The Predictive Utility of MammaPrint and BluePrint in Identifying Patients with Locally Advanced Breast Cancer Who are Most Likely to Have Nodal Downstaging and a Pathologic Complete Response After Neoadjuvant Chemotherapy

BACKGROUND

Neoadjuvant chemotherapy (NCT) increases the feasibility of surgical resection by downstaging large primary breast tumors and nodal involvement, which may result in surgical de-escalation and improved outcomes. This subanalysis from the Multi-Institutional Neo-adjuvant Therapy MammaPrint Project I (MINT) trial evaluated the association between MammaPrint and BluePrint with nodal downstaging.

PATIENTS AND METHODS

The prospective MINT trial (NCT01501487) enrolled 387 patients between 2011 and 2016 aged ≥ 18 years with invasive breast cancer (T2-T4). This subanalysis includes 146 patients with stage II-III, lymph node positive, who received NCT. MammaPrint stratifies tumors as having a Low Risk or High Risk of distant metastasis. Together with MammaPrint, BluePrint genomically (g) categorizes tumors as gLuminal A, gLuminal B, gHER2, or gBasal.

RESULTS

Overall, 45.2% (n = 66/146) of patients had complete nodal downstaging, of whom 60.6% (n = 40/66) achieved a pathologic complete response. MammaPrint and combined MammaPrint and BluePrint were significantly associated with nodal downstaging (p = 0.007 and p < 0.001, respectively). A greater proportion of patients with MammaPrint High Risk tumors had nodal downstaging compared with Low Risk (p = 0.007). When classified with MammaPrint and BluePrint, more patients with gLuminal B, gHER2, and gBasal tumors had nodal downstaging compared with HR+HER2-, gLuminal A tumors (p = 0.538, p < 0.001, and p = 0.013, respectively).

CONCLUSIONS

Patients with genomically High Risk tumors, defined by MammaPrint with or without BluePrint, respond better to NCT and have a higher likelihood of nodal downstaging compared with patients with gLuminal A tumors. These genomic signatures can be used to select node-positive patients who are more likely to have nodal downstaging and avoid invasive surgical procedures.

Abstract P4-02-20: Utility of the 70-gene signature and 10 year follow up in patients with early-stage breast cancer in a single institution study

Introduction: Genomic tests are routinely used by clinicians to guide treatment decisions in early-stage breast cancer (EBC). The 70-gene MammaPrint assay (MP) assesses the risk of distant recurrence in untreated patients with EBC and categorizes the tumors as High Risk (HR, MP index: -1 to ≤0) or Low Risk (LR, &gt;0 to +1). The LR category is further divided into Low but non-UltraLow (LNUL; &gt;0 to ≤0.355) and UltraLow Risk (UL; &gt;0.355 to 1). Here, we report the risk of distant recurrence by MP and 10-year outcomes in patients with EBC diagnosed at Magee Women’s Hospital of the University of Pittsburgh Medical Center.

Methods: In this retrospective analysis, 259 women diagnosed with EBC between 2005 and 2008, who received a MP result, were included. Patient clinical and tumor characteristics were collected. The median FU was 13.1 year among patients with clinical data. Treatment received, 10-year Distant Metastasis Free Interval (DMFI) and 10-year Breast Cancer Specific Survival (BCSS) are reported according to the MP groups. Differences in DMFI and BCSS between MP risk groups were assessed by log-rank. Patients were treated at the physician’s discretion. Treatment was started prior to obtaining MammaPrint results.

Results: Among the 259 patients, 69% were post-menopausal women (mean [range] age: 58 [31-81] years) and diagnosed with hormone receptor-positive HER2-negative tumors (90%), grade 1 or 2 (64%), and without lymph node invasion (93%). In this cohort, 69% (n = 159) had a MP LR result and 31% (n = 100) had a MP HR result. Overall, 14% (n = 35) of patients had a MP UL risk of recurrence of whom 74% (n = 26/35) were post-menopausal women. Substantially more patients received chemotherapy in the HR group (57%, n = 57) compared with the LR group (15%, n = 24) (table). Considering that treatment was initiated before MammaPrint results were known, MP results might have allowed chemotherapy de-escalation in in 15% (n = 24) of patients with a MP LR. Similarly, in the 39% (n = 39) of women with a MP HR treated with endocrine therapy only, knowledge of MP results could have provided important information to add chemotherapy to the treatment plan.. In the MP UL group, 74 % (n = 26) of patients were treated with endocrine therapy only compared with those who received chemotherapy (9%, n = 3) and no adjuvant treatment (9%, n = 3). The 10-year DMFI and 10-year BCSS were higher in the LR compared with the HR group (table). When further stratifying the MP LR group in LNUL and UL, the 10y DMFI was 0.97 (95% CI; 0.94 – 1.00) and 1.00 for the MP LNUL and UL groups, respectively. Within the first 10 years, 8 of the 10 distant recurrences observed were in the MP HR group, and 2 were in the MP LNUL group. Among the 18 recorded deaths, 5 were breast cancer-related, 4 in the MP HR and 1 in the MP LR (LNUL) groups.

Discussion: In this single-institution retrospective analysis, all patients showed excellent BCSS and DMFI outcomes confirming the ability of MP to correctly predict the good prognosis (LR) and poor prognosis (HR) in patients with EBC. In this analysis, as observed in other cohorts, women with a MP UL risk result had an excellent prognosis at 10 years while being treated mostly with endocrine therapy only. Taken together, with the low endocrine therapy adherence reported in the literature, these data suggest that patients with a MP UL result may be candidates for further treatment de-escalation to optimize the risk/benefit ratio of endocrine therapy in future studies.

Table 1. Clinical outcomes and treatment received in patients stratified by MammaPrint results * Patients were treated at the physician’s discretion a p = 0.011, MP LR vs MP HR. b p = 0.032, MP UL vs LNUL vs HR. c p = 0.061 MP LR vs MP HR. d p = 0.170, MP UL vs LNUL vs HR.

Citation Format: Azadeh Nasrazadani, Juan L. Gomez Marti, Margaret Q. Rosenzweig, Meghan McGuire, Katie Quinn, Josien Haan, Alexandre Houzelle, Rohit Bhargava, William Audeh, Adam M. Brufsky. Utility of the 70-gene signature and 10 year follow up in patients with early-stage breast cancer in a single institution study [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P4-02-20.

Abstract PD9-08: ImPrint immune signature in 10,000 early-stage breast cancer patients from the real-world FLEX database

BACKGROUND: Immune checkpoint inhibitors in combination with chemotherapy have demonstrated an improvement of pathologic complete response (pCR) in patients with HR-HER2- and MammaPrint (MP) High Risk, HR+HER2- tumors in the I-SPY2 TRIAL. However, not all patients benefit from immune checkpoint blockade and these new agents come with additional financial burden and significant long-lasting side effects such as adrenal insufficiency. Thus, it is imperative to better understand who benefits. Response Predictive Subtypes (RPS) were developed in the I-SPY2 TRIAL using pre-treatment expression data from 987 MP High Risk patients; 39% of HR+HER2- tumors and 63% of HR-HER2- tumors were identified as immune sensitive. In I-SPY2.2, RPS tumor classification uses ImPrint, a 53-gene signature that has been independently validated to predict the likelihood of a pCR with PD1-PDL1 immune checkpoint inhibitors with high sensitivity and specificity. Using a real-world dataset of 10,000 patients enrolled in the FLEX trial, we identified immune sensitive (ImPrint+) patients within immunohistochemistry (IHC) subtypes and within MP and BluePrint (BP) subgroups.

METHODS: FLEX (NCT03053193) is an ongoing registry trial with 97 sites open in the United States and 2 international sites. Patients enrolled in FLEX have early-stage breast cancer and receive standard of care MP testing with or without BP molecular subtyping and consent to clinically annotated full genome data collection. MP is a 70-gene risk of distant recurrence signature that classifies patients as Low Risk or High Risk. MP High Risk can be further stratified into High 1 and High 2, which have demonstrated differences in chemosensitivity and pCR rates in the I-SPY2 TRIAL (NCT01042379). BP, an 80-gene molecular subtyping signature, categorizes patients’ tumors as Luminal-, HER2- or Basal-Type.

RESULTS: Of the 10,021 patients, 9.1% of the FLEX patient population are ImPrint+ and are predicted to have a meaningful pCR rate with immune checkpoint inhibitors. Younger (≤ 50 years) or pre/peri-menopausal patients, patients with larger or node-positive tumors, and patients of Black or Latin race/ethnicity independently had a higher likelihood of having ImPrint+ tumors (Table 1). ImPrint+ tumors were identified in all clinical subtypes by IHC. There is a higher likelihood of ImPrint+ tumors being MP High 2 or BP Basal-Type tumors. Within BP Basal tumors, 74.7% of HR+ and 66.0% of HR- tumors were ImPrint+.

CONCLUSIONS: The focus of immune therapy trials has been on patients with HR-HER2-, MP High Risk patients. Indeed, most patients who are predicted to benefit have MP High 2 or BP Basal-Type tumors, including some HR+ patients, which is consistent with I-SPY2 results. Importantly, this large real-world dataset enables the identification of populations who may benefit from immune therapy outside of traditional clinical trial populations and supports the testing of checkpoint inhibitors in the immune-positive subtype. Younger women and patients of Black or Latin race/ethnicity who typically have more aggressive tumors also have higher proportions of ImPrint+ tumors. Thus, it is critical that these populations be included in clinical trials. This first look at immune sensitivity in over 10,000 FLEX patients with ImPrint generates preliminary data and hypotheses that will be explored in future FLEX substudies, including an analysis of lobular cancers and long-term outcomes in ImPrint+ patients across all races and ages.

Table 1. Clinical characteristics of ImPrint+ and ImPrint- tumors.

Citation Format: Adam M. Brufsky, Midas Kuilman, Rita Mukhtar, Denise M. Wolf, Christina Yau, Joyce O’Shaughnessy, Cathy Graham, Vijayakrishna K. Gadi, Pat Whitworth, Alexander Hindenburg, Ian Grady, Gordon Srkalovic, Kent Hoskins, Ajay Dhakal, Cynthia Ma, Natasha Hunter, Jennifer Crozier, Blanche Mavromatis, Lorenza Mittempergher, Christine Finn, Shraddha Modh, Erin B. Yoder, Patricia Dauer, Andrea Menicucci, Bas van der Baan, William Audeh, Laura J. Esserman. ImPrint immune signature in 10,000 early-stage breast cancer patients from the real-world FLEX database [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr PD9-08.

Abstract OT2-05-01: FLEX: 30K Full Transcriptome, Real-World Evidence Database for Early-Stage Breast Cancer, and Investigator-Initiated Protocols

FLEX: 30K Full Transcriptome, Real-World Evidence Database for Early-Stage Breast Cancer, and Investigator-Initiated Protocols BACKGROUND: Genomic signatures, such as the 70-gene MammaPrint, provide additional prognostic information for early-stage breast cancer (EBC), such as tumor metastatic potential, and expand the available information clinicians and patients require for shared treatment decision making beyond the standard clinicopathologic factors. EBC tumors are further stratified into clinically actionable subtypes by molecular assays such as the 80-gene molecular profiling assay BluePrint, improving pathological complete response rates (pCR) and outcomes as evidenced by a number of recent trials. The ongoing FLEX Study (NCT03053193) is designed to expand the genomic information available for EBC cases, and to increase the speed of data generation for rare and underserved research areas. To date, FLEX is the largest international multicenter real-world evidence (RWE) EBC registry, with more than 10,000 patients enrolled in fewer than five years since opening. FLEX pairs full genome data with more than 800 clinical data points collected over 10 years of follow up to provide the most comprehensive big data database available for early-stage breast cancer. The FLEX enrollment has a goal of a minimum of 30,000 patients within 10 years. METHODS: The FLEX study is a multicenter, prospective, observational trial for patients ≥18 years old with histologically proven stage I-III invasive breast cancer that is node negative or positive (up to three nodes) who receive MammaPrint, with or without BluePrint as standard of care management. Patients consent to the collection of clinically annotated full transcriptome data. Additionally, this study protocol allows physicians to investigate targeted populations or clinical trial investigator-initiated studies (IIS) upon approval by a peer-driven Scientific Review Committee. As patients enrolled in the FLEX study meet all eligibility criteria for inclusion no additional consent is required. The FLEX enrollment has already surpassed 1/3 of set target goal. As of April 2022, the trial has surpassed 10,000 patients enrolled at nearly 100 trial sites across the US and Europe, Middle East and Africa (EMEA). To date, there have been 36 publications in international scientific congresses with 39 FLEX IIS and in progress. With over 250 active and collaborating physicians leveraging the shared infrastructure, the IIS have enabled the ability to address disparities in treatment to underrepresented populations, rare subtypes, age, and patient centered specific topics. Current and future questions investigated via this platform will continue to strive to improve outcomes for early-stage breast cancer patients.

Citation Format: Alejandra Perez, Hannah Linden, Nathalie Johnson, Sami Diab, Chirag Jani, Chelsea D. Gawryletz, Richard Fine, Laura Lawson, Megan Baker, Victoria Poillucci, Lisa E. Blumencranz, William Audeh. FLEX: 30K Full Transcriptome, Real-World Evidence Database for Early-Stage Breast Cancer, and Investigator-Initiated Protocols [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr OT2-05-01.

Abstract P5-09-02: Impact of neoadjuvant endocrine therapy on tumor transcriptome in patients with early-stage breast cancer from the FLEX trial

BACKGROUND: Neoadjuvant Endocrine Therapy (NET) is seldom used in breast cancer management except in patients with several comorbidities or in elderly patients in which chemotherapy is not an option. Clinical response with NET is not typically achieved until after several months of treatment. In the NET setting, reduction of Ki67 (&lt; 10%) after 2-4 weeks has been used as a predictor of positive response, but studies such as ALTERNATE have questioned this association. It remains uncertain whether a single gene or protein can adequately predict outcomes or inform how NET alters a variety of cancer genes and global tumor biology. This study evaluated the effect of short-term NET on the tumor genomics of patients with early-stage breast cancer (EBC) by comparing whole transcriptome gene expression changes in matched pre- and post-NET tumor samples. METHODS: In this single-institution FLEX substudy performed at Johns Hopkins, patients (n=30) with matched pre- and post-treatment specimens who received at least two weeks of NET between 2019 – 2021 were included. Premenopausal and male patients with breast cancer received Tamoxifen (n=10) and postmenopausal women received either Letrozole (n=10) or Exemestane (n=10). Limma R package was used for quantile normalization and differential gene expression analysis. Significant differentially expressed genes (DEGs) had a false discovery rate of &lt; 0.05 and &gt;2-fold change. Pathway enrichment analysis was performed using Reactome. For patients with available clinical information, changes in immunohistochemistry (IHC) between pre- and post-NET were quantified using absolute values, and the median percent change was reported, with significance assessed using the Wilcoxon test. The observational FLEX trial (NCT03053193) enrolls patients with EBC who have MammaPrint (MP) with or without BluePrint testing and consent to clinically annotated full transcriptome data collection. MammaPrint classifies tumors as having a Low Risk (LR) or High Risk (HR) of distant recurrence. BluePrint is a molecular subtyping assay, and together with MammaPrint, tumors are classified as Luminal A-Type (MP LR), Luminal B-Type (MP HR), HER2-Type, or Basal-Type. RESULTS: Transcriptional profiles between pre- and post-NET samples were distinct with short-term NET inducing 774 DEGs. The majority of significant DEGs (n=748) such as MGAT1, IQGAP3, and PRC1, which are associated with tumor aggressiveness and metastasis, were downregulated in post-NET samples. Upregulated genes in post-NET tumors, such as FOS, JUN, and EGR1, are involved in estrogen signaling and NF-κB pathways and are associated with better outcomes. Among the 30 patients, 7 (6 Luminal B and 1 Basal) remained MP HR and 16 remained MP LR (Luminal A) pre- and post-NET, 1 changed from LR (Luminal A) pre-NET to HR (Luminal B) post-NET, and 6 changed from HR (Luminal B) pre-NET to LR post-NET (Luminal A). The median percent change by IHC in matched pre- and post-NET tissue was 2.5% for estrogen receptor (ER) (range: 0-50%; p=0.750), 22% for progesterone receptor (PR) (range: 0-81%; p=0.097), and 9% for Ki67 (range: 0-43%; p=0.026). CONCLUSIONS: In this study, significant gene expression changes were discovered within a shorter timeframe than when clinical responses are usually observed in the NET setting. This could indicate biological complexity and diverse response pathways, which may be more informative when combined with a single IHC biomarker (ER/PR/Ki67). Results from this study should be confirmed using a larger cohort. Future studies will determine the significance of these DEGs and their impact on outcomes, and will further define gene expression changes by endocrine therapy type (tamoxifen versus aromatase inhibitors). ACKNOWLEDGMENTS: We would like to thank Lynn and Robert Downing for their generous support of our study.

Citation Format: Mehran Habibi, Danijela Jelovac, Rima Couzi, Cesar Augusto Santa-Maria, Catherine Klein, Marissa White, Nivali Naik, Jennifer Wei, Yen Huynh, Architha Ellappalayam, Lisa E. Blumencranz, Erin B. Yoder, Patricia Dauer, Bas van der Baan, William Audeh. Impact of neoadjuvant endocrine therapy on tumor transcriptome in patients with early-stage breast cancer from the FLEX trial [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P5-09-02.

Abstract GS5-10: Utility of the 70-gene MammaPrint test for prediction of extended endocrine therapy benefit in patients with early-stage breast cancer in the IDEAL Trial

Background: The IDEAL trial showed no significant benefit of 5 years extended endocrine therapy (EET) using letrozole in postmenopausal patients with hormone receptor positive (HR+) breast cancer (BC) versus 2.5 years. Genomic classifiers may assist with treatment decisions by predicting EET benefit. The 70-gene MammaPrint (MP) test classifies tumors as having a higher or lower risk of distant metastasis in HR+ early-stage BC. A MP lower risk result can be further classified as either Ultra-Low risk or Low risk of distant metastasis. In the NSABP B42 trial, MP predicted a statistically significant absolute benefit from EET in patients with a MP Low Risk result. Here, we aimed to determine the utility of MP in identifying a subgroup of patients enrolled in the IDEAL trial for which 5 years of EET is beneficial compared to 2.5 years.

Methods: A total of 869 patients had available primary tumor tissue for testing. MP results were available for 545/869 patients, of which 515 did not have an event at 2.5 year after randomization and were used for our analyses. The MP result for each patient was calculated by Agendia while blinded to patient clinical outcomes. The primary endpoint was distant recurrence (DR). Secondary endpoints were recurrence free interval (RFI) and breast cancer free interval (BCFI) as defined by STEEP criteria. Patients were classified as higher risk (score -1.000 - 0) or lower risk (score 0.001 - 1.000). Lower risk tumors were further classified as either MP Ultra-Low (score &gt; 0.355) or MP Low Risk (score ≥ 0.001, ≤ 0.355). Likelihood ratio test based on stratified Cox proportional hazards (PH) model were used to evaluate treatment by risk group interaction. Differences in endpoints between treatment groups were assessed by stratified log-rank tests. Hazard ratios (HR) and 95% Confidence Intervals (CI) were computed based on the stratified Cox PH model.

Results: The clinical characteristics of the 515 IDEAL samples with a MP result were comparable to the whole IDEAL cohort (n=1820). Within the 2.5 year EET group, 50.6% (n=134) were MP higher risk and 49.4% (n=131) MP lower risk, of which 14.5% (n=19/131) were MP Ultra-Low. Within the 5 year EET group, 50.0% (n=125) were MP higher risk and 50.0% (n=125) MP lower risk, of which 11.2% (n=14/125) were MP Ultra-Low. Among patients with MP lower risk tumors, 5 years vs. 2.5 years of EET resulted in a significant absolute benefit of 9.8% for DR (HR=0.42, [95% CI 0.174-0.996]), 9.8% for RFI (HR=0.43, [95% CI 0.198-0.934]), and 8.8% (HR=0.53, [95% CI 0.264-1.055]) for BCFI, whereas patients with MP higher risk tumors did not derive significant benefit (Table 1). Within the MP lower risk group, 5 year vs 2.5 year EET benefit was more pronounced in MP Low tumors, which exhibited a significant benefit of 10.1% for DR (HR=0.32, [95% CI 0.116-0.866]), 11.7% for RFI (HR=0.35, [95% CI 0.147-0.824]), and 9.7% for BCFI (HR=0.48, [95% CI 0.225-1.015]); MP Ultra Low tumors did not derive significant benefit. Treatment-by-risk group interaction was statistically significant for RFI.

Conclusion: A significant EET benefit was observed for MammaPrint lower risk tumors but not for MP higher risk tumors. MammaPrint Low tumors exhibited the largest absolute benefit of 5 years of EET compared to 2.5 years. Consistent with the findings in the NSABP B42 trial, the results from this second randomized trial provide clinically meaningful implications in patient selection for extended endocrine therapy.

Table 1. IDEAL: 10-year outcome analysis comparing 5 years vs. 2.5 years of EET using letrozole stratified by MP risk. **MammaPrint Lower Risk & Higher Risk (n=515) and *** MammaPrint Low Risk & High Risk (n=482)

Citation Format: Gerrit-Jan Liefers, Elma Meershoek-Klein Kranenbarg, Marjolijn Duijm-de Carpentier, Cornelis J.H. van de Velde, Miranda Kleijn, Christa Dreezen, Andrea Menicucci, Laura van’t Veer, William Audeh. Utility of the 70-gene MammaPrint test for prediction of extended endocrine therapy benefit in patients with early-stage breast cancer in the IDEAL Trial [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr GS5-10.

Genomic Classification of HER2-Positive Patients With 80-Gene and 70-Gene Signatures Identifies Diversity in Clinical Outcomes With HER2-Targeted Neoadjuvant Therapy

The prospective Neoadjuvant Breast Registry Symphony Trial compared the 80-gene molecular subtyping signature with clinical assessment by immunohistochemistry and/or fluorescence in situ hybridization in predicting pathologic complete response (pCR) and 5-year outcomes in patients with early-stage breast cancer.

HER2-low breast cancers: incidence, HER2 staining patterns, clinicopathologic features, MammaPrint and BluePrint genomic profiles

Recently, clinical trials have demonstrated promising efficacy for novel HER2-targeted therapies in HER2-low breast cancers, raising the prospect of including a HER2-low category (immunohistochemical [IHC] score of 1+, or 2+ with non-amplified in-situ hybridization [ISH]) in the HER2 evaluation of breast cancers. In order to better understand this newly-proposed HER2 category, we investigated the incidence, HER2 staining patterns, clinicopathologic features, and genomic profile of HER2-low breast cancers. HER2-stained slides of 281 consecutive breast cancers were re-reviewed and the clinicopathologic information, MammaPrint, and BluePrint results of these cases were retrospectively analyzed. HER2-low breast cancers were identified in 31% of cases and were more common in estrogen receptor (ER)-positive than ER-negative breast cancers (33.6% vs 15%, p = 0.017). HER2-low cancers were generally clinical stages I-II (79%), ER-positive (93.1%), had homogenous HER2 staining (59.2%), HER2 IHC score of 1+ (87.4%), ductal phenotype (81.6%), histologic grades of 1 or 2 (94.2%) and luminal molecular subtypes (94.3%). Three HER2-low patients received neoadjuvant chemotherapy and none of them achieved pathologic complete response. When compared to HER2-negative (IHC of 0+) and HER2-positive (IHC of 3+ or IHC of 2+ with amplified ISH) cancers, HER2-low breast cancers had significantly lower Ki-67 (p = 0.03 and p < 0.01, respectively) and higher ER positivity (p = 0.01 and p = 0.03, respectively). HER2-low breast cancers were less likely to be basal molecular subtype when compared to HER2-negative cancers (p < 0.01) and were less likely to have a HER2 molecular subtype when compared to HER2-positive cancers (p < 0.01). When adjusted for ER status, there was no significant difference on all the examined variables between HER2-low and HER2-negative groups. Our study provides valuable baseline characteristics of HER2-low breast cancers deriving from consecutive, real-world cases with a consensus confirmation of HER2 status, and would help to increase our understanding of this newly-proposed HER2 category in breast cancers.

Distinct Neoadjuvant Chemotherapy Response and 5-Year Outcome in Patients With Estrogen Receptor-Positive, Human Epidermal Growth Factor Receptor 2-Negative Breast Tumors That Reclassify as Basal-Type by the 80-Gene Signature

The 80-gene molecular subtyping signature (80-GS) reclassifies a proportion of immunohistochemistry (IHC)-defined luminal breast cancers (estrogen receptor–positive [ER+], human epidermal growth factor receptor 2–negative [HER2–]) as Basal-Type. We report the association of 80-GS reclassification with neoadjuvant treatment response and 5-year outcome in patients with breast cancer.

Identity exposed: genomic assay unmasks TNBC-like breast cancer tumors disguised as HR+ #NBRST

Age-Independent Preoperative Chemosensitivity and 5-Year Outcome Determined by Combined 70- and 80-Gene Signature in a Prospective Trial in Early-Stage Breast Cancer

BACKGROUND

The Neoadjuvant Breast Symphony Trial (NBRST) demonstrated the 70-gene risk of distant recurrence signature, MammaPrint, and the 80-gene molecular subtyping signature, BluePrint, precisely determined preoperative pathological complete response (pCR) in breast cancer patients. We report 5-year follow-up results in addition to an exploratory analysis by age and menopausal status.

METHODS

The observational, prospective NBRST (NCT01479101) included 954 early-stage breast cancer patients aged 18-90 years who received neoadjuvant chemotherapy and had clinical and genomic data available. Chemosensitivity and 5-year distant metastasis-free survival (DMFS) and overall survival (OS) were assessed. In a post hoc subanalysis, results were stratified by age (≤ 50 vs. > 50 years) and menopausal status in patients with hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-) tumors.

RESULTS

MammaPrint and BluePrint further classified 23% of tumors to a different subtype compared with immunohistochemistry, with more precise correspondence to pCR rates. Five-year DMFS and OS were highest in MammaPrint Low Risk, Luminal A-type and HER2-type tumors, and lowest in MammaPrint High Risk, Luminal B-type and Basal-type tumors. There was no significant difference in chemosensitivity between younger and older patients with Low-Risk (2.2% vs. 3.8%; p = 0.64) or High-Risk tumors (14.5% vs. 11.5%; p = 0.42), or within each BluePrint subtype; this was similar when stratifying by menopausal status. The 5-year outcomes were comparable by age or menopausal status for each molecular subtype.

CONCLUSION

Intrinsic preoperative chemosensitivity and long-term outcomes were precisely determined by BluePrint and MammaPrint regardless of patient age, supporting the utility of these assays to inform treatment and surgical decisions in early-stage breast cancer.

Abstract P5-07-05: Deciphering the inferior prognosis of young women with estrogen receptor-positive early-stage breast cancer through full transcriptome analysis: A FLEX database sub-study

Introduction: Young women with estrogen receptor (ER)-positive early-stage breast cancer (EBC) frequently present with larger, higher-grade, more aggressive tumors, with lower ER expression than older women. Post-hoc analyses within the MINDACT and TAILORx cohorts reported that women aged 40-50 with ER+ EBC exhibited a chemotherapy benefit independent of genomic risk compared to women aged &gt; 50, possibly due to chemotherapy-induced ovarian suppression (CIOS). However, women &lt; 40 years, who seldom develop CIOS, are under-represented in clinical trials. Therefore, it is important to distinguish the tumor biological profile within this younger age group. To understand the biological basis underlying why younger women have poorer outcomes than older women, this study aimed to identify genes that distinguish tumors in younger women from older women. Methods: EBC patients enrolled in the FLEX study (NCT03053193) undergo standard of care MammaPrint (MP) and BluePrint (BP) tests, and consent to clinically annotated whole transcriptome data collection. MP categorizes tumors as High Risk (HR) or Low Risk (LR) of recurrence. Together, MP and BP classify molecular subtype as Luminal A, Luminal B, HER2, or Basal. Whole transcriptome gene expression differences were compared among ER+ tumor specimens from three age groups: &lt; 40 years old (n=283), 40-54 years old (n=1535), and ≥ 55 years old (n=4355). Differentially expressed genes (DEGs) were identified from 100 randomized iterations for each comparison; an equal number of LR and HR samples were analyzed in each group. DEGs found in &gt; 20 iterations indicated gene stability within the age group. DEGs were also identified within each age group after correcting for BP subtype, and between pre-menopausal (n=1314) and post-menopausal (n=4859) women. Gene expression data were quantile normalized using R package ‘limma’. DEGs with an adjusted p&lt;0.05 and fold change ≥ 2 were considered significant. Results: Overall, 76.0% of women &lt; 40 years, 53.6% of women aged 40-54, and 48.5% of women ≥ 55 years had MP HR tumors. In addition, women &lt; 40 years had higher frequencies of BP Basal and HER2 tumors (20.5% and 9.2%, respectively) compared with women ≥ 55 years (8.0% and 2.4%, respectively; p&lt;0.0001). In line with unsupervised hierarchical clustering and previous studies, tumors from patients aged 40-54 exhibited limited DEGs in comparison to women ≥ 55 years. In contrast, most gene expression differences were primarily observed between women &lt; 40 and ≥ 55 years (11 DEGs; 10 downregulated and 1 upregulated). We identified limited DEGs within BP Luminal A (n=23), Luminal B (n=10), and Basal (n=7) tumors from women aged &lt; 40 relative to women ≥ 55 years. No DEGs were found in HER2-type tumors. Within Luminal A and Basal tumors, upregulated genes in women &lt; 40 years are important for proliferation and immune responses. Luminal B tumors in women aged &lt; 40 have increased DEGs involved in ER and HER2 signaling. Within all subtypes, DEGs supporting metabolic functions were downregulated in tumors of women &lt; 40 years compared with women ≥ 55 years. Few DEGs were observed between pre-menopausal women and post-menopausal women, though none with &gt; 2-fold change. Conclusion: Tumors from women aged 40-54 had few DEGs, suggesting observed chemotherapy benefit represents differences in host biology rather than intrinsic tumor biology as compared with tumors from women &lt; 40 or ≥ 55, where detected DEGs are associated with proliferation, receptor signaling, and metabolism. Fewer DEGs were observed by menopausal status than age, indicating age is a more relevant cutoff. Future studies will aim to further characterize this high risk young patient population.

Citation Format: Shubhada Dhage, Mina Gendy, Nina D'Abreo, Ruth Oratz, Sami G. Diab, VK Gadi, Cathy Graham, Midas Kuilman, Shiyu Wang, Patricia Dauer, Andrea Menicucci, William Audeh, Douglas K. Marks. Deciphering the inferior prognosis of young women with estrogen receptor-positive early-stage breast cancer through full transcriptome analysis: A FLEX database sub-study [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-07-05.

Abstract P3-14-11: Mammaprint and Blueprint identify genomic differences in HR+ HER2- breast cancers from young Black and White women

Introduction: Hormone receptor positive (HR+), human epidermal growth factor receptor 2-negative (HER2-) breast cancer, the most common immunohistochemical subtype, remains the dominant contributor to annual breast cancer deaths worldwide across all racial and ethnic groups. Black women are 41% more likely to die from breast cancer compared to White women, predominantly among women diagnosed ≤ 50 years of age. Yet, Black women remain underrepresented in clinical trials and population-based studies. Thus, it is critical to better characterize tumor molecular features from young Black women to identify factors contributing to the existing racial survival disparity. In the current study, we compared risk of distant recurrence signature, MammaPrint (MP), molecular subtyping signature, BluePrint (BP), and whole transcriptome differences between young Black women with HR+ HER2- breast cancer compared to matched White controls. Methods: This study included 156 Black women aged ≤ 50 with stage I-III, HR+ HER2- breast cancer of whom, 68 were recruited from 2009-2014 as part of the BEST study (5R01CA204819-04) with follow-up data available (median 114.5 months). The remaining 88 Black women were enrolled in the ongoing FLEX Study (NCT03053193) from 2017. White women (n=156) were randomly selected from FLEX and matched by age, tumor stage, and receptor status. Tumors were classified through MP as Low Risk (MP-LR) versus High Risk (MP-HR), with MP-HR further stratified into HR1 and HR2; HR2 tumors exhibit superior chemosensitivity as demonstrated in a prior large clinical trial of breast cancer patients (ISPY2). All women had MP and BP to classify tumors as Luminal A (MP-LR), Luminal B (MP-HR), HER2, or Basal, and full transcriptomic analyses. Differential gene expression analysis was performed with R package ‘limma’ to compare Black and White women and further compare within each molecular subtype. Differentially expressed genes (DEGs) with a false discovery rate &lt;0.05 were significant. Results: Of 312 young women with localized, HR+ HER2- breast cancer, high grade tumors were more frequent among Black compared to White women (34.6% vs 25.6%; p=0.08). MP-HR tumors were significantly more frequent among Black compared to White women (67.3% vs. 50.0%; p=0.002). Among MP-HR tumors, more HR2 tumors were seen in Black (25.6%) compared to White women (14.1%). Among women with MP-HR tumors and known treatment information, most Black women (94.3%) and all White women received chemotherapy. There were more Luminal B tumors in Black compared to White women (51.9% vs. 41.7%; p=0.07). BP reclassified a larger proportion of ER+ tumors as Basal in Black compared to White women (14.1% vs. 8.3%). Of 68 Black women with available survival data, 7 had death and/or distant recurrence events, of whom 6 (85.7%) had MP-HR tumors (4 Luminal B, 1 HER2, and 1 Basal) and 1 had MP-LR Luminal A tumor. Compared to White women, Black women with: 1) Luminal B tumors had 192 DEGs with upregulation of suspected poor prognosis genes, PSPH and IGHG1; 2) Luminal A tumors had upregulation of PSPH; and 3) Basal tumors had downregulation of POTEH. Conclusion: Among young women with localized HR+ HER2- breast cancer, MP and BP molecular signatures more robustly identified racial disparities in risk and subtype distribution beyond that identified through clinical factors adjusted for age and tumor characteristics. The transcriptomic differences among Black compared to White women across all BP subtypes provide novel insights about tumor biological differences. These findings have tremendous translational potential to identify etiologic underpinnings of racial survival disparities which may guide therapeutic strategies to improve outcomes.

Citation Format: Sonya Reid, Tuya Pal, Ingrid A. Mayer, Xiao-Ou Shu, Ann L. Tezak, Kent Hoskins, Dipali Sharma, Patricia Robinson, Jennifer Wei, Jake Ruby, Shiyu Wang, Josien Haan, Andrea Menicucci, William Audeh, FLEX Investigators Group. Mammaprint and Blueprint identify genomic differences in HR+ HER2- breast cancers from young Black and White women [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P3-14-11.

Abstract P2-08-06: Defining transcriptomic profiles of breast cancer with early lymph node metastases: A FLEX database sub-study

Introduction: Early lymph node (LN) metastasis often precedes systemic metastasis and corresponds with a 35% decrease in 10-year survival compared to patients without LN metastasis. Understanding biologic pathways involved in early LN metastasis could identify promising drug targets for early-stage breast cancer (EBC) treatment. While tumor size modestly correlates with the probability of clinical or occult axillary metastases, small T1 tumors are often found to have LN metastases, while many large T3 tumors are often LN negative. We compared large tumors without evidence of LN metastasis (pT2-3N0) and small tumors with LN metastasis (pT1N1-3) by whole transcriptome analysis to elucidate molecular biological differences associated with the presence or absence of early LN metastasis at diagnosis. Methods: The FLEX study enrolls patients with EBC who undergo standard of care MammaPrint (MP) and BluePrint (BP) testing, and consent to clinically annotated full transcriptome data collection. MP classifies tumors as having High Risk (HR) or Low Risk (LR) of distant recurrence. MP combined with BP subtype tumors as either Luminal A-type (MP LR; 56.8%), Luminal B-type (MP HR; 36.4%), HER2-type (1.5%), or Basal-type (5.3%). In our primary comparison, 355 patients had small, LN-positive tumors (pT1N1-3), and 235 had large, LN-negative tumors (pT2-3N0). Control groups consisted of small, LN-negative (pT1N0; n=1384) and large, LN-positive tumors (pT2-3N1-3; n=375). R package ‘limma’ was used for quantile normalization and differential gene expression (DGE) analyses. Differentially expressed genes (DEGs) with a false discovery rate &lt; 0.05 and fold change &gt; 1.4 were considered significant for this exploratory analysis. Gene set enrichment analysis (GSEA) was performed using the Hallmark gene sets (MSigDB). Results: Principal component analysis of each sample’s transcriptional profile revealed more gene profile differences between MP risk and BP subtype than by pathological stage. Overall, DGE analysis comparing pT2-3N0 to pT1N1-3 identified 30 DEGs; of which MUCL1 had a &gt; 2-fold decrease. Within MP LR and BP Luminal A-type tumors, no DEGs were identified comparing pT2-3N0 to pT1N1-3. HR tumors had 73 DEGs (33 upregulated and 40 downregulated); two of 40 downregulated genes, including MUCL1, had &gt; 2-fold change. Within Luminal B tumors, 34 DEGs were identified when comparing pT2-3N0 with pT1N1-3, although none had &gt; 2-fold change. Basal and HER2 tumors were not analyzed due to small numbers. GSEA showed upregulated proliferation and oxidative phosphorylation pathways, and downregulated epithelial to mesenchymal transition (EMT) and immune signatures in pT2-3N0 relative to pT1N1-3 tumors overall, within HR, and Luminal B tumors. Conclusion: This study provides a foundation for understanding the mechanisms that promote LN metastasis. Studies have previously correlated LN metastasis with EMT, immune function, and gene overexpression (BCL2L1, AURKA, CDKN2A, MCL1, MYC). While we did not find significant differences in these genes, our results indicated similar differentially regulated pathways. Overall, we found most biological differences within MP HR and Luminal B tumors. Proliferation-related pathways were upregulated and EMT and immune-related pathways were downregulated in pT2-3N0 tumors compared with pT1N1-3 tumors, suggesting these pathways distinguish small, node-positive tumors from large, node-negative tumors and are involved in early LN metastasis. Interestingly, MUCL1 which has been reported to promote migration and invasion in breast cancer, was consistently downregulated in pT2-3N0 relative to pT1N1-3 tumors. Future studies will investigate antineoplastic therapies for EBC that modulate these dysregulated pathways to reduce early LN metastasis and subsequent systemic metastasis.

Citation Format: Nibash Budhathoki, Faisal Fa'ak, Nina D'Abreo, Shubhada Dhage, Phyu Phyu Soe, Dennis Holmes, Josien Haan, Shiyu Wang, Patricia Dauer, Andrea Menicucci, William Audeh, Douglas K Marks. Defining transcriptomic profiles of breast cancer with early lymph node metastases: A FLEX database sub-study [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P2-08-06.

Abstract OT2-07-01: The FLEX real-world data platform explores new gene expression profiles and investigator-initiated protocols in early-stage breast cancer

Background: Genomic signatures provide prognostic information such as tumor metastatic potential, outperforming information gained from clinicopathologic factors alone, which has transformed personalized treatment of early breast cancer. Together, whole genome expression data can accurately stratify tumors into clinically actionable molecular subtypes. The FLEX study (NCT03053193) is a large, real-world dataset, combining whole genome expression with clinically annotated data, to accelerate research and discovery of subgroups in breast cancer. The goal of FLEX is to discover and develop novel genomic profiles that bring precision oncology to the clinic, and ultimately improve treatment and outcomes of patients with breast cancer. Study Design: The FLEX study is a multicenter, prospective observational trial of patients aged ≥ 18 years with stage I-III breast cancer who receive MammaPrint, with or without BluePrint. Patients consent to the acquisition of clinical data and clinically annotated full transcriptome tumor analysis. The FLEX study creates a unique infrastructure to develop and investigate hypotheses for targeted subset analyses and/or clinical trials. In doing so, the adaptive protocol allows submission of investigator-initiated sub-studies. Upon approval by a Scientific Review Committee, FLEX investigators can query the clinical and genomic data in FLEX to investigate hypotheses for targeted subset analysis. Patients enrolled in the original study who meet all eligibility criteria may be enrolled in sub-studies without additional consent. All necessary clinical data will be collected from patients, starting from diagnosis through 10-year follow-up. Sub-study research categories include: breast cancer and age, optimization of therapeutic strategies, breast cancer tumor types, biomarker identification and quality of care. The FLEX enrollment goal is a minimum of 30,000 patients within 10 years; since April 2017, more than 7,500 patients were enrolled at more than 90 sites. Participating sites include community hospitals and nine National Cancer Institute-designated comprehensive cancer centers to ensure inclusion of diverse populations, particularly patient populations that are underrepresented in traditional clinical trials. Most recently, the FLEX network was expanded globally with additional sites in Greece and Israel. To date, 38 investigator-initiated sub-studies have been approved, resulting in 23 published abstracts at national congresses. Trial contact information: NCT03053193, FLEX@agendia.com

Citation Format: Cynthia X. Ma, Robert Maganini, Pat Whitworth, Jennifer A. Crozier, Joyce O'Shaughnessy, Sami Diab, Beth-Ann Lesnikoski, June Lee, Nina D'Abreo, Blanche H. Mavromatis, Pond Kelemen, Darina Pronin, Hiwot Abebe, Michelle L. Bolner, Christine Finn, Lisa Blumencranz, William Audeh, FLEX Investigators Group. The FLEX real-world data platform explores new gene expression profiles and investigator-initiated protocols in early-stage breast cancer [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr OT2-07-01.

Abstract P4-07-32: Mammaprint and Blueprint are prognostic of outcome following neoadjuvant chemotherapy

Introduction: Neoadjuvant chemotherapy (NAC) is used increasingly in patients with early-stage breast cancer (EBC). Genomic approaches are necessary to identify better biomarkers that can more accurately predict response to NAC and outcome. The impact of NAC on the molecular biology of breast tumors, especially in patients that fail to respond to treatment, is not well studied. We evaluated the risk of distant recurrence signature, MammaPrint (MP), and the molecular subtyping signature, BluePrint (BP), in the neoadjuvant setting and how the prognostic capability of MP and BP, which was validated on untreated biopsies, is affected by NAC in patients with residual disease. Methods: This analysis included 128 women (median age 50) with stage I-III EBC, of any clinical subtype, diagnosed from 2007-2016 who received NAC at Cedars-Sinai Medical Center. Of 128 patients, 73 had core needle biopsies available before treatment, 17 had surgically resected biopsies with residual disease available after treatment, and 38 with residual disease had paired pre- and post-treatment biopsies. MP classified tumors as High Risk (MP-HR) or Low Risk (MP-LR). MP and BP classified tumors as Luminal A (MP-LR), Luminal B (MP-HR), HER2, or Basal. The primary outcomes were pathological complete response (pCR), recurrence free survival (RFS), distant-metastasis free survival (DMFS), and overall survival (OS). The median follow-up time was 5.2 years. Differences between unmatched independent variables were examined by chi-square test or Fisher’s exact test. Differences in pCR rate were assessed by two-sided proportional z-test. For matched samples, McNemar’s test or Bowker’s test was used. Univariate survival analyses were evaluated using a Cox proportional hazards model. Results: Among 111 pre-treated tumors, MP-HR tumors had a significantly higher pCR rate (40.6%; n=39/96) compared to MP-LR tumors (13.3%; n=2/15; p=0.042). BP HER2 and Basal tumors had significantly higher pCR rates (56.8%; n=25/44 and 45.8%; n=11/24, respectively), compared to Luminal B tumors (16.1%; n=5/31; p&lt;0.001 and p = 0.016 respectively). No Luminal A tumors (n=12) achieved pCR. MP-HR and BP Basal tumors were associated with a higher risk of recurrence or death than MP-LR and Luminal tumors. Among all 128 tumors, post-treated samples were more likely to be classified as MP-LR than pre-treated samples (45.5%; n=25/55 vs. 13.5%; n=15/111), whereas BP subtype distribution was similar. Among 38 paired samples with residual disease, BP subtype did not significantly differ after NAC (p=0.392) and had 92% concordance between pre- and post-treated samples. Basal tumors, whether classified before or after NAC was significantly associated with a RFS (HR=4.79, 95% CI 1.09-20.97; p=0.038), DMFS (HR=4.86, 95% CI 1.28-18.64; p=0.018), or OS event (HR=22.03, 95% CI 2.41-201.26; p=0.006) compared to luminal tumors. A significant number (26%; n=10; p=0.002) of tumors had discordant MP risk between pre- and post-treatment samples, all of which reclassified from MP-HR (Luminal B) to MP-LR (Luminal A). MP-HR tumors that reclassified as MP-LR after NAC had better 5-year OS (100%) compared to tumors that remained MP-HR (73.3%, 95% CI 47.2 - 87.9). Conclusions: MP and BP accurately predicted chemosensitivity and outcomes in NAC treated patients. Activated molecular pathways captured by BP classification remained stable after NAC treatment, with a high concordance rate. Approximately 1 out of 4 Luminal B (MP-HR) tumors reclassified as Luminal A (MP-LR) after NAC and have improved 5-year outcome compared to tumors that remained MP-HR. Together, this preliminary evidence supports the prognostic accuracy of MP and BP on tumors following NAC treatment. Future studies will study the effect of NAC on the BC transcriptome and their association with clinical outcomes.

Citation Format: Alice P. Chung, Marissa Srour, Farnaz Dadmanesh, Sungjin Kim, Armando Giuliano, Jennifer Wei, Yen Huynh, Shiyu Wang, Andrea Menicucci, William Audeh. Mammaprint and Blueprint are prognostic of outcome following neoadjuvant chemotherapy [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P4-07-32.

High concordance of 70-gene recurrence risk signature and 80-gene molecular subtyping signature between core needle biopsy and surgical resection specimens in early-stage breast cancer

BACKGROUND AND OBJECTIVES

With increased neoadjuvant therapy recommendations for early-stage breast cancer patients due to the COVID-19 pandemic, it is imperative that molecular diagnostic assays provide reliable results from preoperative core needle biopsies (CNB). The study objective was to determine the concordance of MammaPrint and BluePrint results between matched CNB and surgical resection (SR) specimens.

METHODS

Matched tumor specimens (n = 121) were prospectively collected from women enrolled in the FLEX trial (NCT03053193). Concordance is reported using overall percentage agreement and Cohen's kappa coefficient. Correlation is reported using Pearson correlation coefficient.

RESULTS

We found good concordance for MammaPrint results between matched tumor samples (90.9%, κ = 0.817), and a very strong correlation of MammaPrint indices (r = 0.94). The concordance of BluePrint subtyping in matched samples was also excellent (98.3%).

CONCLUSIONS

CNB samples demonstrated high concordance with paired SR samples for MammaPrint risk classification and BluePrint molecular subtyping, suggesting that physicians are provided with accurate prognostic information that can be used to guide therapy decisions.

Niraparib for Advanced Breast Cancer with Germline BRCA1 and BRCA2 Mutations: the EORTC 1307-BCG/BIG5-13/TESARO PR-30-50-10-C BRAVO Study

PURPOSE

To investigate the activity of niraparib in patients with germline-mutated BRCA1/2 (gBRCAm) advanced breast cancer.

PATIENTS AND METHODS

BRAVO was a randomized, open-label phase III trial. Eligible patients had gBRCAm and HER2-negative advanced breast cancer previously treated with ≤2 prior lines of chemotherapy for advanced breast cancer or had relapsed within 12 months of adjuvant chemotherapy, and were randomized 2:1 between niraparib and physician's choice chemotherapy (PC; monotherapy with eribulin, capecitabine, vinorelbine, or gemcitabine). Patients with hormone receptor-positive tumors had to have received ≥1 line of endocrine therapy and progressed during this treatment in the metastatic setting or relapsed within 1 year of (neo)adjuvant treatment. The primary endpoint was centrally assessed progression-free survival (PFS). Secondary endpoints included overall survival (OS), PFS by local assessment (local-PFS), objective response rate (ORR), and safety.

RESULTS

After the pre-planned interim analysis, recruitment was halted on the basis of futility, noting a high degree of discordance between local and central PFS assessment in the PC arm that resulted in informative censoring. At the final analysis (median follow-up, 19.9 months), median centrally assessed PFS was 4.1 months in the niraparib arm (n = 141) versus 3.1 months in the PC arm [n = 74; hazard ratio (HR), 0.96; 95% confidence interval (CI), 0.65-1.44; P = 0.86]. HRs for OS and local-PFS were 0.95 (95% CI, 0.63-1.42) and 0.65 (95% CI, 0.46-0.93), respectively. ORR was 35% (95% CI, 26-45) with niraparib and 31% (95% CI, 19-46) in the PC arm.

CONCLUSIONS

Informative censoring in the control arm prevented accurate assessment of the trial hypothesis, although there was clear evidence of niraparib's activity in this patient population.

Abstract CT260: The FLEX real-world data platform explores new gene expression profiles and investigator initiated protocols in early stage breast cancer

Background: Genomic expression profiling has significantly improved the personalized treatment of breast cancer, by providing prognostic information about the metastatic potential of tumors and enabling the classification of breast cancers into molecular subtypes, beyond clinical and pathological factors. When combined with comprehensive clinical data, full genome expression data can precisely stratify tumors into clinically actionable subgroups. The FLEX Study is aggregating a large, real-world dataset, which will enable the discovery of novel genomic profiles to improve precision in the management of breast cancer, particularly in patient subsets that are underrepresented in traditional clinical trials. Methods: The FLEX Study (NCT03053193) is a multicenter network, prospective, observational trial for patients with stage I-III breast cancer whose primary tumor is analyzed by MammaPrint, with or without BluePrint. The primary objective of FLEX is to create a large scale, population-based registry that links comprehensive clinical data with full genome expression data to elucidate new prognostic and/or predictive gene associations in a real-world setting. The FLEX Study employs a shared study infrastructure to develop and investigate hypotheses for targeted subset analyses and/or clinical trials based on genomic data. The adaptable protocol is designed to be amended to include additional targeted sub-studies, allowing participating investigators to leverage data within the FLEX network to investigate research questions. Patients enrolled in the initial study are eligible for inclusion in sub-studies for which they meet all eligibility criteria and additional consent is not required. Data is collected on patients from diagnosis through 10 years of follow-up and any necessary additional clinical data will be collected as specified in the appendix protocols. The target enrollment of FLEX is a minimum of 10,000 patients; over 6,500 patients have enrolled since April 2017 at more than 85 sites, including eight National Cancer Institute-designated comprehensive cancer centers. The FLEX collaborative platform allows participating investigators the opportunity to author their own sub-study protocols, as approved by the FLEX Scientific Review Committee. Sub-study research categories include: Age and Breast Cancer, Optimizing Therapy Strategies, Breast Cancer and Metabolic Syndrome, ctDNA and Liquid Biopsy, Genomics and Subtypes, Social and Ancestry, and Neoadjuvant Therapy and Surgery. To date, 29 investigator-initiated sub-studies have been approved. Trial contact information: NCT03053193 FLEX@agendia.com

Citation Format: Adam Brufsky, Jennifer A. Crozier, Nina D'Abreo, Mehran Habibi, Sami Diab, Joyce O'Shaughnessy, Douglas K. Marks, VK Gadi, Amy M. Truitt, Lisa Blumencranz, Erin Yoder, William Audeh, Sarah Untch, Bastiaan van der Baan, FLEX Investigators' Group. The FLEX real-world data platform explores new gene expression profiles and investigator initiated protocols in early stage breast cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr CT260.

Disparities within luminal breast cancer: Clinical and molecular features of African American and non-Hispanic white patients

1009

Background: African American breast cancer patients (AA) are diagnosed younger, have more high-risk features, and poorer clinical outcomes than non-Hispanic White patients (NHW), despite similar treatments. Although comorbidities such as obesity and metabolic syndrome may contribute to differences, ancestry-specific factors and effects of structural violence that disproportionately afflict AA individuals may influence tumor biology and outcomes. We previously reported differentially expressed genes (DEGs) associated with tumor aggressiveness in Basal tumors from AA compared with NHW (Sharma et al., 2020). Here, we compare DEGs in luminal tumors between AA and NHW. Methods: The prospective, observational FLEX study (NCT03053193) includes stage I-III breast cancer patients who receive 70-gene signature (MammaPrint/MP)/80-gene signature (BluePrint/BP) testing and consent to full transcriptome and clinical data collection. AA (n=364) and NHW (n=400, random selection) with BP luminal tumors, enrolled from 2017 to present, were included. Race/ethnicity was self-reported. AA were younger than NHW (mean, 59 vs. 62 years, p=0.001); thus, an age-matched subset (n= 360 AA, NHW) was compared. Differential gene expression analysis was performed with R limma package. Comparisons were made between AA and age-matched or randomly selected NHW in: (1) all, (2) luminal A, (3) luminal B, and (4) luminal B, obese. DEGs with FDR<0.05 were significant. Different fold change (FC) thresholds were evaluated. Results: Compared with age-matched NHW, AA were similar in menopausal status, T stage, grade, and tumor type; obesity, T2DM status, and nodal stage were significantly different ( p<0.01). Tumors from AA were more often MP high risk ( p<0.001), regardless of age matching. Luminal B AA vs. age-matched NHW comparison resulted in more DEGs (n=1070) than other comparisons; however, most were FC<2. Notably, 5/6 DEGs ( PSPH, NOTCH2NL, POLR1A, MAP1LC3P and RPS26P10) in basal tumors (Nunes et al. 2019) were also identified here. Of 9 DEGs (FC>1.7) in the luminal B age-matched comparison, 2 ( PSPH and LINC01139) were also found in the luminal B, obese subset. Consistently upregulated DEGs in AA were associated with metabolism, translation, and cellular stress response pathways. Conclusions: We found significant transcriptomic differences between luminal tumors from AA and NHW, when controlling for age, obesity, and genomic classification. A subset of DEGs in luminal B tumors were consistent with those in Basal tumors, suggesting that similar race-associated factors drive DEGs regardless of tumor subtype. DEGs that may be unique to AA luminal tumors were also found. This study suggests that some biological differences in breast tumors may result from patient ancestry or shared adverse socioeconomic exposures and underscores the need for inclusion of diverse patient groups in clinical trials. Clinical trial information: NCT03053193.

Molecular profiles of genomically high risk ER+ HER2- breast cancer tumors classified as functionally basal or luminal B by the 80-gene signature

563

Background: The 80-gene signature (BluePrint/BP) classifies early-stage breast cancers based on functional molecular pathways as luminal, HER2, or Basal-type. In the NBRST study, 13% of immunochemistry (IHC) defined ER+ HER2- cancers reclassified as Basal-type by the BP assay (ER+ Basal), and these had worse prognosis but responded better to neoadjuvant chemotherapy than ER+ HER2- cancers classified as genomically luminal-type. The 70-gene risk of recurrence signature (MammaPrint/MP) further stratifies luminal-type cancers into low risk luminal A or high risk (HR) luminal B. HR cancers can be further stratified into High 1 (H1) or High 2 (H2), and the I-SPY2 trial has shown higher pCR rates in ER+ cancers classified as H2. Here, we investigated biological differences among ER+ Basal, ER- Basal, H1 luminal B, and H2 luminal B cancers by full transcriptome analysis. Methods: From the FLEX Study (NCT03053193), 1501 breast cancers with known IHC ER status were classified by MP and BP: 103 ER+ Basal, 210 ER- Basal and 1188 luminal B (H1 n=1034, H2 n=154). Clinical factors were assessed by either the Chi-square or Fisher’s exact tests; ANOVA or t test were used to analyze age. Differentially expressed genes (DEGs) were detected using Limma and pathway analyses were performed with GSEA. DEGs with a fold change >2 and FDR < 0.05 were considered significant. Results: Basal-type cancers (ER+/ER-) were larger and higher grade than luminal B cancers. Clustering analysis showed similar transcriptional profiles between ER+ Basal and ER- Basal cancers, distinct from luminal B cancers. Few DEGs were detected between ER+ Basal and ER- Basal cancers, and significantly more DEGs were found between ER+ Basal and luminal B cancers. Only three upregulated genes were detected in ER+ Basal compared to ER- Basal cancers: ESR1 and two immune-related genes ( FDCSP and LTF). Enrichment analysis of DEGs indicated increased immune activation and cell proliferation in ER+ Basal and ER- Basal cancers, and decreased estrogen response between ER+ Basal and luminal B cancers. Enrichment analysis between luminal B H1 and H2 cancers showed H2 cancers had higher immune activation and cell proliferation and lower estrogen response. Conclusions: Reclassification by BP of IHC defined ER+ HER2- cancers identified a subgroup of ER+ cancers that are biologically closer to ER- Basal than luminal-type cancers. Significant differences in response to neoadjuvant chemotherapy that have been seen between ER+ Basal and luminal B breast cancers lend support to the clinical importance of these findings. These data explain the poor prognosis observed in patients with ER+ Basal cancers and suggest that optimized chemotherapy, such as that for triple negative cancer, might be of benefit. BP provides clinically actionable information beyond pathological subtyping, which may guide neoadjuvant treatment recommendations. Clinical trial information: NCT03053193.

Whole transcriptome analysis comparing HR+ HER2- breast cancer tumors from patients < 50 years and >50 years

565

Background: Recent prospective clinical trials have demonstrated a differential chemotherapy effect based on age (≤ 50 vs. > 50 years) or menopausal status (pre- vs. post-) in a genomic low risk group. Whether this is a direct anti-tumor effect of chemotherapy or a secondary ovarian function suppression effect caused by chemotherapy is unclear. We aimed to compare the biological characteristics of breast cancer tumors from patients aged ≤ 50 years and from patients aged > 50 years using whole transcriptome analysis to provide insights into this differential chemotherapy response. Methods: The FLEX Registry (NCT03053193) enrolls stage I-III breast cancer patients who receive 70-gene signature (MammaPrint/MP) test with or without 80-gene signature (BluePrint/BP) test and consent to clinically annotated transcriptome data collection. 3868 patients with HR+HER2- tumors were evaluated, of whom 808 were aged ≤ 50 years and 3060 were aged > 50 years. Clinical risk was assessed based on the MINDACT algorithm. MP classified tumors as low risk (LR) or high risk (HR). HR was stratified to H1 or H2; H2 exhibits a greater chemotherapy response. BP and MP classified tumors as luminal A-, luminal B-, HER2-, or basal-type. Differences in MP, BP, and clinical features were assessed by chi-squared or t test. For gene expression analysis, older patients were randomly selected to obtain an equal sample size as younger patients. Differentially expressed genes (DEGs) were detected using limma and considered significant with FDR <0.05 and fold change ≥ 2. Results: Approximately 70% of patients aged ≤ 50 were pre or peri-menopausal, whereas 90% of patients aged > 50 were post-menopausal. A higher proportion of patients aged ≤ 50 had tumors of high clinical risk (54%) compared to patients aged > 50 (39%) (p < 0.001). Approximately 53% of patients aged ≤ 50 had a HR tumor, of whom 25% classified as H2, while patients aged > 50 had a lower frequency (44%) of HR tumors (p<0.001). Additionally, younger patients had more tumors that classified as BP Luminal B and Basal-type than older patients (p<0.001). Principal component analysis of the top 500 genes with the highest variance revealed no distinct clustering by age group. Accordingly, only 5 DEGs were detected in tumors from patients aged ≤ 50 compared to patients aged > 50, and even fewer DEGs were detected when adjusting for MP risk and BP subtype group. Conclusions: Whole transcriptome analysis identified no substantial differences in gene expression between tumors, including Low Risk Luminal-type tumors, from women aged ≤ 50 (mostly pre or peri-menopausal) and women aged > 50 (mostly post-menopausal). These data support the likely explanation that the observed age-dependent difference in chemotherapy benefit in women ≤ 50 or >50 years of age is not due to intrinsic biological differences in breast cancers due to age, but rather to differences in the effect of chemotherapy on the host. Clinical trial information: NCT03053193.

Gene expression associated with lymphovascular invasion and genomic risk in early-stage breast cancer

559

Background: Lymphovascular invasion (LVI), the passage of carcinoma cells through lymphatic and blood vessels, is an important early step in metastasis; however, LVI is excluded from most breast cancer (BC) clinical risk assessments. Previous studies assessed the prognostic value of LVI to estimate clinical outcomes. To gain understanding of the molecular basis of LVI, we evaluated differentially expressed genes (DEGs) between tumors with LVI versus those without LVI, stratified by the 70-gene signature (MammaPrint/MP) and 80-gene molecular subtyping signature (BluePrint/BP). Methods: The prospective, observational FLEX Study (NCT03053193) includes stage I-III BC patients who receive MP/BP testing and consent to full transcriptome and clinical data collection. Patients with LVI (n=581) and without LVI (n=600, randomly selected), enrolled from 2017 to present, were included. LVI was assessed by local pathology laboratories. Differential gene expression analysis of 44k Agilent microarray data was performed with R limma package. DEGs were compared within all samples, BP Luminal subtype, MP risk groups (Low Risk [LR]/Luminal A and High Risk [HR]/Luminal B), and by lymph node (LN) status. DEGs with FDR<0.05 were considered significant. Results: Of tumors with LVI (LVI+), 66% were MP HR; notably, 51% of tumors without LVI (LVI-) were MP HR. LVI was associated with larger T stage, LN involvement, high grade, negative ER status by IHC, and younger patient age (LVI+ vs. LVI-, p<0.05 for all comparisons). Patient ethnicity, obesity, and tumor type did not differ by LVI status; however, prevalence of type 2 diabetes trended higher in patients with LVI+ HR tumors (21%), compared with LVI- HR (15%, p=0.09) and LVI+ LR (11%, p=0.004). There were significant transcriptomic differences between LVI+ and LVI, with most DEGs evident in the Luminal B subset. DEGs in LVI+, LN-negative (LN-) tumors overlapped substantially with the overall Luminal group analysis. Functional enrichment analysis showed dysregulation of cell cycle, extracellular matrix (ECM) organization, cell adhesion, and cytokine receptor pathways. Gene sets related to insulin growth factor pathways were also enriched in LVI+ tumors. Conclusions: DEGs associated with LVI were primarily found in MP HR Luminal, LN-negative tumors; enrichment analysis suggested dysregulation of ECM organization and cell adhesion pathways, consistent with previous reports. DEGs were not associated with LVI presence in LN+ tumors, suggesting that LVI assessment may be less relevant in LN+ breast cancer. Future studies will assess clinical outcomes, as well as LVI-associated gene expression in BP Basal- and HER2-type tumors. However, the current analysis indicates few DEGs in LVI+ MP LR tumors; thus, the potential prognostic information gained from LVI-associated gene expression is likely already captured by the MP and BP signatures. Clinical trial information: NCT03053193.

Genomic risk classification by the 70-gene signature and 21-gene assay in African American early-stage breast cancer patients

e12568

Background: Genomic risk of recurrence assays provide prognostic information that aids treatment recommendations for patients with early breast cancer (EBC). The 70-gene signature (MammaPrint/MP) and the 21-gene assay (Oncotype DX/ODX) were both validated by large prospective, randomized clinical trials. Previous studies have directly compared the performance of these assays, with moderate discordance widely observed. Recent work showed poor prognostic performance of the ODX in African American (AA) EBC patients (Hoskins et al., 2021). Here, we compare the concordance of two genomic tests in an AA cohort. Methods: This retrospective analysis included a subset of self-reported AA patients from the prospective, observational PROMIS (n = 66, NCT01617954) and FLEX (n = 29, NCT03053193) registry trials, for whom both MP and ODX RS results were available. Patients were enrolled from 2012 to present. The PROMIS trial enrolled patients with intermediate ODX RS (18-30). MP results were obtained from standard testing performed centrally by Agendia, Inc. (Irvine, CA); ODX testing was performed by Genomic Health, Inc. (Redwood City, CA) and results were collected from patient case report forms. Results: AA patients (n = 95) included in this sub-analysis were 77% post-menopausal with a median age of 60 years. 28% of patients were aged ≤ 50 years. Of patients with metabolic data (n = 49), 63% were obese by body mass index (BMI > 30), and 23% had type 2 diabetes mellitus (T2DM). Tumors were predominantly invasive ductal carcinoma (82%), T1 (73%), grade 2 (53%), ER+ (99%), HER2-negative (94%), lymph node-negative (86%), and MP High Risk (66%). By ODX clinical RS, 17% of tumors classified as low (0-17), 80% intermediate (18-30), and 3% high (31-100) risk. As expected, the large intermediate RS group reflects the inclusion of PROMIS patients. By TAILORx RS ranges, 76% of tumors were low risk (≤25). Of these, 61% (n = 44/72) classified as MP High Risk. Notably, 62% (n = 41/66) of tumors with TAILORx intermediate RS (11-25) were MP High Risk. Conclusions: The overall discordance between MP and ODX was 51% in these AA patients. Discordance was more frequent (61%) in ODX low risk (RS≤25) cases. Notably, of tumors with TAILORx intermediate RS (11-25), the majority (62%) classified as MP High Risk. Combined with previously published data in AA patients (Nunes et al., 2016), this results in a total of 57% (n = 52/91) of ODX low RS (≤25) tumors classified as MP High Risk. Recent data indicate that AA patients who receive a low (≤10 or ≤25) or intermediate RS (11-25) have higher recurrence rates and lower survival than White/Caucasian EBC patients with the same RS. In the current study, clinical outcomes are forthcoming; however, these data suggest the majority of AA patients are more likely to receive a MP High Risk result, which may capture the diversity of pathways driving tumor metastasis. Clinical trial information: NCT01617954, NCT03053193.

Abstract PS6-41: Mammaprint and blueprint as prognostic indicators for elderly patients with early stage breast cancer

Background: Elderly breast cancer (BC) patients are an understudied population, with limited evidence regarding treatment options and outcomes and a lack of research involving prognostic multigene assays for this group. One study in patients 65-89 years old with an Oncotype DX Recurrence Score ≥ 26 concluded that gene expression profiling tests have limited utility in elderly patients, and should only be used for patients aged 65-74 with no/low to moderate comorbidities and not for patients ≥ 75. In this study, the 70-gene risk of distant recurrence signature, MammaPrint (MP), and 80-gene molecular subtyping signature, BluePrint (BP), were evaluated in both the neoadjuvant and adjuvant settings in elderly patients with early stage BC.

Methods: This analysis included 211 BC patients classified as cT2-4N0-3M0 (T2 &gt; 3.5 cm if N0) who received neoadjuvant chemotherapy and enrolled in the Multi-Institutional Neoadjuvant Therapy MammaPrint Project (MINT) study from 2011-2016. Lymph node (LN) involvement was established following neoadjuvant treatment. The second analysis included 517 early stage BC patients with 0-3 positive LNs who enrolled in the community based cohort study (COPPER) from 2009-2016. Patients were given adjuvant treatment following standard of care. Patients from both cohorts were divided into age at diagnosis groups: &lt; 65, 65-74, and &gt; 74. MP stratified patients into either Low Risk (LR) or High Risk (HR) groups. BP classified patient samples into Luminal, HER2, or Basal subtype. Kaplan Meier analysis and log-rank test were used to assess differences in overall survival (OS) and distant metastasis free survival (DMFS). Clinical risk assessment based on the MINDACT trial algorithm was performed.

Results: From MINT, 35 patients were ≥ 65 years old; 80% were HR and 20% were LR. Pathological complete response (pCR) was achieved in 36% (10/28) of elderly HR patients, of whom 70% were HER2 and 30% were Basal by BP. Nodal downstaging occurred in 55% (11/20) of LN positive elderly HR patients, of whom 64% (7/11) achieved pCR. BP classified patients with nodal downstaging as HER2 (55%), Basal (36%), or Luminal (9%). Importantly, pCR and nodal downstaging were more likely to be achieved in HR tumors and correlated with BP subtype in both young and elderly patients. From the COPPER cohort, 77% of HR patients 65-74 years old received chemotherapy (CT), whereas 74% of LR patients omitted CT. Of patients &gt; 74, 49% of HR patients received CT, whereas 75% of LR patients omitted CT. OS and DMFS probabilities indicated good survival outcomes in LR patients that omitted CT and HR patients that received CT, with no significant difference between age groups. A majority of HR patients treated with CT and over 1/3 of LR patients that omitted CT had high clinical risk. Interestingly, among all patients that had a metastasis event, mortality was less likely to occur in patients that received dose dense AC (doxorubicin and cyclophosphamide).

Conclusion: MP and BP may identify HR elderly patients who are likely to achieve nodal downstaging and pCR. Elderly patients were safely spared or assigned adjuvant CT based on MP results independent of clinical risk. Furthermore, these data are in line with previous studies that suggest similar survival benefits between older and younger patients who are candidates for aggressive CT regimens. MP and BP elucidate information about tumor biology and provide prognostic value, which may help inform treatment decisions, independent of patient age.

MINTAge group&lt; 6565-74&gt; 74MP resultHRLRHRLRHR# of patients152242177% of patients with pCR35% (53/152)033% (7/21)043% (3/7)# of LN+ patients103181456% of LN+ patients with nodal downstaging49.5% (51/103)22% (4/18)50% (7/14)067% (4/6)% of LN+ patients with pCR & nodal downstaging65% (33/51)071%(5/7)050%(2/4)COPPERAge group&lt; 6565-74&gt; 74MP resultHRLRHRLRHRLR# of patients1409988665569# of patients received CT121256813278# of patients omitted CT116717492052# of patients with unknown treatment873489Groups treated based on MPHR treated with CTLR omitted CTAge group&lt; 6565-74&gt; 74&lt; 6565-74&gt; 745-yr DMFS probability (95% CI)91% (80.2-96.7)87% (60.2-91.4)87% (55.2-96.6)100%98% (84.3-99.7)94% (75.9-98.5)5-yr OS probability (95% CI)94% (80.2-98.2)96% (83.4-98.9)86% (54.7-96.5)100%98% (84.3-99.7)97% (80.4-99.6)% Clinical high risk83% (100/121)76% (52/68)63% (17/27)34% (23/67)37% (18/49)35% (18/52)

Citation Format: Peter W. Blumencranz, Mehran Habibi, Lisa Blumencranz, Andrea Menicucci, Shiyu Wang, Amy Truitt, William Audeh, Jolanta L. Baginski, Steven Shivers, Geza Acs, Charles E. Cox, MINT Investigators Group. Mammaprint and blueprint as prognostic indicators for elderly patients with early stage breast cancer [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS6-41.

Abstract PS18-03: Differential gene expression in luminal-type invasive lobular carcinoma and invasive ductal carcinoma by MammaPrint risk stratification

Background: Invasive lobular carcinoma (ILC) comprises 10-15% of breast tumors and is the second most common histological type after invasive ductal carcinoma (IDC). Patients with ILC are often diagnosed at an older age and more advanced stage than those with IDC. Late recurrences and worse long-term survival suggest the need for improved approaches to treatment optimization and exploration of molecular pathways unique to ILC. Although previous reports have described comprehensive transcriptomic profiling of ILC, these were limited by small sample sizes. Furthermore, differential gene expression between ILC and IDC within genomic risk groups and molecular subtypes has yet to be explored. Here we characterize differential gene expression between ILC and IDC in a large, age-matched patient subset categorized by 70-gene signature/MammaPrint (MP) risk and 80-gene signature/BluePrint (BP) subtype.

Methods: The prospective FLEX Registry (NCT03053193) includes stage I-III primary invasive breast cancer patients who receive MP/BP testing and consent to full transcriptome and clinical data collection. This sub-analysis included 450 ILC patients enrolled from 2017 to present. Compared with a random selection of IDC patients (n=450, mean age, 60 years), ILC patients were older (mean, 63 years, p&lt;0.001). Thus, we selected an age-matched subset for differential gene expression analysis. There were few non-Luminal ILCs; thus, gene expression analyses were limited to BP Luminal tumors. A subset of 413 age-matched pairs (n=826) of ILC and IDC were used for analysis. Gene expression data were quantile normalized using R limma package, and differentially expressed genes (DEGs) were compared between groups. DEGs with an adjusted p&lt;0.05 and log2 fold change &gt; ± 1.0 were considered significant.

Results: ILC represented 13% of FLEX cases (n=450/3562), and were 81% lymph node-negative, 99% ER+, 94% HER2-negative, and 68% MP Low Risk (LR). By BP, ILC were 99% Luminal, 1% HER2, and &lt;1% Basal type. BP Luminal ILC were predominantly grade 2 (63%), T1 (61%), node-negative (84%), and MP LR (69%). Menopausal status, nodal status, ethnicity, BMI distribution, and frequency of type 2 diabetes mellitus were similar between ILC and IDC. However, IDC were more likely to be MP HR (46% IDC vs. 31% ILC, p&lt;0.001) and grade 3 (15% IDC vs. 4% ILC, p&lt;0.001). ILC were more likely to be T3 (10% ILC vs. 1% IDC, p&lt;0.001). We found 4 DEGs common to all comparisons: all Luminal ILC vs. IDC, MP LR ILC vs. IDC, and MP HR ILC vs. IDC. ILC had lower expression of CDH1 (E-cadherin) than IDC, regardless of MP risk. Including CDH1, 6 unique genes were differentially expressed in LR ILC compared with IDC, and 21 genes were differentially expressed in HR ILC compared with IDC. Genes with increased expression in HR ILC were related to immune cell migration/chemotaxis, hormone signaling, and growth factor signaling. HR ILCs were also enriched for TGFβ signaling and angiogenesis pathway genes.

Conclusions: Here we report differential clinical and molecular characteristics between ILC and IDC in a large, age-matched patient subset. Regardless of MP risk, expression of CDH1 was lower in ILC compared with IDC. Approximately one-third of ILCs were MP HR, and we report a greater number and diversity of DEGs between HR ILC and HR IDC compared with LR tumors, in particular genes related to TGFβ signaling. TGFβ pathway genes play a variety of roles in the tumor microenvironment, including induction of angiogenesis, fibroblast growth factor stimulation, and inhibition and/or exclusion of an immune response. These results suggest that therapeutic strategies targeting the TGFβ pathway may be future avenues of exploration in ILC, although further studies are warranted to characterize underlying molecular mechanisms.

Citation Format: Beth-Ann Lesnikoski, Jennifer A. Crozier, Gordan Srkalovic, Patricia Robinson, Clodia Osipo, Kaylan Banda, Heather M. Kling, Josien Haan, William Audeh, FLEX Investigators Group. Differential gene expression in luminal-type invasive lobular carcinoma and invasive ductal carcinoma by MammaPrint risk stratification [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS18-03.

Abstract PS6-19: Comparing MammaPrint and BluePrint results between core needle biopsy and surgical resection breast cancer specimens

Background: Pre-operative/neoadjuvant treatment utilization in early-stage breast cancer has been increasing, particularly during the COVID-19 pandemic. With goals of minimizing potential exposure to SARS-COV-2, as well as resource rationing, physicians are urged to triage breast cancer patients by identifying those that require urgent surgical care vs. those who may delay surgical treatment.1 Accurate risk assessment is an integral component of this triaging process, which has recommended that genomic testing on diagnostic core needle biopsy (CNB) samples be used to assist with the identification of patients with low risk tumor biology who may be candidates for surgical delay.1 MammaPrint (MP), a 70-gene risk of recurrence signature, and BluePrint (BP), an 80-gene molecular subtyping signature, have been routinely used in formalin-fixed paraffin embedded (FFPE) CNB and surgical resection (SR) samples since MammaPrint obtained FDA clearance for FFPE tissue in 20152. In addition, over 1,500 CNB tumor samples from patients enrolled in prospective neoadjuvant treatment trials (NBRST and ISPY-2) have received successful MP & BP testing. This study compares the gene expression results between CNB and SR specimens to better elucidate how these tests perform across specimen type. Methods: Routine diagnostic samples submitted to Agendia, Inc. (Irvine, CA) between Feb 2017 and May 2020 for MP and BP testing were processed according to standard FFPE microarray procedures. MP was used to stratify samples into Ultra Low Risk (UL), Low Risk (LR), and High Risk (HR). BP was used to classify samples into Luminal, HER2, or Basal-type. This study included 13,603 CNB and 25,684 SR specimens. MP Index (MPI) distribution on BP defined Luminal-type tumors were compared between CNB and SR samples. Comparative “logistics metrics” (average turnaround time [TAT] and success rate) were also assessed between these specimen types. Results: Of the 39,287 samples included in this analysis, 35% were CNB and 65% were SR (Table). BluePrint Luminal, HER2 and Basal-type distributions were 86%, 4%, and 10% respectively for CNB samples and 94%, 1%, and 5% respectively for SR samples. Within BP defined Luminal-type tumors, the frequency of UL, LR, and HR results were 18%, 58%, and 42% for CNB, and 16%, 60%, and 40% for SR, respectively. Overall, MP Index distributions were similar between samples tested from CNB vs. SR. Average TAT between CNB and SR were 4.52 and 4.55 days, respectively. For specimens that met the minimum tumor % threshold, successful testing rates for CNB and SR were 97.5% and 98.4%, respectively. Conclusions: MP and BP testing were successfully performed on both CNB and SR samples in approximately 98% of all eligible specimens with rapid TAT allowing for timely pre-operative decision-making. The frequency of each MP risk group as well as the distribution pattern of MP Index were nearly identical between CNB and SR specimens, indicating comparable performance regardless of specimen type. With no meaningful difference in MPI distribution, TAT or success rate between CNB specimens and SR specimens, pre-operative use of MP+BP genomic testing is feasible, in alignment with recent COVID-19 pandemic guidelines.

References1.Dietz, J.R., et al. Breast Cancer Res Treat 181, 487-497 (2020).2.FDA 510(k) Clearance K141142, January 2015.

Table:(n)(%)(n)(%)Total CNB13,60335%Total SR25,68465%CNB Ultralow2,17516%SR Ultralow3,89815%CNB Low Risk6,90051%SR Low Risk14,65657%CNB High Risk6,70349%SR High Risk11,02843%(n)(%)(n)(%)Luminal-type CNB10,05738%Luminal-type SR16,31162%CNB Ultralow1,82418%SR Ultralow2,58816%CNB Low Risk5,78558%SR Low Risk9,77160%CNB High Risk4,27242%SR High Risk6,54040%

Citation Format: Joseph McKelley, Jennifer Wei, Brian Hoxeng, Andrea Menicucci, Erin Yoder, Shiyu Wang, William Audeh. Comparing MammaPrint and BluePrint results between core needle biopsy and surgical resection breast cancer specimens [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS6-19.

Abstract PS18-05: Using blueprint to elucidate the molecular heterogeneity of triple negative breast cancers

Background: Triple negative breast cancers (TNBC) are more aggressive, have a worse prognosis, and few targeted therapies compared to other BC subtypes. TNBC is molecularly heterogeneous, with at least 4 distinct subtypes: basal-like immune-activated (BLIA), basal-like immunosuppressed (BLIS), luminal androgen receptor (LAR), and mesenchymal (MES). The molecular subtyping gene signature, BluePrint (BP), classifies breast tumors into Luminal, HER2, or Basal subtype based on the assessment of downstream signaling pathways and independently of IHC expression. Compared to IHC-defined TNBC, a higher frequency of BLIS or BLIA subtypes and fewer LAR or MES tumors were reported in BP-defined Basal tumors. To advance our understanding of TNBC heterogeneity, we evaluated the relationship between gene expression signatures, TNBC subtype and BluePrint, in IHC-defined TNBC.

Methods: The FLEX registry (NCT03053193) is an ongoing, prospective study evaluating primary tumors from stage I-III BC patients who receive the risk of distant recurrence gene signature, MammaPrint (MP), and BP testing and consent to clinically annotated full transcriptome data collection. This analysis includes 204 IHC-defined TNBC patients. TNBC subtypes BLIA, BLIS, LAR, and MES were derived using an adjusted version of the Burstein centroid signature. BP classified patient samples into Luminal, HER2, and Basal subtypes. A proportion of tumors may exhibit a secondary but less pronounced activated pathway or BP subtype. Therefore, each BP subtype was divided into single activated or mixed subtype based on BP indices.

Results: Of 204 TNBC tumors, 84% were classified as Basal by BP, most of which were BLIS (65%), followed by BLIA (22%), with a low frequency of MES (8%) and LAR (5%) subtypes (Table). Approximately 14% of TNBCs were reclassified as Luminal by BP, most of which were LAR (76%), whereas 24% were MES. Clustering analysis revealed similar gene expression profiles between Basal-BLIS and Basal-BLIA tumors. Interestingly, the transcriptional profile of Basal-MES and Basal-LAR tumors were similar to Luminal-MES and Luminal-LAR tumors. BP Basal indices distinguished between different TNBC subtypes. The Basal pathway was predominantly activated in 90% of BP Basals (single activated tumors), most of which were either BLIS or BLIA (96%), whereas 10% of BP Basals were mixed subtype and more likely to classify as LAR (53%) or MES (35%). Approximately 25% of the BP Basal gene signature overlapped with the TNBC subtype gene signature. Expression of 18 and 12 genes out of 28 genes that make up the BP basal signature were significantly different in Basal-BLIA/BLIS compared to LAR or MES, respectively (P &lt; 0.05). PRR15 and CAPN13 were significantly differentially expressed between LAR and MES within Basals.

Conclusion: BP reclassified a subgroup of TNBC tumors to Luminal, explaining the discrepancy in the distribution of TNBC subtypes between IHC-defined TNBC and BP Basal tumors. Furthermore, BP indices distinguished between single activated and mixed subtypes, which correlated with different TNBC subtypes. These data suggest that molecular classification by BP adds further precision in classifying TNBC patients and sheds new light on the heterogeneity of these tumors. These findings have clinical implications in stratifying patients and identifying successful targeted treatment options. Future studies are warranted to investigate treatment response and prognosis in these molecular subgroups.

BasalLuminalHER2TotalBLIA370037BLIS11100111LAR1022335MES147021Total172 (84.3%)29 (14.2%)3 (1.5%)204 (100%)

Citation Format: Virginia G. Kaklamani, Cathy Graham, Karen L. Tedesco, Abirami Sivapiragasam, Jennifer A. Crozier, Apurva N. Shah, Yuan Yuan, Josien Haan, Andrea Menicucci, Michelle L. Bolner, Shiyu Wang, Lorenza Mittempergher, Erin Yoder, William Audeh, FLEX Investigators' Group. Using blueprint to elucidate the molecular heterogeneity of triple negative breast cancers [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS18-05.

Abstract OT-12-01: The FLEX real-world data platform explores new gene expression profiles and investigator-initiated protocols in early stage breast cancer

Background: Genomic expression profiles have implications for the personalized treatment of breast cancer beyond clinical and pathological features by enabling the classification of breast cancers into molecular subtypes and providing prognostic information about the metastatic potential of tumors. However, full genome expression data should be combined with comprehensive clinical information to precisely stratify tumors into clinically actionable subgroups. The FLEX Registry aims to aggregate a large, real-world dataset, which will enable the discovery of novel genomic profiles to improve precision in the management of breast cancer, particularly in underrepresented patient subsets in traditional clinical trials. Trial Design: The FLEX Registry (NCT03053193) is a multi-center, prospective, observational trial for patients with stage I-III breast cancer whose primary tumor is analyzed by MammaPrint, with or without BluePrint.The primary objective of FLEX is to create a large scale, population-based registry that links comprehensive clinical data with full genome expression data to elucidate new prognostic and/or predictive gene associations in a real-world setting. The FLEX Registry employs a shared study infrastructure to develop and investigate hypotheses for targeted subset analyses and/or clinical trials based on full genome expression data. The adaptable protocol is designed to be amended with the inclusion of additional targeted sub-studies. Patients enrolled in the initial study are eligible for inclusion in sub-studies for which they meet all eligibility criteria and additional consent is not required. Data will be collected on patients from diagnosis through10 years of follow-up and any necessary additional clinical data will be collected as specified in the appendix protocols. The target enrollment of FLEX is a minimum of 10,000 patients; over 5,000 patients have enrolled sinceApril 2017 at more than 85 sites, including eight National Cancer Institute-designated comprehensive cancer centers. The FLEX collaborative platform allows participating investigators the opportunity to author their own sub-study protocols, as approved by the FLEX Review Committee. Sub-study research categories include: Age and Breast Cancer, Optimizing Therapy Strategies, Breast Cancer and MetabolicSyndrome, ctDNA and Liquid Biopsy, Genomics and Subtypes, Social and Ancestry, and Neoadjuvant Therapy and Surgery. To date, twenty-five investigator-initiated sub-studies have been approved.Trial contact information: NCT03053193FLEX@agendia.com

Citation Format: Laura Lee, Sami Diab, Julie Barone, Jennifer A Crozier, Margret Chen, Rakshanda L Rahman, Robert Maganini, Douglas Marks, Amy M Truitt, Lisa Blumencranz, Erin Yoder, Sarah Untch, William Audeh, Bastiaan van der Baan, FLEX Investigators Group. The FLEX real-world data platform explores new gene expression profiles and investigator-initiated protocols in early stage breast cancer [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr OT-12-01.

Abstract PS7-68: Racial disparities within basal-type breast cancer: Clinical and molecular features of African American and Caucasian obese patients

Background: African American breast cancer patients (AA) are diagnosed at a younger age and present more frequently with triple-negative/Basal tumors than Caucasian American patients (CA). High prevalence of obesity, type 2 diabetes mellitus (T2DM), and metabolic syndrome in AA may confound attempts to evaluate the influence of race on gene expression. Previously we showed that differentially expressed genes (DEGs) between AA and CA within the Basal subtype were related to metabolism, translation, and cell signaling pathways (Nunes et al. 2019). However, AA had higher obesity and T2DM rates than CA, and we were unable to distinguish between the influence of metabolic factors and race. In the current analysis, we aim to dissect these factors by comparing clinical and molecular features of Basal-type breast tumors in obese AA and CA.

Methods: The prospective, observational FLEX Registry (NCT03053193) includes stage I-III breast cancer patients who receive 70-gene signature (MammaPrint, MP)/80-gene signature (BluePrint, BP) testing and consent to full transcriptome and clinical data collection. This interim substudy included 50 AA and 96 CA (n=146), enrolled from 2017 to present, all obese by body mass index (BMI, ≥30) and whose tumors were MP High Risk and BP Basal subtype. AA were significantly younger (mean, 55 years) than CA (mean, 60 years, p=0.02); thus, an age distribution-matched subset (n=49 AA, n=49 CA) was added for comparison. Gene expression data were quantile normalized using R limma package; DEGs were compared between groups in the following: (1) all AA (n=50) and CA (n=96), (2) AA and 3 random selections of CA (n=50 pairs), and (3) age-matched AA and CA (n=49 pairs).

Results: Clinical factors, including tumor stage, nodal stage, and T2DM status were similar between AA and CA, regardless of age-matching. Most tumors were T1/2 (83% AA, 88% CA) and negative for nodal involvement (77% AA, 68% CA). 94% of tumors from AA and 74% of tumors from CA were grade 3 (p=0.17). Notably, 32% of tumors from AA and 46% of tumors from CA were ER+ by immunohistochemistry. Age-matched AA and CA had a 20% rate of T2DM. 152 DEGs were significant (adjusted p&lt;0.05) in at least one comparison, with 115 genes more highly expressed in AA and 37 genes more highly expressed in CA. Across all comparisons, 6 genes were consistently more highly expressed in AA: PSPH, NOTCH2NL, POLR1A, AC069240.1, ORAI1, and RPS26P10. Except ORAI1, these genes were also found in the previous comparison between Basal-type AA and CA, and the current analysis confirmed 11/16 DEGs previously reported (Nunes et al. 2019). Genes more highly expressed in AA are associated with transcription, angiogenesis, and Notch signaling pathways, as well as breast cancer aggressiveness and treatment resistance.

Conclusions: Higher prevalence of obesity/T2DM in AA has been proposed as a key factor to explain racial disparities in breast cancer incidence and prognosis, but the current results suggest that race may influence DEGs more than differences in tumor subtype, age, or metabolic factors. This comparison also emphasizes the importance of matched clinical features for DEG analysis and suggests disparities in AA beyond those attributable to clinical differences within the population. DEGs in AA suggest upregulation of Notch-associated aggressiveness, which may be particularly relevant under hypoxic conditions (e.g., obesity), and pathways associated with stemness, metastasis, and chemotherapy resistance. Notch pathway also interacts with key oncogenic pathways, and future studies will reveal the molecular networks underlying racial disparity in AA and CA breast cancer patients.

Citation Format: Dipali Sharma, Lisa E. Blumencranz, Heather M. Kling, Sahra Uygun, Sarah Untch, William Audeh, Jennifer A. Crozier, Mehran Habibi, Raquel Nunes. Racial disparities within basal-type breast cancer: Clinical and molecular features of African American and Caucasian obese patients [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS7-68.

Abstract PS7-69: Molecular profiles and clinical-pathological features of Asian early-stage breast cancer patients

Background/Objective: Breast cancer incidence in Asian populations has increased in recent years, and variation in prognosis and tumor subtypes indicates that further study is warranted to characterize these differences and identify actionable targets. Patients of Asian ancestry are underrepresented in US registries, and few studies have characterized molecular profiles for these patients. In the current analysis, we assess clinical, pathological, and molecular profiles from self-reported Asian breast cancer patients (AS), in comparison with age-matched Caucasian (CA) and African American patients (AA), to evaluate the influence of Asian ancestry on differential gene expression in breast tumors.

Methods: This meta-analysis included cohorts of self-reported AS, CA, and AA with early-stage, invasive breast cancer (EBC) prospectively enrolled in the US from 2011 to 2020 in FLEX (NCT03053193), MINT (NCT01501487), or IMPACt (NCT02670577) trials. AS were significantly younger (mean, 55 years) than CA (mean, 61 years, p&lt;0.001) or AA (mean, 59 years, p=0.005); thus, an age-matched subset was selected for analyses. 70-gene signature (MammaPrint, MP), 80-gene signature (BluePrint, BP), and clinical-pathological features were compared among age-matched AS (n=103), CA (n=103), and AA (n=100). Whole-genome expression data were quantile normalized using R limma package, and differentially expressed genes (DEGs) were compared among AS (n=90), CA (n=102), and AA (n=96). DEGs with adjusted p-value &lt;0.05 and log2 fold change &gt; ± 0.5 were considered significant.

Results: AS tumors were classified as 59% MP HR, compared with 44% HR in age-matched CA (p=0.08) and 64% HR in age-matched AA (p=0.17). AS had a significantly lower rate of obesity (16%, body mass index ≥30) compared with CA (41%) and AA (67%) (p&lt;0.001). Tumors of AS were predominantly ductal carcinoma (84%), T1 (59%), grade 1 or 2 (70%), lymph-node negative (69%), ER+ (95%), and HER2-negative (89%). Distribution of ER, PR, and HER2 pathology and BP subtypes for AS were similar to CA but significantly different from AA (Table). Histologic tumor type, tumor grade, tumor stage, nodal stage, menopausal status, and frequency of Type 2 diabetes mellitus were not significantly different between AS and CA or AA. Whole-genome expression comparisons revealed 19 significant DEGs between AS and CA, and 45 significant DEGs between AS and AA. Immune-related genes, primarily those involved with B cell responses and signaling, were more highly expressed in AS compared with CA. Expression of genes related to cell-cycle pathways was greater in AS compared with AA.

Conclusions: AS were significantly younger and more often pre/peri-menopausal at diagnosis compared with CA and AA, consistent with the literature. Most clinical-pathological factors were similar between age-matched groups, except for the obesity rate, which was significantly lower in AS than in CA or AA. Although not significant, AS had EBC that was more often MP HR than CA and less often HR than AA; studies with larger patient groups will help confirm these trends. The current analysis revealed different underlying gene expression pathways in AS compared with other ethnic groups, which may result in differential clinical outcomes. As genomic profiling data are not widely available for Asian American EBC patients, further analyses are warranted to elucidate these outcomes and identify appropriate therapeutic strategies.

Pathology and Genomic Results(unknowns excluded)Asian (n=103)Caucasian (n=103)African American (n=100)p-valueAS vs. CAp-value AS vs. AAER status (IHC)ER Positive69 (94.5%)89 (98.9%)73 (80.2%)0.1250.010ER Negative4 (5.5%)1 (1.1%)18 (19.8%)PR status (IHC)PR Positive64 (87.7%)81 (90%)63 (69.2%)0.4110.006PR Negative9 (12.3%)9 (10%)28 (30.8%)HER2 (IHC/FISH)HER2 Positive3 (4.2%)3 (3.4%)13 (14.3%)0.1720.017HER2 Negative64 (88.9%)85 (95.5%)77 (84.6%)Equivocal5 (6.9%)1 (1.1%)1 (1.1%)MP/BP resultsLuminal A39 (40.6%)51 (53.7%)28 (30.4%)0.2320.043Luminal B45 (46.9%)38 (40.0%)42 (45.7%)HER2 (MP HR)7 (7.3%)4 (4.2%)5 (5.4%)Basal (MP HR)5 (5.2%)2 (2.1%)17 (18.5%)

Citation Format: Margaret Chen, Ava Kwong, Carolyn Hendricks, Nina D'Abreo, Laura Lee, Hatem H. Soliman, Charles Cox, Heather M. Kling, Rajith Bhaskaran, Shiyu Wang, Andrea Menicucci, Sarah Untch, William Audeh, FLEX Investigators Group. Molecular profiles and clinical-pathological features of Asian early-stage breast cancer patients [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS7-69.

Abstract PS14-11: Differential gene expression analysis and clinical utility of MammaPrint and BluePrint in male breast cancer patients

Background: Male breast cancer (MaBC) is rare, comprising &lt;1% of all breast cancers in the United States. The low incidence of MaBC limits the ability to conduct clinical trials specifically for this population. Due to the paucity of research on MaBC, current understanding regarding MaBC biology, pathology, and treatment strategies has been primarily based on evidence extrapolated from research on female breast cancer (FBC) patients. Traditional diagnostic biomarkers such as ER, PR, and HER2, as well as newer multi-gene prognostic signatures, are employed when making treatment decisions for both MaBC and FBC. However, limited empirical data is available to support the use of identical laboratory biomarkers and molecular signatures in both MaBC and FBC. The 70-gene risk of distant recurrence signature, MammaPrint (MP), and the 80-gene molecular subtyping signature, BluePrint (BP), are commonly used to help make treatment decisions for both MaBC and FBC patients. To support the clinical utility of MP and BP in MaBC, this study aims to elucidate whether significant molecular biological differences exist between MaBC and FBC. To address this knowledge gap, we evaluated and compared 1) MP index results within Low Risk (LR) and High Risk (HR) groups, 2) MP and BP gene expression, and 3) differentially expressed genes within the full genome and their associated biological pathways between tumors from MaBC and FBC.

Methods: This analysis included a total of 817 breast tumor samples sent to Agendia, Inc. (Irvine, CA) for MP and BP testing. Full-transcriptome microarray data were available for 1) a subset of 400 post-menopausal FBC patients enrolled in the FLEX Registry (NCT03053193) and 2) 80 MaBC pateints, 32 of whom enrolled in the FLEX registry and 48 non-study patients for whom data were limited to metadata and quality metrics routinely captured for diagnostic testing. Data from all patients were de-indentified.

Differences in mean MP indices between FBC (N=400) and MaBC (N=417) according to MP Risk classification (LR or HR) were analyzed using a Z-test. Differential gene expression analysis was performed using the R-limma package in which gene expression data were quantile normalized. Pathway analyses were performed using GOseq. Differentially expressed genes (DEGs) were identified between FBC (N=400) and MaBC (N=80) for whom full transcriptome microarray data were available. DEGs were defined as those with a fold change of &gt; 2 and an adjusted P value of &lt; 0.05.

Results: All patients in this study had hormone positive, HER2 negative early-stage breast cancer. There was no statistical difference in the average MP index between MaBC and FBC classified as MP LR (P=0.273) or those classified as MP HR (P=0.692). Gene expression comparison revealed 166 DEGs between MaBC (N=80) and FBC (N=400), 99 DEGs between MP HR MaBC (N=42) and MP HR FBC (N=200), and 290 DEGs between MP LR MaBC (N=38) and MP LR FBC (N=200). Pathway analyses revealed that downregulated genes in MaBC compared to FBC enriched to immune-related functions, including B-cell mediated immunity, whereas upregulated genes were associated with hormone metabolic processes. In all comparisons, expression of MP or BP genes was not significantly different.

Conclusions: We found similar MP index distributions between MaBC and FBC. Importantly, differential gene expression between MaBC and FBC provides novel insight into the mechanisms underlying MaBC. Although these data reveal biological distinctions between male and female breast cancer, MP and BP assay performance is preserved across both groups. Further studies are needed to assess clinical outcomes; however, these findings support the use of MP risk of recurrence assay and BP molecular subtyping assay for prognosis and informing treatment decisions in MaBC.

Citation Format: Jennifer Crozier, Julie Barone, Kalyan Banda, Beth Lesnikoski, Robert Maganini, Sami Diab, Ian Grady, Thomas Lomis, Charles Cox, Amy Truitt, Benjamin Nota, Andrea Menicucci, Erin Yoder, William Audeh, FLEX Investigators Group. Differential gene expression analysis and clinical utility of MammaPrint and BluePrint in male breast cancer patients [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS14-11.

Abstract PD9-01: 5-year outcomes in the NBRST trial: Preoperative MammaPrint and BluePrint breast cancer subtype is associated with neoadjuvant treatment response and survival

Background: MammaPrint (MP) is used to identify breast cancer (BC) patients who can safely forego adjuvant chemotherapy. MP combined with the BluePrint (BP) molecular subtyping signature identifies BC subtypes with distinct therapeutic response rates and survival outcomes. In the Neoadjuvant Breast Symphony Trial (NBRST), MP and BP (MP/BP) predicted rates of pathologic complete response to neoadjuvant chemotherapy (NCT) and partial response to neoadjuvant endocrine therapy (NET). Here, we report 5-year overall survival (OS) and distant metastasis-free survival (DMFS) in patients from the NBRST registry according to MP/BP molecular classification. Methods: The NBRST trial (NCT01479101) prospectively enrolled 1072 patients from 2011 to 2014, who received MP and BP testing. Patients were assigned to receive NCT or NET according to NCCN guidelines and consented to 5 years post-surgery follow-up (FU). Clinical outcomes were available for 913 patients from 67 US institutions. Median FU for OS and DMFS was 5 and 4.6 years, respectively. Tumors classified by MP as High Risk (HR) or Low Risk (LR) were further stratified into four molecular subtypes by BP: Luminal A, Luminal B, HER2, and Basal. Differences in OS and DMFS at 3 and 5 years were assessed by Kaplan Meier analysis and log-rank test. Results: MP results from neoadjuvant patients (N=913) classified 16% of tumors as MP LR and 84% as MP HR. MP and BP classified 15.7% (143/913) of tumors as Luminal A, 32.5% (297/913) as Luminal B, 17.1% (156/913) as HER2, and 34.7% (317/913) as Basal. The 5-year OS and DMFS probabilities were significantly lower in HR compared to LR patients (p &lt; 0.001 for OS and DMFS), and lowest in Basal and Luminal B compared to Luminal A and HER2 subtypes (p &lt; 0.001 for OS and DMFS). Most DMFS events in BP Basal tumors occurred within the first 3 years. Of 841 patients that received NCT with or without HER2-targeted therapy, 12.2% (103/841) were LR and 87.8% (738/841) were HR. MP and BP classified 11.9% (100/841) of these patients as Luminal A, 32.6% (274/841) as Luminal B, 8.3% (154/841) as HER2 subtype, and 37.2% (313/841) as Basal. The 5-year OS and DMFS probabilities were lowest in HR, Basal or Luminal B patients (p &lt; 0.001). In 59 patients who received NET alone, 5-year OS and DMFS were significantly worse in HR patients that had Luminal B or HER2 tumors compared to LR Luminal A patients. In the 39 patients with Luminal A tumors, response to NET at the time of surgery was: 46.2% partial response, 41.0% stable disease, 5.1% progressive disease, 2.6% not reported. Five year DMFS in patients with Luminal A tumors treated with NCT or NET was not significantly different (p=0.67).Conclusions: MammaPrint remained prognostic in BC patients undergoing neoadjuvant therapy. Long -term prognosis was excellent in LR groups who received NCT or NET alone. MP and BP can accurately classify patients into specific subtypes with distinct OS and DMFS outcomes at five years, with BP Basals having the worst outcomes, followed by Luminal B, HER2, and Luminal A subtypes. BP Basal patients had the highest frequency of events within the first 3 years post-surgery, suggesting a genomic risk timeline distinct from other BP subtypes and a potential benefit from a secondary therapeutic immediately post-surgery. Additionally, Luminal A patients had a very low risk of progressive disease while on NET alone prior to surgery, with similar DMFS outcomes to Luminal A-types who received NCT.

Number of patientsObserved events% at 5 year (95% CI)p-valueAll patients - MammaPrint Risk GroupOS913134p&lt;0.001Low Risk146794.7 (88.4-97.6)High Risk76712781.1 (77.7-84.0)DMFS913182p&lt;0.001Low Risk1461191.2 (84.2-95.2)High Risk76717175.5 (71.9-78.7)All patients - MammaPrint + BluePrint SubtypeOS913134p&lt;0.001Luminal A143794.6 (88.3-97.6)Luminal B2974484.5 (80.0-88.7)Basal3177472.2 (66.2-77.3)HER2156993.4 (87.1-96.7)DMFS913182p&lt;0.001Luminal A1431191.1 (82.1-94.3)Luminal B2976975.2 (68.0-80.4)Basal3178570.4 (64.6-75.5)HER21561787.2 (79.7-92.0)NCT patients - MammaPrint Risk GroupOS841121p&lt;0.001Low Risk103397.4 (90.1-99.4)High Risk73811881.7 (78.3-84.7)DMFS841167p&lt;0.001Low Risk103792.6 (84.1-96.6)High Risk73816076.2 (72.5-79.4)NCT patients - MammaPrint + BluePrint SubtypeOS841121p&lt;0.001Luminal A100395.5 (86.2-98.6)Luminal B2743978.9 (71.7-84.5)Basal3137168.7 (57.9-77.2)HER2154892.8 (85.9-96.4)DMFS841167p&lt;0.001Luminal A100792.4 (83.8-96.5)Luminal B2746375.7 (65.6-76.5)Basal3138171.4 (65.6-76.5)HER21541687.7 (80.2-92.5)NET alone patients - MammaPrintOS597p=0.01Low Risk39293.0 (74.6-98.2)High Risk20580.0 (55.1-92.0)DMFS598p=0.003Low Risk39293.0 (74.6-98.2)High Risk20674.7 (49.4-88.6)NET alone patients - MammaPrint +BluePrint SubtypeOS597p=0.008Luminal A39293.0 (74.6-98.2)Luminal B18483.3 (56.8-94.3)Basal00N/AHER221N/ADMFS598p=0.005Luminal A39293.0 (74.6-98.2)Luminal B18577.4 (50.3-90.9)Basal00N/AHER221N/A

Citation Format: Pat Whitworth, James V Pellicane, Jr., Paul Baron, Peter Beitsch, Laura Lee, Michael Rotkis, Angela Mislowsky, Carrie Dul, Charles Nash, Bichlien Nguyen, Mary Murray, Paul Richards, Mark Gittleman, Stephanie Akbari, Shiyu Wang, Erin B Yoder, Andrea Menicucci, Lisa Blumencranz, William Audeh, NBRST Investigators Group. 5-year outcomes in the NBRST trial: Preoperative MammaPrint and BluePrint breast cancer subtype is associated with neoadjuvant treatment response and survival [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PD9-01.

Abstract PS4-04: Molecular subtyping by BluePrint improves prediction of treatment responses and survival outcomes in patients with discordant clinical and genomic classification

Background: The risk of distant recurrence gene signature, MammaPrint (MP), together with the molecular subtyping gene signature, BluePrint (BP), stratifies breast tumors into Luminal A, Luminal B, HER2, and Basal subtypes, independent of immunohistochemistry (IHC) or fluorescent in situ hybridization (FISH) expression. In the Neoadjuvant Breast Registry Symphony Trial (NBRST), MP and BP identified patients likely to respond to neoadjuvant treatment with higher accuracy compared to conventional methods. Here, we report 5-year follow up (FU) data in breast cancer (BC) patients from the NBRST registry with discordant clinical and genomic subtyping.Methods: This prospective study enrolled 1072 early-stage BC patients from 2009-2014 who received MP and BP testing. Patients received neoadjuvant therapy following standard of care and consented to 5 years post-surgery FU. IHC determined hormone receptor (HR) status, including ER and PR, and IHC and/or FISH determined HER2 status. Median FU for distant metastasis free survival (DMFS) and overall survival (OS) was 4.6 and 5 years, respectively. Differences in DMFS and OS was assessed by Kaplan Meier analysis and log-rank test.Results: Overall, BP reclassified 22% of tumors into different molecular subtypes compared to IHC/FISH (Table). BP reclassified 17% of ER+HER2- tumors as BP Basal, with higher pathological complete response (pCR) rates compared to ER+/BP Luminal tumors (36% vs. 4%). ER+/BP Basal patients had similar pCR rates as triple negative BC (TNBC)/BP Basal patients (36% vs. 37%) following neoadjuvant treatment, and pCR correlated with improved survival outcomes. The 5-year DMFS and OS probabilities were lower in ER+/BP Basal patients compared to TNBC/BP Basal patients and were substantially lower compared to ER+/BP Luminal patients (P &lt; 0.001). There were 106 HR-HER2+ patients, of whom BP reclassified 23.6% to Basal and 2.8% as Luminal B; the remaining 73.6% were confirmed HER2 by BP. The 5-year DMFS and OS probabilities were worse in HER2+/BP Basal patients compared to HER2+/BP HER2 patients. Of 142 triple positive (TP, ER+PR+HER2+) patients, BP classified 55% as Luminal, 39% as HER2, and 6% as Basal, with higher pCR rates observed in BP Basal and BP HER2 tumors compared to BP Luminal. The 5-year DMFS and OS probabilities were substantially lower in TP/BP Basal patients compared to TP/BP HER2 and TP/BP Luminal patients (P &lt; 0.05 and P &lt; 0.04). Of clinical HER2+ patients (HR+ or HR-) that received pertuzumab, patients that reclassified as BP Basal had worse OS compared to BP HER2 patients (P &lt; 0.04).Conclusion: ER+HER2- and HER2+ patients that reclassified as BP Basal are more likely to achieve pCR and have improved survival, demonstrating the clinical utility of BP in the neoadjuvant setting. These patients may benefit from optimized chemotherapy used for TNBC, including novel emerging treatments such as PD-1 and PARP1 inhibitors, in addition to HER2-targeted therapy. Furthermore, HER2+ tumors that were confirmed HER2 by BP may have high response rates to regimens containing TDM-1. Lastly, BP identified a subgroup of triple positive BC patients, who reclassified as BP Luminal, that may avoid overtreatment. Overall, molecular subtyping using MP and BP is more accurate in stratifying patients and predicting treatment responses and 5-year disease outcomes than conventional methods and thus, facilitates successful treatment decisions.

Clinical subtypeFrequency of BP classificationBluePrint subtypepCR%5-yr DMFS (95% CI)5-yr OS (95% CI)TNBC (n=236)0.42% (1/236)Luminal A100% (1/1)N/AN/A2.54% (6/236)Luminal B16.67% (1/6)N/AN/A1.27% (3/236)HER233.33% (1/3)N/AN/A95.76% (226/236)Basal36.73% (83/226)100% (pCR)100% (pCR)60.5% (50.5-69.1)(non-PCR)64.3%(52.1-71.2) (non-PCR)ER+HER2- (n=520)28.84% (152/520)Luminal A1.97% (3/152)91.1% (84.0-95.2)94.6% (88.3-97.6)52.37% (276/520)Luminal B5.43% (15/276)75.2% (69.0-80.4)84.5% (79.0-88.7)1.33% (7/520)HER214.29% (1/7)N/AN/A17.46% (92/520)Basal35.9%(33/92)84.1% (67.8-92.5) (pCR)86.3% (70.1-94.1) (pCR)54.6% (42.0-65.5) (non-pCR)57% (43.7-68.2) (non-pCR)HR-HER2+ (n=106)2.83% (3/106)Luminal B66.67% (2/3)100%100%73.59% (78/106)HER269% (54/78)82.8% (69.9-90.5)88.6% (76.0-94.8)23.58% (25/106)Basal40% (10/25)79.0% (52.5-91.7)79.0% (52.5-91.7)Triple Positive (n=142)12.68% (18/142)Luminal A22.22% (4/18)88.8% (76.5-94.8)94.5% (83.8-98.2)42.25% (60/142)Luminal B11.67%(7/60)38.73% (55/142)HER244.44% (24/55)87.5% (72.0-94.7)97.9% (83.8-98.2)6.34% (9/142)Basal55.56%(5/9)62.5% (22.9-86.1)70.0% (22.5-91.8)HER2+ (HR+ or HR-)treated with pertuzumab (n=105)28.6%(30/105)Luminal37% (11/30)84.7% (63.8-94.1)92.2%(71.8-98.0)57%(60/105)HER282% (49/60)91.4% (78.3-96.8)92.9%(79.2-97.7)14%(15/105)Basal40% (6/15)66.0%(31.1-86.3)64.0% (29.1-85.1)

Citation Format: Pat Whitworth, James Pellicane, Jr, Paul Baron, Peter Beitsch, Laura Lee, Michael Rotkis, Angela Mislowsky, Carrie Dul, Charles Nash, Bichlien Nguyen, Mary Murray, Paul Richards, Mark Gittleman, Stephanie Akbari, Shiyu Wang, Andrea Menicucci, Erin B Yoder, Lisa Blumencranz, William Audeh. Molecular subtyping by BluePrint improves prediction of treatment responses and survival outcomes in patients with discordant clinical and genomic classification [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS4-04.

Abstract LB-320: The FLEX real world data platform explores new gene expression profiles and investigator-initiated protocols in early stage breast cancer

Background: Genomic expression profiles have enabled the classification of breast cancers into molecular subtypes and provide prognostic information about the metastatic potential of the tumor, both of which have implications for the personalized treatment of breast cancer beyond clinical and pathological features. However, to precisely stratify tumors into actionable subgroups, full genome expression data should be combined with comprehensive clinical information. The FLEX Registry aims to aggregate a large, real-world dataset, which will enable the discovery of novel genomic profiles, particularly for patient subsets that are poorly represented in traditional clinical trials and will contribute to improved precision in the management of breast cancer. Trial Design: The FLEX Registry (NCT03053193) is a multicenter, prospective, observational trial for patients with Stage I, II, and III breast cancer. Patients with stage I-III breast cancer who receive MammaPrint, with or without BluePrint, on a primary tumor are eligible for enrollment. The primary objective of FLEX is to create a large scale, population-based registry that links complete clinical data with full genome expression data to elucidate new prognostic and/or predictive gene associations in a real-world setting. The FLEX Registry employs a shared study infrastructure to develop and investigate hypotheses for targeted subset analyses and/or clinical trials based on full genome expression data. The adaptable protocol is designed to be amended with the inclusion of targeted sub-studies. Patients enrolled in the initial study are eligible for inclusion in sub-studies for which they meet all eligibility criteria and additional consent is not required. Data will be collected on patients from diagnosis through 10 years of follow-up and any necessary additional clinical data will be collected as specified in the appendix protocols. Target enrollment is a minimum of 10,000 patients; &gt;4,000 patients have enrolled since April 2017 at more than 80 sites, including seven National Cancer Institute-designated comprehensive cancer centers. The FLEX collaborative platform enables participating investigators the opportunity to author their own sub-study protocols, as approved by the FLEX Steering Committee. Fifteen sub-studies have been approved for investigation within the FLEX Registry.

Citation Format: Jennifer Crozier, Adam Brufsky, Ian Grady, Sami Diab, Blanche Mavromatis, Nina D'Abreo, Carrie Dul, Rakhshanda Layeequr Rahman, Sarah Untch, Erin Yoder, Heather M. Kling, Amy M. Truitt, William Audeh, Bastiaan van der Baan, FLEX Investigators Group. The FLEX real world data platform explores new gene expression profiles and investigator-initiated protocols in early stage breast cancer [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-320.

High-risk breast cancer genes at 8q22-24 and their role in over 5,000 patients evaluated with the 70-gene risk of recurrence assay

3569

Background: Previous studies have shown that CCNE2 expression is higher in patients’ cancers resistant to CDK4/6 inhibitors. Increased expression of CCNE2, MTDH, or TSPYL5, genes contained within the 70-gene risk of distant recurrence signature (70GS), has also been implicated in breast oncogenesis, poor prognosis, and chemoresistance. These genes are located on chromosome region 8q22.1, one of the most recurrently amplified regions out of all 70GS genes in breast tumors (Fatima et al. 2017). MYC, located on 8q24, is overexpressed in 40% of all breast cancers (BC). Here we examined the expression of CCNE2, MTDH, and TSPYL5 in relation to 70GS risk and the 80-gene molecular subtype signature (80GS), and their correlation with MYC expression in early stage BC patients. Methods: CCNE2, MTDH, TSPYL5, and MYC mRNA expression was measured in 5022 BC samples sent to Agendia (Irvine, CA) for 70GS and 80GS testing, which included FFPE microarray full-transcriptome data. 70GS was used to stratify patients into Ultra Low Risk (UL), Low Risk (LR), High Risk (HR), and Ultra High Risk (UH). Both 70GS and 80GS were used to classify patient samples into Luminal A, Luminal B, HER2, or Basal type. Wilcoxon rank sum test was used to assess expression differences. Results: The expression of CCNE2, MTDH, and TSPYL5 significantly correlated with each other and was higher in HR patients compared to LR patients (p < 0.001) and higher in UH patients compared to HR patients (p < 0.001). Expression of these genes was highest in Basal type tumors, 83% of which were UH, followed by Luminal B type tumors, and lowest in Luminal A type tumors. CCNE2 and MYC expression was elevated in LR compared to UL patients (p < 0.001 and p = 0.0043). There was no difference in MYC expression between HR vs. LR or UH vs. HR. Lastly, there was no association between the expression of 8q22.1 genes and MYC in any 70GS subgroup. Conclusions: Within the 70GS, CCNE2, MTDH, and TSPYL5 have similar expression patterns and when overexpressed may identify an UH cohort of BC. This observation, in addition to their physical proximity at 8q22.1 suggests a possible amplicon in this region. The highest expression of CCNE2, MTDH, and TSPYL5 associated with UH patients and is concordant with previous studies that support the role of these genes in BC metastasis. Furthermore, this analysis suggests MYC may not stratify patients based on metastatic potential. These data may be clinically relevant for stratifying patients in ongoing clinical trials evaluating response and resistance to targeted therapies in early stage BC.

12-chemokine gene expression score in breast cancer patients treated with neoadjuvant chemotherapy

591

Background: Although advances in immunotherapy for the treatment of breast cancer have been minimal compared with other cancers, studies demonstrating tumor-infiltrating lymphocytes and immunomodulatory gene activation in the tumor microenvironment suggest the importance of antitumor immune responses in clinical outcomes. A 12-chemokine gene score has been shown to predict the presence of ectopic lymph node-like structures (ELN) in the tumor microenvironment and improved survival in melanoma, colon cancer, and breast cancer patients (Prabhakaran, 2017). Here, we evaluated this signature in an independent dataset of breast cancer patients treated with neoadjuvant chemotherapy. Methods: Tumor specimens used in this retrospective analysis (n = 92) were from breast cancer patients enrolled in either MINT (NCT0151487) or NBRST (NCT01479101) neoadjuvant registry trials from 2011 to 2016. Clinical data were captured with informed consent, and 70-gene signature (70-GS), 80-gene signature (80-GS), and full transcriptome data were generated by Agendia, Inc. Gene expression data were quantile normalized using R limma package. Principal component analysis (PCA) was performed on the normalized dataset using R princomp package. Chemokine score (CS) was defined as the first principal component values resulting from PCA. 70-GS/80-GS and clinical data were evaluated in relation to CS. CS were compared using Mann-Whitney test. Results: Of 92 breast tumors available for analysis, 84% were 70-GS High Risk (HR). Tumors were 39% Luminal-type, 24% HER2-type, and 32% Basal-type by 80-GS. HR tumors had higher CS than 70-GS Low Risk (LR) tumors (p < 0.001). 80-GS Basal-type, HER2-type, and Luminal B tumors had higher CS than Luminal A tumors (p < 0.01 for each comparison). High grade and ER-negative tumors seemed to have a high CS, although not significantly. Tumors from patients who achieved a pathological complete response (pCR) following neoadjuvant chemotherapy had higher CS than patients with residual cancer burden (p = 0.048). Conclusions: The current study demonstrated a significantly higher CS in 70-GS HR tumors and those which achieved pCR following neoadjuvant chemotherapy. Although further study is needed to evaluate the association of high CS with tumor-associated ELN, these results support previous work demonstrating that, although high CS is associated with aggressive clinical features, it also predicts superior clinical outcomes. The current study suggests validation of the 12-chemokine gene score in an independent dataset of breast cancer patients.

Basal subtype and clinical estrogen receptor status of genomically basal breast tumors in Caucasian, African American, and Latin American patients

556

Background: Genomically basal-type breast cancer is a heterogenous subtype that occurs at higher frequencies in non-Caucasian patients. Triple negative breast cancer (TNBC) has been refined into distinct transcriptomic groups including the basal-like immunoactive (BLIA), basal-like immunosuppressed (BLIS), luminal androgen receptor (LAR), and mesenchymal (MES) subtypes. Here we report the distribution of triple negative subtypes in genomically basal cancers from Caucasian (CA), African American (AA), and Latin American (LA) patients and the association of clinical estrogen receptor (ER) status. Methods: The FLEX Registry (NCT03053193) is an ongoing, prospective study evaluating primary tumors from patients with stage I-III breast cancer who receive the 70-gene signature and 80-gene signature (80-GS) molecular testing and consent to clinically annotated full transcriptome data collection. This sub-analysis evaluated 143 80-GS Basal type tumors from patients with self-reported ethnicity (60 CA, 59 AA, 24 LA). TNBC subtypes BLIA, BLIS, LAR, and MES were derived using an adjusted version of the 80-gene centroid signature published by Burstein and colleagues (2015). Differences in clinical ER expression between ethnicities were assessed by Fisher’s exact test. Results: Basal indices from 80-GS were not influenced by patient ethnicity (one-way ANOVA, p = 0.182). The frequency of BLIA, BLIS, LAR, and MES subtypes did not vary significantly by ethnicity (Fisher’s exact test, p = 0.671). The majority of tumors in all ethnic groups were BLIS (67% CA, 75% AA, 63% LA), followed by BLIA (22% CA, 22% AA, 25% LA), with low frequency of LAR (5% CA, 1.7% AA, 8.3% LA) and MES (5% CA, 1.7% AA, 4.2% LA) subtypes. Nearly one third (31%) of Basal type tumors were defined by IHC as ER positive and were present in all TNBC subtypes (39% BLIA, 29% BLIS, 33% LAR, and 20% MES); ER receptor expression ranged from 1-90% and was not associated with specific basal subtype (p = 0.8) nor ethnicity (P = 0.76). Progesterone receptor expression ranged from 1-50%. Conclusions: This analysis demonstrated that genomic Basal type tumor classification by 80-GS encompasses all TNBC subtypes evaluated regardless of ethnicity. Additionally, we show that IHC ER positive tumors occur in all TNBC subtypes assessed. These findings confirm the heterogeneous nature of basal breast tumors in CA, AA, and LA patients and highlight the clinical need to delineate basal biology in the ER+ cohort to advance treatment for basal-like tumors. Clinical trial information: NCT03053193 .

Molecular profiles and treatment recommendations for invasive lobular carcinoma in a real-world prospective breast cancer registry

e19291

Background: Invasive lobular carcinoma (ILC), the second most common histological breast cancer type, comprises 10-15% of breast tumors. Although ILC is thought to have a low risk of relapse, patient survival data vary. ILC treatment guidelines mirror those for invasive ductal carcinoma (IDC); however, these were extrapolated from trials that included predominantly IDC. The FLEX Registry captures data from real-world breast cancer patients; the current sub-study evaluated molecular profiles and treatment recommendations in ILC. Methods: The FLEX Registry (NCT03053193) includes stage I-III primary invasive breast cancer patients who receive 70-gene signature (70-GS)/80-gene signature (80-GS) testing and consent to full transcriptome and clinical data collection. This sub-analysis includes 335 ILC patients and 2,179 IDC patients enrolled from 2017 to present. 70-GS stratified tumors by risk of distant metastasis [High (HR), Low (LR), Ultralow (UL) Risk], and 80-GS classified tumors by molecular subtype (Luminal, HER2, or Basal type). Genomic and clinical data were compared for ILC and IDC using chi-square or Fisher’s exact test. Results: ILC represented 13% of FLEX cases (n = 335/2752); 81% were lymph node-negative, 99% ER-positive, and 94% HER2-negative. Clinical risk assessment (MINDACT criteria) classified 61% of ILC and 53% of IDC as low risk (p = 0.03). 70-GS risk distribution in ILC (18% UL, 51% LR, 31% HR) differed from IDC (13% UL, 34% LR, 53% HR; p < 0.01). Discordance between 70-GS and clinical risk was greater in ILC (43%) than IDC (30%, p < 0.01). 80-GS results also differed between ILC (98% Luminal, 1.6% HER2, 0.3% Basal) and IDC (85% Luminal, 3% HER2, 11% Basal; p < 0.01). Inclusion of chemotherapy (CT) in treatment plans was associated with 70-GS risk (p < 0.01); treatment plans that disagreed with 70-GS results were associated with discordant clinical risk (p < 0.05). Treatment plans disagreed with 70-GS in 15% of ILC and 12% of IDC total cases (p = 0.08), and more frequently in HR cases (25% in ILC, 14% in IDC; p < 0.01). Conclusions: The FLEX Registry includes patients with ILC consistent with real-world breast cancer frequencies. ILC demonstrated genomic risk and subtype profiles distinct from IDC. Treatment plans largely agreed with 70-GS results; however, discordance was more frequent in HR ILC than IDC. Future studies will evaluate clinical outcomes; however, these results demonstrate the added value of molecular profiling in ILC and the utility of real-world registry data in evaluating uncommon breast tumor types. Clinical trial information: NCT03053193 .

The FLEX real-world data platform explores new gene expression profiles and investigator-initiated protocols in early stage breast cancer

TPS7088

Background: Genomic expression profiles have enabled the classification of breast cancers into molecular sub-types and provide prognostic information about the metastatic potential of the tumor, both of which have implications for the personalized treatment of breast cancer beyond clinical and pathological features. However, to precisely stratify tumors into actionable subgroups, full genome expression data should be combined with comprehensive clinical information. The FLEX Registry aims to aggregate a large, real-world dataset, which will enable discovery of novel genomic profiles, particularly for patient subsets that are underrepresented in traditional clinical trials and will contribute to improved precision in the management of breast cancer. Methods: The FLEX Registry (NCT03053193) is a multi-center, prospective, observational trial for patients with Stage I, II, and III breast cancer. Patients with stage I-III breast cancer who receive the 70-gene signature risk of recurrence test, with or without the 80-gene signature molecular sub-typing test, on a primary tumor are eligible for enrollment. The primary objective of FLEX is to create a large scale, population-based registry that links complete clinical data with full genome expression data to elucidate new prognostic and/or predictive gene associations in a real-world setting. The FLEX Registry employs a shared study infrastructure to develop and investigate hypotheses for targeted subset analyses and/or clinical trials based on full genome expression data. The adaptable protocol is designed to be amended with the inclusion of targeted sub-studies. Patients enrolled in the initial study are eligible for inclusion in sub-studies for which they meet all eligibility criteria and additional consent is not required. Data will be collected on patients from diagnosis through 10 years of follow-up and any necessary additional clinical data will be collected as specified in the appendix protocols. Target enrollment is a minimum of 10,000 patients; >4,000 patients have enrolled since April 2017 at more than 80 sites, including seven National Cancer Institute-designated comprehensive cancer centers. The FLEX collaborative platform enables participating investigators the opportunity to author their own sub-study protocols, as approved by the FLEX Steering Committee. Fifteen sub-studies have been approved for investigation within the FLEX Registry. Clinical trial information: NCT03053193 .

Adding precision to 2018 ASCO/CAP HER2 testing guidelines in breast cancer with genomic profiling

3570

Background: Biological heterogeneity of HER2 positive breast cancers has been suggested by a modest benefit of HER2-targeted therapies reported in the APHINITY and ExteNET trials. This highlights the need for improved biomarkers that more precisely identify patients who benefit from HER2-directed agents. The 80-gene molecular subtyping signature (80GS) classifies breast tumors into Luminal, HER2 or Basal type based on the gene expression of downstream signaling pathways. Previous work showed a substantial proportion of tumors identified as HER2 equivocal or HER2 positive by 2013 ASCO/CAP guidelines may be reclassified as non-HER2 type by 80GS. In 2018, ASCO/CAP HER2 IHC/ISH classification guidelines were revised to reduce the frequency of HER2 equivocal cases, for which treatment recommendations have been ambiguous. Here we evaluated concordance between HER2 status by 2018 ASCO/CAP guideline classification and 80GS molecular subtyping. Methods: Pathology reports are provided by physicians for samples that are tested with the 70-gene risk of distant recurrence signature (70GS) and 80GS as part of routine diagnostic care. This analysis includes data sent to Agendia (Irvine, CA) from January 2019 to January 2020. HER2 IHC/ISH results based on ASCO/CAP 2018 guidelines were available for 1453 samples. Results: Of 1453 samples, 1336 (92%) were HER2 negative, 99 (7%) were HER2 positive, and 18 (1.2%) were HER2 equivocal under 2018 guidelines. 80GS reclassified 57 of 99 (58%) HER2 positive tumors as Luminal and 11 of 99 (11%) as Basal; the remaining 31% were confirmed HER2. Furthermore, 55 of 99 (55%) HER2 positive tumors were also ER and PR positive by IHC, with 48 (87%) of these reclassified as Luminal type. Of HER2 negative tumors, 80GS classified 94 of 1336 (7%) as Basal and 2 of 1336 (0.15%) as HER2. Of HER2 equivocal tumors, 16 of 18 (89%) reclassified as Luminal and 2 of 18 (11%) as Basal. Conclusions: In this real-world diagnostic dataset, 2018 ASCO/CAP guidelines resulted in few HER2 equivocal tumors overall, confirming the positive impact of the revised guidelines. However, 80GS reclassified 69% of HER2 positive tumors to non-HER2 molecular subtypes, suggesting these tumors may have suboptimal responses to HER2-directed therapy compared to HER2 enriched. All HER2 equivocal tumors reclassified to non-HER2 subtypes. Molecular classification by 80GS adds further precision in classifying HER2 positive patients and potential to predict responsiveness to HER2-targeted therapies. Further studies are warranted to validate the utility of HER2 status based on 80GS.

Distinct molecular profiles of interval and screen-detected tumors in a real-world breast cancer registry

e15587

Background: Interval breast cancers (BC) are detected between routine screening mammograms and are associated with worse prognosis, requiring more aggressive treatment compared to screen-detected BC identified during scheduled mammograms. Identifying molecular differences between interval BC and screen detected BC may lay the foundation for developing novel therapies. In this study, we compared gene expression profiles of interval BC to screen-detected BC. Methods: This analysis included a subset of 2260 patients enrolled in the FLEX Registry (NCT03053193), an ongoing, prospective study evaluating primary tumor samples from stage I-III BC patients who receive 70-gene risk of recurrence testing (70GS), 80-gene molecular subtyping (80GS), and consent to collection of clinically annotated full genome data. Interval BC were diagnosed < 12 months following a normal screening mammogram. Breast tumors were classified by 70GS as having a Low Risk (LR) or High Risk (HR) of distant metastases. Tumors were classified as Luminal, HER2, or Basal type by 80GS. Differential gene expression analysis was performed with limma and subsequent pathway analysis with DAVID and GSEA. Differences in the proportion of 70GS or 80GS results, 70GS index, and Ki67 were assessed by Chi-squared test or t-test. Results: In this study, 81% (1834/2260) of patients had screen-detected BC and 19% (426) had interval BC. A higher proportion of interval BC (51%) were HR compared to screen-detected BC (44%; p = 0.01). Most LR tumors were invasive ductal carcinoma (78% interval and 73% screen-detected) and over 99% were Luminal type. Between the two LR groups, 70GS indices were similar and there was no significant difference in transcriptional profiles. Basal and HER2 subtypes were more frequent among HR interval BC compared to screen-detected BC (p = 0.03). HR interval BC had 70GS indices of higher risk compared to HR screen-detected BC (p = 0.02). Differentially expressed genes in HR interval BC compared to HR screen-detected BC were associated with MYC signaling and mitosis, which was concordant with higher Ki67 by IHC (p = 0.007). Conclusions: This real-world data analysis shows interval BC are not all biologically High Risk and can be further stratified by the 70GS, aiding in treatment decisions. Preliminary results suggest that following 70GS LR classification, there is no biological difference between interval BC and screen-detected BC. In contrast, there are distinct biological processes associated with HR interval BC, which may have implications in the management of these cancers. Clinical trial information: NCT03053193.

Abstract P2-10-15: Different MammaPrint and BluePrint molecular profiles and clinical-pathological features of early stage breast cancer in Chinese patients in the United States and Hong Kong

Background: Breast cancer rates among Asian women are relatively low compared with Western populations; however, rates have increased in recent years, and it is the most common cancer in women. Few studies have characterized clinical-pathological features and molecular subtypes in Asian breast cancer patients, although substantial variation in occurrence among Asian subpopulations has been reported. Here, we report clinical factors, pathology, and molecular profiles of Chinese patients enrolled in the US and Hong Kong, HKSAR, and from age-matched Caucasian (Cau) and African-American (AA) patients.

Methods: This analysis included Chinese patients enrolled in the United States (n=24) and Hong Kong (n=42) with early stage, invasive breast cancer for whom clinical characteristics were captured with informed consent, enrolled between 2013 and 2018 (NCT02669745). For ethnicity comparisons, age-matched Cau (n=132) and AA (n=33) patients were included from the prospective, US-based FLEX registry (NCT03053193). Ethnicity was patient reported. Clinical characteristics, pathology, and results from the 70-gene (70-GS, MammaPrint) risk of recurrence and 80-gene (80-GS, BluePrint) molecular subtyping signatures were compared.

Results: The median age at diagnosis for Chinese was 51 years, and patients were 53.8% pre-/peri-menopausal. Tumors were predominantly ductal carcinoma, not otherwise specified (NOS) (81.3%), T1 (73.0%), grade 2 (59.7%), estrogen receptor-positive (ER+) (90.6%), and node-negative (82.0%). 70-GS and 80-GS results were available from 64 patients; 56.3% of tumors classified as 70-GS High Risk, and 82.8% Luminal-type, 9.4% HER2-type, and 7.8% Basal-type by 80-GS. Although pooled here for analysis, when examined separately, tumors of Chinese American patients were more frequently HER2-type (5/23; 21.7%) than tumors of Hong Kong patients (1/41, 2.4%) and had some higher risk clinical features, such as greater frequency of grade 3 classification (54.2%). Age-matched Cau patients were similar to Chinese patients in menopausal status (50.4% pre/peri-menopausal); age-matched AA patients were predominantly (58.6%) post-menopausal. Tumors of Cau patients were predominantly ductal carcinoma NOS (87.9%), T1 (73.4%), grade 2 (51.6%), node-negative (82.4%), ER+ (89.2%), 70-GS Low Risk (52.3%) and 80-GS Luminal-type (86.4%). Tumors of AA patients were predominantly ductal carcinoma, not otherwise specified (NOS) (86.7%), T1 (50.0%), grade 3 (53.6%), node-negative (79.3%), ER+ (72.7%), 70-GS High Risk (75.8%), and 80-GS Luminal-type (60.6%), although a substantial portion were 80-GS Basal-type (33.3%).

Conclusions: The current study revealed some disparities in clinical and molecular features of breast tumors between Chinese American and Hong Kong patients, and among Chinese, Cau, and AA patients. Some features typically associated with higher clinical risk were more prevalent in Chinese American patients than in Hong Kong patients, suggesting a possible interaction between genetics and lifestyle factors, and perhaps influenced by immigration. However, given the small sample size, these data should be interpreted with caution, and further study is needed to validate these trends. Future studies of breast cancer in ethnic subpopulations should incorporate evaluation of genomic profiling.

BluePrint Molecular Subtyping ResultsPatients Luminal-typeHER2-typeBasal-typeTotalChinese (pooled)53 (82.8%)6 (9.4%)5 (7.8%)64Hong Kong37 (90.2%)1 (2.4%)3 (7.3%)41US16 (69.6%)5 (21.7%)2 (8.7%)23Caucasian (US)114 (86.4%)6 (4.6%)12 (9.1%)132African-American (US)20 (60.6%)2 (6.1%)11 (33.3%)33

Citation Format: Ava Kwong, Sarah Untch, Heather M. Kling, Christine Finn, William Audeh, Vivian Shin, Sherene Ishtihar, Margaret Chen. Different MammaPrint and BluePrint molecular profiles and clinical-pathological features of early stage breast cancer in Chinese patients in the United States and Hong Kong [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 P2-10-15.

Abstract P4-10-24: Cyclin E overexpression is associated with high risk 70 gene signature, and may indicate intrinsic resistance to CDK4/6 inhibitors

Background: The use of CDK4/6 inhibitors (CDK4/6i) is a promising therapeutic strategy for recurrent ER+, HER2- breast cancers that have escaped previous treatment targeting the endocrine pathway. A number of adjuvant and neoadjuvant trials with CDK4/6i in early breast cancer are also underway. However, no clear predictive biomarkers for resistance are available other than loss of the RB1 tumor suppressor gene. Cyclin E genes (CCNE1, CCNE2) play a critical role in cell cycle control with approximately 25% of breast cancer overexpressing CCNE2, and there is a significant correlation between CCNE2 and CCNE1 expression1,2,3. Furthermore, a positive correlation between expression of CCNE1 and resistance to CDK4/6i (i.e. palbociclib)4,5,6 has been reported in vitro and in a gene expression analysis from the PALOMA-3 trial. The 70-gene MammaPrint® (MP) signature is a prognostic assay that stratifies early-stage breast cancer (ESBC) patients into Low and High-risk of distant relapse. One of the 70 genes is CCNE2, shown to be associated with resistance to both endocrine therapy and CDK4 inhibition7. Considering the potential role of the Cyclin E genes as biomarker of resistance for CDK4/6i, we assessed the expression of CCNE2 in a large series of ESBCs with respect to their MP risk profile.

Methods: The mRNA expression of CCNE2 (and CCNE1) was measured in a series of 5022 breast cancer samples for which FFPE microarray full-transcriptome data were available for testing. Intensities were Lowess normalized and log2 transformed. The 80-gene BluePrint (BP) profile8 was used in combination with MP to stratify patient samples based on their molecular subtype: Luminal A- (MP Low Risk, BP Luminal), Luminal B- (MP High Risk, BP Luminal), HER2- and Basal-type. Wilcoxon rank sum test was used to examine expression differences.

Results: CCNE2 is significantly higher expressed in MP High risk compared to MP Low risk tumors when all samples are included (p &lt;0.0001). When looking at the clinically ER positive BP Luminal group, CCNE2 is significantly higher expressed in the Luminal B compared to the Luminal A tumors (p &lt;0.0001). However, we observe a broad distribution of CCNE2 expression within the Luminal group, indicating a biological diversity in both Luminal A and B tumors, which CCNE2 may help to further define. Specifically, Luminal B tumors have a range of CCNE2 expression with the highest levels being in the highest MP risk interval (Ultra High). Of importance, CCNE1 showed similar expression patterns as CCNE2.

Conclusions: Our preliminary results show that MP could help in stratifying BP Luminal-type B tumors in subgroups with differential CCNE2 expression. Since increased CCNE1 expression is correlated with CDK4/6i resistance, and CCNE2 is a functionally-related gene whose expression correlates to some extent with CCNE1, it is reasonable to speculate that high CCNE2 expression and MP High Risk Luminal B may also correlate with CDK4/6i resistance, a testable hypothesis.

This warrants additional analyses that integrate treatment response data to support and further investigate these observations. Our exploratory analysis also highlights the added value of a multi-gene signature profile versus single-gene testing as tool for patient stratification and treatment recommendation.

Refernces:

1. Asghar U et al Nat Rev Drug Discov. 2015

2. Caldon C et al Cell Div. 2010

3. Payton M et al Oncogene. 2002

4. Turner N et al. J Clin Oncol. 2019

5. Chandarlapaty S et al J Clin Oncol. 2019

6. Guarducci C et al NPJ Br Cancer. 2018

7. Caldon CE et al Mol Cancer Ther. 2012

8. Krijgsman O et al BRCT 2012

Citation Format: Lorenza Mittempergher, Joe McKelley, Rajith Bhaskaran, Sahra Uygun, René Bernards, Laura J van 't Veer, Annuska M Glas, William Audeh. Cyclin E overexpression is associated with high risk 70 gene signature, and may indicate intrinsic resistance to CDK4/6 inhibitors [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 P4-10-24.

Abstract P2-14-11: Treatment recommendations in ER+ patients ≤ 50 years: Comparison of the 21-gene assay and 70-gene signature in the PROMIS study

Background: The PROMIS trial (NCT01617954) previously evaluated how a definitive result from the MammaPrint 70-gene signature (70-GS) can impact treatment recommendations for patients with an intermediate range recurrence score (RS 18-30) from the 21-gene assay (21-GA, Oncotype DX). Since publication of this study, TAILORx results published in 2018 (and further explored in June 2019) suggested an interaction between the 21-GA, patient age (≤50 yrs), and clinical risk. Initially, chemotherapy (CT) was recommended for all women ≤50 with a RS&gt;16. Based on the recent clinical-risk analysis, Ovarian Function Suppression (OFS) + endocrine therapy (ET) has been suggested as an alternative treatment for low clinical risk (clin-low) women ≤50yrs with an RS 16-25. This current analysis examines the updated treatment recommendations based on the interaction between patient age and clinical risk, and explores the impact that the 70-GS can have on adjuvant chemotherapy decisions for women ≤50 years of age. Methods: 70-GS risk of recurrence was determined for 21-GA intermediate patients by standard diagnostic testing (Agendia, Irvine, CA). Clinical risk was assessed using the MINDACT, modified Adjuvant Online! algorithm (Cardoso, NEJM 2016). The 70-GS High and Low Risk classification were subdivided by RS groups 18-20, 21-25, and 26-30 and by clinical risk stratification. Results: 181 patients in PROMIS were ≤50 yrs. Of those, 64% (116/181) were clin-low, and 35% (63/181) were high clinical risk (clin-high) (2 unknown). Among patients ≤50 yrs with RS 18-20, 60% (27/45) of clin-low and 56% (15/27) of clin-high were found to be 70-GS Low Risk. Among patients with RS 21-25, 55% (30/55) of clin-low and 30% clin-high (8/26) were Low Risk by the 70-GS. For patients ≤50 yrs with RS 26-30, 15% (4/27) were found to be 70-GS Low Risk. Of all patients with RS 26-30, 21% (32/156) were Low Risk by 70-GS. Conclusions: With the follow-up publication for TAILORx, incorporation of clinical risk in addition to age, RS group, and the assumed benefit of chemotherapy-induced menopause, has presented additional layers of complexity for physicians treating breast cancer. The current analysis demonstrates that 46% of women ≤ 50yrs with a RS 21-25 are 70-GS Low Risk, and based upon the prospective, randomized MINDACT* trial data, can safely avoid CT. Overall, the 70-GS can precisely identify 20-60% of women ≤ 50yrs with intermediate RS (18-30) as genomic Low Risk with excellent survival with ET alone (&gt;95% 5-yr DMFI [MINDACT]), who may otherwise be candidates for treatment with CT or OFS. *(Microarray in Node Negative and 1-3 Lymph Node Positive Disease May Avoid Chemotherapy)

AgeRS GroupClinical RiskNMP Low Risk% recommended ET alone based on 70 -GStreatment recommendation based on 21-GA≤ 50RS 18-20Clin-low452760%OFS+ET or ET aloneClin-high / (unknown)26 (1)1556%CT+ET or OFS+ETTotal724258% ≤ 50RS 21-25Clin-low553055%OFS+ETClin-high / (unknown)26 (1)830%CT+ET or OFS+ETTotal823846% ≤ 50RS 26-30Clin-low1616%All receive CT+ET regardless of clinical riskClin-high11327%Total27415%

Citation Format: Michaela Tsai, Hatem Soliman, Shelly Lo, Rubina Qamar, Raye Budway, Ellis Levine, Pat Whitworth, Blanche Mavromatis, Robin Zon, Sarah Untch, Lisa Blumencranz, Joseph McKelley, William Audeh, PROMIS Investigators Group. Treatment recommendations in ER+ patients ≤ 50 years: Comparison of the 21-gene assay and 70-gene signature in the PROMIS study [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 P2-14-11.

Abstract P2-10-08: Racial disparities in breast cancer: Identifying predisposing clinical and molecular features associated with African American patients

Background: Breast cancer (BC) mortality is higher in African-American women (AA) than in Caucasian women (CA). AA are also diagnosed at a younger age, have more aggressive subtypes and greater incidence of metabolic dysfunction such as obesity and diabetes. These disparities have been attributed to a confluence of socioeconomic, genetic and epigenetic factors. However, the distinctive tumor biology of AA BC is not yet fully elucidated, as AA remain underrepresented in breast cancer studies and databases. Here, we compared clinical and molecular BC features of AA and CA patients for insights into mechanisms associated with these racial disparities.

Methods: The FLEX Registry Trial (NCT03053193) is a prospective study evaluating tissue collected from patients with stage I-III BC who have consented to receive MammaPrint(MP)/BluePrint(BP) and clinically annotated full genome (FG) data. FLEX subset analyses investigate new gene associations that may be relevant to BC biology. This sub-study includes 160 AA and 199 CA patients (n=359) enrolled since April 2017. Clinical characteristics used in the analysis include menopausal status, metabolic factors, stage, grade and IHC results. A comprehensive publication search (PubMed) was conducted to validate candidate genes involved in BC in AA, BC genes associated with epigenetic regulation and genes associated with metabolic syndrome. Hierarchical clustering was performed on FG microarray (Agilent) intensity data focusing on the candidate genes (50 probes targeting 37 unique genes). Gene expression was compared between race and MP/BP risk groups.

Results: AA were predominately MP High Risk (HR) 67.5%, BP Luminal B 40.2%, BP Basal 22.6%. Interestingly, 40.0% of AA BP Basal were classified by IHC as ER+. CA were MP Low Risk (LR) 54.8%, BP Luminal A 54.8%. Clinically, there were no differences in histology, tumor size, nodal or menopausal status between AA and CA patients. AA had higher grade tumors, higher rates of type 2 diabetes and obesity. Three comparisons were performed: 1) AA and CA, irrespective of MP result, 2) AA and CA HR MP only, and 3) AA only, irrespective of MP result. In comparison 1) 15 unique genes showed significant differences (p&lt;0.05) in gene intensities between AA and CA patients. A heatmap showed four major clusters; cluster 1 included mainly (72.3%) AA patients, and these were predominantly MP HR and BP Basal. In comparison 2) 9 unique genes had significant differences (p&lt;0.05) in gene intensities, 6 of which were common to the first comparison. This showed three major clusters and a significant cluster among AA patients. Finally, comparison 3) found 20 unique genes with significant differences (p&lt;0.05) in gene intensities, 10 of which were common to the previous comparisons. This showed two major clusters with one significant cluster among AA MP HR, BP Basal patients. In total, 11 unique genes showed significant differences in intensities in AA or AA with HR MP including CYP4F8, TWIST1/2, CCND2, FOXA1 and GSTP1. These genes have known functions regulating metabolism and cell cycle.

Conclusions: Here we show AA patients were enriched for genes associated with metabolic syndrome and epigenetic regulation of metabolic syndrome. The aberrant function of these genes has been implicated in tumorigenesis, dysregulation of metabolism and drug resistance. Further validation is warranted to fully understand the association of these genes with the unique biology of breast cancer in AA.

Citation Format: Raquel Nunes, Lisa E. Blumencranz, Heather M Kling, Sahra Uygun, Sarah Untch, Erin B Yoder, Jennifer A Crozier, William Audeh. Racial disparities in breast cancer: Identifying predisposing clinical and molecular features associated with African American patients [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 P2-10-08.

MammaPrint guides treatment decisions in breast Cancer: results of the IMPACt trial

Background

Increased usage of genomic risk assessment assays suggests increased reliance on data provided by these assays to guide therapy decisions. The current study aimed to assess the change in treatment decision and physician confidence based on the 70-gene risk of recurrence signature (70-GS, MammaPrint) and the 80-gene molecular subtype signature (80-GS, BluePrint) in early stage breast cancer patients.

Methods

IMPACt, a prospective, case-only study, enrolled 452 patients between November 2015 and August 2017. The primary objective population included 358 patients with stage I-II, hormone receptor-positive, HER2-negative breast cancer. The recommended treatment plan and physician confidence were captured before and after receiving results for 70-GS and 80-GS. Treatment was started after obtaining results. The distribution of 70-GS High Risk (HR) and Low Risk (LR) patients was evaluated, in addition to the distribution of 80-GS compared to IHC status.

Results

The 70-GS classified 62.5% (n = 224/358) of patients as LR and 37.5% (n = 134/358) as HR. Treatment decisions were changed for 24.0% (n = 86/358) of patients after receiving 70-GS and 80-GS results. Of the LR patients initially prescribed CT, 71.0% (44/62) had CT removed from their treatment recommendation. Of the HR patients not initially prescribed CT, 65.1% (41/63) had CT added. After receiving 70-GS results, CT was included in 83.6% (n = 112/134) of 70-GS HR patient treatment plans, and 91.5% (n = 205/224) of 70-GS LR patient treatment plans did not include CT. For patients who disagreed with the treatment recommended by their physicians, most (94.1%, n = 16/17) elected not to receive CT when it was recommended. For patients whose physician-recommended treatment plan was discordant with 70-GS results, discordance was significantly associated with age and lymph node status.

Conclusions

The IMPACt trial showed that treatment plans were 88.5% (n = 317/358) in agreement with 70-GS results, indicating that physicians make treatment decisions in clinical practice based on the 70-GS result. In clinically high risk, 70-GS Low Risk patients, there was a 60.0% reduction in treatment recommendations that include CT. Additionally, physicians reported having greater confidence in treatment decisions for their patients in 72% (n = 258/358) of cases after receiving 70-GS results.

Trial registration

“Measuring the Impact of MammaPrint on Adjuvant and Neoadjuvant Treatment in Breast Cancer Patients: A Prospective Registry” (NCT02670577) retrospectively registered on Jan 27, 2016.