http://orcid.org/0000-0002-5710-1692
Figures
Abstract
Purpose
Black/African American (AA) women are twice as likely to be diagnosed with triple negative breast cancer (TNBC) compared to whites, an aggressive breast cancer subtype associated with poor prognosis. There are no routinely used targeted clinical therapies for TNBC; thus there is a clear need to identify prognostic markers and potential therapeutic targets.
Methods
We evaluated expression of 27,016 genes in 155 treatment-naïve TN tumors from AA women in Detroit. Associations with survival were evaluated using Cox proportional hazards models adjusting for stage and age at diagnosis, and p-values were corrected using a false discovery rate. Our validation sample consisted of 494 TN tumors using four publically available data sets. Meta-analyses were performed using summary statistics from the four validation results.
Results
In the Detroit AA cohort, CLCA2 [Hazard ratio (HR) = 1.56, 95% confidence interval (CI) 1.31–1.86, nominal p = 5.1x10-7, FDR p = 0.014], SPIC [HR = 1.47, 95%CI 1.26–1.73, nominal p = 1.8x10-6, FDR p = 0.022], and MIR4311 [HR = 1.57, 95% CI 1.31–1.92, nominal p = 2.5x10-5, FDR p = 0.022] expression were associated with overall survival. Further adjustment for treatment and breast cancer specific survival analysis did not substantially alter effect estimates. CLCA2 was also associated with increased risk of death in the validation cohorts [HR = 1.14, 95% CI 1.05–1.24, p = 0.038, p-heterogeneity = 0.88].
Citation: Purrington KS, Knight J III, Dyson G, Ali-Fehmi R, Schwartz AG, Boerner JL, et al. (2020) CLCA2 expression is associated with survival among African American women with triple negative breast cancer. PLoS ONE 15(4): e0231712. https://doi.org/10.1371/journal.pone.0231712
Editor: Qing-Xiang Amy Sang, Florida State University, UNITED STATES
Received: December 17, 2019; Accepted: March 30, 2020; Published: April 16, 2020
Copyright: © 2020 Purrington et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All raw data and processed files used for analyses are available from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/, accession number(s) GSE142102).
Funding: KSP: American Cancer Society Institutional Research Grant 11-053-01-IRG https://www.cancer.org/ AGS: National Institutes of Health P30CA022453 https://www.nih.gov/ AGS: National Institutes of Health HHSN261201300011I https://www.nih.gov/ The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Triple negative breast cancers are a distinct histopathologic subtype of breast cancer (BC) that accounts for approximately 15% of all invasive BCs [1, 2] where <1% of tumors cells stain positive for estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor-2 (HER2) protein expression. Clinical outcomes among TNBC patients have a unique pattern, including a peak risk of death and recurrence within the first three years following treatment [3] followed by a 50% decrease in risk beyond five years compared to hormone receptor-positive BC patients [4]. There are no routinely used targeted clinical therapies for TNBC or effective approaches for reducing high cancer mortality among this subtype, partly because we have yet to identify the underlying etiologic factors. Thus, there is a clear need to better identify the molecular processes in TN tumors related to clinical outcomes that could be explored as therapeutic targets.
Black/African American (AA) women are twice as likely to be diagnosed with TNBC compared to white/European American women [5]. Poor TNBC survival contributes to the racial disparity in overall breast cancer outcomes because are approximately 40% more likely to die from overall BC as white women, due in large part to the increased incidence of aggressive TNBC among AA women [6]. Several studies further suggest that AA women with TNBC specifically experience poorer clinical outcomes compared to white women [7–12], although this relationship is less clear. While AA women would receive substantial benefit from targeted TNBC therapies, few studies exist evaluating molecular processes related to survival specifically in TN tumors from AA women.
Evaluation of the gene expression profiles of TN tumors is a valuable tool for evaluating prognostic biomarkers. This was first robustly demonstrated through the identification of four main intrinsic breast cancer subtypes with important prognostic implications, now evaluated using the PAM50 expression microarray: two luminal epithelial groups (A and B); a HER2 over-expressing group; and a basal-like group that is largely TNBC [12, 13]. Although the majority of TN tumors are classified as basal-like (80–85%), they have been shown to have significant biological heterogeneity [14]. Consensus across multiple TNBC subtyping studies appears to classify TN tumors into 3 to 4 molecular subtypes: basal-like, immune enriched, mesenchymal (M), and luminal androgen receptor (LAR) [13]. With the exception of the immune-enriched subtype, which consistently is associated with better survival across studies, there is discrepancy in the prognostic implications across subtypes [14]. While expression of individual genes or gene signatures have been evaluated with respect to clinical outcomes [15–24], none of these are used clinically to identify patients at higher risk or recurrence or death or to guide treatment decision making. Thus, there remains a need to identify strong prognostic biomarkers in TNBC that can be used, particularly in AA women who have a higher burden of this aggressive breast cancer subtype. Here we evaluated associations between gene expression and survival in AA women with TNBC to identify potential prognostic factors and potential therapeutic targets.
Methods
Detroit AA cohort and sample selection
Inclusion criteria for the TNBC cohort required that participants were (1) African American, (2) female, (3) diagnosed with primary invasive breast cancer, (4) negative for ER, PR, and HER2, and (5) underwent surgery format the Karmanos Cancer Center in Detroit, MI from 2004–2013. Women meeting inclusion criteria were identified by the Karmanos Cancer Institute (KCI) Epidemiology Research Core using registry data from the Metropolitan Detroit Cancer Surveillance System (MDCSS). ER and PR status was recorded in MDCSS data while HER2 status was determined by pathology review for cases diagnosed prior to 2010 and using MDCSS data for cases diagnosed 2010 or later. Tumor blocks were identified and retrieved by the Karmanos Cancer Institute Biobanking and Correlative Sciences Core. Clinical data (stage, grade, age at diagnosis), treatment data (surgery type, first line systemic therapy type, radiation, sequence of surgery and first line systemic therapy), and outcomes data (vital status at last contact, cause of death, and active follow-up time) were obtained via linkage with the MDCSS registry. Among 239 eligible patients, tumor samples were obtained for 226 of these patients for expression profiling and 155 treatment-naïve tumors were utilized for analysis. This study was approved for exemption by Wayne State University Institutional Review Board.
Detroit AA cohort tumor processing & expression profiling
Formalin-fixed paraffin-embedded (FFPE) tumor blocks were processed in two batches. For all tumors, hematoxylin and eosin (H&E) slides were created and unstained tissue curls were cut from four 10mm unstained slides. Tissue curls were generated to correspond to the pathologist-defined tumor area and collected in DNAse/RNAse free microcentrifuge tubes. Batch 1 total RNA (n = 60) was extracted using the QIASymphony Automated system (Qiagen, Germany) and Batch 2 RNA (n = 166) was extracted using the Qiagen RNeasy FFPE Kit according to the manufacturer protocol. All tumors were profiled using Affymetrix Human Gene ST 2.0 arrays after amplification of RNA using the Affymetrix WT Pico Kit (Santa Clara, CA) in two batches (Batch 1 n = 60 tumors, Batch 2 n = 166 tumors). Raw probe intensity data were exported for statistical analysis. Raw and normalized, log-2 transformed expression data and outcomes data for these 226 tumors are publically available in the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) (accession number GSE142102).
Validation datasets
We downloaded mRNA expression data for 1,084 breast tumors from The Cancer Genome Atlas (TCGA, PanCancer Atlas) and 318 triple negative breast tumors from the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) using cBioPortal (http://www.cbioportal.org/). We curated relevant datasets from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) by searching for “breast cancer expression” and “survival” and filtering results to include on entries for human primary breast cancers (non-cell line) with expression profiling by array or high throughput sequencing (n = 14). We then excluded datasets that did not have appropriate survival information (follow-up time or vital status) or included only patients treated with neoadjuvant chemotherapy (n = 9). Finally, we restricted our analyses to include datasets with at least ten triple negative breast cancers, resulting in two usable GEO datasets: GSE35629-GPL1390 and GSE69031. All expression data were downloaded as Z scores from either RNA sequencing data (TCGA: batch normalized/merged from Illumina HiSeq_RNASeqV2 data) or expression microarrays (METABRIC, GEO datasets). Data were not available for MIR4311 in any of the four validation datasets. For TCGA, corresponding clinical and demographic data were obtained from the Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/). For METABRIC and GEO datasets, clinical and demographic data were obtained simultaneously with the expression data.
We subset our analyses to 494 TNBC (12 GSE35629-GPL1390, 21 GSE69031, 158 TCGA, 303 METABRIC) defined by ER, PR, and HER2 negative status with available data for survival (vital status, survival time), age, and stage at diagnosis. METABRIC and GEO tumors were considered TNBC when indicated negative by ER, PR, and HER2 status. TCGA Tumors were considered ER negative when negative by IHC staining, PR negative when negative by IHC staining, and HER2 negative when both 1) negative by either IHC or FISH staining and 2) not positive for IHC staining, FISH staining, or copy number status.
Statistical methods
All statistical analyses were performed in R (https://cran.r-project.org/). Raw probe intensity data from the Detroit AA cohort were normalized separately by batch as implemented by the “rma” function to perform background subtraction, quantile normalization, summarization of probe sets using median-polish, and log2-transformation. We evaluated batch effects using principal components analysis as implemented by the “princomp function” (S1 Fig). Batch effects were corrected by standardization of probes (subtracting the mean expression value and dividing by standard deviation) by batch (Batch 1 n = 60, Batch 2 n = 166) [25]. There was no evidence for batch effects after standardization (S1 Fig). Differences in expression of significant genes in the Detroit AA cohort was evaluated using both t-tests and Wilcox rank sum tests (S2 Fig). We selected only the 155 treatment naïve tumors and 27,016 annotated gene transcript probes for analysis, which were combined into a single dataset for subsequent analyses.
For the Detroit AA cohort, overall survival was evaluated in Cox proportional hazards (CoxPH) models as implemented in the “survival” package and all models were adjusted for stage and age at diagnosis. Grade was homogenous in the cohort and surgery type was strongly correlated with stage, so these variables were not included as covariates. For the initial Detroit AA cohort survival analysis of 27,016 genes, nominal p-values were corrected using the false discovery rate (FDR) method using the “p.adjust” function, and FDR-corrected p-values <0.05 were considered statistically significant. We performed additional adjustment for statistically significant genes to evaluate chemotherapy (none vs. adjuvant) and radiation therapy (none vs. adjusted) as potential confounders. We also evaluated breast cancer specific survival for the Detroit AA cohort using CoxPH models adjusting for stage, age at diagnosis, chemotherapy, and radiation therapy.
For the validation cohorts, overall survival was evaluated using CoxPH models adjusting for age and stage. Two TCGA values (Z score>20) for SPIC were excluded as outliers, which were from Asian women and exclusion did not substantially affect the overall association. Associations were estimated within each of the four individual validation cohorts using all participants as well as subset to African American and white participants. We then performed meta-analyses of the age- and stage-adjusted effect estimates from the validation cohorts for both CLCA2 and SPIC to obtain a summary effect estimate, 95% confidence intervals, and p-value from tests of heterogeneity of effects for the overall, African American, and white participants. Based on a threshold of p<0.10 for the test of heterogeneity, a fixed effects model was used for all CLCA2 and SPIC meta-analyses except for the SPIC overall analysis, where a random effects models was used.
Results
We evaluated expression of 27,016 well-annotated genes in 155 treatment-naïve TN tumors from the Detroit AA Cohort (Table 1). Approximately half of these women were <55 years of age at diagnosis. Almost 90% of tumors were intraductal carcinomas, and the vast majority of women presented with either localized (63.2%) or regional (31.0%) TNBC. Nearly all (96.8%) tumors were either moderately or poorly differentiated. There was considerable variation in treatment decisions, with slightly more women choosing breast-conserving surgery compared to mastectomy. A quarter of women did not receive adjuvant chemotherapy and about a third of women did not undergo adjuvant radiation therapy, although 92% of women received at least one of these treatments. Approximately 36% of the 155 women died during a mean follow-up time of 5.5 years, where ~60% of deaths were due to breast cancer.
Among the 27,016 genes evaluated for association with overall survival adjusting for stage and grade (S1 Table), three genes remained statistically after FDR correction (FDR p<0.05) (Table 2, Fig 1). CLCA2 (Chloride Channel Accessory 2) expression was associated with a 56% increased risk of death [Hazard ratio (HR) = 1.56, 95% confidence interval (CI) 1.31–1.86, nominal p = 5.1x10-7, FDR p = 0.014], SPIC (Spi-C Transcription Factor) expression was associated with a 47% increase in mortality [95%CI 1.26–1.73, nominal p = 1.8x10-6, FDR p = 0.022], and MIR4311 expression was associated with a 59% increased risk of death [95% CI 1.31–1.92, nominal p = 2.5x10-5, FDR p = 0.022]. Additional adjustment for adjuvant chemotherapy and radiation did not change the effect estimates for any of these three genes (Table 2). We also evaluated associations with breast cancer-specific survival using the fully adjusted model for stage, age, and treatment (Table 2). While statistical significance was slightly reduced due to the decrease in the number of events (33 breast cancer deaths compared to 55 deaths from any cause), the effects estimates for all three genes remained stable compared to the overall survival analysis [CLCA2 HR = 1.66, SPIC HR = 1.52, MIR4311 HR = 1.52].
Kaplan−Meier plots for overall survival by (a) CLCA2 quartile, (b) SPIC quartile, and (c) MIR4311 quartile among 155 treatment-naïve TN tumors (56 deaths) in the Detroit AA cohort. Solid lines represent curves for Q1, dashed lines represent curves for Q2, dotted lines represent curves for Q3, and dash-dotted lines represent curves for Q4 in each panel.
We next sought to validate these associations using publically available expression data from TN breast tumors in TCGA (n = 158), METABRIC (n = 303), GSE35629-GPL1390 (n = 12), and GSE69031 (n = 21). Only CLCA2 expression data was available for all four validation cohorts, and SPIC expression data was available for only three (GSE69031, TCGA, METABRIC). While CLCA2 was not significantly associated with survival in the four individual validation datasets, the magnitude and direction of effect were consistent with the association observed for CLCA2 in the Detroit cohort (Fig 2). Indeed, CLCA2 was significantly associated with overall survival in the validation meta-analysis [HR = 1.14, 95% CI 1.05–1.24, p = 0.038, p-heterogeneity = 0.88]. SPIC was not significantly associated with overall survival in individual analyses or in the meta-analysis. We were only able to estimate race-specific hazard ratios using TCGA data, which has a substantially reduced AA sample size and number of deaths compared to the Detroit AA cohort (54 AA cases, 11 deaths). While not statistically significant, the hazard ratio estimates for both CLCA2 and SPIC were comparable to those seen in the Detroit AA cohort [CLCA2: HR = 1.41, 95% CI 0.77–2.58; SPIC: HR = 1.24, 95% CI 0.52–2.97]. The CLCA2 association with survival was similar among European Americans in TCGA (92 EA cases, 13 deaths) [HR = 1.26, 95% CI 0.82–1.93], while SPIC was associated with a non-significant reduction in death among European Americans [HR = 0.46, 95% CI 0.063–3.31].
Forest plots for analyses of GSE35629-GPL1390, GSE69031, TCGA, and METABRIC are shown for CLCA2 and SPIC. Only overall, rather than race-specific, analyses are shown because of small sample size (GPL1390, GSE69031) or lack of race data (METABRIC). All estimates are adjusted for age and stage at diagnosis. Study-specific hazard ratios (HR) are denoted by black boxes and 95% confidence intervals (CI) are denoted by corresponding black lines. Box heights are inversely proportional to precision of the HR estimate as influenced by sample size, such that a larger HR box indicates larger sample size and better precision. Summary estimates are denoted as diamonds, where the width of the diamond corresponds to the 95% CI. Estimates with confidence intervals that do not overlap the null line at 1.0 indicate significance at p<0.05.
Discussion
Here we evaluated potential prognostic biomarkers for TNBC by analyzing associations between transcriptome-wide tumor expression profiles and survival in 155 treatment-naïve tumors from AA women with TNBC in Detroit. Three genes (CLCA2, SPIC, MIR4311) were associated with overall and breast cancer specific survival in this discovery cohort. While associations between CLCA2 and SPIC in validation cohorts were not statistically significant, meta-analysis revealed that CLCA2 was significantly associated with overall survival when combining four TNBC validation cohorts.
CLCA2 encodes a calcium-activated chloride channel regulator family member, proteins that regulate the transport of chloride across the plasma membrane. CLCA2 confers an antiproliferative role, where expression is upregulated by the tumor suppressor protein p53 in response to DNA damage [26]. CLCA2 expression has been found to be downregulated in breast tumors, whereas CLCA2 expression is known to inhibit migration and invasion while simultaneously promoting mesenchymal-to-epithelial transition in cancer cell lines [27–30]. Interestingly, we found that increased expression of this putative tumor suppressor was associated with worse survival. Despite this in vitro evidence for a favorable prognostic role of CLCA2 in cancers, very few studies have been reported associations between CLCA2 expression in cancer with respect to clinical outcomes in humans [31, 32]. The one study that reported associations between CLCA2 and survival in cancer reported worse disease-free survival associated with increased CLCA2 expression in early stage lung adenocarcinoma [31], which is comparable to our findings. While the CLCA2 has clear implications for invasion and metastasis in breast cancer, the mechanism by which increased CLCA2 tumor expression may be related to worse clinical outcomes for women with TNBC is unclear.
SPIC is a transcription factor that controls the development of red pulp macrophages, splenic macrophages which are critical for blood homeostasis via red blood cell recycling and iron homeostasis [33]. SPIC has also been shown to act as a lymphoid-specific enhancer and regulates VCAM1 [34], a gene that has been associated with progression, angiogenesis, and metastasis in breast cancer [35]. VCAM1 is also critical for macrophage-mediated retention of hematopoietic stem cells in the spleen, and there is evidence that these splenic stem cells are a continuous source of tumor associated macrophages throughout tumor progression [36]. SPIC is also involved in the genomic stability of pre-B cells, where DNA double strand breaks were found to inhibit pre-B cell receptor signaling through induction of SPIC [37]. While few reports of the relevance of SPIC to cancer development or progression exist, one study found that SPIC transcription factor binding sites were enriched among lncRNAs found to be involved in the neuroendocrine transdifferentiation process through which Treatment-induced neuroendocrine prostate cancers arise [38]. Given the known importance of tumor infiltrating lymphocytes to breast cancer overall and TNBC specifically [39–41], the mechanisms relating SPIC, tumor associated macrophages, and B cells to tumor progression should be further explored.
Much less is known about the function or expression targets of MIR4311, a microRNA gene located on chromosome 15q22. However, using the NCBI Phenotype-Genotype Integrator (https://www.ncbi.nlm.nih.gov/gap/phegeni), an intergenic variant in the MIR4311/DIS3L region (rs6494560) was associated with obesity (4.917 x 10–6), waist circumference (4.798 x 10–5), and body mass index (4.863 x 10–5) in the Family Heart Study genome wide association study [42]. Obesity is a known risk factor for TNBC [43], and there is some evidence that overweight or obesity is related to TNBC progression [44].
Here we identified CLCA2 as a potential prognostic marker for TNBC. While we were somewhat limited in our ability to validate race-specific findings in TCGA due to small race-specific sample sizes and lack of treatment data, we were able to show that CLCA2 is associated with increased risk of death for TNBC considering the discovery and validation data sets. We were also unable to directly evaluate racial comparisons in effects in our Detroit cohort due to our selection criteria, which will be important to evaluate in future studies. It is also possible due to limited sample size that we were unable to detect associations between expression of additional important genes and survival. Nevertheless, our findings suggest potential new mechanisms for TN tumor progression and identify possible new therapeutic targets that are relevant for AA women.
Supporting information
S1 Table. Associations between 27,016 annotated genes and survival in the Detroit AA cohort.
https://doi.org/10.1371/journal.pone.0231712.s001
(XLSX)
S2 Table. Associations between CLCA2 and SPIC expression and stage in TCGA.
https://doi.org/10.1371/journal.pone.0231712.s002
(XLSX)
S1 Fig. Principal components analysis of batch effects in Detroit AA cohort.
Plots of the first two principal components are shown pre- and post-standardization. Red dots correspond to batch 1 samples and black dots correspond to batch 2 samples.
https://doi.org/10.1371/journal.pone.0231712.s003
(TIFF)
S2 Fig. Boxplots of CLCA2, SPIC, and MIR4311 expression by batch.
Boxplots of CLCA2, SPIC, and MIR4311 expression are shown by batch. The bold line within each boxplot represents median expression, and the upper and lower bounds of the boxplot represent the 75th and 25th percentiles, respectively. Outer edges reflect the 10th (lower) and 90th (upper) percentiles, respectively, and outliers are denoted in shaded circles.
https://doi.org/10.1371/journal.pone.0231712.s004
(TIFF)
References
- 1. Foulkes WD, Smith IE, Reis-Filho JS. Triple-negative breast cancer. N Engl J Med. 2010 Nov 11;363(20):1938–48. pmid:21067385. Epub 2010/11/12. eng.
- 2.
American Cancer Society. Breast Cancer Facts & Figures, 2011–2012. Atlanta, GA: American Cancer Society, 2012.
- 3. Dent R, Trudeau M, Pritchard KI, Hanna WM, Kahn HK, Sawka CA, et al. Triple-negative breast cancer: clinical features and patterns of recurrence. Clin Cancer Res. 2007 Aug 1;13(15 Pt 1):4429–34. pmid:17671126. Epub 2007/08/03. eng.
- 4. Nofech-Mozes S, Trudeau M, Kahn HK, Dent R, Rawlinson E, Sun P, et al. Patterns of recurrence in the basal and non-basal subtypes of triple-negative breast cancers. Breast Cancer Res Treat. 2009 Nov;118(1):131–7. pmid:19189211. Epub 2009/02/04. eng.
- 5. Millikan RC, Newman B, Tse CK, Moorman PG, Conway K, Dressler LG, et al. Epidemiology of basal-like breast cancer. Breast Cancer Res Treat. 2008 May;109(1):123–39. pmid:17578664. Pubmed Central PMCID: 2443103. Epub 2007/06/21. eng.
- 6. Kohler BA, Sherman RL, Howlader N, Jemal A, Ryerson AB, Henry KA, et al. Annual Report to the Nation on the Status of Cancer, 1975–2011, Featuring Incidence of Breast Cancer Subtypes by Race/Ethnicity, Poverty, and State. J Natl Cancer Inst. 2015 Jun;107(6):djv048. pmid:25825511. Pubmed Central PMCID: 4603551.
- 7. Komenaka IK, Martinez ME, Pennington RE Jr., Hsu CH, Clare SE, Thompson PA, et al. Race and ethnicity and breast cancer outcomes in an underinsured population. J Natl Cancer Inst. 2010 Aug 4;102(15):1178–87. pmid:20574040.
- 8. Christiansen N, Chen L, Gilmore J, Pechar D, Szabo S. Association between African American race and outcomes in patients with nonmetastatic triple-negative breast cancer: a retrospective analysis by using results from the Georgia Cancer Specialist Database. Clinical breast cancer. 2012 Aug;12(4):270–5. pmid:22683281.
- 9. Chu QD, Henderson AE, Ampil F, Li BD. Outcome for patients with triple-negative breast cancer is not dependent on race/ethnicity. International journal of breast cancer. 2012;2012:764570. pmid:22645687. Pubmed Central PMCID: 3356882.
- 10. Dawood S, Broglio K, Kau SW, Green MC, Giordano SH, Meric-Bernstam F, et al. Triple receptor-negative breast cancer: the effect of race on response to primary systemic treatment and survival outcomes. J Clin Oncol. 2009 Jan 10;27(2):220–6. pmid:19047281.
- 11. Bauer KR, Brown M, Cress RD, Parise CA, Caggiano V. Descriptive analysis of estrogen receptor (ER)-negative, progesterone receptor (PR)-negative, and HER2-negative invasive breast cancer, the so-called triple-negative phenotype: a population-based study from the California cancer Registry. Cancer. 2007 May 1;109(9):1721–8. pmid:17387718. Epub 2007/03/28. eng.
- 12. Lund MJ, Trivers KF, Porter PL, Coates RJ, Leyland-Jones B, Brawley OW, et al. Race and triple negative threats to breast cancer survival: a population-based study in Atlanta, GA. Breast Cancer Res Treat. 2009 Jan;113(2):357–70. pmid:18324472.
- 13. Lehmann BD, Bauer JA, Chen X, Sanders ME, Chakravarthy AB, Shyr Y, et al. Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies. J Clin Invest. 2011 Jul 1;121(7):2750–67. pmid:21633166. Pubmed Central PMCID: 3127435. Epub 2011/06/03. eng.
- 14. Ahn SG, Kim SJ, Kim C, Jeong J. Molecular Classification of Triple-Negative Breast Cancer. Journal of breast cancer. 2016 Sep;19(3):223–30. pmid:27721871. Pubmed Central PMCID: 5053306.
- 15. Wang S, Beeghly-Fadiel A, Cai Q, Cai H, Guo X, Shi L, et al. Gene expression in triple-negative breast cancer in relation to survival. Breast Cancer Res Treat. 2018 Aug;171(1):199–207. pmid:29748761.
- 16. Jin Z, Xu L, Zhang L, Zhao M, Li D, Ye L, et al. Interleukin enhancer binding factor 2 is a prognostic biomarker for breast cancer that also predicts neoadjuvant chemotherapy responses. American journal of translational research. 2018;10(6):1677–89. pmid:30018709.
- 17. Zhang M, Zhang X, Zhao S, Wang Y, Di W, Zhao G, et al. Prognostic value of survivin and EGFR protein expression in triple-negative breast cancer (TNBC) patients. Targeted oncology. 2014 Dec;9(4):349–57. pmid:24233638.
- 18. Yue Y, Astvatsaturyan K, Cui X, Zhang X, Fraass B, Bose S. Stratification of Prognosis of Triple-Negative Breast Cancer Patients Using Combinatorial Biomarkers. PloS one. 2016;11(3):e0149661. pmid:26930401. Pubmed Central PMCID: 4773063.
- 19. Lu L, Mao X, Shi P, He B, Xu K, Zhang S, et al. MicroRNAs in the prognosis of triple-negative breast cancer: A systematic review and meta-analysis. Medicine. 2017 Jun;96(22):e7085. pmid:28562579. Pubmed Central PMCID: 5459744.
- 20. Liu Y, Baglia M, Zheng Y, Blot W, Bao PP, Cai H, et al. ALDH1A1 mRNA expression in association with prognosis of triple-negative breast cancer. Oncotarget. 2015 Dec 1;6(38):41360–9. pmid:26462023. Pubmed Central PMCID: 4747411.
- 21. Baglia ML, Cai Q, Zheng Y, Wu J, Su Y, Ye F, et al. Dual specificity phosphatase 4 gene expression in association with triple-negative breast cancer outcome. Breast Cancer Res Treat. 2014 Nov;148(1):211–20. pmid:25281216. Pubmed Central PMCID: 4200532.
- 22. Kim JY, Jung HH, Do IG, Bae S, Lee SK, Kim SW, et al. Prognostic value of ERBB4 expression in patients with triple negative breast cancer. BMC cancer. 2016 Feb 22;16:138. pmid:26907936. Pubmed Central PMCID: 4763414.
- 23. Park YH, Jung HH, Do IG, Cho EY, Sohn I, Jung SH, et al. A seven-gene signature can predict distant recurrence in patients with triple-negative breast cancers who receive adjuvant chemotherapy following surgery. Int J Cancer. 2015 Apr 15;136(8):1976–84. pmid:25537444.
- 24. Lee HJ, Lee JJ, Song IH, Park IA, Kang J, Yu JH, et al. Prognostic and predictive value of NanoString-based immune-related gene signatures in a neoadjuvant setting of triple-negative breast cancer: relationship to tumor-infiltrating lymphocytes. Breast Cancer Res Treat. 2015 Jun;151(3):619–27. pmid:26006068.
- 25. Purrington KS, Visscher DW, Wang C, Yannoukakos D, Hamann U, Nevanlinna H, et al. Genes associated with histopathologic features of triple negative breast tumors predict molecular subtypes. Breast Cancer Res Treat. 2016 May;157(1):117–31. pmid:27083182.
- 26. Sasaki Y, Koyama R, Maruyama R, Hirano T, Tamura M, Sugisaka J, et al. CLCA2, a target of the p53 family, negatively regulates cancer cell migration and invasion. Cancer Biol Ther. 2012 Dec;13(14):1512–21. pmid:22990203. Pubmed Central PMCID: 3542243.
- 27. Gruber AD, Pauli BU. Tumorigenicity of human breast cancer is associated with loss of the Ca2+-activated chloride channel CLCA2. Cancer Res. 1999 Nov 1;59(21):5488–91. pmid:10554024.
- 28. Li X, Cowell JK, Sossey-Alaoui K. CLCA2 tumour suppressor gene in 1p31 is epigenetically regulated in breast cancer. Oncogene. 2004 Feb 19;23(7):1474–80. pmid:14973555.
- 29. Walia V, Ding M, Kumar S, Nie D, Premkumar LS, Elble RC. hCLCA2 Is a p53-Inducible Inhibitor of Breast Cancer Cell Proliferation. Cancer Res. 2009 Aug 15;69(16):6624–32. pmid:19654313. Pubmed Central PMCID: 2745301.
- 30. Walia V, Yu Y, Cao D, Sun M, McLean JR, Hollier BG, et al. Loss of breast epithelial marker hCLCA2 promotes epithelial-to-mesenchymal transition and indicates higher risk of metastasis. Oncogene. 2012 Apr 26;31(17):2237–46. pmid:21909135. Pubmed Central PMCID: 4154589.
- 31. Ludovini V, Bianconi F, Siggillino A, Piobbico D, Vannucci J, Metro G, et al. Gene identification for risk of relapse in stage I lung adenocarcinoma patients: a combined methodology of gene expression profiling and computational gene network analysis. Oncotarget. 2016 May 24;7(21):30561–74. pmid:27081700. Pubmed Central PMCID: 5058701.
- 32. Qiang YY, Li CZ, Sun R, Zheng LS, Peng LX, Yang JP, et al. Along with its favorable prognostic role, CLCA2 inhibits growth and metastasis of nasopharyngeal carcinoma cells via inhibition of FAK/ERK signaling. Journal of experimental & clinical cancer research: CR. 2018 Feb 20;37(1):34. pmid:29463274. Pubmed Central PMCID: 5819171.
- 33. Kurotaki D, Uede T, Tamura T. Functions and development of red pulp macrophages. Microbiology and immunology. 2015 Feb;59(2):55–62. pmid:25611090.
- 34. Haldar M, Kohyama M, So AY, Kc W, Wu X, Briseno CG, et al. Heme-mediated SPI-C induction promotes monocyte differentiation into iron-recycling macrophages. Cell. 2014 Mar 13;156(6):1223–34. pmid:24630724. Pubmed Central PMCID: 4010949.
- 35. Lee JY, Park K, Lee E, Ahn T, Jung HH, Lim SH, et al. Gene Expression Profiling of Breast Cancer Brain Metastasis. Scientific reports. 2016 Jun 24;6:28623. pmid:27340107. Pubmed Central PMCID: 4919653.
- 36. Cortez-Retamozo V, Etzrodt M, Newton A, Rauch PJ, Chudnovskiy A, Berger C, et al. Origins of tumor-associated macrophages and neutrophils. Proc Natl Acad Sci U S A. 2012 Feb 14;109(7):2491–6. pmid:22308361. Pubmed Central PMCID: 3289379.
- 37. Bednarski JJ, Pandey R, Schulte E, White LS, Chen BR, Sandoval GJ, et al. RAG-mediated DNA double-strand breaks activate a cell type-specific checkpoint to inhibit pre-B cell receptor signals. The Journal of experimental medicine. 2016 Feb 8;213(2):209–23. pmid:26834154. Pubmed Central PMCID: 4749927.
- 38. Ramnarine VR, Alshalalfa M, Mo F, Nabavi N, Erho N, Takhar M, et al. The long noncoding RNA landscape of neuroendocrine prostate cancer and its clinical implications. GigaScience. 2018 Jun 1;7(6). pmid:29757368. Pubmed Central PMCID: 6007253.
- 39. Stanton SE, Adams S, Disis ML. Variation in the Incidence and Magnitude of Tumor-Infiltrating Lymphocytes in Breast Cancer Subtypes: A Systematic Review. JAMA oncology. 2016 Oct 1;2(10):1354–60. pmid:27355489.
- 40. Ku YJ, Kim HH, Cha JH, Shin HJ, Baek SH, Lee HJ, et al. Correlation Between MRI and the Level of Tumor-Infiltrating Lymphocytes in Patients With Triple-Negative Breast Cancer. AJR American journal of roentgenology. 2016 Nov;207(5):1146–51. pmid:27556407.
- 41. Ono M, Tsuda H, Shimizu C, Yamamoto S, Shibata T, Yamamoto H, et al. Tumor-infiltrating lymphocytes are correlated with response to neoadjuvant chemotherapy in triple-negative breast cancer. Breast Cancer Res Treat. 2012 Apr;132(3):793–805. pmid:21562709.
- 42. Higgins M, Province M, Heiss G, Eckfeldt J, Ellison RC, Folsom AR, et al. NHLBI Family Heart Study: objectives and design. American journal of epidemiology. 1996 Jun 15;143(12):1219–28. pmid:8651220.
- 43. Sun H, Zou J, Chen L, Zu X, Wen G, Zhong J. Triple-negative breast cancer and its association with obesity. Molecular and clinical oncology. 2017 Dec;7(6):935–42. pmid:29285353. Pubmed Central PMCID: 5740844.
- 44. Choi Y, Park SK, Ahn KJ, Cho H, Kim TH, Yoon HK, et al. Being Overweight or Obese Increases the Risk of Progression in Triple-Negative Breast Cancer after Surgical Resection. Journal of Korean medical science. 2016 Jun;31(6):886–91. pmid:27247497. Pubmed Central PMCID: 4853667.