Background: Endocrine therapy (ET) is an essential tool for the treatment of the ER-positive (ER+) subtype of breast cancer (BC). However, about 40% of patients develop acquired endocrine therapy resistance (ETR). This leads to relapse and, potentially, the development of metastatic disease, which accounts for the majority of BC-related deaths.

Abstract Background: ER+/HER2+ accounts for up to 10% of all breast cancers (BCs) and most are treated with endocrine therapy (ET) after surgery to reduce the recurrence risk. We developed and validated an immunohistochemistry (IHC) based test (EA2Clin) that incorporates baseline IL6ST, clinical variables and on-treatment measurement of MCM4. Responders (Rs) and non-responders (NRs) to ET are identified and it accurately estimates recurrence-free survival (RFS) and BC-specific overall survival (BCSS). The aim was to determine if EA2Clin could accurately predict ER+/HER2+ patients likely to benefit from ET and to determine if it can identify those for whom HER2-targeted therapies are required. Methods: 3 cohorts were studied: A: 32 post-menopausal women (PMW) with large ER+/HER2+ BC treated with neoadjuvant (3-6 months) then adjuvant letrozole. 5 also received adjuvant chemotherapy plus Herceptin. Neoadjuvant clinical response was assessed by changes in tumour volume. Tumour core biopsies were taken at 0, 14 days and 3 months. Gene expression analysis using Illumina HT12 whole-genome beadarrays was performed on a subset (n=17) where fresh tissue was available. B: 13 PMW with ER+/HER2+ BC who were treated by surgery without neoadjuvant therapy. RNA was extracted from excision tissues and analysed using whole-genome Affymetrix U133A microarrays. C: 15 PMW with ER+/HER2+ BC treated with 2-weeks of pre-operative letrozole (n=7) or anastrozole (n=8). All received adjuvant letrozole. Tissues were collected at pre-treatment and at surgery. None received Herceptin or chemotherapy. All patients were followed-up after surgery (median follow-up = 6.4 years). Results: In cohort A, half (16/32) of the patients responded to ET with tumour volume reductions of >70% with neoadjuvant treatment. Innate resistance was apparent in 3 patients with continued tumour growth on ET, whereas 13 patients developed resistance after a period of response. EAClin2 predicted neoadjuvant response with a 92% accuracy. There was increased expression of phospho-AKT and phospho-ERK in NRs, not seen in Rs. Half (8/16) of the NR cancers expressed phospho-ER; but was not seen in any responsive cancer. Gene expression analysis in 17 patients showed increased MAPK and PI3K pathway activity in the 9 NR compared with the 8 R tumours. These results were recapitulated in cohort B where MAPK and PI3K activity were associated with low levels of IL6ST. In the 16/32 patients who responded well to neoadjuvant ET the actuarial recurrence rate was 0% at 5 and 10 years. The rate of recurrence in the NR was 30% at both 5 and 10 years. Of the 5 patients who received chemotherapy plus Herceptin, none recurred despite a poor response to neoadjuvant letrozole (median length to last follow-up was 6.1 years). Initial data suggest that in cohort B EA2Clin identifies a group of ER+/HER2+ cancers that can be managed by ET alone. Conclusions: · The EA2Clin test identifies ER+/HER2+ BCs who respond well to ET alone and those with a poor clinical response who have higher risk of recurrence. · NR to ET have increased expression of PI3K and MAPK pathways, consistent with active HER2 signalling. · There is potential role for EA2Clin in selecting ER+/HER2+ patients that require and benefit from HER2-targeted therapies. Citation Format: Turnbull AK, Webber V, McStay D, Arthur L, Martinez-Perez C, Fernando A, Renshaw L, Keys J, Clarke R, Sims AH, Dixon JM. Predicting benefit from HER2-targeted therapies in patients with ER+/HER2+ breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-10-26.

Abstract Background: The risk of recurrence for oestrogen receptor positive (ER+) breast cancer patients treated with 5 years of adjuvant endocrine therapy persists for many years or even decades following surgery and apparently successful adjuvant therapy. This period of dormancy and acquired resistance is inherently difficult to investigate. Therefore, previous efforts have been limited to in vitro or in vivo approaches. In this study sequential tumour samples from patients receiving extended neoadjuvant endocrine treatment were characterised as a novel clinical model of ER+ breast cancer dormancy and acquired resistance. Methods: Consecutive tumour samples from 62 patients undergoing extended (4-45 months) neoadjuvant letrozole therapy were subjected to transcriptomic and proteomic analysis, representing pre- (before treatment), early-on (13-120 days) and long-term (>120 days) neoadjuvant letrozole treatment. Patients with at least a 40% initial reduction in tumour size by 4 months of treatment were included. Of these, 42 patients with no subsequent progression were classified as “dormant”, and the remaining 20 patients as “acquired resistant”. Expression analysis was performed by using Illumina BeadChips. R and BioConductor packages were used for analysis. Differentially expressed genes were determined by using paired Rank Products (FDR, 5%). Results: Multidimensional scaling using most variant 500 genes demonstrated that long-term treated dormant samples clustered separately from their matched pre- and early-on samples whereas long-term treated resistant samples were indistinguishable from their pre-treatment counterparts. Therapy-induced changes in resistant tumours were common features of treatment, rather than being specific to resistant phenotype. Comparative analysis of long-term treated dormant and resistant tumours highlighted changes in epigenetics pathways including DNA methylation and histone acetylation. DNA methylation marks 5-methylcytosine and 5-hydroxymethylcytosine were significantly reduced in resistant tumours compared to dormant tissues after extended letrozole treatment. Decrease in 5-hydroxymethylcytosine were significant early-on. Conclusions: This is the first patient-matched gene expression study investigating long-term aromatase inhibitor-induced dormancy and acquired resistance in breast cancer. Dormant tumors exhibit distinct molecular changes under extended treatment whereas acquired resistant tumors are more similar to matched diagnostic samples supporting the molecular concordance between primary tumors and metastases. Global loss of DNA methylation was observed in resistant tumours under extended treatment which can be predicted within first 4 months of therapy. Epigenetic alterations may lead to escape from dormancy and drive acquired resistance in a subset of patients supporting a potential role for therapy targeted at these epigenetic alterations in the management of endocrine resistant breast cancer. Funding: This work was supported by Marie Skłodowska-Curie Individual Fellowship [H2020-MSCA-IF, 658170] and Welcome Trust Institutional Fund (ISSF3) to CS and AHS, Breast Cancer Now to AHS. Citation Format: Selli C, Turnbull AK, Pearce D, Fernando A, Renshaw L, Thomas JS, Dixon MJ, Sims AH. Molecular characterisation of ER+ breast cancer dormancy and acquired resistance using a clinical model: Potential involvement of epigenetic regulation [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P5-04-03.

Abstract Background: Approximately 70% of breast cancers (BCs) are estrogen receptor positive (ER+). Not all ER+ cancers respond to endocrine therapy (ET) and many eventually develop acquired resistance. Next-generation sequencing (NGS) has shown ESR1 mutations (ESRMs) are present in 10-50% of recurrent/metastatic cancers treated with aromatase inhibitors (AIs). Many of these mutations are located in the ligand-binding domain of ER, so they can lead to constitutive activation. This suggests ESRMs are a major mechanism of acquired resistance to endocrine therapy (ET) and numerous studies have shown a link between ESRMs and reduced sensitivity to 2nd line ET. The aim of this project was to investigate the incidence and clonal evolution of common ESRMs in BCs resistant to multiple sequential ETs using NGS, as well as novel PCR and in situ mutation detection methods. Methods: We have optimised an allele-specific real-time PCR (rtPCR) assay and an in situ mutation detection method (ER-ISMD) for the assessment of ESRMs. Both have been designed to identify a missense gain-of-function D538G mutation with a single nucleotide-resolution in formalin-fixed paraffin-embedded (FFPE) BC tissues. Two unique cohorts have been studied: (A) 20 post-menopausal women (PMW) with ER+ BC who acquired resistance to letrozole and were treated with up to 4 subsequent lines of ET. Serial RNA and DNA from 3-5 cancer samples per patient (58 samples from 20 patients) were analysed by Ribo0-RNAseq, DNA exome sequencing, rtPCR and ER-ISMD. (B) 150 PMW with ER+ BC who developed local (n=79), lymph node (n=59) or distant (n=12) recurrences on 1st line adjuvant letrozole, anastrozole or tamoxifen. Of these, 48 patients developed subsequent recurrences on 2nd line ET. Tissue samples from each recurrence and matched primary BC were collected. Results: In cohort A, 5/20 patients (20%) had expansion of a D538G ESR1 mutation clone at time of resistance 1st line ET (3:letrozole, 1:anastrozole, 1:tamoxifen). The mutant allele frequency (MAF) increased further in the 4 BCs treated with 2nd line ET (2:tamoxifen, 2:exemestane) and further still in the 1 BC who received 3rd line exemestane. 0/6 patients with ESRM responded to subsequent ET. Allele-specific rtPCR and ER-ISMD have been used to validate these findings and also identified low frequency ESRM clones in the sequential samples prior to the development of clinical resistance, that were not reported by NGS. Both methods have also been applied to screen tissues from patients in cohort B, where ESRMs have also been identified in recurrent samples. Complete analysis is currently ongoing. Conclusions: · ESRMs develop and expand in some BCs as a mechanism for acquired resistance to ET and are associated with a lack of response to subsequent standard ETs. · Allele-specific rtPCR can detect ESRMs and is more cost-effective and easier to use than NGS for ER mutation analysis. · Some ESRMs predate clinical resistance. · ER-ISMD is a novel approach that allows for identification and visualisation of the distribution of mutant clones in morphologically intact FFPE tissue. · ER-ISMD has the potential to become a clinically useful tool to help direct the use of 2nd line ET in routine care. Citation Format: Martinez-Perez C, Turnbull AK, Tanioka M, Fernando A, Renshaw L, Keys J, Wheless A, Garrett A, Parker J, He X, Sims AH, Carey LA, Perou CM, Dixon JM. Tracking ESR1 mutation clonal evolution in breast cancer using in situ mutation detection [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P5-04-14.

Background: Around 70% of all breast cancers (BCs) are estrogen receptor positive (ER+), but some do not respond to endocrine therapy (ET) and many eventually develop resistance. ESR amplification (ESRA) linked to an increase in ESR1 gene expression is known to occur in some cancers that are endocrine resistant. However, the incidence of ESRA has been the object of debate and its clinical significance remains unclear. This study aimed to investigate the incidence of ESRA in BCs resistant to multiple sequential ETs and optimise a fluorescence in-situ hybridisation (FISH) methodology to robustly detect ESRA. Methods: Two unique cohorts have been studied: (A) 20 post-menopausal women with ER+ BC with acquired resistance to letrozole, subsequently treated with up to 4 different lines of ET. Serial RNA and DNA from 3-5 cancer samples per patient (58 samples from 20 patients) were analysed by Ribo0-RNAseq and DNA exome sequencing; (B) 18 post-menopausal women who developed ER+ BC recurrences on 1st line adjuvant letrozole, then on 2nd line tamoxifen and subsequently on 3rd line exemestane. Tissues were collected at the time of each surgery. We have optimised a FISH method to assess ESRA in these tissues. Results: In cohort A, 6/20 patients developed ESR1 gene amplification (ESRA) at some point during treatment. In 5 of these cases, ESRA was only found while on 2nd or 3rd line exemestane but was not present on acquired resistance to previous letrozole or tamoxifen. 1 patient had ESRA at the time of first recurrence on letrozole. The FISH method showed concordance with the genomic analysis. This suggests that ESRA may be associated with BCs that are treated with and then become resistant to exemestane. ESRA is also evident in samples from Cohort B, which includes 18 exemestane resistant cases. The complete analysis is ongoing. Conclusions: · ESRA can be seen in ER+ recurrent BCs. · ESRA may be associated with BCs treated with 2nd or 3rd line exemestane. · The frequency of ESRA in endocrine and exemestane resistance can now be ascertained using an optimised FISH-based method, which is more cost-effective than alternative genomic and biochemical methods. Citation Format: Turnbull AK, Martinez-Perez C, Mok S, Tanioka M, Fernando A, Renshaw L, Keys J, Wheless A, Garrett A, Parker J, He X, Sims AH, Carey LA, Perou CM, Dixon JM. Investigating the incidence of ESR1 gene amplification in breast cancers resistant to multiple endocrine agents [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P5-04-27.

Abstract Background: The majority of patients with early-stage estrogen receptor positive (ER+) breast cancer (BC) are treated with adjuvant endocrine therapy (ET) after surgery to reduce the risk of recurrence. Recently, we have developed and validated an immunohistochemistry (IHC) based assay (EndoAdjuvant2 Clinical; EA2Clin) that measures pre-treatment IL6ST level together with clinical variables and on-treatment MCM4 to assess proliferation. We have previously shown that it can accurately identify responders and non-responders to ET and predicts recurrence-free survival (RFS) and BC-specific overall survival (BCSS). We postulated that measuring on-treatment proliferation in lymph node metastases (LN+) rather in the primary cancer might further improve the accuracy of the test for these patients. The aim was to test and validate this in cohorts of pre- and post-menopausal women (preMW & PMW) treated with preoperative ET (tamoxifen (T), fulvestrant (F), letrozole (L) or anastrozole (A)) and subsequent adjuvant ET. Methods: Cohorts: (1) 137 PMW with ER+ BC, 59 were LN+, treated with neoadjuvant L (median duration 4.8 months, range 1-33), then surgery followed by adjuvant L (n=109) or other ET (n=28); (2) 148 PMW with ER+ BC, 55 were LN+, treated with 2 weeks of preoperative L (n=76) or A (n=72), then surgery followed by adjuvant L (n=69) or T (n=79); (3) 52 preMW with ER+ BC, 24 were LN+, treated with 2 weeks of preoperative T (n=26) or 1x750mg dose of F (n=26), then surgery followed by adjuvant T. All LN+ patients had sentinel node biopsies or clearance. The median follow-up was 6.5 years (cohort 1), 6.3 years (cohort 2) and 10.2 years (cohort 3). EA2Clin: Patients are classified as: · Low risk: ER+ and LN-negative and <2cm or pre-treatment IL6ST 2+/3+ (IHC) and post-treatment MCM4 in the primary has <20% positive nuclear staining. · High risk: ER+ LN+ grade 3 BCs >2cm or pre-treatment IL6ST is 0 or 1+, or IL6ST is 2+ or 3+ and MCM4 in the primary has >10% positive nuclear staining. EA2CliN uses the post-treatment level of MCM4 in the nodes, rather than the primary cancer. Results: In cohort 1, EA2Clin (using primary tumour MCM4) was significantly associated with both RFS (P=0.0003, HR=13.17, 95%CI=5.48-13.61) and BCSS (P=0.005, HR=11.91, 95%CI=8.73-31.42). The 5 and 10 year actuarial recurrence rates were 5%/5% and 48%/64% for the low and high-risk groups respectively. In the same cohort, using the MCM4 level in the node (EA2CliN) there was an even more significant association with both RFS (P<0.00009, HR=18.16, 95%CI=12.59-19.46) and BCSS (P=0.002, HR=12.93, 95%CI=5.43-25.62). The 5 and 10 year actuarial recurrence rates were 0%/0% and 48%/72% for the low and high-risk groups respectively. Further validation of EA2CliN in cohorts 2 and 3 is underway. Discussion: · Direct measurement of on-treatment proliferation biomarkers in LN metastases improves prediction of outcomes to ET in women with BC. · This tests identifies a group of low risk women that are node negative and node positive with a 100% RFS and BCSS. · This is the most impressive predictive test for patients with ER+ breast cancer yet developed. Citation Format: Turnbull AK, Mok S, Martinez-Perez C, Fernando A, Renshaw L, Keys J, Sims AH, Dixon JM. Measurement of on-treatment proliferation biomarkers in nodal metastasis improves prediction of endocrine therapy response using the EA2CliN test [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P5-11-03.

Abstract Background: Transcriptomic analyses of clinical samples can help improve our understanding of disease aetiology, drug effectiveness, assign molecular subtypes and derive prognostic signatures for clinical decision-making. The success of early microarray studies relied heavily on sample quality and predominantly fresh frozen (FF) tissues to generate reliably robust data. The emergence of next-generation microarray and sequencing-based technologies from formalin-fixed paraffin-embedded (FFPE) tissues provides an opportunity to study archival clinical tissues with long-term follow-up. Here we assess 9 technologies, which vary in resolution, cost and RNA requirements, with matched FF and FFPE tissues from the same patient. Methods: Sequential tumour biopsies were taken pre-treatment and on-treatment (at 14-days and 3-months) from 11 postmenopausal patients with oestrogen receptor positive breast cancer treated with 3 months of neoadjuvant letrozole. Half of each sample was snap frozen in liquid nitrogen and half was FFPE, RNA was extracted from both. Transcriptomic analyses were performed using 9 technologies: Illumina Beadarray, Affymetrix U133A, Affymetrix Clariom S, NanoString nCounter, AmpliSeq Transcriptome, Lexogen QuantSeq and IonXpress RNAseq, Tempo-Seq BioSpyder and Qiagen UPX3'. Results: Success rates for generating robust expression profiles from FFPE tissues were 100% all except the Illumina BeadChip (22%) and AmpliSeq Transcriptome (83%) , which varied by the age of tissue. With the total number and position of probes/primers/counts varying widely between approaches, in total 7305 genes were represented across all of the whole-genome technologies tested. Clear batch effects were evident when comparing data from FF and FFPE tissues and when comparing between different technologies. Standard batch correction approaches such as XPN and ComBat minimised technical bias effect and increased the correlations between matched samples (FF and FFPE) to R>0.9, irrespective of the technology used. When analysed by multi-dimensional scaling following batch correction, samples clustered by treatment time-point. When ranked by expression of 60 proliferation genes, reported by us to change with letrozole treatment, samples ordered again by time-point, consistent with our previous findings, and paired samples clustered together. Conclusions: · Robust gene expression profiles can be reliably generated from FFPE tissues and are comparable to those derived from FF tissue using established transcriptomic approaches. · A range of new technologies are available for the study of FFPE tissues; these vary in cost, resolution and RNA requirements to fit the user's needs. · Gene expression data from biologically similar studies, generated using different technologies, can be reliably integrated for robust meta-analysis, subject to appropriate batch correction analysis. Citation Format: Turnbull AK, Selli C, Martinez-Perez C, Fernando A, Renshaw L, Keys J, Figueroa JD, He X, Tanioka M, Munro A, Murphy L, Fawkes A, Clark R, Coutts A, Perou CM, Carey LA, Dixon JM, Sims AH. Unlocking the transcriptomic potential of formalin-fixed paraffin embedded breast cancer tissues for high-throughput genomic analysis [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-06-17.

Abstract Background: The majority of patients with early-stage estrogen receptor positive (ER+) breast cancer (BC) are treated with adjuvant endocrine therapy (ET) after primary surgery to reduce the risk of recurrence. A variety of tests are available to predict outcomes on ET but most require gene-level measurements and are expensive. Recently, we developed an immunohistochemistry (IHC) based test (EA2Clin) using levels of pre-treatment IL6ST together with clinical variables and on-treatment proliferation. The aim was to validate this test in cohorts of both pre- and post-menopausal women treated with two weeks of a variety of endocrine treatments (tamoxifen, fulvestrant or an aromatase inhibitor) prior to surgery. Methods: The cohorts are: (A) 186 post-menopausal women (PMW) with ER+ BC treated with at least 2 weeks of preoperative or neoadjuvant letrozole or anastrozole, then surgery followed by adjuvant letrozole (n=132) or tamoxifen (n=54); (B) 51 pre-menopausal women (preMW) with ER+ BC treated with 2 weeks of either neoadjuvant tamoxifen (n=24) or one 750mg dose of faslodex (n=27), then surgery followed by adjuvant tamoxifen. The median follow-up was 5.4 years for cohort A and 10.2 years for cohort B. IHC analysis was performed using a Leica BOND III autostainer and the EA2Clin algorithm was used to stratify patients in binary high or low-risk groups. Results: In the cohort of PMW, EA2Clin was highly significantly associated with both recurrence-free survival (RFS) (P<0.0001, HR=13.26, 95%CI=5.59-13.46) and breast cancer specific survival (BCSS) (P<0.0001, HR=12.93, 95%CI=4.43-37.72). The 5 and 10 year actuarial recurrence rates were 7%/22% and 46%/73% for the low and high risk groups, respectively. The actuarial breast cancer-related death rate for the low risk group was 5% at both 5 and 10 years, whereas for the high risk group was 33%/38%. Confounding factors were not found to be significant. In the cohort of preMW, our test was significantly associated with both RFS (P=0.002, HR=5.71, 95%CI=1.91-17.05) and BCSS (P=0.016, HR=4.81, 95%CI=1.34-17.26). The 5 and 10 year actuarial recurrence rates were 12%/29% and 27%/77% for the low and high risk groups, respectively. The 5 and 10 year actuarial breast cancer-related death rates were 7%/19% and 9%/58% for low and high risk groups, respectively. Discussion: · This study has validated EA2Clin as the first predictive tool to incorporate clinical data with pre and on-treatment immunohistochemical biomarkers to predict accurately the outcome of patients with ER positive breast cancer treated with adjuvant ET. · This test predicts both RFS and BCSS in pre- and PMW treated with a variety of endocrine agents. · Because this test incorporates clinical variables with simple IHC, it can be performed locally in any pathology lab. Citation Format: Turnbull AK, Fernando A, Renshaw L, Keys J, Thomas JS, Sims AH, Dixon JM. EA2Clin: A novel immunohistochemical prognostic and predictive test for patients with estrogen receptor-Positive breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P4-08-03.

Abstract Background:Late recurrence (emergence from dormancy) is characteristic of ER+ breast cancers. Despite adjuvant endocrine therapy, many breast cancers recur decades after their initial diagnosis and treatment. Why this occurs is poorly understood. Methods: We studied 2 independent datasets of endocrine treated, ER+ breast cancers with up to 20 years follow-up. The 1st comprised matched samples from the primary tumor pretreatment at diagnosis and the first recurrence after or during adjuvant endocrine therapy (all FFPE). The 2nd dataset comprised pretreatment biopsies only (all snap frozen). For both datasets, high quality RNA was amplified, labelled, and subjected to transcriptome analysis using the Affymetrix technology (U133 Plus 2.0). Low quality data were identified using 'simpleaffy' and 'ffpe', and removed; all tools were from the R package unless otherwise noted. Remaining data were normalized using 'frma'. Genes differentially expressed between early (≤3 years) and late (≥ 5 years) were selected using limma. Unsupervised hierarchical clustering and PCA explored the structure of the data. A similar molecular analysis was done on the 2nd dataset. A classification scheme that robustly separated early from late recurrences was validated in an independent public dataset of comparable patients, array platform, and frozen tissues. We also explored features in pretreatment samples that predetermined response duration. Results: Genes that separated pretreatment specimens by recurrence time did not separate posttreatment specimens. Specimens did not cluster in patient pairs or by site of recurrence. 8245 genes were differentially expressed between early and late recurrences in the FFPE samples, while 2400 genes were significantly different in the same comparison in the frozen samples. Initial pathway analysis was done on each dataset independently using IPA (Ingenuity® Systems, www.ingenuity.com). 70 canonical pathways were identified in common between the two datasets (pretreatment). We then looked for genes regulated in both datasets (ignores FFPE and frozen tissue as source). There were 279 genes in common that differentially regulated in the same direction (upregulated; downregulated). IPA analysis of these genes identified 49 canonical pathways. We also explored the differentially expressed gene sets using 'GSEA' (www.software.broadinstitute.org/gsea/index.jsp). Pathways consistently associated with early vs. late recurrence include integrin signaling, the unfolded protein response, endoplasmic reticulum stress, actin-based motility, and estrogen biosynthesis. Conclusion: Analysis of pretreatment tumors can predict early recurrences from those that will remain dormant and recur much later. Recurrent tumors exhibit a remodeled molecular landscape that likely reflects the effects of treatments and/or a recreation of a niche with potentially common features at the site of recurrence. Changes in molecular signaling associated with duration of recurrence are consistent with our experimental model studies in vitro implicating UPR signaling as a major integrator of response to endocrine therapy and duration of survival. Additional data sets are being arrayed and more detailed molecular signaling studies are in progress. Citation Format: Clarke R, Dixon MJ, Jin L, Turnbull A, Hu R, Zwart A, Wang Y, Xuan J, Sengupta S, Renshaw L, Sims A, Liu MC. Molecular features of dormancy in ER+ breast cancers [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P4-04-10.

Abstract Background Identifying breast cancer (BC) patients likely to recur on endocrine therapy (ET) is a challenge. Several methods and tests based on clinical parameters and multi-gene or protein classifiers have been shown to predict those likely to recur. Tests that incorporate baseline and on-treatment markers are likely to be more accurate than tests based on baseline characteristics alone. 4 genes were identified by microarray that predicted for to ET: 2 at diagnosis and 2 at 14 days. The EndoAdjuvant2 (EA2) test uses 2 of these genes: IL6ST at diagnosis and on-treatment MCM4 at transcript level or by graded immunohistochemistry (IHC). The aim of this study was to compare EA2 with currently used clinical parameters in 4 different cohorts of pre and postmenopausal women. Patients The cohorts are (1) 73 post-menopausal women (PMW) with ER+ BC treated with neoadjuvant letrozole (L) then surgery, (2) 39 PMW with ER+ BC treated with neoadjuvant anastrozole (A) then surgery, (3) 108 PMW who received 2-weeks of A or L prior to surgery (4) 25 preMW with ER+ BC who received one 750mg dose of fulvestrant prior to surgery. All had adjuvant ET and 5-10 years follow up. Neoadjuvant response was assessed by periodic 3D ultrasound. Results The 4 gene assay had 96% (training; cohort 1) and 93% (validation; cohort 2) accuracy of response prediction to neoadjuvant L or A. In cohort 1, clinical parameters were available. On univariate regression analysis intrinsic subtype (luminal A/B; defined using PAM50) (P=0.002), histological grade (P=0.033) and HER2 status (P=0.001) significantly predicted clinical response. EA2 out-performed all these in both univariate (P<0.0001) and multivariate regression analysis (P<0.0001). The final model only retained EA2 as significant. Node status, tumour size and PR expression were not associated with response to endocrine therapy. EA2 predicted recurrence free in cohorts 1 and 3 combined: RFS (P=0.0004, HR=17.63 95%CI: 4.95-17.6), BCSS (P=0.0007, HR=16.60: 3.36-45.7). The Nottingham Prognostic Index (NPI) also predicted RFS (P=0.0002) and BCSS (P=0.0017) in univariate analysis but in the Cox analysis NPI was not found to be significant, although in the low risk group there were only 1/46 events compared to 19/62 in the moderate/high risk group. Histological grade (P<0.0001), Ki67 (%) (P=0.02) and tumour size (P=0.007) were significant on univariate analysis. In the Cox regression analysis, only EA2 (P=0.012) and histological grade (P=0.016) were significant predictors of RFS and BCSS. Using EA2 and NPI together a high and low risk subgroup could be identified. In the Cox regression analysis, combined EA2 plus NPI (EA2clin) outperformed both NPI and EA2 alone. EA2clin was validated in cohort 4 and predicted RFS (P=0.002, HR=8.43 95%CI: 2.18-32.53) and BCSS (P=0.008, HR=10.95 95%CI: 1.48-64.9). Conclusions • EA2 predicts clinical response, RFS and BCSS • EA2clin combines NPI and EA2, outperforms either alone and predicts outcome in a new validation cohort of preMW treated with Fulvestrant • EA2clin works in preM and PM women regardless of ET. Citation Format: Turnbull AK, Pearce DA, Arthur LM, Martinez-Perez C, Thomas JS, Fernando A, Renshaw L, Sims AH, Dixon JM. The EA2clin test significantly predicts response and survival in both pre and post-menopausal women with ER-positive breast cancers [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P6-09-27.

Abstract Introduction Currently the primary breast tumour is used for prognostic profiling and as a monitor of response to therapy but how often does the molecular profile of the primary cancer reflect the molecular profile of nodal metastases? No previous study has investigated in detail the genomic profile of matched primary breast cancer (P) and nodal metastases (N) and correlated these with outcome. The aim of this study was to investigate whether the mRNA profiles of matched P and N differ significantly. Methods RNA was extracted from core biopsies from primary breast tumours and paired metastatic axillary lymph node samples from both FFPE blocks and fresh frozen samples. RNA was labelled and hybridised to Illumina HT-12 BeadChips to create a dataset consisting of one primary and one or two matched nodal metastasis, totalling 68 samples from 31 patients. Data was processed and corrected for batch effects, then analysed using the statistical programming language R. Clinical data on progression free and overall survival was collected from electronic and medical case note review. Results Unsupervised hierarchical clustering of the 500 most variable genes in each sample grouped only 12 of 31 P&Ns (39%) together, meaning in the majority of patients their P or N more resembled a cancer from another patient than its own paired P or N. The number of genes with greater than 2 fold change (>2FC) between P&N was used to categorise paired samples into 'least changed' (<130 genes with >2 FC) and 'most changed' (>370 genes with >2FC) groups. Multidimensional scaling of the 500 most variable genes in the most changed group (n-=10) showed consistently that nodal metastases differed molecularly from the primary cancer. When categorised by Sorlie centroid, 12 of 31 patients (39%) had a different molecular subtype in N compared with P. N tended to be a poorer prognostic subtype than P. 50% had luminal A primaries paired with luminal B nodes. The remaining 50% changed in other non-consistent patterns. 6 patients had 2 N samples to analyse alongside P. 4 of these (67%) had the same subtype in all 3 samples, and a further 1 the same 2Ns (luminal B) which differed from P (luminal A). The final had luminal A P paired with 1 luminal A and 1 luminal B Ns. There was no evident correlation between the least changed and most changed groups and progression free and overall survival. This may however reflect the short term follow up. Discordance between P and N in expression of ESR1 was 32%; PGR 19% and ERBB2 16%. Conclusions This study of gene expression change in matched primary breast cancers and synchronous metastatic paired axillary lymph nodes shows that molecular subtype differs in 39%. 50% of nodes had a poorer prognostic subtype than their primary. Expression of ESR, PGR and ERBB2 differs in up to 32% Classifying cancer molecular phenotype and estimating prognosis based only on the primary cancer misclassifies significant numbers of patients. Classification of prognosis, and treatment based on the nodal metastasis may provide better information on which to base treatment. Citation Format: Arthur LM, Turnbull AK, Pearce DA, Renshaw L, Thomas JS, Sims AH, Dixon JM. Molecular characterisation, subtype concordance and prognostic group assignment between patient-matched primary breast tumours and axillary lymph node metastases [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-06-04.

Background: Aromatase inhibitors (AIs) have an established role in the treatment of estrogen receptor alpha positive post-menopausal breast cancer. Recently we have developed and validated a microarray-derived 4-gene test (Edinburgh EndoResponse4) to predict response to AIs in the neoadjuvant setting. We have also demonstrated the translational potential of this test in predicting accurately clinical response when mRNA is measured for these genes by polymerase chain reaction (PCR) or the gene protein is measured by immunohistochemistry (IHC). There is a major clinical need for biomarkers to predict which patients are likely to recur on adjuvant endocrine therapy so alternative or additional treatments can be provided to reduce recurrence and improve outcome. The aim of this study was to determine if Endoresponse4 and IHC of these gene proteins could do this. Methods: The original microarray assay used pre- and on-treatment (14-days) biopsies from 73 post-menopausal women with ER-rich breast cancer receiving 3 months of neoadjuvant letrozole prior to surgery with 10 years follow-up after adjuvant letrozole. Matched formalin-fixed paraffin embedded (FFPE) tissue sections from 42 of these patients were used for IHC and antibodies were optimised against 3 of the 4 proteins (where validated antibodies were available) using Envision technology. The ability of our test to estimate recurrence-free (RFS) and breast cancer specific overall survival (OS) using both PCR and IHC was then tested in a unique validation cohort of 140 post-menopausal women with ER-rich breast cancer treated with 2 weeks of neoadjuvant letrozole or anastrozole prior to surgery followed by adjuvant endocrine therapy and 10 years of follow up.. Results: Within our training cohort (n=73) using Kaplan-Meier analysis our 4-gene test predicted neoadjuvant clinical response and demonstrated a significant association with both RFS (P=0.029) and OS (P=0.009). This approach predicts outcomes within 2-weeks rather than 4-months of treatment required in other studies such as P024. Using IHC in the training cohort (n=42), two gene markers in combination (IL6ST at diagnosis and MCM4 after 2-weeks treatment) predicted both RFS (P=0.017) and OS (P=0.009) with great accuracy. The 140 patient group is being analysed and the findings are so far are consistent with the initial training cohort and indicate a significant association with outcomes. Conclusion: • A 4 gene model with clinical potential has been developed and validated to predict response to neoadjuvant aromatase inhibitors. • This 4 gene model predicts for response and also predicts relapse free and overall survival. • Proteins encoded by 2 of these 4 genes measured by IHC in an initial test set of 42 patients predict accurately both EFS and OS • A validation cohort (n=140) with over 10-years of follow-up will be available at SABCS 2015 to determine if this 2 biomarker test can predict outcome on adjuvant endocrine therapy. Citation Format: Turnbull AK, Arthur LM, Webber V, Thomas J, Uddin S, Webb H, Dunbier A, Dowsett M, Renshaw L, Sims AH, Dixon JM. A validated test for neoadjuvant clinical response to endocrine therapy in breast cancer that estimates accurately recurrence-free and overall survival. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P3-07-20.

The phosphatidylinositol-3-kinase pathway plays an important role in proliferation, migration and survival in breast cancer and may play a role in resistance to endocrine therapy. Pathway activation occurs as a result of mutations in PIK3CA or loss of functional PTEN. Matched primary and recurrent samples from 120 breast cancer patients treated with endocrine therapy were profiled with a qPCR-based mutation assay covering eight mutational hotspots in PIK3CA. PTEN was assayed by immunohistochemistry. Samples were well characterized with respect to anatomic location of recurrence (metastatic nodal or local recurrence as opposed to contralateral or ipsilateral new primary cancers). In total, 43 % of patients had at least one PIK3CA mutation at diagnosis, and 41 % had a mutation at the time of recurrence. Only 8 % of patients with local recurrence, metastatic disease or progression on primary endocrine treatment changed their PIK3CA mutation status (four gains, two losses, total 76). The most common changes in PIK3CA mutation status were seen in patients who developed a new cancer either in the treated or contralateral breast (64 %, three gains, four losses, total 11). PIK3CA mutation status does not change in the majority of patients with breast cancer and the acquisition of mutations in PIK3CA is not responsible for the development of endocrine resistance. PTEN loss at diagnosis is associated with a significantly shorter time to progression compared with tumours in which PTEN was retained. These are the most comprehensive data currently available correlating PIK3CA status, site of recurrence and endocrine resistance.

Background: Invasive lobular carcinoma (ILC) is a special histological subtype accounting for 10–15% of breast cancers. ILC responds poorly to neoadjuvant chemotherapy, but appears to respond well to endocrine therapy. However, it is unclear whether the molecular changes that occur in response to letrozole differ between lobular and ductal carcinomas. Methods: Pre- and on-treatment (after 2 weeks and 3 months) biopsies were obtained from 14 post-menopausal women with ER+ histologically confirmed ILC who responded to 3 months of neoadjuvant letrozole and were compared with a cohort of 14 infiltrating ductal carcinomas (IDCs) matched on clinicopathological features. Microarray gene expression data were generated. Dynamic clinical response was assessed for each patient using periodic 3D ultrasound measurements performed during treatment. Response was defined as a reduction of >70% in tumour volume by 3 months. Results: Unsupervised analysis using the 500 most variable genes across samples at pre-treatment was able to distinguish between IDC and ILC with 89% accuracy. Supervised analysis (Rank Products, FDR = 0.01) identified 207 genes differentially expressed: 136 with significantly lower expression and 71 (including e-cadherin) with higher expression in IDC compared with ILC. The 136 genes were functionally enriched for immune and extra-cellular matrix (ECM) remodelling. Differences in gene expression between IDCs and ILCs were maintained during treatment. The most consistently changed genes (n = 377) over 3 months on letrozole from a pairwise analysis in responding IDCs were enriched for down-regulation of proliferation and up-regulation of immune/ECM remodelling genes. This pattern of changed expression was mirrored in the lobulars suggesting a similar molecular response to letrozole. Genes differentially expressed between ductal and lobular tumours were unique from those most changed in response to letrozole. Conclusions: This is the first study of molecular changes in ILC in response to endocrine therapy to date. The genes which change on letrozole are similar in ILC and IDC. Differences in gene expression can distinguish between patient-matched ILC and IDC at diagnosis and are maintained at each time point on treatment. Disclosure: All authors have declared no conflicts of interest.