Proteogenomic characterization of difficult-to-treat breast cancer with tumor cells enriched through laser microdissection

BACKGROUND

Breast cancer (BC) is the most commonly diagnosed cancer and the leading cause of cancer death among women globally. Despite advances, there is considerable variation in clinical outcomes for patients with non-luminal A tumors, classified as difficult-to-treat breast cancers (DTBC). This study aims to delineate the proteogenomic landscape of DTBC tumors compared to luminal A (LumA) tumors.

METHODS

We retrospectively collected a total of 117 untreated primary breast tumor specimens, focusing on DTBC subtypes. Breast tumors were processed by laser microdissection (LMD) to enrich tumor cells. DNA, RNA, and protein were simultaneously extracted from each tumor preparation, followed by whole genome sequencing, paired-end RNA sequencing, global proteomics and phosphoproteomics. Differential feature analysis, pathway analysis and survival analysis were performed to better understand DTBC and investigate biomarkers.

RESULTS

We observed distinct variations in gene mutations, structural variations, and chromosomal alterations between DTBC and LumA breast tumors. DTBC tumors predominantly had more mutations in TP53, PLXNB3, Zinc finger genes, and fewer mutations in SDC2, CDH1, PIK3CA, SVIL, and PTEN. Notably, Cytoband 1q21, which contains numerous cell proliferation-related genes, was significantly amplified in the DTBC tumors. LMD successfully minimized stromal components and increased RNA-protein concordance, as evidenced by stromal score comparisons and proteomic analysis. Distinct DTBC and LumA-enriched clusters were observed by proteomic and phosphoproteomic clustering analysis, some with survival differences. Phosphoproteomics identified two distinct phosphoproteomic profiles for high relapse-risk and low relapse-risk basal-like tumors, involving several genes known to be associated with breast cancer oncogenesis and progression, including KIAA1522, DCK, FOXO3, MYO9B, ARID1A, EPRS, ZC3HAV1, and RBM14. Lastly, an integrated pathway analysis of multi-omics data highlighted a robust enrichment of proliferation pathways in DTBC tumors.

CONCLUSIONS

This study provides an integrated proteogenomic characterization of DTBC vs LumA with tumor cells enriched through laser microdissection. We identified many common features of DTBC tumors and the phosphopeptides that could serve as potential biomarkers for high/low relapse-risk basal-like BC and possibly guide treatment selections.

Association of clinicopathologic and molecular factors with the occurrence of positive margins in breast cancer

PURPOSE

To explore the association of clinicopathologic and molecular factors with the occurrence of positive margins after first surgery in breast cancer.

METHODS

The clinical and RNA-Seq data for 951 (75 positive and 876 negative margins) primary breast cancer patients from The Cancer Genome Atlas (TCGA) were used. The role of each clinicopathologic factor for margin prediction and also their impact on survival were evaluated using logistic regression, Fisher's exact test, and Cox proportional hazards regression models. In addition, differential expression analysis on a matched dataset (71 positive and 71 negative margins) was performed using Deseq2 and LASSO regression.

RESULTS

Association studies showed that higher stage, larger tumor size (T), positive lymph nodes (N), and presence of distant metastasis (M) significantly contributed (p ≤ 0.05) to positive surgical margins. In case of surgery, lumpectomy was significantly associated with positive margin compared to mastectomy. Moreover, PAM50 Luminal A subtype had higher chance of positive margin resection compared to Basal-like subtype. Survival models demonstrated that positive margin status along with higher stage, higher TNM, and negative hormone receptor status was significant for disease progression. We also found that margin status might be a surrogate of tumor stage. In addition, 29 genes that could be potential positive margin predictors and 8 pathways were identified from molecular data analysis.

CONCLUSION

The occurrence of positive margins after surgery was associated with various clinical factors, similar to the findings reported in earlier studies. In addition, we found that the PAM50 intrinsic subtype Luminal A has more chance of obtaining positive margins compared to Basal type. As the first effort to pursue molecular understanding of the margin status, a gene panel of 29 genes including 17 protein-coding genes was also identified for potential prediction of the margin status which needs to be validated using a larger sample set.

Abstract P1-05-04: Novel breast cancer proteomic subtyping with connection to cell of origin

Introduction: In clinical practice, immunohistochemistry (IHC) is combined with clinicopathologic information to stratify patients into subtypes and guide treatment decisions. In our mass spectrometry study of 116 breast cancers (BrCAs), we identified 34 significant proteins which provide additional prognostic information. This novel proteomic subtyping, combining IHC with proteomics, separates HER2 negative (Her2-) BrCA patients into Luminal-like and TN-like subtypes. A novel subtype, luminal by IHC and TN-like by proteomics (L/T), is associated with unfavorable outcome compared with the luminal subtype (L/L) by both IHC and proteomics. These IHC-based Her2- subtypes are described as L/L, L/T, T/L and T/T subtypes. We investigated the cell-of-origin of our enhanced subtypes using gene set enrichment analysis with gene sets consisting of breast cancer cell-of-origin signatures. We further identified genes representing our 34 proteins in the significant gene sets and investigated their association with survival outcomes across The Cancer Genome Atlas (TCGA) pan-cancers. Method: The single-cell RNA-sequencing data analyses of health breast samples identified 23 breast cell-of-origin signatures (Bhat-Nakshatri et al., 2021) with 46 up and down-regulated gene sets based on the provided fold changes. Using our cohort, we performed differential analysis for each enhanced subtype (L/L, L/T, T/T) versus the other subtypes separately. Pre-ranked gene set enrichment analyses were used to identify significantly enriched gene sets. The significance of enriched gene sets was reported at FDR<0.05. We further identified genes of 34 biomarkers that overlapped in the identified significant gene sets and investigated the impact of the genes on survival across TCGA pan-cancers. Univariate overall survival (OS) and progression-free interval (PFI) analyses were performed on each cancer cohort and the significance of the association with survival outcome was reported by log-rank p-value <0.05. Results: Our signature was compared to the clusters from healthy tissues. We found up-regulated genes in L/L subtype were enriched in gene sets elevated in mature luminal annotated clusters (N8 and N12). Down-regulated genes in L/L subtype were enriched in the elevated genes in the N3 (luminal progenitor) and N9 (luminal progenitor/basal) clusters. Up-regulated genes in L/T subtype were enriched in the elevated genes in the N19 (luminal progenitor) and N11 (basal) clusters. Down-regulated genes in L/T were enriched in the elevated genes in the N8 (mature luminal) cluster and down-regulated genes in the N9 (luminal progenitor/basal) cluster. Up-regulated genes in T/T subtype were enriched in genes highly expressed in the luminal progenitor (N3) and luminal progenitor/basal (N9) clusters. Down-regulated genes in T/T subtype were enriched in the elevated genes in the mature luminal annotated clusters (N8, N12). One gene, KPNA2, was up-regulated in both the N9 (luminal progenitor/basal) cluster and the Basal-like (TN-like) subtype. The higher expression of this gene had significant association with worse OS outcomes in most of the cancers. Conclusions: T/T and L/T subtypes likely arise from progenitor and basal cells. The L/L subtype may arise from mature luminal cells. The novel L/T subtype appears to have characteristics that result in poorer survival. Disclaimers The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions or policies of USUHS, HJF, the DOD or the Departments of the Army, Navy, or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. Government.

Citation Format: Guisong Wang, Punit Shah, Rick Searfoss, J. Leigh Fantacone-Campbell, Jeffrey A. Hooke, Brenda Deyarmin, Rebecca N. Zingmark, Stella Somiari, Jianfang Liu, Leonid Kvecher, Lori A. Sturtz, Praveen-Kumar Raj-Kumar, Elder Granger, Linda Vahdat, Niven R. Narain, Mary L. Cutler, Rangaprasad Sarangarajan, Hai Hu, Michael A. Kiebish, Albert J. Kovatich, Craig D. Shriver. Novel breast cancer proteomic subtyping with connection to cell of origin [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P1-05-04.

Assessing the quality of RNA isolated from human breast tissue after ambient room temperature exposure

High quality human tissue is essential for molecular research, but pre-analytical conditions encountered during tissue collection could degrade tissue RNA. We evaluated how prolonged exposure of non-diseased breast tissue to ambient room temperature (22±1°C) impacted RNA quality. Breast tissue received between 70 to 190 minutes after excision was immediately flash frozen (FF) or embedded in Optimal Cutting Temperature (OCT) compound upon receipt (T0). Additional breast tissue pieces were further exposed to increments of 60 (T1 = T0+60 mins), 120 (T2 = T0+120 mins) and 180 (T3 = T0+180 mins) minutes of ambient room temperature before processing into FF and OCT. Total exposure, T3 (T0+180 mins) ranged from 250 minutes to 370 minutes. All samples (FF and OCT) were stored at -80°C before RNA isolation. The RNA quality assessment based on RNA Integrity Number (RIN) showed RINs for both FF and OCT samples were within the generally acceptable range (mean 7.88±0.90 to 8.52±0.66). No significant difference was observed when RIN at T0 was compared to RIN at T1, T2 and T3 (FF samples, p = 0.43, 0.56, 0.44; OCT samples, p = 0.25, 0.82, 1.0), or when RIN was compared between T1, T2 and T3. RNA quality assessed by quantitative real-time PCR (qRT-PCR) analysis of beta-actin (ACTB), glyceraldehyde-3-phosphate dehydrogenase (GAPDH), cyclophilin A (CYPA), and porphobilinogen deaminase (PBGD) transcripts showed threshold values (Ct) that indicate abundant and intact target nucleic acid in all samples (mean ranging from 14.1 to 25.3). The study shows that higher RIN values were obtained for non-diseased breast tissue up to 190 minutes after resection and prior to stabilization. Further experimental exposure up to 180 minutes had no significant effect on RIN values. This study strengthens the rationale for assessing RIN and specific gene transcript levels as an objective method for determining how suitable RNA will be for a specific research purpose (“fit-for purpose”).

Abstract 2342: Multidimensional metabolomic stratification of ER+/HER2- compared to ER-/HER2- breast tumors

Introduction: In the United States, breast cancer represents the leading cancer diagnosis among women and can readily be classified as a metabolic disease based on its distinct metabolic activity within the tumor microenvironment. Compared to other omics technologies, extensive lipidomic and metabolomic studies are lacking. Here in, we evaluated a cohort of 109 tumors characterized as ER+/HER2- and ER-/HER2- based on immunohistochemistry (IHC) and performed comprehensive structural lipidomic, signaling lipidomic, and global metabolomic analyses for an extensive characterization of the biophysical, signaling, and metabolic interplay between these tumors.

Methods: Clinical IHC subtyping of core biopsies was used to select a cohort of patients with ER+/HER2- or ER-/HER2- primary tumors from flash-frozen surgical samples. The positive/negative status of ER/PR/HER2 was defined using updated ASCO 2020 guidelines. Ki-67 status was determined using the 2011 St. Gallen's International Expert Consensus recommendations. ER low (1-10%) cases were excluded from this analysis. Structural lipidomic analysis was employed through the use of MS/MSALL high resolution shotgun lipidomics using a SCIEX 5600+ TripleTOF micro LC approach characterizing 23 lipid classes and over 1200 molecular species. Signaling lipids were analyzed using a SCIEX 6600 TripleTOF microLC platform characterizing 106 lipid analytes across octadecanoid, eicosanoid and docosanoid species. Metabolomics analysis was performed using LECO PEGASUS GC TOF, SCIEX 5500 HILIC LC MS/MS analysis, and SCIEX 6600 High resolution RP-LC-MS analysis detecting 450 metabolite Metabolomics data was further interpreted using MetaboAnalyst software.

Results/Conclusions: Compared to their ER+ counterparts, ER-/HER2- tumors exhibited a significant decrease in triacylglycerides, and a corresponding increase in cholesterol ester, phosphatidylcholine, phosphatidylethanolamine, and phosphatidylglycerol species demonstrating a signature of biophysical and metabolic rewiring with alterations in Kennedy pathway lipid shuttling. One signaling lipid was decreased and six were increased (predominantly arachidonic species) in ER-/HER2- tumors compared to ER+/HER2- ones. Metabolomic analysis revealed distinct alterations in cysteine/methionine, arginine/proline, purine, butanoate, and tryptophan metabolism. Utilizing a multidimensional metabolic integration approach, we identified distinct biophysical, signaling, and biochemical alterations in ER+/HER2- compared to ER-/HER2- breast tumors, which may impact selection of therapy and outcome in the future.

Citation Format: Valerie Bussberg, Vladimir Tolstikov, Guisong Wang, Punit Shah, Rick Searfoss, Leigh Fantacone-Campbell, Jeffrey A. Hooke, Brenda Deyarmin, Rebecca N. Zingmark, Stella Somiari, Jianfang Liu, Leonid Kvecher, Bradley Mostoller, Lori Sturtz, Praveen-Kumar Raj-Kumar, Elder Granger, Linda Vahdat, Mary L. Cutler, Chas Bountra, Rangaprasad Sarangarajan, Hai Hu, Albert J. Kovatich, Michael A. Kiebish, Niven R. Narain, Craig D. Shriver. Multidimensional metabolomic stratification of ER+/HER2- compared to ER-/HER2- breast tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2342.

Abstract 1188: Reclassification of ER+ (luminal A/luminal B1 minus ER low)-like and ER- like breast tumors based on proteomic/gene and clinical outcome signatures

Introduction: Classification of breast cancer can incorporate immunohistochemical (IHC) detection of ER/PR/HER2/KI67 to stratify the subtypes. High throughput proteomics analysis allows for the expansion of biomarker discovery within the subtypes. We evaluated a cohort of 109 tumors characterized as ER+ (Luminal A and Luminal B1; HER2+ and ER low (1-10%) cases were excluded) compared to ER-/HER2- tumors. Utilizing an integrated bioinformatics approach, we developed a proteomic marker signature to reclassify tumors into ER+(like) and ER-(like) tumors. CPTAC (Proteomic)/TCGA (RNAseq) datasets and larger METBRIC and GSE96058 cohorts were used to validate this marker signature. The selected biomarkers demonstrated significant differences impacting survival outcome.

Methods: Clinical IHC subtyping of core biopsies was used to select a cohort of patients with ER+/HER2- and ER-/HER2- primary tumors from flash-frozen surgical samples. The positive/negative status of ER/PR/HER2 was defined using updated ASCO 2020 guidelines. Ki-67 status was determined using the 2011 St. Gallen's International Expert Consensus recommendations. Proteomic analysis was performed using Thermo Q-Exactive+ LC MS/MS analysis. Differential analysis was applied to select the significantly altered proteins between ER+ and ER- cases, Univariate survival analysis was engaged to filter informative protein/genes using TCGA RNA-Seq data. Nearest centroid analysis was deployed to define the classifier to predict novel molecular subtypes.

Results/Conclusions: We selected 34 proteins/genes from 164 significantly differentially expressed proteins for further analysis. The centroid model constructed with the 34 proteins defined 2 groups: ER+(like) and ER-(like). An additional 4 groups were defined across subtypes: luminal tumors classified both by IHC and marker signature (LL), luminal tumors classified by IHC but marker signature more like triple negative (LT), triple negative tumors classified by IHC but marker signature more like luminal (TL), and triple negative classified by both IHC and marker signature (TT). This marker signature segregated close to 5000 tumors across CPTAC, TCGA, METABRIC and GSE96058 cohorts. Survival analysis in these groups of patients revealed differences in radiation, hormone/radiation, hormone therapy, and hormone/radiation/chemotherapy treatments. In summary using proteomics data we identified a 34 gene/protein marker signature, validated in large external cohorts and exhibited impact on survival and response to therapy. Further, this signature was enriched in metabolism and microenvironmental associated factors that could represent novel targets or development combination strategies based on this signature.

Citation Format: Guisong Wang, Punit Shah, Rick Searfoss, Leigh Fantacone-Campbell, Jeffrey A. Hooke, Brenda Deyarmin, Rebecca N. Zingmark, Stella Somiari, Jianfang Liu, Leonid Kvecher, Bradley Mostoller, Lori A. Sturtz, Praven-Kumar Raj-Kumar, Elder Granger, Linda Vahdat, Mary L. Cutler, Chas Bountra, Rangaprasad Sarangarajan, Hai Hu, Michael A. Kiebish, Albert J. Kovatich, Niven R. Narain, Craig D. Shriver. Reclassification of ER+ (luminal A/luminal B1 minus ER low)-like and ER- like breast tumors based on proteomic/gene and clinical outcome signatures [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1188.

Abstract PS5-34: Identification of proteomics-based biomarkers for ER+/HER2- breast cancer stratification: Implications on clinical outcome

Introduction: Improving stratification of breast cancer (BC) patients based on molecular signatures for treatment responses and clinical outcomes is a critical unmet need. Currently, endocrine therapy is first-line for hormone receptor positive (HR+) BC. Chemotherapy is added to patients with high-risk luminal BC. In practice, levels of Ki-67 are used to distinguish luminal tumors as low-risk luminal A (LA) and high-risk luminal B (LB) for adjuvant therapy decisions. Herein, proteomic tumor assessment from ER-positive HER2-negative (ER+/HER2-) BC patients was utilized to define molecular subtyping, estimate congruency between proteomic subtyping and traditionally used Ki67 marker, and define a new set of potential predictive and prognostic therapeutic biomarkers for ER+/HER2- BC patients.

Method/Result: Clinical immunohistochemistry (IHC) subtyping of core biopsies was used to select a cohort of 86 BC patients with ER+/HER2- primary tumors from flash-frozen surgical samples. The positive/negative status of ER/PR/HER2 was defined using updated ASCO 2020 guidelines. Ki-67 status was determined using the 2011 St. Gallen’s International Expert Consensus recommendations. The cohort includes 28 LA (Ki67 < 14%) cases and 58 LB1 (Ki67 >= 14%) cases. Integrated consensus clustering algorithms with the most varying proteins in our cohort were applied to identify proteomic subtypes. Two distinct separations were observed from the analysis, resulting in one cluster enriched with LA (40 cases) and the other enriched with LB1 (46 cases) called by Fisher’s exact test. These clusters matched 100% with the clusters generated using 900+ proteins common to the 1500+ proteins used in the CPTAC-BC proteomics-based subtyping analysis (Mertins et al. Nature 2016). The differential analyses demonstrated that there is no significant difference between Ki67-defined subtypes and proteomics-defined subtypes (LA-enriched vs. LA cases, LB1-enriched vs. LB1 cases),indicating they are consistent in the molecular profile. Differential analysis was performed to compare LB1-enriched versus LA-enriched cases, resulting in 672 significantly differentially expressed proteins defined at false discovery rate (FDR) < 0.05 and |log2(fold change)|>1. 353 of the 672 proteins were correlated with mRNA at Pearson correlation > 0.39 as reported in the CPTAC-BC study or cBioPortal for Cancer Genome, and their coding genes were used for progression free interval (PFI) analysis based on TCGA RNA-seq data in the TCGA ER+/HER2- cases (662 cases, c.f. Huo et al. JAMA Oncology 2017). 90 of the 353 coding genes significantly associated with PFI were detected at p-value<0.05. Unsupervised hierarchical clustering method and principal component analysis (PCA) of the 90 genes were applied to our cohort to investigate the clustering performance and 94.2% of the cases were clustered correctly using support vector machine (SVM) method after PCA analysis. Biological process and molecular function GO term over-representation analyses of the 90 coding genes were performed separately. Some significant and biologically meaningful GO terms were identified at FDR<0.05.

Conclusions: We identified a set of biomarkers that can be potentially employed as proteomic or gene signatures to stratify ER+/HER2- BC into low risk and high-risk groups.

Disclaimers The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions, or policies of Uniformed Services University of the Health Sciences, The Henry M Jackson Foundation for the Advancement of Military Medicine Inc., the Department of Defense or the Departments of the Army, Navy or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. Government.

Citation Format: Guisong Wang, Punit Shah, Rick Searfoss, J. Leigh Fantacone Campbell, Jeffrey A. Hooke, Brenda Deyarmin, Rebecca N. Zingmark, Stella Somiari, Jianfang Liu, Leonid Kvecher, Lori A. Sturtz, Praveen-Kumar Raj-Kumar, Elder Granger, Linda Vahdat, Mary L. Cutler, Rangaprasad Sarangarajan, Hai Hu, Michael A. Kiebish, Albert J. Kovatich, Niven R. Narain, Craig D. Shriver. Identification of proteomics-based biomarkers for ER+/HER2- breast cancer stratification: Implications on clinical outcome [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS5-34.

Abstract PS18-38: Comparative analysis of differentially abundant proteins quantified by LC-MS/MS between flash frozen and laser microdissected OCT-embedded breast tumor samples

Background: Proteomic studies are typically conducted using flash-frozen (FF) samples utilizing tandem mass spectrometry. However, FF samples are comprised of multiple cell types, making it difficult to ascertain the proteomic profiles of specific cells. Conversely, OCT-embedded (Optimal Cutting Temperature compound) specimens can undergo laser microdissection (LMD) to capture and study specific cell types separately from the cell mixture. In the current study, we compared proteomic data obtained from FF and OCT samples to determine if samples that are stored and processed differently produce comparable results. Methods: Proteins were extracted from FF and OCT-embedded invasive breast tumors from 5 female patients. FF samples were lysed via homogenization (FF/HOM) while OCT-embedded specimens underwent LMD to collect only tumor cells (OCT/LMD-T) or both tumor and stromal cells (OCT/LMD-TS) followed by incubation at 37°C. Proteins were extracted using the illustra triplePrep kit and then trypsin-digested, TMT-labeled, and processed by two-dimensional liquid chromatography-tandem mass spectrometry (2D LC-MS/MS). Proteins were identified and quantified with Proteome Discoverer v1.4 and comparative analyses performed to identify proteins that were significantly differentially expressed amongst the different processing methods. Results: Among 4,950 proteins consistently quantified across all samples, 216 and 171 proteins were significantly differentially expressed (adjusted p-value < 0.05; |log2 FC| > 1) between FF/HOM vs. OCT/LMD-T and FF/HOM vs. OCT/LMD-TS, respectively, with most proteins being more highly abundant in the FF/HOM samples. PCA and unsupervised hierarchical clustering analysis with these 216 and 171 proteins were able to distinguish FF/HOM from OCT/LMD-T and OCT/LMD-TS samples, respectively. Likewise, PCA analysis and unsupervised clustering analysis using the 402 and 60 significantly differentially enriched GO terms (adjusted p-value (BH) < 0.2) in the FF/HOM vs. OCT/LMD-T and FF/HOM vs OCT/LMD-TS comparisons, respectively, not only distinguished OCT/LMD from FF/HOM samples but also separated LA and LB1 breast cancer subtypes within each storage/preparation method from one another. Although FF/HOM appears to be more similar to OCT/LMD-TS than OCT/LMD-T based on the number of differentially enriched proteins (216 vs. 171; p=0.022) and GO terms (402 vs. 60; p < 2.2 x 10-16), FF/HOM shows no greater similarity to OCT/LMD-TS than OCT/LMD-T based on PCA analysis with either proteins or GO terms ( based on weighted distance for pairwise samples, p = 0.97 from paired t-test). No significantly differentially enriched proteins or GO terms were detected between the OCT/LMD-T and OCT/LMD-TS samples but trended differences were detected. Conclusions: The proteomic profiles of the OCT/LMD-TS samples were more similar to those from OCT/LMD-T samples than FF/HOM samples, suggesting a strong influence from the sample processing methods. These results indicate that in LC-MS/MS proteomic studies, FF/HOM samples exhibit different protein profiles from OCT/LMD samples and thus, results from these two different methods cannot be directly compared. Our study also provides preliminary data for designing new studies to explore why OCT/LMD-TS samples are more similar to OCT/LMD-T than to FF/HOM samples, and to separate LA from LB1 samples. Disclaimer: The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions or policies of USUHS, HJF, the DOD or the Departments of the Army, Navy or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. Government.

Citation Format: Lori A. Sturtz, Guisong Wang, Punit Shah, Richard Searfoss, Praveen-Kumar Raj-Kumar, Jeffrey A. Hooke, J. Leigh Fantacone-Campbell, Brenda Deyarmin, Mary Lou Cutler, Rangaprasad Sarangarajan, Niven R. Narain, Hai Hu, Michael A. Kiebish, Albert J. Kovatich, Craig D. Shriver. Comparative analysis of differentially abundant proteins quantified by LC-MS/MS between flash frozen and laser microdissected OCT-embedded breast tumor samples [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS18-38.

Comparative analysis of differentially abundant proteins quantified by LC-MS/MS between flash frozen and laser microdissected OCT-embedded breast tumor samples

Background

Proteomic studies are typically conducted using flash-frozen (FF) samples utilizing tandem mass spectrometry (MS). However, FF specimens are comprised of multiple cell types, making it difficult to ascertain the proteomic profiles of specific cells. Conversely, OCT-embedded (Optimal Cutting Temperature compound) specimens can undergo laser microdissection (LMD) to capture and study specific cell types separately from the cell mixture. In the current study, we compared proteomic data obtained from FF and OCT samples to determine if samples that are stored and processed differently produce comparable results.

Methods

Proteins were extracted from FF and OCT-embedded invasive breast tumors from 5 female patients. FF specimens were lysed via homogenization (FF/HOM) while OCT-embedded specimens underwent LMD to collect only tumor cells (OCT/LMD-T) or both tumor and stromal cells (OCT/LMD-TS) followed by incubation at 37 °C. Proteins were extracted using the illustra triplePrep kit and then trypsin-digested, TMT-labeled, and processed by two-dimensional liquid chromatography-tandem mass spectrometry (2D LC–MS/MS). Proteins were identified and quantified with Proteome Discoverer v1.4 and comparative analyses performed to identify proteins that were significantly differentially expressed amongst the different processing methods.

Results

Among the 4,950 proteins consistently quantified across all samples, 216 and 171 proteins were significantly differentially expressed (adjusted p-value < 0.05; |log₂ FC|> 1) between FF/HOM vs. OCT/LMD-T and FF/HOM vs. OCT/LMD-TS, respectively, with most proteins being more highly abundant in the FF/HOM samples. PCA and unsupervised hierarchical clustering analysis with these 216 and 171 proteins were able to distinguish FF/HOM from OCT/LMD-T and OCT/LMD-TS samples, respectively. Similar analyses using significantly differentially enriched GO terms also discriminated FF/HOM from OCT/LMD samples. No significantly differentially expressed proteins were detected between the OCT/LMD-T and OCT/LMD-TS samples but trended differences were detected.

Conclusions

The proteomic profiles of the OCT/LMD-TS samples were more similar to those from OCT/LMD-T samples than FF/HOM samples, suggesting a strong influence from the sample processing methods. These results indicate that in LC–MS/MS proteomic studies, FF/HOM samples exhibit different protein expression profiles from OCT/LMD samples and thus, results from these two different methods cannot be directly compared.

Abstract P2-18-03: Breast cancer treatment and association with clinicopathologic features in TCGA

Background: For a decade, The Cancer Genome Atlas (TCGA) program collected clinicopathologic annotation data along with multi-platform molecular profiles of more than 11,000 human tumors across 33 different cancer types. TCGA clinical data were analyzed and a standardized dataset named the TCGA Pan-Cancer Clinical Data Resource was created. However, TCGA treatment data are not systematically analyzed. Here we focus on TCGA primary breast cancer (TCGA-BC) treatment data to assess their completeness, regimen patterns, and association with clinicopathologic features.

Method: 814 TCGA-BC patients with treatment data diagnosed from 2001 to 2013 were selected for this study. The treatment data were prepared and classified to be adjuvant Chemotherapy (CT), adjuvant Radiation Therapy (RT), and adjuvant Hormone Therapy (HT). An in-house Clinical Breast Care Project (CBCP) treatment dataset (n=1051 from 2001 to 2017), which were relatively complete, were used as a reference for data completeness. Multinomial logistic regression was used to analyze the associations between different therapies and clinical features.

Results: There is no consistent treatment data difference between TCGA-BC and CBCP. In TCGA-BC, 68.7%, 64.4%, and 64.5% of patients received CT, RT, and HT respectively, whereas in CBCP, the corresponding percentages were 59.0%, 71.2%, and 77.1%. The percentages of patients receiving combined therapies are even more comparable between the two cohorts (data not shown due to many combinations).

The associations between treatment and clinicopathologic features were analyzed using multinomial logistic regressions with age, race, menopausal status, AJCC stage, and PAM50 subtype as covariates. Several of these covariates significantly associate with the use of therapies. Compared to patients with age &lt; 50, patients with age ≥ 65 were significantly more likely to receive HT, HT+RT, or RT than CT only (OR=11.34, 9.25, 4.82; p=0.001, 0.002, and 0.024 respectively); Compared to pre-menopausal patients, post-menopausal patients were significantly more likely to receive HT+RT than CT only (OR=4.12, p=0.034); Compared to patients with early stages (I, II), patients with advanced stages (III, IV) were more likely to receive CT+HT+RT (OR=1.92, p=0.038) or CT+RT (OR=2.38, p=0.010) than CT only, and less likely to receive CT+HT (OR=0.39, p=0.033) or HT (OR=0.30, p=0.008) than CT only. PAM50 subtype also significantly associates with the use of therapies. Compared to Luminal A patients, patients with Basal subtype were significantly less likely to receive CT+HT (OR=0.14, p=1.3e-05), CT+HT+RT (OR=0.06, p=3.0e-10), HT+RT (OR=0.01, p=9.2e-06), or HT (OR=0.02, p=3.8e-07) than CT alone. Patients with HER2+ subtype showed similar patterns (note targeted therapies were classified as CT). In addition, patients with Basal subtype were significantly more likely to receive CT+RT (OR=2.47, p=0.01) than CT only.

Conclusion: TCGA-BC patient treatment data are relatively as complete as our in-house CBCP patient treatment data, enabling us to perform a preliminary analysis of the former for association with clinicopathologic features of the patients. The significant associations of age, menopausal status, AJCC stage, and PAM50 subtype with treatment regimens are consistent with clinical knowledge, suggesting potential validity of TCGA BC treatment data for research use.

Disclaimer: The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions or policies of Uniformed Services University of the Health Sciences (USUHS), The Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc., the Department of Defense (DoD), the Departments of the Army, Navy, or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. Government.

Citation Format: Jianfang Liu, J. Leigh Fantacone-Campbell, Albert J. Kovatich, Brenda Deyarmin, Bradley J. Mostoller, Jeffrey A. Hooke, Hallgeir Rui, Craig D. Shriver, Hai Hu. Breast cancer treatment and association with clinicopathologic features in TCGA [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P2-18-03.

Abstract P6-06-09: Evaluation of laser microdissected primary breast tumors for RNA Seq over bulk processing and validated with cohort control

Introduction: Laser microdissection (LMD) is a valuable method to isolate target populations of cells for molecular analysis. LMD of breast tumor samples can isolate breast tumor cells whereas bulk processing of tumor tissue will incorporate surrounding non-cancerous cells and bias tumor expression profiling. Here, we evaluated the advantage of using LMD breast tumors for RNA-Seq over bulk processing.

Methods: Tissue samples for the in-house dataset were from breast cancer patients consented by a HIPAA-compliant, IRB-approved protocol of the Clinical Breast Care Project. A total of 118 primary breast tumors embedded in OCT (Optimum Cutting Temperature) were selected and processed by LMD. Total RNA and protein were extracted using the illustra triplePrep kit. Paired-end RNA sequencing of 118 cases was performed using the Illumina HiSeq platform and the reads were preprocessed using a PERL-based pipeline involving PRINSEQ, GSNAP and HTSeq. The Cancer Genome Atlas (TCGA) primary breast cancer RNA-Seq data for 1097 tumors, bulk processed was downloaded. Differential expression of genes (DEG) was assessed using DESeq2. Significance was described for DEG with fold change &gt;2 and p-adjusted value of 0.05.

Results: A total of 24,518 genes with a mean expression of ≥ 10 (~9%) raw counts across 118 tumor samples were identified in the in-house LMD dataset. In TCGA breast cancer RNA-Seq, 14,281 genes with a mean expression of ≥ 100 (~9%) raw counts across 1097 tumor samples were identified. The conventional PAM50 classifier was used for intrinsic subtyping of in-house data, yielding 36 Basal-like, 14 HER2-enriched, 43 Luminal A, 22 Luminal B and 3 Normal-like calls. The provided PAM50 calls for TCGA were 192 Basal-like, 82 HER2-enriched, 566 Luminal A, 217 Luminal B and 40 Normal-like calls. Within commonly expressed 13,165 genes, LMD (in-house) and bulk (TCGA) processing exhibited approximately 40-78% non-overlap in significantly differentially expressed genes (SDEG) among the conventional intrinsic subtypes. 21 unique stromal genes were present in SDEG unique to TCGA whereas there were only 5 SDEG unique to in-house dataset. We validated the results with 34 patients that had both LMD and bulk processing RNA-Seq data and found the non-overlap genes percentage to be even greater from 46-85%. The observed percentages of non-overlapping genes in the whole datasets were also validated in the 34 overlapping cases when using IHC subtypes. Overall high positive correlation is observed among the stromal genes present in SDEG unique to TCGA suggesting strong stromal contribution in bulk processing. Pathway analysis of SDEG unique to LMD data suggested alterations in known cancer pathways (B-cell immune response, RNA metabolism and splicing, phagocytosis, and signaling components).

Conclusion: Analysis of The Cancer Genome Atlas breast cancer RNA-Seq data set (based on bulk tissue processing) suggested contribution of stromal signature genes and important differences from LMD specimens. Thus, tumor selection via LMD may allow us to unveil signals that are more specific to cancer cells.

Disclaimer: The contents of this publication are the sole responsibility of the author(s) and do not necessarily reflect the views, opinions or policies of Uniformed Services University of the Health Sciences (USUHS), The Henry M. Jackson Foundation for the Advancement of Military Medicine, Inc., the Department of Defense (DoD), the Departments of the Army, Navy, or Air Force. Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. Government.

Citation Format: Praveen-Kumar Raj-Kumar, Lori A. Sturtz, Albert J. Kovatich, Brenda Deyarmin, Jeffrey A. Hooke, Leigh Fantacone-Campbell, Anupama Praveen-Kumar, Jianfang Liu, James Craig, Leonid Kvecher, Jennifer Kane, Jennifer Melley, Stella Somiari, Stephen C. Benz, Justin Golovato, Shahrooz Rabizadeh, Patrick Soon-Shiong, Richard Mural, Craig D. Shriver, Hai Hu. Evaluation of laser microdissected primary breast tumors for RNA Seq over bulk processing and validated with cohort control [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P6-06-09.

Abstract 3402: Evaluation of laser microdissected primary breast tumors for RNA-Seq over bulk processing

Introduction: RNA-Seq based gene expression profiling of breast tumor samples is widely used to subgroup patients and to identify gene signatures of prognostic value. However, tumor samples are highly heterogeneous, and so bulk processing of tumor tissue will consist of several different cell types. Here, we evaluated the advantage of using laser microdissected (LMD) breast tumors for RNA-Seq over bulk processing.

Methods: Patients for the in-house dataset were duly consented under an IRB-approved protocol of the Clinical Breast Care Project. A total of 118 primary breast tumors embedded in OCT (Optimum Cutting Temperature) were selected and processed by LMD. Total RNA and protein were extracted using the Illustra triplePrep kit. Paired-end RNA sequencing of 118 cases was performed using the Illumina HiSeq platform and the reads were preprocessed using a PERL-based pipeline involving PRINSEQ, GSNAP and HTSeq. The Cancer Genome Atlas (TCGA) primary breast cancer RNA-Seq data for 1097 samples was downloaded. Differential expression of genes (DEG) was assessed using DESeq2. Significance was described for DEG with fold change &gt;2 and p-adjusted value of 0.05.

Results: A total of 24,518 genes with a mean expression of ≥ 10 raw counts across 118 tumor samples were identified in the in-house LMD dataset. In TCGA breast cancer RNA-Seq, 14,281 genes with a mean expression of ≥ 100 raw counts across 1097 tumor samples were identified. The conventional PAM50 classifier was used for intrinsic subtyping of in-house data, yielding 36 Basal-like, 14 HER2-enriched, 43 Luminal A, 22 Luminal B and 3 Normal-like calls. The provided PAM50 calls were used for TCGA which are 192 Basal-like, 82 HER2-enriched, 566 Luminal A, 217 Luminal B and 40 Normal-like calls. Within commonly expressed 13,165 genes, LMD and bulk processing exhibited approximately 40-78% non-overlap in significantly differentially expressed genes (SDEG) among the intrinsic subtypes. 21 unique stromal genes were present in SDEG unique to TCGA whereas there were only 5 SDEG unique to in-house dataset. Overall high positive correlation is observed among the stromal genes present in SDEG unique to TCGA suggesting strong stromal contribution in bulk processing. Pathway analysis of SDEG unique to LMD data suggested alterations in known cancer pathways (B-cell immune response, RNA metabolism and splicing, phagocytosis, and signaling components).

Conclusion: Analysis of The Cancer Genome Atlas breast cancer RNA-Seq data set (based on bulk processing tissue) suggested contribution of stromal signature genes and important differences from LMD specimens. Thus, tumor selection via LMD can result in better expression profiling by RNA-Seq which has the potential to uncover many cancer genes and pathways. The views expressed in this abstract are those of the author and do not reflect the official policy of the Department of Army/Navy/Air Force, Department of Defense, or U.S. Government.

Citation Format: Praveen-Kumar Raj-Kumar, Lori A. Sturtz, Albert J. Kovatich, Brenda Deyarmin, Jeffrey A. Hooke, Leigh Fantacone-Campbell, Anupama Praveen-Kumar, Jianfang Liu, James Craig, Leonid Kvecher, Jennifer Kane, Jennifer Melley, Stella Somiari, Stephen C. Benz, Justin Golovato, Shahrooz Rabizadeh, Patrick Soon-Shiong, Richard J. Mural, Craig D. Shriver, Hai Hu. Evaluation of laser microdissected primary breast tumors for RNA-Seq over bulk processing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3402.

Abstract 284: Integrated proteogenomic analysis of laser microdissected primary breast tumors define proteome clusters

Introduction: Breast tumors have 4 well-established intrinsic subtypes based on transcriptome profiling. However, clusters defined by proteomics are often in disagreement with those defined by transcriptomics. Here, we report the findings of proteogenomic profiling of 118 laser microdissected (LMD) breast tumors using RNA-Seq and mass-spectrometry (MS)-based proteomic technologies.

Methods: Cases used in this study were drawn from the Clinical Breast Care Project, with patients consented using an IRB-approved protocol. A total of 118 primary breast tumors embedded in OCT were selected and processed by LMD. Total RNA and protein were extracted using the Illustra triplePrep kit. Paired-end RNA sequencing of 118 cases was performed using the Illumina HiSeq platform, and the reads were preprocessed using a PERL-based pipeline involving the preprocessing tool PRINSEQ, splice-aligner GSNAP and HTSeq for quantifying expression. Quantitative global proteomics analyses were performed on 113 cases using isobaric TMT 6-plex labeling with the “universal reference” strategy. MS data were acquired using a Q-Exactive instrument and analyzed using Proteome Discoverer with Byonic node. Sample-to-sample normalization was conducted to remove pipetting errors and ComBat was used to remove batch effect. K-means clustering was done using Bioconductor package Consensusclustering.

Results: The number of preprocessed RNA sequencing reads for the 118 cases ranged from over 43 to 295 million. An average of 83% of reads was mapped, and 24,518 genes with a mean expression of ≥ 10 counts across 118 tumor samples were identified. The PAM50 algorithm was used for intrinsic subtyping, yielding 37 Basal-like, 16 HER2-enriched, 39 Luminal A and 26 Luminal B calls. Unsupervised clustering of 3,000 highly varying genes reflected 4 intrinsic subtypes. In the global proteomics data, 840 proteins were identified across all 113 cases. Unsupervised K-means consensus clustering on all 840 or just using the top 210 highly varying proteins indicated the optimal number of clusters to be 3. These 3 clusters were identified as Basal-enriched, Luminal A-enriched and Luminal B-enriched. HER2-enriched cases were distributed among these clusters.

We did not observe a stromal-enriched cluster in this analysis of LMD-prepared samples that selected against stromal components of the tumor.

Conclusion: Analysis of LMD breast tumors using proteogenomic technologies resulted in 3 clusters for proteome data: basal-enriched, luminal A-enriched and luminal B-enriched. Unlike a recent report on proteomics clustering using bulk processing of tumors, a stromal-enriched cluster was not observed in this analysis which excluded stromal components of the samples.

The views expressed in this abstract are those of the author and do not reflect the official policy of the Department of Army/Navy/Air Force, Department of Defense, or U.S. Government.

Citation Format: Praveen-Kumar Raj-Kumar, Tao Liu, Lori A. Sturtz, Albert J. Kovatich, Marina A. Gritsenko, Vladislav A. Petyuk, Brenda Deyarmin, Viswanadham Sridhara, James Craig, Jason E. McDermott, Anil K. Shukla, Ronald J. Moore, Matthew E. Monroe, Bobbie-Jo M. Webb-Robertson, Jeffrey A. Hooke, J.Leigh Fantacone-Campbell, Leonid Kvecher, Jianfang Liu, Jennifer Kane, Jennifer Melley, Stella Somiari, Stephen C. Benz, Justin Golovato, Shahrooz Rabizadeh, Patrick Soon-Shiong, Richard D. Smith, Richard J. Mural, Karin D. Rodland, Craig D. Shriver, Hai Hu. Integrated proteogenomic analysis of laser microdissected primary breast tumors define proteome clusters [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 284.

Abstract 5660: Proteomic comparative assessment of flash-frozen and OCT embedded breast cancer tissues for utilization in precision oncology discovery studies

Proteomic analysis is an important tool in classifying molecular architecture of tissue samples for subtyping and precision oncology applications. Tissues are commonly either flash frozen (FF) or embedded in optimal cutting temperature (OCT) compound for long-term storage. OCT provides a convenient specimen matrix for cryostat sectioning but contains glycols and resins which are incompatible with mass spectrometric analysis of proteins. In contrast, flash frozen preservation of tissue is historically considered to be more suitable and compatible with mass spectrometry applications. As a pilot study, we evaluated different tissue preservation methods, sample processing methods, and protein extraction methods, and their impact on quantitation and coverage of the proteome. In this pilot study, 30 breast cancer cases with matching flash frozen and OCT-embedded samples were used for proteomic analysis. Tissues were procured from the Clinical Breast Care Project (CBCP) in partnership with the Chan Soon-Shiong Institute of Molecular Medicine at Windber, and the Walter Reed National Military Medical Center (WRNMMC). Breast cancer subtypes were determined by clinical and expanded immunohistochemical panels. These subtypes were Her2, Luminal A (LA), Luminal B (LB) and Triple Negative (TN). Flash frozen tissues were cryosectioned to isolate 60-100 mg total tissue per case, and proteins were extracted using an 8M urea lysis buffer. The OCT specimens were laser microdissected to isolate primarily tumor cells from the samples. Proteins were extracted using the illustra TriplePrep kit. Digested proteins were then multiplexed using Tandem-Mass-Tag (TMT) 10plex isobaric labeling reagents, and chromatographically resolved using a Waters 2D nanoAcquity liquid chromatographer (LC) coupled to a Thermo Q Exactive Plus mass spectrometer (MS). Data was then analyzed using Proteome Discoverer. In total, 6130 proteins were identified using a 1% peptide FDR confidence filter in the 30 samples analyzed. Unsupervised hierarchical clustering of proteins across the samples resulted in two primary clusters, OCT and FF. Within the primary clustering, there were sub-clusters of Luminal, Her2, and Triple Negatives. At 2-fold threshold, 246 proteins were differentially expressed between OCT and FF tissues. However, the differential proteins between ER+ (LA and LB) and ER- (Her2 and TN) tumors correlate irrespective of the cryopreservation and processing methods. In conclusion, proteome analysis of samples stored under two different conditions provide subtyping information; however, there is differential expression of proteins between the two types of sample handling techniques which can impact biological interpretation of breast cancer precision oncology studies.

Citation Format: Kiki Panagopoulos, Punit Shah, Albert Kovatich, Michael A. Kiebish, Jeffrey Hooke, Leigh Campbell, Mary Lou Cutler, Audrey Kovatich, Hai Hu, Brenda Deyarmin, Lori Sturtz, Praveen Kumar Raj Kumar, Stella Somiari, Emily Y. Chen, Elder Granger, Viatcheslav Akmaev, Rangaprasad Sarangarajan, Niven Narain, Craig D. Shriver. Proteomic comparative assessment of flash-frozen and OCT embedded breast cancer tissues for utilization in precision oncology discovery studies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5660.

Abstract P4-09-14: Analysis of breast cancer recurrence using gene set enrichment analysis

Background: Even after successful treatment of primary breast tumors, there is a continued risk of recurrence. The risk varies between subtypes and there are ongoing efforts that aim to improve prediction of such risks for individual patients. Detection of subclinical metastases might be achieved by biomarkers in blood. In this study, we profiled protein expression in blood plasma from patients with known clinical outcome (recurrence vs no recurrence) to identify prognostic markers of breast cancer recurrence.

Methods: The subjects and specimens were made available through the Clinical Breast Care Project using IRB-approved protocols. We analyzed blood plasma samples taken at the time of diagnosis from consented patients who subsequently relapsed (33 cases) as well as those with no disease recurrence (31 controls). Based on hormone receptor and lymph node status the samples were grouped as: ER-/HER2- (17 cases/15 controls), ER+/LN+ (10/10) and ER+/LN- (6/6). We used aptamer-based SOMAscan assay platform to study the expression of 1252 proteins. We analyzed the protein expression data by using their coding genes in order to apply the Gene Set Enrichment Analysis method (GSEA v.2, Broad Institute). Pathway databases of KEGG, REACTOME, BIOCARTA and C4 collection were used. Significant gene sets were called at 5% FDR, and overlaps and low coverage gene sets (Tags &lt;70%) were removed. Statistical analysis and clustering were done using R.

Results: Unsupervised clustering showed some difference in signal in the ER+/LN- group. Even though there was a lack of significantly differentiated proteins between the cases and controls of this group, many significant gene sets were identified. After applying the cutoff filters and removing the overlaps, there were 5 gene sets enriched with the pathway collection, involved in B-cell receptor signaling, mRNA metabolism, tight junction and SCF-KIT signaling. Similarly, 9 gene sets from the MORF compendium were differentially expressed with the C4 collection and included neighborhood genes of NME2, ACTG1, EIF3S2, AP2M1, DAP3, UBE2I, NPM1, AATF and NPM1. In contrast, neither differentially expressed proteins nor gene sets were identified from the ER+/LN+ and ER-/HER2- groups. Since the sample size of the ER+/LN- group was small, we conducted a similar analysis by randomly choosing 6 case and control samples in the other two groups respectively. There were still no differentially expressed proteins or gene sets identified above the specified cutoff parameters.

Conclusion: Using plasma protein expression data we identified underlying gene sets differentially expressed between ER+/LN- patients who had cancer recurrence and no recurrence. Many genes in these sets were already known biomarkers (e.g. PTEN, AKT1, STAT3, SET etc.). These results can be used for understanding patterns of recurrence in different cancer subtypes. Further research is needed to estimate the clinical significance of these gene products.

The views expressed in this article are those of the author and do not reflect the official policy of the Department of Army/Navy/Air Force, the Department of Defense, or U.S. Government.

Citation Format: Praveen Kumar A, Kovatich AJ, Biancotto A, Cheung F, Davidson-Moncada JK, Kvecher L, Liu J, Ru Y, Kovatich AW, Deyarmin B, Fantacone-Campbell JL, Hooke JA, Raj Kumar PK, Rui H, Hu H, Shriver CD. Analysis of breast cancer recurrence using gene set enrichment analysis [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 P4-09-14.

Abstract 213: Integrated proteogenomic analysis of laser capture microdissected breast tumors

Introduction: Molecular characteristics of breast tumors play an important role in determining patients’ survival outcome. Here, we report preliminary findings of proteogenomic profiling of 50 breast tumors using RNA-Seq and mass-spectrometry (MS) based proteomic technologies. An additional 60 tumors are being analyzed, including WGS for all samples. We are also collecting patient survival outcome data.

Methods: Cases used in this study were drawn from the Clinical Breast Care Project, where patients were consented using an IRB-approved protocol. A total of 50 breast tumors were selected and processed by laser capture microdissection (LCM). This cohort includes 36 Caucasian Americans (CA) and 8 African Americans (AA), and the age of the patients is 57 ± 13 years. Protein and RNA were extracted using the Illustra triplePrep kit, which isolates DNA from the same cells as well. Quantitative global proteomics and phosphoproteomics analyses were performed using isobaric TMT 6-plex labeling with the “universal reference” strategy and IMAC enrichment of phosphopeptides. Mass spectrometry data were acquired using a Q-Exactive instrument and analyzed using Proteome Discoverer with Byonic node. Phosphopeptide abundance was normalized to abundance measurements of the parent protein for all of the phosphorylation analyses. Phosphoproteomic data was also searched for the presence of O-GlcNAc modifications. RNA-Seq analyses were done on Illumina HiSeq and the data were analyzed using GSNAP.

Results: There were 19 Luminal A, 7 Luminal B, 8 HER2-enriched, and 16 basal-like subtypes based on the PAM50 algorithm. In the global proteomics data, we were able to quantitate &gt;8600 proteins. Unsupervised clustering on the highly varying proteins across the samples resulted in two primary clusters, with one being luminal-enriched. The other cluster contains a basal-like tumor sub-cluster and a sub-cluster of mixed subtypes. Differential protein expression analyses between the two primary clusters confirmed known markers (e.g., overexpression of KRT8/KRT18 in luminal-enriched cluster). The luminal-enriched cluster is primarily CA with post-menopausal status.

A similar search of the phosphoproteomic data yielded quantitation of &gt;12500 phosphopeptides. Unsupervised clustering of the phosphoproteins resulted in four primary groups, with one being basal-enriched and another being luminal-enriched. We also observed &gt;50 overexpressed phosphopeptides. While some of these phosphosites have been previously reported (e.g., on RANBP2), other phosphosites appeared to be novel (e.g., on IRF2BP2).

Conclusion: Analysis of LCM breast tumors using proteogenomic technologies resulted in basal- and luminal-enriched clusters, thus enabling us to study protein and phosphopeptide markers across multiple platforms.

The views expressed in this article are those of the author and do not reflect the official policy of the Department of Defense, or U.S. Government.

Citation Format: Viswanadham Sridhara, Tao Liu, Marina A. Gritsenko, Lori A. Sturtz, Albert J. Kovatich, Vladislav A. Petyuk, Brenda Deyarmin, Jason E. McDermott, Anil K. Shukla, Ronald J. Moore, Matthew E. Monroe, Bobbie-Jo M. Webb-Robertson, Jeffrey A. Hooke, Leigh Fantacone-Campbell, Praveen Kumar Raj Kumar, Leonid Kvecher, Jianfang Liu, Jennifer Kane, Jennifer Melley, Stella Somiari, Joji Iida, Stephen C. Benz, Justin Golovato, Shahrooz Rabizadeh, Patrick Soon-Shiong, Richard D. Smith, Richard J. Mural, Craig D. Shriver, Hai Hu, Karin D. Rodland. Integrated proteogenomic analysis of laser capture microdissected breast tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 213. doi:10.1158/1538-7445.AM2017-213

Molecular Heterogeneity in Primary Breast Carcinomas and Axillary Lymph Node Metastases Assessed by Genomic Fingerprinting Analysis

Molecular heterogeneity within primary breast carcinomas and among axillary lymph node (LN) metastases may impact diagnosis and confound treatment. In this study, we used short tandem repeated sequences to assess genomic heterogeneity and to determine hereditary relationships among primary tumor areas and regional metastases from 30 breast cancer patients. We found that primary carcinomas were genetically heterogeneous and sampling multiple areas was necessary to adequately assess genomic variability. LN metastases appeared to originate at different time periods during disease progression from different sites of the primary tumor and the extent of genomic divergence among regional metastases was associated with a less favorable patient outcome (P = 0.009). In conclusion, metastasis is a complex process influenced by primary tumor heterogeneity and variability in the timing of dissemination. Genomic variation in primary breast tumors and regional metastases may negatively impact clinical diagnostics and contribute to therapeutic resistance.

Abstract P3-07-20: Survival comparative analysis of patients with invasive breast cancer treated by a military medical center and matched patients of the US general population

BACKGROUND

U.S. military beneficiaries differ from the U.S. general population with regards to access to health care as care is provided at no or much lower cost in the military health system. Other differences also exist. Many of these differences are known factors affecting invasive breast cancer outcomes. Thus it is desirable to conduct a comparative analysis of breast cancer patient outcomes between these two populations to find out whether there is any outcome difference, and if yes what the contributing factors are.

METHODS

We compared overall survival (OS), disease-specific survival (DSS), and 5-year OS and DSS rates in breast cancers between 399 patients from the Clinical Breast Care Project at the Walter Reed National Military Medical Center (CBCP-WR) and 1,000 sets of 1596 matched patients from the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute. All patients were diagnosed between 2001 and 2010. Each CBCP-WR patient was randomly matched to four SEER patients on six demographic and clinicopathologic variables (age at diagnosis in 5-year groups, race, diagnosis year, estrogen receptor (ER), progesterone receptor, and AJCC stage).

RESULTS

The CBCP-WR cohort had better survival than the SEER population. At the whole cohort level, the mean hazard ratios (HRs) from 1,000 matched comparisons for OS and DSS were 0.774 and 0.708, with mean log-rank P-values of 0.124 and 0.125. The numbers of 175 and 141 comparisons showing a log-rank P-value &lt;0.05 out of the 1,000 tests were significantly higher than what would be expected from a random distribution of these P-values (P&lt;0.00001, exact binomial test). By stratifying the cohorts we identified that this survival disparity was mainly contributed by patients with a diagnosis age ≥50 years (for DSS, mean HR=0.550, mean P=0.049, and 642 of 1,000 tests showed a P&lt;0.05; for OS, mean HR=0.713, mean P=0.081, and 377 of 1,000 tests showed a P&lt;0.05), but not by patients with a diagnosis age &lt;50 years. The absolute differences in 5-year DSS rates were 4.4% (94.6% in CBCP-WR vs. 90.2% in SEER; mean P=0.010) for all matched patients and 4.8% (95.2% vs. 90.4%; mean P=0.015) for patients diagnosed at an age ≥50 years. Again there was no significant difference for patients diagnosed at an age &lt;50 years. When stratified by race, ER, stage or grade, most of the patient subpopulations showed favorable 5-year OS and DSS rates in the CBCP-WR cohorts.

CONCLUSION

Overall, these results suggested that breast cancer patients, especially older patients seen in the CBCP-WR, carried more favorable outcomes than those from the general population. The findings warrant further analyses of the contributing factors, such as health care access, treatments, population characteristics, additional pathologic characteristics, and socioeconomic statuses, to this outcome disparity.

The views expressed in this article are those of the author and do not reflect the official policy of the Department of Defense, or U.S. Government.

Citation Format: Yuanbin Ru, Jianfang Liu, Jamie Leigh Campbell, Kangmin Zhu, Albert J Kovatich, Jeffrey A Hooke, Leonid Kvecher, Brenda Deyarmin, Audrey W Kovatich, Frank Cammarata, Hallgeir Rui, Richard J Mural, Craig D Shriver, Hai Hu. Survival comparative analysis of patients with invasive breast cancer treated by a military medical center and matched patients of the US general population [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P3-07-20.

Abundance and distribution of polychlorinated biphenyls (PCBs) in breast tissue

Many environmental chemicals accumulate in human tissues and may contribute to cancer risk. Polychlorinated biphenyls (PCBs) are associated with adverse health effects, but relationships between PCB exposure and breast cancer are unclear. In this study, we sought to determine whether bioaccumulation of PCBs differs within regions of the human breast and whether PCB levels are associated with clinical and pathological characteristics in breast cancer patients. Tissue sections (n=245) were collected from breast quadrants from 51 women with a diagnosis ranging from disease-free to metastatic breast cancer. Ninety-seven PCB congeners were assayed by high resolution gas chromatography. ANOVA was used to examine PCB distribution within the breast and relationships with clinical/pathological variables. Pearson product-moment correlations assessed relationships between age at mastectomy and PCB levels. PCBs were abundant in breast tissues with a median concentration of 293.4ng/g lipid (range 15.4-1636.3ng/g). PCB levels in breast tissue were significantly different (p<0.001) among functional groupings of congeners defined by structure-activity properties: Group I (28.2ng/g), Group II (96.6ng/g), Group III (166.0ng/g). Total PCB concentration was highly correlated with age at mastectomy, but the distribution of PCBs did not differ by breast quadrant. PCB levels were not associated with patient status or tumor characteristics. In conclusion, PCB congeners with carcinogenic potential were present at high levels in the human breast, but were not associated with clinical or pathological characteristics in breast cancer patients.

Abstract B49: Molecular characterization of breast tumor-associated adipose

Background: Long thought to function only as an inert energy storage depot, the role of adipose tissue in breast tumorigenesis has been largely ignored. In light of increasing rates of obesity and use of breast conserving therapy and autologous fat grafting, improved understanding of the role of adipose in tumor etiology is crucial.

Methods: Adipose adjacent to and distant from invasive breast tumors (n=20), or adjacent to non-malignant diagnoses (n=20) was laser microdissected from post-menopausal women. Gene expression data were generated using microarrays and data analyzed to identify significant patterns of differential expression between adipose groups, at the individual gene and molecular pathway level.

Results: Pathway analysis revealed significant differences in immune response between non-malignant, distant and tumor adjacent adipose, with the highest response in tumor-adjacent and lowest in non-malignant adipose. Adipose from invasive breasts exhibits increased expression in anti-inflammatory genes, such as MARCO and VSIG4, while genes differentially expressed between tumor-adjacent and distant adipose such as SPP1, RRM2 and MMP9, are associated with increased cellular proliferation, invasion, and angiogenesis.

Conclusions: Gene expression levels differ in breast adipose, depending on presence of or proximity to tumor cells. Heightened immunotolerance in adipose from invasive breasts provides a microenvironment favorable to tumorigenesis. In addition, tumor-adjacent adipose demonstrates expression of genes associated with tumor growth and progression. Together, these data suggest that adipose is not an inert component of the breast microenvironment but plays an active role in tumorigenesis.

Citation Format: William C. Yarina, Lori A. Field, Brenda Deyarmin, Ryan van Laar, Jeffrey A. Hooke, Craig D. Shriver, Rachel E. Ellsworth. Molecular characterization of breast tumor-associated adipose. [abstract]. In: Abstracts: AACR Special Conference on Cellular Heterogeneity in the Tumor Microenvironment; 2014 Feb 26-Mar 1; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(1 Suppl):Abstract nr B49. doi:10.1158/1538-7445.CHTME14-B49

Gene expression differences in adipose tissue associated with breast tumorigenesis

Long thought to function only as an inert energy storage depot, the role of adipose tissue in breast tumorigenesis has been largely ignored. In light of increasing rates of obesity and use of breast conserving therapy and autologous fat grafting, improved understanding of the role of adipose tissue in tumor etiology is crucial. Thus, adipose tissue adjacent to and distant from invasive breast tumors (n = 20), or adjacent to non-malignant diagnoses (n = 20) was laser microdissected from post-menopausal women. Gene expression data were generated using microarrays and data analyzed to identify significant patterns of differential expression between adipose tissue groups at the individual gene and molecular pathway level. Pathway analysis revealed significant differences in immune response between non-malignant, distant, and tumor-adjacent adipose tissue, with the highest response in tumor-adjacent and lowest in non-malignant adipose tissue. Adipose tissue from invasive breasts exhibits increased expression of anti-inflammatory genes such as MARCO and VSIG4 while genes differentially expressed between tumor-adjacent and distant adipose tissue such as SPP1, RRM2, and MMP9, are associated with increased cellular proliferation, invasion, and angiogenesis. These data suggest that molecular profiles of adipose tissue differ depending on presence of or proximity to tumor cells. Heightened immunotolerance in adipose tissue from invasive breasts provides a microenvironment favorable to tumorigenesis. In addition, tumor-adjacent adipose tissue demonstrates expression of genes associated with tumor growth and progression. Thus, adipose tissue is not an inert component of the breast microenvironment but plays an active role in tumorigenesis.

Abstract P4-06-04: Effect of genomic heterogeneity on breast cancer progression and metastatic spread

Purpose: DNA fingerprinting has revealed discordant patterns of chromosomal alterations in primary breast tumors (PBT) compared to matched metastatic axillary lymph node tumors (MLNT); however, whether these genetic differences reflect the timing of dissemination of cells with metastatic potential from the primary tumor, intratumoral heterogeneity of the primary breast tumor, or independent seeding of lymph nodes with metastatic cells remains unclear.

Patients and Methods: From the 30 node-positive patients evaluated in this study, allelic imbalance (AI) data was generated using 52 microsatellite markers from 5-19 areas of each PBT as well as from available MLNT. Data were analyzed using pairwise correlations, ANOVA and PHYLIP.

Results: The frequency of genomic changes was significantly higher (P&lt;0.001) in PBT areas (13%) than MLNT (9%). No two PBT areas had identical patterns of AI and the percent concordance (PC) of AI events between tumor regions ranged from 0-65% (average 33%). Of the 196 MLNT from 28 patients, PC between MLNT from the same patient ranged from 0-88% (average = 32%). Neither the overall frequency of AI nor the PC differed significantly between sentinel (SLN) and non-SLNs. Phylogenetic analysis revealed that within patients, many MLNT appeared to be descended from different areas of the PBT, but patterns of descent were complex.

Conclusions: Both PBT and MLNT are characterized by extensive molecular heterogeneity, MLNT appear to originate from different areas of the PBT, and SLN metastases are not genomically more advanced than non-SLN metastases. These data suggest that metastatic dissemination may be influenced by both spatial and temporal factors, with cells with metastatic potential colonizing lymph nodes throughout the development of the PBT, and that SLN metastases do not appear to be a source of metastatic cells for non-SLN but rather MLNT arise by independent colonization.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-06-04.

Tissue imprints: assessing their potential for routine biobanking collection

Biomedical research depends on the availability of good quality biospecimens. Unfortunately, certain specimens are scarce due to disease rarity or size restrictions of surgical materials. To increase access to limited surgical specimens, Biobanks need to reassess and adjust their collection programs. We evaluated the feasibility of adapting "touch imprints" to gain access to limited surgical specimens as well as to maximize the use of "precious" specimens. We utilized 12 kidney samples for touch imprints on defined areas of microscope glass slides and FTA paper. DNA was isolated from glass slides on the day of preparation, Day 0, and from glass slide and FTA paper preparations after two weeks of storage at room temperature and -80°C. Yield and purity of DNA from reference kidney samples were compared to DNA from the touch imprints and the quality determined by real-time PCR using the amplification of Cyclophilin A (Cyc A) as an index. DNA quality for glass slides at Day 0 was not significantly different from DNA after two weeks at room temperature (glass at room temperature; p=0.111 and 0.097, yield and purity, respectively) and after two weeks at -80°C (glass -80°C; p=0.358 and 0.281, yield and purity, respectively). Glass slide DNA at room temperature and -80°C were not significantly different (p=0.795 and 0.146 for yield and purity, respectively). DNA from FTA paper at room temperature and from FTA paper at -80°C were significantly different from glass at room temperature and glass at -80°C (p=0.002, respectively). Threshold values for Cyc A were ≤28 for the reference DNA and ≤32 for DNA from glass and FTA paper. This study demonstrates that touch preparations on microscope glass slides and FTA paper can provide sufficient and good quality DNA suitable for PCR. Touch imprints could therefore be adopted by biobanks to collect and bank biological materials from limited surgical specimens.

Abstract A105: Identification of gene expression profiles associated with different types of breast adipose and their relationship to tumorigenesis

Background: Research over the past decade has shown the importance of the stroma in tumorigenesis, however, having long been thought to function only as an inert energy storage depot, the role of adipose tissue in tumor etiology has been largely ignored. Improved understanding of the role of adipose in tumor development and progression is crucial given the increasing rates of obesity and the use of autologous fat transfer in breast reconstruction.

Methods: Adipose, adjacent to and distant from invasive breast tumors, was laser microdissected from 20 post-menopausal women, and from 20 post-menopausal women with non-malignant breast disease. Gene expression data were generated using U133 2.0 microarrays. After quality control and visualization steps, the data were analyzed to identify significant patterns of differential expression between adipose classes, at the individual gene and molecular pathway level. A subset of genes were further analyzed by qRT-PCR in out-of-sample adipose specimens.

Results: Pathway analysis revealed significant differences in immune response between non-malignant, distant and tumor adjacent adipose. 141 genes were differentially expressed (FDR &lt;0.05, &gt;2-fold difference) between tumor-adjacent and non-malignant breasts including FCGR2A, FOLR2, LGMN and NLRP3. These four genes were also differentially expressed (FDR &lt;0.05, &gt;2-fold difference) between distant and non-malignant adipose. Within invasive breasts, no genes were differentially expressed using FDR &lt;0.05, however, RRM2, PLA2G7, MMP9, MMP12, CHI3L1, SPP1 were expressed at &gt;3-fold higher levels (P&lt;0.05) in tumor-adjacent compared to distant adipose.

Conclusions: Gene expression levels differ in breast adipose, depending on presence of or proximity to tumor cells. Tumor-adjacent and distant adipose from invasive breasts both exhibit increased expression in genes involved in the M2 anti-inflammatory response, suggesting that the microenvironment in an invasive breast has a decreased immune response compared to the non-malignant microenvironment. Genes expressed at higher levels in tumor-adjacent compared to distant adipose are associated with increased cellular proliferation, invasion, migration, angiogenesis and metastasis, suggesting that tumor-adjacent adipose promotes the growth and progression of the tumor. Together, these data suggest that adipose is not an inert component of the breast microenvironment but plays an active role in tumorigenesis.

Citation Format: Lori Field, Brenda Deyarmin, Ryan van Laar, Craig Shriver, Rachel Ellsworth. Identification of gene expression profiles associated with different types of breast adipose and their relationship to tumorigenesis. [abstract]. In: Proceedings of the AACR Special Conference on Advances in Breast Cancer Research: Genetics, Biology, and Clinical Applications; Oct 3-6, 2013; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Res 2013;11(10 Suppl):Abstract nr A105.

Abstract P6-02-08: Molecular drivers of adipogenotoxicosis in breast tumor-associated adipose

Background: Having long been thought to function only as an inert energy storage depot, the role of adipose tissue in tumorigenesis has been largely ignored; however, adipose is an active endocrine organ that can directly influence tumor growth. Improved understanding of the role of adipose in tumorigenesis is crucial given the association between obesity and breast cancer risk in post-menopausal women, increasing rates of obesity and use of autologous fat transfer in breast reconstruction.

Methods: Adipose, adjacent to and distant from (&gt;3 cm from the closest tumor margin) invasive breast tumors, was laser microdissected from 20 post-menopausal women, and from 22 post-menopausal women with non-malignant breast disease. Gene expression data were generated using U133A 2.0 microarrays. Data were analyzed to identify significant patterns of differential expression between adipose classes at the individual gene and molecular pathway level. Gene expression differences were validated using qRT-PCR in an additional set of 29 specimens.

Results: SPP1, RRM2, MMP9 and PLA2G7 were expressed at &gt;3-fold (P &lt; 0.01) higher levels in adjacent adipose compared to distant adipose from the same breast. A number of immune response genes including MARCO, FABP7, ELF5, MYBPC1, MMP7, CLDN8, HLA-DQB1 and HLA-DQA1 were differentially expressed in distant adipose compared to adipose from non-malignant breasts. The most significant gene expression differences were detected between tumor-adjacent and non-malignant adipose with &gt;3-fold higher expression of EGFL6 and ITGB2 and &gt;3-fold lower levels of PIP, which are involved in growth, proliferation, and cellular adhesion in adjacent compared to non-malignant adipose. Pathway analysis revealed that immune response differs between non-malignant, distant and tumor-adjacent adipose with an enhanced B- and T-cell response detected in adjacent compared to distant or non-malignant adipose. Inflammatory response as well as DNA transcription and replication pathways were differentially expressed in distant compared to non-malignant adipose.

Conclusions: Gene expression levels differ in breast adipose depending on presence of and proximity to tumor cells. Adipose adjacent to the tumor demonstrated the largest immune response, supporting the idea of adipogenotoxicosis, which through pro-inflammatory and genotoxic responses, promotes tumor development. In addition, genes involved in cellular proliferation, degradation of the extracellular matrix and angiogenesis are differentially expressed in adjacent compared to distant or non-malignant adipose, thus tumor-adjacent adipose may be contributing to the growth and invasion of the primary tumor. These data thus suggest that adipose is not an inert component of the breast microenvironment but plays an active role in tumorigenesis.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P6-02-08.

Abstract 4266: Identification of gene expression profiles associated with different types of breast adipose and their relationship to tumorigenesis

Background: Research over the past decade has shown the importance of the stroma in tumorigenesis, however, having long been thought to function only as an inert energy storage depot, the role of adipose tissue in tumorigenesis has been largely ignored. Improved understanding of the role of adipose in tumorigenesis is crucial given the increasing rates of obesity and the use of autologous fat transfer in breast reconstruction. Methods: Adipose, adjacent to and distant from invasive breast tumors, was laser microdissected from 20 post-menopausal women, and from 20 post-menopausal women with non-malignant breast disease. Gene expression data were generated using U133 2.0 microarrays. After quality control and visualization steps, the data were analyzed to identify significant patterns of differential expression between adipose classes, at the individual gene and molecular pathway level. Results: Pathway analysis revealed that immune response differs between non-malignant, distant and tumor adjacent adipose; this response is seen as a gradient with the largest response closest to the tumor. Gene expression differed significantly in adipose from invasive compared to non-malignant breasts with FCGR2A, FOLR2, LGMN, MARCO and NLRP3 expressed at significantly higher levels and HLA-DQB1 and HLA-DQA1 at significantly lower levels in adipose from invasive breasts. Within the invasive breasts, MMP9, PLA2G7, RRM2 and SPP1 were expressed at &gt;3-fold higher levels in adjacent compared to distant adipose. Conclusions: Gene expression levels differ in breast adipose, depending on presence of or proximity to tumor cells. Adipose adjacent to the tumor demonstrated the largest immune response; this response may reflect a reaction to surgical insult from the original biopsy; however, response to surgical injury has been associated with increased ability to metastasize. In addition, within breasts with invasive breast cancer, genes involved in cellular proliferation, degradation of the extracellular matrix and angiogenesis were expressed at higher levels in adjacent compared to distant adipose. Together, these data suggest that adipose is not an inert component of the breast microenvironment but plays an active role in tumorigenesis.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4266. doi:1538-7445.AM2012-4266

P3-03-03: Congruence between Patterns of microRNA Expression and Histologic Grading of Invasive Breast Carcinomas

Background

Histologic grading may be used as an indicator of prognosis in breast cancer; patients with low-grade carcinomas have ∼85% ten-year survival compared to just 45% survival in patients with high-grade disease. Although useful for risk stratification, assigning nuclear grade is subjective, and a large proportion of carcinomas are classified as intermediate-grade with uncertain prognosis, thus limiting clinical utility. MicroRNAs (miRNAs) regulate gene expression and serve an important role in breast cancer development. In this study we examined miRNA expression profiles in low-grade and high-grade breast carcinomas to determine if miRNA expression is associated with pathological classifications of tumor grade.

Methods: Breast tumors were obtained from 69 patients enrolled in the Clinical Breast Care Project. Samples were partitioned into low-grade (n=30) or high-grade (n=39) categories using the Nottingham Histologic Score. Following laser microdissection of frozen tissue sections, miRNA was isolated from pure populations of breast tumor cells and hybridized to Affymetrix GeneChip® miRNA arrays containing over 800 human miRNA probes. Expression profiles were analyzed with Partek Genomics Suite using the miRNA Expression Module.

Results: We identified 30 unique miRNAs that showed differential expression at a False Discovery Rate (FDR) p&lt;0.05 between low-grade and high-grade breast carcinomas. Gene targets for these miRNAs function primarily in metabolism and cell communication. Expression of hsa-miR-18a and hsa-miR-572 was significantly different between histologic grades at an FDR p&lt;1×10−8 and hierarchical clustering based on these miRNAs correctly classified 97% (29/30) of low-grade and 90% (35/39) of high-grade tumors. miR-18a has been shown to inhibit ER signaling and promote cellular differentiation, while the role of miR-572 in breast carcinogenesis is not well known.

Conclusions: Dysregulation of miRNAs may accompany changes in cellular morphology typically used in histologic classification of breast carcinomas. Patterns of miRNA expression may improve reproducibility and clinical utility of tumor grading and may prove useful for prediction of recurrence and survival for patients with intermediate-grade carcinomas.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P3-03-03.

P3-06-06: Comparison of Gene Expression Profiles of Lymph Node Positive and Lymph Node Negative ER Positive Breast Tumors in Pre- and Postmenopausal Women

Background

Breast cancer is the most common female cancer in US and is the second leading cause of cancer related death in women. Metastases are the primary cause of cancer morbidity and mortality. Axillary lymph node (LN) status has long been used as a prognostic factor for breast cancer. The molecular mechanisms that control LN metastasis remains poorly understood. To better understand the various genes and regulatory pathways that drive breast cancer LN metastasis, we compared the gene expression profiles between breast tumors that have metastasized to the LNs and those which have not in pre- and postmenopausal women.

Material and Methods: Tumor cells were isolated from the primary tumors (ER+) of postmenopausal node positive (PMNP; N=20), postmenopausal node negative (PMNN; N=19), premenopausal node positive (PRNP; N=18) and premenopausal node negative (PRNN; N=16) women using laser capture microdissection. RNA was isolated using the RNAqueous®-micro kit (Ambion, Austin, TX). Total RNA was converted to Biotin-labelled aRNA using two rounds of amplification with MessageAmp II aRNA amplification kit (Applied Biosystems, Foster City, CA). The aRNA concentration was determined by Nanodrop 1000 and the quality was assessed with a Bioanalyzer. The aRNA was fragmented and hybridized to Human Genome U133 Plus 2.0 GeneChip (Affymetrix, Santa Clara, CA). Microarray raw data were analyzed using a variety of R programming packages for probe density processing, background correction, normalization, quality control/quality assessment, and calculation of gene expression value, etc. To identify differentially expressed genes, Wilcoxon rank sum test with FDR (false discovery rate) control was performed for pair-wise comparison between different groups. Functional analyses were performed on the identified statistically significant differentially expressed genes to search for the functional categories and pathways in which they are involved and further understand their potential roles in breast cancer metastatic process.

Results: Multivariate data mining (hierarchical clustering analysis and principal component analysis, etc) revealed that in postmenopausal women, the node positive and node negative women are well separated while this was not the case in premenopausal women. Further analysis of the PMNN and PMNP groups to identify differentially expressed genes (with at least a 1.5 fold difference) at FDR =0.1 showed that 232 genes were upregulated and 470 genes were downregulated in PMNP vs PMNN groups. Gene function analysis revealed that genes down regulated in the PMNP group compared to PMNN are related to extracellular matrix, cell adhesion, EGF-like pathway, cytoskeleton etc, while the over-expressed genes are related to cell cycle and cell division, chromosome condensation, etc.

Discussion: The ability to differentiate lymph node positive cases from lymph node negative cases in ER+ breast cancer based on transcriptional profiling may have an impact on the clinical management of ER+ breast cancer cases. Having transcriptional profiles that identify ER+ tumors likely to have poor outcomes would suggest more aggressive treatment for such patients.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P3-06-06.

Identification of differentially expressed genes in breast tumors from African American compared with Caucasian women

BACKGROUND

Breast tumors from African American women have less favorable pathological characteristics and higher mortality rates than those of Caucasian women. Although socioeconomic status may influence prognosis, biological factors are also likely to contribute to tumor behavior.

METHODS

Patients with invasive breast cancer were matched by age, grade, and estrogen receptor status; patients with benign disease were matched by age and diagnosis type. RNA from laser microdissected tumors and whole-sectioned nonmalignant breast tissues was hybridized to HG U133A 2.0 microarrays. Data were analyzed using Partek Genomics Suite using a cutoff of P < .001, >1.5-fold change, and results were validated by quantitative real-time polymerase chain reaction.

RESULTS

Clinicopathological factors did not differ significantly between groups for age at diagnosis, tumor size or stage, lymph node or human epidermal growth receptor 2 status, intrinsic subtype, or mortality. Two-way analysis of the tumor specimens revealed 25 probes representing 23 genes differentially expressed between populations; hierarchical clustering classified 24 of 26 African American women and 25 of 26 Caucasian women correctly. In the nonmalignant specimens, 15 probes representing 13 genes were differentially expressed, including 5 genes that also differed in the tumor specimens; these genes were able to correctly classify nonmalignant breast specimens from 20 of 22 of African American women and all of the Caucasian women.

CONCLUSIONS

Despite matching of tumors by pathological characteristics, molecular profiles differed between African American women and Caucasian women in both invasive tumors and benign breast tissues. These differentially expressed genes, including CRYBB2, PSPHL, and SOS1, are involved in cellular growth and differentiation, invasion, metastasis, and immune response and thus may contribute to the poor outcome in African American women.

Abstract 3741: Identification of differentially expressed genes in breast tumors from African American compared with Caucasian women

Background: Breast tumors from African American women (AAW) have less favorable pathological characteristics and higher mortality rates than Caucasian women. While socioeconomic status may influence prognosis, biological factors are also likely to contribute to tumor behavior.

Methods: Patients treated within an equal-access military medical center were matched by age, grade and ER status. Tumors were subjected to laser microdissection and RNA hybridized to HG U133A 2.0 microarrays. Data was analyzed using Partek Genomics Suite using a cutoff of P&lt;0.001, 1.5-fold change. Microarray results were validated by qRT-PCR.

Results: Twenty-six pairs of matched tumor samples were identified. Clinico-pathological factors did not differ significantly for age at diagnosis, tumor size or stage, lymph node, hormone receptor, or HER2 status. Mortality was similar between the two groups. One putative and nineteen known genes were differentially expressed between populations with five genes expressed at higher and 15 genes at lower levels in tumors from AAW; hierarchical clustering classified 23/26 AAW and 26/26 CW correctly. Differential expression was validated (P&lt;0.001) for PSPHL (higher in AAW) and LTF and PSD3 (lower in AAW).

Conclusions: Despite matching of tumors by pathological characteristics and provision of medical care in an equal-access health care system, a molecular signature differentiating AAW from CW tumors was detected, supporting a model in which aggressive tumor phenotypes are driven by molecular differences within the tumors. These differentially expressed genes are involved in immune response, cell cycle and metastasis and thus may contribute to the poor outcome in AAW.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 3741. doi:10.1158/1538-7445.AM2011-3741

Laser microdissection for gene expression profiling

Microarray-based gene expression profiling is revolutionizing biomedical research by allowing expression profiles of thousands of genes to be interrogated in a single experiment. In cancer research, the use of laser microdissection (LM) to isolate RNA from tissues provides the ability to accurately identify molecular profiles from different cell types that comprise the tumor and its surrounding microenvironment. Because RNA is an unstable molecule, the quality of RNA extracted from tissues can be affected by sample preparation and processing. Thus, special protocols have been developed to isolate research-quality RNA after LM. This chapter provides detailed descriptions of protocols used to generate micro-array data from high-quality frozen breast tissue specimens, as well as challenges associated with formalin-fixed paraffin-embedded specimens.

Abstract P4-07-09: The Impact of Matrix Metalloproteinase-1 Promoter 1G/2G Polymorphism on Breast Diseases

Background: Matrix Metalloproteinase-1 (MMP-1) is a ubiquitously expressed interstitial collagenase. Overexpression of MMP-1 has a role in initiating mammary tumorigenesis by degrading stroma and by releasing growth factors. A single guanine insertion polymorphism in the MMP-1 promoter creates the binding site, 5'-GGAA-3', for the Ets transcription factor, and increases transcription of MMP-1. The MMP-1 2G polymorphism is linked to early onset, increased risk or aggressiveness of several cancers. Its relationship with other potential markers in invasion and metastasis of breast cancer is unknown.

Experimental Design: To study the impact of the 2G polymorphism on breast cancer we analyzed the genotypes of 109 patients (52 invasive breast cancer [IBC], 29 ductal carcinoma in situ [DCIS], 13 atypical ductal hyperplasia [ADH] and 15 benign breast disease). Immunohistochemical (IHC) data for MMP-1, HER2, ER/PR and P53 from these donors were also analyzed. IHC results for MMP-1 were scored as 0 (no expression) or increasing expression of+1, +2 or +3. Data were analyzed using Pearson's chi-square test to identify statistical significance.

Results: A significantly higher number of patients in the IBC group expressed high MMP-1 (+2 and +3; p &lt;0.001) while the benign group had the least number of patients expressing higher MMP-1 (score +3; p = 0.0075). In the IBC group, among patients with low levels of MMP-1 (+1), 57% had the 1G/1G phenotype, and among those expressing high levels of MMP-1 (+2 and +3), over 70% were 1G/2G heterozygotes or 2G/2G homozygotes. The 2G allele frequency in the ADH group was 0.62 and these patients had higher MMP-1 expression (+2 and +3). Further analyses of HER2, ER/PR and P53 in relation to the MMP-1 polymorphism within the IBC group showed MMP-1 allelic variation in Her-2 positive group was significantly different compared with Her-2 negative group (p = 0.039), with a distribution curve shifted to a greater frequency of 2G homozygosity. A similar result was also observed in P53 positive group when compared with P53 negative group (p = 0.043).

Conclusions: 1) In the IBC group, the 2G insertion polymorphism contributes to MMP-1 over expression. 2) Increased expression of MMP-1 in ADH and higher 2G allele frequency are consistent with the hypothesis that increased MMP-1 2G polymorphism plays a role in initiation of ADH through up regulation of MMP-1 expression. 3) Earlier studies show prognostic role for the coexistence of increased expression of HER2 and P53 in breast cancer. Our observation of a significant increase in the 2G homozygotes in HER2 and P53 positive patients supports a prognostic role for this polymorphism and suggests its possible association with other breast cancer markers. Thus, the MMP-1 2G polymorphism may both contribute to breast disease onset and serve as a prognostic marker for breast cancer.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P4-07-09.

Abstract P6-04-10: Genetic Discrimination of Aggressive from Indolent DCIS

Background: Treatment options for DCIS vary from surgical excision with or without radiation and/or chemopreventive therapy, or mastectomy. Intuitively, more aggressive treatment options should lead to improved survival rates, however, studies have shown no difference in breast cancer mortality between women treated with wide excision only versus those with excision plus radiation and treatments can be costly, lengthy and associated with side effects. To avoid over-treating women with indolent disease, while intensively treating women with aggressive disease, new molecular tools must be developed to supplement pathological information to classify DCIS lesions and predict clinical outcome.

Methods: Formalin-fixed paraffin-embedded (FFPE) pure DCIS biopsy specimens were collected from the pathology archives of the Anne Arundel Medical Center. Samples included those with poor prognosis characterized by either recurrence of DCIS or progression to invasive cancer (n=7) and those good prognosis, having ≥5-year disease-free survival (n=10). RNA was isolated after laser-microdissection of pure tumor cells and hybridized to Breast Cancer DSA™ microarrays (Almac Diagnostics). S-way ANOVA was used to account for batch effects and then Support Vector Machine (SVM) was used to identify candidate genes effective at discriminating good from poor prognosis DCIS. Pathway analysis was performed using MetaCore (GeneGeo).

Results: 328 genes were found to be differentially expressed between good and poor prognosis specimens (P&lt;0.01). Preliminary analysis with SVM found that a 70-gene candidate signature from these 328 genes wasoptimal under the tested conditions for discriminating favorable from poor prognosis DCIS. This candidate signature included genes such as MEF2C, PTK2 and ZBTB2. Pathway analysis revealed that genes involved in cytoskeleton modeling, apoptosis and survival, DNA damage repair and cell adhesion are expressed at lower levels in poor prognosis DCIS while those involved in cell cycle, immune response and cell proliferation are expressed at higher levels.

Conclusions: While studies have attempted to identify molecular profiles associated with aggressive DCIS by comparing DCIS co-occurring with invasive disease to pure DCIS, to our knowledge, this is the first study that identified a candidate molecular signature of prognosis in pure DCIS. Although many of the 70 genes found to differ between favorable and poor prognosis DCIS have not been previously associated with breast cancer or have unknown function, MEF2C and PTK2 have been implicated in invasion and migration, while ZBTB2 is a master regulator of p53 and stimulates cellular proliferation. These data demonstrate aggressive DCIS do differ from indolent DCIS at the genetic level and that these differences may be useful in developing molecular tools to classify DCIS lesions and guide appropriate treatment.

Citation Information: Cancer Res 2010;70(24 Suppl):Abstract nr P6-04-10.

Abstract 2971: Genomic instability of breast stroma

Background: Research in the past decade has demonstrated the importance of the tumor microenvironment in breast cancer development and progression. Allelic imbalance evaluation of tumor and stromal components suggests a limited number of alterations in the tumor may lead to cellular transformation while tumor behavior may be driven by molecular changes in the surrounding stroma. Recent studies evaluating copy number changes in frozen stromal tissue, however, found no chromosomal alterations and suggested that earlier studies represent artifacts from the use of archive tissues.

Methods: High-quality flash-frozen specimens were collected from breast quadrants from 14 patients undergoing mastectomy; patient characteristics ranged from prophylactic mastectomy to stage IV breast cancer. DNA was isolated (Gentra) from benign tissue sections and hybridized to GeneChip Human Mapping 250K arrays (Affymetrix). Genomic DNA from peripheral blood for each patient was genotyped to assess copy number variation (CNV) by paired analysis. CNVs were identified using Genome Console 3.0.2 (Affymetrix) with a minimum genomic size/fragment of 500 Kb and a minimum of 50 markers/segment.

Results: Copy number changes ranging in size from 586-5061 kb were detected in 50% (7/14) patients in at least one quadrant, with a single patient having copy number changes present in each quadrant. 21% of quadrants (12/57) showed detectable copy number changes affecting 1 - 18 chromosomal arms. Copy number changes were detected not only in stroma from diseased breasts, but in prophylactic mastectomies, as well. Copy number changes were observed in all quadrants irrespective of tumor location. While copy number alterations on chromosomes 1q21, 2p21, 12p13, 14q12 and 18q12 were shared between patients, no patient had the same patterns of CNVs across all quadrants.

Conclusions: Chromosomal alterations were detected in high-quality frozen breast stroma specimens utilizing genome-wide SNP array data, supporting earlier claims that breast stroma may be genetically unstable. Half of the patients in this study demonstrated copy number alterations in breast stroma distant from the tumor, suggesting that for a subgroup of breast cancer patients, widespread chromosomal instability may increase risk of recurrence.

Note: This abstract was not presented at the AACR 101st Annual Meeting 2010 because the presenter was unable to attend.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 2971.

Genetic Characterization of Columnar Cellular Lesions and Atypical Ductal Hyperplasia of the Breast

Aims: Columnar cell lesions (CCL) and atypical ductal hyperplasia (ADH) frequently coexist and share molecular changes with in situ and invasive components, suggesting that CCL and ADH may be precursors to breast cancer. These conclusions are, however, largely based on studies examining CCL and/or ADH from patients diagnosed with more advanced disease. Thus, we assessed allelic imbalance (AI) in pure CCL or ADH specimens to characterize molecular changes in early breast lesions.Methods and results: DNA samples were obtained from laser microdissected lesions from CCL with (n=42) or with ADH (n=31) without concurrent in situ or invasive disease. AI was assessed at 26 chromosomal regions commonly altered in breast cancer. The average AI frequency was 6.2% (range 0-20%) in CCL and 6.1% (range 0-25%) in ADH with no significant difference in levels of AI between CCL and ADH. The highest levels of AI were on chromosomes 8q24 and 17q21 in ADH (23%) and CCL (15%), respectively.Conclusions: In conjunction with low overall levels of AI, chromosomal alterations that characterize low- and high-grade in situ and invasive disease were not frequent in pure CCL and ADH. Thus, pure lesions are not genetically advanced and are molecularly distinct from synchronous lesions.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 5156.

Proteomic Discrimination of Well- from Poorly-Differentiated Breast Carcinomas

Background: Pathological grade is a useful prognostic factor for stratifying breast cancer patients into favorable (well-differentiated tumors) and less favorable (poorly-differentiated tumors) outcome groups. The current system of tumor grading, however, is highly subjective and a large proportion of tumors are characterized as intermediate-grade, making determination of optimal treatments difficult.Methods: Primary breast tumor specimens from patients diagnosed with well- (n=27) and poorly-differentiated (n=51) invasive ductal carcinoma were obtained from patients enrolled in the Clinical Breast Care Project. Frozen tissues were sectioned and mounted on gold coated MALDI target plates for protein expression profiling. Hematoxylin and eosin (H&E) stained slides were prepared from serial sections for histological characterization. MALDI matrix was deposited as individual spots on the tissue sections in a histology directed manner to assay specific areas and tissue types of interest. Mass spectral data were then acquired from multiple sites across each tissue section.Results: 129 features were observed in well-differentiated and 132 in poorly-differentiated tumors. While the majority of features detected were similar between the two groups, 6 protein features were expressed at significantly lower and 12 at significantly higher levels in the poorly-differentiated tumors, including increased expression of Calgranulin A and Calgizzarin.Conclusions: Protein expression differences detected here suggest that well- and poorly-differentiated invasive breast tumors are molecularly distinct diseases and that these protein changes may contribute to the structural integrity of the tumor cell. In particular, calgranulin A and calgizzarin are members of the S100 protein family, and function in processes such as cell proliferation and differentiation. Further refinement of this differentiation protein signature may not only improve our understanding of the biological processes involved with tumor grade but provide pathologists with new molecular tools to classify breast tumors and reduce the subjectivity associated with current grading criteria.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 6126.

Fingerprinting genomic heterogeneity in primary breast carcinomas and among sentinel lymph node metastases: implications for clinical management of breast cancer patients

Abstract #1059

Background: Sentinel lymph node (SLN) status is a key prognostic factor for breast cancer patients. The SLN hypothesis postulates that cancer cells initially spread from the primary tumor to the first-draining (sentinel) node(s) before reaching higher echelon nodes. Although SLN biopsy is clinically useful for staging patients, little is known about the genomic heritage of cells that define the genetic makeup and clinical behavior of nodal metastases. We used allelic imbalance (AI) coupled with high-density SNP genotyping to examine the extent of genetic heterogeneity in primary tumors and relationships among metastases in sentinel and non-sentinel axillary lymph nodes in 30 patients with node positive breast cancer.&#x2028; Material and Methods: Pathologically positive nodes were identified by H&E/IHC, while sentinel nodes were localized by standard scintigraphic techniques. For each patient, nodal metastases and 6-15 areas from multiple paraffin blocks of the primary carcinoma were isolated by microdissection. AI was assessed by microsatellite typing at 26 chromosomal regions; genome-wide copy number variation was examined using Affymetrix GeneChip Human Mapping 500K arrays and the Genotyping Consol. The genomic heritage of sentinel and non-sentinel nodal metastases in relation to multiple areas of the primary tumor was assessed by hierarchical clustering and phylogenetic analyses.&#x2028; Results: Extensive genomic heterogeneity was observed in primary tumors and among nodal metastases (0-44% in primaries, 0-36% in metastases). Overall levels of variation were significantly higher (P&lt;0.01) in primary tumors (17%) than metastases (10%), but did not differ between sentinel (10.2%) and non-sentinel (10.0%) metastases. Within patients, many (but not all) metastases appeared to be descended from different areas of the primary tumor, but patterns of descent were complex. As a group, sentinel node metastases were not genomically distinct from other axillary node metastases, and there was no evidence that SLN metastases are seeded by cells that colonize the axilla early in tumorigenesis.&#x2028; Discussion: Although metastases develop in a sequential manner that maintains the known capability of SLN biopsy to accurately stage the axilla, metastases appear to develop from genomically-diverse progenitor cells descended from different areas of the primary tumor. Genomic diversity in primary tumors and metastases may be an important factor in prognosis and treatment resistance in breast cancer patients.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 1059.

Role of phosphoserine phosphatase-like in breast disease in African American women

Abstract #2095

Background: Although breast cancer incidence is higher among Caucasian women (CW), mortality is higher in African American women (AAW). AAW also tend to be diagnosed at a younger age and with larger and more aggressive tumors. Previously, we have found that gene expression patterns in AAW and CW differ in both breast tissues from women without breast disease as well as those from women with invasive breast cancer. Characterization of these differentially expressed genes is critical to understanding breast pathogenesis in AAW.&#x2028; Methods: Total RNA was isolated from breast tissue from 29 AAW and 26 CW without evidence of breast disease using the RNeasy Lipid Tissue Midi kit (Qiagen) and from laser microdissected primary breast tumors from 23 AAW and 24 CW using the RNAqueous Micro kit (Ambion). Based on earlier microarray results, genes were selected for validation by quantitative real-time PCR using TaqMan Gene Expression Assays (Applied Biosystems). All TaqMan assays were performed on duplicate samples using the iCycler Real-Time Detection System (Bio-Rad). Relative expression levels were determined using the Comparative Ct Method. A Wilcoxon test was used to determine significance.&#x2028; Results: Among those genes that were confirmed to be differentially expressed in breast tissues from AAW and CW was phosphoserine phosphatase-like, or PSPHL. PSPHL showed significantly higher expression in both disease-free samples as well as primary tumors from AAW when compared to the same sample type from CW. In tissues from disease-free women, the relative level of PSPHL expression was significantly higher in AAW (median = 17.15) compared to CW (median = 0; P&lt;0.000001). Similarly, primary tumors from AAW also had higher levels of PSPHL expression (median = 42.22) compared to those from CW (median = 0; P&lt;0.0005).&#x2028; Discussion: The current study has confirmed that expression of PSPHL is significantly higher in both disease-free and invasive breast tissues from AAW than those from CW. Recently, PSPHL has also been shown to have higher expression in prostate tumors from African American men than those from Caucasian men. The function of PSPHL is unknown; however, its increased expression has been associated with aberrant cellular proliferation and cell cycle progression in Fanconi anemia. Higher expression of PSPHL in both normal and diseased breast tissues may, therefore, increase cellular proliferation, contributing to the initial development and/or aggressiveness of breast tumors in AAW.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2095.

Amplification of HER2 is a marker for global genomic instability

Background

Genomic alterations of the proto-oncogene c-erbB-2 (HER-2/neu) are associated with aggressive behavior and poor prognosis in patients with breast cancer. The variable clinical outcomes seen in patients with similar HER2 status, given similar treatments, suggests that the effects of amplification of HER2 can be influenced by other genetic changes. To assess the broader genomic implications of structural changes at the HER2 locus, we investigated relationships between genomic instability and HER2 status in patients with invasive breast cancer.

Methods

HER2 status was determined using the PathVysion^® assay. DNA was extracted after laser microdissection from the 181 paraffin-embedded HER2 amplified (n = 39) or HER2 negative (n = 142) tumor specimens with sufficient tumor available to perform molecular analysis. Allelic imbalance (AI) was assessed using a panel of microsatellite markers representing 26 chromosomal regions commonly altered in breast cancer. Student t-tests and partial correlations were used to investigate relationships between genomic instability and HER2 status.

Results

The frequency of AI was significantly higher (P < 0.005) in HER2 amplified (27%) compared to HER2 negative tumors (19%). Samples with HER2 amplification showed significantly higher levels of AI (P < 0.05) at chromosomes 11q23, 16q22-q24 and 18q21. Partial correlations including ER status and tumor grade supported associations between HER2 status and alterations at 11q13.1, 16q22-q24 and 18q21.

Conclusion

The poor prognosis associated with HER2 amplification may be attributed to global genomic instability as cells with high frequencies of chromosomal alterations have been associated with increased cellular proliferation and aggressive behavior. In addition, high levels of DNA damage may render tumor cells refractory to treatment. In addition, specific alterations at chromosomes 11q13, 16q22-q24, and 18q21, all of which have been associated with aggressive tumor behavior, may serve as genetic modifiers to HER2 amplification. These data not only improve our understanding of HER in breast pathogenesis but may allow more accurate risk profiles and better treatment options to be developed.

Genomic heritage of sentinel lymph node metastases: Implications for clinical management of breast cancer patients

571

Background: Sentinel lymph node (SLN) biopsy status is a key prognostic factor for breast cancer patients. Sentinel nodes are believed to receive early disseminating cells from the primary tumor, but little is known about the origin of metastases colonizing the sentinel nodes. We used allelic imbalance (AI) to examine genomic relationships among metastases in the sentinel and non-sentinel axillary lymph nodes from complete axillary dissections in 15 patients with lymph node positive breast cancer. Methods: Sentinel nodes were localized by standard scintigraphic and gamma probe techniques using 1.0 mCi technetium-99m sulfur colloid. Pathologically positive nodes were identified by H&E histology and immunohistochemistry. Primary breast tumors and metastases in sentinel and axillary nodes were isolated by laser microdissection. AI was assessed at 26 chromosomal regions and used to examine the timing and molecular mechanisms of metastatic spread to the sentinel and axillary nodes. Results: Overall AI frequencies were significantly higher (p<0.05) in primary breast tumors compared to lymph node metastases. A high level of discordance was observed in patterns and frequencies of AI events between metastases in the sentinel and non-sentinel axillary nodes. Phylogenetic analyses showed that 1) multiple genetically-divergent lineages of metastatic cells independently colonize the lymph nodes; 2) some lymph node metastases appeared to acquire metastatic potential early in tumorigenesis, while other metastases evolved later; and 3) importantly, lineages colonizing the sentinel nodes appeared to originate at different times and to progress by different molecular mechanisms. Conclusions: Genomic diversity and timing of metastatic nodal spread may be important factors in determining outcomes of breast cancer patients. Metastases colonizing the sentinel nodes appear to arise at different times during disease progression and may not be descendants of progenitor cells that colonize the lymph nodes early in tumorigenesis. Metastatic growth in the sentinel nodes thus may be a consequence of stimulating factors from the primary tumor that affect proliferation of previously disseminated cells rather than the timing of metastatic spread.

No significant financial relationships to disclose.

Clinical implications of the molecular relationship between HER2 and BRCA1

10536

Background: While HER2 status is used prognostically, the effects of HER2 can be influenced by other genes; co- amplification of TOP2A modulates the effectiveness of treatment with Herceptin. BRCA1, deleted in >50% of sporadic breast tumors, is located distal to HER2 on chromosome 17q and is marked by structural instability. To determine how alterations at the BRCA1 locus may modify amplification of HER2, FISH analysis of BRCA1 was performed in breast tumors with HER2 amplification and/or allelic imbalance on chromosome 17q12. Methods: HER2 status (n=68) was determined using the PathVysion HER2 kit. For BRCA1 analysis, DNA probes were generated by nick translation from a BAC clone containing the entire BRCA1 gene in conjunction with a CEP 17 probe. Copy number gains and deletions were defined as BRCA1:CEP17 >1.3 and <0.8, respectively; aneusomy was defined when >10% of tumor cells had copy numbers > (polysomy) or < (monosomy) than two. Results: Amplification of the HER2 gene was detected in 57% of tumors; an additional 16% of tumor were polysomic for chromosome 17q. The remaining 27% of tumors had previously detected allelic imbalance at the 17q12 region. The majority of samples (47%) were characterized by an amplification of HER2 with a downstream deletion of BRCA1, while HER2 and BRCA1 were amplified concomitantly in 21% of tumors, including those with polysomy of 17q. Fifteen percent of samples were monosomic for chromosome 17q; using current clinical scoring criteria, each of those samples were reported as having, despite the altered copy number, negative HER2 status. Levels of overall genomic instability, which were previously measured at 26 chromosomal regions, were significantly higher (P<0.05) in tumors with deletion of BRCA1 compared to those without loss of BRCA1 (38% and 24%, respectively). Conclusions: Alterations, especially deletions, of BRCA1 are highly correlated with copy number changes in HER2. Because deletion of BRCA1 has been associated with overall genomic instability, patients with amplification of HER2 and deletion of BRCA1 may prove unresponsive treatment with Herceptin. Loss of BRCA1 should, therefore, be considered as a genetic modifier to HER2 amplification and the status of BRCA1 should be considered in selecting treatment options for patients with HER2 amplified tumors.

No significant financial relationships to disclose.

Comparison of genomic changes between tumors detected by screening mammography versus other methods

10033

Background: The prognosis in patients with breast cancer detected by screening mammography has been found to be significantly better than those with similar-sized tumors detected by other methods (e.g. physical examination). (Joensuu, et al. 2004). Given the similar pathological characteristics, outcome differences may be explained by molecular differences within the tumors. To address this question, the relationship between genomic instability and methods of detection was assessed using allelic imbalance (AI) in similarly sized primary breast tumors from patients with breast cancer identified by screening mammography versus other methods. Methods: Patients with stage I breast cancer identified by screening mammography (n=50) versus other methods (n=50) were selected. Samples were matched between the two groups for clinical information. DNA was extracted from pure populations of tumor cells obtained by laser microdissection. Referent DNA was obtained from blood or disease-free skin samples from each patient. A panel of 52 microsatellite markers from 26 chromosomal regions commonly altered in breast cancer was used to assess AI, determined by the peak height ratio (≤0.35) between alleles from tumor and referent samples. Fisher’s exact test and Student’s t-tests were used to test for significant differences in levels of AI. Results: When clinical parameters such as age, ethnicity, menopause and status of tumor hormone receptors or HER2 were examined between tumor groups, no significant differences were detected. The overall frequency of AI between tumors detected by screening mammography (21.8%) versus other methods (22.3%) was similar (P=0.93). When patterns of AI were examined, none of the 26 chromosomal regions detected here showed a significant difference in AI events between the two groups. Conclusions: In this study, we have shown that global chromosomal differences are likely not the cause for the difference in tumor detection and prognosis between patients whose tumors were detected mammographically versus those whose were not. Prognostic factors known to correlate with disease aggressiveness and recurrence (e.g. hormone receptor or HER2 status) were also shown to not be associated with this difference in tumor detection among our patients.

No significant financial relationships to disclose.

Timing of Critical Genetic Changes in Human Breast Disease

BACKGROUND

Breast cancer development has been characterized as a nonobligatory sequence of histological changes from normal epithelium through invasive malignancy. Although genetic alterations are thought to accumulate stochastically during tumorigenesis, little is known about the timing of critical mutations. This study examined allelic imbalance (AI) in tissue samples representing a continuum of breast cancer development to examine the evolution of genomic instability.

METHODS

Laser-microdissected DNA samples were collected from histologically normal breast specimens (n = 25), atypical ductal hyperplasia (ADH, n = 16), ductal carcinoma-in-situ (DCIS, n = 37), and stage I to III invasive carcinomas (n = 72). Fifty-two microsatellite markers representing 26 chromosomal regions commonly deleted in breast cancer were used to assess patterns of AI.

RESULTS

AI frequencies were <5% in histologically normal and ADH specimens, 20% in DCIS lesions, and approximately 25% in invasive tumors. Mann-Whitney tests showed (1) that levels of AI in ADH samples did not differ significantly from those in histologically normal tissues and (2) that AI frequencies in DCIS lesions were not significantly different from those in invasive carcinomas. ADH and DCIS samples, however, differed significantly (P < .0001).

CONCLUSIONS

DCIS lesions contain levels of genomic instability that are characteristic of advanced invasive tumors, and this suggests that the biology of a developing carcinoma may already be predetermined by the in situ stage. Observations that levels of AI in ADH lesions are similar to those in disease-free tissues provide a genomic rationale for why prevention strategies at the ADH level are successful and why cases with ADH involving surgical margins do not require further resection.

Allelic imbalance in primary breast carcinomas and metastatic tumors of the axillary lymph nodes

Axillary lymph node status is the most important prognostic factor in predicting disease outcome in women with breast cancer. A number of chromosomal aberrations in primary breast tumors have been correlated with lymph node status and clinical outcome, but chromosomal changes particular to metastatic lymph node tumors have not been well studied. DNA samples isolated from laser-microdissected primary breast and metastatic axillary lymph node tumors from 25 women with invasive breast cancer were amplified using 52 microsatellite markers defining 26 chromosomal regions commonly deleted in breast cancer. Levels and patterns of allelic imbalance (AI) within and between breast and lymph node tumors were assessed to identify chromosomal alterations unique to primary or metastatic tumors and to examine the timing of metastatic potential. The overall frequency of AI in primary breast tumors (0.24) was significantly greater (P < 0.001) than that in lymph node tumors (0.10), and congruent AI events were observed for < 20% of informative markers. AI at chromosomes 11q23.3 and 17p13.3 occurred significantly more frequently (P < 0.05) in primary breast tumors alone; no chromosomal regions showed a significantly higher AI frequency in lymph nodes. Higher rates of AI in primary versus metastatic lymph node tumors suggest that acquisition of metastatic potential may be an early event in carcinogenesis, occurring before significant levels of AI accumulate in the primary tumor. In addition, patterns of AI were highly discordant between tumor types, suggesting that additional genetic alterations accumulated independently in the two cell populations.