Treatment-Related Outcome in Triple-Negative Breast Cancer (TNBC): A Multi-institutional Retrospective Analysis

Volume 87  Number 2S  Supplement 2013

Oral Scientific Sessions

S99

interval with higher incidence of visceral metastasis including lung, liver, and brain. Few reports have suggested similar rates of loco-regional failure compared with receptor positive subtypes. We undertook this retrospective study to analyze the demographic characteristics and pattern of failures in TNBC breast cancer. Materials/Methods: Tumor registry queries identified 476 patients with TNBC diagnosed between 1996 and 2011. Demographic and clinical information including age, race, menopausal status, stage, insurance status and tumor characteristics were collected after obtaining IRB approval. Information regarding sites of failure was collected. The Kaplan-Meier method was used to compute local failure-free survival (LFFS) and distal failure-free survival (DFFS) probabilities with p values for group comparisons computed using the log-rank statistic. Fisher’s exact test was used to compare frequencies for categorical variables. Results: Fifty-eight percent of patients (n Z 275) were between 41 and 60 years, with 20% (n Z 93) and 23% (n Z 103) being  40 and > 61 years, respectively. Forty-nine percent (n Z 231) patients were white, 47% (n Z 223) were AA and 4% was others. Fifty-five percent (n Z 262) were premenopausal and 83% (n Z 396) had grade 3 tumors. Stage II (48%, n Z 230) was most frequent stage at diagnosis followed by stage III (28%, n Z 131), stage I (19%, n Z 91) and stage IV (5%, n Z 24). Twenty-eight percent (n Z 133) had lymphovascular invasion versus 65% with none and 7% not reported. Ninety-three percent of the patients received either neoadjuvant or adjuvant chemotherapy, 93% patients received surgery including mastectomy or lumpectomy and 73% was treated with radiation. The 2, 5, and 8-year Kaplan-Meier LFFS: DFFS was 77%:73%, 64%:62% and 58%:56%, respectively. Black and white patients did not differ statistically with respect to OS (p Z 0.38), LFFS (p Z 1.0), or DFFS (p Z 0.77). Sixty-five (14%) patients failed locally either with IBTR or CW, 20 (4%) failed in the regional lymphatics. Lung (n Z 83, 17%) was the most frequent distant failure site followed by liver in 50 (11%), brain in 53 (11%), and bone in 53 (11%). Fifty-nine patients (12%) failed in nonregional lymph nodes. Additionally, 22 (5%) patients failed in skin or soft tissue outside breast or chest wall. AA and white patients did not differ statistically with respect to failure sites. Conclusions: Survival and failure site outcomes for AA and white patients did not differ in this large series of TNBC cases. Lung is the predominate site of distant failure and few patients failed in the regional lymph nodes. Even beyond 5 years, the probability of failing locally or distally continues to occur steadily. Author Disclosure: S. Prasad: None. T.M. Zagar: None. S. James: None. P. Walker: None. R. Raab: None. J. Efird: None. T. Biswas: None.

n Z 35). Pathologic complete response (pCR) was defined as no residual invasive cancer in breast and nodes. OS and EFS were assessed using Kaplan-Meier analyses and Log-Rank tests with SPSS software (IBM SPSS 20). Results: The median age was 52 years (range, 25-85 years). The median follow-up time was 29 months (range, 6-144 months). The majority of patients had clinical stage (AJCC, 2010) II or III disease; 50% and 46% respectively. Most patients, 84% (108 of 129), were treated with an anthracycline-taxane based regimen and 94% of HER2+ patients received trastuzumab. All patients underwent surgical resection, 72% mastectomy and 28% lumpectomy, followed by RT. Thirteen percent of patients had a pCR (17/129), by subtype: 21% of HER2+, 6% of HR+, and 14% of TN. The 3-year actuarial OS and EFS rates for the entire cohort were 82% and 58% respectively. The 3 year EFS for patients who achieved a pCR was 100% versus 57% (p < 0.01) for those who did not (non-pCR). Univariate analyses showed pathological stages, lymphovascular invasion and number of positive nodes to be significant prognostic factors for EFS. The overall rate of DM was 15.5% (20 of 129) for the entire cohort, with equivalent rates by subtypes in non-pCR patients; 17% HER2+, 18% HR+, and 18.5% TN respectively. The overall rate of LRR was 8% (10/129); however, the incidence of LRR was significantly higher for TN patients (18%) versus HER2+ (6.7%) or HR+ (4.5%) in non-pCR group. No patient with a pCR had a DM or LRR. Conclusions: Patients with TN breast cancer with less than pCR to neoadjuvant chemotherapy had a higher rate of LRR as compared to HR + or HER2+ patients. While further study is needed to validate these findings, our data support that there may be a need to further intensify local therapy in this subgroup of patients. Author Disclosure: C. Zhang: None. S.X. Yan: None. S. Crockford: None. E.P. Connolly: None.

242

Purpose/Objective(s): TNBC is a distinct breast cancer subtype and accounts for about 15%-20% of all breast cancer cases and has particularly aggressive clinical course with a disproportionate number of breast cancer relapses and deaths. This retrospective study aims to investigate various treatment factors and related outcome in TNBC. Materials/Methods: Four hundred fifty-two Stage I-III TNBC patients who were treated between April 1996 and September 2011 were included in this analysis. As part of an IRB-approved study, records were reviewed to extract demographic, pathologic, treatment and outcome data. Two hundred three patients (45%) received neoadjuvant chemotherapy (NACT) vs 218 patients (48%) receiving adjuvant chemotherapy (ACT) while 32 patients (7%) did not receive any chemotherapy. Two hundred ten (46%) patients underwent lumpectomy compared with 226 (50%) patients with mastectomy. Sixteen patients had neither lumpectomy nor mastectomy due to disease progression or patient refusal. Overall Survival (OS) was computed from the date of diagnosis to the date of death or last FU. For disease-free survival (DFS), patients were scored if they failed either locally or distally. Statistical analysis was performed using SAS v9.0. Results: The median age was 51 yrs (21-88 yrs). The median follow-up was 3.0 yrs (0.14-14 yrs). The 2, 5, and 8-year Kaplan-Meier OS: DFS was 85%:77%, 71%:66%, and 67%:59%, respectively. On univariate analysis, both lumpectomy and mastectomy were associated with better OS and DFS compared with no surgery (p < 0.0001). NACT had significantly

Triple-Negative Breast Cancer Is Associated With Higher Rate of Locoregional Recurrence for Patient Achieving Less Than a pCR Following Neoadjuvant Chemotherapy, Surgery, and Radiation Therapy C. Zhang,1 S.X. Yan,2 S. Crockford,2 and E.P. Connolly1; 1New York Presbyterian Hospital - Columbia, New York, NY, 2Columbia University School of Medicine, New York, NY Purpose/Objective(s): Breast cancer subtype, determined by expression of estrogen/progesterone receptor (ER/PR) and human epidermal growth factor receptor (HER)-2, has been demonstrated to be predictive for overall survival (OS), event-free survival (EFS) and distant metastasis (DM) following neoadjuvant chemotherapy. The importance of subtype to rates of locoregional recurrence (LRR) following neoadjuvant chemotherapy is less well characterized, particularly in patients who receive adjuvant radiation therapy (RT). Materials/Methods: Using an institutional review board approved protocol, we retrospectively identified 129 breast cancer patients treated from 2001-2011 with neoadjuvant chemotherapy followed by surgery and adjuvant RT. Patients were grouped according to receptor status: hormone receptor positive (HR+) (ER or PR+/HER2-; n Z 49), HER2+ (HER2+/ HR+ or HR-; n Z 38), or triple negative (TN) (ER(-)PR(-)HER2(-);

243 Treatment-Related Outcome in Triple-Negative Breast Cancer (TNBC): A Multi-institutional Retrospective Analysis T. Biswas,1 J. Efird,2 S. Prasad,3 T. Zagar,4 S. James,5 P. Walker,2 R. Raab,2 L. Carey,6 and L. Marks7; 1University Hospital at Case Western Reserve University, Cleveland, OH, 2Brody School of Medicine, East Carolina University, Greenville, NC, 3University of North Carolina at Chapel Hill, Chapel Hill, NC, 4University of North Carolina at Chapel Hill, Chapel Hill, NC, 5East Carolina University, Greenville, NC, 6 University of North Carolina at Chapel Hill, Chapel Hill, NC, 7University of North Carolina at Chapel Hill, North Carolina, NC

S100

International Journal of Radiation Oncology  Biology  Physics

inferior outcome than ACT on univariate analysis (OS: HR Z 2.3, p < 0.0001; DFS: HR Z 1.9, p Z 0.0002). Among patients who underwent NACT, 33% had pathological complete response (pCR) versus 67% patients with residual disease. Both OS and DFS were significantly better among patients achieving pCR than those with residual disease (OS: HR Z 0.07, p < 0.0001; DFS: HR Z 0.19, p < 0.0001). Patients receiving ACT or NACT with pCR had superior OS and DFS than patients receiving neoadjuvant with partial response (p < 0.0001). Presence of LVI, stage III and having Medicaid had significantly inferior OS and DFS outcome. Adjuvant radiation, type of chemotherapy, age, race, grade and year of treatment had no effect on OS or DFS. Results did not substantively change after multivariate adjustment for age, race, stage, surgery, chemotherapy, radiation, LVI, institution and insurance type. Conclusions: This is a large triple negative breast cancer series showing survival outcome and various prognostic factors. Future studies are needed to improve the pathological complete response rate. More insight is needed to explore the relationship between different insurance types and outcome in this distinct breast cancer cohort. Author Disclosure: T. Biswas: None. J. Efird: None. S. Prasad: None. T. Zagar: None. S. James: None. P. Walker: None. R. Raab: None. L. Carey: None. L. Marks: None.

human breast cancer and therefore represents “value added” information over traditional BC subtyping. This signature has the clinical potential to identify patients with tumors refractory to standard RT for whom other strategies or treatment intensification are needed. Author Disclosure: C. Speers: None. F. Feng: None. M. Liu: None. J. Brenner: None. O. Balbin: None. L. Pierce: None.

244 Identification and Validation of a Radiation Sensitivity Signature in Human Breast Cancer C. Speers, F. Feng, M. Liu, J. Brenner, O. Balbin, and L. Pierce; University of Michigan Hospital, Ann Arbor, MI Purpose/Objective(s): The development of a specific radiation (RT) sensitivity signature to predict likelihood of response to RT in human breast cancer (BC) is an attractive goal. Previous BC signature development efforts have focused on prognosis or response to adjuvant chemotherapy. We hypothesized that pairing post-RT clonogenic survival data with gene expression data across a large spectrum of BC cell lines would generate a BC-specific RT sensitivity signature predictive for RT response in BC patients and allow identification of patients with tumors refractive to conventional therapy. Materials/Methods: Using clonogenic survival assays, we identified the range of surviving fraction (SF) after 2 Gy of RT across 22 BC cell lines. Using SF as a continuous variable, the RT sensitivity score (RSS) was correlated to gene expression using Spearman’s correlation method. Supervised hierarchical clustering identified differences in gene expression across resistant and sensitive cell lines to generate a radiation sensitivity (RS) signature. This signature was validated in a separate human breast tumor dataset (185 patients) containing early stage, nodenegative patients treated with surgery and RT alone without adjuvant chemotherapy to assess the predictive effect of the radiation signature on recurrence risk after RT. Results: Clonogenic survival identifies a range of radiation sensitivity in human BCC lines (SF 77%-17%) with no significant correlation (r value < 0.3) to the intrinsic BC subtype. Using Spearman’s correlation method, a total of 126 genes were identified as being associated with radiation sensitivity (72 positively correlated, 54 negatively correlated). Supervised hierarchical expression discriminates gene expression patterns in the RT resistant and sensitive cell lines and is enriched for genes involved in cell cycle arrest and DNA damage response, including RAD51, TOP2A, BUB1 (enrichment p value 5.0 E-22). Application of this RS signature to an independent breast cancer dataset with clinical outcomes validates the signature and accurately identifies patients with decreased rates of recurrence compared to patients with high expression of the radioresistant signature (p value < 0.0001, misclassification error rate .31, 12 of 13 patients with locoregional recurrence accurately identified). Conclusions: In this study, we derive a human breast cancer-specific radiation sensitivity signature (RSS) with biologic relevance from preclinical studies and validate this RSS for prediction of recurrence in a clinical data set. The RSS is not correlated to the intrinsic subtypes of

245 Cytosolic Mislocalization of BRCA1 Is Associated With Increased Metastatic Risk in Breast Cancer W.L. Santivasi,1 T. Wang,2 Q. Yang,3 X. Mo,1 E. Brogi,4 B.G. Haffty,3 A.B. Chakravarthy,2 and F. Xia1; 1The Ohio State University College of Medicine, Columbus, OH, 2Vanderbilt University School of Medicine, Nashville, TN, 3Robert Wood Johnson Medical School, New Brunswick, NJ, 4Memorial Sloan-Kettering Cancer Center, New York, NY Purpose/Objective(s): To assess the relationship between subcellular mislocalization of BRCA1 to the cytosol and metastasis in patients with breast cancers. Materials/Methods: Tissue microarrays were assembled from 106 patients treated from 1986-2006. These samples were analyzed for BRCA1 localization via immunofluorescence staining and categorized as “cytosolic,” “nuclear-cytosolic,” or “nuclear” BRCA1 localization. Wild type and BRCA1-sequestering mutant (BRCA1 5382insC) MCF-7 cells were subjected to in vitro invasion and migration assays. Tissue microarrays were assembled from 504 patients treated from 1970-2005. These samples were analyzed for BRCA1 localization via immunofluorescence staining and compared to recurrence-free survival (RFS) and distant metastasis-free survival (MFS). BRCA1 localization analyses were conducted via paired and unpaired t-test. Association of cytosolic BRCA1 RFS and MFS were analyzed via log-rank test using a Cox regression to adjust for other prognostic factors (age, tumor size, and lymph node, ER, PR, and Her2 statuses). Results: Previous literature reported that breast cancer cells contain 8.214.8% cytosolic BRCA1. Tissue microarray analysis revealed that the cytosolic BRCA1 content of lung metastases (m Z 36.0%, 95% CI Z [31.7%, 40.3%]) from patients with breast cancer was markedly higher. Intriguingly, lung metastases and their corresponding primary tumors were similarly rich in cytosolic BRCA1 in both unpaired (36.0%  18.9% vs 38.5%  10.5%, p Z 0.36) and paired analyses (35.8%  19.2% vs 37.8%  12.3%, p Z 0.70). In vitro studies demonstrated that genetically induced BRCA1 cytosolic sequestration increases the invasion efficiency of tumor cells (5.04% vs 1.24%, p Z 0.03). Correlation of tissue microarray data with clinical outcomes indicated an inverse relationship between cytosolic BRCA1 and metastasis-free survival (p Z 0.07), which was statistically significant in patients over age 40 (p Z 0.02). Conclusions: Enrichment of cytosolic BRCA1 in primary tumors may play a causative role in breast cancer metastasis development, making it a potential therapeutic target and biomarker for metastatic risk. Author Disclosure: W.L. Santivasi: None. T. Wang: None. Q. Yang: None. X. Mo: None. E. Brogi: None. B.G. Haffty: None. A.B. Chakravarthy: None. F. Xia: None.

246 Complementing Ipsilateral Breast Tumor Recurrence (IBTR!) Estimates With the 21-Gene Recurrence Score (RS) in Early-Stage, Hormone Receptor-Positive, HER2-Normal, Lymph Node-Negative Breast Cancers N. Thaker,1 K.E. Hoffman,1 L. Pusztai,2 T.A. Buchholz,1 and W.A. Woodward1; 1Department of Radiation Oncology, University of Texas MD Anderson Cancer Center, Houston, TX, 2Department of Medical Oncology, Yale Cancer Center, New Haven, CT Purpose/Objective(s): The 21-gene RS quantifies the risk of distant recurrence in early-stage, hormone-receptor positive, lymph node-negative