Balázs Győrffy

Online survival analysis software to assess the prognostic value of biomarkers using transcriptomic data in non-small-cell lung cancer.

In the last decade, optimized treatment for non-small cell lung cancer had lead to improved prognosis, but the overall survival is still very short. To further understand the molecular basis of the disease we have to identify biomarkers related to survival. Here we present the development of an online tool suitable for the real-time meta-analysis of published lung cancer microarray datasets to identify biomarkers related to survival. We searched the caBIG, GEO and TCGA repositories to identify samples with published gene expression data and survival information. Univariate and multivariate Cox regression analysis, Kaplan-Meier survival plot with hazard ratio and logrank P value are calculated and plotted in R. The complete analysis tool can be accessed online at: www.kmplot.com/lung. All together 1,715 samples of ten independent datasets were integrated into the system. As a demonstration, we used the tool to validate 21 previously published survival associated biomarkers. Of these, survival was best predicted by CDK1 (p<1E-16), CD24 (p<1E-16) and CADM1 (p = 7E-12) in adenocarcinomas and by CCNE1 (p = 2.3E-09) and VEGF (p = 3.3E-10) in all NSCLC patients. Additional genes significantly correlated to survival include RAD51, CDKN2A, OPN, EZH2, ANXA3, ADAM28 and ERCC1. In summary, we established an integrated database and an online tool capable of uni- and multivariate analysis for in silico validation of new biomarker candidates in non-small cell lung cancer.

Cutoff Finder: A Comprehensive and Straightforward Web Application Enabling Rapid Biomarker Cutoff Optimization

Gene or protein expression data are usually represented by metric or at least ordinal variables. In order to translate a continuous variable into a clinical decision, it is necessary to determine a cutoff point and to stratify patients into two groups each requiring a different kind of treatment. Currently, there is no standard method or standard software for biomarker cutoff determination. Therefore, we developed Cutoff Finder, a bundle of optimization and visualization methods for cutoff determination that is accessible online. While one of the methods for cutoff optimization is based solely on the distribution of the marker under investigation, other methods optimize the correlation of the dichotomization with respect to an outcome or survival variable. We illustrate the functionality of Cutoff Finder by the analysis of the gene expression of estrogen receptor (ER) and progesterone receptor (PgR) in breast cancer tissues. This distribution of these important markers is analyzed and correlated with immunohistologically determined ER status and distant metastasis free survival. Cutoff Finder is expected to fill a relevant gap in the available biometric software repertoire and will enable faster optimization of new diagnostic biomarkers. The tool can be accessed at http://molpath.charite.de/cutoff.

Cross-validation of survival associated biomarkers in gastric cancer using transcriptomic data of 1,065 patients.

Introduction Multiple gene expression based prognostic biomarkers have been repeatedly identified in gastric carcinoma. However, without confirmation in an independent validation study, their clinical utility is limited. Our goal was to establish a robust database enabling the swift validation of previous and future gastric cancer survival biomarker candidates. Results The entire database incorporates 1,065 gastric carcinoma samples, gene expression data. Out of 29 established markers, higher expression of BECN1 (HR = 0.68, p = 1.5E-05), CASP3 (HR = 0.5, p = 6E-14), COX2 (HR = 0.72, p = 0.0013), CTGF (HR = 0.72, p = 0.00051), CTNNB1 (HR = 0.47, p = 4.3E-15), MET (HR = 0.63, p = 1.3E-05), and SIRT1 (HR = 0.64, p = 2.2E-07) correlated to longer OS. Higher expression of BIRC5 (HR = 1.45, p = 1E-04), CNTN1 (HR = 1.44, p = 3.5E- 05), EGFR (HR = 1.86, p = 8.5E-11), ERCC1 (HR = 1.36, p = 0.0012), HER2 (HR = 1.41, p = 0.00011), MMP2 (HR = 1.78, p = 2.6E-09), PFKB4 (HR = 1.56, p = 3.2E-07), SPHK1 (HR = 1.61, p = 3.1E-06), SP1 (HR = 1.45, p = 1.6E-05), TIMP1 (HR = 1.92, p = 2.2E- 10) and VEGF (HR = 1.53, p = 5.7E-06) were predictive for poor OS. MATERIALS AND METHODS We integrated samples of three major cancer research centers (Berlin, Bethesda and Melbourne datasets) and publicly available datasets with available follow-up data to form a single integrated database. Subsequently, we performed a literature search for prognostic markers in gastric carcinomas (PubMed, 2012–2015) and re-validated their findings predicting first progression (FP) and overall survival (OS) using uni- and multivariate Cox proportional hazards regression analysis. Conclusions The major advantage of our analysis is that we evaluated all genes in the same set of patients thereby making direct comparison of the markers feasible. The best performing genes include BIRC5, CASP3, CTNNB1, TIMP-1, MMP-2, SIRT, and VEGF.

Gene signature of the metastatic potential of cutaneous melanoma: Too much for too little?

It was expected that with the advent of genomics, oncology may defeat the deadliest forms of cancer including malignant melanoma, but the past years have indicated that this is not the case. Despite the stunning success of genomics in defining markers or gene signatures for breast cancer prognosis and predicting therapies, there is virtually no progression in malignant melanoma. This is happening when experimental oncology or metastasis research is using several rodent and human melanoma models, when our knowledge on the metastatic cascade is actually derived from these models. Our critical analysis of these studies revealed several factors which might be responsible for this failure. First, it is evident, that these studies must be based on rigorous sample collection and basic pathological considerations, where divergent histological types of melanoma cannot be analysed universally. Secondly, without following basic consideration of metastasis biology, the majority of these studies were rarely based on primary tumors but frequently on various types of regional metastases. Third, successful expression profiling studies on other tumors such as breast cancer, provided evidences that the homogeneity of the patient cohort at least by clinicopathological stage is a critical element when defining prognostic signatures. Four studies attempted to define the prognostic signature of skin melanoma but only one based the study on the primary tumor resulting in heterogenous signatures with a minimal overlap (MCM3 and NFKBIZ). Four study attempted to define the invasiveness-signature in the primary tumor based on thickness or growth pattern discrimination identifying a 9-gene overlap which proved to be different from the prognostic signatures. On the other hand, seven studies analyzed various types of metastatic tissues (rarely visceral-, mostly cutaneous or lymphatic metastases) to define the metastasis-signatures, again with minimal overlap (AQP3, LGALS7 and SFN). Using seven GEO-based melanoma datasets we have performed a meta-analysis of the metastasis-gene signatures using normalization protocols. This analysis identified a 350-gene signature, the core of which was a 17-gene signature characterizing locoregional metastases where the individual components occurred in 3 studies: several members of this signature were extensively studied before in context of melanoma metastasis including WNT5A, EGFR, BCL2A1 and OPN. These data suggest that only efficient inter-disciplinary collaboration throughout genomic analysis of human skin melanoma could lead to major advances in defining relevant gene-sets appropriate for clinical prognostication or revealing basic molecular pathways of melanoma progression.

Tumor-selective proteotoxicity of verteporfin inhibits colon cancer progression independently of YAP1

In an oxygen- and nutrient-deprived environment, verteporfin kills colon cancer cells by inducing proteotoxicity. Aggregates kill cancer cells The drug verteporfin is used clinically to enhance phototherapy and may also inhibit the transcription factor YAP1, which is often active in cancers. However, Zhang et al. found a different path to toxicity for verteporfin-mediated death of colorectal cancer cells. Verteporfin triggered the accumulation of toxic amounts of protein oligomers that selectively killed colorectal cancer cells in mice and in cells cultured under hypoxic and nutrient-deprived conditions. Normal cells in culture and in tumor-adjacent tissue sections from mice cleared these aggregates through autophagy and survived. Thus, verteporfin produces tumor-selective proteotoxicity, which may be a useful therapeutic for patients with solid tumors. Yes-associated protein 1 (YAP1) is a transcriptional coactivator in the Hippo signaling pathway. Increased YAP1 activity promotes the growth of tumors, including that of colorectal cancer (CRC). Verteporfin, a drug that enhances phototherapy to treat neovascular macular degeneration, is an inhibitor of YAP1. We found that verteporfin inhibited tumor growth independently of its effects on YAP1 or the related protein TAZ in genetically or chemically induced mouse models of CRC, in patient-derived xenografts, and in enteroid models of CRC. Instead, verteporfin exhibited in vivo selectivity for killing tumor cells in part by impairing the global clearance of high–molecular weight oligomerized proteins, particularly p62 (a sequestrome involved in autophagy) and STAT3 (signal transducer and activator of transcription 3; a transcription factor). Verteporfin inhibited cytokine-induced STAT3 activity and cell proliferation and reduced the viability of cultured CRC cells. Although verteporfin accumulated to a greater extent in normal cells than in tumor cells in vivo, experiments with cultured cells indicated that the normal cells efficiently cleared verteporfin-induced protein oligomers through autophagic and proteasomal pathways. Culturing CRC cells under hypoxic or nutrient-deprived conditions (modeling a typical CRC microenvironment) impaired the clearance of protein oligomers and resulted in cell death, whereas culturing cells under normoxic or glucose-replete conditions protected cell viability and proliferation in the presence of verteporfin. Furthermore, verteporfin suppressed the proliferation of other cancer cell lines even in the absence of YAP1, suggesting that verteporfin may be effective against multiple types of solid cancers.

miR-34a Silences c-SRC to Attenuate Tumor Growth in Triple-Negative Breast Cancer.

Triple-negative breast cancer (TNBC) is an aggressive subtype with no clinically proven biologically targeted treatment options. The molecular heterogeneity of TNBC and lack of high frequency driver mutations other than TP53 have hindered the development of new and effective therapies that significantly improve patient outcomes. miRNAs, global regulators of survival and proliferation pathways important in tumor development and maintenance, are becoming promising therapeutic agents. We performed miRNA-profiling studies in different TNBC subtypes to identify miRNAs that significantly contribute to disease progression. We found that miR-34a was lost in TNBC, specifically within mesenchymal and mesenchymal stem cell-like subtypes, whereas expression of miR-34a targets was significantly enriched. Furthermore, restoration of miR-34a in cell lines representing these subtypes inhibited proliferation and invasion, activated senescence, and promoted sensitivity to dasatinib by targeting the proto-oncogene c-SRC. Notably, SRC depletion in TNBC cell lines phenocopied the effects of miR-34a reintroduction, whereas SRC overexpression rescued the antitumorigenic properties mediated by miR-34a. miR-34a levels also increased when cells were treated with c-SRC inhibitors, suggesting a negative feedback exists between miR-34a and c-SRC. Moreover, miR-34a administration significantly delayed tumor growth of subcutaneously and orthotopically implanted tumors in nude mice, and was accompanied by c-SRC downregulation. Finally, we found that miR-34a and SRC levels were inversely correlated in human tumor specimens. Together, our results demonstrate that miR-34a exerts potent antitumorigenic effects in vitro and in vivo and suggests that miR-34a replacement therapy, which is currently being tested in human clinical trials, represents a promising therapeutic strategy for TNBC.

Differential epigenetic reprogramming in response to specific endocrine therapies promotes cholesterol biosynthesis and cellular invasion

Endocrine therapies target the activation of the oestrogen receptor alpha (ERα) via distinct mechanisms, but it is not clear whether breast cancer cells can adapt to treatment using drug-specific mechanisms. Here we demonstrate that resistance emerges via drug-specific epigenetic reprogramming. Resistant cells display a spectrum of phenotypical changes with invasive phenotypes evolving in lines resistant to the aromatase inhibitor (AI). Orthogonal genomics analysis of reprogrammed regulatory regions identifies individual drug-induced epigenetic states involving large topologically associating domains (TADs) and the activation of super-enhancers. AI-resistant cells activate endogenous cholesterol biosynthesis (CB) through stable epigenetic activation in vitro and in vivo. Mechanistically, CB sparks the constitutive activation of oestrogen receptors alpha (ERα) in AI-resistant cells, partly via the biosynthesis of 27-hydroxycholesterol. By targeting CB using statins, ERα binding is reduced and cell invasion is prevented. Epigenomic-led stratification can predict resistance to AI in a subset of ERα-positive patients.

APOBEC3B-Mediated Cytidine Deamination Is Required for Estrogen Receptor Action in Breast Cancer

Summary Estrogen receptor α (ERα) is the key transcriptional driver in a large proportion of breast cancers. We report that APOBEC3B (A3B) is required for regulation of gene expression by ER and acts by causing C-to-U deamination at ER binding regions. We show that these C-to-U changes lead to the generation of DNA strand breaks through activation of base excision repair (BER) and to repair by non-homologous end-joining (NHEJ) pathways. We provide evidence that transient cytidine deamination by A3B aids chromatin modification and remodelling at the regulatory regions of ER target genes that promotes their expression. A3B expression is associated with poor patient survival in ER+ breast cancer, reinforcing the physiological significance of A3B for ER action.

Aberrant DNA methylation impacts gene expression and prognosis in breast cancer subtypes

DNA methylation has a substantial impact on gene expression, affecting the prognosis of breast cancer (BC) patients dependent on molecular subtypes. In this study, we investigated the prognostic relevance of the expression of genes reported as aberrantly methylated, and the link between gene expression and DNA methylation in BC subtypes. The prognostic value of the expression of 144 aberrantly methylated genes was evaluated in ER+/HER2−, HER2+, and ER−/HER2− molecular BC subtypes, in a meta‐analysis of two large transcriptomic cohorts of BC patients (n = 1,938 and n = 1,640). The correlation between gene expression and DNA methylation in distinct gene regions was also investigated in an independent dataset of 104 BCs. Survival and Pearson correlation analyses were computed for each gene separately. The expression of 48 genes was significantly associated with BC prognosis (p < 0.05), and 32 of these prognostic genes exhibited a direct expression–methylation correlation. The expression of several immune‐related genes, including CD3D and HLA‐A, was associated with both relapse‐free survival (HR = 0.42, p = 3.5E‐06; HR = 0.35, p = 1.7E‐08) and overall survival (HR = 0.50, p = 5.5E‐04; HR = 0.68, p = 4.5E‐02) in ER‐/HER2‐ BCs. On the overall, the distribution of both positive and negative expression–methylation correlation in distinct gene regions have different effects on gene expression and prognosis in BC subtypes. This large‐scale meta‐analysis allowed the identification of several genes consistently associated with prognosis, whose DNA methylation could represent a promising biomarker for prognostication and clinical stratification of patients with distinct BC subtypes.

TP53 mutation‐correlated genes predict the risk of tumor relapse and identify MPS1 as a potential therapeutic kinase in TP53‐mutated breast cancers

Breast cancers (BC) carry a complex set of gene mutations that can influence their gene expression and clinical behavior. We aimed to identify genes driven by the TP53 mutation status and assess their clinical relevance in estrogen receptor (ER)‐positive and ER‐negative BC, and their potential as targets for patients with TP53 mutated tumors. Separate ROC analyses of each gene expression according to TP53 mutation status were performed. The prognostic value of genes with the highest AUC were assessed in a large dataset of untreated, and neoadjuvant chemotherapy treated patients. The mitotic checkpoint gene MPS1 was the most significant gene correlated with TP53 status, and the most significant prognostic marker in all ER‐positive BC datasets. MPS1 retained its prognostic value independently from the type of treatment administered. The biological functions of MPS1 were investigated in different BC cell lines. We also assessed the effects of a potent small molecule inhibitor of MPS1, SP600125, alone and in combination with chemotherapy. Consistent with the gene expression profiling and siRNA assays, the inhibition of MPS1 by SP600125 led to a reduction in cell viability and a significant increase in cell death, selectively in TP53‐mutated BC cells. Furthermore, the chemical inhibition of MPS1 sensitized BC cells to conventional chemotherapy, particularly taxanes. Our results collectively demonstrate that TP53‐correlated kinase MPS1, is a potential therapeutic target in BC patients with TP53 mutated tumors, and that SP600125 warrant further development in future clinical trials.