Citations

Abstract Background: It has become increasingly clear that effective treatment of metastatic breast cancer (MBC) requires an in-depth understanding of the molecular differences between primary tumors and metastases. The AURORA US Network was established to collect primary breast cancer-metastasis pairs for multi-platform genomic profiling in order to identify the molecular drivers of metastatic disease. AURORA US has both a retrospective and prospective phase. This is the first report of the retrospective phase. Methods: Archived tissue samples from the primary tumor and at least one distant metastasis were retrospectively collected from 83 MBC patients. Following internal quality assessment, samples from 55 pts, including 105 distinct metastatic lesions, were subject to DNA low pass whole genome and exome sequencing, DNA methylation arrays, and RNA sequencing. Early analyses of these multi-platform data include: DNA methylation, tumor gene expression and microenvironmental signatures, somatic and germline variants, DNA copy number changes, and structural variants between breast primaries and matched metastases. Results: Median age at diagnosis was 49 years (25-76); 32 (58%) were Stage I or II at presentation, 27 (49%) had a family history of breast cancer, and 20 (36%) had a second breast primary. Median disease-free interval before developing MBC was 2 years (range 0-36, 5 patients presented with Stage IV). Median overall survival from initial presentation was 4 years (range 0-37). Median survival after developing MBC was 1 year (range 0-13), with a median of three treatments. Primary tissue samples were banked from 1977-2017 and metastases were banked from 1999-2017. Clinical phenotypes of the primaries included 27 HR+ (49%), 15 triple negative (TNBC, 27%), and 11 HER2+ (20%, 12 missing HER2 status). Intrinsic subtype distribution of the primaries included 17 Basal-like (31%), 17 Luminal A (31%), 7 Normal-like (13%), 5 HER2-enriched (9%), and 1 Luminal B, with 8 pending. All metastases from the Basal-like cases remained Basal-like, while metastases from luminal primaries tended to gain HER2-Enriched subtype features (5/18, p = 0.01). Overall, we identified significant metastasis-enriched alterations in metabolism pathways, an increase in proliferation, and the loss of differentiation signatures and immune infiltrates with progression; the latter being the most pronounced in brain metastases. The most frequent somatic mutations in this cohort were in TP53, NCOR1, and RUNX1. Interestingly, ERBB2, EGFR, and ATM were also mutated in ≥10% of the tumors sequenced. In almost all cases, CpG island hypermethylation was clonally present in the primary tumor and persisted stably in the majority of metastatic lesions. Promoter CpG island hypermethylation was also identified in some metastatic lesions at JAM3, an important cellular adhesion molecule,and this was accompanied by reduced mRNA expression. CONCLUSIONS: Collection of banked primary and metastatic tissue pairs identified a young MBC cohort with a high frequency of breast cancer family history and second breast primaries. Molecular characterization of luminal tumor pairs highlighted acquisition of aggressive traits including increased proliferation and loss of differentiation in the metastases. In contrast, basal-like pairs remained relatively unchanged, except for the loss of immune activation. Ongoing analyses to be presented include clonal heterogeneity and phylogeny, novel metastasis signature discovery, gene fusion, and endogenous retrovirus detection. Citation Format: Tari A King, Minetta C Liu, Marni B McClure, Toshinori Hinoue, Benjamin J Kelly, Chad J Creighton, Jay Bowen, Kristen Leraas, Robyn T Burns, Sara Coppens, Salma Rezk, Amy L Garrett, Justin M Balko, Joel S Parker, Ben H Park, Ian Krop, Carey Anders, Katherine A Hoadley, Julie Gastier-Foster, Mothaffar F Rimawi, Rita Nanda, Nancy U Lin, Claudine Isaacs, P. Kelly Marcom, Anna Maria Storniolo, Fergus J Couch, Elaine R Mardis, Adrian V Lee, Uma Chandran, Peter W Laird, Susan G Hilsenbeck, Larry Norton, Andrea L Richardson, W. Fraser Symmans, Lisa A Carey, Antonio C Wolff, Nancy E Davidson, Charles M Perou, the AURORA US Network. Multiplatform analysis of matched primary and metastatic breast tumors from the AURORA US Network [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr GS3-08.

520 Background: There is considerable interest in developing HER2 testing criteria for adjuvant H. We used the 21-gene assay to examine the relationship of HER2 mRNA to benefit from H. Methods: N9831 compared adjuvant chemotherapy AC-T to concurrent chemotherapy-trastuzumab AC-TH in stage I-III HER2+ breast cancer. Recurrence Score (RS) and HER2 mRNA expression were determined by Oncotype DX (neg<10.7, equiv 10.7 to <11.5, and pos ≥11.5 log2 expression units). Cox regression was used to assess the association of HER2 expression with H benefit for distant recurrence. Results: Median follow-up: 7.4 yrs. Of 1,940 total pts, 901 had consent and sufficient tissue. HER2 by RT-PCR was neg in 130 (14%), equiv in 85 (9%), and pos in 686 (76%) pts. Concordance between HER2 assessments was 95% for RT-PCR vs central IHC (>10% + cells = +), 91% for RT-PCR vs central FISH (≥2.0 = pos) and 94% for central IHC vs central FISH. In the primary analysis, the association of HER2 expression with H benefit was marginally non-significant (P=0.057). In hormone receptor pos pts (local IHC) the association was significant (P=0.002). The association was nonlinear with the greatest estimated benefit at lower and higher HER2 mRNA expression levels. The observed treatment benefit in low HER2 pts was not due to imbalance between arms in RS and individual gene expression values. Conclusions: Concordance among HER2 assessments by central IHC, FISH, and RT-PCR was high. Association of HER2 mRNA expression with H benefit was marginally non-significant. A consistent benefit of trastuzumab irrespective of mHER2 levels was observed in the pts with either IHC+ or FISH+ tumors. Benefit was observed in pts with high HER2 by RT-PCR but also observed for the small groups of pts with negative results by quantitative RT-PCR or FISH (Table). Plausible mechanisms for this observation will be discussed. [Table: see text]

1021 Background: At 5 years, AT did not improve disease free survival or overall survival and RS was a more accurate predictor of relapse than standard clinicopathologic characteristics for patients with hormone receptor (HR) positive tumors. Methods: A Phase III Intergroup trial tested adjuvant AT vs. AC. Women with 1-3 N + or N - and T-size > 1cm were randomized to 4 cycles of AT (60 mg/m2/60 mg/ m2) or AC (60 mg/m2/600 mg/m2) q 3 wk x 4. Patients(pts) with ER + and/ or PR + tumors received tam for 5 yrs. Pts were stratified by nodal, HR (ER+ PR+, ER+PR-, ER-PR+, ER-PR-, ER/PR unk) and menopausal status. The primary endpoint was DFS. A sample of 465 pts with HR + breast cancer with 0 to 3 positive axillary nodes who did (N =116) or did not have a recurrence had tumor tissue evaluated using the 21- gene assay. Grade and HR expression were evaluated locally and centrally. Results: 2952 pts were randomized between 7/30/98 and 1/21/00. 2883 were eligible and analyzable. Arms were balanced for age, HR, menopause, nodes, surgery, grade and T-size: median age 51; 64% ER +; 65% LN-; grade: 10% low, 38% int., 46% high; and median T-size - 2.0 cm. At a median follow-up of 11.5 years the DFS/OS results are shown in the table below. RS was a highly significant predictor of recurrence including node negative and node positive disease (P < .0001) and predicted recurrence more accurately than clinical variables. Conclusions: At 11.5 yrs. median follow-up, there remains no difference in DFS or OS, although there continue to be fewer events in the AT arm in the prespecified ER/PR negative subgroup. At 10 years, the RS continues to be a more accurate predictor of relapse than standard clinical features. [Table: see text]

1005 Background: We previously reported that in the E5194 clinical trial patients with ductal carcinoma in situ (DCIS) treated with wide local excision (WLE) without radiation (RT), the DCIS Score was significantly associated with 10 year risk of an ipsilateral breast event (IBE - recurrence of in situ or invasive carcinoma), whether evaluated as a continuous or categorical variable (P=0.02 for both). Here we evaluate correlation between DCIS Score and clinicopathologic (CP) features and if DCIS Score provides independent recurrence risk information. Methods: The study population included 327 women with DCIS prospectively selected for treatment with WLE without RT, including low-intermediate grade tumors ≤2.5 cm or high-grade ≤1 cm. CP variables included age, menopausal status, tamoxifen treatment (used in 29%) and expert centrally determined tumor size, grade, comedo necrosis, tumor type, and margin status. The association between DCIS Score and CP variables was examined by spearman rank correlation, and proportional hazards regression models were used to determine variables significantly associated with IBE. Results: Lesion size (p=0.009) and menopausal status (p=0.03) were significantly associated with IBE, while grade (p=0.69) and comedo necrosis (p=0.47) were not. DCIS Score was significantly associated with IBE after adjustment for CP features and tamoxifen use (p=0.02). DCIS Score was moderately correlated with grade (rs=0.46; 95% CI 0.37,0.54), percentage comedo necrosis (rs=0.49; CI 0.41,0.57), and lesion size (rs=0.18; CI 0.08,0.29) but not other features. Exploratory analyses in all CP subgroups, including the multicomponent Van Nuys Prognostic Index, showed a wide range of DCIS Scores in each subgroup. Concordance of the grades among readers was low: local vs parent central, 68%; local vs central nuclear grade, 45%; parent central vs central nuclear grade, 37%. Conclusions: DCIS Score is moderately correlated with grade, comedo necrosis, and tumor size. DCIS Score provides recurrence risk information independent of these features and identifies subjects with DCIS who are at high risk for local invasive and in-situ local recurrence after WLE alone.

579 Background: N9831 and other clinical trials have demonstrated the additive survival benefit of trastuzumab (H) therapy with chemotherapy for patients with resected HER-2 positive breast cancer. Cardiac toxicity (CTox) is increased by this strategy. Prediction of or early detection of CTox is important. Methods: 95 patients provided informed consent. Brain natriuretic peptide (BNP), C-reactive protein (CRP), troponin T (TnT) and I (TnI), TNF-alpha, IL-1b, and IL-6 were to be drawn at baseline, after initiation of doxorubicin/cyclophosphamide (AC), after initiation of paclitaxel (T), and after the initiation of H (may have been given concurrently with T). 67 patients with at least a baseline sample, a post-treatment sample, and a baseline and subsequent evaluation of LVEF were studied. No patient in this group had a grade 3 decline in LVEF. Only 3 of these patients did not receive H. Values above the 99th percentile were considered abnormal for troponin. 40 ng/ml was considered elevated for BNP. Results: Values of CRP, TNF-alpha, IL- 6 and IL1b changed in very few patients and without a discernible pattern. For further evaluation of BNP, TnI, and TnT, 2 definitions of CTox were tested: 1. a >10% drop in LVEF from baseline (n=27) and 2. a drop in LVEF of 10–15% to less than the institutional lower limit of normal (LLN) OR a >15% decline in LVEF (n=14). No pre-treatment marker appeared to predict a > 10% decline in LVEF (definition 1). For patients fulfilling definition 2, pre-treatment BNP > 40 was seen in 21% of patients vs 8% of normals (nls) (p= 0.18) and an abnormal TnI (>0.04) was seen in 21% vs 6% of nls (p=0.10). Only doubling of BNP appeared promising as a serial marker of CTox using definition 2 (27% vs 7% for nls (p = 0.09). Conclusions: Baseline BNP and TnI and serial measures of BNP may be promising for further investigation for the prediction and detection of treatment-related CTox. Due to limitations of the study including the small numbers of patients analyzed, the difficulty in defining CTox, and the variation in the timing of sample procurement and assessment, these data are considered exploratory. Supported by CA-25224, CA-2115, CA 38926, CA31946. No significant financial relationships to disclose.

512 Background: The joint efficacy analysis of NCCTG N9831 and NSABP B-31 demonstrated improved outcomes with the addition of trastuzumab (H) to doxorubicin and cyclophosphamide (AC) followed by paclitaxel (T) in women with surgically removed HER2- positive breast cancer (NEJM 2005). Following this report, patients randomized to AC→T and less than 6 months from completion of chemotherapy were eligible to receive H. We have updated the initial combined results of these two trials. Methods: Primary and secondary efficacy endpoint were disease free survival (DFS) and overall survival (OS). Cox modeling was performed. Hazard ratios are adjusted for nodal status, tumor size, T schedule, hormone receptor status, and trial (also age for OS). Results: Among the 3,969 women enrolled (ages: 22 to 80yrs), there have been 619 events (H group: 222, non-H: 397). First events were: recurrence (511), contralateral breast disease (18), other second primary cancers (48), and death without recurrence or second primary cancers (42). The median follow-up among the 3,711 women still alive is 2.9 years (range up to 6,4 years). The 4 yr DFS rate and 4 yr OS rate respectively were: 85.9% (95%CI: 84.0–87.8%) and 92.6% (95%CI: 91.2–94.2%) in H group and 73.1% (95%CI: 70.6–75.8%) and 89.4% (95%CI: 87.6–91.2%) in non- H group. Hazard ratio for adjuvant (H/non-H) was 0.49 (P<0.0001; 95%CI: 0.41–0.58) for DFS and 0.63 (P=0.0004; 95%CI: 0.49–0.81) for OS.The rates of a first event per 1,000 women/year during yrs 1–4 were: 26.7, 52.9, 49.6 and 23.2 for H and 42.3, 102.3, 107.6 and 61.5 for non-H. The impact of crossover on clinical outcome will be explored. Conclusions: With an increase in the median follow-up of 11 months and with 225 additional events, the demonstration of substantial improvement in outcomes with the addition of trastuzumab to chemotherapy persist. This improvement continues in spite of some degree of cross-over occurring after the initial results were reported.Updated data regarding crossover outcomes and H duration (as they relate to disease free survival) will be presented. [Table: see text]