BACKGROUND Gene expression (expr) analyses are increasingly used for characterising the pharmacodynamic response of primary BC. This includes assessing ER+ BC9s dependence on estrogen (E) by measuring gene expr changes after AI-treatment. However, differences in tissue sampling and other preanalytic procedures between samples taken at diagnosis (D) and surgery (S), may lead to systematic artifactual changes that are falsely ascribed to the intervention. To identify genes whose expr is truly affected by AI, we measured global gene expr changes from paired core-cut biopsies at D and S from patients in the POETIC presurgical window trial. METHODS In POETIC, 4486 postmenopausal women with primary ER+ BC were randomised 2:1 to receive perioperative AI (2 weeks pre + 2 weeks post surgery, termed Tr) or no perioperative treatment (termed Con), allowing gene expr changes to be compared between Tr and Con. RNA was extracted from paired RNA-later stored core-cuts of 56 Con and 157 Tr patients and arrayed on Illumina whole genome expr BeadChips. Raw data was extracted, transformed, normalised and batch-corrected. Probes not detected (pu003e0.01) in u003e=25% of samples were discarded. Impact of AI on genes was evaluated based on difference of the expr mean changes (log 2 (S/D)) of the Tr and Con samples. RESULTS In the Con group, expr of 73 genes significantly changed (FDR MYC increase) but was unaffected in Tr. Such artifactual gene changes in Con tumors conceal true AI-induced changes that would not be detected in the absence of comparison with Con. 615 genes were down-regulated and 472 up-regulated in Tr but not Con. The majority of down-regulated genes were cell cycle or proliferation-associated or E-regulated, including ESR1, PDZK1, GREB1, HSPB1 . Functional mapping showed changes in the regulation of cyclins and cyclin dependent kinases impacting on G1/S and G2/M. Of note, up-regulated genes included CDK6 (target for CDK4/6 inhibitors) and CCND2, involved in G1/S checkpoint regulation ; SNAI2, TGFB3, TGFBR2, associated with tumour invasion and metastasis ; and other genes involved in aryl hydrocarbon receptor, Glioblastoma Multiforme, HIPPO and p53 signalling. CONCLUSION Expr of certain genes is altered by processes involved in presurgical window studies. In the absence of a Con group, these may be wrongly ascribed to an experimental intervention or wrongly considered as unaffected by the intervention (eg MYC in this study). Down-regulation of E-responsive and proliferation genes was an expected response to AI but increased expr of genes such as SNAI2 , CCND2 and CDK6 indicates immediate tumour re-wiring and provides mechanistic support for benefit from combination therapy with a CDK4/6 inhibitor. Citation Format: Gao Q, Lopez-Knowles E, Cheang MCU, Morden J, Martin L-A, Sidhu K, Evans D, Martins V, Dodson A, Skene A, Holcombe C, Mallon E, Abigail E, Bliss J, Robertson J, Smith I, Dowsett M. True effect of aromatase inhibitor (AI) treatment on global gene expression (expr) changes in postmenopausal ER+ breast cancer (BC) patients: A POETIC study (CRUK/07/015) [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 P2-09-02.

Abstract BACKGROUND Gene expression (expr) analyses are increasingly used for characterising the pharmacodynamic response of primary BC. This includes assessing ER+ BC's dependence on estrogen (E) by measuring gene expr changes after AI-treatment. However, differences in tissue sampling and other preanalytic procedures between samples taken at diagnosis (D) and surgery (S), may lead to systematic artifactual changes that are falsely ascribed to the intervention. To identify genes whose expr is truly affected by AI, we measured global gene expr changes from paired core-cut biopsies at D and S from patients in the POETIC presurgical window trial. METHODS In POETIC, 4486 postmenopausal women with primary ER+ BC were randomised 2:1 to receive perioperative AI (2 weeks pre + 2 weeks post surgery, termed Tr) or no perioperative treatment (termed Con), allowing gene expr changes to be compared between Tr and Con. RNA was extracted from paired RNA-later stored core-cuts of 56 Con and 157 Tr patients and arrayed on Illumina whole genome expr BeadChips. Raw data was extracted, transformed, normalised and batch-corrected. Probes not detected (p>0.01) in >=25% of samples were discarded. Impact of AI on genes was evaluated based on difference of the expr mean changes (log2(S/D)) of the Tr and Con samples. RESULTS In the Con group, expr of 73 genes significantly changed (FDR<5%); 70 of these changed by a similar magnitude in the Tr group, indicating their change was independent of AI therapy but would have been artifactually discovered as changed by AI in the absence on Con. The 8 genes most up-regulated in Tr were all among the 20 genes most up-regulated in Con: many were early-response or stress-associated genes. Three of the 8 most down-regulated in AI were the most down-regulated in Con: all were haemoglobin-related. Expr of some genes was changed in Con (eg MYC increase) but was unaffected in Tr. Such artifactual gene changes in Con tumors conceal true AI-induced changes that would not be detected in the absence of comparison with Con. 615 genes were down-regulated and 472 up-regulated in Tr but not Con. The majority of down-regulated genes were cell cycle or proliferation-associated or E-regulated, including ESR1, PDZK1, GREB1, HSPB1. Functional mapping showed changes in the regulation of cyclins and cyclin dependent kinases impacting on G1/S and G2/M. Of note, up-regulated genes included CDK6 (target for CDK4/6 inhibitors) and CCND2, involved in G1/S checkpoint regulation; SNAI2, TGFB3, TGFBR2, associated with tumour invasion and metastasis; and other genes involved in aryl hydrocarbon receptor, Glioblastoma Multiforme, HIPPO and p53 signalling. CONCLUSION Expr of certain genes is altered by processes involved in presurgical window studies. In the absence of a Con group, these may be wrongly ascribed to an experimental intervention or wrongly considered as unaffected by the intervention (eg MYC in this study). Down-regulation of E-responsive and proliferation genes was an expected response to AI but increased expr of genes such as SNAI2, CCND2 and CDK6 indicates immediate tumour re-wiring and provides mechanistic support for benefit from combination therapy with a CDK4/6 inhibitor. Citation Format: Gao Q, López-Knowles E, Cheang MCU, Morden J, Martin L-A, Sidhu K, Evans D, Martins V, Dodson A, Skene A, Holcombe C, Mallon E, Abigail E, Bliss J, Robertson J, Smith I, Dowsett M. True effect of aromatase inhibitor (AI) treatment on global gene expression (expr) changes in postmenopausal ER+ breast cancer (BC) patients: A POETIC study (CRUK/07/015) [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 P2-09-02.

Abstract Aim: To assess the performance of the risk of recurrence (ROR) score in predicting RFS and DRFS for postmenopausal patients with primary ER+ breast cancer treated with anastrozole or tamoxifen and compare this with the performance of Oncotype Dx and IHC4 for both N0 and N+ patients. Background: OncotypeDx Recurrence Score (RS) is used to assess RR after endocrine therapy in primary ER+ breast cancer and is valid for both tamoxifen and anastrozole (Dowsett ref). IHC4 is a 4-panel set of IHC markers (ER, PgR, HER2, Ki67) that was shown to provide as much prognostic accuracy as RS in the translational arm of the ATAC trial (TransATAC) of anastrozole versus tamoxifen alone or combined and subsequently independently validated (Cuzick et al, JCO, 2011, in press). PAM50 is 50-gene test that has been optimised to separate intrinsic disease subtypes and is used to generate a ROR score. Good correlation has been shown between the ROR and RS but no large scale comparisons of their ability to predict clinical outcome have been conducted. Methods: PAM50 analysis was conducted using NanoString technology on 125 ng RNA extracted from all 1026 samples from the TransATAC study that had at least 500ng RNA available. Nodal status was known on 986: N+, 271; N-neg 715. ROR scores were calculated and their relationship with RFS and DRFS assessed in both the N-neg and N+ categories. The prognostic value of ROR when added to standard clinical variables was assessed using the likelihood ratio chi-square. Finally, the relative accuracy of prognosis in this patient population using ROR, RS and IHC4 was evaluated by comparing the C-index for each test. Results: Results are currently being derived and will be presented in full. Discussion: This is the first large-scale clinical comparison of 2 prominent multi-gene predictors for use in FFPE material. The results will be instrumental in defining the preferred method for use in assessing residual risk of recurrence in patients treated with endocrine therapy. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr S4-5.

PURPOSE:Risk of distant recurrence (DR) among women with estrogen receptor (ER) -positive early breast cancer is the major determinant of recommendations for or against chemotherapy. It is frequently estimated using the Oncotype DX recurrence score (RS). The PAM50 risk of recurrence (ROR) score provides an alternative approach, which also identifies intrinsic subtypes. PATIENTS AND METHODS:mRNA from 1,017 patients with ER-positive primary breast cancer treated with anastrozole or tamoxifen in the ATAC trial was assessed for ROR using the NanoString nCounter. Likelihood ratio (LR) tests and concordance indices (c indices) were used to assess the prognostic information provided beyond that of a clinical treatment score (CTS) by RS, ROR, or IHC4, an index of DR risk derived from immunohistochemical assessment of ER, progesterone receptor, human epidermal growth factor receptor 2 (HER2), and Ki67. RESULTS:ROR added significant prognostic information beyond CTS in all patients (Δ LR-χ(2) = 33.9; P < .001) and in all four subgroups: node negative, node positive, HER2 negative, and HER2 negative/node negative; more information was added by ROR than by RS. C indices in the HER2-negative/node-negative subgroup were 0.73, 0.76, and 0.78 for CTS, CTS plus RS, and CTS plus ROR, respectively. More patients were scored as high risk and fewer as intermediate risk by ROR than by RS. Relatively similar prognostic information was added by ROR and IHC4 in all patients but more by ROR in the HER2-negative/node-negative group. CONCLUSION:ROR provides more prognostic information in endocrine-treated patients with ER-positive, node-negative disease than RS, with better differentiation of intermediate- and higher-risk groups.