H-G Moon

P5-23-01: The Impact of Primary Tumor Resection on the Survival of Patients with Stage IV Breast Cancer According to Molecular Subtype.

Purpose: The main treatment for stage IV breast cancer is currently systemic therapy. Surgical resection of the primary tumor is usually done for treating the tumor-related complications. Recent studies have suggested that surgery may improve the long-term survival of stage IV breast cancer patients. We evaluated the impact of the primary surgical resection site on the survival of stage IV breast cancer patients according to molecular subtype using nationwide Korean breast cancer registry data. Methods: We analyzed the records of the stage IV breast cancer patients from Korean Central Cancer Registry (KCCR) between 1999 and 2008. We used clinical assays to distinguish luminal A (HR+/HER2−, n=290), luminal B (HR+/HER2+, n=154), Basal-like (HR-/HER2−, n=107) and HER2 (HR-/HER2+, n=145). The clinical and tumor characteristics, the type of treatments and the overall survival were compared between the surgically versus nonsurgically treated patients according to molecular subtype. Results: Of the 1091 identified patients, 719 (65.9%) received surgical excision of their primary tumor and 372 (34.1%) did not. The mean survival was 86 months versus 43 months for the surgically treated patients vs. the patients without surgery, respectively (p Conclusion: Surgical resection of the primary tumor in stage IV breast cancer patients was independently associated with improved survival only in luminal A subtype. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-23-01.

Abstract P5-03-04: To excise or not?: Scoring system for predicting malignancy in patients diagnosed with intraductal papilloma at ultrasound-guided core needle biopsy

Background: The management of benign intraductal papillomas on core biopsy is controversial. The aim of this study was to determine factors that predict under-evaluation of atypical lesion or malignancy in patients diagnosed with benign papilloma at ultrasound-guided core needle biopsy (CNB), and to develop a prediction algorithm for scoring the possibility of a diagnosis upgrade to atypical lesion or malignancy based on clinical, radiological and pathological factors. Methods: The study enrolled patients diagnosed with benign papilloma at ultrasound-guided CNB who subsequently underwent surgical excision of the lesion. Multivariate analysis was used to identify relevant clinical, radiological and pathological factors that may predict malignancy. Results: A total of 520 CNBs led to a diagnosis of benign papilloma (including benign and atypical papillary lesion), of which 452 CNBs were benign papilloma without atypia. Of the 250 lesions in 234 women were underwent subsequent surgical excision, 44 (17.6%) were diagnosed with atypia or malignancy. Multivariate analysis revealed that bloody nipple discharge, size on imaging ≥15 mm, BIRADS≥4b, peripheral location, and a palpable lesion were independent predictors of atypical lesion or malignancy. A scoring system was developed based on logistic regression models and beta coefficients for each variable. The area under the ROC curve was 0.830 (95% CI: 0.665-0.996), and the negative predictive value was 100% for a score ≤4. Conclusions: A scoring system to predict malignancy in patients diagnosed with benign papilloma at CNB was developed based on five factors: bloody nipple discharge, size on imaging ≥15 mm, BIRADS≥4b, peripheral location, and a palpable lesion. This system was able to identify a subset of patients with lesions likely to be benign, indicating that imaging follow-up rather than surgical excision may be appropriate. Citation Format: Ahn Sk, Moon H-G, Han W, Noh D-Y, Ko E. To excise or not?: Scoring system for predicting malignancy in patients diagnosed with intraductal papilloma at ultrasound-guided core needle biopsy [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 P5-03-04.

Abstract P4-07-08: The prognostic significance of ataxia-telangiectasia-mutated (ATM) and p53 expression in breast cancer

The purpose of this study was to investigate the correlation of ataxia-telangiectasia-mutated (ATM) protein and p53 expression with clinicopathological features and prognosis in patients with sporadic breast cancers. The expression of ATM and p53 was determined by immunohistochemistry in 420 surgically resected breast cancers. Loss of ATM was observed in 126 out of 407 evaluable cases (31.0%), and was significantly associated with aggressive features with large tumor size, lymph node metastasis, higher tumor grade, and negativity of ER and/or PR. ATM loss was associated with a significantly shorter disease-free survival (DFS) (p = 0.019). Abnormal p53 expression was found in 39.3% of tumors (157 out of 400), conferring a worse DFS as well (p = 0.002). When investigated together, combined ATM and p53 expression status were associated with a worse DFS (p = 0.002). On multivariate analysis, ATM loss and abnormal p53 expression status was an independent predictor of poorer DFS (intact ATM and normal p53 vs. ATM loss and abnormal p53, HR 3.350; 95% CI 1.496 - 7.502; p = 0.003). Furthermore, in patients treated with adjuvant anthracyclines, either p53 alone or p53 combined with ATM significantly influenced DFS (p = 0.004, p = 0.015, respectively). The present study demonstrates that expression of ATM and p53 is an independent prognostic marker in breast cancers, and might be a practical tool for predicting benefits from anthracycline-based adjuvant therapy. Citation Format: Suh KJ, Ryu HS, Lee K-H, Im S-A, Kim T-Y, Kim H, Yang Y, Moon H-G, Han S-W, Oh D-Y, Han W, Kim T-Y, Park IA, Noh D-Y, Bang Y-J. The prognostic significance of ataxia-telangiectasia-mutated (ATM) and p53 expression in 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 P4-07-08.

Abstract P6-04-02: Identification of ESR1 splice variants associated with prognosis in estrogen receptor positive breast cancer

Background: Splice variants play a major role in carcinogenesis and disease progression. It is well known that androgen receptor splice variants are associated with resistance to prostate cancer treatment. Estrogen receptor (ER)-positive breast cancers constitute about 70% of all breast cancers and have better prognosis compared to ER-negative cancers. However, there are ER-positive breast cancers that acquire resistance to anti-estrogen therapy, and 12-55% of those tumors were shown to possess ESR1 mutations. The aim of this study was to identify common splice variants in the ESR1 gene and investigate their association with disease outcome. Methods: Whole transcriptome sequencing was performed on breast cancer specimens from 120 invasive breast cancer patients who underwent operation at Seoul National University Hospital (SNUH) and data from SNUH, GEO, and The Cancer Genome Atlas (TCGA) was used for normal breast tissue sequencing. Exon-exon junctions were identified on aligned RNA sequencing data and was used to construct exon graphs. Splice variant candidates were selected from exon graphs and were merged according to variant subtypes of samples. Subtypes were accessed differentially in relation to how frequent the junctions appear in tumor samples and common exon skipping types with frequent junctions were identified. TCGA RNA sequencing data was then used to search for the common exon skipping subtypes detected from SNUH RNA sequencing data. Results: Of the 120 tumor samples, 50 were clinically ER-positive by immunohistochemistry. Among exon paths logically possible, 125 paths were not observed in normal breast tissues. Exon 4-5 junction was the most commonly observed junction in the tumor samples. In a search for exon skipping type that results in missing ligand-binding domain of ER, three exon skipping types were identified. Exon skipping with exon 5-10 junction (type 1), exon 9-12 junction (type 2), and exon 10-12 (type 3) was seen in 4 (8%), 4 (8%), and 10 (20%) ER-positive samples, respectively. Retrospective medical chart review of the 18 patients showed recurrence in 4 (100%), 2 (50%), and 4 (40%) patients with type 1, 2, and 3 exon skipping, respectively. Evaluation of TCGA RNA sequencing data of 872 ER-positive samples suggested exon 4-5 junction as the most common junction. A search for exon skipping types in TCGA revealed 1 (0.1%), 9 (1.0%), and 454 (52.1%) samples with type 1, 2, and 3 exon skipping, respectively. However, none of the patients with type 1 or 2 had metastasis or had expired. Of the 454 patients with type 3 exon skipping, 54 patients had died, constituting 61.4% of 88 mortalities in the whole ER-positive population. Conclusion: Certain splice variants of ESR1 gene yields exon skipping subtypes commonly observed in the ER-positive breast cancer. Estrogen receptor-positive breast cancer with these exon skipping types resulting in a missing ligand-binding domain of ER may be associated with poorer disease outcome. Further investigation is warranted to validate the role of ESR1 exon skipping subtypes in the disease progression of breast cancer. Citation Format: Lee H-B, Han W, Ko S, Kim M-S, Lim S, Lee K-M, Kang YJ, Han JH, Kim Y, Yoo T-K, Moon H-G, Noh D-Y, Kim S, Han W. Identification of ESR1 splice variants associated with prognosis in estrogen receptor positive 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 P6-04-02.

Abstract P1-02-01: Genomic analysis of single cells isolated by a pulse laser retrieval system

Background: Isolating tumor cells of interest and harvesting histologically pure samples is important for genomic studies. Laser capture microdissection (LCM) is an established method to obtain such purified cell populations for various applications including DNA, gene expression, and single cell analyses. However, LCM possesses problems such as limited optical resolution, cell fragmentation from dissection, and adherence of adjacent tissue to the cells which interrupts with single cell isolation from tissue sections. To overcome these obstacles, we developed a high-throughput pulse laser retrieval system which uses a wavelength that minimizes damage to the cellular content and is processed with a sacrificial layer that provides applicable optical resolution. The aim of this study was to evaluate the performance of the pulse laser retrieval system to provide appropriate samples for genomic analysis using breast cancer tissue. Methods: An indium tin oxide (ITO) coated glass slide was prepared using fresh frozen breast cancer tissue sections of 4㎛ thickness and stained by hematoxylin and eosin. The slide was mounted on the cell isolation machine and imaging was performed with a charge-coupled device camera using a 20× lens. Following identification of the target cells by a pathologist, nano-second pulsed laser (wavelength= 1064nm) was irradiated on the target. Isolated cells were collected in a polymerase chain reaction tube and whole genome amplification (WGA) was carried out using Illustra GenomiPhi V2 DNA Amplification Kit (GE Healthcare Life Sciences, Pittsburgh, PA, USA). Amplified genomic DNA was fragmented and Illumina sequencing libraries were constructed. Sequencing was carried out to generate data with 0.1∼0.2× depth throughout the whole genome for each sample. Copy number variation (CNV) was analyzed by the Variable binning algorithm. Results: Whole genome amplification was performed using bulk tissue and 10 captured single cells from the same specimen. No difference in amplification coverage was observed between the two samples. A CNV analysis of captured single cells revealed similar CNV profiles with those in a matched bulk tumor. Whole exome sequencing (WES) of captured single cells yielded a variant frequency of 15% at a read depth of 15× and 50M base coverage, compared to 0% at 100× and 50M for WES using bulk tumor and 0.5% at 1200× and 100K for targeted sequencing using bulk tumor. Laser capture was performed for DCIS and stromal cells from the same slide. CNV analysis of the two samples showed minimal CNV in normal stromal cells in contrast to DCIS where multiple CNVs were observed. Conclusions: Newly developed pulse laser retrieval system is suitable for capturing single cells for genomic analysis of breast cancer. WGA, WES, and CNV analysis was successfully carried out using the captured single cells and showed no difference in profile compared to those performed with bulk tissue. This method may have the potential to replace LCM for certain applications such as single cell analyses. Citation Format: Lee H-B, Kim S, Lee K-M, Jung Y, Lee AC, Kim J, Bae S, Ryu HS, Yoo T-K, Moon H-G, Noh D-Y, Kwon S, Han W. Genomic analysis of single cells isolated by a pulse laser retrieval system. [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 P1-02-01.

Abstract P2-02-15: Discovery of putative circulating tumor cells through somatic mutation profile of epithelial cell adhesion molecule positive single cells from blood of metastatic breast cancer patients

Background: Circulating tumor cell (CTC) enumeration provides prognostic information for chemotherapy in metastatic breast cancer. However, due to its rarity and heterogeneity, it is difficult to distinguish true CTCs from normal blood cells and perform genomic analysis on them for use in therapeutic strategies. The main application of most currently available CTC detection systems consists of an enumeration of putative CTCs without further analysis. The aim of this study was to evaluate the feasibility of single cell picking and target sequencing of epithelial cell adhesion molecule (EpCAM)-positive cells for detecting CTCs. Methods: Whole blood sampled from metastatic breast cancer patients who were newly diagnosed with metastasis or who had disease progression during palliative treatment were used for this study. After applying IsoFlux Circulating Tumor Cell Enrichment Kit (Fluxion, South San Francisco, CA, USA), single CTC candidates were picked from a pool of EpCAM-positive cells. Genomic DNA from the picked cells was whole genome amplified and target sequencing was performed using Ion AmpliSeq™ Cancer Hotspot Panel (Life Technologies, Carlsbad, CA, USA). Target sequencing reads were mapped to human genome reference (hg19) using BWA-MEM (0.7.10). Single nucleotide variants (SNVs) were annotated using dbSNP, Variome Data 0.2, and COSMIC databases. Results: A total of 172 EpCAM-positive cells were selected according to size and EpCAM status from whole blood of 11 patients. The remaining cells were grouped into a pooled sample for each patient. The mean read depth of the target genes was 13455×. A mean 7.82 mutations as determined by SNVs listed in the COSMIC database but not in dbSNP and Variome Data 0.2 were detected in each patient. Cells with multiple mutated genes, or those with a mutated gene repeatedly observed in another cell from the same patient were judged to be putative CTCs. At least 2 putative CTCs were detected in 7 patients while no CTCs were detected in 2 patients. Mutated genes observed in the putative CTCs were ABL1, AKT1, APC, CDH1, CDKN2A, ERBB2, FGFR3, HRAS, IDH1, JAK2, KDR, NPM1, RB1, RET, SMARCB1, STK11, and TP53. Conclusions: Potential CTCs were successfully identified by single cell picking and target sequencing of EpCAM-positive cells from whole blood of metastatic breast cancer patients. Unique mutations not detected in other single cells and pooled samples can be used to distinguish putative CTCs from normal cells. Genomic profiling of corresponding primary tumor and metastatic site biopsy is warranted to verify the CTCs and investigate their role in disease progression. Citation Format: Lee H-B, Jeon S, Kim BC, Jho S, Kim J, Kang YJ, Yoo T-K, Han JH, Kim Y, Im S-A, Moon H-G, Noh D-Y, Han W. Discovery of putative circulating tumor cells through somatic mutation profile of epithelial cell adhesion molecule positive single cells from blood of metastatic breast cancer patients. [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 P2-02-15.

Abstract P1-10-11: Prognostic value of non-alcoholic fatty liver disease in stage II/III breast cancer patients who received neoadjuvant chemotherapy

Background Metabolic syndrome is associated with various malignancies, including breast cancer. Non-alcoholic fatty liver disease (NAFLD) is a hepatic manifestation of metabolic syndrome. NAFLD is frequently observed during chemotherapy, but its clinical implication is unclear. The purpose of this study is to evaluate the prognostic value of NAFLD in patients with stage II/III breast cancer who received neoadjuvant chemotherapy (NAC Tx). Methods The patients with clinical stage II/III breast cancer who received NAC Tx with docetaxel and doxorubicin were enrolled. Treatment sequences were: 3 cycles of NAC Tx → surgery → 3 cycles of adjuvant chemotherapy (AC Tx), or 6 cycles of NAC Tx → surgery, allowing AC Tx at the physician9s discretion. NAFLD was determined by ultrasonography (radiologists9 decision), non-constrast computed tomography [CT] (the attenuation of liver is less than that of spleen or Results Between 2002 and 2008, 269 patients were enrolled. The median age was 45 (range 18-69), and 51.7% and 28.6% were with hormone receptor and HER2 positive tumors, respectively. The median number of NAC Tx was 3 cycles (range 1-6). NAFLD was observed in 33 (12.4%) patients at diagnosis, 52 (19.3%) after NAC Tx and 71 (26.4%) at the completion of AC Tx. The patients with NAFLD at diagnosis showed shorter overall survival than those without NAFLD (HR=2.688, 95% CI 1.259-5.747, P =0.011). The improvement of NAFLD during NAC Tx was observed in 28 (10.4%, group A) and persistence of NAFLD observed in 24 (8.9%, group B). Group A showed better prognosis in both PFS (HR 0.125, 95 % CI 0.016-0.962) and OS (no event), whereas group B showed worse PFS (HR 2.329, 95% CI 1.280-4.237) and OS (HR 3.721, 95% CI 1.727-8.015) compared to the patients without NAFLD at diagnosis. Conclusions NAFLD at diagnosis was a poor prognostic marker of OS in patients who received NAC Tx. Improvement of NAFLD during NAC TX was associated with good prognosis in terms of PFS and OS. Citation Format: Yang Y, Lee K-H, Kim T-Y, Han S-W, Oh D-Y, Kim T-Y, Han W, Moon H-G, Park IA, Noh D-Y, Im S-A. Prognostic value of non-alcoholic fatty liver disease in stage II/III breast cancer patients who received neoadjuvant chemotherapy. [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 P1-10-11.

Abstract P5-08-23: Ki-67 expression is not a valuable predictive prognostic factor when progesterone receptor expression is high in estrogen receptor-positive breast cancer

Background Immunohistochemistry markers are recognized as a predictive prognostic factor for women with breast cancer. Ki-67 and progesterone receptor (PgR) expression are reported to be independently associated with breast cancer prognosis. Some studies report high Ki-67 expression as a negative predictive marker. Whereas other studies report tendency of similar survival between high and low Ki67 cancers when PgR expression is high. In this study, we examined the prognostic significance of Ki-67 expression under PgR expression status. Methods The records of 2,366 patients were retrospectively reviewed. The patients underwent surgery for primary breast cancer from July 2009 to December 2012 at a single institution. We studied the prognostic significance of Ki-67 expression under PgR expression. We used 20% and 10% as the cut-off value for PgR and Ki-67, respectively. The end point was recurrence-free survival (RFS) evaluated by use of Kaplan-Meier analysis. Result Of the 2,366 analyzed patients, the median follow-up time was 43 months. During follow-up, 44 patients had recurrence, loco-regional recurrence developed in 23 patients and distant recurrence developed in 21 patients. In patients with low PgR expression, high Ki-67 expression group showed significantly worse prognosis compared to low Ki-67 expression group (p=0.005). On the other hand, no significant difference was shown between low and high Ki-67 expression group when PgR expression was high (p=0.637). Also multivariate analysis demonstrated that high Ki-67 expression was an independent prognostic factor only when PgR expression was low. (95% confidence interval [CI], 1.35-10.48; p=0.011) Conclusion This is the largest reported study that prognostic significance of Ki-67 expression is defined by PgR expression. Our study presents that high Ki-67 expression is inversely correlated with recurrence risk in early breast cancer patients only under low PgR expression. At high PgR expression, Ki-67 expression has no influence on breast cancer prognosis. Therefore, attention should be paid to correlation between PgR and Ki-67 expression. Citation Format: Han JH, Kang YJ, Han W, Lee H-B, Kim Y, Yoo T-K, Moon H-G, Noh D-Y. Ki-67 expression is not a valuable predictive prognostic factor when progesterone receptor expression is high in estrogen receptor-positive 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 P5-08-23.

Abstract PD4-2: Whole exome and transcriptome sequencing of 120 primary breast cancer to discover novel therapeutic target

Introduction: Many somatic mutations, structural alterations, and gene expression changes are causally implicated in oncogenesis and tumor progression, and as a result, affect clinical outcome. Although majority of breast cancer patients have benefits from therapeutics targeting tumor biology, such as estrogen receptor and HER-2, still many patients suffer from disease recurrence and metastasis. More kind of specific target therapies are needed, especially for hormone-resistant tumor and triple-negative breast cancer. Materials and Method: To find novel therapeutic target in breast cancer, here we examine the both whole exome and whole transcriptome of fresh-frozen primary breast cancer tissues from 120 patients whose clinical, pathological, and survival data are available. Patients with Stage IV disease or who received neoadjuvant chemotherapy were excluded. 36 patients had distant metastasis within 5 years from surgery, and 84 patients were NED at least 5 years. RNA and DNA were extracted and qualities were assessed in all samples. Exome and transcriptome sequencing were done using NGS technology (Illumina HiSeq 2000). As a control, exome sequencing was done for 93 normal DNA from matched patients. Single nucleotide variations (SNV) identified in cancer samples on exonic region, nonsynonymous SNV or stop gain/loss, whose quality ≥20, and not found in 93 normal samples were included. SNVs registered in dbSNP135_common or 1000 genome allele frequency >0.001 were excluded.Results and Discussion: We identified 11,684 putative somatic mutations in 7,373 genes. Of them, 6,547 were deleterious or damaging mutation by Provean or SIFT analysis. Mutations were found in potential drug target genes, such as PIK3CA(25), PTEN(3), AKT1(3), ALK(3), ROS1(2), FGFR4(3), FGFR3(2), ERBB2(2), and IDH1(1) etc. In a pathway analysis, mutations in insulin signaling pathway were most dominant. We hypothesized that driver gene and therapeutic target has to have recurrent mutation and gene expression at least more than average expression. We calculated expression “Volume” according to the median normalized FPKM value of individual gene9s RNA-seq data. With a cut-off of 3 or more mutations in each gene, 1,116 genes were selected. After the filtering of Volume Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr PD4-2.

Abstract P4-04-09: Extensive novel hybrid isoforms revealed by RNA sequencing of 120 primary breast cancer samples

Introduction Recent studies of next generation sequencing (NGS) have highlighted the extensive transcriptional heterogeneity of cancer cells. Alternative splicing is one of the evolutionary process by which cells and tissues achieve their specificity within central dogma. Also it is highly assumed to contribute to oncogenesis and thought to be a critical mechanism how cancer cells gain resistance to therapeutic agents and adapt to various circumstances. Relevance of differential splicing in breast cancer biology is mostly unknown. We performed whole transcriptome sequencing (RNA-Seq) to reveal novel splicing alterations among 120 primary breast cancer samples. Materials and Method Total RNA was prepared using the Illumina TrueSeq™ RNA sample Preparation Kit and TrueSeq mRNA library was constructed. Clustering and sequencing was done using Illumina HiSeq 2000. RNA-Seq reads were aligned to human reference genome(hg19) using TopHat software and expression was measured using cufflinks software. We used tissues extracted from previously collected 120 fresh-frozen primary breast cancer samples obtained after surgical resection whose clinicopathological data are available. Patients undergone neoadjuvant systemic therapies or stage lV disease at diagnosis were excluded. Thirty-six(30%) cases occurred distant metastasis during follow up. Hormone receptor(HR) was positive in 61(50.8%) samples, 20(16.7%) had HER2 oncogene overexpression and 36(30%) were triple negative breast cancer. Results and discussion Total 11345 novel isoforms were detected among 120 tumors. Isoforms of pseudo-genes and exon skipping of the ‘non-coding exon’ were excluded. Splice variants detected in normal reference were sorted out as well. 4045 were in-frame exon skipping and 4960 were off-frame exon skipping which may lead to protein truncation. 5036 were private exon skipping and 3969 isoforms were detected recurrently in more than 2 samples. To minimize false positivity we confined ‘exon skipping’ analysis to those with the expression level (Fragments per kilo-base of exon per million fragments mapped, FPKM) of the skipped exon below 0.1 compared to the adjacent exons. Mean number of exon skipping events per sample was 196.8 (range 75-299, SD 35.9). There were no differences in numbers of exon skipping event among breast cancer subtypes nor distant metastasis. We have identified novel exon skipping in ESR1, CHEK2, EIF3E, FGFR, MAP2K, PIK3R2, TERT, VAV3 genes which is strongly suspected to be novel driver isoforms and is under validation process. Conclusion We performed whole-transcriptome sequencing with a large set of primary breast cancer samples and revealed extensive transcriptional heterogeneity by isoform profiling. As distinguishing the natural transcriptomic dynamics from oncogenic ‘driver’ isoform is a major challenge, validation and functional studies are ongoing. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-04-09.