationswith disease-free and overall survival (DFS and OS,</p><p>respectively) were analysed by KaplanMeier plots and log</p><p>rank test. P-values were two-sided, and P \ 0.05 wasconsidered significant.</p><p>Results</p><p>A total of 514 cases of breast carcinoma were studied,</p><p>including 251 males and 263 females. The median age for</p><p>the male cohort was 66 years (range 3094) and 59 years</p><p>(range 2792) for females. Patient characteristics are</p><p>shown in Table 2. As this was a matched cohort, no sig-</p><p>nificant differences were observed in grade, or lymph node</p><p>status between genders. Significant differences were</p><p>observed in the distribution of histopathological subtypes</p><p>(P \ 0.0001). There was an even distribution of ductalphenotype whilst lobular carcinomas found in 9% of the</p><p>female cohort was only seen in a single male case. Papil-</p><p>lary and mucinous phenotypes were restricted to males. A</p><p>significantly higher proportion of males expressed ERacompared to females (80 and 68%, respectively), although</p><p>no differences in the frequency of PR was observed (71 and</p><p>72%, respectively). Follow up data was available on 183</p><p>(73%) male and 237 (90%) female cases.</p><p>Both cohorts were classified into molecular subtypes by</p><p>IHC: luminal A (ERa?, and/or PR?, HER2-), luminal B(ERa?, and/or PR?, HER2?), HER2 (ERa, PR-, HER2?)and basal-like (ERa-, PR-, HER2-, CK5/6?) according toprevious studies [1517]. Representative immunopro-</p><p>files for each subgroup are shown in Fig. 1. Significant</p><p>Table 1 Patient characteristics of male and female breast cancers</p><p>Characteristic Male</p><p>(n = 251)no (%)</p><p>Female</p><p>(n = 263)no (%)</p><p>P values</p><p>Histology \0.0001Ductal 208 (83) 220 (84)</p><p>Lobular 1 (0.4) 23 (9)</p><p>Papillary 11 (4) 0</p><p>Intraductal papillary 4 (1.5) 0</p><p>Micropapillary 1 (0.4) 0</p><p>Mucinous 9 (4) 0</p><p>Mixed 13 (5) 14 (7)</p><p>Unknown 4 (1.6) 0</p><p>Grade 0.948</p><p>1 25 (10) 29 (11)</p><p>2 128 (51) 140 (53)</p><p>3 81 (33) 94 (36)</p><p>Unknown 17 (6) 0</p><p>Lymph node 0.521</p><p>? 76 (30) 131 (50)</p><p>- 80 (32) 121 (46)</p><p>Unknown 95 (48) 11 (4)</p><p>ERa \0.0001? 201 (80) 180 (68)</p><p>- 22 (9) 56 (21)</p><p>Unknown 28 (11) 27 (10)</p><p>PR 0.838</p><p>? 177 (71) 190 (72)</p><p>- 39 (15) 44 (17)</p><p>Unknown 35 (14) 29 (11)</p><p>Subtypea 0.0004</p><p>Luminal A 199 (98) 197 (90)</p><p>Luminal B 0 (0) 14 (6)</p><p>Basal 4 (2) 4 (2)</p><p>HER2?/ERa- 0 5 (2)</p><p>a Subtype classification was not possible in cases from 48 males and</p><p>43 females due to core loss from the section for at least one of the</p><p>biomarkers, or less commonly, exhaustion of the TMA core</p><p>Breast Cancer Res Treat (2012) 133:949958 951</p><p>123</p></li><li><p>differences were observed between molecular subtypes</p><p>(P = 0.0004). Luminal A was seen in 98% of males and 90%</p><p>of females. Luminal B or HER2 subgroups were not</p><p>observed in males but found in 6 and 2% of females,</p><p>respectively. Basal-like tumours (ERa-, PR-, HER2-,CK5/6?) were infrequent in both cohorts (2% in each).</p><p>We then examined the frequencies of expression</p><p>between genders of other hormone receptors including AR,</p><p>nuclear and cytoplasmic ERb1 and ERb2, nuclear ERb5,plus the PR isoforms A and B (Table 3). AR immunore-</p><p>activity was expressed in 64% of males and 93% females,</p><p>respectively, (P \ 0.0001). For ERb1 and -2, both nuclearand cytoplasmic immunoreactivity was assessed [12].</p><p>ERb1 nuclear immunoreactivity was significantly expres-sed in FBC whilst cytoplasmic ERb1 and ERb2 immuno-reactivity were associated with MBC. No associations were</p><p>observed for ERb5. Of the PR isoforms, only PRA wassignificantly expressed in MBC. As the male cohort con-</p><p>tained cases of European and Canadian origin, we tested if</p><p>there were differences between these; none were found.</p><p>In luminal A carcinomas, no differences in overall sur-</p><p>vival were observed between genders at either 5 or</p><p>10 years (Fig. 2a, b). This was also reflected in the unse-</p><p>lected cohorts (data not shown). When hormone receptor</p><p>expression was considered, only AR significantly associ-</p><p>ated with survival. AR-positive luminal A MBC had sig-</p><p>nificantly improved overall survival over the equivalent</p><p>FBC at 5 (P = 0.01, HR = 0.39, 95% CI = 0.260.87) but</p><p>not 10 (P = 0.29, HR = 0.75, 95% CI = 0.461.26) years</p><p>(Fig. 2c, d). In the unselected group, ERa and AR-positiveMBC had significantly improved overall survival over ERaand AR-negative cases at both 5 (P = 0.04, HR = 0.37,</p><p>95% CI = 0.070.97) and 10 (P = 0.04, HR = 0.43, 95%</p><p>CI = 0.120.97) years (Fig. 2e, f) with ERa and AR-</p><p>positive MBC also having significantly improved overall</p><p>survival over the equivalent FBCs at 5 (P = 0.05,</p><p>HR = 0.48, 95% CI = 0.291.00) but not 10 (P = 0.37,</p><p>HR = 0.79, 95% CI = 0.481.32) years (Fig. 2g, h).</p><p>Hierarchical clustering based on hormone receptor profiles</p><p>classified MBC and FBC into three distinct groups (Fig. 3).</p><p>The cytoplasmic ERb cluster was common to both genders. InFBC an ERa/PR cluster was observed, grouping ERa and PRisoforms, whilst ERb isoforms clustered with AR (ERb/ARcluster). In MBC, there were striking changes in the position</p><p>of ERa; AR and ERa clustered with ERb (ERa/b AR cluster)whilst PR isoforms formed an independent cluster (PR clus-</p><p>ter). This was also reflected in a PCA-based plot of variable</p><p>factor co-ordinates (data not shown).</p><p>Discussion</p><p>Currently, MBC is treated based on the assumption that it</p><p>is essentially the same disease as FBC. In this the largest</p><p>comparative study to date, directly comparing the immu-</p><p>nohistochemical profile of matched MBCs an...</p></li></ul>


View more >