Abstract P3-10-01: Single-cell profiling identifies hypoxic carcinoma cells as source of an immunosuppressive VEGFA metagene

Background:

We have previously shown that expression of the IL8/VEGFA-metagene eliminates the good prognostic effect of TILs in TNBC (PMID 21978456, 28750120). We also showed that the VEGFA metagene predicted response to neoadjuvant bevacizumab in the GeparQuinto trial (Karn 2017 SABCS #851166). The main cellular sources of the transcripts that comprise the VEGFA metagene are unknown since mRNA profiling of bulk biopsies contains signals from different cell types.

Methods:

Individual genes that comprise the VEGFA metagene were measured in bulk tissue- and single cell-RNA-Seq from breast cancer subtypes and normal cells on different platforms (Affymetrix n=4915, Agilent n=597, Illumina n=2433, RNA-Seq n=1215, Exome Capture RNA-Seq n=226, HTG-Seq n=243, sc-RNA-Seq n=24710). For blinded, orthogonal validation we performed immunohistochemistry. Effect of neoadjuvant chemotherapy with or without bevacizumab was studied by RNA-Seq and IHC on samples from GeparQuinto trial. SWOG S0800 (GSE114403), PROMIX (GSE87455), and GeparSixto trials were used for validation. TCGA was mined for mutations and somatic CNA. RNA-Seq from GeparNuevo was used for correlation with checkpoint inhibitor treatment.

Results:

We identified a stable core of six genes (VEGFA, ANGPTL4, ADM, NDRG1, DDIT4, CSTB) in different cohorts. Strong expression of this signature was mainly restricted to TNBC subtype and associated with poor prognosis within this subgroup. Single cell RNA-Seq of breast epithelial cells from 4 reduction mammoplasties and 4 TNBC revealed that these genes are coexpressed in individual epithelial cells and not associated with endothelial cells. In line with their presumed functions in cellular stress and hypoxia, immunohistochemistry revealed strong para-necrotic expression in TNBC. Moreover, high gene expression in TNBC was associated with mutations in DNA damage control pathways, somatic copy number alterations, and lower TILs. While chemotherapy led to downregulation, bevacizumab increased expression. In multivariate analysis, high pretreatment values predict pCR to both bevacizumab and chemotherapy (OR 2.40, P=0.006), which may be explained by sensitivity of tumors which are already under cellular stress. On the other hand, expression of the VEGFA metagene seems to create an immunosuppressive environment that counteracts the positive prognostic effect of TILs. In pre-treatment biopsies from the GeparNuevo checkpoint inhibitor trial we found a negative correlation of VEGFA metagene expression with the amount of the recently identified tissue-resident memory T cell subset (CD8TRM, PMID 29942092; P=0.002), while the subsequent increase of CD8TRM during treatment was larger in tumors with high VEGFA (P=0.019).

Conclusions:

Perinecrotic carcinoma cells under stress from hypoxia and or chromosomal instability are the source of the VEGFA metagene signature. Its predictive value in TNBC suggests estimating and reporting the amount of necrosis in the pathology report may be helpful in predicting response to preoperative chemotherapy, and could be used as stratification factor in clinical trials. The signature indicates an immunosuppressive environment and should be further studied in the context of immune therapies in combinations with anti-angiogenic treatment.

Citation Format: Karn T, Denkert C, Sinn BV, Weber K, Nekljudova V, Rody A, Meissner T, Hatzis C, El-Balat A, Becker S, Solbach C, Untch M, Schneeweiss A, von Minckwitz G, Loibl S, Pusztai L, Holtrich U. Single-cell profiling identifies hypoxic carcinoma cells as source of an immunosuppressive VEGFA metagene [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-10-01.

Abstract P4-03-01: Pathway level complementarity of germline and somatic events in breast cancer

Background: Progression from a normal cell state to cancer requires multiple genomic hits in key regulatory pathways. In the case of hereditary cancer syndromes, some of these hits occur in the germline, but additional somatic mutations are required for malignant transformation. We hypothesize that this paradigm could be extended to sporadic cancers as well. What somatic mutation function as a cancer driver event may be determined by the constellation of germline variants a person is born with. We propose that even rare, non-recurrent, high functional impact germline variants in genes involved in cancer-related pathways could influence the biological impact of somatic mutations in other cancer-related genes. The goal of the current analysis was to examine associations between pathway alterations caused by high functional impact germline variants or somatic mutations in the “hallmarks of cancer” pathways in breast cancer.

Methods: We obtained germline DNA sequencing and copy number variation (CNV) data from the breast cancer TCGA cohort. After population clustering with the HapMap cohort, we selected a homogeneous group of 796 patients of Western European ancestry and downloaded the matching somatic mutations (SNVs and INDELs) that were available for 750 cases, that comprise the current study population. Germline CNVs were classified as recurrent or rare losses or gains. Potentially pathogenic germline variants (SNPs) were obtained from the PanCancer Altas project. All germline or somatic mutations were mapped at the gene level to the 50 Cancer Hallmarks pathway collection. We designated a pathway mutated if at least 1 gene had a germline or a somatic mutation. Complementarity between pathway alterations by germline and somatic events were evaluated using the Fisher exact test adjusted for multiple comparisons.

Results: At the germline level, 2,057 genes were affected by CNVs (mean 30, range 3-151 genes/patient), and a total of 43 genes carried germline pathogenic SNPs that affected 13.8% of the patients. At the somatic level, we detected 40,881 high functional impact mutations (mean 54.3, range 1-3889 mutations/patient) in 13,080 genes (mean 50.8, range 1-3166 genes/patient). The 50 Cancer Hallmark pathways contained 4386 genes (mean 146.5, range 32-200 genes/pathway), and were mutated in the majority of the patients (85% germline, 93% somatic). Several pathways, such as HEME_METABOLISM, INTERFERON_ALPHA_RESPONSE, and KRAS_SIGNALING, were frequently affected by germline alterations, while the somatic mutations were most frequently involved in the COMPLEMENT, E2F_TARGET, and UV_RESPONSE_UP. Interaction analysis revealed co-occurrence between MYC_TARGETS_V1 (germline) and UV_RESPONSE_DN (somatic) or MTORC1_SINGALING (somatic) (p<0.01), and TNFA_SIGNALING_VIA_NFKB (germline) and IL6_JAK_STAT3_SIGNALING (germline) with E2F_TARGETS (somatic) (p<0.01). We also observed an exclusive relationship between germline alterations in BILE_ACID_METABOLISM and somatic mutations in COMPLEMENT pathway (p<0.01).

Conclusions: Our results highlight the importance of pathway-level analysis of germline alterations in breast cancer, which might help to understand the interrelationship between germline and somatic alterations in breast cancer.

Citation Format: Qing T, Marczyk M, Wali V, Gunasekharan V, Patwardhan G, Pusztai L, Hatzis C. Pathway level complementarity of germline and somatic events in breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P4-03-01.

Abstract P2-07-03: Refining neoadjuvant predictors of three year distant metastasis free survival: Integrating volume change as measured by MRI with residual cancer burden

Background: Patients achieving a pathologic complete response (pCR) following neoadjuvant therapy have significantly improved event-free survival relative to those who do not; and pCR is an FDA-accepted endpoint to support accelerated approval of novel agents/combinations in the neoadjuvant treatment of high risk early stage breast cancer. Previous studies have shown that recurrence risk increased with increasing burden of residual disease (as assessed by the RCB index). As well, these studies suggest that patients with minimum residual disease (RCB-I class) also have favorable outcomes (comparable to those achieving a pCR) within high risk tumor subtypes. In this study, we assess whether integrating RCB with MRI functional tumor volume (FTV), which in itself is prognostic, can improve prediction of distant recurrence free survival (DRFS); and identify a subset of patients with minimal residual disease with comparable DRFS as those who achieved a pCR. Imaging tools can then be used to identify the subset that will do well early and guide the timing of surgical therapy.

Method: We performed a pooled analysis of 596 patients from the I-SPY2 TRIAL with RCB, pre-surgical MRI FTV data and known follow-up (median 2.5 years). We first assessed whether FTV predicts residual disease (pCR or pCR/RCB-I) using ROC analysis. We applied a power transformation to normalize the pre-surgical FTV distribution; and assessed its association with DRFS using a bi-variate Cox proportional hazard model adjusting for HR/HER2 subtype. We also fitted a bivariate Cox model of RCB index adjusting for subtype; and assessed whether adding pre-surgical FTV to this model further improves association with DRFS using a likelihood ratio (LR) test. For the Cox modeling, penalized splines approximation of the transformed FTV and RCB index with 2 degrees of freedom was used to allow for non-linear effects of FTV and RCB on DRFS.

Result: Pre-surgical MRI FTV is significantly associated with DRFS (Wald p<0.00001), and more effective at predicting pCR/RCB-I than predicting pCR alone (AUC: 0.72 vs. 0.65). Larger pre-surgical FTV remains associated with worse DRFS adjusting for subtype (Wald p <0.00001). The RCB index is also significantly associated with DRFS adjusting for subtype (Wald p<0.00001). Adding FTV to a model containing RCB and subtype further improves association with DRFS (LR p=0.0007). RCB-I patients have excellent DRFS (94% at 3 years compared to 95% in the pCR group). Efforts are underway to identify an optimal threshold for dichotomizing pre-surgical FTV and FTV change measures for use in combination with pCR/RCB-I class to generate integrated RCB (iRCB) groups as a composite predictor of DRFS.

Conclusion: Pre-surgical MRI FTV is effective at predicting minimal residual disease (RCB0/I) in the I-SPY 2 TRIAL. Despite the association between FTV and RCB, FTV appears to provide independent added prognostic value (to RCB and subtype), suggesting that integrating MRI volume measures and RCB into a composite predictor may improve DRFS prediction.

Citation Format: Hylton NM, Symmans WF, Yau C, Li W, Hatzis C, Isaacs C, Albain KS, Chen Y-Y, Krings G, Wei S, Harada S, Datnow B, Fadare O, Klein M, Pambuccian S, Chen B, Adamson K, Sams S, Mhawech-Fauceglia P, Magliocco A, Feldman M, Rendi M, Sattar H, Zeck J, Ocal I, Tawfik O, Grasso LeBeau L, Sahoo S, Vinh T, Yang S, Adams A, Chien AJ, Ferero-Torres A, Stringer-Reasor E, Wallace A, Boughey JC, Ellis ED, Elias AD, Lang JE, Lu J, Han HS, Clark AS, Korde L, Nanda R, Northfelt DW, Khan QJ, Viscusi RK, Euhus DM, Edmiston KK, Chui SY, Kemmer K, Wood WC, Park JW, Liu MC, Olopade O, Tripathy D, Moulder SL, Rugo HS, Schwab R, Lo S, Helsten T, Beckwith H, Haugen PK, van't Veer LJ, Perlmutter J, Melisko ME, Wilson A, Peterson G, Asare AL, Buxton MB, Paoloni M, Clennell JL, Hirst GL, Singhrao R, Steeg K, Matthews JB, Sanil A, Berry SM, Abe H, Wolverton D, Crane EP, Ward KA, Nelson M, Niell BL, Oh K, Brandt KR, Bang DH, Ojeda-Fournier H, Eghtedari M, Sheth PA, Bernreuter WK, Umphrey H, Rosen MA, Dogan B, Yang W, Joe B, I-SPY 2 TRIAL Consortium, Yee D, Pusztai L, DeMichele A, Asare SM, Berry DA, Esserman LJ. Refining neoadjuvant predictors of three year distant metastasis free survival: Integrating volume change as measured by MRI with residual cancer burden [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-07-03.

Abstract PD5-10: Immune characterization of matched primary and multiple metastatic samples issued from an institutional autopsy cohort

Introduction

While immune infiltrates have already been extensively characterized in primary tumors (P), data on breast cancer metastases (M) remains limited. To this end we quantified and qualified the immune cells in a unique cohort of multiple matched P and M samples selected from an institutional breast cancer autopsy cohort.

Patients and methods

Twenty-three patients were selected from an institutional autopsy program (Semmelweis University, Budapest, Hungary) based on matched P and M sample availability (124 samples). All samples were centrally characterized for estrogen (ER), progesterone (PR) and HER2 receptors. The primary molecular subtypes were as follows: 9 ER+/PR+/HER2-, 8 triple negative and 6 HER2+. Ten patients relapsed ≤1 year after diagnosis and were further referred to as “early relapsers”, as opposed to the remaining qualified as “late relapsers”. Immunohistochemistry (IHC) was carried out against CD3/CD20 and CD4/CD8 in 21 patients (119 samples). Tumor infiltrating lymphocytes (TILs) were assessed on hematoxylin and eosin (H&E) and CD3-stained slides. Gene expression data were generated using the NanoString nCounter assay (PanCancer Immune Profiling Panel) for 11 patients (35 samples) and analyzedusing the R package NanoStringQCPro. The scores from published immune gene signatures were calculated as a weighted sum of the expressions of their genes. All samples were analyzed for 22 immune cell subtypes relative abundance using CIBERSORT.

Results

TILs assessed on H&E and CD3-stained slides were weakly correlated (Rho= 0.38, p<.001). TIL levels as well as the number of tertiary lymphoid structures (TLS) were significantly lower in Ms as compared to Ps (p<.001). Among the different metastatic sites, the lung was more infiltrated when considering CD3+ and CD4+ cells (p=.01 and .02, respectively). We further observed significantly higher levels of TILs, CD3+, CD4+ and CD8+ cells in the Ms but not in the Ps from late relapsers as opposed to those from early relapsers. Gene expression analyses further confirmed these observations as several immune gene signatures displayed significantly higher scores in the Ms from late compared to early relapsers. An unsupervised analysis identified 13 genes significantly differentially expressed between Ps and Ms: CSF1R, CXCL14, CYBB, IL21R, IL2RB, TNF and TNFSF15 were upregulated in Ps while BCL2L1, C7, HSD11B1, and PSMB7 were upregulated in Ms. The matched P/M CIBERSORT analyses revealed a distinct composition of immune cell types between P and M of a same patient. Apart from a potential increase in M0 macrophages, no common trait was observed in immune cell composition between the Ms from the different patients.

Conclusion

This is to the best of our knowledge the first study characterizing the immune infiltration in patients with multiple matched P and M samples. The results suggest that Ms have not only a globally lower immune infiltration as compared to Ps, but also a different immune composition. Additionally, Ms from late relapsers are more infiltrated as compared to early relapsers. The present data also uncovers not only important inter-patient but also intra-patient immune heterogeneity, which should be taken into consideration for optimal treatment decision.

Citation Format: Szekely B, Bareche Y, Van den Eynden G, Salgado R, Buisseret L, Garaud S, Willard-Gallo K, Hatzis C, Szasz M, Kulka J, Larsimont D, Sotiriou C, Pusztai L, Desmedt C. Immune characterization of matched primary and multiple metastatic samples issued from an institutional autopsy cohort [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr PD5-10.

Abstract P3-05-01: Age-related methylation signals of breast cancer risk in blood

BACKGROUND: Age is the biggest risk factor for developing breast cancer, which suggests that the biological aging process is a direct driver of cancer etiology. In all normal tissues, DNA methylation status changes systematically with age and is believed to mediate the biological consequences of aging. DNA methylation patterns, commonly referred to as 'Epigenetic Clocks', can be used as measures of aging. However, the chronologic and epigenetic ages can have subtle differences in different individuals. We hypothesize that accelerated epigenetic aging (i.e. DNA methylation indicating an older age than the chronologic age of the individual) is a risk factor for breast cancer development.

METHODS: We used DNA methylation data from blood samples and clinical data from n=2,107 participants in the Women's Health Initiative (WHI) and tested the association between breast cancer risk and two of the most commonly used epigenetic clocks by Horvath (based on 353 CpGs) and Hannum et al. (based on 71 CpGs). DNA methylation in whole blood was measured using the Illumina HumanMethylation450 BeadChip. We used Cox proportional hazard models to assess the association between two epigenetic age predictors (calculated using the algorithms by Horvath and Hannum et al.) and subsequent risk of breast cancer. The model was adjusted for several breast cancer risk factors including: chronologic age at the time of blood sampling, observational vs. clinical trial, clinical trial arm, race/ethnicity, education, BMI, waist-hip ratio, smoking, alcohol, age at menopause, age at menarche, number of pregnancies, age at first birth, previous mastectomy, months breastfed, and cell count estimates. Family history data and BRCA mutation status was incomplete in the WHI and therefore could not be included in our analysis.

RESULTS: Increased epigenetic age determined by the Horvath clock relative to chronological age was associated with increased future incidence of invasive breast cancer, even after adjusting for known risk factors (HR=1.04, P=0.03). Utilizing the Hannum clock, we found no significant association between epigenetic age and breast cancer risk (HR=1.01, P=0.568). When we included both age predictors as independent variables in a single model, the strength of the association between the Horvath epigenetic age and breast cancer risk increased (Horvath HR=1.09, P=6.3e-5; Hannum HR=0.95, p=0.077), such that every one year increase in epigenetic age relative to chronological age was associated with a 9% increased risk of future breast cancer. These results suggest that the aging signal in the Horvath clock that is unique from that captured by Hannum is what drives the specific association with breast cancer.

CONCLUSIONS: Our results support the hypothesis that “accelerated” epigenetic aging measured in the blood increases breast cancer risk. We also demonstrate that the two epigenetic clocks capture different aspects of aging, only some of which have implications for breast cancer risk. Epigenetic clocks may assist in targeting breast cancer screening to higher risk populations in the future, and understanding the biological mechanisms that are altered by the epigenetic changes may lead to new risk reduction strategies.

Citation Format: Hofstatter EW, Levine M, Hatzis C, Pusztai L. Age-related methylation signals of breast cancer risk in blood [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P3-05-01.

Abstract P2-06-06: Targeting loss of isoenzyme diversity as a novel therapeutic strategy in breast cancer

Background: Several metabolic steps are mediated by distinct proteins or isoenzymes that catalyze the same reaction, providing redundancy of metabolic functions. Metabolic states are frequently altered in cancer to support survival and proliferation in hypoxic and otherwise hostile microenvironments, and metabolic re-wiring often involve loss of isoenzyme diversity. We hypothesize that targeting enzymes that have lost isoenzyme diversity in cancer, but not in normal cells, provides an opportunity to selectively target cancers. In this study, we assessed mRNA expression of all known human isoenzyme families in breast cancer and normal breast tissue and identified isoenzymes with loss of diversity within each breast cancer subtype.

Methods: We obtained RNAseq data from cancer and patient-matched normal breast tissues from the TCGA (N=66 HR+, N=24 HER2+, and N=15 TNBC tumors). We retrieved annotated human isoenzyme families from the ENZYME nomenclature database. We compared expression in cancer and matched normal samples from the same patient to identify isoenzymes that had i) same or increased expression of the target isoenzyme in cancer vs normal and ii) reduced expression of the complementary isoenzymes in cancer. We developed five scores that capture various elements of these characteristics and prioritized candidates as targets based on clustering and their combined ranking based on the five scores. We validated overexpression of the candidate isoenzymes relative to other isoforms in breast cancer microarray data from ArrayExpress (E-GEOD-76250: 33 TNBC, and E-GEOD-70951: 30 TNBC, 108 HR+, 10 HER2+).

Results: We identified 321 enzymes in the TCGA discovery cohort that correspond to 829 unique isoenzymes. Overall, 636, 483 and 429 isoenzymes were differentially expressed in HR+, HER2+ and TNBC cancers, respectively, compared to corresponding normal samples. Of these, 308 isoenzymes were differentially expressed relative to normal in all 3 subtypes. In all, 112 and 92, and 84 were selected as candidate isoenzyme therapeutic targets in HR+, HER2+ and TNBC, respectively. 23 isoenzymes prioritized in clustering step were further validated. Finally, 6 isoenzymes were validated in HR+ (ALDOA, GUSB, GYG1, MIF, P3H1, PCK2), 10 in HER2+ (ALDH1L2, ALDOA, GLYATL2, GUSB, GYG1, GYS1, MIF, P3H1, PCK2, PTGS1) and 12 in TNBC (ADSS, ALAS1, ALDH1L2, ALDOA, ART3, GLYATL2, GUSB, GYS1, HS3ST1, MIF, PCK2, SOAT1), as potential targets for breast cancer treatment. Of these, 5 potential isoenzyme targets (ALDH1L2, GUSB, GLYATL2, MIF, PCK2), which were mostly hydrolases and transferases, were further selected for ongoing experimental validation in the laboratory. Decreased expression of the complementary isoforms of these 5 targets were primarily due to DNA methylation of the genes in cancer.

Conclusions: We found that loss of isoenzyme diversity is a broad phenomenon in breast cancers that may be explored therapeutically. We identified several instances of “isoenzyme addiction” in which cancers depend exclusively on a single isoenzyme while downregulating via methylation the complementary isoenzymes, providing cancer-specific targeting opportunities. We are currently validating several of these targets in cell line models.

Citation Format: Marczyk M, Gunasekharan V, Wali VB, Shi W, Patwardhan G, Qing T, Pusztai L, Hatzis C. Targeting loss of isoenzyme diversity as a novel therapeutic strategy in breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-06-06.

Abstract P4-08-20: Pre-analytical effects of FFPE extraction methods on targeted and whole transcriptome sequencing assays for endocrine sensitivity in metastatic breast cancer

Background: The clinical management of patients with metastatic HR-positive breast cancer is often uncertain due to decreased sensitivity to anti-estrogen therapy over time. Recently, we developed a targeted RNAseq based 18-transcript SET ER/PR assay of endocrine sensitivity from biopsies of metastatic cancer. In this work we assess the effect of pre-analytical factors, specifically RNA extraction methods for FFPE tissue samples, on the reliability of the targeted RNAseq assay.

Methods: FFPE blocks and matched fresh frozen (FF) sections from 12 tumors were collected at MD Anderson Cancer Center. RNA from FFPE slides was extracted in duplicate using three kits (Norgen, Qiagen, Roche), and RNAseq libraries from all samples were prepared using Kapa Total RNAseq kit. Targeted RNA libraries were prepared using droplet-based PCR (RainDance), and also by transcriptome-wide RNAseq for comparison. Reads were mapped to genomic sequence using STAR and expression was quantified using RSEM. Expression data were normalized based on expression of 10 reference genes. The effect of FFPE RNA extraction kit on the reliability of the SET index was assessed using linear mixed effects model (LME) analysis, and agreement with FF was assessed using the concordance correlation coefficient (CCC).

Results: Analysis of the whole transcriptome RNAseq data confirmed minimal 3'-end transcript bias from FFPE samples, irrespective of transcript size or FFPE kit. All 18 genes included in the SET index had high overall concordance between FFPE and FF (median CCC percentile=98.8, range 57.2-99.9 for Norgen; similar for the other two kits) and relatively consistent bias across genes, as estimated by the random effects of the LME model. Furthermore, compared to random 18-gene indices, concordance in the SET index values between FF and FFPE was higher than 99.8% of the random samples, verifying the analytical reliability of the selected genes. For the targeted RNAseq assay, RNA from FFPE extracted with the Norgen kit showed the highest concordance compared to FF (CCC=0.956, 95%CI 0.871-0.985). In general, the analytical variation of SET from FFPE samples was greater than that from FF (1.71-2.71 fold greater), with the lowest variation associated with the Norgen kit. The SET index values from targeted RNAseq for both FF and FFPE samples were consistently lower compared to transcriptome-wide RNAseq but were highly correlated, with the Norgen kit having the highest correlation between targeted and transcriptome-wide RNAseq (rho=0.915).

Conclusions: All three FFPE RNA extraction kits have excellent analytical performance compared to FF samples. The Norgen kit may be marginally better yielding higher concordance with FF and lower analytical variation between replicates. All genes in the SET ER/PR showed very good analytical performance in comparison to random indices and individual genes. Targeted gene RNA sequencing appears very promising as a platform for clinical deployment of quantitative assays, showing only a small (fixable) bias compared to RNAseq.

Citation Format: Marczyk M, Fu C, Lau R, Du L, Trevarton AJ, Sinn BV, Gould RE, Symmans WF, Hatzis C. Pre-analytical effects of FFPE extraction methods on targeted and whole transcriptome sequencing assays for endocrine sensitivity in metastatic breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P4-08-20.

Pathway level alterations rather than mutations in single genes predict response to HER2-targeted therapies in the neo-ALTTO trial

Background

We performed whole-exome sequencing of pretreatment biopsies and examined whether genome-wide metrics of overall mutational load, clonal heterogeneity or alterations at variant, gene, and pathway levels are associated with treatment response and survival.

Patients and Methods

Two hundred and three biopsies from the NeoALTTO trial were analyzed. Mutations were called with MuTect, and Strelka, using pooled normal DNA. Associations between DNA alterations and outcome were evaluated by logistic and Cox-proportional hazards regression.

Results

There were no recurrent single gene mutations significantly associated with pathologic complete response (pCR), except PIK3CA [odds ratio (OR) = 0.42, P = 0.0185]. Mutations in 33 of 714 pathways were significantly associated with response, but different genes were affected in different individuals. PIK3CA was present in 23 of these pathways defining a ‘trastuzumab resistance-network’ of 459 genes. Cases with mutations in this network had low pCR rates to trastuzumab (2/50, 4%) compared with cases with no mutations (9/16, 56%), OR = 0.035; P < 0.001. Mutations in the ‘Regulation of RhoA activity’ pathway were associated with higher pCR rate to lapatinib (OR = 14.8, adjusted P = 0.001), lapatinib + trastuzumab (OR = 3.0, adjusted P = 0.09), and all arms combined (OR = 3.77, adjusted P = 0.02). Patients (n = 124) with mutations in the trastuzumab resistance network but intact RhoA pathway had 2% (1/41) pCR rate with trastuzumab alone (OR = 0.026, P = 0.001) but adding lapatinib increased pCR rate to 45% (17/38, OR = 1.68, P = 0.3). Patients (n = 46) who had no mutations in either gene set had 6% pCR rate (1/15) with lapatinib, but had the highest pCR rate, 52% (8/15) with trastuzumab alone.

Conclusions

Mutations in the RhoA pathway are associated with pCR to lapatinib and mutations in a PIK3CA-related network are associated with resistance to trastuzumab. The combined mutation status of these two pathways could define patients with very low response rate to trastuzumab alone that can be augmented by adding lapatinib or substituting trastuzumab with lapatinib.

Abstract P2-04-02: Comparison of DNA methylation patterns in normal breast tissue from women with and without breast cancer

BACKGROUND: Increasing evidence suggests that epigenetic mechanisms play critical roles in the development of breast cancer. However, precise DNA methylation signatures associated with breast cancer susceptibility remain unknown. We sought to compare DNA methylation changes in the normal breast tissue of women with and without breast cancer to identify patterns of aberrant DNA methylation in women with breast cancer.

METHODS:Samples of normal breast tissue were collected from four cohorts of women: age &lt; 50 years with and without breast cancer, and age ≥50 years with and without breast cancer. Normal breast tissue from healthy women was obtained from the Komen Tissue Bank at IU Simon Cancer Center and from women presenting for reduction mammoplasty at Yale New Haven Hospital. Normal breast tissue from women with breast cancer was obtained from patients undergoing adjuvant total mastectomy at Yale Breast Center. DNA was extracted using Qiagen AllPrep Universal kit. Raw data files in idat format were imported to Partek Genomics Suite 6.6 for normalization and differential methylation analysis. Raw intensities were normalized using With Array Normalization (SWAN) method. Principal component analysis (PCA) were performed as quality control. Differentially methylated loci (DML) between control and breast cancer groups were detected when False discovery rate (FDR) &lt; 0.05 and fold change &gt; 1.5. Functional enrichment analysis of genes with DML in the gene body were conducted using METACORE™. Pathways with FDR &lt; 0.05 were selected.

RESULTS: Ninety-three normal breast tissue samples from 89 subjects were analyzed (breast cancer=40, unaffected=53). Comparison of DNA methylation patterns between women with and without breast cancer revealed 200 DMLs. The majority of DMLs (186) were hyper-methylated in breast cancer patients, and 48 DMLs locate in enhancers of genes. 170 DMLs locate in 134 genes, enriched in two pathways: (1) Cell adhesion_Endothelial cell contacts by junctional mechanisms, and (2) Neurophysiological process_Constitutive and regulated NMDA receptor trafficking. Genes associated with cell adhesion and cell contacts included: ACTN2, GJA4, GJA7 and MAGI1. Two hyper-methylated loci were found in enhancers of ACTN2. In addition, one hyper-methylated locus in GJA4, one hyper-methylated and one hypo-methylated loci in GJA7, and two hyper-methylated loci in MAGI1 were detected in breast cancer patients. Genes associated with NMDA receptor trafficking include: TPK1, ADCY4 and LIN7C. One and two loci were found in TPK1 and ADCY4, respectively, that were hyper-methylated in normal breast tissue from cancer patients in the gene body, while a hypo-methylated locus in breast cancer patients was identified in LIN7C.

CONCLUSIONS: Comparison of DNA methylation patterns of normal breast tissue from women with and without breast cancer reveal specific mechanistic pathways and genes that are differentially methylated in women with breast cancer. DNA methylation of normal breast tissue deserves further study as a potential biomarker for breast cancer risk stratification and may lend new insight into mechanisms of breast cancer development.

Citation Format: Hofstatter EW, Zhu Y, Horvath S, Chagpar AB, Wali VB, Bossuyt V, Storniolo AM, Hatzis C, Patwardhan G, Von Wahlde M-K, Butler M, Epstein L, Stavris K, Sturrock T, Au A, Kwei S, Pusztai L. Comparison of DNA methylation patterns in normal breast tissue from women with and without breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-04-02.

Abstract P1-07-10: Molecular subtypes of triple negative breast cancer have different sensitivity signature to established chemotherapy agents

Background: Triple negative breast cancer (TNBC, ER/PR/HER2 negative) patients have the worst prognosis among all breast cancer patients. Recently, six molecular subtypes have been proposed to sub-divide TNBC into clinically relevant sub-cohorts. It is unknown whether these molecular subtypes have different sensitivity against known chemotherapy agents. Here, we identified the cross-section of predictive and prognostic genes in a large TNBC dataset.

Methods: An integrated database comprising 2,071 TNBC patients with transcriptome-level gene expression data and clinical follow-up was set up from 52 different GEO and EGA datasets. The database included n=418 basal-like 1, n=170 basal-like 2, n=465 immunomodulatory, n=384 mesenchymal, n=207 mesenchymal stem-like, and n=427 luminal androgen receptor samples. The prognosis was computed for all genes (n=54,675) in each of the six subtypes separately using Cox proportional hazards regression. Another database (n=1,641) was set up for tumors with known therapy response as defined by pathological complete response (PCR) – in this, the correlation to response against a taxanes/anthracycline chemotherapy was computed using Mann-Whitney analysis. The overlap between the two datasets was determined at p&lt;0.0001.

Results: The proportion of responder patients was lowest in basal like 2 subtype. There were more genes significant for response to anthracyclines (n=6,118) than to taxanes (n=3,927). When comparing each subtype separately, most genes were prognostic in basal-like 1 and luminal androgen receptor subtypes (n=73/110 and n=73/111 respectively for taxane/anthracycline response genes). Almost no genes were overlapping to the immunmodulatory and mesenchymal stem-like subtypes (n=8/14 and n=3/7, respectively). When comparing anthracycline and taxane treatment across all six subtypes, there was no significant difference. We have to note that the different sample sizes in the different TNBC cohorts provide different power to detect an association.

Discussion: Based on our results, TNBC tumors in the basal-like 1 and luminal androgen receptors are most likely to benefit from chemotherapy while there is no difference for the selection between the two most common agents.

Citation Format: Győrffy B, Nagy Á, Santarpia L, Hatzis C. Molecular subtypes of triple negative breast cancer have different sensitivity signature to established chemotherapy agents [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P1-07-10.

Abstract PD6-02: Immunological differences between primary and metastatic breast cancer

Background: Little is known about how the immune microenvironment of breast cancer evolves during disease progression. Immunological differences between primary and metastatic lesions may explain discordant results of clinical trials that showed low tumor response rates with immune checkpoint therapy in metastatic breast cancer but high rates of pathologic complete response in early stage disease. The goal of this project was to examine TIL counts, PD-L1 protein expression and immune gene mRNA expression in primary tumors (P) and metastatic lesions (M).

Methods: FFPE blocks of primary breast cancers and core needle biopsies of matching distant metastases from 54 patients (n=104 samples). TIL count was assessed on H&E slides for 39 paired tissues and is reported as % of TIL in the stroma. PD-L1 protein expression was detected with immunohistochemistry (IHC, E1L3N antibody) in 36 pairs, samples with &gt; 1% cells showing staining were considered PD-L1 positive which was determined separately for tumor and stroma. The expression of 770 immune-related genes was measured using the Nanostring PanCancer Immune Gene Panel in 31 P and 17 M, including 10 paired cases. Genes were organized into 14 immune cell type (total T, Th1, Treg, Total CD8, exhausted CD8, Cytotoxic T, B, NK, NK-CD56, Mast cell, CD45, Dendritic cell, macrophage, neutrophil) and 22 immune function metagenes. Differences in mean expression in P and M were assessed using Fisher exact and Mann-Whitney tests without adjustment for multiple comparisons due to overlap in metagene membership.

Results: Mean TIL counts (14% vs 20%, p=0.026) and stromal PD-L1 positivity by IHC (14% vs 54%, p=0.004) were significantly lower in M. PD-L1 positivity in tumor cells was similar (25% in M vs 42%, in P p=0.14). The total TIL gene expression score (2.48 vs 2.8 p=0.018) and all immune cell type metagenes, except neutrophils, had lower absolute expression levels in M. The relative abundance of neutrophils (0.035 vs -0.38, p=0.0001) and macrophages (0.62 vs 0.38 p=0.0013) increased in M. Among the 22 immune function metagenes, T, B and NK cell functions, cytotoxicity, chemokine and TNF superfamily expression, regulation and pathogen defense were significantly lower in M and none showed significantly increased expression in M. Pro-inflammatory/immune-activating cytokines of IL-6, CCL-5,-12,-19,-22 and CXCL-5,-9,-10,-11 were all significantly lower in M. The greatest drop was seen for CXCL-9 (2243 vs 422, p&lt;0.0001) and CCL-19 (2537 vs 309, p&lt;0.0001). No cytokine showed increased expression in M. Only 6 immune genes (C7, GPI, MAPK1, TAB1, TLR5, PVR) showed significantly higher (p&lt;0.05) expression in M, the greatest increase was for Complement C7 (member of the terminal complement pathway membrane attack complex; 484 vs 3499, p=0.03) and GPI (glucose 6-phosphate isomerase that induces immunoglobulin secretion; 1967 vs 3011, p=0.01).

Conclusions: Breast cancer metastases exist in an attenuated immune microenvironment. Most immune cell subtypes, immune functions, and immune-associated gene expression are lower in M compared to P, consistent with immune escape. Metastatic lesions have higher relative abundance of macrophages and neutrophils, which suggest new therapeutic opportunities.

Citation Format: Szekely B, Bossuyt V, Li X, Baine M, Silber A, Sanft T, Hofstatter E, Mougalian S, Baghwagar S, Neumeister V, Pelekanou V, Hatzis C, Pusztai L. Immunological differences between primary and metastatic breast cancer [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr PD6-02.

Abstract P2-09-01: T-cell receptor beta chain variable region (TRBV) expression patterns predict response to combined trastuzumab/lapatinib treatment in the NeoALTTO/BIG-1-06 trial

Background: Dual anti-HER2 blockade resulted in increased pathologic complete response rate (pCR) in the 3 arm NeoALTTO trial. High immune gene expression and the absence of PIK3CA pathway mutations are predictive of pCR in all treatment arms but no markers were identified that could predict which patients require dual HER2 targeted therapy. The goal of this analysis was to examine if TRBV expression could add to the predictive function of previously identified immune markers.

Patients and Methods: We analyzed RNA and Whole Exome sequencing data from 245 cancers (54% of all patients) included in the trial. The TRBV reference sequences were obtained from the International ImMunoGeneTics information system. Reads were aligned using a custom BLAST mapping pipeline and normalized by the total number of aligned reads in each sample. We calculated 3 T cell receptor metrics for each tumor including (i) total TRBV chain expression level, (ii) Shannon entropy of the normalized unique TRBV-expression frequencies which reflect TCR diversity and (iii) we also used non-negative matrix factorization (NMF) to define TRBV co-expression metagenes (TRBVMG). We evaluated correlation between these metrics and immune and proliferation gene expression signatures and genomic features of the cancer including clonal heterogeneity and mutation load. We assessed association between TRBV and pCR using multivariate logistic regression.

Results: 65 distinct TRBV variants showed heterogeneous expression levels across cancers with strong co-expression patterns. Total TRBV expression correlated strongly with immune metagene expression (Spearman's ρ=0.93, P&lt;0.001), but entropy had a weaker, inverse correlation with immune metagene expression (Spearman's ρ=-0.40, P&lt;0.001). Associations between TRBV metrics and mutation load and clonal heterogeneity were weak. pCR correlated with higher total TRBV expression (Spearman's ρ=0.17, P&lt;0.05). Correlation between entropy and pCR was non-significant (odds ratio (OR) for regressing entropy with pCR was &lt;1). NMF identified 4 distinct TRBVMGs that showed substantial expression variation within immune cell rich cancers. ER-status, proliferation and immune-gene expression adjusted logistic regression analysis including a treatment-arm interaction term revealed that TRBVMG-2, characterized by high expression of TRBV4.3, TRBV6.3 and TRBV7.2 variants, was associated with higher pCR rate in patients treated with trastuzumab plus lapatinib (Interaction OR=3.23 adjusted P=0.03). In immune-rich cancers, TRBVMG-2 expression above the median was associated with higher pCR rate in the dual HER2 targeted treatment arm compared to the other arms (68% vs 21%, Fisher exact test P&lt;0.001). Patients with immune cell rich cancers but TRBVMG-2 expression below the median had similar pCR rates in all arms (42% monotherapy vs. 28% dual therapy, P=0.46).

Conclusions: TRBV expression pattern can provide predictive information beyond known immune gene expression signatures. High expression of TRBV4.3, TRBV6.3 and TRBV7.2 variants is associated with higher pCR rate with dual HER2 targeted and paclitaxel neoadjuvant therapy.

Citation Format: Powles R, Redmond D, Sotiriou C, Loi S, Fumagalli D, Nuciforo P, Harbeck N, de Azambuja E, Sarp S, Di Cosimo S, Huober J, Baselga J, Piccart-Gebhart M, Elemento O, Hatzis C, Pusztai L. T-cell receptor beta chain variable region (TRBV) expression patterns predict response to combined trastuzumab/lapatinib treatment in the NeoALTTO/BIG-1-06 trial [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P2-09-01.

Abstract P1-05-01: Landscape of somatic mutations in inflammatory breast cancer whole-genome sequences

Goal: Inflammatory breast cancer (IBC) is a rare, aggressive form of breast cancer that is characterized by a highly metastatic phenotype. Numerous previous attempts failed to identify, recurrent, IBC-specific gene expression or DNA copy number alterations. We performed whole genome sequencing (WGS) of IBC biopsies obtained before any therapy to define a comprehensive genomic landscape of this disease.

Methods: Illumina paired-end whole genome sequencing (WGS) of 20 IBC (n=9 ER+, n=11 ER-) and matched normal samples were performed with median coverage of 60X and 40X for cancer and normal, the percentages of mapped reads were 99.3% and 99.2%, respectively. We identified germ-line and somatic variants, indels as well as large scale structural variants, using GATK Haplotype Caller, MuTect and CREST, respectively. We performed the same analysis on WGS data from 23, age, race and ER and HER2 matched, non-IBC (n=12 ER+, n=11 ER-) from the TCGA for comparison. Variants in both coding and noncoding sequences were categorized by FunSeq to identify potential drivers. Mutation clustering in each gene, as well as significantly mutated non-coding regulatory modules, were identified using LARVA. DeconstructSigs were used to decompose the mutational spectrum of each cancer into 30 validated, mutational signatures provided by COSMIC. Contributions of each validated signature to mutations in IBC vs. non-IBC were compared using Welch's t-test.

Results: We identified 118,818 somatic variants in the IBC samples (median: 3,856; minimum: 1,109; maximum: 24,815) including 1,060 variants (~0.9%) in coding regions. 5,287 somatic indels and 5,959 large scale structural variants were detected including 1,028 insertions and 1,857 deletions. Recurrent, non-synonymous mutations were detected in the coding region of GRIN2A gene in 3/20 IBC samples (15%), (previously reported as a potential driver mutation in 1.7% of breast cancers). Other significant mutations in coding regions included GRHL1, PIK3R2, ESR1, FLG2 and etc. Three DNase I hypersensitive sites (DHSs) in non-coding regions were altered in 20% (4/20) IBC samples vs. fewer than 8.7% (2/23) in non-IBC. Mutational frequency of GATA3 is 80.0% vs. 47.8% (p=0.03), and PTEN is 45.0% vs. 73.9% (p=0.05), in IBC vs. non-IBC samples when including both coding and non-coding variants. Contributions of mutational signature 9, that is associated with polymerase η , were significantly higher in IBC cohort than non-IBC cohort (p-value=0.056).

Conclusion: This is the first whole genome sequencing analysis of IBC and comparison with the results from non-IBC. We identified promising candidate drivers in the coding sequence and in non-coding regulatory modules of expressed genes. We also identified mutational signature 9, and mutations in several DHS as significantly more frequent alterations in IBC compared to non-IBC.

Citation Format: Li X, Krishnamurthy S, Kumar S, Reddy S, Woodward W, Reuben J, Hatzis C, Ueno NT, Gerstein M, Pusztai L. Landscape of somatic mutations in inflammatory breast cancer whole-genome sequences [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 P1-05-01.

Abstract P2-09-15: A multi-institutional, prospective study of incorporating the genomic platform breast cancer index as a tool for decision-making regarding extension of adjuvant endocrine therapy

Background: Extending adjuvant endocrine therapy (AET) for hormone responsive breast cancer (HRBC) from 5 to 10 years is beneficial for many in preventing late relapse. Current decision-making regarding extension relies on a decision-making process that weighs non-personalized recurrence risks against risks and benefits of extended AET. The Breast Cancer Index (BCI, BioTheranostics Inc) has been validated to quantify the risk of late recurrence and to predict likelihood of benefit from AET extension based on an individual's tumor genomic profile. The purpose of this study was to conduct a multi-institutional study to prospectively assess the impact of BCI i) on provider's recommendation using the BCI results; 2) the confidence with decision-making; and 3) patient's satisfaction regarding extension of AET.

Methods: Patients with stage I-III HRBC treated at Yale Cancer Center and University of Pittsburgh Medical Center (UPMC), who had completed at least 3.5 years of AET were eligible. BCI was performed on FFPE samples from the original tumor sample (bioTheranostics Inc.). Patients and physicians completed pre- and post-test questionnaires examining preferences for extending AET, patients also completed anxiety and decision-conflict surveys.

Results: 140 patients [mean age 61, 80% postmenopausal, 73% stage I] were included. No extended AET was recommended for 35.3% patients' pre-testing. Reasons physicians did not recommend extended AET were perceived low risk of recurrence (87%), risk of osteoporosis (25%) and side effects (13%). Extended therapy was recommended for 65.7% patients pre-testing. Integration of BCI resulted in a change in physician treatment recommendation in 29% of patients. The recommendation for no extended AET rose to 48% and recommendation for extended AET dropped to 52% (OR=1.76 95% CI 1.08-2.85; p=.003). Of the recommendations that changed (N=41), the majority (73%) was for not extending endocrine therapy. However, 27% of recommendations were to extend endocrine therapy because of high risk or high likelihood of benefit results. More physicians felt strongly confident in their recommendation after the test result (26.4%) than before (9.3%) (OR= 3.5 95% CI 1.77-6.95; p&lt;.0001). Satisfaction of decision increased in 23% of patients (OR=2.72 95% CI 1.66-4.46; p&lt;.0001). Patient reported concerns including the cost, safety and benefit of extended AET decreased from pre- to post-testing (p=.025; p&lt;.0001; p=.0012 respectively)

Conclusions: Overall, incorporation of BCI into clinical practice resulted in significant changes in physician recommendations regarding AET duration, with the majority of recommendations for no extended AET. Physicians reported increased confidence for their recommendation when incorporating the test result. There was also a significant increase in patient satisfaction and decrease in patient reported concerns regarding cost, safety and benefit of extended AET. The BCI is a tool that could be incorporated into decision-making algorithms to enhance physician confidence and patient satisfaction with recommendations for extending AET.

Citation Format: Sanft T, Berkowitz A, Schroeder B, Hatzis C, Schnabel C, Aktas B, Brufsky A, Pusztai L, vanLonden GJ. A multi-institutional, prospective study of incorporating the genomic platform breast cancer index as a tool for decision-making regarding extension of adjuvant endocrine therapy [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 P2-09-15.

Abstract P6-09-49: Three-gene signature predictive of high residual risk of recurrence after adjuvant chemotherapy in ER-positive HER2-negative breast cancer

Background: A proportion of ER-positive, HER2-negative patients treated with both adjuvant endocrine and chemotherapy continue to remain high risk for recurrence. These individuals are the ideal patient population for future adjuvant trials.

Methods: We used gene expression data from ER+/HER2- patients who received adjuvant endocrine and chemotherapy. We utilized a training cohort of multiple gene expression datasets (N=131; GSE16391, GSE21653, GSE17907, GSE19615, GSE16716, GSE45255) and a validation cohort (N=277; GSE25066) from similarly treated ER+ patients. All data sets were generated with Affymetrix HGU133 arrays. Cox proportional hazards regression was performed across all genes using all possible cutoff values between the lower and upper quartile. Genes were ranked according to the estimated hazard rates for relapse. The top three genes were combined into a final signature that used the mean expression of these genes as the predictive score. Statistical analyses were performed in the R environment.

Results: The final signature capable to predict survival consisted of three genes, MX1, MBD4, and ZWINT. Each of these genes achieved high significance in a univariate analysis: MX1 HR=5.7 (p=1.8E-04) and HR=4.7 (p=1.3E-03), MBD4 HR=3.4 (p=2.8E-04) and HR=4.3 (p=4.4E-04), and ZWINT HR=4.4 (p=1E-03) and HR=3.7 (p=8.5E-03) in the training and validation cohorts, respectively. Using the mean expression of the three genes, the estimated hazard rates for survival were 20.5 (p=2.5E-05) and 17.5 (p=1E-04) in the two cohorts, respectively. Only one patient relapsed in each cohort. In a multivariate analysis including grade and lymph node status as covariates (size was not available), the three-gene signature retained significance (HR=13.2, p=0.011; and HR=13.9, p=0.01 in the two cohorts, respectively), while only lymph node status (HR=3.6, p=0.007) was significant in the validation cohort and no clinical variable was significant in the training cohort.

Conclusions: High expression of a 3-gene signature can identify ER+ patients who remain at very high risk for recurrence despite current adjuvant endocrine and chemotherapy.

Citation Format: Györffy B, Ocana A, Herman P, Hatzis C, Pandiella A, Pusztai L. Three-gene signature predictive of high residual risk of recurrence after adjuvant chemotherapy in ER-positive HER2-negative breast cancer [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 P6-09-49.

Abstract S1-07: Immune sculpting of the triple negative breast cancer genome

Background: Tumors with infiltrating lymphocytes (TIL) demonstrate a better prognosis particularly in TNBC and HER2 positive breast cancer. Two competing hypothesis predict contrasting relationships of TILs and genomic heterogeneity. On one hand, a strong immune response may lead to “pruning” of intratumor heterogeneity by eliminating immunogenic clones resulting in a near equilibrium, hence better prognosis, while cancers that escape the surveillance may evolve towards greater clonal heterogeneity and genomic complexity. In some cancers, the predicted neoantigens are less frequent than expected by chance also suggesting immune mediated elimination of neoplastic clones (Rooney et al. 2015). Studies also showed an inverse association between immune cell infiltration and intratumor clonal heterogeneity (Morris et al. 2016). On the other hand, cancers with greater genomic instability and mutational burden will have larger clonal heterogeneity and therefore more neoantigens and greater immune infiltration. Indeed, a positive correlation between overall mutation load and immune activity in the tumor microenvironment was observed in pooled data across a broad range of cancer types (Brown et al. 2014, Rooney et al. 2015, Schumacher and Schreiber 2015).

Methods: We assessed these two competing hypothesis and examined the relationship between genomic complexity and immune gene expression in different breast cancer subtypes. We used previously described immune metagene expression (DNA microarray n=655) as measures of immune infiltration in the TCGA data set (RNA-Seq n=1215). We compared somatic mutations, mutation count, neoantigen load, clonal heterogeneity metrics and the distribution of mutations in 119 canonical cancer genes and 12 cancer pathways between good and poor prognosis TNBC (n=208) corresponding to high and low immune infiltration.

Results:A positive but weak correlation between mutation count and immune metagene expression was observed when all breast cancer subtypes were analyzed together (P=0.08). This was driven by the generally higher mutation count and immune infiltration in TNBC. When TNBC was analyzed separately, good prognosis TNBC with high immune infiltration had lower total mutation count (P=0.021) and predicted neo-antigen count (P=0.035). Clonal heterogeneity was also lower in good prognosis TNBC (P=0.001). There was a strong inverse relationship of dispersion in mutation variant allele frequencies and immune metagene expression. CASP8 was the top enriched mutation in TNBC with high immune infiltration (P=0.007 with no adjustment for multiple testing).

Conclusions:High immune infiltration is associated with reduced intratumor heterogeneity in TNBC suggesting immune sculpting of the tumor and a near equilibrium between the cancer and immune surveillance. Surgical resection of the primary tumor may tilt the balance towards the immune system resulting in the better prognosis of high-TIL TNBC. TNBC with low immune infiltration has greater clonal heterogeneity and mutation load and may represent the consequence of escape from immune surveillance. Mutation of CASP8 may be one way to evade tumor cell killing in high-TIL TNBC as previously noted.

Citation Format: Karn T, Jiang T, Hatzis C, Sänger N, El-Balat A, Holtrich U, Becker S, Bianchini G, Pusztai L. Immune sculpting of the triple negative breast cancer genome [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 S1-07.

Abstract P6-09-23: SETER/PR - A robust 18-gene predictor of sensitivity to endocrine therapy in metastatic breast cancer

Rationale: A robust index for gene expression related to activity of estrogen (ESR1) and progesterone (PGR) receptors could predict sensitivity to endocrine therapy in metastatic breast cancer.

Methods: Transcripts correlated with ESR1 and PGR expression in 389 hormone receptor-positive breast cancer samples (Affymetrix U133A microarrays) were ranked for reliability according to their pre-analytical (intratumoral heterogeneity, biopsy type) and analytical reproducibility. Eighteen target and ten reference genes were selected and summarized as the SETER/PR index. The SETER/PR index was evaluated in a different set of 140 biopsies from distant metastases of hormone receptor-positive and HER2-negative (HR+/HER2-) breast cancer, and in additional pre-analytical and analytical sample cohorts. Thereafter, SETER/PR was translated to a customized format for application to formalin-fixed and paraffin-embedded (FFPE) sections.

Results: Higher SETER/PR in a metastasis was associated with longer progression-free survival (PFS, 9 vs. 2 months) and overall survival (OS, 50 vs. 19 months) following endocrine therapy in the cohort with metastatic breast cancer (MBC) and relapsed disease (n=79), so a cut point was defined in that cohort. SETER/PR was also significantly associated with PFS after adjusting for PR status of the metastasis, presence of visceral metastases, number of previous relapse events, and clinical history of previous sensitivity to endocrine therapy (HR 0.485, 95%CI 0.265 – 0.889, p = 0.019). Technically, SETER/PR was highly reproducible under different pre-analytical and analytical conditions, including host organ contamination. The translated SETER/PR assay used a single 10 µm FFPE tissue section, did not require RNA purification, and represented the microarray results from matched fresh samples with excellent agreement (correlation = 0.980, n = 31).

Conclusion: The SETER/PR index is a new biomarker to predict PFS and OS for patients with HR+/HER2- MBC who receive endocrine therapy. The assay is applicable to FFPE tissue sections from small biopsies of metastases. Additional independent (blinded) validation studies will be necessary to confirm these results.Rationale: A robust index for gene expression related to activity of estrogen (ESR1) and progesterone (PGR) receptors could predict sensitivity to endocrine therapy in metastatic breast cancer.

Methods: Transcripts correlated with ESR1 and PGR expression in 389 hormone receptor-positive breast cancer samples (Affymetrix U133A microarrays) were ranked for reliability according to their pre-analytical (intratumoral heterogeneity, biopsy type) and analytical reproducibility. Eighteen target and ten reference genes were selected and summarized as the SETER/PR index. The SETER/PR index was evaluated in a different set of 140 biopsies from distant metastases of hormone receptor-positive and HER2-negative (HR+/HER2-) breast cancer, and in additional pre-analytical and analytical sample cohorts. Thereafter, SETER/PR was translated to a customized format for application to formalin-fixed and paraffin-embedded (FFPE) sections.

Results: Higher SETER/PR in a metastasis was associated with longer progression-free survival (PFS, 9 vs. 2 months) and overall survival (OS, 50 vs. 19 months) following endocrine therapy in the cohort with metastatic breast cancer (MBC) and relapsed disease (n=79), so a cut point was defined in that cohort. SETER/PR was also significantly associated with PFS after adjusting for PR status of the metastasis, presence of visceral metastases, number of previous relapse events, and clinical history of previous sensitivity to endocrine therapy (HR 0.485, 95%CI 0.265 – 0.889, p = 0.019). Technically, SETER/PR was highly reproducible under different pre-analytical and analytical conditions, including host organ contamination. The translated SETER/PR assay used a single 10 µm FFPE tissue section, did not require RNA purification, and represented the microarray results from matched fresh samples with excellent agreement (correlation = 0.980, n = 31).

Conclusion: The SETER/PR index is a new biomarker to predict PFS and OS for patients with HR+/HER2- MBC who receive endocrine therapy. The assay is applicable to FFPE tissue sections from small biopsies of metastases. Additional independent (blinded) validation studies will be necessary to confirm these results.

Citation Format: Sinn BV, Tsai T-H, Lau R, Fu C, Gould R, Murthy R, King TA, Hatzis C, Kwiatkowski DN, Valero V, Symmans WF. SETER/PR - A robust 18-gene predictor of sensitivity to endocrine therapy in metastatic breast cancer [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 P6-09-23.

Abstract S5-01: Whole exome sequencing of pre-treatment biopsies from the neoALTTO trial to identify DNA aberrations associated with response to HER2-targeted therapies

Background: We examined if alterations in nucleic acid variants, genes, pathways, and overall mutational load and clonal entropy are associated with pathologic complete response (pCR) and survival after neoadjuvant anti-HER2 therapies in the NeoALTTO trial.

Methods: Whole exome sequencing was performed of 203 baseline biopsies with outcome information. The mean nucleotide coverage was 150x with &gt;90% of target bases showing &gt; 30x coverage in &gt; 99% of samples. Somatic mutations were called by MuTect and indels by Strelka, using pooled reference normal DNA. Significantly mutated genes (FDR&lt;10%) were identified by MutSigCV. Mutations in 714 canonical biological pathways were assessed and mutational load and genome clonal entropy (MATH) were calculated. Association with pCR and survival were evaluated by logistic regression adjusted for ER status and Cox-proportional hazards regression.

Results: Only 12 genes had mutation rates significantly above background and among these only PI3KCA was associated with lower pCR rate (OR=0.42, p=0.019). Genes with somatic mutations in more than 10 patients were also assessed, but none were associated with pCR or survival. Clonal entropy or adjusted mutation load also did not correlate with response. Mutations in 33 pathways showed significant association with response in the entire cohort. In the trastuzumab arm, 23 of the 33 pathways showed an association with response but none was independent of PIK3CA mutation. We constructed "PIK3CA-gene network" that included all unique genes (n=439) from theese 23 pathways. Of the 66 patients in the trastuzumab arm, 50 carried at least one mutation in one of the 439 genes and among these only 2 achieved pCR (4%) compared to 9 of 16 pCR (56%) among the wild type (OR=0.035; p &lt; 0.001). The same genes/mutations had little impact on pCR in the lapatinib arm (pCR 20%). In the lapatinib arm, mutations in 3 pathways conferred higher probability of pCR. The "Regulation of RhoA activity" pathway, had the most significant association with pCR in the entire cohort (OR=3.77, p=0.0009) and in the lapatininb (pCR 67% vs 17%, OR=14.8, p=0.008) and lapatinib + trastuzumab (OR=3.0, p=0.06) arms, but not in the trastuzumab arm (OR=1.4, p=0.7). Event free and overall survival were also significantly higher in patients who had mutations in this pathway. Twenty seven of the 48 genes in this pathway had mutations affecting 33 patients but different genes were affected in different individuals.

Conclusions: There are no high frequency recurrent single mutations associated with response to HER2-targeted therapies, other than PIK3CA. We identified several biological pathways, including RhoA activity, and a network of PIK3CA associated genes that are significantly associated with response when affected by mutations, however, different genes are mutated in different individuals.

Citation Format: Pusztai L, Shi W, Jiang T, Nuciforo P, Holmes E, Harbeck N, Sotiriou C, Rimm D, Hatzis C, de la Peña L, Armour A, Piccart-Gebhart M, Baselga J. Whole exome sequencing of pre-treatment biopsies from the neoALTTO trial to identify DNA aberrations associated with response to HER2-targeted therapies. [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 S5-01.

Abstract P6-13-06: Novel combination therapies for triple negative breast cancer identified by high-throughput screening

Background: Finding effective novel therapies for triple negative breast cancer (TNBC) remains an unmet challenge. Our objective was to assess the sensitivity of TNBC to single drugs and to combinations using a systematic screen of existing FDA approved drugs combined with other approved or experimental agents and to characterize the mechanism of action of the most promising combinations.

Methods: We conducted a comprehensive high-throughput drug combination screen in six TNBC cell lines with different genetic backgrounds. We tested a panel of 128 agents with known targets for their growth inhibitory potential individually and in pairwise combinations with six FDA approved anticancer drugs in MDA-MB-231, MDA-MB-436, MDA-MB-468, BT-20, BT-549, and HCC-38 TNBC cell lines. High-throughput assays were performed in 384-well plates and dose response curves were generated using 5 concentrations of the secondary drug, while keeping the first drug at fixed dose. Cell viability was measured 72h after treatment exposure using CellTiter-Glo®. Drug combinations were assessed for overall inhibitory effect and also for superadditivity, assessed as deviations from non-interaction using Bliss model of synergy. Selected synergistic and effective drug combinations identified from the high-throughput screen were subsequently validated in a 96-well low-throughput format in all TNBC cell lines.

Results: Cell cycle and apoptosis regulators were more inhibitory as single agents across TNBC cell lines relative to other drug classes. Bortezomib, carfilzomib, YM155, Flavopiridol, KP372-1, and dactinomycin were highly toxic to all TNBC lines as single agents. Combinations with either everolimus or erlotinib were particularly effective in BT-20 and MDA-MB-468, and combinations with crizotinib in MDA-MB-231 cells. Bay-11-7082/erlotinib and MK-1775/everolimus were few of the promising combinations that elicited potentiated response in all the cell lines. ABT-263/crizotinib was one of the top synergistic combination most effective in MDA-MB-231 cells that express highest relative mRNA and protein levels of respective drug targets, Bcl-xL, and AXL. The combination treatment with ABT-263 and crizotinib also resulted in apoptosis in MDA-MB-231 cells, indicated by marked PARP cleavage in these cells, while MDA-MB-436 cells with lowest expression of Bcl-xL were resistant to apoptosis or growth inhibition, indicating a potential efficacy of this combination in a subset of TNBCs.

Conclusions: This study reveals novel combinations of cell cycle or apoptosis regulators and crizotinib, everolimus, or erlotinib with enhanced anticancer activity. Combinations of Bay-11-7082 and erlotinib; and MK-1775 and everolimus had mild synergistic activity in all TNBC lines, while ABT-263 and Crizotinib showed large synergistic antiproliferative activity in a subset of TNBCs. These promising drug combinations have great potential to improve cure rates in TNBC patients. Since crizotinib, everolimus, and erlotinib are US FDA-approved while ABT-263 and MK-1775 are in advanced clinical trials, there is a rapid path for clinical translation for these drug-combinations after validation in animal studies.

Citation Format: Wali VB, Langdon CG, Held MA, Platt JT, Safonov A, Aktas B, Stern DF, Pusztai L, Hatzis C. Novel combination therapies for triple negative breast cancer identified by high-throughput screening. [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 P6-13-06.

Abstract P4-04-20: Subtype specific differential expression and immunogenicity of endogenous retrovirus elements in breast cancer

Background: Endogenous retroviruses (ERVs) are germline encoded DNA sequences that entered the human genome millions of years ago. While they are mostly inactivated due to accumulated termination codons and deletions, previous studies have demonstrated overexpression and antibody-targeted immunotherapeutic potential, of ERV-related env proteins in breast cancer. We sought to elucidate subtype specificity, immunogenicity, and correlation with innate immunity related gene signatures of ERVs in breast cancer.

Methods: We utilized publically available RNASeq gene expression data of breast cancer samples along and corresponding matched normals from TCGA. The dataset included 191 ER-/HER2- (TN), 197 HER2+, and 627 ER+/HER2- (Luminal) breast cancers. ERV expression was obtained by mapping bowtie2-aligned reads of recently annotated to be transcriptionally active to the RNAseq bam files (Rooney et al 2015, Mayer et al 2011). ERVs preferentially expressed in tumors compared normal tissue were identified as those for which the 5th percentile of ERV expression in the tumors exceeded the 95th percentile of ERV expression in the normal samples. A gene signature involving GZMB, PRF1, CXCL13, IRF1, IKZF1, and HLA-E was used as a measure of immune activity. To assess the immunogenic potential of the tumor-specific ERVs, we compared the expression level of the ERV within the lower and higher immune signature tertiles using the Wilcoxon signed-rank test. To elucidate mechanism of potential immune response, ERVs found to be significantly associated with immune response at a false discovery rate of &lt; 10% were further analyzed for association with specific toll-like receptor (TLR) gene expression.

Results: Out of the 66 original annotated ERVs, 47 were found to be expressed at significantly higher levels in breast cancer compared to normal tissue and 22 were immunogenic. Examples include members of the ERV-K family, as have also been previously detected by flow cytometry and IHC. Subtype-specific immunogenic potential was demonstrated in 4 ERVs in TNBC (ERVK10, ERVK17, ERVFRD.1, ERVPABLB.1) and in 7 ERVs in the luminal subtype (ERV3.1, ERVE.4, ERVFRD.2, ERVK.15, ERVK.19, ERVK.20, ERVK.25, ERVW.3). Twelve of the 22 immunogenic ERVs were significantly correlated with expression of all ten TLR evaluated, while four ERVs showed more specific correlation patterns with TLRs. High ERV3.1 expression was associated with high TLR3, TLR8, TLR9 that specifically target double stranded or single stranded RNA, suggesting a potential mechanism for mediation of ERV related immune response.

Conclusion: Our results suggest breast cancer subtype specific ERV dysregulation and immunogenicity. The potentially immunogenic ERVs were generally not self-correlated or located in the same amplicon as HLA genes, suggesting an independent immune response pathway. Furthermore, ERV expression correlates with specific endosomal nucleic-acid recognizing toll-like receptors, which may prompt further investigation into subtype-specific TLR-targeted therapy.

Citation Format: Safonov A, Bianchini G, Jiang T, Pusztai L, Hatzis C. Subtype specific differential expression and immunogenicity of endogenous retrovirus elements 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 P4-04-20.

Abstract P6-11-03: A cost effectiveness analysis of baseline left ventricular function assessment for breast cancer patients undergoing anthracycline chemotherapy

Background: It is unclear if all breast cancer (BC) patients require baseline left ventricular function (LVEF) assessment prior to anthracycline based chemotherapy (ABC), and the approach is variable in clinical practice. Our objective is to determine the cost effectiveness of obtaining a baseline LVEF assessment prior to (neo) adjuvant ABC in clinical practice.

Methods: We performed a retrospective analysis of the Yale Equilibrium Radionuclide Angiography (ERNA) database for 701 breast cancer patients who had a baseline ERNA scan prior to systemic therapy for an initial diagnosis of stages I-IV BC between July 2003 and May 2013. We found that 14 of 701 (2%) patients had a baseline LVEF &lt; 50%. Age, pre-existing cardiac risk factors and coronary artery disease did not predict an abnormal baseline LVEF &lt;50 %. To evaluate the benefit of obtaining a baseline echocardiogram or ERNA before ABC, we considered the screening scenario in which BC patients with a baseline LVEF &lt; 50% on screening echocardiogram are treated with a second (2nd) generation non-ABC and those with baseline LVEF ≥ 50% receive a third (3rd) generation ABC, and compared this with a non-screening scenario with uniform 3rd generation ABC treatment for all patients who do not have a baseline echocardiogram. We used Adjuvant Online to obtain estimates of the disease free (DFS) and overall survival (OS) for a 3rd generation ABC regimen vs a 2nd generation non-ABC regimen for 50 year old patients with a T2N1 hormone receptor positive BC. We implemented these oncologic clinical outcomes (in addition to cardiotoxicity-related clinical outcomes, costs of screening echocardiogram and treatment of congestive heart failure (CHF), quality of life metrics, as reported in the literature) into a simplified decision-analytic cost-effectiveness analysis that accounts for the different disease states and their associated costs and quality of life outcomes.

Results: Assuming that 20% of the unscreened patients with a LVEF &lt; 50% will develop CHF if treated with ABC regimen without management of baseline cardiac dysfunction, the base case incremental cost effectiveness ratio (ICER) was determined to be 18,520 $USD/QALY. Sensitivity analysis suggested that the cost-effectiveness of baseline LVEF assessment is primarily driven by the prevalence of patients with LVEF &lt; 50%, the incidence of CHF in this high-risk patient group if treated with ABC regimen, and time to CHF development. While our analysis did not reveal risk factors predictive of low baseline LVEF, our model's dependence on prevalence of LVEF &lt; 50% demonstrates the importance of risk factor stratification. A hypothetical predictive marker which enriches the prevalence of an abnormal baseline LVEF 5-fold to 10% would result in a cost-effectiveness of 10,990 $USD/QALY. The model is less sensitive to the cost of baseline echocardiogram testing.

Conclusion: Baseline LVEF assessment was found to be cost-effective under a willingness-to-pay threshold of $50,000/QALY. Our sensitivity analysis suggests that risk factor-guided LVEF baseline LVEF screening may increase the number of high-risk patients in the treatment population, thus further increasing the cost-effectiveness of baseline LVEF assessment.

Citation Format: Safonov A, Hatzis C, Stratton J, Gross CP, Russell R, Pusztai L, Abu-Khalaf MM. A cost effectiveness analysis of baseline left ventricular function assessment for breast cancer patients undergoing anthracycline 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 P6-11-03.

Abstract P5-07-09: Heterogeneity of tumor infiltrating lymphocytes in breast cancer and its impact for use as a biomarker

Background: In breast cancer, elevated tumor infiltrating lymphocytes (TILs) is associated with PD-L1 expression, hormone receptors negativity, and better outcome. The presence of numerous CD8+ cytotoxic T cells in pre-treatment specimens is associated with clinical benefit from PD-1 axis blockade in melanoma and lung cancer, suggesting its predictive value. Despite recent efforts to standardize the pathologist evaluation of TILs in breast cancer, objective determination of lymphocyte subpopulations and their distribution/uniformity within tumor tissues remains largely unexplored. Here, we simultaneously measured diverse TIL subpopulations using quantitative immunofluorescence (QIF) in different areas of breast tumors to determine the heterogeneity of TILs and its possible impact for use as biomarker.

Methods: Using a multiplexed QIF-based assay for simultaneous detection of DAPI (all cells), Cytokeratin (epithelial cells, M3515-DAKO), CD3 (T lymphocytes, E272--Novus), CD8 (cytotoxic T cells, C8/144B--DAKO), and CD20 (B cells, clone L26-DAKO), we measured the levels of TIL subpopulations in whole tissue section slides of 3 tumor cores obtained from different areas of 31 breast carcinomas. The levels of the markers were measured using the AQUA method of QIF and the heterogeneity was studied using numerical correlations of log2 transformed scores and variance component analysis with linear mixed effects (LME). The concordance (kappa index [κ]) between binarized scores obtained measuring 1 vs 3 cores of the same tumor was also evaluated.

Results: As expected, we found a positive correlation between CD3 and CD8 levels across all patients (Pearson correlation coefficient [CC]=0.827). The levels of CD3 and CD8 showed weaker association with CD20 signal (CC=0.446 and 0.363, respectively). For all the TIL markers, the intra-tumor variation was higher than the inter-tumor differences with intraclass correlation coefficients (ICC) of 0.411 for CD3, 0.324 for CD8, and 0.252 for CD20. In the variance component analysis, 66-69% of the variance was attributable to signal differences between areas of the same tumor core and 30-33% was due to differences between cores from different areas. Consistent with this and using the median score as cutpoint to stratify cases in high/low marker levels, the concordance of measuring TILs in 1 vs 3 cores of the same tumor was κ=0.705 for CD3, κ=0.655 for CD8, and κ=0.603 for CD20.

Conclusion: Objective measurement of TIL markers indicates that T and B lymphocytes show heterogeneity in breast cancer. The tumor variation of the markers is driven predominantly by differences within the same tumor core. The data from our study suggests that although a single core biopsy of tumors provides considerable information regarding the degree of lymphocyte infiltration in breast cancer patients, caution should be taken when using this as a clinical biomarker.

Citation Format: Mani NL, Schalper K, Hatzis C, Chagpar A, Pusztai L, Rimm DL. Heterogeneity of tumor infiltrating lymphocytes in breast cancer and its impact for use as a biomarker. [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 P5-07-09.

Abstract P4-04-04: Triple-negative (TN) and HER2+ breast cancers (BC) have different immune milieu in primary and metastatic tumors

Background: In TN and HER2+ early BCs, a high immune infiltration is linked to good prognosis and improved treatment benefit. Little is known about the characteristics of the immune milieu of BCs in metastatic disease. We aimed to investigate differences of the immune microenvironment between cohorts of primary early breast cancer (EBC) and metastatic (MBC) tumors according to molecular subtypes.

Methods: We identified publicly available gene expression profiles (GEPs) of MBCs profiled either on Agilent (n=12, cohort I) or AffymetrixU133A (n=36, cohort II). These included 21 ER-/HER2- (TN), 10 HER2+ and 17 ER+/HER2- (Luminal). From GEPs of EBC profiled on the same platforms, we randomly selected two cohorts with the same molecular subtype composition (n=65 and n=230) and compared them with MBCs. We assessed differential expression of 40 pre-selected immune genes belonging to six immune-related metagenes [CD8, IGG and MHC2, related to T cells, plasma cells and antigen presenting cells respectively; MHC1, STAT1 and IF.I related to HLA class I genes; and genes modulated by interferon (Gianni L SABCS 2012)]. We also evaluated β2-microglobulin (B2M), for its role in the MHC1 complex, and an immune signature associated with benefit from pembrolizumab in melanoma (Ribas A ASCO 2015).

Results: In cohort I (Agilent), only 33 genes were annotated. Overall, 16/33 (48.4%) genes had a significantly lower expression in MBC (p&lt;0.05). In TN and HER2+ MBCs 18 and 11 genes were significantly lower than in EBC, respectively (p&lt;0.05) (6 in both), while only one was lower in luminal MBCs (IGHM). In cohort II (Affymetrix), 26/40 genes (65%) had lower expression in MBC (p&lt;0.05). Considering molecular subtypes, 25 and 19 genes were lower in TN and HER2+, respectively (17 in both), and only one in ER+/HER2- (IL7R). In ER+/HER2- one gene was higher in MBC (IFIT2). In TN and HER2+ the genes with lower expression in MBC belong to all immune functional categories, in particular MHC1 (HLA-A, B and C), STAT1 (STAT1, CXCL10, CXCL11, GBP1), MHC2 (HLA-DQB1 and DRB4) and T cells (CD52, IL7R and TRBC1). B2M was significantly lower in all MBC patients, and in HER2+ and TN groups both in cohort I (p=0.0002; p=0.006 and p=0.0005, respectively) and in cohort II (p&lt;1E10-6; p=0.0008 and p=0.00004, respectively), while it was modestly lower in ER+/HER2- in cohort II only (p=0.027). The signature associated with benefit from pembrolizumab in melanoma was significantly lower in TN and HER2+ MBC in both cohort I (p=0.003) and cohort II (p=0.001), but not in luminal cases.

Conclusions: TN and HER2+ MBCs have a "colder" immune microenvironment than primary tumors, with significantly lower expression of genes related to immune response and to antigen presentation (B2M and MHC1). This is consistent with the lower TILs we have described in a small series of paired EBC-MBC (Ogiya R ASCO 2015), suggesting the engagement of mechanisms of immune escape during the metastatic process. However, the "cold" immune milieu observed in MBC could also result from selection of low immunogenic tumors more likely to relapse. Our findings suggest that use of immune-checkpoints inhibitors in MBCs may require the combination with agents able to turn on an immunogenic response.

Citation Format: Bianchini G, Riba M, Zambelli S, Safonov A, Ogiya R, Jiang T, Hatzis C, Niikura N, Zambetti M, Iwamoto T, Pusztai L, Gianni L. Triple-negative (TN) and HER2+ breast cancers (BC) have different immune milieu in primary and metastatic tumors. [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 P4-04-04.

Abstract P4-07-01: DNA repair deficiency enhances immune response and correlates with excellent clinical outcome in triple negative breast cancer

Background: Mutations or epigenetic silencing of BRCA1/2 genes result in DNA repair deficiency in a large proportion of TNBC cases. Yet it is unclear whether this deficiency is associated with increased chemosensitivity and improved benefit from standard-of-care chemotherapy. We systematically evaluated BRCA deficiency in TNBC using integrated DNA and RNA sequencing data and its association clinical outcome, exploring the potential role of tumor immune response.

Patients and Methods: Whole-exome, DNA methylation, copy number variation, and RNA sequencing data from 102 stage I-III TNBCs were retrieved from The Cancer Genome Atlas (TCGA). Almost all patients received adjuvant taxane-anthracycline-cyclophosphamide (T-AC) chemotherapy and had &gt;30 days of follow-up. The number of predicted neoantigens and an estimate of the level of immune cell activity for 77 of these tumors were previously published. Deleterious germline or somatic BRCA1/2 mutations were identified by majority voting on predictions of 5 variant scoring algorithms. Definition of BRCA1/2 deficiency (BRCA-D) included carrier of deleterious BRCA1/2 mutations or BRACA1/2 normal (BRCA-N) with wild type BRCA1/2 expression less than the maximum observed in mutation carriers. Normalized genomic mutation rate and mutant allele tumor heterogeneity (MATH) were computed as broad measures of genomic instability. Characteristics of BRCA-D vs N tumors were compared using the Wilcoxon rank test.

Results: Twenty tumors (19.6%) had mutations in BRCA1, 6 (5.8%) in BRCA2, and 2 (1.9%) in both. Based on the expanded definition, 39 cases (38%) were characterized as BRCA1 deficient, 5 (4.9%) as BRCA2 deficient and 4 (3.9%) as deficient in both. BRCA-D tumors (47%) were associated with a significantly higher mutation rate (P=8x10-4) but had similar clonal heterogeneity (P=0.55) as BRCA-N tumors. BRCA-D tumors had excellent 4-year overall survival (100%) compared to 79.5% (95%CI: 66.6- 94.9) for BRCA-N tumors (log-rank P=0.02). BRCA-D tumors also presented a significantly higher number of predicted neoantigents (P=0.003), which resulted in increased level of immune cell activity. In contrast, low immunogenic TNBC tumors were underrepresented in BRCA-D (p=0.05) and showed potential signs of immunoediting (observed/expected number of predicted neoantigens &lt; 1; p=0.07).

Conclusions: Deleterious mutations in BRCA1/2 genes occur in 25% of TNBC tumors, but parallel quantification of wild-type BRCA1/2 expression identifies 47% of TNBC samples with double strand break DNA repair deficiency. These BRCA-D TNBC tumors are characterized by a significantly higher mutation rate and present a significantly greater number of neoantigens that result in increased immune cell activity. Our analysis suggested that enhanced immune activation could explain to a large extent the excellent clinical outcome in patients with BRCA-D tumors treated with standard-of-care T-AC chemotherapy.

Citation Format: Jiang T, Safonov A, Bianchini G, Shi W, Wali VB, Pusztai L, Hatzis C. DNA repair deficiency enhances immune response and correlates with excellent clinical outcome in 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 P4-07-01.

Abstract A32: MCL1 gene amplification in breast cancer is associated with TNBC status and can respond to a sorafenib/vorinostat regimen

Background: MCL1 encodes the induced myeloid leukemia cell differentiation protein Mcl-1, a member of the BCL-2 family which functions to inhibit apoptosis. Mcl-1 over-expression has been associated with high tumor grade and adverse prognosis in triple negative breast cancer (TNBC) but therapies specifically leading to inhibition of MCL-1 have not been identified.

Methods: Comprehensive genomic profiling (CGP) using hybridization capture of 3,769 exons from 315 cancer-related genes and 47 introns of 19 genes commonly rearranged in cancer was applied to ≥50ng of DNA extracted from 2,824 consecutive BC and sequenced to high, uniform median coverage (&gt;600X). The original primary BC was assayed in 44% of cases and a sample from a metastatic focus was assayed in 56% of cases.

Results: Of 2824 consecutive BC cases, 200 (7.1%) cases harbored MCL1 amplification. Of these MCL1-amplified cases, 146 (73%) were TNBC and 54 were non-TNBC (p&lt;0.0001). Twelve of the latter cases (22%) were ERBB2 (HER2) amplified and slide-based HER2 status concordance with CGP was 99%. MCL1 amplification was also observed in the TCGA dataset for 32/123 (26%) of TNBC (p=0.008). Of the MCL1 amplified TNBC cases, 88% were high grade and 98% were stage IV at the time of CGP. Genes co-altered within MCL1 amplified TNBC included TP53 (86%), MYC (41%), MYST3 (21%), LYN (20%), CCNE1 (19%), PIK3CA (18%), and AKT3 (15%). Two MCL1 amplified TNBC patients were treated with a multi-drug regimen based on sorafenib and vorinostat and experienced significant clinical benefit.

Conclusions: MCL1 amplification is a frequent feature in advanced stage and high grade TNBC, and correspondingly such MCL1 amplified tumors very seldom harbor co-amplifications of ERBB2. Clinical observation suggests that treatment with sorafenib and vorinostat in heavily pre-treated MCL1 amplified patients may be correlated with clinical benefit, consistent with historic preclinical investigation. These preliminary findings suggest that MCL1 amplified TNBC may be able to benefit from combination targeted therapy, and warrant further systematic investigation.

Citation Format: JS Ross, K Wang, A Johnson, J Watson, C Hatzis, L Pusztai, J Chmielecki, R Yelensky, D Lipson, JA Elvin, J Vergilio, J Suh, VA Miller, K Dicke, PJ Stephens, SM Ali. MCL1 gene amplification in breast cancer is associated with TNBC status and can respond to a sorafenib/vorinostat regimen. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Breast Cancer Research; Oct 17-20, 2015; Bellevue, WA. Philadelphia (PA): AACR; Mol Cancer Res 2016;14(2_Suppl):Abstract nr A32.

Efficacy and tolerability of combination cisplatin and ifosfamide chemotherapy with vaginal cuff brachytherapy in the first line treatment of uterine carcinosarcoma

PURPOSE OF INVESTIGATION

A retrospective study to evaluate six cycles of cisplatin 40 mg/m2 on day 1 and ifosfamide 1,200 mg/m2 daily on days 1 to 4 with Mesna every four weeks as first line treatment for 29 patients with a diagnosis of uterine carcinosarcoma.

MATERIALS AND METHODS

A total of 23 of 29 patients received high dose rate intracavitary vaginal cuff brachytherapy (VCBT) with two fractions of seven Gy each. Median age was 65 years (range 40-82); 13 (44.8%) had Stage I disease, three (10.3%) had Stage II, eight (27.6%) had Stage III, and five (17.2%) patients had Stage IV disease.

RESULTS

Most common toxicities were anemia grade 1 (35%)/grade 2 (45%), and neutropenia grade 3 (17%)/grade 4 (6.9%). Eleven dose modifications, four treatment discontinuations, and one patient withdrawal occurred. At a median follow up of 45 months (range 9 to 144), Progression free survival (PFS) was 20% and overall survival (OS) was 40% for Stage IV, PFS 75% and OS 62.5% for Stage III, compared to a PFS 75% and OS 72.2% for Stages I-II. Median OS for the entire group was 12.43 years (95% CI 3.69 to inf); for Stage I-III 12.4 years (6.1 to inf), and for Stage IV 15.6 months (95% CI 9.4 to inf).

CONCLUSIONS

Cisplatin and ifosfamide chemotherapy with VCBT was well tolerated and has promising activity in uterine carcinosarcoma.

Metabolic isoenzyme shifts in cancer as potential novel therapeutic targets

The functional redundancy of metabolic enzyme expression may present a new strategy for developing targeted therapies in cancer. To satisfy the increased metabolic demand required during neoplastic transformations and proliferation, cancer cells may rely on additional isoforms of a metabolic enzyme to satisfy the increased demand for metabolic precursors, which could subsequently render cancer cells more vulnerable to isoform-specific inhibitors. In this review, we provide a survey of common isoenzyme shifts that have been reported to be important in cancer metabolism and link those to metabolic pathways that currently have drugs in various stages of development. This phenomenon suggests a potentially new therapeutic strategy for the treatment of cancer by identifying shifts in the expression of metabolic isoenzymes between cancer and normal cells. We also delineate other putative metabolic isoenzymes that could be targets for novel targeted therapies for cancer. Changes in isoenzyme expression that occur during neoplastic transformations or in response to environmental pressure in cancer cells may result in isoenzyme diversity that may subsequently render cancer cells more vulnerable to isoform-specific inhibitors due to reliance on a single isoform to perform a vital enzymatic function.

Abstract P4-06-05: Transcriptional heterogeneity correlates with chemotherapy sensitivity in triple negative breast cancer and varies by molecular subtype

Purpose: Different molecular subtypes of breast cancer are believed to have different degrees of molecular heterogeneity. However, the heterogeneity of breast cancer subtypes has not been formally compared. The goal of this analysis was to develop metrics of global transcriptional heterogeneity, compare heterogeneity across breast cancer subtypes and assess whether heterogeneity metrics correlate with sensitivity to chemotherapy in basal-like cancers.

Methods: Affymetrix U133A gene expression data from 836 breast cancers (basal n = 219, HER2-like n = 147, luminal A n = 343, luminal B n = 127) were analyzed. Among the basal-like cancers, 128 received neoadjuvant chemotherapy (pathologic complete response, pCR, n = 44; residual disease, RD, n = 94). Molecular class was assigned by the PAM50 classifier. We used the average pairwise Pearson distance, the sum of eigenvalues of the correlation matrix and multi-dimensional scaling dispersions of the probe set expression values to quantify heterogeneity. Distribution of each metric within the different subtypes was estimated from 500 bootstrapped cohorts of 80 cases in each subtype. We used analysis of variance (ANOVA) to evaluate the significance of the differences in heterogeneity between subtypes.

Results: All three heterogeneity metrics yielded concordant results. Within subtypes, heterogeneity was greatest in basal-like cancers followed by HER2-like, luminal B and luminal A (see figure). Luminal A cancers were distinctly less heterogeneous than the rest. Basal-like cancers with RD were more heterogeneous than those with pCR. The best metric to separate basal tumors with pCR versus RD was the sum of eigenvalues shown below.

Conclusion: Breast cancer molecular subtypes differ in transcriptional heterogeneity. Luminal A cancers have the lowest within group heterogeneity while basal like cancers have the highest. Among the basal-like cancers, those with high chemotherapy sensitivity (pCR) are more homogeneous than those with RD suggesting that transcriptional heterogeneity might be contributing to increased resistance to therapy in this subgroup.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-06-05.

Abstract P6-11-03: Follow up of breast cancer patients who underwent gamma knife radiosurgery for a diagnosis of brain metastases: A single institutional experience

Background: Breast cancer (BC) is the second most common cause of brain metastases in the United States. We performed a retrospective analysis of BC patients (pts) in the Yale Gamma Knife Radiosurgery (GKRS) database treated for brain metastases (mets) to report overall survival (OS) from time of GKRS, and to identify factors that predict improved survival outcomes.

Methods: We performed retrospective review of the Yale database of BC pts treated with standard GKRS technique for brain mets from 2000 to May 2013. Treatment doses typically ranged from 16 to 24 Gy prescribed to the 40-85% isodose surfaces. Overall survival (OS) for each patient was calculated from the date of first GKRS treatment of brain metastases. Variables included age, Karnofsky Performance Status (KPS), pre/post whole-brain radiation treatment (WBRT) and/or surgical resection, and # of brain mets treated (1-4 vs ≥5).

Results: A total of 80 BC pts were identified, with a mean age of 56.2 yrs; KPS 40-60 in 8pts (10%), 70-80 in 13 (16%), 90-100 in 41 (51%), unknown in 18 (23%); ER+ in 38 (48%), ER- in 38 (48%), unknown in 4 (5%); HER2 + in 38 (48%), HER2- in 33 (41%), indeterminate in 3 (4%) and unknown in 6 (8%). BC subtypes are Luminal A (ER+/HER2-) in 16 (20%), Luminal B (ER+/HER2+) in 19 (24%), HER-2(ER-/HER2+) in 17 (21%), Basal (ER-/HER2-) in 17 (21%), unknown in 11 (14%). Number of brain mets at time of first GKRS ≥5 in 32 (40%); prior WBRT in 30 (38%); WBRT post GKRS in 9 (11%); surgical resection for brain mets in 13 (16%) prior to GKRS and post GKRS in 9 (11%); 16% had ≥3 GKRS. Median time from primary BC dx to brain mets is 62.8 months (m) (range 46.2-81.6); with no significant difference among subtypes. Median OS from GKRS is 13.1 m (7.6-21.9); median OS from brain mets is 18.8 m (11.7-23.9). On univariate analysis, factors associated with significant improved OS from time of GKRS included HER-2 subtype (p = 0.005), KPS 90-100 (p = 0.008), good control of systemic disease (p = 0.017), pre (p = 0.041) and post GKRS surgical resection (p = 0.02). Age, pre or post WBRT, # of brain mets (1-4 vs. 5) were not significant predictors of OS from time of GKRS. In a multivariate Cox analysis (46 cases), HER-2 subtype remained a significant predictor of improved OS from time of GKRS; (HR) 0.26, 95%CI 0.08 to 0.8, p-value = 0.02. There was a trend in favor of surgical resection of brain mets pre GKRS (HR 0.14, 95% CI 0.02 to 1.15, p-value = 0.067

Conclusion: GKRS offers good local control for BC pts with brain mets, and is a reasonable option for pts with ≥5 brain mets. Our results show no associated survival benefit for administration of WBRT pre/post GKRS. HER-2 subtype is a significant predictor of improved survival in BC pts undergoing GKRS for brain mets.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P6-11-03.

Abstract P2-10-17: SET index predicts response to endocrine therapy rather than prognosis independently of other genomic signatures in a blinded validation study

Background: A genomic index for sensitivity to endocrine therapy (SET) was previously derived from genes strongly positively and negatively coexpressed with estrogen receptor (Symmans et al. 2010, JCO 28:4111). This SET index has been reported to predict survival benefit from adjuvant endocrine therapy independently of prognosis.

Materials and Methods: Affymetrix gene expression profiling was performed on 307 ER positive primary breast cancers from a retrospective cohort treated solely with endocrine therapy. In a blinded study the SET index' ability to predict survival was analyzed in this previously unpublished dataset. Affymetrix profiles (CEL files) were anonymized and provided to the developers of SET index (Nuvera Biosciences) without any clinical information. The 261 profiles that passed QC were categorized into SET classes based on the published prespecified cutoffs. Samples categorized as ER negative based on gene expression were excluded. Subsequently classifications were unblinded and clinical associations analyzed according to a predefined protocol. Exploratory analyses were performed on the relationship of SET index to other genomic signatures. We also assessed a potential pure prognostic value of SET index in an additional dataset of 164 node negative ER positive patients that did not receive any adjuvant treatment.

Results: A lower SET index significantly correlated with higher grade (p = 0.009) and negative PgR status (p = 0.016). We detected no significant differences for age, tumor size, lymph node status, and HER2 status between patients with high, intermediate, and low SET index. In the lymph node negative (LNN) cohort (n = 120) we observed a significant difference in DFS (5yr DFS 77.1±10.2% vs 94.4±2.2%; HR 4.20, 95% CI 1.72–10.2; p = 0.002) and DMFS (HR 3.18, 95% CI 1.20–8.47; p = 0.020) for patients with low SET index. In contrast, we found no prognostic value of SET index in lymph node positive patients (n = 95). In multivariate analyses of LNN patients including SET, age, tumor size, histological grade, and PgR status, only SET was significantly associated with DFS (HR 3.37, 95% CI 1.25–9.01; p = 0.016) and DMFS (HR 3.03, 95% CI 1.08–8.55; p = 0.036). In exploratory analyses SET index was not correlated to other genomic signatures related to proliferation (as Recurrence Score, GGI, and NKI70) or immune response (e.g. 7IGS, SDPP). When we included all these genomic signatures as continous scores in a multivariate stepwise Cox regression model in the LNN endocrine treated cohort, only SET remained as significant (p = 0.025) while Recurrence Score displayed a strong trend (p = 0.054). We also verified that SET had no pure prognostic value in an additional dataset of 164 ER positive patients that did not receive any adjuvant treatment.

Conclusions: In a blinded analysis the predictive ability of SET index was prospectively validated in an independent cohort of node-negative patients. Our exploratory analysis demonstrates that SET index is unrelated to other genomic signatures and delivers independent information on the response of patients to endocrine therapy rather than prognosis.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P2-10-17.

P1-07-09: Estrogen Receptor (ER) mRNA and ER-Related Gene Expression in Breast Cancers That Are 1%-10% ER-Positive by Immunohistochemistry

Purpose: Our goal was to examine whether borderline estrogen receptor (ER)-positive cancers, defined as 1–10% positivity by immunohistochemistry (IHC), show the same global gene expression pattern and high ESR1 mRNA expression as ER-positive cancers or are more similar to ER-negative cancers.

Patients and methods: ER status was determined by IHC in 465 primary breast cancers and with Affymetrix U133A gene chip (ESR1 mRNA gene expression: Probe set = 205225_at). We compared expressions of ESR1 mRNA and a 106-probe set ER-associated gene signature score between ER-negative (n=183), 1–9% (n=25), exactly 10% (n=6), and &gt; 10% ER-positive (n=251) cancers. We also assessed the molecular class of the borderline ER-positive cases using the PAM-50 classifier.

Results: Among the 1–9%, 10% and &gt; 10% IHC positive cases, 24%, 67% and 92% were also ER-positive by ESR1 mRNA expression. The average ESR1 expression was significantly higher in the &gt; 10% IHC-positive cohorts compared to the 1–9% or completely negative cases but in these latter two cohorts ER expression was similarly low. The average ER gene signature scores were similar for the ER-negative and 1–9% IHC-positive cases, but significantly lower than in &gt; 10% ER-positive cases. None of the 1–9% ER-positive cases were classified as Luminal A, 2 were Luminal B and 12 were Basal-like. Among the 10% ER-positive cases, 2 were Luminal A and 1 was Luminal B. Conclusion: Overall, 24% of the 1–9% and 67% of the 10% ER-positive cancers show ESR1 mRNA levels and gene signatures that are consistent with ER-positive, potentially endocrine sensitive tumors.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P1-07-09.

Abstract PD07-03: A Genomic Predictor of Survival Following Taxane-Anthracycline Chemotherapy for Breast Cancer

Background: There is currently no predictive assay for patients with clinical Stage II-III breast cancer from which predicted sensitivity to treatment is associated with high probability of survival following chemotherapy.

Patients & Methods: We performed Affymetrix gene expression microarrays of prospectively collected tumor biopsies from 508 patients with newly diagnosed HER2-normal invasive breast cancer prior to neoadjuvant taxane-anthracycline chemotherapy followed by adjuvant endocrine therapy (if hormone receptor-positive). The predictor was developed from 310 samples (from MDACC & I-SPY) by combining: 1) a signature to predict sensitivity to endocrine therapy (SET); 2) estrogen receptor (ER)-stratified predictive signatures of resistance to chemotherapy, defined as extensive residual cancer burden (RCB-III) or relapse within 3 years; and 3) ER-stratified predictive signatures of response to chemotherapy, defined as pathologic complete response (pCR) or minimal RCB (RCB-I). The predictor classified tumors as treatment sensitive if high or intermediate SET, or if predicted to be responsive (and not resistant) to chemotherapy. Otherwise, tumors were classified as treatment insensitive. The predictor was then tested on an independent cohort (N= 198, 98% with clinical Stage II-III) who received neoadjuvant (N= 180) or adjuvant (N= 18) taxane-anthracycline chemotherapy (from MDACC, USO, GEICAM, Peru, LBJ). Distant relapse-free survival (DRFS) was evaluated at a 3-year median follow up using negative predictive value (NPV, absence of event if predicted to be sensitive), and absolute risk reduction (ARR) for those predicted to be sensitive (versus insensitive), with 95% confidence interval (CI). The independent predictive value was assessed in multivariate Cox regression analysis based on the likelihood ratio test (P≥0.05). Results: Patients in the independent validation cohort who were predicted to be treatment sensitive (28%) had excellent DRFS, with NPV 92% (CI 85-100) and significant absolute risk reduction (ARR 18%, CI 6-28) at 3 years, compared to those predicted to be insensitive. This was similar to the DRFS observed in patients who achieved pCR after they completed neoadjuvant chemotherapy (NPV 93%, CI 85-100). Predictions were accurate in each phenotypic subset: ER+/HER2- (30% predicted sensitive, NPV 97%, CI 91-100; ARR 11%, CI 0.1-21) and ER-/HER2- (26% predicted sensitive, NPV 83%, CI 68-100; ARR 26%, CI 4-28). Predicted treatment sensitivity (HR 0.20, CI 0.07-0.57), ER+ status (HR 0.32, CI 0.17-0.63), clinical tumor stage T3-4 (HR 2.04, CI 1.07-3.88) and age &gt;50 (HR 0.50, CI 0.25-0.98) were significant in a multivariate model that also included clinical nodal status, grade, and type of taxane used.

Conclusion: We report validation results for the first molecular predictor of sensitivity to neoadjuvant/adjuvant systemic therapy for clinical Stage II-III breast cancer that is independently associated with excellent DRFS in those predicted to be sensitive. Predictions were accurate for both ER+/HER2- and ER-/HER2- invasive breast cancer.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr PD07-03.

From the Lab to the Clinic: Gene-Expression Profiles That Are Associated with Mek-Inhibitor Sensitivity In Vitro Are Coordinately Co-Expressed in Breast Cancer Biopsy Samples from the I-SPY Trial (CALGB 150007/150012, ACRIN 6657)

There is increasing appreciation that response to breast cancer therapy is dependent on the specific characteristics of each tumor, as has also been observed in the first analyses of 221 patients treated by standard neo-adjuvant chemotherapy in the I-SPY 1 TRIAL (CALGB 150007/150012, ACRIN 6657). I-SPY 1 has collected full-genome gene expression data on pre-treatment biopsies as well as outcome data for initial tumor response (pathological assessment) and 3-year outcome data. In addition, we have established an in vitro high-throughput system using 50 breast cancer cell lines that can be used to identify gene-expression patterns predictive of drug sensitivity using 'Adaptive Splines'. Though several groups are using cell lines to develop drug response predictors, there is understandable skepticism as to whether the cancer cell lines have retained sufficient similarity to in vivo tumors to make such an endeavor successful.Methods: We used the in vitro cell lines to identify molecular markers of sensitivity to the MEK inhibitor CI1040. We next used the molecular data collected in I-SPY 1, in particular Affymetrix U133A data from 121 patients, to see whether subgroups could be identified among the ISPY patients that vary according to the expression of genes in this signature. In addition, we used the network visualization tool, EGAN, to identify biological networks among the significant predictors of CI-1040 sensitivity or resistance.Results: Analysis of in vitro data using adaptive splines identified 406 genes predictive of response to CI1040, of which 135 and 271 were expressed more highly in CI1040-resistant or -sensitive cell lines respectively. EGAN analyses of these predictors revealed multiple subnetworks and pathways as important. Some of these expression patterns could be corroborated in additional datasets. When we clustered the gene expression data from biopsies using the best linear-classifier based gene predictors of CI1040 sensitivity, the genes and patients each divided into two classes. Notably, most of the genes expressed more highly in the CI-1040-resistant cell lines formed one cluster, while most of the genes expressed more highly in the sensitive cell lines formed another cluster (p&lt;&lt;0.001; Fisher Exact Test), a strong indicator that these genes are similarly co-expressed in both cell lines and tumor samples. Preliminary results indicate that roughly a third of patients show a pattern of gene-expression that is similar to the CI1040-sensitive cell lines.Conclusion: The expression of genes identified to associate with CI1040 sensitivity in vitro show a similar pattern of co-expression in I-SPY patient samples. These data suggest that this CI1040-sensitivity predictor should be tested for its ability to predict the therapeutic response among patients receiving CI1040 (or other drugs in this class). The I-SPY 2 trial will include phase II targeted drugs for which predictor gene sets will be derived from our in vitro system and prospectively tested in patients receiving that treatment.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 2042.

The Molecular Anatomy of Breast Cancer Stroma; Independent Prognostic Role in ER-Positive and ER-Negative Cancers

BackgroundIn this paper we examine the various molecular components of a breast cancer stromal gene signature and correlate these with clinical phenotype and outcome including prognosis and response to preoperative chemotherapy.MethodsWe compared Affymetrix HGU133A-based gene expression profiles of 37 matching core needle biopsies (CNB) and fine-needle aspirations (FNA) from the same cancers. Genes over-expressed in CNB relative to the FNA were defined as the “stromal signature”. Gene expression data from 56 breast cancer cell lines, two separate neoadjuvant data sets (n=233, n=103), 3 independent cohorts of node negative, untreated patients (n=286, n=198, n=200) and 259 estrogen receptor-positive (ER+) tamoxifen-treated patients were used to assess the prognostic and predictive values of these genes in ER+ and ER- cancers separately. Univariate and multivariate Cox analyses were performed. Metagenes were defined as average expression of co-clustered genes.Results293 probe sets (206 genes) were significantly over-expressed in the CNBs (false discovery rate ≤ 0.001, fold-change ≥ 3). These genes overlapped with previously reported stromal signatures and fell into several co-expression clusters including a B-cell/Plasma Cell (B-cell), Dendritic cell, extracellular matrix (ECM), and TGFb-receptor metagenes. Interestingly, ER+ and ER- cancers showed a significantly different stroma-gene expression pattern, and many stromal genes were also differentially expressed between ER+ and ER- breast cancer cell lines. The ECM and TGFb metagenes had modest and variable prognostic value across different datasets in both ER groups. The Dendritic and B cell metagenes were highly co-expressed, but the B-cell metagene had more robust and consistent prognostic value. The B-cell metagene was statistically significant prognostic in univariate and multivariate analysis in ER+/High proliferative and ER- tumors, but it was not prognostic in ER+/Low proliferative tumors. In the 3 different node negative, untreated patient cohorts, the ER+/High proliferative cancers in the lowest B-cell metagene tertile had 10-year distant metastasis free survival (DMFS) of 0.18 (0.07-0.46), 0.22 (0.06-0.75), and 0.44 (0.21-0.92) compared to 0.71 (0.54-0.94), 0.89 (0.71-1.00) and 0.89 (0.71-1.00) in the highest tertile. Among the ER- cancers, the lowest B cell metagene group had 10-year DMFS of 0.57 (0.44-0.75), 0.63 (0.48-0.83), and 0.44 (0.25-0.76) compared to 0.93 (0.81-1.00), 0.83 (0.64-1.00) and 0.83 (0.58-1.00) in the highest B-cell metagene group in each prognostic dataset, respectively. The B-cell metagene was also prognostic in univariate (HR 0.83 (0.71-0.98) p=0.03) and multivariate (HR 0.80 (0.68-0.94) p=0.007) analysis in the Tamoxifen-treated cohort. None of the above stromal metagenes showed reproducible association with response to chemotherapy.ConclusionsThe B-cell/Plasma cell metagene component of the “stroma-related genes” is a robust and reproducible prognostic marker in ER+/High proliferative and also in ER- cancers. Other stromal genes are variably expressed in ER+ and ER- tumors and many are also expressed by neoplastic cells in culture and by primary tumors and carry less reproducible prognostic value.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 105.

Adipose tissue fatty acids in breast cancer patients versus healthy control women from Crete

BACKGROUND

Few studies have implemented biomarkers of fatty acid intake in relation to breast cancer.

AIMS

To examine possible differences in adipose tissue fatty acid composition between breast cancer patients and healthy control women. The relationship between tumor promotion and adipose tissue fatty acid synthesis was also investigated.

METHODS

The study was conducted at the University of Crete. Subjects included 94 women with clinically diagnosed cancer of the breast and 131 healthy control women. Histological tumor grading and breast cancer staging were assessed. Fatty acids were determined by gas chromatography in gluteal adipose tissue.

RESULTS

Conditional logistic regression analysis controlling for potential confounders indicated that elevated adipose monounsaturated fatty acids and oleic acid are associated with reduced odds of breast cancer [OR (T2 vs. T1) 0.15; 95% CI 0.03-0.64, and OR (T2 vs. T1) 0.18; 95% CI 0.04-0.71, respectively]. Adipose myristic acid was associated with an increase in breast cancer risk [OR (T3 vs. T1) 5.66; 95% CI 1.3-23.9].

CONCLUSIONS

Adipose oleic acid is inversely related, whereas adipose myristic acid is positively related to breast cancer risk. These relations could be mediated by Her-2/neu and FAS oncogenes.

Correlations of estrogen receptor (ER) related genomic transcription and ER gene expression with increasing AJCC stage of ER-positive breast cancer

1044

Background: Advanced stages of ER-positive breast cancer may have decreased tumor dependence on estrogen, possibly due to biological selection and/or progression. The sensitivity to endocrine therapy (SET) index measures transcriptional activity of the ER genomic pathway (165 ER-related genes). We investigated whether expression of genes for receptors ER (ESR1), progesterone receptor (PGR), or HER-2 (ERBB2), the SET index, or a housekeeper gene (GAPDH) vary by stage of ER-positive breast cancer. Methods: We evaluated gene expression profiles (Affymetrix U133 microarrays, Affymetrix, Santa Clara, CA) from 956 patients’ clinical samples of ER-positive breast cancer, including 290 new samples profiled at MDACC and 666 samples from published datasets. Microarray data were uniformly normalized, log-transformed, and expression levels for single genes and the SET index were compared to pathologic AJCC stage (315 patients were stage I, 362 stage IIA, 151 stage IIB, 29 stage III, 27 stage IV at initial presentation, and 72 stage IV previously treated and/or at relapse) using a median ordered regression analysis (p < 0.05 was significant). Results: SET index significantly decreased with advancing pathologic stage (p < 0.001), and PGR expression levels showed a similar, but lesser, effect (p = 0.014). However, expression levels of ESR1 and ERBB2 did not vary by stage. Overall, GAPDH gene expression increased with stage (p < 0.001), but that effect was only observed in stages III and IV. Expression of these genes was not significantly different between stage IV disease at initial presentation or at relapse. Conclusions: Expression levels of ESR1 and ERBB2 receptor genes did not vary, but ER-related genomic transcription (SET index and, to a lesser extent, PGR) declined significantly with increasing pathologic stage. This suggests that ER-positive breast cancer tends to have less transcriptional dependence on estrogen with increasing pathologic stage. The observed increase in expression of GAPDH in stages III and IV might reflect higher metabolic activity in advanced ER-positive breast cancer and deserves further study.

[Table: see text]

Hormone receptor status and pathologic response of HER2-positive breast cancer treated with neoadjuvant chemotherapy and trastuzumab

BACKGROUND

The aim of this study was to compare the extent of pathologic response in patients with HER2-positive (HER2+) breast cancer treated with standard neoadjuvant chemotherapy, with or without trastuzumab (H), according to hormone receptor (HR) status.

PATIENTS AND METHODS

We included 199 patients with HER2+ breast cancer from three successive cohorts of neo-adjuvant chemotherapy on the basis of paclitaxel (Taxol) (P) administered weekly (w) or three weekly (3-w), followed by 5-fluorouracil (F), doxorubicin (A) or epirubicin (E), and cyclophosphamide (C). Residual cancer burden (RCB) was determined from pathologic review of the primary tumor and lymph nodes and was classified as pathologic complete response (pCR) or minimal (RCB-I), moderate (RCB-II), or extensive (RCB-III) residual disease.

RESULTS

In HR-positive (HR+) cancers, a higher rate of pathologic response (pCR/RCB-I) was observed with concurrent H + 3-wP/FEC (73%) than with 3-wP/FEC (34%, P = 0.002) or wP/FAC (47%; P = 0.02) chemotherapy alone. In HR-negative (HR-) cancers, there were no significant differences in the rate of pathologic response (pCR/RCB-I) from 3-wP/FAC (50%), wP/FAC (68%), or concurrent H + 3-wP/FEC (72%).

CONCLUSIONS

Patients with HR+/HER2+ breast cancer obtained significant benefit from addition of trastuzumab to P/FEC chemotherapy; pathologic response rate was similar to that seen in HR-/HER2+ breast cancers.