Accepted Article
Received Date : 18-Dec-2013 Revised Date : 12-Feb-2014 Accepted Date : 19-Feb-2014 Article type
: Original Article
The prognostic significance of the aberrant extremes of p53 immunophenotypes in breast cancer
Running title: Aberrant extremes of p53 IHC in breast cancer
David P. Boyle1 and Darragh G. McArt1, Gareth Irwin1, Charlotte S. Wilhelm-Benartzi2, Tong F. Lioe3, Elena Sebastian1, Stephen McQuaid1, Peter W. Hamilton1, Jacqueline A. James1, Paul B. Mullan1, Mark A. Catherwood4, D. Paul Harkin1, Manuel Salto-Tellez1
Authors’ affiliation 1. Centre for Cancer Research and Cell Biology, Queen’s University Belfast, Belfast, United Kingdom 2. Epigenetics Unit, Department of Surgery and Cancer, Imperial College, London, United Kingdom 3. Department of Histopathology, Belfast City Hospital, Belfast, United Kingdom This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/his.12398 This article is protected by copyright. All rights reserved.
Accepted Article
4. Department of Haematology, Belfast City Hospital, Belfast, United Kingdom
Corresponding author: Dr David P Boyle Centre for Cancer Research and Cell Biology Queen’s University Belfast, 97 Lisburn Road, Belfast BT9 7BL Telephone: +44(0)2890972243 Fax: +44(0)2890972776 Email: [email protected]
Disclosure/Conflict of interest None: Darragh G. McArt, Charlotte S. Wilhelm-Benartzi, Tong F. Lioe, Elena Sebastian, Peter W. Hamilton, Paul B. Mullan, Mark A. Catherwood, Manuel Salto-Tellez Scientific Director of the Northern Ireland Biobank: Jacqueline James; Deputy Director of the Northern Ireland Biobank: Stephen McQuaid; President & Managing Director, Almac Diagnostics: D. Paul Harkin
This article is protected by copyright. All rights reserved.
Accepted Article
Cancer Research UK and the Pathological Society funding: David P. Boyle; Northern Ireland Health and Social Care Research and Development Division funding: Gareth Irwin
Abstract Introduction: Utility of p53 as a prognostic assay has been elusive. We describe a novel, reproducible scoring system and assess the relationship between differential p53 IHC expression patterns, TP53 mutation status and patient outcomes for breast cancer. Methods and results: Tissue microarrays were used to study p53 IHC expression patterns: expression was defined as extreme positive (EP), extreme negative (EN) and intermediate patterns as non-extreme (NE). Overall survival was used to define patient outcome. A representative subgroup (n=30) displaying the various p53 immunophenotypes was analysed for TP53 hotspot mutation status (exons 4-9). Extreme expression of any type occurred in 176/288 (61%) cases. EP p53 compared to NE p53 was significantly associated (p = 0.039) with poor OS. In addition, EN p53 compared to NE p53 was associated (p = 0.059) with poor OS. Combining cases displaying either EP/EN expression better predicted OS than either pattern alone (p = 0.028). This combination immunophenotype was significant in univariate but not multivariate analysis. In subgroup analysis six substitution exon mutations were detected, all corresponding to extreme IHC phenotypes. Five missense mutations corresponded to EP staining while the nonsense mutation corresponded to EN staining. None were detected in the NE group. Conclusions: Patients with extreme p53 IHC expression have a worse OS compared to those with NE expression. Accounting for EN as well as EP p53 improves its prognostic impact. Extreme expression positively correlates with nodal stage and histological grade and
This article is protected by copyright. All rights reserved.
Accepted Article
negatively with hormone receptor status. Extreme expression may relate to specific mutational status. Keywords: Breast cancer, prognosis, p53 protein, immunohistochemistry, mutation
Introduction Extensively researched, p53 has been referred to as ‘the guardian of the genome’.1 As such it is mutated at various generally high frequencies across a range of tumour and tissue types2 not only resulting in loss of tumour suppressor activity but also possibly causing gain of function activity including increased invasive potential and poor therapeutic response3. In breast cancer TP53 is the second most commonly mutated gene with an estimated rate of 23%4. Despite an extensive research effort, characterisation of a p53 prognostic assay has been elusive in breast cancer and consequently it is not currently recommended as a tumour biomarker in routine clinical management5.
Many studies have previously focussed on p53 immunohistochemistry (IHC) overexpression alone as a marker of poor prognosis. In an attempt to explain the biological significance of p53 overexpression and its possible correlation with mutation status the significance of p53 protein absence by IHC may be overlooked. Given the undoubted inherent intra- and intertumoural heterogeneity in breast cancer when examined across a variety of assay techniques it is likely that a range of mutational and epigenetic phenomena contribute to abnormal p53 function and protein expression whether this manifests as absent or strong positive expression by IHC. Aberrant expression, whatever the cause, may then be a marker of poor prognosis.
This article is protected by copyright. All rights reserved.
Accepted Article
A previous lack of appreciation that complete absence of p53 expression may be a particular staining pattern, as opposed to part of a spectrum of normal or wildtype expressions, has prevented study of its potential significance. Indeed, IHC expression patterns and intensities may be better thought of as a measure of spectra of functioning protein status, thresholds above and below which significant impaired function has occurred, rather than a direct or at best subtotal correlate of mutation status. Negating consideration of absent IHC expression in comparisons with prognosis, associations with therapeutic response, and mutational status may in part account for conflicting reports in the literature5. Although different biomarker platforms demonstrate particular and unique qualities, the application of sequencing techniques to clinical practice requires substantial technical and financial investment. Additionally, the inactivation of TP53 via alternative aberrations resulting in similar altered functional status may be overlooked. Using a revised p53 IHC platform may help address these issues. In the current study we aim to correlate reproducible p53 expression patterns with patient outcome for breast cancer. We hypothesise that inclusion of negative expression patterns will improve prognostic significance. We record in a reproducible way abnormal p53 protein expression to include extreme positive (EP) and negative (EN) expression in breast cancer and its relation to outcome. These abnormal p53 expression patterns are together compared against non-extreme (NE) expression patterns and the relationship to patient outcome described. Materials and methods Study design and patient selection We conducted a retrospective case-control study involving a cohort of breast cancer patients with ethical approval obtained and tissue acquired through the Northern Ireland Biobank
This article is protected by copyright. All rights reserved.
Accepted Article
(NIB ref: 12-00017). A dataset was compiled for a cohort of 303 female patients with de novo breast cancer and included a range of clinical, pathological and outcome parameters (table 1). All of the patients were diagnosed and received treatment in Northern Ireland. The vast majority of resection specimens were processed and reported in one of two hospitals in the Belfast catchment area. Surgical resection of tumours was performed between September 1997 and May 2009: resection specimens comprised total or partial mastectomies with axillary node clearance. The entire cohort received anthracycline based chemotherapy with or without radiotherapy. Administration of hormone therapy and trastuzumab were directed by hormone receptor (HR) and HER2 status. None of the patients were treated neoadjuvantly. Original HR and HER2 scores were available and were used to assess representativeness of tissue microarray (TMA) cores for full face sections. Patient exclusion criteria included male sex, no anthracycline administration and past history of cancer of any type. Cases were defined as patients who died during follow-up at 60 months compared to surviving controls. The study was performed with reference to REMARK guidelines6.
Tissue microarray construction From a total of 303 identified patients, slides and paraffin blocks were retrieved for 293 patients (in 10 cases slides were available but corresponding paraffin blocks could not be located). Original haemotoxylin and eosin (H&E) sections of all of the slides were reviewed for tumour block selection. Following block location a new section was cut for H&E and the slides annotated (DPB) for TMA construction. For each selected donor block three representative areas were annotated for targeted coring. The TMA was constructed using a manual tissue arrayer (Beecher Instruments, Silver Spring, MD) as indicated before7. The
This article is protected by copyright. All rights reserved.
Accepted Article
manual arrayer was used to extract 1mm diameter tissue cores from donor blocks for insertion into recipient blocks. Three cases had extremely focal tumour and were not included in the TMA yielding a final total of 290 individual breast cancer cases.
Immunohistochemistry and haemotoxylin and eosin All IHC was performed in a hybrid laboratory (Northern Ireland Molecular Pathology Laboratory) which has UK Clinical Pathology Accreditation, and the infrastructure to process both clinical patient samples and research materials. Sections were cut from the TMA blocks for H&E and IHC. The initial section was used for H&E staining to assess TMA quality and appropriate tumour content for subsequent IHC localisation and analysis. Sections for IHC were cut at 4 microns on a rotary microtome, dried at 37oC overnight and then used for IHC tests which were performed on an automated immunostainer (Leica BOND-MAX™). Validated and optimised protocols, used in local diagnostics, were selected for each biomarker with inclusion of carefully selected control tissues during antibody application. Antigen binding sites were detected with a polymer based detection system (Bond cat no. DS 9800). All sections were visualized with DAB, counterstained in haematoxylin and mounted in DPX. Biomarker conditions were as follows: p53 (clone DO-7 mouse monoclonal antibody, Dako) was used at a 1:100 dilution with epitope retrieval solution 1 pre-treatment for 30 minutes. ER (clone 6F11 mouse monoclonal antibody, Leica) was used at a 1:200 dilution with epitope retrieval solution 2 pre-treatment for 30 minutes. PR (clone PgR 636 mouse monocloncal antibody, Dako) was used at a 1:150 dilution with epitope retrieval solution 1 pre-treatment for 20 minutes. HER2 (clone CB11 mouse monoclonal antibody, Leica) was used at the manufacturer’s pre-set optimum dilution using epitope retrieval solution 1 pre-treatment for 25 minutes. Control tissues included breast carcinoma sections
This article is protected by copyright. All rights reserved.
Accepted Article
known to be positive for ER, PR and p53 and cell line blots provided by the manufacturer for HER2.
HER2 dual in situ hybridisation Sections for Dual Hapten, Dual Colour In Situ Hybridization (DDISH) were cut at 4 microns on a rotary microtome and dried overnight at 37oC. DDISH was performed on a fully automated Ventana BENCHMARK® XT. Following dewaxing, sections were incubated in Protease 3 (Ventana Cat. No. 760-2020) for 12 minutes, denatured and then hybridized for 6 hours with an inform HER2 Dual ISH DNA probe cocktail (Ventana Cat. No. 800-4422). Ultraview SISH DNP detection kit (Ventana Cat. No. 800-098) was used to detect HER2 signals and Ultraview red ISH DIG detection kit (Ventana Cat. No. 800-505) was used to detect chromosome 17 signals. Sections were counterstained in haematoxylin and mounted in DPX.
Scoring and assessment Only cores with identifiable tumour as confirmed by pathology assessment of H&E slides were used in IHC analysis. All IHC was scored independently by at least 2 histopathologists (DPB and JJ or TFL) blinded to patient clinicopathological and outcome data. ER and PR expression was assessed using the quick (Allred) score method8. Only nuclear expression was considered in interpretation. Provided a proportional score of at least 2 was obtained, total quick scores of 3 or more were considered positive. This threshold is equivalent to at least 1% of cells demonstrating nuclear expression of any intensity9.
This article is protected by copyright. All rights reserved.
Accepted Article
Original ER, PR and HER2 scores were available for 287, 252, and 265 cases respectively. Scores generated from TMA assessment were compared to original clinically assigned scores to assess concordance and representativeness of the TMA. Additionally, a random subset of full sections from 60 cases used in the TMA had p53 IHC applied to assess concordance and representativeness of the TMA. Analysis of HER2 expression was based on the current USA/UK guidelines.10,11 Only membranous expression was considered in interpretation. Four scores (0 to 3+) are possible. A score of 3+ was considered as indicative of HER2 overexpression (positive) and scores of 0 and 1+ were considered negative. Tumours indeterminate for HER2 overexpression were assigned a score of 2+. These prompted application of HER2 DDISH to determine HER2 status. Assessment was performed based on UK clinical guidance.12
For p53 IHC a simple 3 tier scoring metric was applied wherein complete confluent negativity of staining was considered EN, strong diffuse confluent positivity was considered EP and all intermediate expression of any intensity considered NE (figure 1). Only one of the cores per case was required to show extreme expression to be considered as such. This approach approximates to a focality of expression status previously described for a different tissue type.13 If cases showed different extremes of expression then the final expression status was assigned according to the predominant expression pattern. If NE expression was present then all replicate cores were required to show NE expression to be considered as such. To be classified as EP, the nuclei of cells were stained very dark brown to black (i.e. moderate to strong expression) and generally had a “painted-on” appearance. Almost all of the cells expressed p53 when this degree of staining was apparent. At the other end of the spectrum, to be classified as EN, the nuclei of cells appeared blue due to visible counterstain as the
This article is protected by copyright. All rights reserved.
Accepted Article
predominant pattern. When assessable, normal lobular epithelium or stromal cells were checked to rule out false negativity: these tissues showed faint p53 expression serving as an internal control against which EN expression could be confirmed. All expression patterns between these extremes prompted interpretation as NE. When present, very low NE expression generally involved the majority of visualised cells. When disagreement on staining interpretation was detected the relevant cores were re-reviewed by 2 pathologists on a multihead microscope to reach consensus opinion.
Breast cancer subtypes According to the results of biomarker expression, breast cancers were classified as HR positive (ER+ and/or PR+, HER2-), HER2 positive (ER+/- and PR+/-, HER2+) or triple negative (TNBC) (ER-, PR-, HER2-). This categorisation was termed biomarker profile A. An alternative categorisation (biomarker profile B) dividing the cohort into luminal (ER+ and/or PR+) and non-luminal (ER- and PR-) subgroups was also applied. These categories were then used to analyse the subtype specific expression levels of p53.
Histopathology Grading and tumour morphological subtypes were derived from the original pathological information. Grade was confirmed on whole face assessment using the method described by Elston and Ellis.14 Tumour size was derived from macroscopic assessment as recorded in original reports with thresholds at less than 2cm and greater than 5cm.15
This article is protected by copyright. All rights reserved.
Accepted Article
Mutation analysis DNA extracted from 30 cases from whole face tumour sections was available for TP53 mutation analysis. These comprised equal proportions of cases demonstrating the EN, NE and EP phenotypes. Four x 10μm sections were cut for each case, deparaffinised in xylene, cleared with absolute ethanol and macrodissected according to tumour annotations (DPB). Standard procedures were used according to the manufacturer’s protocol for DNA extraction (QIAamp® DNA FFPE Tissue Kit, QIAGEN®) except that incubation at 56oC was extended to 3 days to optimise purification of genomic DNA. Quantification of DNA was performed by an absorbance method using the NanoDrop 2000c spectrophotometer.
Exons 4-9 were amplified and sequenced since they contain the most common TP53 mutations in breast cancer.16 Exons were amplified with primers as previously published by Gonzalez et al.17 For each exon, 100ng of DNA was amplified by PCR in a 25µL reaction containing 3mM MgCl2, 200µM dNTP, 1 U Platinum Taq (Invitrogen, UK) and 10 µM of each primer.17 Cycling conditions were as follows: 94oC for 10 minutes, 40 cycles at 94oC for 1 minute, 55oC for 1 minute and 72oC for 1 minute, and a final extension at 72oC for 10 minutes.
PCR products were directly sequenced by Sanger sequencing with Big Dye terminators v3.1 (Applied Biosystems, UK) using both forward and reverse primers. Mutations were confirmed by amplifying a duplicated PCR product. Polymorphism status and the confirmation of mutations were assessed with reference to The IARC TP53 database.16
This article is protected by copyright. All rights reserved.
Accepted Article
Statistical analytic methods Correlation (Kendall) and Fisher’s exact test were performed using the R statistical package. Results were considered statistically significant if p-values under p=50 = 1,
Received Date : 18-Dec-2013 Revised Date : 12-Feb-2014 Accepted Date : 19-Feb-2014 Article type
: Original Article
The prognostic significance of the aberrant extremes of p53 immunophenotypes in breast cancer
Running title: Aberrant extremes of p53 IHC in breast cancer
David P. Boyle1 and Darragh G. McArt1, Gareth Irwin1, Charlotte S. Wilhelm-Benartzi2, Tong F. Lioe3, Elena Sebastian1, Stephen McQuaid1, Peter W. Hamilton1, Jacqueline A. James1, Paul B. Mullan1, Mark A. Catherwood4, D. Paul Harkin1, Manuel Salto-Tellez1
Authors’ affiliation 1. Centre for Cancer Research and Cell Biology, Queen’s University Belfast, Belfast, United Kingdom 2. Epigenetics Unit, Department of Surgery and Cancer, Imperial College, London, United Kingdom 3. Department of Histopathology, Belfast City Hospital, Belfast, United Kingdom This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/his.12398 This article is protected by copyright. All rights reserved.
Accepted Article
4. Department of Haematology, Belfast City Hospital, Belfast, United Kingdom
Corresponding author: Dr David P Boyle Centre for Cancer Research and Cell Biology Queen’s University Belfast, 97 Lisburn Road, Belfast BT9 7BL Telephone: +44(0)2890972243 Fax: +44(0)2890972776 Email: [email protected]
Disclosure/Conflict of interest None: Darragh G. McArt, Charlotte S. Wilhelm-Benartzi, Tong F. Lioe, Elena Sebastian, Peter W. Hamilton, Paul B. Mullan, Mark A. Catherwood, Manuel Salto-Tellez Scientific Director of the Northern Ireland Biobank: Jacqueline James; Deputy Director of the Northern Ireland Biobank: Stephen McQuaid; President & Managing Director, Almac Diagnostics: D. Paul Harkin
This article is protected by copyright. All rights reserved.
Accepted Article
Cancer Research UK and the Pathological Society funding: David P. Boyle; Northern Ireland Health and Social Care Research and Development Division funding: Gareth Irwin
Abstract Introduction: Utility of p53 as a prognostic assay has been elusive. We describe a novel, reproducible scoring system and assess the relationship between differential p53 IHC expression patterns, TP53 mutation status and patient outcomes for breast cancer. Methods and results: Tissue microarrays were used to study p53 IHC expression patterns: expression was defined as extreme positive (EP), extreme negative (EN) and intermediate patterns as non-extreme (NE). Overall survival was used to define patient outcome. A representative subgroup (n=30) displaying the various p53 immunophenotypes was analysed for TP53 hotspot mutation status (exons 4-9). Extreme expression of any type occurred in 176/288 (61%) cases. EP p53 compared to NE p53 was significantly associated (p = 0.039) with poor OS. In addition, EN p53 compared to NE p53 was associated (p = 0.059) with poor OS. Combining cases displaying either EP/EN expression better predicted OS than either pattern alone (p = 0.028). This combination immunophenotype was significant in univariate but not multivariate analysis. In subgroup analysis six substitution exon mutations were detected, all corresponding to extreme IHC phenotypes. Five missense mutations corresponded to EP staining while the nonsense mutation corresponded to EN staining. None were detected in the NE group. Conclusions: Patients with extreme p53 IHC expression have a worse OS compared to those with NE expression. Accounting for EN as well as EP p53 improves its prognostic impact. Extreme expression positively correlates with nodal stage and histological grade and
This article is protected by copyright. All rights reserved.
Accepted Article
negatively with hormone receptor status. Extreme expression may relate to specific mutational status. Keywords: Breast cancer, prognosis, p53 protein, immunohistochemistry, mutation
Introduction Extensively researched, p53 has been referred to as ‘the guardian of the genome’.1 As such it is mutated at various generally high frequencies across a range of tumour and tissue types2 not only resulting in loss of tumour suppressor activity but also possibly causing gain of function activity including increased invasive potential and poor therapeutic response3. In breast cancer TP53 is the second most commonly mutated gene with an estimated rate of 23%4. Despite an extensive research effort, characterisation of a p53 prognostic assay has been elusive in breast cancer and consequently it is not currently recommended as a tumour biomarker in routine clinical management5.
Many studies have previously focussed on p53 immunohistochemistry (IHC) overexpression alone as a marker of poor prognosis. In an attempt to explain the biological significance of p53 overexpression and its possible correlation with mutation status the significance of p53 protein absence by IHC may be overlooked. Given the undoubted inherent intra- and intertumoural heterogeneity in breast cancer when examined across a variety of assay techniques it is likely that a range of mutational and epigenetic phenomena contribute to abnormal p53 function and protein expression whether this manifests as absent or strong positive expression by IHC. Aberrant expression, whatever the cause, may then be a marker of poor prognosis.
This article is protected by copyright. All rights reserved.
Accepted Article
A previous lack of appreciation that complete absence of p53 expression may be a particular staining pattern, as opposed to part of a spectrum of normal or wildtype expressions, has prevented study of its potential significance. Indeed, IHC expression patterns and intensities may be better thought of as a measure of spectra of functioning protein status, thresholds above and below which significant impaired function has occurred, rather than a direct or at best subtotal correlate of mutation status. Negating consideration of absent IHC expression in comparisons with prognosis, associations with therapeutic response, and mutational status may in part account for conflicting reports in the literature5. Although different biomarker platforms demonstrate particular and unique qualities, the application of sequencing techniques to clinical practice requires substantial technical and financial investment. Additionally, the inactivation of TP53 via alternative aberrations resulting in similar altered functional status may be overlooked. Using a revised p53 IHC platform may help address these issues. In the current study we aim to correlate reproducible p53 expression patterns with patient outcome for breast cancer. We hypothesise that inclusion of negative expression patterns will improve prognostic significance. We record in a reproducible way abnormal p53 protein expression to include extreme positive (EP) and negative (EN) expression in breast cancer and its relation to outcome. These abnormal p53 expression patterns are together compared against non-extreme (NE) expression patterns and the relationship to patient outcome described. Materials and methods Study design and patient selection We conducted a retrospective case-control study involving a cohort of breast cancer patients with ethical approval obtained and tissue acquired through the Northern Ireland Biobank
This article is protected by copyright. All rights reserved.
Accepted Article
(NIB ref: 12-00017). A dataset was compiled for a cohort of 303 female patients with de novo breast cancer and included a range of clinical, pathological and outcome parameters (table 1). All of the patients were diagnosed and received treatment in Northern Ireland. The vast majority of resection specimens were processed and reported in one of two hospitals in the Belfast catchment area. Surgical resection of tumours was performed between September 1997 and May 2009: resection specimens comprised total or partial mastectomies with axillary node clearance. The entire cohort received anthracycline based chemotherapy with or without radiotherapy. Administration of hormone therapy and trastuzumab were directed by hormone receptor (HR) and HER2 status. None of the patients were treated neoadjuvantly. Original HR and HER2 scores were available and were used to assess representativeness of tissue microarray (TMA) cores for full face sections. Patient exclusion criteria included male sex, no anthracycline administration and past history of cancer of any type. Cases were defined as patients who died during follow-up at 60 months compared to surviving controls. The study was performed with reference to REMARK guidelines6.
Tissue microarray construction From a total of 303 identified patients, slides and paraffin blocks were retrieved for 293 patients (in 10 cases slides were available but corresponding paraffin blocks could not be located). Original haemotoxylin and eosin (H&E) sections of all of the slides were reviewed for tumour block selection. Following block location a new section was cut for H&E and the slides annotated (DPB) for TMA construction. For each selected donor block three representative areas were annotated for targeted coring. The TMA was constructed using a manual tissue arrayer (Beecher Instruments, Silver Spring, MD) as indicated before7. The
This article is protected by copyright. All rights reserved.
Accepted Article
manual arrayer was used to extract 1mm diameter tissue cores from donor blocks for insertion into recipient blocks. Three cases had extremely focal tumour and were not included in the TMA yielding a final total of 290 individual breast cancer cases.
Immunohistochemistry and haemotoxylin and eosin All IHC was performed in a hybrid laboratory (Northern Ireland Molecular Pathology Laboratory) which has UK Clinical Pathology Accreditation, and the infrastructure to process both clinical patient samples and research materials. Sections were cut from the TMA blocks for H&E and IHC. The initial section was used for H&E staining to assess TMA quality and appropriate tumour content for subsequent IHC localisation and analysis. Sections for IHC were cut at 4 microns on a rotary microtome, dried at 37oC overnight and then used for IHC tests which were performed on an automated immunostainer (Leica BOND-MAX™). Validated and optimised protocols, used in local diagnostics, were selected for each biomarker with inclusion of carefully selected control tissues during antibody application. Antigen binding sites were detected with a polymer based detection system (Bond cat no. DS 9800). All sections were visualized with DAB, counterstained in haematoxylin and mounted in DPX. Biomarker conditions were as follows: p53 (clone DO-7 mouse monoclonal antibody, Dako) was used at a 1:100 dilution with epitope retrieval solution 1 pre-treatment for 30 minutes. ER (clone 6F11 mouse monoclonal antibody, Leica) was used at a 1:200 dilution with epitope retrieval solution 2 pre-treatment for 30 minutes. PR (clone PgR 636 mouse monocloncal antibody, Dako) was used at a 1:150 dilution with epitope retrieval solution 1 pre-treatment for 20 minutes. HER2 (clone CB11 mouse monoclonal antibody, Leica) was used at the manufacturer’s pre-set optimum dilution using epitope retrieval solution 1 pre-treatment for 25 minutes. Control tissues included breast carcinoma sections
This article is protected by copyright. All rights reserved.
Accepted Article
known to be positive for ER, PR and p53 and cell line blots provided by the manufacturer for HER2.
HER2 dual in situ hybridisation Sections for Dual Hapten, Dual Colour In Situ Hybridization (DDISH) were cut at 4 microns on a rotary microtome and dried overnight at 37oC. DDISH was performed on a fully automated Ventana BENCHMARK® XT. Following dewaxing, sections were incubated in Protease 3 (Ventana Cat. No. 760-2020) for 12 minutes, denatured and then hybridized for 6 hours with an inform HER2 Dual ISH DNA probe cocktail (Ventana Cat. No. 800-4422). Ultraview SISH DNP detection kit (Ventana Cat. No. 800-098) was used to detect HER2 signals and Ultraview red ISH DIG detection kit (Ventana Cat. No. 800-505) was used to detect chromosome 17 signals. Sections were counterstained in haematoxylin and mounted in DPX.
Scoring and assessment Only cores with identifiable tumour as confirmed by pathology assessment of H&E slides were used in IHC analysis. All IHC was scored independently by at least 2 histopathologists (DPB and JJ or TFL) blinded to patient clinicopathological and outcome data. ER and PR expression was assessed using the quick (Allred) score method8. Only nuclear expression was considered in interpretation. Provided a proportional score of at least 2 was obtained, total quick scores of 3 or more were considered positive. This threshold is equivalent to at least 1% of cells demonstrating nuclear expression of any intensity9.
This article is protected by copyright. All rights reserved.
Accepted Article
Original ER, PR and HER2 scores were available for 287, 252, and 265 cases respectively. Scores generated from TMA assessment were compared to original clinically assigned scores to assess concordance and representativeness of the TMA. Additionally, a random subset of full sections from 60 cases used in the TMA had p53 IHC applied to assess concordance and representativeness of the TMA. Analysis of HER2 expression was based on the current USA/UK guidelines.10,11 Only membranous expression was considered in interpretation. Four scores (0 to 3+) are possible. A score of 3+ was considered as indicative of HER2 overexpression (positive) and scores of 0 and 1+ were considered negative. Tumours indeterminate for HER2 overexpression were assigned a score of 2+. These prompted application of HER2 DDISH to determine HER2 status. Assessment was performed based on UK clinical guidance.12
For p53 IHC a simple 3 tier scoring metric was applied wherein complete confluent negativity of staining was considered EN, strong diffuse confluent positivity was considered EP and all intermediate expression of any intensity considered NE (figure 1). Only one of the cores per case was required to show extreme expression to be considered as such. This approach approximates to a focality of expression status previously described for a different tissue type.13 If cases showed different extremes of expression then the final expression status was assigned according to the predominant expression pattern. If NE expression was present then all replicate cores were required to show NE expression to be considered as such. To be classified as EP, the nuclei of cells were stained very dark brown to black (i.e. moderate to strong expression) and generally had a “painted-on” appearance. Almost all of the cells expressed p53 when this degree of staining was apparent. At the other end of the spectrum, to be classified as EN, the nuclei of cells appeared blue due to visible counterstain as the
This article is protected by copyright. All rights reserved.
Accepted Article
predominant pattern. When assessable, normal lobular epithelium or stromal cells were checked to rule out false negativity: these tissues showed faint p53 expression serving as an internal control against which EN expression could be confirmed. All expression patterns between these extremes prompted interpretation as NE. When present, very low NE expression generally involved the majority of visualised cells. When disagreement on staining interpretation was detected the relevant cores were re-reviewed by 2 pathologists on a multihead microscope to reach consensus opinion.
Breast cancer subtypes According to the results of biomarker expression, breast cancers were classified as HR positive (ER+ and/or PR+, HER2-), HER2 positive (ER+/- and PR+/-, HER2+) or triple negative (TNBC) (ER-, PR-, HER2-). This categorisation was termed biomarker profile A. An alternative categorisation (biomarker profile B) dividing the cohort into luminal (ER+ and/or PR+) and non-luminal (ER- and PR-) subgroups was also applied. These categories were then used to analyse the subtype specific expression levels of p53.
Histopathology Grading and tumour morphological subtypes were derived from the original pathological information. Grade was confirmed on whole face assessment using the method described by Elston and Ellis.14 Tumour size was derived from macroscopic assessment as recorded in original reports with thresholds at less than 2cm and greater than 5cm.15
This article is protected by copyright. All rights reserved.
Accepted Article
Mutation analysis DNA extracted from 30 cases from whole face tumour sections was available for TP53 mutation analysis. These comprised equal proportions of cases demonstrating the EN, NE and EP phenotypes. Four x 10μm sections were cut for each case, deparaffinised in xylene, cleared with absolute ethanol and macrodissected according to tumour annotations (DPB). Standard procedures were used according to the manufacturer’s protocol for DNA extraction (QIAamp® DNA FFPE Tissue Kit, QIAGEN®) except that incubation at 56oC was extended to 3 days to optimise purification of genomic DNA. Quantification of DNA was performed by an absorbance method using the NanoDrop 2000c spectrophotometer.
Exons 4-9 were amplified and sequenced since they contain the most common TP53 mutations in breast cancer.16 Exons were amplified with primers as previously published by Gonzalez et al.17 For each exon, 100ng of DNA was amplified by PCR in a 25µL reaction containing 3mM MgCl2, 200µM dNTP, 1 U Platinum Taq (Invitrogen, UK) and 10 µM of each primer.17 Cycling conditions were as follows: 94oC for 10 minutes, 40 cycles at 94oC for 1 minute, 55oC for 1 minute and 72oC for 1 minute, and a final extension at 72oC for 10 minutes.
PCR products were directly sequenced by Sanger sequencing with Big Dye terminators v3.1 (Applied Biosystems, UK) using both forward and reverse primers. Mutations were confirmed by amplifying a duplicated PCR product. Polymorphism status and the confirmation of mutations were assessed with reference to The IARC TP53 database.16
This article is protected by copyright. All rights reserved.
Accepted Article
Statistical analytic methods Correlation (Kendall) and Fisher’s exact test were performed using the R statistical package. Results were considered statistically significant if p-values under p=50 = 1,
