Advertisement

Stromal Expression of miR-21 Identifies High-Risk Group in Triple-Negative Breast Cancer

Open AccessPublished:November 05, 2014DOI:https://doi.org/10.1016/j.ajpath.2014.08.020
      Triple-negative breast cancer (TNBC) is an aggressive subtype defined by the lack of estrogen receptor, progesterone receptor, and human epidermal growth factor receptor 2 expression. Expression of miR-21, an oncomiR, is frequently altered and may be distinctly expressed in the tumor stroma. Because tumor lesions are a complex mixture of cell types, we hypothesized that analysis of miR-21 expression at single-cell resolution could provide more accurate information to assess disease recurrence risk and BC-related death. We implemented a fully automated, tissue slide–based assay to detect miR-21 expression in 988 patients with BC. The miR-21High group exhibited shorter recurrence-free survival [hazard ratio (HR), 1.71; P < 0.001] and BC-specific survival (HR, 1.96; P < 0.001) in multivariate regression analyses. When tumor compartment and levels of miR-21 expression were considered, significant associations with poor clinical outcome were detected exclusively in tumor epithelia from estrogen receptor– and/or progesterone receptor–positive human epidermal growth factor receptor 2–negative cases [recurrence-free survival: HR, 3.67 (P = 0.006); BC-specific survival: HR, 5.13 (P = 0.002)] and in tumor stroma from TNBC cases [recurrence-free survival: HR, 2.59 (P = 0.013); BC-specific survival: HR, 3.37 (P = 0.003)]. These findings suggest that the context of altered miR-21 expression provides clinically relevant information. Importantly, miR-21 expression was predominantly up-regulated and potentially prognostic in the tumor stroma of TNBC.
      Breast cancer (BC) is a highly heterogeneous disease consisting of several molecular subtypes that correlate with clinical outcome. Expression levels of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) are used to define the major subtypes, with ER+ tumors having a significantly better prognosis than ER tumors.
      Cancer Genome Atlas Network
      Comprehensive molecular portraits of human breast tumours.
      • Curtis C.
      • Shah S.P.
      • Chin S.F.
      • Turashvili G.
      • Rueda O.M.
      • Dunning M.J.
      • Speed D.
      • Lynch A.G.
      • Samarajiwa S.
      • Yuan Y.
      • Graf S.
      • Ha G.
      • Haffari G.
      • Bashashati A.
      • Russell R.
      • McKinney S.
      • Langerod A.
      • Green A.
      • Provenzano E.
      • Wishart G.
      • Pinder S.
      • Watson P.
      • Markowetz F.
      • Murphy L.
      • Ellis I.
      • Purushotham A.
      • Borresen-Dale A.L.
      • Brenton J.D.
      • Tavare S.
      • Caldas C.
      • Aparicio S.
      The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups.
      Patients with BC have benefited from recent advances in molecular diagnostic assays. For ER+ cases, assays that are gene expression based (ie, OncoType Dx, Mammaprint, and PAM50) and tissue slide based [eg, immunohistochemistry (IHC) 4 and Mammostrat] can refine risk assessment for recurrence and identify patients who may benefit from intensive adjuvant treatments in addition to hormonal therapy.
      Cancer Genome Atlas Network
      Comprehensive molecular portraits of human breast tumours.
      • Curtis C.
      • Shah S.P.
      • Chin S.F.
      • Turashvili G.
      • Rueda O.M.
      • Dunning M.J.
      • Speed D.
      • Lynch A.G.
      • Samarajiwa S.
      • Yuan Y.
      • Graf S.
      • Ha G.
      • Haffari G.
      • Bashashati A.
      • Russell R.
      • McKinney S.
      • Langerod A.
      • Green A.
      • Provenzano E.
      • Wishart G.
      • Pinder S.
      • Watson P.
      • Markowetz F.
      • Murphy L.
      • Ellis I.
      • Purushotham A.
      • Borresen-Dale A.L.
      • Brenton J.D.
      • Tavare S.
      • Caldas C.
      • Aparicio S.
      The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups.
      • Bartlett J.M.
      • Bloom K.J.
      • Piper T.
      • Lawton T.J.
      • van dV.
      • Ross D.T.
      • Ring B.Z.
      • Seitz R.S.
      • Beck R.A.
      • Hasenburg A.
      • Kieback D.
      • Putter H.
      • Markopoulos C.
      • Dirix L.
      • Seynaeve C.
      • Rea D.
      Mammostrat as an immunohistochemical multigene assay for prediction of early relapse risk in the tamoxifen versus exemestane adjuvant multicenter trial pathology study.
      • Schnitt S.J.
      Classification and prognosis of invasive breast cancer: from morphology to molecular taxonomy.
      For HER2+ cases, overexpression of HER2 is a predictive biomarker for response to targeted treatment against HER2 signaling.
      • Yan M.
      • Parker B.A.
      • Schwab R.
      • Kurzrock R.
      HER2 aberrations in cancer: implications for therapy.
      However, for ER cases, there are no salient prognostic markers.
      • Blows F.M.
      • Driver K.E.
      • Schmidt M.K.
      • Broeks A.
      • van Leeuwen F.E.
      • Wesseling J.
      • et al.
      Subtyping of breast cancer by immunohistochemistry to investigate a relationship between subtype and short and long term survival: a collaborative analysis of data for 10,159 cases from 12 studies.
      This is particularly relevant for triple-negative BC (TNBC), which accounts for 15% to 20% of BC cases and is defined by the lack of ER, PR, and HER2 expression.
      • Bayraktar S.
      • Gluck S.
      Molecularly targeted therapies for metastatic triple-negative breast cancer.
      • Kashiwagi S.
      • Yashiro M.
      • Takashima T.
      • Aomatsu N.
      • Ikeda K.
      • Ogawa Y.
      • Ishikawa T.
      • Hirakawa K.
      Advantages of adjuvant chemotherapy for patients with triple-negative breast cancer at Stage II: usefulness of prognostic markers E-cadherin and Ki67.
      TNBCs are associated with shorter recurrence free survival (RFS) and BC-specific survival (CSS) relative to other major BC subtypes. TNBC is still a heterogeneous group of diseases, and gene expression profiling methods have suggested the existence of distinct subgroups.
      • Engebraaten O.
      • Vollan H.K.
      • Borresen-Dale A.L.
      Triple-negative breast cancer and the need for new therapeutic targets.
      • Mayer I.A.
      • Abramson V.G.
      • Lehmann B.D.
      • Pietenpol J.A.
      New strategies for triple-negative breast cancer: deciphering the heterogeneity.
      Currently, TNBC is treated with different cytotoxic combination chemotherapy; complete response to neoadjuvant chemotherapy, depending on specific treatment regimens, varies widely, between 8% and 83%.
      • Engebraaten O.
      • Vollan H.K.
      • Borresen-Dale A.L.
      Triple-negative breast cancer and the need for new therapeutic targets.
      Thus, predictive biomarkers for therapeutic response prediction and novel therapeutic targets that address distinct biological features of TNBC subgroups are needed for these patients.
      • Bayraktar S.
      • Gluck S.
      Molecularly targeted therapies for metastatic triple-negative breast cancer.
      • Kashiwagi S.
      • Yashiro M.
      • Takashima T.
      • Aomatsu N.
      • Ikeda K.
      • Ogawa Y.
      • Ishikawa T.
      • Hirakawa K.
      Advantages of adjuvant chemotherapy for patients with triple-negative breast cancer at Stage II: usefulness of prognostic markers E-cadherin and Ki67.
      Interactions between tumor epithelia (cancer cells) and the tumor microenvironment (TME) play a major role in BC progression. The TME is composed of a complex network of stromal cells, immune cells, extracellular matrix, and cytokines/chemokines. Stromal gene expression signatures have been identified that predict outcome and treatment response in BC.
      • Finak G.
      • Bertos N.
      • Pepin F.
      • Sadekova S.
      • Souleimanova M.
      • Zhao H.
      • Chen H.
      • Omeroglu G.
      • Meterissian S.
      • Omeroglu A.
      • Hallett M.
      • Park M.
      Stromal gene expression predicts clinical outcome in breast cancer.
      • Farmer P.
      • Bonnefoi H.
      • Anderle P.
      • Cameron D.
      • Wirapati P.
      • Becette V.
      • Andre S.
      • Piccart M.
      • Campone M.
      • Brain E.
      • Macgrogan G.
      • Petit T.
      • Jassem J.
      • Bibeau F.
      • Blot E.
      • Bogaerts J.
      • Aguet M.
      • Bergh J.
      • Iggo R.
      • Delorenzi M.
      A stroma-related gene signature predicts resistance to neoadjuvant chemotherapy in breast cancer.
      The tumor/stroma ratio of patients with TNBC is particularly predictive of relapse.
      • de Kruijf E.M.
      • van Nes J.G.
      • van de Velde C.J.
      • Putter H.
      • Smit V.T.
      • Liefers G.J.
      • Kuppen P.J.
      • Tollenaar R.A.
      • Mesker W.E.
      Tumor-stroma ratio in the primary tumor is a prognostic factor in early breast cancer patients, especially in triple-negative carcinoma patients.
      Despite the overwhelming data on the influence of the TME, cancer diagnostics and therapeutics are directed primarily at targeting the tumor epithelia.
      miRNAs are a class of short, noncoding, regulatory RNAs that modulate gene expression in important developmental, physiological, and pathological processes.
      • Mendell J.T.
      • Olson E.N.
      MicroRNAs in stress signaling and human disease.
      A single miRNA can down-regulate the expression of hundreds of target genes by binding to a partially complementary site in the 3′-untranslated region of their cognate mRNAs. The clinical implications of miRNA biological features are generating great interest in the areas of cancer research and cancer medicine.
      • Ventura A.
      • Jacks T.
      MicroRNAs and cancer: short RNAs go a long way.

      Sempere LF, Kauppinen S: Translational Implications of MicroRNAs in Clinical Diagnostics and Therapeutics: Handbook of Cell Signaling. Edited by RA Bradshaw, EA Dennis. Oxford, Academic Press, 2009, pp 2965–2981

      • Nana-Sinkam S.P.
      • Croce C.M.
      Clinical applications for microRNAs in cancer.
      • Garzon R.
      • Marcucci G.
      • Croce C.M.
      Targeting microRNAs in cancer: rationale, strategies and challenges.
      • Ling H.
      • Fabbri M.
      • Calin G.A.
      MicroRNAs and other non-coding RNAs as targets for anticancer drug development.
      miRNA-based diagnostics is an emerging area of novel prognostic and predictive indicators, and some miRNAs have provided leads for the development of novel targeted therapies.
      • Nana-Sinkam S.P.
      • Croce C.M.
      Clinical applications for microRNAs in cancer.
      • Garzon R.
      • Marcucci G.
      • Croce C.M.
      Targeting microRNAs in cancer: rationale, strategies and challenges.
      • Sempere L.F.
      Integrating contextual miRNA and protein signatures for diagnostic and treatment decisions in cancer.
      Quantitative real-time RT-PCR expression analysis of miRNAs in formalin-fixed tissue specimens are now offered commercially as ancillary tests for identifying the organ site of cancers of unknown primary, for performing differential diagnosis of pancreatic cancer, and for obtaining subtype classification of lung and renal cell cancers.
      • Sempere L.F.
      Recent advances in miRNA-based diagnostic applications.
      Altered expression of miR-21 has been associated with poor disease outcome in various cancer types.
      • Radojicic J.
      • Zaravinos A.
      • Vrekoussis T.
      • Kafousi M.
      • Spandidos D.A.
      • Stathopoulos E.N.
      MicroRNA expression analysis in triple-negative (ER, PR and Her2/neu) breast cancer.
      • Ota D.
      • Mimori K.
      • Yokobori T.
      • Iwatsuki M.
      • Kataoka A.
      • Masuda N.
      • Ishii H.
      • Ohno S.
      • Mori M.
      Identification of recurrence-related microRNAs in the bone marrow of breast cancer patients.
      • Gong C.
      • Yao Y.
      • Wang Y.
      • Liu B.
      • Wu W.
      • Chen J.
      • Su F.
      • Yao H.
      • Song E.
      Up-regulation of miR-21 mediates resistance to trastuzumab therapy for breast cancer.
      • Qian B.
      • Katsaros D.
      • Lu L.
      • Preti M.
      • Durando A.
      • Arisio R.
      • Mu L.
      • Yu H.
      High miR-21 expression in breast cancer associated with poor disease-free survival in early stage disease and high TGF-beta1.
      • Yan L.X.
      • Huang X.F.
      • Shao Q.
      • Huang M.Y.
      • Deng L.
      • Wu Q.L.
      • Zeng Y.X.
      • Shao J.Y.
      MicroRNA miR-21 overexpression in human breast cancer is associated with advanced clinical stage, lymph node metastasis and patient poor prognosis.
      • Lee J.A.
      • Lee H.Y.
      • Lee E.S.
      • Kim I.
      • Bae J.W.
      Prognostic implications of microRNA-21 overexpression in invasive ductal carcinomas of the breast.
      • Anastasov N.
      • Hofig I.
      • Vasconcellos I.G.
      • Rappl K.
      • Braselmann H.
      • Ludyga N.
      • Auer G.
      • Aubele M.
      • Atkinson M.J.
      Radiation resistance due to high expression of miR-21 and G2/M checkpoint arrest in breast cancer cells.
      Both in vitro and in vivo work has demonstrated the significant role that miR-21 plays in tumorigenesis and the potential of miR-21 as a therapeutic target.
      • Medina P.P.
      • Nolde M.
      • Slack F.J.
      OncomiR addiction in an in vivo model of microRNA-21-induced pre-B-cell lymphoma.
      • Hatley M.E.
      • Patrick D.M.
      • Garcia M.R.
      • Richardson J.A.
      • Bassel-Duby R.
      • van R.E.
      • Olson E.N.
      Modulation of K-Ras-dependent lung tumorigenesis by MicroRNA-21.
      miR-21 also exerts fibrogenic functions, promoting fibrosis in mouse models of renal and cardiac injury–induced failure as well as in in vitro co-culture studies of colon and pancreatic cancer cell lines.
      • Kadera B.E.
      • Li L.
      • Toste P.A.
      • Wu N.
      • Adams C.
      • Dawson D.W.
      • Donahue T.R.
      MicroRNA-21 in pancreatic ductal adenocarcinoma tumor-associated fibroblasts promotes metastasis.
      • Bullock M.D.
      • Pickard K.M.
      • Nielsen B.S.
      • Sayan A.E.
      • Jenei V.
      • Mellone M.
      • Mitter R.
      • Primrose J.N.
      • Thomas G.J.
      • Packham G.K.
      • Mirnezami A.H.
      Pleiotropic actions of miR-21 highlight the critical role of deregulated stromal microRNAs during colorectal cancer progression.
      • Chung A.C.
      • Yu X.
      • Lan H.Y.
      MicroRNA and nephropathy: emerging concepts.
      • Cheng Y.
      • Zhang C.
      MicroRNA-21 in cardiovascular disease.
      We and others have demonstrated that miRNA expression in the tumor epithelia versus the stroma can result in distinct clinical implications.
      • Sempere L.F.
      Integrating contextual miRNA and protein signatures for diagnostic and treatment decisions in cancer.
      • Sempere L.F.
      • Preis M.
      • Yezefski T.
      • Ouyang H.
      • Suriawinata A.A.
      • Silahtaroglu A.
      • Conejo-Garcia J.R.
      • Kauppinen S.
      • Wells W.
      • Korc M.
      Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered MicroRNA expression in solid tumors.
      • Nelson P.T.
      • Wilfred B.R.
      In situ hybridization is a necessary experimental complement to microRNA (miRNA) expression profiling in the human brain.
      • Zhang X.
      • Lu X.
      • Lopez-Berestein G.
      • Sood A.
      • Calin G.
      In situ hybridization-based detection of microRNAs in human diseases.
      • Nielsen B.S.
      • Holmstrom K.
      Combined microRNA in situ hybridization and immunohistochemical detection of protein markers.
      • Nielsen B.S.
      • Jorgensen S.
      • Fog J.U.
      • Sokilde R.
      • Christensen I.J.
      • Hansen U.
      • Brunner N.
      • Baker A.
      • Moller S.
      • Nielsen H.J.
      High levels of microRNA-21 in the stroma of colorectal cancers predict short disease-free survival in stage II colon cancer patients.
      It is not clear whether miR-21 exerts its tumorigenic functions in BC primarily within cancer cells or in other cellular elements of the TME. Research using in situ hybridization (ISH) has suggested that miR-21 expression is predominantly up-regulated in the tumor stroma.
      • Sempere L.F.
      • Preis M.
      • Yezefski T.
      • Ouyang H.
      • Suriawinata A.A.
      • Silahtaroglu A.
      • Conejo-Garcia J.R.
      • Kauppinen S.
      • Wells W.
      • Korc M.
      Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered MicroRNA expression in solid tumors.
      • Sempere L.F.
      • Christensen M.
      • Silahtaroglu A.
      • Bak M.
      • Heath C.V.
      • Schwartz G.
      • Wells W.
      • Kauppinen S.
      • Cole C.N.
      Altered MicroRNA expression confined to specific epithelial cell subpopulations in breast cancer.
      • Jorgensen S.
      • Baker A.
      • Moller S.
      • Nielsen B.S.
      Robust one-day in situ hybridization protocol for detection of microRNAs in paraffin samples using LNA probes.
      • Rask L.
      • Balslev E.
      • Jorgensen S.
      • Eriksen J.
      • Flyger H.
      • Moller S.
      • Hogdall E.
      • Litman T.
      • Schnack N.B.
      High expression of miR-21 in tumor stroma correlates with increased cancer cell proliferation in human breast cancer.
      The few BC cases in these studies precluded prognostic evaluation of this marker.
      Herein, we conducted an adequately powered ISH-based study on 901 BC cases to test the hypothesis that both the levels and tumor compartment of miR-21 expression in primary BC tumor tissues are informative for predicting RFS and CSS. We observed that miR-21 expression was significantly associated with clinical outcome in different BC subtypes on the basis of ER-PR/HER2 status. This association was tumor compartment specific and was influenced by the intrinsic biological features of major BC subtypes. Specifically, high expression of miR-21 in the tumor epithelia in ER+ and/or PR+HER2 cases was associated with poor clinical outcome, and high expression of miR-21 in tumor stroma was associated with a much poorer clinical outcome in TNBC.

      Materials and Methods

      Patient Cohort

      Tissue microarrays (TMAs) representing tissue cores from 988 female patients diagnosed with nonmetastatic invasive BC between 1985 and 1997 (stage I, sets 9 to 13; and stage II, sets 14 to 17) were acquired from the National Cancer Institute (NCI) Cancer Diagnosis Program (CDP; http://cdp.nci.nih.gov/breast/prognostic_cs.html, last accessed September 19, 2014). This NCI CDP 2008 Series Prognostic TMAs include 590 stage I and 398 stage II cases. The Institutional Review Board (Dartmouth College Committee for the Protection of Human Subjects) approved a waiver for the entire consent process and use of a consent form (Committee for Protection of Human Subjects number 22042).

      ISH and IHC

      A fluorescence-based ISH/IHC assay was conducted as previously described,
      • Sempere L.F.
      • Preis M.
      • Yezefski T.
      • Ouyang H.
      • Suriawinata A.A.
      • Silahtaroglu A.
      • Conejo-Garcia J.R.
      • Kauppinen S.
      • Wells W.
      • Korc M.
      Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered MicroRNA expression in solid tumors.
      with slight modifications, on a Leica BOND-MAX automated staining station (Leica Biosystems, Buffalo Grove, IL) in the Dartmouth Pathology Translational Research Laboratory, a College of American Pathologists’–accredited Clinical Laboratory Improvement Amendments–certified facility. miR-21 staining was calibrated by adjusting probe concentration and fluorescent substrate incorporation time so that no signal was detectable in adjacent normal tissue of BC tumor lesions. Briefly, double-tagged miR-21 (FAM2X) and snRNA U6 (biotin2X) locked nucleic acid–modified DNA probes at 50 nmol/L each were hybridized to tissue slides for 75 minutes at 45°C. Expression of miR-21, U6, and cytokeratin (CK) 19 was assessed with appropriate antibody combinations, followed by sequential rounds of HRP-mediated deposition of appropriate fluorochrome-conjugated tyramine substrates for 20 minutes. Fluorescent images were captured with an EXi Aqua QImaging camera (QImaging, Surrey, BC, Canada) mounted on a BX51 microscope (Olympus, Center Valley, PA). Image-Pro Plus software version 7.0 (Media Cybernetics, Rockville, MD) was used for histogram-based image segmentation analysis. Eighty-seven cores were excluded because of insufficient representation of tumor lesions and/or lack of stain for all markers. miR-21 signal intensity was scored in two locations, tumor epithelium and stroma, on a scale from 0 (no expression) to 3 (high expression). For assessing correlations with clinicopathological factors, sample tissue cores were considered low (0 or 1) or high (2 or 3) for each location (Supplemental Table S1). This scoring system was independently validated by multiviewer (L.F.S. and W.A.W.) and computer-assisted image segmentation analyses on digital images captured from a representative subset of 399 cases (Supplemental Figure S1).

      Statistical Analysis

      We followed the REMARK guidelines for reporting of prognostic markers
      • Altman D.G.
      • McShane L.M.
      • Sauerbrei W.
      • Taube S.E.
      Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK): explanation and elaboration.
      • McShane L.M.
      • Altman D.G.
      • Sauerbrei W.
      • Taube S.E.
      • Gion M.
      • Clark G.M.
      Reporting recommendations for tumor marker prognostic studies (REMARK).
      and the NCI CDP's recommendations for data set analysis (http://cdp.nci.nih.gov/breast/prognostic_dm.html, last accessed September 19, 2014). These TMAs have a robust statistical design to detect associations between tumor markers and clinical outcome in nonmetastatic BC cases. TMA sample sizes were determined by NCI statisticians to detect a hazard ratio (HR) of 2.0 with 80% or greater power. Associations between miR-21 expression and clinicopathological factors were assessed by t-test, χ2 test, or Fischer's exact test (Supplemental Appendix S1). Kaplan-Meier curves (which estimate survival if death from other causes were eliminated as a possibility) of RFS and CSS were constructed. RFS was defined as the time to any documented recurrence event or time of death with evidence of BC in the absence of an earlier documented recurrence event; CSS was defined as time before death with evidence of BC. Kaplan-Meier curves were compared using the log-rank test. Cause-specific HRs, subdistribution HRs, and their 95% CIs were derived using the Cox proportional hazards model. Unless otherwise noted, reported HRs correspond to results from the multivariate Cox proportional hazards regression model, adjusted for age, tumor size, lymph node involvement, and grade. All statistical analyses were performed using the R software package (http://www.r-project.org).

      Results

      miR-21 Is Predominantly Expressed in the Tumor Stroma in Breast Cancer

      To investigate the prognostic value of tumor compartment and levels of miR-21 expression in BC, we determined the expression of miR-21, snRNA U6, and CK19 protein on tissue cores represented in the NCI CDP Prognotic TMAs, using a fully automated fluorescence-based multiplex ISH/IHC assay.
      • Sempere L.F.
      • Preis M.
      • Yezefski T.
      • Ouyang H.
      • Suriawinata A.A.
      • Silahtaroglu A.
      • Conejo-Garcia J.R.
      • Kauppinen S.
      • Wells W.
      • Korc M.
      Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered MicroRNA expression in solid tumors.
      Expression of snRNA U6 was used to assess RNA quality, and expression of CK19 was used to identify the tumor epithelia in each tissue core. In total, 901 cases were included in the analysis. Molecular and clinicopathological characteristics are summarized in Table 1, and the complete data set with clinical outcomes is provided in Supplemental Table S1.
      Table 1Patient Characteristics and miR-21 Expression Data
      CharacteristicConditionn% of Total% of Available
      miR-21 compartmentTumor epithelia expression455.005.00
      Tumor stroma expression34137.8037.80
      No expression51557.2057.20
      miR-21 scoreHigh20723.0023.00
      Low69477.0077.00
      Age (years)<5533136.7036.70
      >5557063.3063.30
      Tumor size (cm)<264671.7071.70
      >225528.3028.30
      LN statusN064271.3071.30
      N125928.7028.70
      Grade124927.6027.60
      241245.7045.70
      324026.6026.60
      StageI52057.6057.60
      II38142.4042.40
      ER statusNegative18220.2023.82
      Positive58264.6076.18
      Data not available13715.20NA
      PR statusNegative23325.9030.46
      Positive53259.0069.54
      Data not available13615.10NA
      HER statusNegative65873.0085.34
      Positive11312.5014.66
      Data not available13014.40NA
      ER/PR/HER2 subtypeER+ and/or PR+HER254960.9372.14
      (any ER or PR) HER2+10711.8814.06
      ERPRHER210511.6513.80
      Data not available14015.54NA
      Data are given as number and percentage of cases in each category for miR-21 expression data and patient characteristics currently used in the clinic as prognostic factors. Additional patient characteristics are provided in Supplemental Table S1.
      ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; LN, lymph node; NA, not applicable because data are missing for these categories; PR, progesterone receptor.
      miR-21 expression was detected in 386 (42.8%) of the 901 cases. miR-21 was predominantly or exclusively expressed within the tumor stroma compartment (CK19 cells) in most cases (88.3%), whereas it was confined to the tumor epithelia compartment (CK19+ cells) in a few cases (11.7%). We previously determined by microarray and Northern blot analyses that miR-21 was expressed at high levels in the MCF7 BC cell line and was not expressed in the MCF10A BC line.
      • Sempere L.F.
      • Christensen M.
      • Silahtaroglu A.
      • Bak M.
      • Heath C.V.
      • Schwartz G.
      • Wells W.
      • Kauppinen S.
      • Cole C.N.
      Altered MicroRNA expression confined to specific epithelial cell subpopulations in breast cancer.
      Therefore, we used cores of these BC cell lines represented in TMA slides as a reference for scoring miR-21 expression levels (Figure 1). Cases were reviewed under fluorescent microscopy and assigned a score of low (694 cases) or high (207 cases) on the basis of signal intensity. This miR-21 score provided a threshold that further separated nonexpressing and low-expressing cases from those with higher miR-21 expression.
      Figure thumbnail gr1
      Figure 1Patterns, quantification, and scoring of miR-21 expression in breast cancer lesions. A and B: Representative images of multicolor detection of miR-21, snRNA U6, and CK19 in BC cell lines (A) and tissue cores (B). miR-21 heat map images were generated using histogram-based image segmentation analysis; the background signal was set on the basis of exposed slide areas (eg, tissue holes). Other categories were defined relative to the highest background intensity (black) as onefold (blue), twofold (cyan), threefold (orange), and fourfold or more (red) over background. Cases with a score of 1 to 2 and 3 to 4 were categorized low and high miR-21 expression, respectively. C: Summary of miR-21 expression levels, tumor compartment of expression, and molecular subtype on the basis of ER/PR/HER2 status in entire patient cohort. Original magnification, ×200 (A and B). Ca, tumor epithelia; Stroma, tumor stroma.

      miR-21 Score Is an Independent Risk Factor Associated with Poor Disease Outcome

      The miR-21High group had an increased risk for shorter RFS (Supplemental Tables S2 and S3) and CSS (Tables 2 and 3) at 5 years after diagnosis and during the overall follow-up period (median, 10.33 years). Stratification of cases by standard clinicopathological characteristics indicated that the miR-21 score was an independent risk factor associated with poor disease outcome (Tables 2 and 3 and Supplemental Tables S4 and S5). Notably, the miR-21High grade 2 subgroup had a much higher risk relative to the miR-21Low grade 2 subgroup for shorter RFS (5 years: HR, 3.85; P < 0.001) and CSS (5 years; HR, 5.51; P < 0.001) and a similar risk as the miR-21Low grade 3 subgroup (Supplemental Table S4).
      Table 2CSS Risk Associated with miR-21 Expression for Disease Outcome 5 Years after Diagnosis
      GroupVariableUnivariateMultivariate
      HR95% CIP valueHR95% CIP value
      All cases (n = 901)miR-21High/Low2.691.65–4.39<0.0012.311.41–3.79<0.001
      miR-21HiCancer/Low3.861.35–11.030.0103.041.05–8.860.037
      miR-21HiStroma/Low2.571.54–4.28<0.0012.221.32–3.730.002
      Grade 1/20.400.132–1.2350.1050.480.16–1.480.195
      Grade 3/25.533.10–9.85<0.0014.512.50–8.13<0.001
      Stage II/I4.422.52–7.74<0.0011.040.36–3.050.937
      ER+ and/or PR+HER2 (n = 549)miR-21High/Low3.151.46–6.810.0032.571.18–5.610.016
      miR-21HiCancer/Low9.462.65–33.80<0.0017.051.75–28.440.005
      miR-21HiStroma/Low2.621.15–6.010.0202.150.93–4.990.069
      ERPR HER2 (n = 105)miR-21High/Low2.501.03–6.080.0392.821.10–7.260.028
      miR-21HiCancer/LowNANANANANANA
      miR-21HiStroma/Low2.841.17–6.900.0193.091.20–7.980.017
      Data are given as HRs and their 95% CIs for miR-21 expression and standard clinical indicators using Cox proportional hazard models. The following variables were adjusted for in multivariate analyses: age, tumor size, grade, and lymph node involvement. Risks associated with miR-21High score are reported as a whole group (miR-21High) and as subgroups (miR-21HiCancer and miR-21HiStroma) divided by tumor compartment of expression.
      NA, not applicable because not enough data entries for this category to perform analysis.
      Table 3CSS Risk Associated with miR-21 Expression for Overall Disease Outcome
      GroupVariableUnivariateMultivariate
      HR95% CIP valueHR95% CIP value
      All cases (n = 901)miR-21High/Low1.941.38–2.75<0.0011.961.38–2.78<0.001
      miR-21HiCancer/Low2.611.13–6.050.0232.741.17–6.390.018
      miR-21HiStroma/Low1.881.31–2.70<0.0011.881.31–2.72<0.001
      Grade 1/20.490.29–0.840.0070.560.33–0.960.032
      Grade 3/22.571.81–3.66<0.0012.161.51–3.10<0.001
      Stage II/I3.152.23–4.43<0.0010.970.48–1.940.925
      ER+ and/or PR+HER2 (n = 549)miR-21High/Low1.701.05–2.750.0281.811.11–2.950.016
      miR-21HiCancer/Low4.251.51–11.980.0055.131.75–15.070.002
      miR-21HiStroma/Low1.530.91–2.550.1001.610.96–2.700.068
      ERPR HER2 (n = 105)miR-21High/Low2.621.24–5.550.0103.291.47–7.370.003
      miR-21HiCancer/Low2.250.29–17.560.4312.710.32–22.710.338
      miR-21HiStroma/Low2.661.23–5.770.0113.371.45–7.810.004
      Data are given as HRs and their 95% CIs for miR-21 expression and standard clinical indicators using Cox proportional hazard models. The following variables were adjusted for in multivariate analyses: age, tumor size, grade, and lymph node involvement. Risks associated with miR-21High score are reported as a whole group (miR-21High) and as subgroups (miR-21HiCancer and miR-21HiStroma) divided by tumor compartment of expression.

      Tumor Compartment–Specific Expression of miR-21 Defines Risk in Distinct Molecular Subtypes

      To assess the contribution of tumor compartment of miR-21 expression to the risk associated with disease outcome, the miR-21High group was subdivided into a miR-21HiCancer subgroup of 20 cases, which included only cases with tumor epithelia–specific expression, and a miR-21HiStroma group of 187 cases with predominant expression in the tumor stroma. The miR-21HiCancer subgroup had a higher risk for shorter RFS and CSS (Tables 2 and 3 and Supplemental Tables S4 and S5) than the miR-21HiStroma subgroup relative to the miR-21Low group. Because the ER/PR/HER2 status of breast cancer tumors drives their biological features and evolution,
      Cancer Genome Atlas Network
      Comprehensive molecular portraits of human breast tumours.
      • Curtis C.
      • Shah S.P.
      • Chin S.F.
      • Turashvili G.
      • Rueda O.M.
      • Dunning M.J.
      • Speed D.
      • Lynch A.G.
      • Samarajiwa S.
      • Yuan Y.
      • Graf S.
      • Ha G.
      • Haffari G.
      • Bashashati A.
      • Russell R.
      • McKinney S.
      • Langerod A.
      • Green A.
      • Provenzano E.
      • Wishart G.
      • Pinder S.
      • Watson P.
      • Markowetz F.
      • Murphy L.
      • Ellis I.
      • Purushotham A.
      • Borresen-Dale A.L.
      • Brenton J.D.
      • Tavare S.
      • Caldas C.
      • Aparicio S.
      The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups.
      • Blows F.M.
      • Driver K.E.
      • Schmidt M.K.
      • Broeks A.
      • van Leeuwen F.E.
      • Wesseling J.
      • et al.
      Subtyping of breast cancer by immunohistochemistry to investigate a relationship between subtype and short and long term survival: a collaborative analysis of data for 10,159 cases from 12 studies.
      • Carey L.A.
      • Perou C.M.
      • Livasy C.A.
      • Dressler L.G.
      • Cowan D.
      • Conway K.
      • Karaca G.
      • Troester M.A.
      • Tse C.K.
      • Edmiston S.
      • Deming S.L.
      • Geradts J.
      • Cheang M.C.
      • Nielsen T.O.
      • Moorman P.G.
      • Earp H.S.
      • Millikan R.C.
      Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study.
      we determined if ER/PR/HER2 status influenced the tumor compartment–specific effects of altered miR-21 expression. When cases were stratified by the status of individual receptors, poor clinical outcome for the miR-21HiCancer subgroup was only statistically significant in ER+ cases or in PR+ cases, and for the miR-21HiStroma subgroup, in ER cases or PR cases (Supplemental Tables S4 and S5). In addition, poor clinical outcome associated with the miR-21HiCancer and miR-21HiStroma subgroups was statistically significant only for HER2 cases (Figures 2 and 3 and Supplemental Tables S4 and S5). When cases were stratified by the combined status of ER/PR/HER2 as molecular surrogate of major gene expression subtypes,
      • Carey L.A.
      • Perou C.M.
      • Livasy C.A.
      • Dressler L.G.
      • Cowan D.
      • Conway K.
      • Karaca G.
      • Troester M.A.
      • Tse C.K.
      • Edmiston S.
      • Deming S.L.
      • Geradts J.
      • Cheang M.C.
      • Nielsen T.O.
      • Moorman P.G.
      • Earp H.S.
      • Millikan R.C.
      Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study.
      poor clinical outcome associated with the miR-21HiCancer subgroup in ER+ and/or PR+HER2 cases (5-year CSS: HR, 7.05; P = 0.005). The miR-21HiStroma subgroup (Figures 2 and 3 and Tables 2 and 3) was associated with poor outcome in ERPRHER2 cases (5-year CSS: HR, 3.09; P = 0.017).
      Figure thumbnail gr2
      Figure 2Kaplan-Meier plots of disease recurrence on the basis of miR-21 expression. Kaplan-Meier survival curves for breast cancer–specific recurrence were plotted by miR-21 category for all cases or major subtypes on the basis of ER/PR/HER2 status. Curves were also plotted by tumor compartment (miR-21HiCancer, miR-21HiStroma, and miR-21Low).
      Figure thumbnail gr3
      Figure 3Kaplan-Meier plots for disease outcome on the basis of miR-21 expression. Kaplan-Meier survival curves for breast cancer–specific deaths were plotted by miR-21 category for all cases or major subtypes on the basis of ER/PR/HER2 status. Curves were also plotted by tumor compartment (miR-21HiCancer, miR-21HiStroma, and miR-21Low).

      Discussion

      miRNA diagnostics is an emerging area of novel prognostic and predictive indicators for BC. Tissue slide–based assays are routine procedures in clinical laboratories to diagnose, prognosticate, and guide treatment selection. To facilitate clinical integration of ancillary miRNA detection assays, we implemented a tissue slide–based fully automated staining assay in a Clinical Laboratory Improvement Amendments–certified environment. Herein, we used this multiplex fluorescent assay to determine molecular and cellular characteristics of BC tumors by measuring the tumor compartment and levels of miR-21 expression. In this biomarker discovery study, we found that contextual expression of miR-21 provided clinically relevant information.
      We determined altered expression of miR-21 at single-cell resolution and extracted this contextual information from the tumor epithelia and stroma (Figure 1). This large retrospective study in a patient cohort of 901 early-stage BC cases indicated that both expression levels and compartment-specific expression of miR-21 contained prognostic information. This information was independent of standard clinicopathological characteristics (Tables 2 and 3), including tumor grade, which is indicative of the differentiation status and proliferative rate of cancer cells. Tumor grade is an important factor in prognostic tools, such as the Nottingham Prognostic Index and Adjuvant! Online.
      • Rakha E.A.
      • Reis-Filho J.S.
      • Baehner F.
      • Dabbs D.J.
      • Decker T.
      • Eusebi V.
      • Fox S.B.
      • Ichihara S.
      • Jacquemier J.
      • Lakhani S.R.
      • Palacios J.
      • Richardson A.L.
      • Schnitt S.J.
      • Schmitt F.C.
      • Tan P.H.
      • Tse G.M.
      • Badve S.
      • Ellis I.O.
      Breast cancer prognostic classification in the molecular era: the role of histological grade.
      The prognostic value of grade is limited in ER cases (HER2-overexpressing and TNBC subtypes) because these tumors frequently have a high grade (grades 2/3).
      • Rakha E.A.
      • Reis-Filho J.S.
      • Baehner F.
      • Dabbs D.J.
      • Decker T.
      • Eusebi V.
      • Fox S.B.
      • Ichihara S.
      • Jacquemier J.
      • Lakhani S.R.
      • Palacios J.
      • Richardson A.L.
      • Schnitt S.J.
      • Schmitt F.C.
      • Tan P.H.
      • Tse G.M.
      • Badve S.
      • Ellis I.O.
      Breast cancer prognostic classification in the molecular era: the role of histological grade.
      Our results suggest that a combination of miR-21 score (tumor stroma characteristic) and grade (tumor epithelia characteristic) may be useful for identifying tumors having more aggressive features.
      By using RNA extracted from whole tissue biopsy specimens, altered expression of miR-21 has been frequently observed in BC and other cancer types.
      • Sempere L.F.
      Integrating contextual miRNA and protein signatures for diagnostic and treatment decisions in cancer.
      • Volinia S.
      • Calin G.A.
      • Liu C.G.
      • Ambs S.
      • Cimmino A.
      • Petrocca F.
      • Visone R.
      • Iorio M.
      • Roldo C.
      • Ferracin M.
      • Prueitt R.L.
      • Yanaihara N.
      • Lanza G.
      • Scarpa A.
      • Vecchione A.
      • Negrini M.
      • Harris C.C.
      • Croce C.M.
      A microRNA expression signature of human solid tumors defines cancer gene targets.
      Functional studies have uncovered oncogenic properties of miR-21 in cancer cell lines and in mouse models of cancer.

      Sempere LF, Kauppinen S: Translational Implications of MicroRNAs in Clinical Diagnostics and Therapeutics: Handbook of Cell Signaling. Edited by RA Bradshaw, EA Dennis. Oxford, Academic Press, 2009, pp 2965–2981

      • Kumarswamy R.
      • Volkmann I.
      • Thum T.
      Regulation and function of miRNA-21 in health and disease.
      Implicit in these findings is the idea that miR-21–mediated regulation is confined to cancer cell–specific processes. We identified a small subgroup of cases with high miR-21 expression exclusively within the tumor epithelia (Figure 1) that were associated with poor clinical outcome (Figures 2 and 3). However, our results indicate that most BC tumors up-regulate miR-21 expression in the tumor stroma (Figure 1). Thus, miR-21–mediated regulation of biological processes in the reactive tumor stroma is likely a more frequent mechanism by which miR-21 affects BC tumor biological features. The fact that high levels of expression in the tumor stroma were associated with poor disease outcome in ERPRHER2/TNBC cases (Figures 2 and 3) suggests that the intrinsic biological features of different subtypes contextualizes the effect and influence of miR-21 activity. Consequently, miR-21 expression levels, tumor compartment of miR-21 expression, and molecular characteristics (eg, ER/PR/HER2 status) appear to contribute to the biological effects of miR-21 and its value as a prognostic indicator and potential therapeutic target in BC.
      The association between miR-21 expression and TNBC subtype is highly relevant. There are currently a few specific prognostic markers and no effective targeted therapies for TNBC.
      • Bayraktar S.
      • Gluck S.
      Molecularly targeted therapies for metastatic triple-negative breast cancer.
      • Kashiwagi S.
      • Yashiro M.
      • Takashima T.
      • Aomatsu N.
      • Ikeda K.
      • Ogawa Y.
      • Ishikawa T.
      • Hirakawa K.
      Advantages of adjuvant chemotherapy for patients with triple-negative breast cancer at Stage II: usefulness of prognostic markers E-cadherin and Ki67.
      • Engebraaten O.
      • Vollan H.K.
      • Borresen-Dale A.L.
      Triple-negative breast cancer and the need for new therapeutic targets.
      • Mayer I.A.
      • Abramson V.G.
      • Lehmann B.D.
      • Pietenpol J.A.
      New strategies for triple-negative breast cancer: deciphering the heterogeneity.
      Seminal studies using gene expression profiling identified an aggressive basal-like subtype, which was highly enriched for TNBC cases, on the basis of intrinsic cancer cell characteristics.
      • Perou C.M.
      • Sorlie T.
      • Eisen M.B.
      • van de R.M.
      • Jeffrey S.S.
      • Rees C.A.
      • Pollack J.R.
      • Ross D.T.
      • Johnsen H.
      • Akslen L.A.
      • Fluge O.
      • Pergamenschikov A.
      • Williams C.
      • Zhu S.X.
      • Lonning P.E.
      • Borresen-Dale A.L.
      • Brown P.O.
      • Botstein D.
      Molecular portraits of human breast tumours.
      • Sorlie T.
      • Perou C.M.
      • Tibshirani R.
      • Aas T.
      • Geisler S.
      • Johnsen H.
      • Hastie T.
      • Eisen M.B.
      • van de R.M.
      • Jeffrey S.S.
      • Thorsen T.
      • Quist H.
      • Matese J.C.
      • Brown P.O.
      • Botstein D.
      • Eystein L.P.
      • Borresen-Dale A.L.
      Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications.
      More recently, systems biology and integrative pathway analysis approaches have defined gene expression signatures that classify TNBC into four to six subgroups.
      • Curtis C.
      • Shah S.P.
      • Chin S.F.
      • Turashvili G.
      • Rueda O.M.
      • Dunning M.J.
      • Speed D.
      • Lynch A.G.
      • Samarajiwa S.
      • Yuan Y.
      • Graf S.
      • Ha G.
      • Haffari G.
      • Bashashati A.
      • Russell R.
      • McKinney S.
      • Langerod A.
      • Green A.
      • Provenzano E.
      • Wishart G.
      • Pinder S.
      • Watson P.
      • Markowetz F.
      • Murphy L.
      • Ellis I.
      • Purushotham A.
      • Borresen-Dale A.L.
      • Brenton J.D.
      • Tavare S.
      • Caldas C.
      • Aparicio S.
      The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups.
      • Kristensen V.N.
      • Vaske C.J.
      • Ursini-Siegel J.
      • Van L.P.
      • Nordgard S.H.
      • Sachidanandam R.
      • Sorlie T.
      • Warnberg F.
      • Haakensen V.D.
      • Helland A.
      • Naume B.
      • Perou C.M.
      • Haussler D.
      • Troyanskaya O.G.
      • Børresen-Dale A.L.
      Integrated molecular profiles of invasive breast tumors and ductal carcinoma in situ (DCIS) reveal differential vascular and interleukin signaling.
      • Kreike B.
      • van Kouwenhove M.
      • Horlings H.
      • Weigelt B.
      • Peterse H.
      • Bartelink H.
      • van de Vijver M.J.
      Gene expression profiling and histopathological characterization of triple-negative/basal-like breast carcinomas.
      • Lehmann B.D.
      • Bauer J.A.
      • Chen X.
      • Sanders M.E.
      • Chakravarthy A.B.
      • Shyr Y.
      • Pietenpol J.A.
      Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies.
      Stromal expression of mesenchymal cells and infiltrating immune cells are main attributes of some of these subgroups. It will be important to determine whether stromal expression of miR-21 correlates with one of more of these TNBC subgroups with tumorigenic stromal features or is an independent risk factor.
      We previously showed that miR-21 was expressed in tumor stromal cells that also expressed smooth muscle actin and vimentin in BC tissues.
      • Sempere L.F.
      • Preis M.
      • Yezefski T.
      • Ouyang H.
      • Suriawinata A.A.
      • Silahtaroglu A.
      • Conejo-Garcia J.R.
      • Kauppinen S.
      • Wells W.
      • Korc M.
      Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered MicroRNA expression in solid tumors.
      Although we did not costain with these (myo)fibroblast markers in this study, the morphological features of miR-21–expressing cells were consistent with those of tumor-associated fibroblasts.
      miR-21 expression in tumor stromal fibroblast-like cells has been recently associated with poor clinical outcome in colon and pancreatic cancers.
      • Kadera B.E.
      • Li L.
      • Toste P.A.
      • Wu N.
      • Adams C.
      • Dawson D.W.
      • Donahue T.R.
      MicroRNA-21 in pancreatic ductal adenocarcinoma tumor-associated fibroblasts promotes metastasis.
      • Nielsen B.S.
      • Jorgensen S.
      • Fog J.U.
      • Sokilde R.
      • Christensen I.J.
      • Hansen U.
      • Brunner N.
      • Baker A.
      • Moller S.
      • Nielsen H.J.
      High levels of microRNA-21 in the stroma of colorectal cancers predict short disease-free survival in stage II colon cancer patients.
      • Kjaer-Frifeldt S.
      • Hansen T.F.
      • Nielsen B.S.
      • Joergensen S.
      • Lindebjerg J.
      • Soerensen F.B.
      • dePont C.R.
      • Jakobsen A.
      The prognostic importance of miR-21 in stage II colon cancer: a population-based study.
      Moreover, in vitro co-culture studies of cancer cell lines and fibroblasts (either from established cell lines or freshly derived from patients) indicate that miR-21 influences fibrogenic processes that enhance cancer cell aggressiveness and invasion.
      • Kadera B.E.
      • Li L.
      • Toste P.A.
      • Wu N.
      • Adams C.
      • Dawson D.W.
      • Donahue T.R.
      MicroRNA-21 in pancreatic ductal adenocarcinoma tumor-associated fibroblasts promotes metastasis.
      • Bullock M.D.
      • Pickard K.M.
      • Nielsen B.S.
      • Sayan A.E.
      • Jenei V.
      • Mellone M.
      • Mitter R.
      • Primrose J.N.
      • Thomas G.J.
      • Packham G.K.
      • Mirnezami A.H.
      Pleiotropic actions of miR-21 highlight the critical role of deregulated stromal microRNAs during colorectal cancer progression.
      In the colon cancer co-culture model, the reversion-inducing cysteine-rich protein with kazal motifs, a negative regulator of prometastatic matrix metalloprotease 2, was identified as a key target of miR-21.
      • Bullock M.D.
      • Pickard K.M.
      • Nielsen B.S.
      • Sayan A.E.
      • Jenei V.
      • Mellone M.
      • Mitter R.
      • Primrose J.N.
      • Thomas G.J.
      • Packham G.K.
      • Mirnezami A.H.
      Pleiotropic actions of miR-21 highlight the critical role of deregulated stromal microRNAs during colorectal cancer progression.
      In cardiac and/or renal mouse models of injury-induced fibrosis, key targets of miR-21 were identified as sprouty homolog 1/2 of Drosophila gene, programmed cell death 4, peroxisome proliferator–activated receptor α, and phosphatase and tensin homolog.
      • Chung A.C.
      • Yu X.
      • Lan H.Y.
      MicroRNA and nephropathy: emerging concepts.
      • Cheng Y.
      • Zhang C.
      MicroRNA-21 in cardiovascular disease.
      miR-21 has also been shown to regulate these same target mRNAs in cancer cell lines,
      • Buscaglia L.E.
      • Li Y.
      Apoptosis and the target genes of microRNA-21.
      suggesting that there is at least some overlap of miR-21 activity in different cell types and diseases. This knowledge could be exploited to implement tumor compartment–specific anti–miR-21–based therapies in BC, especially in TNBC cases for which effective targeted therapies are still lacking.
      • Bayraktar S.
      • Gluck S.
      Molecularly targeted therapies for metastatic triple-negative breast cancer.
      Computer-assisted morphological analyses identified stromal features that correlated with poor outcome in BC.
      • Yuan Y.
      • Failmezger H.
      • Rueda O.M.
      • Ali H.R.
      • Graf S.
      • Chin S.F.
      • Schwarz R.F.
      • Curtis C.
      • Dunning M.J.
      • Bardwell H.
      • Johnson N.
      • Doyle S.
      • Turashvili G.
      • Provenzano E.
      • Aparicio S.
      • Caldas C.
      • Markowetz F.
      Quantitative image analysis of cellular heterogeneity in breast tumors complements genomic profiling.
      • Beck A.H.
      • Sangoi A.R.
      • Leung S.
      • Marinelli R.J.
      • Nielsen T.O.
      • van dV.
      • West R.B.
      • van de R.M.
      • Koller D.
      Systematic analysis of breast cancer morphology uncovers stromal features associated with survival.
      • Moorman A.M.
      • Vink R.
      • Heijmans H.J.
      • van der P.J.
      • Kouwenhoven E.A.
      The prognostic value of tumour-stroma ratio in triple-negative breast cancer.
      It is an intriguing possibility that altered miR-21 expression may cause some of these morphological manifestations. We determined contextual miR-21 expression and generated a miR-21 score on the basis of tissue core stains (Figure 1). Although we discarded tissue cores that did not contain tumor lesions, it is possible that not all cores that we analyzed provided an optimal sampling of the entire tumor. Patients in this study cohort were diagnosed with BC between 1985 and 1997 and were treated in different health care centers. Treatments varied: 47.4% of patients received only local therapy, and 52.6% of patients also received adjuvant treatment (hormone therapy and/or chemotherapy) (Supplemental Table S1). Adjuvant treatments did not match current protocols, namely, use of third-generation aromatase inhibitors (eg, exemestane and anastrozole) for ER+ cases, anti–HER2-targeted therapies (eg, herceptin) for HER2-overexpressing cases, and modifications in chemotherapy combinations and modalities for specific patient subtypes and subgroups. Thus, one should be cautious when interpreting the prognostic value of altered miR-21 expression due to diverse treatment histories. An important first step to address these caveats and to pursue clinical implementation of a tissue slide–based miR-21 detection assay will be to validate these findings in whole-tissue specimens via multi-institutional clinical studies. Future studies are also needed to investigate whether miR-21 expression could serve as a predictive indicator of treatment response in clinical trials evaluating specific treatments. Such studies would reinforce the concept that miR-21 activity in the TME is biologically relevant in BC.
      In conclusion, this study provides both a promising miRNA prognostic biomarker and a robust method to bring tissue slide–based miRNA detection assays closer to routine clinical practice. If these findings are supported and validated by independent clinical studies, the tumor compartment and level of miR-21 expression could become useful ancillary indicators for risk stratification in specific BC subtypes or their subgroups.

      Acknowledgments

      We thank Carol Valentine, Rebecca O'Meara, and Eric York (Dartmouth Pathology Translational Research Laboratory) for their technical support and assistance; Drs. Angeline Andrew, Elena Bryleva, Andrew DeCastro, Arti Gaur, and Jennifer Westerhuis for critical reading of the manuscript; and the NCI Cancer Diagnosis Program for providing TMA slides. (Because of the nature of the program, other investigators may receive and analyze slides from these same array blocks.)

      Supplemental Data

      • Supplemental Figure S1

        Multiviewer quantification of miR-21 expression. L.F.S. and W.A.W. independently scored digital images of 399 representative cases using the same categories (1 to 4) for intensity of miR-21 stain as in Figure 1. Computer-assisted (CA) digital analysis of the same 399 images rendered scores of 1, 3, and 4 for histogram-based segmented images with <1%, >1%, and >1.5% of the combined pixels for miR-21 signal, respectively, which were located in the 3 and 4+ fold over background categories. Consensus score was the arithmetic average of ocular (L.F.S.), digital (L.F.S. or W.A.W.), and CA digital scores. Agreement between initial scores from different viewers was measured with κ coefficients. After group review and discussion, scores of one viewer (W.A.W. digital), who had applied a more stringent criterion to assign a score of miR-21High, became more similar to those of other viewers (data not shown). Ultimately, cases with a score of 1 to 2 and 3 to 4 received a miR-21 score of low and high, respectively.

      References

        • Cancer Genome Atlas Network
        Comprehensive molecular portraits of human breast tumours.
        Nature. 2012; 490: 61-70
        • Curtis C.
        • Shah S.P.
        • Chin S.F.
        • Turashvili G.
        • Rueda O.M.
        • Dunning M.J.
        • Speed D.
        • Lynch A.G.
        • Samarajiwa S.
        • Yuan Y.
        • Graf S.
        • Ha G.
        • Haffari G.
        • Bashashati A.
        • Russell R.
        • McKinney S.
        • Langerod A.
        • Green A.
        • Provenzano E.
        • Wishart G.
        • Pinder S.
        • Watson P.
        • Markowetz F.
        • Murphy L.
        • Ellis I.
        • Purushotham A.
        • Borresen-Dale A.L.
        • Brenton J.D.
        • Tavare S.
        • Caldas C.
        • Aparicio S.
        The genomic and transcriptomic architecture of 2,000 breast tumours reveals novel subgroups.
        Nature. 2012; 486: 346-352
        • Bartlett J.M.
        • Bloom K.J.
        • Piper T.
        • Lawton T.J.
        • van dV.
        • Ross D.T.
        • Ring B.Z.
        • Seitz R.S.
        • Beck R.A.
        • Hasenburg A.
        • Kieback D.
        • Putter H.
        • Markopoulos C.
        • Dirix L.
        • Seynaeve C.
        • Rea D.
        Mammostrat as an immunohistochemical multigene assay for prediction of early relapse risk in the tamoxifen versus exemestane adjuvant multicenter trial pathology study.
        J Clin Oncol. 2012; 30: 4477-4484
        • Schnitt S.J.
        Classification and prognosis of invasive breast cancer: from morphology to molecular taxonomy.
        Mod Pathol. 2010; 23: S60-S64
        • Yan M.
        • Parker B.A.
        • Schwab R.
        • Kurzrock R.
        HER2 aberrations in cancer: implications for therapy.
        Cancer Treat Rev. 2014; 40: 770-780
        • Blows F.M.
        • Driver K.E.
        • Schmidt M.K.
        • Broeks A.
        • van Leeuwen F.E.
        • Wesseling J.
        • et al.
        Subtyping of breast cancer by immunohistochemistry to investigate a relationship between subtype and short and long term survival: a collaborative analysis of data for 10,159 cases from 12 studies.
        PLoS Med. 2010; 7: e1000279
        • Bayraktar S.
        • Gluck S.
        Molecularly targeted therapies for metastatic triple-negative breast cancer.
        Breast Cancer Res Treat. 2013; 138: 21-35
        • Kashiwagi S.
        • Yashiro M.
        • Takashima T.
        • Aomatsu N.
        • Ikeda K.
        • Ogawa Y.
        • Ishikawa T.
        • Hirakawa K.
        Advantages of adjuvant chemotherapy for patients with triple-negative breast cancer at Stage II: usefulness of prognostic markers E-cadherin and Ki67.
        Breast Cancer Res. 2011; 13: R122
        • Engebraaten O.
        • Vollan H.K.
        • Borresen-Dale A.L.
        Triple-negative breast cancer and the need for new therapeutic targets.
        Am J Pathol. 2013; 183: 1064-1074
        • Mayer I.A.
        • Abramson V.G.
        • Lehmann B.D.
        • Pietenpol J.A.
        New strategies for triple-negative breast cancer: deciphering the heterogeneity.
        Clin Cancer Res. 2014; 20: 782-790
        • Finak G.
        • Bertos N.
        • Pepin F.
        • Sadekova S.
        • Souleimanova M.
        • Zhao H.
        • Chen H.
        • Omeroglu G.
        • Meterissian S.
        • Omeroglu A.
        • Hallett M.
        • Park M.
        Stromal gene expression predicts clinical outcome in breast cancer.
        Nat Med. 2008; 14: 518-527
        • Farmer P.
        • Bonnefoi H.
        • Anderle P.
        • Cameron D.
        • Wirapati P.
        • Becette V.
        • Andre S.
        • Piccart M.
        • Campone M.
        • Brain E.
        • Macgrogan G.
        • Petit T.
        • Jassem J.
        • Bibeau F.
        • Blot E.
        • Bogaerts J.
        • Aguet M.
        • Bergh J.
        • Iggo R.
        • Delorenzi M.
        A stroma-related gene signature predicts resistance to neoadjuvant chemotherapy in breast cancer.
        Nat Med. 2009; 15: 68-74
        • de Kruijf E.M.
        • van Nes J.G.
        • van de Velde C.J.
        • Putter H.
        • Smit V.T.
        • Liefers G.J.
        • Kuppen P.J.
        • Tollenaar R.A.
        • Mesker W.E.
        Tumor-stroma ratio in the primary tumor is a prognostic factor in early breast cancer patients, especially in triple-negative carcinoma patients.
        Breast Cancer Res Treat. 2011; 125: 687-696
        • Mendell J.T.
        • Olson E.N.
        MicroRNAs in stress signaling and human disease.
        Cell. 2012; 148: 1172-1187
        • Ventura A.
        • Jacks T.
        MicroRNAs and cancer: short RNAs go a long way.
        Cell. 2009; 136: 586-591
      1. Sempere LF, Kauppinen S: Translational Implications of MicroRNAs in Clinical Diagnostics and Therapeutics: Handbook of Cell Signaling. Edited by RA Bradshaw, EA Dennis. Oxford, Academic Press, 2009, pp 2965–2981

        • Nana-Sinkam S.P.
        • Croce C.M.
        Clinical applications for microRNAs in cancer.
        Clin Pharmacol Ther. 2013; 93: 98-104
        • Garzon R.
        • Marcucci G.
        • Croce C.M.
        Targeting microRNAs in cancer: rationale, strategies and challenges.
        Nat Rev Drug Discov. 2010; 9: 775-789
        • Ling H.
        • Fabbri M.
        • Calin G.A.
        MicroRNAs and other non-coding RNAs as targets for anticancer drug development.
        Nat Rev Drug Discov. 2013; 12: 847-865
        • Sempere L.F.
        Integrating contextual miRNA and protein signatures for diagnostic and treatment decisions in cancer.
        Expert Rev Mol Diagn. 2011; 11: 813-827
        • Sempere L.F.
        Recent advances in miRNA-based diagnostic applications.
        Expert Rev Mol Diagn. 2012; 12: 557-559
        • Radojicic J.
        • Zaravinos A.
        • Vrekoussis T.
        • Kafousi M.
        • Spandidos D.A.
        • Stathopoulos E.N.
        MicroRNA expression analysis in triple-negative (ER, PR and Her2/neu) breast cancer.
        Cell Cycle. 2011; 10: 507-517
        • Ota D.
        • Mimori K.
        • Yokobori T.
        • Iwatsuki M.
        • Kataoka A.
        • Masuda N.
        • Ishii H.
        • Ohno S.
        • Mori M.
        Identification of recurrence-related microRNAs in the bone marrow of breast cancer patients.
        Int J Oncol. 2011; 38: 955-962
        • Gong C.
        • Yao Y.
        • Wang Y.
        • Liu B.
        • Wu W.
        • Chen J.
        • Su F.
        • Yao H.
        • Song E.
        Up-regulation of miR-21 mediates resistance to trastuzumab therapy for breast cancer.
        J Biol Chem. 2011; 286: 19127-19137
        • Qian B.
        • Katsaros D.
        • Lu L.
        • Preti M.
        • Durando A.
        • Arisio R.
        • Mu L.
        • Yu H.
        High miR-21 expression in breast cancer associated with poor disease-free survival in early stage disease and high TGF-beta1.
        Breast Cancer Res Treat. 2009; 117: 131-140
        • Yan L.X.
        • Huang X.F.
        • Shao Q.
        • Huang M.Y.
        • Deng L.
        • Wu Q.L.
        • Zeng Y.X.
        • Shao J.Y.
        MicroRNA miR-21 overexpression in human breast cancer is associated with advanced clinical stage, lymph node metastasis and patient poor prognosis.
        RNA. 2008; 14: 2348-2360
        • Lee J.A.
        • Lee H.Y.
        • Lee E.S.
        • Kim I.
        • Bae J.W.
        Prognostic implications of microRNA-21 overexpression in invasive ductal carcinomas of the breast.
        J Breast Cancer. 2011; 14: 269-275
        • Anastasov N.
        • Hofig I.
        • Vasconcellos I.G.
        • Rappl K.
        • Braselmann H.
        • Ludyga N.
        • Auer G.
        • Aubele M.
        • Atkinson M.J.
        Radiation resistance due to high expression of miR-21 and G2/M checkpoint arrest in breast cancer cells.
        Radiat Oncol. 2012; 7: 206
        • Medina P.P.
        • Nolde M.
        • Slack F.J.
        OncomiR addiction in an in vivo model of microRNA-21-induced pre-B-cell lymphoma.
        Nature. 2010; 467: 86-90
        • Hatley M.E.
        • Patrick D.M.
        • Garcia M.R.
        • Richardson J.A.
        • Bassel-Duby R.
        • van R.E.
        • Olson E.N.
        Modulation of K-Ras-dependent lung tumorigenesis by MicroRNA-21.
        Cancer Cell. 2010; 18: 282-293
        • Kadera B.E.
        • Li L.
        • Toste P.A.
        • Wu N.
        • Adams C.
        • Dawson D.W.
        • Donahue T.R.
        MicroRNA-21 in pancreatic ductal adenocarcinoma tumor-associated fibroblasts promotes metastasis.
        PLoS One. 2013; 8: e71978
        • Bullock M.D.
        • Pickard K.M.
        • Nielsen B.S.
        • Sayan A.E.
        • Jenei V.
        • Mellone M.
        • Mitter R.
        • Primrose J.N.
        • Thomas G.J.
        • Packham G.K.
        • Mirnezami A.H.
        Pleiotropic actions of miR-21 highlight the critical role of deregulated stromal microRNAs during colorectal cancer progression.
        Cell Death Dis. 2013; 4: e684
        • Chung A.C.
        • Yu X.
        • Lan H.Y.
        MicroRNA and nephropathy: emerging concepts.
        Int J Nephrol Renovasc Dis. 2013; 6: 169-179
        • Cheng Y.
        • Zhang C.
        MicroRNA-21 in cardiovascular disease.
        J Cardiovasc Transl Res. 2010; 3: 251-255
        • Sempere L.F.
        • Preis M.
        • Yezefski T.
        • Ouyang H.
        • Suriawinata A.A.
        • Silahtaroglu A.
        • Conejo-Garcia J.R.
        • Kauppinen S.
        • Wells W.
        • Korc M.
        Fluorescence-based codetection with protein markers reveals distinct cellular compartments for altered MicroRNA expression in solid tumors.
        Clin Cancer Res. 2010; 16: 4246-4255
        • Nelson P.T.
        • Wilfred B.R.
        In situ hybridization is a necessary experimental complement to microRNA (miRNA) expression profiling in the human brain.
        Neurosci Lett. 2009; 466: 69-72
        • Zhang X.
        • Lu X.
        • Lopez-Berestein G.
        • Sood A.
        • Calin G.
        In situ hybridization-based detection of microRNAs in human diseases.
        microRNA Diagn Ther. 2013; 1: 12-23
        • Nielsen B.S.
        • Holmstrom K.
        Combined microRNA in situ hybridization and immunohistochemical detection of protein markers.
        Methods Mol Biol. 2013; 986: 353-365
        • Nielsen B.S.
        • Jorgensen S.
        • Fog J.U.
        • Sokilde R.
        • Christensen I.J.
        • Hansen U.
        • Brunner N.
        • Baker A.
        • Moller S.
        • Nielsen H.J.
        High levels of microRNA-21 in the stroma of colorectal cancers predict short disease-free survival in stage II colon cancer patients.
        Clin Exp Metastasis. 2011; 28: 27-38
        • Sempere L.F.
        • Christensen M.
        • Silahtaroglu A.
        • Bak M.
        • Heath C.V.
        • Schwartz G.
        • Wells W.
        • Kauppinen S.
        • Cole C.N.
        Altered MicroRNA expression confined to specific epithelial cell subpopulations in breast cancer.
        Cancer Res. 2007; 67: 11612-11620
        • Jorgensen S.
        • Baker A.
        • Moller S.
        • Nielsen B.S.
        Robust one-day in situ hybridization protocol for detection of microRNAs in paraffin samples using LNA probes.
        Methods. 2010; 52: 375-381
        • Rask L.
        • Balslev E.
        • Jorgensen S.
        • Eriksen J.
        • Flyger H.
        • Moller S.
        • Hogdall E.
        • Litman T.
        • Schnack N.B.
        High expression of miR-21 in tumor stroma correlates with increased cancer cell proliferation in human breast cancer.
        APMIS. 2011; 119: 663-673
        • Altman D.G.
        • McShane L.M.
        • Sauerbrei W.
        • Taube S.E.
        Reporting Recommendations for Tumor Marker Prognostic Studies (REMARK): explanation and elaboration.
        PLoS Med. 2012; 9: e1001216
        • McShane L.M.
        • Altman D.G.
        • Sauerbrei W.
        • Taube S.E.
        • Gion M.
        • Clark G.M.
        Reporting recommendations for tumor marker prognostic studies (REMARK).
        J Natl Cancer Inst. 2005; 97: 1180-1184
        • Carey L.A.
        • Perou C.M.
        • Livasy C.A.
        • Dressler L.G.
        • Cowan D.
        • Conway K.
        • Karaca G.
        • Troester M.A.
        • Tse C.K.
        • Edmiston S.
        • Deming S.L.
        • Geradts J.
        • Cheang M.C.
        • Nielsen T.O.
        • Moorman P.G.
        • Earp H.S.
        • Millikan R.C.
        Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study.
        JAMA. 2006; 295: 2492-2502
        • Rakha E.A.
        • Reis-Filho J.S.
        • Baehner F.
        • Dabbs D.J.
        • Decker T.
        • Eusebi V.
        • Fox S.B.
        • Ichihara S.
        • Jacquemier J.
        • Lakhani S.R.
        • Palacios J.
        • Richardson A.L.
        • Schnitt S.J.
        • Schmitt F.C.
        • Tan P.H.
        • Tse G.M.
        • Badve S.
        • Ellis I.O.
        Breast cancer prognostic classification in the molecular era: the role of histological grade.
        Breast Cancer Res. 2010; 12: 207
        • Volinia S.
        • Calin G.A.
        • Liu C.G.
        • Ambs S.
        • Cimmino A.
        • Petrocca F.
        • Visone R.
        • Iorio M.
        • Roldo C.
        • Ferracin M.
        • Prueitt R.L.
        • Yanaihara N.
        • Lanza G.
        • Scarpa A.
        • Vecchione A.
        • Negrini M.
        • Harris C.C.
        • Croce C.M.
        A microRNA expression signature of human solid tumors defines cancer gene targets.
        Proc Natl Acad Sci U S A. 2006; 103: 2257-2261
        • Kumarswamy R.
        • Volkmann I.
        • Thum T.
        Regulation and function of miRNA-21 in health and disease.
        RNA Biol. 2011; 8: 706-713
        • Perou C.M.
        • Sorlie T.
        • Eisen M.B.
        • van de R.M.
        • Jeffrey S.S.
        • Rees C.A.
        • Pollack J.R.
        • Ross D.T.
        • Johnsen H.
        • Akslen L.A.
        • Fluge O.
        • Pergamenschikov A.
        • Williams C.
        • Zhu S.X.
        • Lonning P.E.
        • Borresen-Dale A.L.
        • Brown P.O.
        • Botstein D.
        Molecular portraits of human breast tumours.
        Nature. 2000; 406: 747-752
        • Sorlie T.
        • Perou C.M.
        • Tibshirani R.
        • Aas T.
        • Geisler S.
        • Johnsen H.
        • Hastie T.
        • Eisen M.B.
        • van de R.M.
        • Jeffrey S.S.
        • Thorsen T.
        • Quist H.
        • Matese J.C.
        • Brown P.O.
        • Botstein D.
        • Eystein L.P.
        • Borresen-Dale A.L.
        Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications.
        Proc Natl Acad Sci U S A. 2001; 98: 10869-10874
        • Kristensen V.N.
        • Vaske C.J.
        • Ursini-Siegel J.
        • Van L.P.
        • Nordgard S.H.
        • Sachidanandam R.
        • Sorlie T.
        • Warnberg F.
        • Haakensen V.D.
        • Helland A.
        • Naume B.
        • Perou C.M.
        • Haussler D.
        • Troyanskaya O.G.
        • Børresen-Dale A.L.
        Integrated molecular profiles of invasive breast tumors and ductal carcinoma in situ (DCIS) reveal differential vascular and interleukin signaling.
        Proc Natl Acad Sci U S A. 2012; 109: 2802-2807
        • Kreike B.
        • van Kouwenhove M.
        • Horlings H.
        • Weigelt B.
        • Peterse H.
        • Bartelink H.
        • van de Vijver M.J.
        Gene expression profiling and histopathological characterization of triple-negative/basal-like breast carcinomas.
        Breast Cancer Res. 2007; 9: R65
        • Lehmann B.D.
        • Bauer J.A.
        • Chen X.
        • Sanders M.E.
        • Chakravarthy A.B.
        • Shyr Y.
        • Pietenpol J.A.
        Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies.
        J Clin Invest. 2011; 121: 2750-2767
        • Kjaer-Frifeldt S.
        • Hansen T.F.
        • Nielsen B.S.
        • Joergensen S.
        • Lindebjerg J.
        • Soerensen F.B.
        • dePont C.R.
        • Jakobsen A.
        The prognostic importance of miR-21 in stage II colon cancer: a population-based study.
        Br J Cancer. 2012; 107: 1169-1174
        • Buscaglia L.E.
        • Li Y.
        Apoptosis and the target genes of microRNA-21.
        Chin J Cancer. 2011; 30: 371-380
        • Yuan Y.
        • Failmezger H.
        • Rueda O.M.
        • Ali H.R.
        • Graf S.
        • Chin S.F.
        • Schwarz R.F.
        • Curtis C.
        • Dunning M.J.
        • Bardwell H.
        • Johnson N.
        • Doyle S.
        • Turashvili G.
        • Provenzano E.
        • Aparicio S.
        • Caldas C.
        • Markowetz F.
        Quantitative image analysis of cellular heterogeneity in breast tumors complements genomic profiling.
        Sci Transl Med. 2012; 4: 157ra43
        • Beck A.H.
        • Sangoi A.R.
        • Leung S.
        • Marinelli R.J.
        • Nielsen T.O.
        • van dV.
        • West R.B.
        • van de R.M.
        • Koller D.
        Systematic analysis of breast cancer morphology uncovers stromal features associated with survival.
        Sci Transl Med. 2011; 3: 108ra13
        • Moorman A.M.
        • Vink R.
        • Heijmans H.J.
        • van der P.J.
        • Kouwenhoven E.A.
        The prognostic value of tumour-stroma ratio in triple-negative breast cancer.
        Eur J Surg Oncol. 2012; 38: 307-313