K-D Yu

Abstract P1-06-07: Clinical and molecular relevance of intra-tumor genetic heterogeneity in breast cancer: Integrative analysis of data from The Cancer Genome Atlas

Background: Intra-tumor heterogeneity (ITH) plays a pivotal role in driving breast cancer progression and therapeutic resistance. Emerging evidence has indicated that the extent of genetic heterogeneity may serve as a clinically useful biomarker. While several studies have suggested the prognostic value of ITH in several cancer types, the clinical significance of genetic ITH and molecular portraits that correlated with different ITH levels were poorly understood in breast cancer. The establishment of algorithms estimating genetic ITH based on sequencing of bulk tumor DNA offered us an opportunity to explore the clinical implication of ITH in large breast cancer cohorts and, for the first time, to use integrative genomic analyses to reveal molecular portraits related to intra-tumor genetic heterogeneity. Methods: We assessed 916 female breast cancer patients from The Cancer Genome Atlas. Mutant-allele tumor heterogeneity (MATH) values were calculated from whole-exome sequencing data. We used integers nearest to the tertiles of the MATH values as cutoff points to divide the patients into three groups nearly equal in size. The association between MATH value and clinical characteristics was evaluated, followed by survival analyses in these different MATH groups. We then compared the rates of total non-silent somatic mutations among the different MATH groups, and further determined the mutations independently associated with high MATH by logistic regression adjusting for T classification and clinical subtypes. Similar methods, superadding somatic copy number alteration (SCNA) burden in logistic model, were used to evaluate SCNA events that were significantly associated with high MATH level. Gene enrichment between the high and rest MATH groups was analyzed using Gene Set Enrichment Analysis. Results: The patients were divided into low (MATH value lower than 33), intermediate (MATH between 33 and 46) and high (MATH higher than 46) MATH groups. High T stage, African American race, and triple-negative or basal-like subtype were associated with a higher MATH level (all P Conclusion: Our study extended the knowledge concerning the clinical role of ITH in breast cancer, especially the distinct pattern of prognostic values in different clinical subtypes, which may help promote the clinical utilization of genetic ITH. Our attempt at exploring the molecular features related to ITH might provide clues for the source and consequences of ITH, inspiring subsequent experiments investigating the laws underling tumor heterogeneity. Citation Format: Ma D, Jiang Y-Z, Liu X-Y, Liu Y-R, Yu K-D, Shao Z-M. Clinical and molecular relevance of intra-tumor genetic heterogeneity in breast cancer: Integrative analysis of data from The Cancer Genome Atlas [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-06-07.

Abstract P6-04-04: Comprehensive transcriptome analysis identifies novel molecular subtypes and subtype-specific lncRNAs of triple-negative breast cancer

Background: Triple-negative breast cancer (TNBC) is a highly heterogeneous group of cancers with no effective therapeutic targets hitherto. Thus molecular subtyping is necessary to better identify molecular-based therapies. While some classifiers have been established, no one has integrated the expression profiles of long-noncoding RNAs (lncRNAs) into such subtyping criterions. Considering the emerging important role of lncRNAs in gene regulation and other cellular processes, a novel classification integrating the transcriptome profiles of both messenger RNA (mRNA) and lncRNA would help us better understand the heterogeneity of TNBC and treat patients accordingly. Methods: Using human transcriptome microarray, we retrieved the transcriptome profiles of 165 consecutive TNBC samples. We used k-means clustering to classify the samples based on the most differentially expressed genes (standard deviation>0.65). Empirical cumulative distribution function was analyzed to determine the optimal number of subtypes. Then the new classifier was compared with the Lehmann/Pietenpol system, and survival analyses were performed to compare the recurrence-free survival (RFS) in different subtypes. Gene Ontology (GO) and pathway analyses were applied to determine the main function of the subtype-specific genes and pathways. We conducted co-expression network analysis to identify interactions between lncRNAs and mRNAs, and to predict possible functions of subtype specific lncRNAs. Results: All 165 TNBC tumors were classified into four distinct clusters, each displaying unique GOs and pathways. These include an immunomodulatory (IM) subtype, a luminal androgen receptor (LAR) subtype, a mesenchymal-like (MES) subtype and a basal-like and immune suppressed (BLIS) subtype, accounting for 17.0%, 17.6%, 33.3%, and 32.2% of the patients, respectively. The IM subtype had unique GOs and pathways involving immune cell process. The LAR subtype was highly enriched in hormonally regulated pathways. Enriched pathways in the MES subtype included ECM-receptor interaction, focal adhesion, and processes linked to growth factor signaling pathways. The BLIS subtype was characterized by downregulation of immune response gene and activation of cell cycle and DNA repair, and patients in this subtype experienced worse RFS compared to other subtypes (log-rank test,P=0.045), which was in concordance with the highly proliferative and immune-suppressed nature of these tumors. When analyzing the distribution of the Lehmann/Pietenpol subtypes in our classification system, we found that the two classification systems were significantly correlated (P=0.039). However, our novel classification was more concise and significantly connected with survival outcome. Subtype-specific lncRNAs were identified and their possible functions were predicted using co-expression network analysis. Conclusions: We developed a novel TNBC classification system integrating the expression profiles of both mRNAs and lncRNAs, and determined subtype-specific lncRNAs that are potential biomarkers and targets of TNBC. If further validated in larger population, our novel classification system could facilitate patient counseling and individualize treatment of TNBC. Citation Format: Liu Y-R, Jiang Y-Z, Xu X-E, Zuo W-J, Yu K-D, Jin X, Hu X, Wu J, Liu G-Y, Di G-H, Shao Z-M. Comprehensive transcriptome analysis identifies novel molecular subtypes and subtype-specific lncRNAs of triple-negative breast cancer. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P6-04-04.

Abstract S4-05: Exome sequencing identifies shift in TP53 and PIK3CA mutation status after paclitaxel-based neoadjuvant chemotherapy in breast cancer

Purpose: The changed gene-expression profiles of breast cancer under neoadjuvant chemotherapy (NCT) have been described. However, it remains unclear whether chemotherapy affects somatic mutation status in breast cancer. We investigated the influence of paclitaxel-based chemotherapy on somatic mutations in tumor tissues from patients with breast cancer. Patients and Methods: Samples were derived from two cohorts: first, 296 patients who underwent paclitaxel-based NCT with matched pre- and post-chemotherapy tumor tissues; second, 81 patients who underwent paclitaxel-based chemotherapy with pre-chemotherapy tumor tissues. First, we performed whole exome sequencing (WES) to examine the exomes (100-fold coverage) of two paired basal-like breast cancer samples (pre-treatment tumor biopsies and post-treatment tumors). Shift in somatic mutation status was validated by Sanger sequencing and analyzed in an extension cohort of 296 patients. Furthermore, we performed microdissection and genetic heterogeneity analysis for samples from 81 patients who underwent paclitaxel-based chemotherapy. Somatic mutation status was determined to reveal potential impact of paclitaxel-based chemotherapy on mutation status. Results: Compared with post-treatment biopsies, the analysis of WES data identified and validated TP53 and PIK3CA — as significantly mutated genes in pre-treatment tumors with a mutation frequency of approximately 10% in an independent extension cohort of 296 paired samples of different molecular subtypes. Somatic mutations of TP53 or PIK3CA were detected in 24.3% of the pre-chemotherapy tumor samples (35 of 296 for TP53 , 32 of 296 for PIK3CA , 5 of 296 for both genes, respectively) but in only 12.2% of the post-chemotherapy tumor samples (18 of 296 for TP53 , 16 of 296 for PIK3CA , 2 of 296 for both genes, respectively). The decrease in mutation rate was statistically significant ( P < 0.001). Patients with TP53 / PIK3CA mutations switched from positive to negative after chemotherapy had better partial response than patients with no change or with a reverse change ( P = .008). Furthermore, patients with TP53 / PIK3CA mutations switched from positive to negative had improved disease-free survival ( P = 0.018) and improved overall survival ( P = 0.032) relative to patients with no change or with a reverse change. In the second cohort, 18.5% of the tumors (15 of 81) showed intratumoral heterogeneity of TP53 or PIK3CA somatic mutations, whereas 81.5% (66 of 81) were homogeneous, either with mutations of certain genes or not. Conclusion: Our data reveal the novel concept that significant evolution might occur during chemotherapy. Chemotherapy may reduce somatic mutation frequency in patients with breast cancer, likely the result of a preferential response of highly fit sub-clones without mutations of certain genes. In addition, the identification of somatic mutations of TP53 and PIK3CA as biomarkers for prediction of treatment response and prognosis might help us optimize choice for sequential therapy and improve patients’ survival. KEY WORDS : Exome sequencing; TP53 ; PIK3CA ; neoadjuvant chemotherapy; breast cancer. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr S4-05.

Abstract P3-14-06: Residual tumor autophagy marker LC3B is a prognostic marker in local advanced breast cancer underwent neoadjuvant chemotherapy

Purpose A membrane-bound microtubule-associated protein chain 3B(LC3B) is known as one of the most specific biomarker for autophagy. The goal of this study is to investigate the prognostic value of autophagy marker LC3B in patients with residual tumors after neoadjuvant chemotherapy(NCT) for local advanced breast cancer(LABC) Patients and Methods Expression of LC3B in residual breast cancer cells were assessed by immunohistochemistry in surgical specimens from 229 patients who were diagnosed with histological proven invasive breast cancer. All patients had undergone neoadjuvant chemotherapy followed by mastectomy, and were considered as non pathological complete responders(non-pCR)after pathological evaluation. The density of positive staining of LC3B was measured with the use of a computerized image system. Prognostic value of various clinicopathologic factors was combined and evaluated by using Cox regression model. Results LC3B density was similar in the peripheral and central area of tumor(P = 0.328), however, it was siginifcantly lower in the extratumoral area(P<0.001 and P<0.001, respectively). High LC3B density was more likely to be observed in patients with PR-negative at diagnosis(P = 0.033), Beclin-1-positive(P<0.001) at surgery and higher Ki-67 index at surgery(P = 0.029). Breast cancer subtype was also correlated to LC3B expression(P = 0.003), an obvious lower proportion of LC3B-high density was detected in luminal-A patients than in other subtypes(32.4% in luminal-A, 59.4% in luminal-B, 65.4% in HER-2+, and 46.7% in TNBC). In survival analysis, only LC3B and residual nodal status after NCT resulted significant predictors of both RFS(HR = 1.180,P = 0.012 and HR = 3.067, 6.103, P<0.001, respectively) and overall survival(HR = 2.428, P = 0.008 and HR = 1.809,4.283, P = 0.003, respectively). Ki-67 was also an indepentent predictor for RFS(HR = 1.122, P = 0.019). The prognostic value of LC3B was most significant in triple-negative patients than in other subtypes. With a combination of LC3B expression and status of residual involved lymph nodes, patients were classified into four groups with different risk of relapse and death(P<0.001 for RFS and P = 0.003 for OS). Patients with no positive node residual and low LC3B expression were at low risk of relapse and death, with 5-year RFS and OS of 97% and 97%, respectively. However, the 5-year RFS and OS for patients with a positive node satus and high LC3B were only 52% and 66%, respectively. Conclusion LC3B is a prognostic marker in non-pCR patient after neoadjuvant chemotherapy for breast cancer, which highlighting the importance of autophagy in the biological behavior of chemo-resistence cancer cells. Evaluating and targeting autophagy in neoadjuvant setting may help in prevention of disease relapse in non-pCR patients. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P3-14-06.

Abstract P2-07-01: Polymorphism A247S of XRCC4 contributes to the risk of breast cancer and impairs DNA damage repair pathway

Genetic variations in a network of genes involved in DNA double strand breaks (DSB) repair pathway have been shown to increase the risk of breast cancer. XRCC4 is identified as a pivotal modulator in non-homologous end joining (NHEJ), a predominant DSB repair pathway. In this study, we evaluated the associations between single nucleotide polymorphisms (SNPs) in XRCC4 and breast cancer, and investigated the impact of XRCC4A247S variant on nuclear localization and DNA damage response functions of XRCC4. A hospital-based case-control study, comprising 412 breast cancer patients and 458 cancer-free controls recruited from Shanghai Cancer Center, was conducted to investigate the relationship between common polymorphisms in XRCC4 and the risk of breast cancer. Our results showed that XRCC4 A247S polymorphism (rs3734091), located near the putative nuclear localization signal (NLS) of XRCC4, was associated with a significantly increased risk of breast cancer in a recessive genetic model (Adjusted odds ratio (OR) = 1.47, 95% confidence interval (CI) = 1.09-4.52; P = 0.042). This homogeneous variant was more often observed in triple negative subtype (Adjusted OR = 2.07, 95% CI = 1.34 - 3.58 in the recessive model; P = 0.036). Although XRCC4 A247S variant did not harm its interactions with XLF and ligase IV (the other two components of DNA ligase complex), we revealed that A247S alteration partially impaired the nuclear localization of XRCC4 in mammary cells. Remarkably, more evidences implied that XRCC4 A247S variant significantly reduced the in vivo NHEJ repair efficiency compared to the wild-type XRCC4 via colony formation assay and NHEJ reporter assay, suggesting that reduced nuclear accumulation by XRCC4 A247S variant impaired its functions in response to DNA damage. Importantly, XRCC4 A247S alternation promoted cellular sensitivity to ionizing radiation and DNA damaging agents in chemotherapy. Taken together, this study demonstrates that genetic variation of XRCC4 A247S could impair NHEJ mediated DSB repair and provide further evidences that genetic variants involved in DNA repair pathways contribute to breast cancer susceptibility. In addition, our results also indicate that XRCC4 A247S variant might be a potential predictive biomarker and therapeutic target in breast cancer. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-07-01.

Abstract P2-14-04: Effect of age at the time of diagnosis on overall and cancer-specific mortality in breast cancer: Analysis of SEER, 1990-2008

Background: As a critical factor, age at the time of diagnosis has a significant impact on cancer therapy and outcomes. The purpose of this study is to investigate the overall mortality (OM) and breast cancer-specific mortality (BCSM) in patients with breast cancer corresponding to age at the time of diagnosis, thus constructing an age-related pattern on breast cancer survival. Methods: We analyzed data for 396,978 women aged 20-79 from 1990 to 2008 diagnosed with breast cancer from the population-based Surveillance, Epidemiology, and End Results (SEER) program of the U.S. national cancer institute. Clinicopathological factors and outcomes were compared using a chi-square test, the Kaplan-Meier method, and Cox proportional hazards regression model. Results: In the multivariable analysis, a significant relationship was found between age at the time of diagnosis and OM (p<0.001) as well as BCSM (p<0.001). Using 70-79 years old as the reference, the hazard ratio (HR) of OM and BCSM presents a U-like shape with the bottom at the group 40-49 years old (OM: HR,0.56; P<0.001; BCSM: HR,0.58; p<0.001) and 50-59 years old (OM: HR,0.59; P<0.001;; BCSM: HR,0.60; p<0.001), increasing to both ends gradually. Analysis in subgroup according to ER and PgR status, group 20-29 years old showed no significant difference with group 70-79 years old on both OM and BCSM in all subgroups except on BCSM in ER+PR+ subgroup (HR,1.25; P = 0.002). Conclusion: Patients at very young and very old age (20-29 years old and 70-79 years old) have the poorest outcomes while prognosis improves with age towards mean level. The very young age (20-29 years old) may be regarded as a significant hazard factor for BCSM in patients with ER+PR+ breast cancer. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-14-04.

Abstract P5-11-05: Overexpression of girdin in nucleus indicates poor disease-free survival in ER-positive breast cancer

Background Girdin is identified as a novel modulator of the PI3K-Akt signaling pathway in regulating neurons integration during neurogenesis process. Recently, several studies demonstrated that Girdin and its phosphorylation type (p-Girdin) were also involved in tumor metastasis and angiogenesis. However, the clinical implications of Girdin and p-Girdin in breast cancer remain unclear. In this study, we detected the expression pattern of p-Girdin (Ser1416) and Girdin in breast cancer to uncover the association between p-Girdin/Girdin status and clinicopathological parameters like estrogen receptor (ER), progestogen receptor (PR), and human epidermal growth factor receptor 2 (Her2), etc. Methods This study included 250 patients who were histologically confirmed as invasive ductal breast cancer and underwent mastectomy in the Department of Breast Surgery in Shanghai Cancer Center during 2001 to 2006. We performed immunohistochemistry of p-Girdin/Girdin on tissue microarrays constructed from above specimens. Results We observed that Girdin was mainly located in the cytoplasm of mammary epithelial carcinoma cells (195/248, 78.63%) while p-Girdin was mainly located in the nucleus (160/250, 64.00%). However, Girdin/p-Girdin was also weakly expressed in the nucleus/cytoplasm (77/248, 31.05% for Girdin and 193/250, 77.2% for p-Girdin). By the Chi-square test, we found that Girdin in the nucleus (GN) was positively associated with ER status (P = 0.012); Girdin in the cytoplasm (GC) was positively associated with ER (P = 0.004), PR (P = 0.028), and Her2 status (P = 0.004); p-Girdin in the cytoplasm (pGC) was positively associated with PR status (P = 0.001) and negatively associated with tumor size (P = 0.025). However, no significant association was observed between p-Girdin in the nucleus (pGN) and other parameters. Using Kaplan-Meier method, we disclosed a trend that GN indicated poor disease-free survival (DFS) in ER-positive breast cancer (P = 0.075). After adjusted for histological grade and tumor size, multivariate cox regression analysis confirmed this trend (HR = 3.051, 95% CI: 1.137-8.191, P = 0.027). Furthermore, GC was prone to associating with poor DFS, especially in stage II (P = 0.079) and grade III (P = 0.058) patients. Conclusions Our work reveals, for the first time, that the location and expression pattern of both Girdin and p-Girdin (Ser1416) in breast cancer, suggesting that p-Girdin/Girdin are associated with ER, PR and Her2 status. In addition, Girdin in the nucleus is associated with poor DFS in ER-positive breast cancer. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P5-11-05.

Abstract P3-06-14: Identification of Prognosis-Relevant Subgroups in Patients with Chemoresistant Triple Negative Breast Cancer

Purpose: Triple-negative breast cancer (TNBC) is a highly heterogeneous disease. TNBC patients with pathologic complete response (pCR) have excellent survival, but those with residual disease after neoadjuvant chemotherapy have significantly worse outcome. However, some patients having extensive residual cancer burden after neoadjuvant chemotherapy do not relapse, and we hypothesize that there may be subgroups with diverse prognosis among these chemoresistant TNBC patients. Methods: Forty-nine cases with residual cancer from 111 TNBC patients treated with neoadjuvant chemotherapy (in M.D. Anderson Cancer Center, 2000–2006) constituted the discovery cohort. Twenty-five chemoresistant samples from 47 neoadjuvant chemotherapy-treated TNBC (in Baylor College of Medicine and Methodist Hospital, 2002–2006) were chosen for validation. Extended validation was performed in 269 operable TNBC predicted to be chemoresistant (using a JAMA-published genomic predictor) from public databases. Results: By comparing the gene expression data from cases in relapse with those from un-relapsed cases, we established a 7-gene prognostic signature (including AR, ESR2, GATA3, GBX2, KRT16, MMP28, and WNT11) using dChip and gene enrichment analyses. In the discovery cohort, the signature showed positive predictive value (PPV; i.e., cumulative relapse rate of patients predicted to relapse in 3 years) of 95.4% and negative predictive values (NPV; i.e., relapse-free survival of patients predicted not to relapse in 3 years) of 100%. In the validation cohort, the classifier predicted correctly with PPV of 75.0% and NPV of 76.9% at 3 years. Compared with patients predicted not to relapse, those predicted to relapse had a hazard ratio of 3.37 (95% CI, 1.15–9.85) for disease recurrence or death in 3 years. In an extended validation cohort of 269 patients, our signature discriminated chemoresistant TNBC in overall cohort (PPV, 52.4%; NPV, 77.7%; log rank P Conclusion: We developed a clinically useful prognostic signature for chemoresistant TNBC. For these chemoresistant TNBC patients, new therapeutic strategies targeting AR-activation or cancer stem cells need to be developed. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P3-06-14.

Abstract P1-13-07: Association Between Delayed Initiation of Adjuvant Chemotherapy and Survival in Breast Cancer: A Single-institution Study and A Systematic Review and Meta-analysis

Objective: Adjuvant chemotherapy (AC) improves survival among patients with operable breast cancer. However, the effect of delay in AC initiation on survival is unclear. We performed a single-institution data analysis and a systematic review and meta-analysis to determine the relationship between time to AC and survival outcomes. Methods: PubMed, EMBASE, Cochrane Database of Systematic Reviews, and Web-of-Science databases (before March-20, 2012) were searched to identify relevant eligible studies. An additional retrospective analysis including 1,408 patients from the prospectively maintained database of Shanghai Cancer Center was performed and results were also included. Hazard ratios (HRs) for overall survival (OS) and disease-free survival (DFS) from each study were converted to a regression coefficient (β) and its standard error corresponding to a continuous representation per 4-week delay of AC. Individual adjusted β were combined using a fixed-effects or random-effects model depending on heterogeneity. Results: We included 8 eligible studies with 10 independent analytical groups involving 21,221 patients, 1 prospective observational study, 2 secondary analyses in randomized trials (4 analytical groups), and 5 hospital-/population-based retrospective studies. In our single-institution study, every 4-week delay in initiation of AC significantly decreased OS (HR = 1.38; 95% confidence interval [CI], 1.03–1.87) and DFS (HR = 1.35; 95% CI, 1.08–1.69) after adjustment for other prognostic variables. The overall meta-analysis demonstrated that a 4-week increase in time to AC was associated with a significant decrease in both OS (HR = 1.20; 95% CI, 1.03–1.39; random-effects model) and DFS (HR = 1.21; 95% CI, 1.08–1.36; fixed-effects model). One study caused a significant between-study heterogeneity for OS (p = 0.001; I 2 =74.1%); after excluding that single study, there was no heterogeneity (p = 0.345; I 2 =11.1%) and the HR was more significant (HR = 1.24; 95% CI, 1.14–1.35; P Conclusion: The present meta-analysis suggests that longer time to AC was associated with worse survival in breast cancer patients. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-13-07.

P2-07-02: Germline Genetic Variants Disturbing the Let-7/LIN28 Double-Negative Feedback Loop Alter Breast Cancer Susceptibility.

Previous studies have shown that let-7 can repress the post-transcriptional translation of LIN28, and LIN28, in turn, could block the maturation of let-7, forming a double-negative feedback loop. In this study, we investigated the effect of germline genetic variants on regulation of the homeostasis of the let-7/LIN28 loop and breast cancer risk. We initially demonstrated that the T/C variants of rs3811463, a single nucleotide polymorphism (SNP) located near the let-7 binding site in LIN28, could lead to differential regulation of LIN28 by let-7. Specifically, the C allele of rs3811463 weakened let-7-induced repression of LIN28 mRNA, resulting in increased production of LIN28 protein, which could in turn downregulate the level of mature let-7. This effect was then validated at the tissue level in that the normal breast tissue of individuals with the rs3811463-TC genotype expressed significantly lower levels of let-7 and higher levels of LIN28 protein than those individuals with the rs3811463-TT genotype. Because previous in vitro and ex vivo experiments have consistently suggested that LIN28 could promote cellular transformation, we then systematically evaluated the relationship between rs3811463 as well as other common LIN28 SNPs and the risk of breast cancer in a stepwise manner. The first hospital-based association study (n = 2,300) demonstrated that two SNPs were significantly associated with breast cancer risk, one of which was rs3811463, while the other was rs6697410. The C allele of the rs3811463 SNP corresponded to an increased risk of breast cancer with an odds ratio (OR) of 1.25 (P = 0.0091), which was successfully replicated in a second independent study (n = 1,156) with community-based controls. The combined P-value of the two studies was 8.0 × 10 −5 . Taken together, our study demonstrates that host genetic variants could disturb the regulation of the let-7/LIN28 double-negative feedback loop and alter breast cancer risk. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-07-02.