A validated gene expression profile for detecting clinical outcome in breast cancer using artificial neural networks

SLC1A5 is a key regulator of glutamine metabolism and a prognostic marker for aggressive luminal breast cancer

Cancer cells exhibit altered metabolism, often relying on glutamine (Gln) for growth. Breast cancer (BC) is a heterogeneous disease with varying clinical outcomes. We investigated the role of the amino acid transporter SLC1A5 (ASCT2) and its association with BC subtypes and patient outcomes. In large BC cohorts, SLC1A5 mRNA (n = 9488) and SLC1A5 protein (n = 1274) levels were assessed and correlated their expression with clinicopathological features, molecular subtypes, and patient outcomes. In vitro SLC1A5 knockdown and inhibition studies in luminal BC cell lines (ZR-75-1 and HCC1500) were used to further explore the role of SLC1A5 in Gln metabolism. Statistical analysis was performed using chi-squared tests, ANOVA, Spearman's correlation, Kaplan-Meier analysis, and Cox regression. SLC1A5 mRNA and SLC1A5 protein expression were strongly correlated in luminal B, HER2 + and triple-negative BC (TNBC). Both high SLC1A5 mRNA and SLC1A5 protein expression were associated with larger tumour size, higher grade, and positive axillary lymph node metastases (P < 0.01). Importantly, high SLC1A5 expression correlated with poor BC-specific survival specifically in the highly proliferative luminal subtype (P < 0.001). Furthermore, SLC1A5 knockdown by siRNA or GPNA inhibition significantly reduced cell proliferation and glutamine uptake in ZR-75-1 cells. Our findings suggest SLC1A5 plays a key role in the aggressive luminal BC subtype and represents a potential therapeutic target. Further research is needed to explore SLC1A5 function in luminal BC and its association with Gln metabolism pathways.

The relationship between carbonic anhydrase IX (CAIX) and patient survival in breast cancer: systematic review and meta-analysis

PURPOSE

Hypoxia is a characteristic of many solid tumours and an adverse prognostic factor for cancer therapy. Hypoxia results in upregulation of carbonic anhydrase IX (CAIX) expression, a pH-regulating enzyme. Many human tissue studies have examined the prognostic value of CAIX expression in breast cancer but have yielded inconsistent results. Therefore, a systematic review and meta-analysis was undertaken to assess the prognostic value of CAIX expression for breast cancer patients.

METHODS

The electronic databases were systematically searched to identify relevant papers. The clinical outcomes included disease-free survival (DFS), recurrence-free survival (RFS) and overall survival (OS) in breast cancer patients. Review Manager version 5.4 was employed to analysis data from 23 eligible studies (containing 8390 patients).

RESULTS

High CAIX expression was associated with poorer RFS [HR = 1.42, 95% CI (1.32-1.51), p < 0.00001], DFS [HR = 1.64, 95% CI (1.34-2.00), p < 0.00001], and OS [HR = 1.48, 95% CI (1.22-1.80), p < 0.0001]. Heterogeneity was observed across the studies. There was an effect of the CAIX antibody employed, scoring methods, and tumour localisation on CAIX expression.

CONCLUSION

CAIX overexpression was significantly associated with poorer RFS, DFS, and OS in breast cancer patients. However, further work in high quantity tissue cohorts is required to define the optimal methodological approach.

Deep Learning Based Methods for Breast Cancer Diagnosis: A Systematic Review and Future Direction

Breast cancer is one of the precarious conditions that affect women, and a substantive cure has not yet been discovered for it. With the advent of Artificial intelligence (AI), recently, deep learning techniques have been used effectively in breast cancer detection, facilitating early diagnosis and therefore increasing the chances of patients' survival. Compared to classical machine learning techniques, deep learning requires less human intervention for similar feature extraction. This study presents a systematic literature review on the deep learning-based methods for breast cancer detection that can guide practitioners and researchers in understanding the challenges and new trends in the field. Particularly, different deep learning-based methods for breast cancer detection are investigated, focusing on the genomics and histopathological imaging data. The study specifically adopts the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), which offer a detailed analysis and synthesis of the published articles. Several studies were searched and gathered, and after the eligibility screening and quality evaluation, 98 articles were identified. The results of the review indicated that the Convolutional Neural Network (CNN) is the most accurate and extensively used model for breast cancer detection, and the accuracy metrics are the most popular method used for performance evaluation. Moreover, datasets utilized for breast cancer detection and the evaluation metrics are also studied. Finally, the challenges and future research direction in breast cancer detection based on deep learning models are also investigated to help researchers and practitioners acquire in-depth knowledge of and insight into the area.

Nonlinear decision-making with enzymatic neural networks

Artificial neural networks have revolutionized electronic computing. Similarly, molecular networks with neuromorphic architectures may enable molecular decision-making on a level comparable to gene regulatory networks^1,2. Non-enzymatic networks could in principle support neuromorphic architectures, and seminal proofs-of-principle have been reported^3,4. However, leakages (that is, the unwanted release of species), as well as issues with sensitivity, speed, preparation and the lack of strong nonlinear responses, make the composition of layers delicate, and molecular classifications equivalent to a multilayer neural network remain elusive (for example, the partitioning of a concentration space into regions that cannot be linearly separated). Here we introduce DNA-encoded enzymatic neurons with tuneable weights and biases, and which are assembled in multilayer architectures to classify nonlinearly separable regions. We first leverage the sharp decision margin of a neuron to compute various majority functions on 10 bits. We then compose neurons into a two-layer network and synthetize a parametric family of rectangular functions on a microRNA input. Finally, we connect neural and logical computations into a hybrid circuit that recursively partitions a concentration plane according to a decision tree in cell-sized droplets. This computational power and extreme miniaturization open avenues to query and manage molecular systems with complex contents, such as liquid biopsies or DNA databases.

Deep Learning and Machine Learning with Grid Search to Predict Later Occurrence of Breast Cancer Metastasis Using Clinical Data

Background: It is important to be able to predict, for each individual patient, the likelihood of later metastatic occurrence, because the prediction can guide treatment plans tailored to a specific patient to prevent metastasis and to help avoid under-treatment or over-treatment. Deep neural network (DNN) learning, commonly referred to as deep learning, has become popular due to its success in image detection and prediction, but questions such as whether deep learning outperforms other machine learning methods when using non-image clinical data remain unanswered. Grid search has been introduced to deep learning hyperparameter tuning for the purpose of improving its prediction performance, but the effect of grid search on other machine learning methods are under-studied. In this research, we take the empirical approach to study the performance of deep learning and other machine learning methods when using non-image clinical data to predict the occurrence of breast cancer metastasis (BCM) 5, 10, or 15 years after the initial treatment. We developed prediction models using the deep feedforward neural network (DFNN) methods, as well as models using nine other machine learning methods, including naïve Bayes (NB), logistic regression (LR), support vector machine (SVM), LASSO, decision tree (DT), k-nearest neighbor (KNN), random forest (RF), AdaBoost (ADB), and XGBoost (XGB). We used grid search to tune hyperparameters for all methods. We then compared our feedforward deep learning models to the models trained using the nine other machine learning methods. Results: Based on the mean test AUC (Area under the ROC Curve) results, DFNN ranks 6th, 4th, and 3rd when predicting 5-year, 10-year, and 15-year BCM, respectively, out of 10 methods. The top performing methods in predicting 5-year BCM are XGB (1st), RF (2nd), and KNN (3rd). For predicting 10-year BCM, the top performers are XGB (1st), RF (2nd), and NB (3rd). Finally, for 15-year BCM, the top performers are SVM (1st), LR and LASSO (tied for 2nd), and DFNN (3rd). The ensemble methods RF and XGB outperform other methods when data are less balanced, while SVM, LR, LASSO, and DFNN outperform other methods when data are more balanced. Our statistical testing results show that at a significance level of 0.05, DFNN overall performs comparably to other machine learning methods when predicting 5-year, 10-year, and 15-year BCM. Conclusions: Our results show that deep learning with grid search overall performs at least as well as other machine learning methods when using non-image clinical data. It is interesting to note that some of the other machine learning methods, such as XGB, RF, and SVM, are very strong competitors of DFNN when incorporating grid search. It is also worth noting that the computation time required to do grid search with DFNN is much more than that required to do grid search with the other nine machine learning methods.

Understanding metabolic alterations and heterogeneity in cancer progression through validated immunodetection of key molecular components: a case of carbonic anhydrase IX

Cancer metabolic heterogeneity develops in response to both intrinsic factors (mutations leading to activation of oncogenic pathways) and extrinsic factors (physiological and molecular signals from the extracellular milieu). Here we review causes and consequences of metabolic alterations in cancer cells with focus on hypoxia and acidosis, and with particular attention to carbonic anhydrase IX (CA IX). CA IX is a cancer-associated enzyme induced and activated by hypoxia in a broad range of tumor types, where it participates in pH regulation as well as in molecular mechanisms supporting cancer cells’ invasion and metastasis. CA IX catalyzes reversible conversion of carbon dioxide to bicarbonate ion plus proton and cooperates with a spectrum of molecules transporting ions or metabolites across the plasma membrane. Thereby CA IX contributes to extracellular acidosis as well as to buffering intracellular pH, which is essential for cell survival, metabolic performance, and proliferation of cancer cells. Since CA IX expression pattern reflects gradients of oxygen, pH, and other intratumoral factors, we use it as a paradigm to discuss an impact of antibody quality and research material on investigating metabolic reprogramming of tumor tissue. Based on the validation, we propose the most reliable CA IX-specific antibodies and suggest conditions for faithful immunohistochemical analysis of molecules contributing to heterogeneity in cancer progression.

A novel prognostic two-gene signature for triple negative breast cancer

The absence of a robust risk stratification tool for triple negative breast cancer (TNBC) underlies imprecise and nonselective treatment of these patients with cytotoxic chemotherapy. This study aimed to interrogate transcriptomes of TNBC resected samples using next generation sequencing to identify novel biomarkers associated with disease outcomes. A subset of cases (n = 112) from a large, well-characterized cohort of primary TNBC (n = 333) were subjected to RNA-sequencing. Reads were aligned to the human reference genome (GRCH38.83) using the STAR aligner and gene expression quantified using HTSEQ. We identified genes associated with distant metastasis-free survival and breast cancer-specific survival by applying supervised artificial neural network analysis with gene selection to the RNA-sequencing data. The prognostic ability of these genes was validated using the Breast Cancer Gene-Expression Miner v4. 0 and Genotype 2 outcome datasets. Multivariate Cox regression analysis identified a prognostic gene signature that was independently associated with poor prognosis. Finally, we corroborated our results from the two-gene prognostic signature by their protein expression using immunohistochemistry. Artificial neural network identified two gene panels that strongly predicted distant metastasis-free survival and breast cancer-specific survival. Univariate Cox regression analysis of 21 genes common to both panels revealed that the expression level of eight genes was independently associated with poor prognosis (p < 0.05). Adjusting for clinicopathological factors including patient's age, grade, nodal stage, tumor size, and lymphovascular invasion using multivariate Cox regression analysis yielded a two-gene prognostic signature (ACSM4 and SPDYC), which was associated with poor prognosis (p < 0.05) independent of other prognostic variables. We validated the protein expression of these two genes, and it was significantly associated with patient outcome in both independent and combined manner (p < 0.05). Our study identifies a prognostic gene signature that can predict prognosis in TNBC patients and could potentially be used to guide the clinical management of TNBC patients.

A parsimonious 3-gene signature predicts clinical outcomes in an acute myeloid leukemia multicohort study

Acute myeloid leukemia (AML) is a genetically heterogeneous hematological malignancy with variable responses to chemotherapy. Although recurring cytogenetic abnormalities and gene mutations are important predictors of outcome, 50% to 70% of AMLs harbor normal or risk-indeterminate karyotypes. Therefore, identifying more effective biomarkers predictive of treatment success and failure is essential for informing tailored therapeutic decisions. We applied an artificial neural network (ANN)-based machine learning approach to a publicly available data set for a discovery cohort of 593 adults with nonpromyelocytic AML. ANN analysis identified a parsimonious 3-gene expression signature comprising CALCRL, CD109, and LSP1, which was predictive of event-free survival (EFS) and overall survival (OS). We computed a prognostic index (PI) using normalized gene-expression levels and β-values from subsequently created Cox proportional hazards models, coupled with clinically established prognosticators. Our 3-gene PI separated the adult patients in each European LeukemiaNet cytogenetic risk category into subgroups with different survival probabilities and identified patients with very high-risk features, such as those with a high PI and either FLT3 internal tandem duplication or nonmutated nucleophosmin 1. The PI remained significantly associated with poor EFS and OS after adjusting for established prognosticators, and its ability to stratify survival was validated in 3 independent adult cohorts (n = 905 subjects) and 1 cohort of childhood AML (n = 145 subjects). Further in silico analyses established that AML was the only tumor type among 39 distinct malignancies for which the concomitant upregulation of CALCRL, CD109, and LSP1 predicted survival. Therefore, our ANN-derived 3-gene signature refines the accuracy of patient stratification and the potential to significantly improve outcome prediction.

AN EMPIRICAL STUDY OF CANCER CLASSIFICATION TECHNIQUES BASED ON THE NEURAL NETWORKS

Cancer is one of the most common dreadful diseases prevailing worldwide, and patients with cancer are rescued only when the cancer is detected at a very early stage. Early detection of cancer is appropriate as in the fourth stage, but the chance of survival is limited. The symptoms of cancers are rigorous, and therefore, all the symptoms should be studied properly before the diagnosis. Thus, an automatic prediction system is necessary for classifying the tumor, i.e. malignant or benign tumor. Over the past few years, cancer classification is increased rapidly, but there is no general technique to find novel cancer classes (class discovery) or to assign tumors to known classes. Accordingly, this survey analyzes distinct cancer classification techniques. Thus, this review article provides a detailed review of 50 research papers presenting the suggested cancer classification techniques, like Deep learning-based techniques, Neural network-based techniques, and Hybrid techniques. Moreover, an elaborative analysis and discussion are made based on the year of publication, utilized datasets, accuracy range, evaluation metrics, implementation tool, and adopted classification methods. Eventually, the research gaps and issues of various cancer classification schemes are presented for extending the researchers towards a better future scope.

Development of a Bioinformatics Framework for Identification and Validation of Genomic Biomarkers and Key Immunopathology Processes and Controllers in Infectious and Non-infectious Severe Inflammatory Response Syndrome

Sepsis is defined as dysregulated host response caused by systemic infection, leading to organ failure. It is a life-threatening condition, often requiring admission to an intensive care unit (ICU). The causative agents and processes involved are multifactorial but are characterized by an overarching inflammatory response, sharing elements in common with severe inflammatory response syndrome (SIRS) of non-infectious origin. Sepsis presents with a range of pathophysiological and genetic features which make clinical differentiation from SIRS very challenging. This may reflect a poor understanding of the key gene inter-activities and/or pathway associations underlying these disease processes. Improved understanding is critical for early differential recognition of sepsis and SIRS and to improve patient management and clinical outcomes. Judicious selection of gene biomarkers suitable for development of diagnostic tests/testing could make differentiation of sepsis and SIRS feasible. Here we describe a methodologic framework for the identification and validation of biomarkers in SIRS, sepsis and septic shock patients, using a 2-tier gene screening, artificial neural network (ANN) data mining technique, using previously published gene expression datasets. Eight key hub markers have been identified which may delineate distinct, core disease processes and which show potential for informing underlying immunological and pathological processes and thus patient stratification and treatment. These do not show sufficient fold change differences between the different disease states to be useful as primary diagnostic biomarkers, but are instrumental in identifying candidate pathways and other associated biomarkers for further exploration.

The multifunctional solute carrier 3A2 (SLC3A2) confers a poor prognosis in the highly proliferative breast cancer subtypes

Breast cancer (BC) is a heterogeneous disease characterised by variant biology, metabolic activity and patient outcome. This study aimed to evaluate the biological and prognostic value of the membrane solute carrier, SLC3A2 in BC with emphasis on the intrinsic molecular subtypes. SLC3A2 was assessed at the genomic level, using METABRIC data (n = 1980), and at the proteomic level, using immunohistochemistry on tissue microarray (TMA) sections constructed from a large well-characterised primary BC cohort (n = 2500). SLC3A2 expression was correlated with clinicopathological parameters, molecular subtypes and patient outcome. SLC3A2 mRNA and protein expression were strongly correlated with higher tumour grade and poor Nottingham prognostic index (NPI). High expression of SLC3A2 was observed in triple-negative (TN), HER2+ and ER+ high-proliferation subtypes. SLC3A2 mRNA and protein expression were significantly associated with the expression of c-MYC in all BC subtypes (p < 0.001). High expression of SLC3A2 protein was associated with poor patient outcome (p < 0.001), but only in the ER+ high-proliferation (p = 0.01) and TN (p = 0.04) subtypes. In multivariate analysis SLC3A2 protein was an independent risk factor for shorter BC-specific survival (p < 0.001). SLC3A2 appears to play a role in the aggressive BC subtypes driven by MYC and could act as a potential prognostic marker. Functional assessment is necessary to reveal its potential therapeutic value in the different BC subtypes.

The amino acid transporter SLC7A5 confers a poor prognosis in the highly proliferative breast cancer subtypes and is a key therapeutic target in luminal B tumours

BACKGROUND

Breast cancer (BC) is a heterogeneous disease characterised by variant biology and patient outcome. The amino acid transporter, SLC7A5, plays a role in BC although its impact on patient outcome in different BC subtypes remains to be validated. This study aimed to determine whether the clinicopathological and prognostic value of SLC7A5 is different within the molecular classes of BC.

METHODS

SLC7A5 was assessed at the genomic level, using Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) data (n = 1980), and proteomic level, using immunohistochemical analysis and tissue microarray (TMA) (n = 2664; 1110 training and 1554 validation sets) in well-characterised primary BC cohorts. SLC7A5 expression correlated with clinicopathological and biological parameters, molecular subtypes and patient outcome.

RESULTS

SLC7A5 mRNA and protein expression were strongly correlated with larger tumour size and higher grade. High expression was observed in triple negative (TN), human epidermal growth factor receptor 2 (HER2)+, and luminal B subtypes. SLC7A5 mRNA and protein expression was significantly associated with the expression of the key regulator of tumour cell metabolism, c-MYC, specifically in luminal B tumours only (p = 0.001). High expression of SLC7A5 mRNA and protein was associated with poor patient outcome (p < 0.001) but only in the highly proliferative oestrogen receptor (ER)+/ luminal B (p = 0.007) and HER2+ classes of BC (p = 0.03). In multivariate analysis, SLC7A5 protein was an independent risk factor for shorter breast-cancer-specific survival only in ER+ high-proliferation tumours (p = 0.02).

CONCLUSIONS

SLC7A5 appears to play a role in the aggressive highly proliferative ER+ subtype driven by MYC and could act as a potential therapeutic target. Functional assessment is necessary to reveal the specific role played by this transporter in the ER+ highly proliferative subclass and HER2+ subclass of BC.

Strategies for power calculations in predictive biomarker studies in survival data

PURPOSE

Biomarkers and genomic signatures represent potentially predictive tools for precision medicine. Validation of predictive biomarkers in prospective or retrospective studies requires statistical justification of power and sample size. However, the design of these studies is complex and the statistical methods and associated software are limited, especially in survival data. Herein, we address common statistical design issues relevant to these two types of studies and provide guidance and a general template for analysis.

METHODS

A statistical interaction effect in the Cox proportional hazards model is used to describe predictive biomarkers. The analytic form by Peterson et al. and Lachin is utilized to calculate the statistical power for both prospective and retrospective studies.

RESULTS

We demonstrate that the common mistake of using only Hazard Ratio's Ratio (HRR) or two hazard ratios (HRs) can mislead power calculations. We establish that the appropriate parameter settings for prospective studies require median survival time (MST) in 4 subgroups (treatment and control in positive biomarker, treatment and control in negative biomarker). For the retrospective study which has fixed survival time and censored status, we develop a strategy to harmonize the hypothesized parameters and the study cohort. Moreover, we provide an easily-adapted R software application to generate a template of statistical plan for predictive biomarker validation so investigators can easily incorporate into their study proposals.

CONCLUSION

Our study provides guidance and software to help biostatisticians and clinicians design sound clinical studies for testing predictive biomarkers.

SPAG5 as a prognostic biomarker and chemotherapy sensitivity predictor in breast cancer: a retrospective, integrated genomic, transcriptomic, and protein analysis

BACKGROUND

Proliferation markers and profiles have been recommended for guiding the choice of systemic treatments for breast cancer. However, the best molecular marker or test to use has not yet been identified. We did this study to identify factors that drive proliferation and its associated features in breast cancer and assess their association with clinical outcomes and response to chemotherapy.

METHODS

We applied an artificial neural network-based integrative data mining approach to data from three cohorts of patients with breast cancer (the Nottingham discovery cohort (n=171), Uppsala cohort (n=249), and Molecular Taxonomy of Breast Cancer International Consortium [METABRIC] cohort; n=1980). We then identified the genes with the most effect on other genes in the resulting interactome map. Sperm-associated antigen 5 (SPAG5) featured prominently in our interactome map of proliferation and we chose to take it forward in our analysis on the basis of its fundamental role in the function and dynamic regulation of mitotic spindles, mitotic progression, and chromosome segregation fidelity. We investigated the clinicopathological relevance of SPAG5 gene copy number aberrations, mRNA transcript expression, and protein expression and analysed the associations of SPAG5 copy number aberrations, transcript expression, and protein expression with breast cancer-specific survival, disease-free survival, distant relapse-free survival, pathological complete response, and residual cancer burden in the Nottingham discovery cohort, Uppsala cohort, METABRIC cohort, a pooled untreated lymph node-negative cohort (n=684), a multicentre combined cohort (n=5439), the Nottingham historical early stage breast cancer cohort (Nottingham-HES; n=1650), Nottingham early stage oestrogen receptor-negative breast cancer adjuvant chemotherapy cohort (Nottingham-oestrogen receptor-negative-ACT; n=697), the Nottingham anthracycline neoadjuvant chemotherapy cohort (Nottingham-NeoACT; n=200), the MD Anderson taxane plus anthracycline-based neoadjuvant chemotherapy cohort (MD Anderson-NeoACT; n=508), and the multicentre phase 2 neoadjuvant clinical trial cohort (phase 2 NeoACT; NCT00455533; n=253).

FINDINGS

In the METABRIC cohort, we detected SPAG5 gene gain or amplification at the Ch17q11.2 locus in 206 (10%) of 1980 patients overall, 46 (19%) of 237 patients with a PAM50-HER2 phenotype, and 87 (18%) of 488 patients with PAM50-LumB phenotype. Copy number aberration leading to SPAG5 gain or amplification and high SPAG5 transcript and SPAG5 protein concentrations were associated with shorter overall breast cancer-specific survival (METABRIC cohort [copy number aberration]: hazard ratio [HR] 1·50, 95% CI 1·18-1·92, p=0·00010; METABRIC cohort [transcript]: 1·68, 1·40-2·01, p<0·0001; and Nottingham-HES-breast cancer cohort [protein]: 1·68, 1·32-2·12, p<0·0001). In multivariable analysis, high SPAG5 transcript and SPAG5 protein expression were associated with reduced breast cancer-specific survival at 10 years compared with lower concentrations (Uppsala: HR 1·62, 95% CI 1·03-2·53, p=0·036; METABRIC: 1·27, 1·02-1·58, p=0·034; untreated lymph node-negative cohort: 2·34, 1·24-4·42, p=0·0090; and Nottingham-HES: 1·73, 1·23-2·46, p=0·0020). In patients with oestrogen receptor-negative breast cancer with high SPAG5 protein expression, anthracycline-based adjuvant chemotherapy increased breast cancer-specific survival overall compared with that for patients who did not receive chemotherapy (Nottingham-oestrogen receptor-negative-ACT cohort: HR 0·37, 95% CI 0·20-0·60, p=0·0010). Multivariable analysis showed high SPAG5 transcript concentrations to be independently associated with longer distant relapse-free survival after receiving taxane plus anthracycline neoadjuvant chemotherapy (MD Anderson-NeoACT: HR 0·68, 95% CI 0·48-0·97, p=0·031). In multivariable analysis, both high SPAG5 transcript and high SPAG5 protein concentrations were independent predictors for a higher proportion of patients achieving a pathological complete response after combination cytotoxic chemotherapy (MD Anderson-NeoACT: OR 1·71, 95% CI, 1·07-2·74, p=0·024; Nottingham-ACT: 8·75, 2·42-31·62, p=0·0010).

INTERPRETATION

SPAG5 is a novel amplified gene on Ch17q11.2 in breast cancer. The transcript and protein products of SPAG5 are independent prognostic and predictive biomarkers that might have clinical utility as biomarkers for combination cytotoxic chemotherapy sensitivity, especially in oestrogen receptor-negative breast cancer.

FUNDING

Nottingham Hospitals Charity and the John and Lucille van Geest Foundation.

Temporal Expression of Peripheral Blood Leukocyte Biomarkers in a Macaca fascicularis Infection Model of Tuberculosis; Comparison with Human Datasets and Analysis with Parametric/Non-parametric Tools for Improved Diagnostic Biomarker Identification

A temporal study of gene expression in peripheral blood leukocytes (PBLs) from a Mycobacterium tuberculosis primary, pulmonary challenge model Macaca fascicularis has been conducted. PBL samples were taken prior to challenge and at one, two, four and six weeks post-challenge and labelled, purified RNAs hybridised to Operon Human Genome AROS V4.0 slides. Data analyses revealed a large number of differentially regulated gene entities, which exhibited temporal profiles of expression across the time course study. Further data refinements identified groups of key markers showing group-specific expression patterns, with a substantial reprogramming event evident at the four to six week interval. Selected statistically-significant gene entities from this study and other immune and apoptotic markers were validated using qPCR, which confirmed many of the results obtained using microarray hybridisation. These showed evidence of a step-change in gene expression from an ‘early’ FOS-associated response, to a ‘late’ predominantly type I interferon-driven response, with coincident reduction of expression of other markers. Loss of T-cell-associate marker expression was observed in responsive animals, with concordant elevation of markers which may be associated with a myeloid suppressor cell phenotype e.g. CD163. The animals in the study were of different lineages and these Chinese and Mauritian cynomolgous macaque lines showed clear evidence of differing susceptibilities to Tuberculosis challenge. We determined a number of key differences in response profiles between the groups, particularly in expression of T-cell and apoptotic makers, amongst others. These have provided interesting insights into innate susceptibility related to different host `phenotypes. Using a combination of parametric and non-parametric artificial neural network analyses we have identified key genes and regulatory pathways which may be important in early and adaptive responses to TB. Using comparisons between data outputs of each analytical pipeline and comparisons with previously published Human TB datasets, we have delineated a subset of gene entities which may be of use for biomarker diagnostic test development.

MYC functions are specific in biological subtypes of breast cancer and confers resistance to endocrine therapy in luminal tumours

BACKGROUND

MYC is amplified in approximately 15% of breast cancers (BCs) and is associated with poor outcome. c-MYC protein is multi-faceted and participates in many aspects of cellular function and is linked with therapeutic response in BCs. We hypothesised that the functional role of c-MYC differs between molecular subtypes of BCs.

METHODS

We therefore investigated the correlation between c-MYC protein expression and other proteins involved in different cellular functions together with clinicopathological parameters, patients' outcome and treatments in a large early-stage molecularly characterised series of primary invasive BCs (n=1106) using immunohistochemistry. The METABRIC BC cohort (n=1980) was evaluated for MYC mRNA expression and a systems biology approach utilised to identify genes associated with MYC in the different BC molecular subtypes.

RESULTS

High MYC and c-MYC expression was significantly associated with poor prognostic factors, including grade and basal-like BCs. In luminal A tumours, c-MYC was associated with ATM (P=0.005), Cyclin B1 (P=0.002), PIK3CA (P=0.009) and Ki67 (P<0.001). In contrast, in basal-like tumours, c-MYC showed positive association with Cyclin E (P=0.003) and p16 (P=0.042) expression only. c-MYC was an independent predictor of a shorter distant metastases-free survival in luminal A LN+ tumours treated with endocrine therapy (ET; P=0.013). In luminal tumours treated with ET, MYC mRNA expression was associated with BC-specific survival (P=0.001). In ER-positive tumours, MYC was associated with expression of translational genes while in ER-negative tumours it was associated with upregulation of glucose metabolism genes.

CONCLUSIONS

c-MYC function is associated with specific molecular subtypes of BCs and its overexpression confers resistance to ET. The diverse mechanisms of c-MYC function in the different molecular classes of BCs warrants further investigation particularly as potential therapeutic targets.

Prognostic Significance of Carbonic Anhydrase IX Expression in Cancer Patients: A Meta-Analysis

Hypoxia is a characteristic of many solid tumors and an adverse prognostic factor for treatment outcome. Hypoxia increases the expression of carbonic anhydrase IX (CAIX), an enzyme that is predominantly found on tumor cells and is involved in maintaining the cellular pH balance. Many clinical studies investigated the prognostic value of CAIX expression, but most have been inconclusive, partly due to small numbers of patients included. The present meta-analysis was therefore performed utilizing the results of all clinical studies to determine the prognostic value of CAIX expression in solid tumors. Renal cell carcinoma was excluded from this meta-analysis due to an alternative mechanism of upregulation. 958 papers were identified from a literature search performed in PubMed and Embase. These papers were independently evaluated by two reviewers and 147 studies were included in the analysis. The meta-analysis revealed strong significant associations between CAIX expression and all endpoints: overall survival [hazard ratio (HR) = 1.76, 95% confidence interval (95%CI) 1.58–1.98], disease-free survival (HR = 1.87, 95%CI 1.62–2.16), locoregional control (HR = 1.54, 95%CI 1.22–1.93), disease-specific survival (HR = 1.78, 95%CI 1.41–2.25), metastasis-free survival (HR = 1.82, 95%CI 1.33–2.50), and progression-free survival (HR = 1.58, 95%CI 1.27–1.96). Subgroup analyses revealed similar associations in the majority of tumor sites and types. In conclusion, these results show that patients having tumors with high CAIX expression have higher risk of locoregional failure, disease progression, and higher risk to develop metastases, independent of tumor type or site. The results of this meta-analysis further support the development of a clinical test to determine patient prognosis based on CAIX expression and may have important implications for the development of new treatment strategies.

HAGE in Triple-Negative Breast Cancer Is a Novel Prognostic, Predictive, and Actionable Biomarker: A Transcriptomic and Protein Expression Analysis

PURPOSE

The expression of HAGE as a novel prognostic and predictive tool was assessed in 1,079 triple-negative breast cancers (TNBC).

EXPERIMENTAL DESIGN

HAGE protein expression was investigated in an early primary TNBC (EP-TNBC; n = 520) cohort who received adjuvant chemotherapy (ACT) and in a locally advanced primary TNBC cohort who received anthracycline combination Neo-ACT (n = 110; AC-Neo-ACT). HAGE-mRNA expression was evaluated in the METABRIC-TNBC cohort (n = 311) who received ACT and in a cohort of patients with TNBC who received doxorubicin/cyclophosphamide Neo-ACT, followed by 1:1 randomization to ixabepilone (n = 68) or paclitaxel (n = 64) as part of a phase II clinical trial. Furthermore, a cohort of 128 tumors with integrated HAGE gene copy number changes, mRNA, and protein levels were analyzed.

RESULTS

In patients with EP-TNBC, who were chemotherapy-naïve, high HAGE protein expression (HAGE(+)) was associated with a higher risk of death [HR, 1.3; 95% confidence interval (CI), 1.2-1.5; P = 0.000005] when compared with HAGE(-) cases. Patients who received ACT and expressed mRNA-HAGE(+) were at a lower risk of death than those who were mRNA-HAGE(-) (P = 0.004). The expression of HAGE was linked to the presence of tumor-infiltrating lymphocytes (TIL), and both features were found to be independent predictors for pathologic complete response (pCR, P < 0.001) and associated with prolonged survival (P < 0.01), following AC-Neo-ACT. In patients with residual disease, HAGE(+) had a 2-fold death risk increase (P = 0.018) compared with HAGE(-).

CONCLUSIONS

HAGE expression is a potential prognostic marker and a predictor of response to anthracycline treatment in TNBC. A prospective clinical trial to examine the therapeutic value of HAGE for TNBC cases is warranted.

Dissecting DNA repair in adult high grade gliomas for patient stratification in the post-genomic era

Deregulation of multiple DNA repair pathways may contribute to aggressive biology and therapy resistance in gliomas. We evaluated transcript levels of 157 genes involved in DNA repair in an adult glioblastoma Test set (n=191) and validated in ‘The Cancer Genome Atlas’ (TCGA) cohort (n=508). A DNA repair prognostic index model was generated. Artificial neural network analysis (ANN) was conducted to investigate global gene interactions. Protein expression by immunohistochemistry was conducted in 61 tumours. A fourteen DNA repair gene expression panel was associated with poor survival in Test and TCGA cohorts. A Cox multivariate model revealed APE1, NBN, PMS2, MGMT and PTEN as independently associated with poor prognosis. A DNA repair prognostic index incorporating APE1, NBN, PMS2, MGMT and PTEN stratified patients in to three prognostic sub-groups with worsening survival. APE1, NBN, PMS2, MGMT and PTEN also have predictive significance in patients who received chemotherapy and/or radiotherapy. ANN analysis of APE1, NBN, PMS2, MGMT and PTEN revealed interactions with genes involved in transcription, hypoxia and metabolic regulation. At the protein level, low APE1 and low PTEN remain associated with poor prognosis. In conclusion, multiple DNA repair pathways operate to influence biology and clinical outcomes in adult high grade gliomas.

Immunoglobulin free light chains are biomarkers of poor prognosis in basal-like breast cancer and are potential targets in tumor-associated inflammation

Inflammation is an important component of various cancers and its inflammatory cells and mediators have been shown to have prognostic potential. Tumor-infiltrating mast cells can promote tumor growth and angiogenesis, but the mechanism of mast cell activation is unclear. In earlier studies, we demonstrated that immunoglobulin free light chains (FLC) can trigger mast cells in an antigen-specific manner. Increased expression of FLC was observed within stroma of various human cancers including those of breast, colon, lung, pancreas, kidney and skin, and FLC expression co-localized with areas of mast cell infiltration. In a large cohort of breast cancer patients, FLC expression was shown associated with basal-like cancers with an aggressive phenotype. Moreover, lambda FLC was found expressed in areas of inflammatory infiltration and its expression was significantly associated with poor clinical outcome. Functional importance of FLCs was shown in a murine B16F10 melanoma model, where inhibition of FLC-mediated mast cell activation strongly reduced tumor growth. Collectively, this study identifies FLCs as a ligand in the pro-tumorigenic activation of mast cells. Blocking this pathway may open new avenues for the inhibition of tumor growth, while immunohistochemical staining of FLC may be helpful in the diagnosis and prognosis of cancer.

Targeting BRCA1-BER deficient breast cancer by ATM or DNA-PKcs blockade either alone or in combination with cisplatin for personalized therapy

BRCA1, a key factor in homologous recombination (HR) repair may also regulate base excision repair (BER). Targeting BRCA1-BER deficient cells by blockade of ATM and DNA-PKcs could be a promising strategy in breast cancer. We investigated BRCA1, XRCC1 and pol β protein expression in two cohorts (n = 1602 sporadic and n = 50 germ-line BRCA1 mutated) and mRNA expression in two cohorts (n = 1952 and n = 249). Artificial neural network analysis for BRCA1-DNA repair interacting genes was conducted in 249 tumours. Pre-clinically, BRCA1 proficient and deficient cells were DNA repair expression profiled and evaluated for synthetic lethality using ATM and DNA-PKcs inhibitors either alone or in combination with cisplatin. In human tumours, BRCA1 negativity was strongly associated with low XRCC1, and low pol β at mRNA and protein levels (p < 0.0001). In patients with BRCA1 negative tumours, low XRCC1 or low pol β expression was significantly associated with poor survival in univariate and multivariate analysis compared to high XRCC1 or high pol β expressing BRCA1 negative tumours (ps < 0.05). Pre-clinically, BRCA1 negative cancer cells exhibit low mRNA and low protein expression of XRCC1 and pol β. BRCA1-BER deficient cells were sensitive to ATM and DNA-PKcs inhibitor treatment either alone or in combination with cisplatin and synthetic lethality was evidenced by DNA double strand breaks accumulation, cell cycle arrest and apoptosis. We conclude that XRCC1 and pol β expression status in BRCA1 negative tumours may have prognostic significance. BRCA1-BER deficient cells could be targeted by ATM or DNA-PKcs inhibitors for personalized therapy.

Artificial neural network inference (ANNI): a study on gene-gene interaction for biomarkers in childhood sarcomas

Objective

To model the potential interaction between previously identified biomarkers in children sarcomas using artificial neural network inference (ANNI).

Method

To concisely demonstrate the biological interactions between correlated genes in an interaction network map, only 2 types of sarcomas in the children small round blue cell tumors (SRBCTs) dataset are discussed in this paper. A backpropagation neural network was used to model the potential interaction between genes. The prediction weights and signal directions were used to model the strengths of the interaction signals and the direction of the interaction link between genes. The ANN model was validated using Monte Carlo cross-validation to minimize the risk of over-fitting and to optimize generalization ability of the model.

Results

Strong connection links on certain genes (TNNT1 and FNDC5 in rhabdomyosarcoma (RMS); FCGRT and OLFM1 in Ewing’s sarcoma (EWS)) suggested their potency as central hubs in the interconnection of genes with different functionalities. The results showed that the RMS patients in this dataset are likely to be congenital and at low risk of cardiomyopathy development. The EWS patients are likely to be complicated by EWS-FLI fusion and deficiency in various signaling pathways, including Wnt, Fas/Rho and intracellular oxygen.

Conclusions

The ANN network inference approach and the examination of identified genes in the published literature within the context of the disease highlights the substantial influence of certain genes in sarcomas.

Cross-species gene expression analysis of species specific differences in the preclinical assessment of pharmaceutical compounds

Animals are frequently used as model systems for determination of safety and efficacy in pharmaceutical research and development. However, significant quantitative and qualitative differences exist between humans and the animal models used in research. This is as a result of genetic variation between human and the laboratory animal. Therefore the development of a system that would allow the assessment of all molecular differences between species after drug exposure would have a significant impact on drug evaluation for toxicity and efficacy. Here we describe a cross-species microarray methodology that identifies and selects orthologous probes after cross-species sequence comparison to develop an orthologous cross-species gene expression analysis tool. The assumptions made by the use of this orthologous gene expression strategy for cross-species extrapolation is that; conserved changes in gene expression equate to conserved pharmacodynamic endpoints. This assumption is supported by the fact that evolution and selection have maintained the structure and function of many biochemical pathways over time, resulting in the conservation of many important processes. We demonstrate this cross-species methodology by investigating species specific differences of the peroxisome proliferator-activator receptor (PPAR) α response in rat and human.

DNA polymerase β deficiency is linked to aggressive breast cancer: a comprehensive analysis of gene copy number, mRNA and protein expression in multiple cohorts

Short arm of chromosome 8 is a hot spot for chromosomal breaks, losses and amplifications in breast cancer. Although such genetic changes may have phenotypic consequences, the identity of candidate gene(s) remains to be clearly defined. Pol β gene is localized to chromosome 8p12-p11 and encodes a key DNA base excision repair protein. Pol β may be a tumour suppressor and involved in breast cancer pathogenesis. We conducted the first and the largest study to comprehensively evaluate pol β in breast cancer. We investigated pol β gene copy number changes in two cohorts (n = 128 &n = 1952), pol β mRNA expression in two cohorts (n = 249 &n = 1952) and pol β protein expression in two cohorts (n = 1406 &n = 252). Artificial neural network analysis for pol β interacting genes was performed in 249 tumours. For mechanistic insights, pol β gene copy number changes, mRNA and protein levels were investigated together in 128 tumours and validated in 1952 tumours. Low pol β mRNA expression as well as low pol β protein expression was associated high grade, lymph node positivity, pleomorphism, triple negative, basal-like phenotypes and poor survival (ps < 0.001). In oestrogen receptor (ER) positive sub-group that received tamoxifen, low pol β protein remains associated with aggressive phenotype and poor survival (ps < 0.001). Artificial neural network analysis revealed ER as a top pol β interacting gene. Mechanistically, there was strong positive correlation between pol β gene copy number changes and pol β mRNA expression (p < 0.0000001) and between pol β mRNA and pol β protein expression (p < 0.0000001). This is the first study to provide evidence that pol β deficiency is linked to aggressive breast cancer and may have prognostic and predictive significance in patients.

Identification and validation of oncologic miRNA biomarkers for luminal A-like breast cancer

INTRODUCTION

Breast cancer is a common disease with distinct tumor subtypes phenotypically characterized by ER and HER2/neu receptor status. MiRNAs play regulatory roles in tumor initiation and progression, and altered miRNA expression has been demonstrated in a variety of cancer states presenting the potential for exploitation as cancer biomarkers. Blood provides an excellent medium for biomarker discovery. This study investigated systemic miRNAs differentially expressed in Luminal A-like (ER+PR+HER2/neu-) breast cancer and their effectiveness as oncologic biomarkers in the clinical setting.

METHODS

Blood samples were prospectively collected from patients with Luminal A-like breast cancer (n = 54) and controls (n = 56). RNA was extracted, reverse transcribed and subjected to microarray analysis (n = 10 Luminal A-like; n = 10 Control). Differentially expressed miRNAs were identified by artificial neural network (ANN) data-mining algorithms. Expression of specific miRNAs was validated by RQ-PCR (n = 44 Luminal A; n = 46 Control) and potential relationships between circulating miRNA levels and clinicopathological features of breast cancer were investigated.

RESULTS

Microarray analysis identified 76 differentially expressed miRNAs. ANN revealed 10 miRNAs for further analysis (miR-19b, miR-29a, miR-93, miR-181a, miR-182, miR-223, miR-301a, miR-423-5p, miR-486-5 and miR-652). The biomarker potential of 4 miRNAs (miR-29a, miR-181a, miR-223 and miR-652) was confirmed by RQ-PCR, with significantly reduced expression in blood of women with Luminal A-like breast tumors compared to healthy controls (p = 0.001, 0.004, 0.009 and 0.004 respectively). Binary logistic regression confirmed that combination of 3 of these miRNAs (miR-29a, miR-181a and miR-652) could reliably differentiate between cancers and controls with an AUC of 0.80.

CONCLUSION

This study provides insight into the underlying molecular portrait of Luminal A-like breast cancer subtype. From an initial 76 miRNAs, 4 were validated with altered expression in the blood of women with Luminal A-like breast cancer. The expression profiles of these 3 miRNAs, in combination with mammography, has potential to facilitate accurate subtype-specific breast tumor detection.

DACH1: its role as a classifier of long term good prognosis in luminal breast cancer

BACKGROUND

Oestrogen receptor (ER) positive (luminal) tumours account for the largest proportion of females with breast cancer. Theirs is a heterogeneous disease presenting clinical challenges in managing their treatment. Three main biological luminal groups have been identified but clinically these can be distilled into two prognostic groups in which Luminal A are accorded good prognosis and Luminal B correlate with poor prognosis. Further biomarkers are needed to attain classification consensus. Machine learning approaches like Artificial Neural Networks (ANNs) have been used for classification and identification of biomarkers in breast cancer using high throughput data. In this study, we have used an artificial neural network (ANN) approach to identify DACH1 as a candidate luminal marker and its role in predicting clinical outcome in breast cancer is assessed.

MATERIALS AND METHODS

A reiterative ANN approach incorporating a network inferencing algorithm was used to identify ER-associated biomarkers in a publically available cDNA microarray dataset. DACH1 was identified in having a strong influence on ER associated markers and a positive association with ER. Its clinical relevance in predicting breast cancer specific survival was investigated by statistically assessing protein expression levels after immunohistochemistry in a series of unselected breast cancers, formatted as a tissue microarray.

RESULTS

Strong nuclear DACH1 staining is more prevalent in tubular and lobular breast cancer. Its expression correlated with ER-alpha positive tumours expressing PgR, epithelial cytokeratins (CK)18/19 and 'luminal-like' markers of good prognosis including FOXA1 and RERG (p<0.05). DACH1 is increased in patients showing longer cancer specific survival and disease free interval and reduced metastasis formation (p<0.001). Nuclear DACH1 showed a negative association with markers of aggressive growth and poor prognosis.

CONCLUSION

Nuclear DACH1 expression appears to be a Luminal A biomarker predictive of good prognosis, but is not independent of clinical stage, tumour size, NPI status or systemic therapy.

A novel artificial neural network method for biomedical prediction based on matrix pseudo-inversion

Biomedical prediction based on clinical and genome-wide data has become increasingly important in disease diagnosis and classification. To solve the prediction problem in an effective manner for the improvement of clinical care, we develop a novel Artificial Neural Network (ANN) method based on Matrix Pseudo-Inversion (MPI) for use in biomedical applications. The MPI-ANN is constructed as a three-layer (i.e., input, hidden, and output layers) feed-forward neural network, and the weights connecting the hidden and output layers are directly determined based on MPI without a lengthy learning iteration. The LASSO (Least Absolute Shrinkage and Selection Operator) method is also presented for comparative purposes. Single Nucleotide Polymorphism (SNP) simulated data and real breast cancer data are employed to validate the performance of the MPI-ANN method via 5-fold cross validation. Experimental results demonstrate the efficacy of the developed MPI-ANN for disease classification and prediction, in view of the significantly superior accuracy (i.e., the rate of correct predictions), as compared with LASSO. The results based on the real breast cancer data also show that the MPI-ANN has better performance than other machine learning methods (including support vector machine (SVM), logistic regression (LR), and an iterative ANN). In addition, experiments demonstrate that our MPI-ANN could be used for bio-marker selection as well.

The potential of hypoxia markers as target for breast molecular imaging--a systematic review and meta-analysis of human marker expression

Background

Molecular imaging of breast cancer is a promising emerging technology, potentially able to improve clinical care. Valid imaging targets for molecular imaging tracer development are membrane-bound hypoxia-related proteins, expressed when tumor growth outpaces neo-angiogenesis. We performed a systematic literature review and meta-analysis of such hypoxia marker expression rates in human breast cancer to evaluate their potential as clinically relevant molecular imaging targets.

Methods

We searched MEDLINE and EMBASE for articles describing membrane-bound proteins that are related to hypoxia inducible factor 1α (HIF-1α), the key regulator of the hypoxia response. We extracted expression rates of carbonic anhydrase-IX (CAIX), glucose transporter-1 (GLUT1), C-X-C chemokine receptor type-4 (CXCR4), or insulin-like growth factor-1 receptor (IGF1R) in human breast disease, evaluated by immunohistochemistry. We pooled study results using random-effects models and applied meta-regression to identify associations with clinicopathological variables.

Results

Of 1,705 identified articles, 117 matched our selection criteria, totaling 30,216 immunohistochemistry results. We found substantial between-study variability in expression rates. Invasive cancer showed pooled expression rates of 35% for CAIX (95% confidence interval (CI): 26-46%), 51% for GLUT1 (CI: 40-61%), 46% for CXCR4 (CI: 33-59%), and 46% for IGF1R (CI: 35-70%). Expression rates increased with tumor grade for GLUT1, CAIX, and CXCR4 (all p < 0.001), but decreased for IGF1R (p < 0.001). GLUT1 showed the highest expression rate in grade III cancers with 58% (45-69%). CXCR4 showed the highest expression rate in small T1 tumors with 48% (CI: 28-69%), but associations with size were only significant for CAIX (p < 0.001; positive association) and IGF1R (p = 0.047; negative association). Although based on few studies, CAIX, GLUT1, and CXCR4 showed profound lower expression rates in normal breast tissue and benign breast disease (p < 0.001), and high rates in carcinoma in situ. Invasive lobular carcinoma consistently showed lower expression rates (p < 0.001).

Conclusions

Our results support the potential of hypoxia-related markers as breast cancer molecular imaging targets. Although specificity is promising, combining targets would be necessary for optimal sensitivity. These data could help guide the choice of imaging targets for tracer development depending on the envisioned clinical application.

Biomarker identification in breast cancer: Beta-adrenergic receptor signaling and pathways to therapeutic response

Recent preclinical studies have associated beta-adrenergic receptor (β-AR) signaling with breast cancer pathways such as progression and metastasis. These findings have been supported by clinical and epidemiological studies which examined the effect of beta-blocker therapy on breast cancer metastasis, recurrence and mortality. Results from these studies have provided initial evidence for the inhibition of cell migration in breast cancer by beta-blockers and have introduced the beta-adrenergic receptor pathways as a target for therapy. This paper analyzes gene expression profiles in breast cancer patients, utilising Artificial Neural Networks (ANNs) to identify molecular signatures corresponding to possible disease management pathways and biomarker treatment strategies associated with beta-2-adrenergic receptor (ADRB2) cell signaling. The adrenergic receptor relationship to cancer is investigated in order to validate the results of recent studies that suggest the use of beta-blockers for breast cancer therapy. A panel of genes is identified which has previously been reported to play an important role in cancer and also to be involved in the beta-adrenergic receptor signaling.

Network inference analysis identifies an APRR2-like gene linked to pigment accumulation in tomato and pepper fruits

Carotenoids represent some of the most important secondary metabolites in the human diet, and tomato (Solanum lycopersicum) is a rich source of these health-promoting compounds. In this work, a novel and fruit-related regulator of pigment accumulation in tomato has been identified by artificial neural network inference analysis and its function validated in transgenic plants. A tomato fruit gene regulatory network was generated using artificial neural network inference analysis and transcription factor gene expression profiles derived from fruits sampled at various points during development and ripening. One of the transcription factor gene expression profiles with a sequence related to an Arabidopsis (Arabidopsis thaliana) ARABIDOPSIS PSEUDO RESPONSE REGULATOR2-LIKE gene (APRR2-Like) was up-regulated at the breaker stage in wild-type tomato fruits and, when overexpressed in transgenic lines, increased plastid number, area, and pigment content, enhancing the levels of chlorophyll in immature unripe fruits and carotenoids in red ripe fruits. Analysis of the transcriptome of transgenic lines overexpressing the tomato APPR2-Like gene revealed up-regulation of several ripening-related genes in the overexpression lines, providing a link between the expression of this tomato gene and the ripening process. A putative ortholog of the tomato APPR2-Like gene in sweet pepper (Capsicum annuum) was associated with pigment accumulation in fruit tissues. We conclude that the function of this gene is conserved across taxa and that it encodes a protein that has an important role in ripening.

Molecular Pathology of the Breast

Information focuses on molecular pathology of breast cancer. Presented are clinical features of breast cancer, detailed discussion of histology and molecular pathology for invasive ductal carcinoma, invasive lobular carcinoma, other subtypes of invasive breast cancer, and breast cancer progression and precursors. Phenotypes and genotypes of breast cancer are presented, along with the role of biomarkers, gene profiling, and hormone receptors. Targeted therapies and prognostic indicators are presented with practical applications of molecular pathology for the surgical pathologist.

Gene expression profiles for predicting metastasis in breast cancer: a cross-study comparison of classification methods

Machine learning has increasingly been used with microarray gene expression data and for the development of classifiers using a variety of methods. However, method comparisons in cross-study datasets are very scarce. This study compares the performance of seven classification methods and the effect of voting for predicting metastasis outcome in breast cancer patients, in three situations: within the same dataset or across datasets on similar or dissimilar microarray platforms. Combining classification results from seven classifiers into one voting decision performed significantly better during internal validation as well as external validation in similar microarray platforms than the underlying classification methods. When validating between different microarray platforms, random forest, another voting-based method, proved to be the best performing method. We conclude that voting based classifiers provided an advantage with respect to classifying metastasis outcome in breast cancer patients.

Predicting two-year quality of life after breast cancer surgery using artificial neural network and linear regression models

The purpose of this study was to validate the use of artificial neural network (ANN) models for predicting quality of life (QOL) after breast cancer surgery and to compare the predictive capability of ANNs with that of linear regression (LR) models. The European Organization for Research and Treatment of Cancer Quality of Life Questionnaire and its supplementary breast cancer measure were completed by 402 breast cancer patients at baseline and at 2 years postoperatively. The accuracy of the system models were evaluated in terms of mean square error (MSE) and mean absolute percentage error (MAPE). A global sensitivity analysis was also performed to assess the relative significance of input parameters in the system model and to rank the variables in order of importance. Compared to the LR model, the ANN model generally had smaller MSE and MAPE values in both the training and testing datasets. Most ANN models had MAPE values ranging from 4.70 to 19.96 %, and most had high prediction accuracy. The ANN model also outperformed the LR model in terms of prediction accuracy. According to global sensitivity analysis, pre-operative functional status was the best predictor of QOL after surgery. Compared with the conventional LR model, the ANN model in the study was more accurate for predicting patient-reported QOL and had higher overall performance indices. Further refinements are expected to obtain sufficient performance improvements for its routine use in clinical practice as an adjunctive decision-making tool.

Calpastatin is associated with lymphovascular invasion in breast cancer

Metastasis of breast cancer is a major contributor to mortality. Histological assessment of vascular invasion (VI) provides important prognostic information and demonstrates that VI occurs predominantly via lymphatics in breast cancer. We sought to examine genes and proteins involved in lymphovascular invasion (LVI) to understand the mechanisms of this key disease process. A gene expression array of 91 breast cancer patients was analysed by an Artificial Neural Network (ANN) approach using LVI to supervise the analysis. 89 transcripts were significantly associated (p<0.001) with the presence of LVI. Calpastatin, a specific calpain inhibitor, had the second lowest selection error and was investigated in breast cancer specimens using real-time PCR (n=56) and immunohistochemistry (n=53). Both calpastatin mRNA and protein levels were significantly associated with the presence of LVI (p=0.014 and p=0.025 respectively). The data supports the hypothesis that calpastatin may play a role in regulating the initial metastatic dissemination of breast cancer.

Luminal and basal-like breast cancer cells show increased migration induced by hypoxia, mediated by an autocrine mechanism

Background

Some breast cancer patients receiving anti-angiogenic treatment show increased metastases, possibly as a result of induced hypoxia. The effect of hypoxia on tumor cell migration was assessed in selected luminal, post-EMT and basal-like breast carcinoma cell lines.

Methods

Migration was assessed in luminal (MCF-7), post-EMT (MDA-MB-231, MDA-MB-435S), and basal-like (MDA-MB-468) human breast carcinoma cell lines under normal and oxygen-deprived conditions, using a collagen-based assay. Cell proliferation was determined, secreted cytokine and chemokine levels were measured using flow-cytometry and a bead-based immunoassay, and the hypoxic genes HIF-1α and CA IX were assessed using PCR. The functional effect of tumor-cell conditioned medium on the migration of neutrophil granulocytes (NG) was tested.

Results

Hypoxia caused increased migratory activity but not proliferation in all tumor cell lines, involving the release and autocrine action of soluble mediators. Conditioned medium (CM) from hypoxic cells induced migration in normoxic cells. Hypoxia changed the profile of released inflammatory mediators according to cell type. Interleukin-8 was produced only by post-EMT and basal-like cell lines, regardless of hypoxia. MCP-1 was produced by MDA-MB-435 and -468 cells, whereas IL-6 was present only in MDA-MB-231. IL-2, TNF-α, and NGF production was stimulated by hypoxia in MCF-7 cells. CM from normoxic and hypoxic MDA-MB-231 and MDA-MB-435S cells and hypoxic MCF-7 cells, but not MDA-MB-468, induced NG migration.

Conclusions

Hypoxia increases migration by the autocrine action of released signal substances in selected luminal and basal-like breast carcinoma cell lines which might explain why anti-angiogenic treatment can worsen clinical outcome in some patients.

Luminal and basal-like breast cancer cells show increased migration induced by hypoxia, mediated by an autocrine mechanism

BACKGROUND

Some breast cancer patients receiving anti-angiogenic treatment show increased metastases, possibly as a result of induced hypoxia. The effect of hypoxia on tumor cell migration was assessed in selected luminal, post-EMT and basal-like breast carcinoma cell lines.

METHODS

Migration was assessed in luminal (MCF-7), post-EMT (MDA-MB-231, MDA-MB-435S), and basal-like (MDA-MB-468) human breast carcinoma cell lines under normal and oxygen-deprived conditions, using a collagen-based assay. Cell proliferation was determined, secreted cytokine and chemokine levels were measured using flow-cytometry and a bead-based immunoassay, and the hypoxic genes HIF-1α and CA IX were assessed using PCR. The functional effect of tumor-cell conditioned medium on the migration of neutrophil granulocytes (NG) was tested.

RESULTS

Hypoxia caused increased migratory activity but not proliferation in all tumor cell lines, involving the release and autocrine action of soluble mediators. Conditioned medium (CM) from hypoxic cells induced migration in normoxic cells. Hypoxia changed the profile of released inflammatory mediators according to cell type. Interleukin-8 was produced only by post-EMT and basal-like cell lines, regardless of hypoxia. MCP-1 was produced by MDA-MB-435 and -468 cells, whereas IL-6 was present only in MDA-MB-231. IL-2, TNF-α, and NGF production was stimulated by hypoxia in MCF-7 cells. CM from normoxic and hypoxic MDA-MB-231 and MDA-MB-435S cells and hypoxic MCF-7 cells, but not MDA-MB-468, induced NG migration.

CONCLUSIONS

Hypoxia increases migration by the autocrine action of released signal substances in selected luminal and basal-like breast carcinoma cell lines which might explain why anti-angiogenic treatment can worsen clinical outcome in some patients.

Alpha- and beta-adrenergic receptor (AR) protein expression is associated with poor clinical outcome in breast cancer: an immunohistochemical study

Breast cancer mortality is frequently associated with metastatic disease. Metastasis models have shown adrenoceptor (AR) stimulation induces cell migration which is inhibited by adrenoceptor antagonist drugs. We investigated adrenoceptor protein expression in clinical breast tumours and its association with disease progression and prognosis. Immunohistochemistry on tissue microarrays was used to characterise α1b, α2c and β(2)2 adrenoceptor protein expression in operable breast tumours. Associations with tumour-relevant biological markers and clinical outcome were statistically assessed. Strong α1b expression occurred in large high grade (P < 0.0001), HER2+ (P < 0.0001) or basal-like (CK5/6, P = 0.0005; CK14, P = 0.0001; EGFR, P = 0.003) cancers, showing increased proliferation (Mib1, P = 0.002), decreased apoptosis (Bcl2, P < 0.0001) and poor NPI membership (P = 0.001). α1b expression correlated with poor cancer-specific survival (LR = 7.628, P = 0.022) and tumour recurrence (LR = 6.128, P = 0.047). Strong α2c was over-expressed in high grade (P = 0.007), HER3+ (P = 0.002) and HER4+ (P < 0.0001) cancers with borderline increase in EGFR, p53 and MIB1 proteins, and inverse association with hormonal (PgR, P = 0.002) phenotype. In contrast, strong β(2) expression occurred in small-size, luminal-like (ER+, P < 0.001) tumours of low grade (P < 0.001) and lymph node stage (P = 0.027) that showed poor prognosis when hormonal treatment was withheld. Adrenoceptors were not found to be independent predictors of clinical outcome. Alpha1b and α2c AR is over-expressed in basal-like breast tumours of poor prognosis. Strong β(2) adrenoceptor expression is seen in patients with a luminal (ER+) tumour phenotype and good prognosis, due to benefits derived from hormonal therapy. These findings suggest a possible role for targeted therapy using adrenoceptor antagonists.

RERG (Ras-like, oestrogen-regulated, growth-inhibitor) expression in breast cancer: a marker of ER-positive luminal-like subtype

Global gene expression profiling studies have classified breast cancer into a number of distinct biological and molecular classes with clinical relevance. The heterogeneous luminal group, which is largely characterised by oestrogen receptor (ER) expression, appears to contain distinct subgroups with differing behaviour. In this study, we analysed 47,293 gene transcripts in 128 invasive breast carcinomas (BC) using Artificial Neural Networks and a cross-validation analysis in combination with an ensemble sample classification to identify genes that can be used to subclassify ER+ luminal tumours. The results were validated using immunohistochemistry on TMAs containing 1,140 invasive breast cancers. Our results showed that the RERG gene is one of the highest ranked genes to differentiate between ER+ luminal-like and ER- non-luminal cancers based on a 10-fold external cross-validation analysis with an average classification accuracy of 89%. This was confirmed in our protein expression studies that showed RERG positive associations with markers of luminal differentiation including ER, luminal cytokeratins (CK19, CK18 and CK7/8) and FOXA1 (P = 0.004) and other markers of good prognosis in BC including small size, lower histologic grade and positive expression of androgen receptor, nuclear BRCA1, FHIT and cell cycle inhibitors p27 and p21. RERG expression was inversely associated with the proliferation marker MIB1 (P = 0.005) and p53. Strong RERG expression showed an association with longer breast cancer specific survival and distant metastasis free interval in the whole series as well as in the ER+ luminal group and these associations were independent of other prognostic variables. In conclusion, we used novel bioinformatics methods to identify candidate genes to characterise ER+ luminal-like breast cancer. RERG gene is a key marker of the luminal BC class and can be used to separate distinct prognostic subgroups.

Biomarkers for Basal-like Breast Cancer

Initially recognized through microarray-based gene expression profiling, basal-like breast cancer, for which we lack effective targeted therapies, is an aggressive form of carcinoma with a predilection for younger women. With some success, immunohistochemical studies have attempted to reproduce the expression profile classification of breast cancer through identification of subtype-specific biomarkers. This review aims to present an in depth summary and analysis of the current status of basal-like breast cancer biomarker research. While a number of biomarkers show promise for future clinical application, the next logical step is a comprehensive investigation of all biomarkers against a gene expression profile gold standard for breast cancer subtype assignment.

The proteins FABP7 and OATP2 are associated with the basal phenotype and patient outcome in human breast cancer

The basal-like or basal phenotype (BP) class of breast cancers have recently attracted attention as a poor prognostic form of breast cancer. However, BP appears to encompass biologically and clinically heterogeneous tumours, resulting in a lack of consensus definition of BP. We analysed 48,000 gene transcripts in 132 invasive breast carcinomas to identify two novel genes (OATP2 and FABP7) significantly associated with BP [defined by cytokeratin (CK)5/6 and/or CK14 positivity]. Using a series of invasive breast carcinoma cases (n = 899), prepared as tissue microarrays, we assessed OATP2 and FABP7 protein expression using immunohistochemistry to investigate associations with clinicopathological variables, patients' outcome and ability to refine BP classification. A total of 7.9 and 15.6% cases were OATP2 and FABP7 positive, respectively. OATP2 was associated with tumours of high histological grade (p < 0.01), ER and PgR negativity (p < 0.01) and shorter breast cancer-specific survival (p = 0.04). FABP7 expression was associated with lower lymph node stage (p < 0.01), ER and PgR negativity (p < 0.01). BP tumours which were FABP7 positive had a significantly longer BCSS (p = 0.05) and disease-free survival (p = 0.01) compared with FABP7 negative basal tumours (p < 0.01). OATP2 positive tumours were associated with adverse survival and increased risk of early recurrence. This study confirms the biological and clinical heterogeneity of the BP in breast cancer. We have identified a novel subgroup of basal tumours showing FABP7 expression that have significantly better clinical outcome. Further studies analysing the role of FABP7 are therefore warranted.

Combined T regulatory cell and Th2 expression profile identifies children with cow's milk allergy

The role of T regulatory cells in spontaneous recovery from cow's milk allergy (CMA) is unclear. We investigated the mRNA expression of 12 T-cell markers and the protein expression of CD4, CD25, CD127, FoxP3 after in vitro beta-lactoglobulin stimulation of peripheral blood mononuclear cells from children with persisting CMA (n=16), early recovery (n=20) or no atopy (n=21). Artificial neural networks with exhaustive search for all marker combinations revealed that markers FoxP3, Nfat-C2, IL-16 and GATA-3 distinguished patients with persisting CMA most accurately from other study groups. FoxP3 mRNA expression following beta-lactoglobulin stimulation was highest in children with persisting CMA. Also the FoxP3 intensity in CD4(+) CD25(high)CD127(low) cells was higher in children with CMA compared with non-atopic children. The expression profile of both Th2- and T regulatory cell-related genes thus reflects the clinical activity of CMA. Tolerance, in contrast, is not characterized by activation of circulating T regulatory cells.

Combinatorial biomarker expression in breast cancer

Current clinical management of breast cancer relies on the availability of robust clinicopathological variables and few well-defined biological markers. Recent microarray-based expression profiling studies have emphasised the importance of the molecular portraits of breast cancer and the possibility of classifying breast cancer into biologically and molecularly distinct groups. Subsequent large scale immunohistochemical studies have demonstrated that the added value of studying the molecular biomarker expression in combination rather than individually. Oestrogen (ER) and progesterone (PR) receptors and HER2 are currently used in routine pathological assessment of breast cancer. Additional biomarkers such as proliferation markers and 'basal' markers are likely to be included in the future. A better understanding of the prognostic and predictive value of combinatorial assessment of biomarker expression could lead to improved breast cancer management in routine clinical practice and would add to our knowledge concerning the variation in behaviour and response to therapy. Here, we review the evidence on the value of assessing biomarker expression in breast cancer individually and in combination and its relation to the recent molecular classification of breast cancer.