Expression profiling using a tumor-specific cDNA microarray predicts the prognosis of intermediate risk neuroblastomas

Lineage specific transcription factor waves reprogram neuroblastoma from self-renewal to differentiation

Temporal regulation of super-enhancer (SE) driven transcription factors (TFs) underlies normal developmental programs. Neuroblastoma (NB) arises from an inability of sympathoadrenal progenitors to exit a self-renewal program and terminally differentiate. To identify SEs driving TF regulators, we use all-trans retinoic acid (ATRA) to induce NB growth arrest and differentiation. Time-course H3K27ac ChIP-seq and RNA-seq reveal ATRA coordinated SE waves. SEs that decrease with ATRA link to stem cell development (MYCN, GATA3, SOX11). CRISPR-Cas9 and siRNA verify SOX11 dependency, in vitro and in vivo. Silencing the SOX11 SE using dCAS9-KRAB decreases SOX11 mRNA and inhibits cell growth. Other TFs activate in sequential waves at 2, 4 and 8 days of ATRA treatment that regulate neural development (GATA2 and SOX4). Silencing the gained SOX4 SE using dCAS9-KRAB decreases SOX4 expression and attenuates ATRA-induced differentiation genes. Our study identifies oncogenic lineage drivers of NB self-renewal and TFs critical for implementing a differentiation program.

Technological advancements in cancer diagnostics: Improvements and limitations

BACKGROUND

Cancer is characterized by the rampant proliferation, growth, and infiltration of malignantly transformed cancer cells past their normal boundaries into adjacent tissues. It is the leading cause of death worldwide, responsible for approximately 19.3 million new diagnoses and 10 million deaths globally in 2020. In the United States alone, the estimated number of new diagnoses and deaths is 1.9 million and 609 360, respectively. Implementation of currently existing cancer diagnostic techniques such as positron emission tomography (PET), X-ray computed tomography (CT), and magnetic resonance spectroscopy (MRS), and molecular diagnostic techniques, have enabled early detection rates and are instrumental not only for the therapeutic management of cancer patients, but also for early detection of the cancer itself. The effectiveness of these cancer screening programs are heavily dependent on the rate of accurate precursor lesion identification; an increased rate of identification allows for earlier onset treatment, thus decreasing the incidence of invasive cancer in the long-term, and improving the overall prognosis. Although these diagnostic techniques are advantageous due to lack of invasiveness and easier accessibility within the clinical setting, several limitations such as optimal target definition, high signal to background ratio and associated artifacts hinder the accurate diagnosis of specific types of deep-seated tumors, besides associated high cost. In this review we discuss various imaging, molecular, and low-cost diagnostic tools and related technological advancements, to provide a better understanding of cancer diagnostics, unraveling new opportunities for effective management of cancer, particularly in low- and middle-income countries (LMICs).

RECENT FINDINGS

Herein we discuss various technological advancements that are being utilized to construct an assortment of new diagnostic techniques that incorporate hardware, image reconstruction software, imaging devices, biomarkers, and even artificial intelligence algorithms, thereby providing a reliable diagnosis and analysis of the tumor. Also, we provide a brief account of alternative low cost-effective cancer therapy devices (CryoPop®, LumaGEM®, MarginProbe®) and picture archiving and communication systems (PACS), emphasizing the need for multi-disciplinary collaboration among radiologists, pathologists, and other involved specialties for improving cancer diagnostics.

CONCLUSION

Revolutionary technological advancements in cancer imaging and molecular biology techniques are indispensable for the accurate diagnosis and prognosis of cancer.

Comparative epigenomics by machine learning approach for neuroblastoma

BACKGROUND

Neuroblastoma (NB) is the second most common pediatric solid tumor. Because the number of genetic mutations found in tumors are small, even in some patients with unfavorable NB, epigenetic variation is expected to play an important role in NB progression. DNA methylation is a major epigenetic mechanism, and its relationship with NB prognosis has been a concern. One limitation with the analysis of variation in DNA methylation is the lack of a suitable analytical model. Therefore, in this study, we performed a random forest (RF) analysis of the DNA methylome data of NB from multiple databases.

RESULTS

RF is a popular machine learning model owing to its simplicity, intuitiveness, and computational cost. RF analysis identified novel intermediate-risk patient groups with characteristic DNA methylation patterns within the low-risk group. Feature selection analysis based on probe annotation revealed that enhancer-annotated regions had strong predictive power, particularly for MYCN-amplified NBs. We developed a gene-based analytical model to identify candidate genes related to disease progression, such as PRDM8 and FAM13A-AS1. RF analysis revealed sufficient predictive power compared to other machine learning models.

CONCLUSIONS

RF is a useful tool for DNA methylome analysis in cancer epigenetic studies, and has potential to identify a novel cancer-related genes.

Chromosome instability in neuroblastoma: A pathway to aggressive disease

For over 100-years, genomic instability has been investigated as a central player in the pathogenesis of human cancer. Conceptually, genomic instability includes an array of alterations from small deletions/insertions to whole chromosome alterations, referred to as chromosome instability. Chromosome instability has a paradoxical impact in cancer. In most instances, the introduction of chromosome instability has a negative impact on cellular fitness whereas in cancer it is usually associated with a worse prognosis. One exception is the case of neuroblastoma, the most common solid tumor outside of the brain in children. Neuroblastoma tumors have two distinct patterns of genome instability: whole-chromosome aneuploidy, which is associated with a better prognosis, or segmental chromosomal alterations, which is a potent negative prognostic factor. Through a computational screen, we found that low levels of the de- ubiquitinating enzyme USP24 have a highly significant negative impact on survival in neuroblastoma. At the molecular level, USP24 loss leads to destabilization of the microtubule assembly factor CRMP2 - producing mitotic errors and leading to chromosome missegregation and whole-chromosome aneuploidy. This apparent paradox may be reconciled through a model in which whole chromosome aneuploidy leads to the subsequent development of segmental chromosome alterations. Here we review the mechanisms behind chromosome instability and the evidence for the progressive development of segmental alterations from existing numerical aneuploidy in support of a multi-step model of neuroblastoma progression.

The PRC2 molecule EED is a target of epigenetic therapy for neuroblastoma

Epigenetic modifications by polycomb repressive complex (PRC) molecules appear to play a role in the tumorigenesis and aggressiveness of neuroblastoma (NB). Embryonic ectoderm development (EED) is a member of the PRC2 complex that binds to the H3K27me3 mark deposited by EZH2 via propagation on adjacent nucleosomes. We herein investigated the molecular roles of EED in MYCN-amplified NB cells using EED-knockdown (KD) shRNAs, EED-knockout sgRNAs, and the EED small molecule inhibitor EED226. The suppression of EED markedly inhibited NB cell proliferation and flat and soft agar colony formation. A transcriptome analysis using microarrays of EED-KD NB cells indicated the de-repression of cell cycle-regulated and differentiation-related genes. The results of a GSEA analysis suggested that inhibitory cell cycle-regulated gene sets were markedly up-regulated. Furthermore, an epigenetic treatment with the EED inhibitor EED226 and the HDAC inhibitors valproic acid/SAHA effectively suppressed NB cell proliferation and colony formation. This combined epigenetic treatment up-regulated cell cycle-regulated and differentiation-related genes. The ChIP sequencing analysis of histone codes and PRC molecules suggested an epigenetic background for the de-repression of down-regulated genes in MYCN-amplified/PRC2 up-regulated NB.

Suppressed miR-128-3p combined with TERT overexpression predicts dismal outcomes for neuroblastoma

BACKGROUND

Molecular and clinical diversity of neuroblastomas is notorious. The activating TERT rearrangements have been associated with dismal prognosis. Suppression of miR-128-3p may complement and enhance the adverse effects of TERT overexpression.

OBJECTIVE

The study aimed at evaluation of prognostic significance of the miR-128-3p/TERT expression in patients with primary neuroblastoma.

METHODS

RNA samples isolated from fresh-frozen tumor specimens (n= 103) were reverse transcribed for evaluation of miR-128-3p and TERT expression by qPCR. The normalized expression levels were tested for correlations with the event-free survival (EFS). ROC-analysis was used to establish threshold expression levels (TLs) for the possible best prediction of the outcomes. The median follow-up was 57 months.

RESULTS

Both TERT overexpression and miR-128-3p downregulation were independently associated with superior rates of adverse events (p= 0.027, TL =-2.32 log10 and p= 0.080, TL =-1.33 log10, respectively). The MYCN single-copy patients were stratified into groups based on the character of alterations in expression of the studied transcripts. Five-year EFS in the groups of patients with elevated TERT/normal miR-128-3p expression and normal TERT/reduced miR-128-3p expression were 0.74 ± 0.08 and 0.60 ± 0.16, respectively. The patients with elevated TERT/reduced miR-128-3p expression had the worst outcomes, with 5-year EFS of 0.40 ± 0.16 compared with 0.91 ± 0.06 for the patients with unaltered levels of both transcripts (p< 0.001). Cumulative incidence of relapse/progression for the groups constituted 0.23 ± 0.08, 0.40 ± 0.16, 0.60 ± 0.16 and 0.09 ± 0.06, respectively. Moreover, the loss of miR-128-3p was qualified as independent adverse predictor which outperformed the conventional clinical and genetic risk factors in the multivariate Cox regression model of EFS.

CONCLUSIONS

Combined expression levels of miR-128-3p and TERT represent a novel prognostic biomarker for neuroblastoma.

A Metastatic Neuroblastic Tumor in a 28-Month-old Boy: Unusual Spontaneous Regression From Neuroblastoma to Ganglioneuroma?

Neuroblastoma with bone metastasis is well known to have an extremely poor prognosis. We experienced the case of a patient with adrenal ganglioneuroblastoma (GNB) with metastases of subcutaneous nodules, a lymph node, and multiple bones. A pathologic examination of tumors from different sites revealed both GNB and ganglioneuroma. A genetic comparison between these tumors identified the same molecular signatures, suggesting the possibility of spontaneous differentiation in the remaining GNB. The patient has been healthy without aggressive chemotherapy, and the patient's pathologic urinary catecholamines normalized. Even if unusual, we have to recognize probable spontaneous differentiation from neuroblastoma to GNB and then to ganglioneuroma, even in sites of bone metastasis.

Deep analysis of neuroblastoma core regulatory circuitries using online databases and integrated bioinformatics shows their pan-cancer roles as prognostic predictors

AIM

Neuroblastoma is a heterogeneous childhood cancer derived from the neural crest. The dual cell identities of neuroblastoma include Mesenchymal (MES) and Adrenergic (ADRN). These identities are conferred by a small set of tightly-regulated transcription factors (TFs) binding super enhancers, collectively forming core regulatory circuitries (CRCs). The purpose of this study was to gain a deep understanding of the role of MES and ADRN TFs in neuroblastoma and other cancers as potential indicators of disease prognosis, progression, and relapse.

METHODS

To that end, we first investigated the expression and mutational profile of MES and ADRN TFs in neuroblastoma. Moreover, we established their correlation with neuroblastoma risk groups and overall survival while establishing their extended networks with long non-coding RNAs (lncRNAs). Furthermore, we analysed the pan-cancer expression and mutational profile of these TFs and their correlation with patient survival and finally their network connectivity, using a panel of bioinformatic tools including GEPIA2, human pathology atlas, TIMER2, Omicsnet, and Cytoscape.

RESULTS

We show the association of multiple MES and ADRN TFs with neuroblastoma risk groups and overall survival and find significantly higher expression of various MES and ADRN TFs compared to normal tissues and their association with overall survival and disease-free survival in multiple cancers. Moreover, we report the strong correlation of the expression of these TFs with the infiltration of stromal and immune cells in the tumour microenvironment and with stemness and metastasis-related genes. Furthermore, we reveal extended pan-cancer networks comprising these TFs that influence the tumour microenvironment and metastasis and may be useful indicators of cancer prognosis and patient survival.

CONCLUSION

Our meta-analysis shows the significance of MES and ADRN TFs as indicators of patient prognosis and the putative utility of these TFs as potential novel biomarkers.

NLRR1 Is a Potential Therapeutic Target in Neuroblastoma and MYCN-Driven Malignant Cancers

Receptor tyrosine kinases (RTKs) receive different modulation before transmitting proliferative signals. We previously identified neuronal leucine-rich repeat 1 (NLRR1) as a positive regulator of EGF and IGF-1 signals in high-risk neuroblastoma cells. Here, we show that NLRR1 is up-regulated in various adult cancers and acts as a key regulator of tumor cell proliferation. In the extracellular domains of NLRR1, fibronectin type III (FNIII) domain is responsible for its function to promote cell proliferation. We generated monoclonal antibodies against the extracellular domains of NLRR1 (N1mAb) and screened the positive N1mAbs for growth inhibitory effect. The treatment of N1mAbs reduces tumor cell proliferation in vitro and in vivo, and sensitizes the cells to EGFR inhibitor, suggesting that NLRR1 is a novel regulatory molecule of RTK function. Importantly, epitope mapping analysis has revealed that N1mAbs with growth inhibitory effect recognize immunoglobulin-like and FNIII domains of NLRR1, which also indicates the importance of FNIII domain in the function of NLRR1. Thus, the present study provides a new insight into the development of a cancer therapy by targeting NLRR1 as a modulator of proliferative signals on cellular membrane of tumor cells.

The Potential Functional Roles of NME1 Histidine Kinase Activity in Neuroblastoma Pathogenesis

Neuroblastoma is the most common extracranial solid tumor in childhood. Gain of chromosome 17q material is found in >60% of neuroblastoma tumors and is associated with poor patient prognosis. The NME1 gene is located in the 17q21.3 region, and high NME1 expression is correlated with poor neuroblastoma patient outcomes. However, the functional roles and signaling activity of NME1 in neuroblastoma cells and tumors are unknown. NME1 and NME2 have been shown to possess histidine (His) kinase activity. Using anti-1- and 3-pHis specific monoclonal antibodies and polyclonal anti-pH118 NME1/2 antibodies, we demonstrated the presence of pH118-NME1/2 and multiple additional pHis-containing proteins in all tested neuroblastoma cell lines and in xenograft neuroblastoma tumors, supporting the presence of histidine kinase activity in neuroblastoma cells and demonstrating the potential significance of histidine kinase signaling in neuroblastoma pathogenesis. We have also demonstrated associations between NME1 expression and neuroblastoma cell migration and differentiation. Our demonstration of NME1 histidine phosphorylation in neuroblastoma and of the potential role of NME1 in neuroblastoma cell migration and differentiation suggest a functional role for NME1 in neuroblastoma pathogenesis and open the possibility of identifying new therapeutic targets and developing novel approaches to neuroblastoma therapy.

Distillation of the clinical algorithm improves prognosis by multi-task deep learning in high-risk Neuroblastoma

We introduce the CDRP (Concatenated Diagnostic-Relapse Prognostic) architecture for multi-task deep learning that incorporates a clinical algorithm, e.g., a risk stratification schema to improve prognostic profiling. We present the first application to survival prediction in High-Risk (HR) Neuroblastoma from transcriptomics data, a task that studies from the MAQC consortium have shown to remain the hardest among multiple diagnostic and prognostic endpoints predictable from the same dataset. To obtain a more accurate risk stratification needed for appropriate treatment strategies, CDRP combines a first component (CDRP-A) synthesizing a diagnostic task and a second component (CDRP-N) dedicated to one or more prognostic tasks. The approach leverages the advent of semi-supervised deep learning structures that can flexibly integrate multimodal data or internally create multiple processing paths. CDRP-A is an autoencoder trained on gene expression on the HR/non-HR risk stratification by the Children’s Oncology Group, obtaining a 64-node representation in the bottleneck layer. CDRP-N is a multi-task classifier for two prognostic endpoints, i.e., Event-Free Survival (EFS) and Overall Survival (OS). CDRP-A provides the HR embedding input to the CDRP-N shared layer, from which two branches depart to model EFS and OS, respectively. To control for selection bias, CDRP is trained and evaluated using a Data Analysis Protocol (DAP) developed within the MAQC initiative. CDRP was applied on Illumina RNA-Seq of 498 Neuroblastoma patients (HR: 176) from the SEQC study (12,464 Entrez genes) and on Affymetrix Human Exon Array expression profiles (17,450 genes) of 247 primary diagnostic Neuroblastoma of the TARGET NBL cohort. On the SEQC HR patients, CDRP achieves Matthews Correlation Coefficient (MCC) 0.38 for EFS and MCC = 0.19 for OS in external validation, improving over published SEQC models. We show that a CDRP-N embedding is indeed parametrically associated to increasing severity and the embedding can be used to better stratify patients’ survival.

The roles of moonlight ribosomal proteins in the development of human cancers

"Moonlighting protein" is a term used to define a single protein with multiple functions and different activities that are not derived from gene fusions, multiple RNA splicing, or the proteolytic activity of promiscuous enzymes. Different proteinous constituents of ribosomes have been shown to have important moonlighting extra-ribosomal functions. In this review, we introduce the impact of key moonlight ribosomal proteins and dependent signal transduction in the initiation and progression of various cancers. As a future perspective, the potential role of these moonlight ribosomal proteins in the diagnosis, prognosis, and development of novel strategies to improve the efficacy of therapies for human cancers has been suggested.

Predicting neuroblastoma using developmental signals and a logic-based model

Genomic information from human patient samples of pediatric neuroblastoma cancers and known outcomes have led to specific gene lists put forward as high risk for disease progression. However, the reliance on gene expression correlations rather than mechanistic insight has shown limited potential and suggests a critical need for molecular network models that better predict neuroblastoma progression. In this study, we construct and simulate a molecular network of developmental genes and downstream signals in a 6-gene input logic model that predicts a favorable/unfavorable outcome based on the outcome of the four cell states including cell differentiation, proliferation, apoptosis, and angiogenesis. We simulate the mis-expression of the tyrosine receptor kinases, trkA and trkB, two prognostic indicators of neuroblastoma, and find differences in the number and probability distribution of steady state outcomes. We validate the mechanistic model assumptions using RNAseq of the SHSY5Y human neuroblastoma cell line to define the input states and confirm the predicted outcome with antibody staining. Lastly, we apply input gene signatures from 77 published human patient samples and show that our model makes more accurate disease outcome predictions for early stage disease than any current neuroblastoma gene list. These findings highlight the predictive strength of a logic-based model based on developmental genes and offer a better understanding of the molecular network interactions during neuroblastoma disease progression.

EZH2 regulates neuroblastoma cell differentiation via NTRK1 promoter epigenetic modifications

The polycomb repressor complex 2 molecule EZH2 is now known to play a role in essential cellular processes, namely, cell fate decisions, cell cycle regulation, senescence, cell differentiation, and cancer development/progression. EZH2 inhibitors have recently been developed; however, their effectiveness and underlying molecular mechanisms in many malignancies have not yet been elucidated in detail. Although the functional role of EZH2 in tumorigenesis in neuroblastoma (NB) has been investigated, mutations of EZH2 have not been reported. A Kaplan–Meier analysis on the event free survival and overall survival of NB patients indicated that the high expression of EZH2 correlated with an unfavorable prognosis. In order to elucidate the functional roles of EZH2 in NB tumorigenesis and its aggressiveness, we knocked down EZH2 in NB cell lines using lentivirus systems. The knockdown of EZH2 significantly induced NB cell differentiation, e.g., neurite extension, and the neuronal differentiation markers, NF68 and GAP43. EZH2 inhibitors also induced NB cell differentiation. We performed a comprehensive transcriptome analysis using Human Gene Expression Microarrays and found that NTRK1 (TrkA) is one of the EZH2-related suppression targets. The depletion of NTRK1 canceled EZH2 knockdown-induced NB cell differentiation. Our integrative methylome, transcriptome, and chromatin immunoprecipitation assays using NB cell lines and clinical samples clarified that the NTRK1 P1 and P2 promoter regions were regulated differently by DNA methylation and EZH2-related histone modifications. The NTRK1 transcript variants 1/2, which were regulated by EZH2-related H3K27me3 modifications at the P1 promoter region, were strongly expressed in favorable, but not unfavorable NB. The depletion and inhibition of EZH2 successfully induced NTRK1 transcripts and functional proteins. Collectively, these results indicate that EZH2 plays important roles in preventing the differentiation of NB cells and also that EZH2-related NTRK1 transcriptional regulation may be the key pathway for NB cell differentiation.

Molecular Classification Substitutes for the Prognostic Variables Stage, Age, and MYCN Status in Neuroblastoma Risk Assessment

BACKGROUND: Current risk stratification systems for neuroblastoma patients consider clinical, histopathological, and genetic variables, and additional prognostic markers have been proposed in recent years. We here sought to select highly informative covariates in a multistep strategy based on consecutive Cox regression models, resulting in a risk score that integrates hazard ratios of prognostic variables. METHODS: A cohort of 695 neuroblastoma patients was divided into a discovery set (n = 75) for multigene predictor generation, a training set (n = 411) for risk score development, and a validation set (n = 209). Relevant prognostic variables were identified by stepwise multivariable L1-penalized least absolute shrinkage and selection operator (LASSO) Cox regression, followed by backward selection in multivariable Cox regression, and then integrated into a novel risk score. RESULTS: The variables stage, age, MYCN status, and two multigene predictors, NB-th24 and NB-th44, were selected as independent prognostic markers by LASSO Cox regression analysis. Following backward selection, only the multigene predictors were retained in the final model. Integration of these classifiers in a risk scoring system distinguished three patient subgroups that differed substantially in their outcome. The scoring system discriminated patients with diverging outcome in the validation cohort (5-year event-free survival, 84.9 ± 3.4 vs 63.6 ± 14.5 vs 31.0 ± 5.4; P < .001), and its prognostic value was validated by multivariable analysis. CONCLUSION: We here propose a translational strategy for developing risk assessment systems based on hazard ratios of relevant prognostic variables. Our final neuroblastoma risk score comprised two multigene predictors only, supporting the notion that molecular properties of the tumor cells strongly impact clinical courses of neuroblastoma patients.

Applications of Machine Learning in Cancer Prediction and Prognosis

Machine learning is a branch of artificial intelligence that employs a variety of statistical, probabilistic and optimization techniques that allows computers to “learn” from past examples and to detect hard-to-discern patterns from large, noisy or complex data sets. This capability is particularly well-suited to medical applications, especially those that depend on complex proteomic and genomic measurements. As a result, machine learning is frequently used in cancer diagnosis and detection. More recently machine learning has been applied to cancer prognosis and prediction. This latter approach is particularly interesting as it is part of a growing trend towards personalized, predictive medicine. In assembling this review we conducted a broad survey of the different types of machine learning methods being used, the types of data being integrated and the performance of these methods in cancer prediction and prognosis. A number of trends are noted, including a growing dependence on protein biomarkers and microarray data, a strong bias towards applications in prostate and breast cancer, and a heavy reliance on “older” technologies such artificial neural networks (ANNs) instead of more recently developed or more easily interpretable machine learning methods. A number of published studies also appear to lack an appropriate level of validation or testing. Among the better designed and validated studies it is clear that machine learning methods can be used to substantially (15–25%) improve the accuracy of predicting cancer susceptibility, recurrence and mortality. At a more fundamental level, it is also evident that machine learning is also helping to improve our basic understanding of cancer development and progression.

Artificial neural network classifier predicts neuroblastoma patients' outcome

Background

More than fifty percent of neuroblastoma (NB) patients with adverse prognosis do not benefit from treatment making the identification of new potential targets mandatory. Hypoxia is a condition of low oxygen tension, occurring in poorly vascularized tissues, which activates specific genes and contributes to the acquisition of the tumor aggressive phenotype. We defined a gene expression signature (NB-hypo), which measures the hypoxic status of the neuroblastoma tumor. We aimed at developing a classifier predicting neuroblastoma patients’ outcome based on the assessment of the adverse effects of tumor hypoxia on the progression of the disease.

Methods

Multi-layer perceptron (MLP) was trained on the expression values of the 62 probe sets constituting NB-hypo signature to develop a predictive model for neuroblastoma patients’ outcome. We utilized the expression data of 100 tumors in a leave-one-out analysis to select and construct the classifier and the expression data of the remaining 82 tumors to test the classifier performance in an external dataset. We utilized the Gene set enrichment analysis (GSEA) to evaluate the enrichment of hypoxia related gene sets in patients predicted with “Poor” or “Good” outcome.

Results

We utilized the expression of the 62 probe sets of the NB-Hypo signature in 182 neuroblastoma tumors to develop a MLP classifier predicting patients’ outcome (NB-hypo classifier). We trained and validated the classifier in a leave-one-out cross-validation analysis on 100 tumor gene expression profiles. We externally tested the resulting NB-hypo classifier on an independent 82 tumors’ set. The NB-hypo classifier predicted the patients’ outcome with the remarkable accuracy of 87 %. NB-hypo classifier prediction resulted in 2 % classification error when applied to clinically defined low-intermediate risk neuroblastoma patients. The prediction was 100 % accurate in assessing the death of five low/intermediated risk patients. GSEA of tumor gene expression profile demonstrated the hypoxic status of the tumor in patients with poor prognosis.

Conclusions

We developed a robust classifier predicting neuroblastoma patients’ outcome with a very low error rate and we provided independent evidence that the poor outcome patients had hypoxic tumors, supporting the potential of using hypoxia as target for neuroblastoma treatment.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-016-1194-3) contains supplementary material, which is available to authorized users.

Transcript signatures that predict outcome and identify targetable pathways in MYCN-amplified neuroblastoma

BACKGROUND

In the pediatric cancer neuroblastoma (NB), patients are stratified into low, intermediate or high-risk subsets based in part on MYCN amplification status. While MYCN amplification in general predicts unfavorable outcome, no clinical or genomic factors have been identified that predict outcome within these cohorts of high-risk patients. In particular, it is currently not possible at diagnosis to determine which high-risk neuroblastoma patients will ultimately fail upfront therapy.

EXPERIMENTAL DESIGN

We analyzed the prognostic potential of most published gene expression signatures for NB and developed a new prognostic signature to predict outcome for patients with MYCN amplification. Network and pathway analyses identified candidate therapeutic targets for this MYCN-amplified patient subset with poor outcome.

RESULTS

Most signatures have a high capacity to predict outcome of unselected NB patients. However, the majority of published signatures, as well as most randomly generated signatures, are highly confounded by MYCN amplification, and fail to predict outcome in subpopulations of high-risk patients with MYCN-amplified NB. We identify a MYCN module signature that predicts patient outcome for those with MYCN-amplified tumors, that also predicts potential tractable therapeutic signaling pathways and targets including the DNA repair enzyme Poly [ADP-ribose] polymerase 1 (PARP1).

CONCLUSION

Many prognostic signatures for NB are confounded by MYCN amplification and fail to predict outcome for the subset of high-risk patients with MYCN amplification. We report a MYCN module signature that is associated with distinct patient outcomes, and predicts candidate therapeutic targets in DNA repair pathways, including PARP1 in MYCN-amplified NB.

Serum-Based Quantification of MYCN Gene Amplification in Young Patients with Neuroblastoma: Potential Utility as a Surrogate Biomarker for Neuroblastoma

We previously developed a method for determining MYCN gene amplification status using cell-free DNA fragments released from cancer cells into the blood of patients with neuroblastoma (NB). Here, we analyzed the relationship between MYCN amplification (MNA) status and neuroblastoma prognosis. We screened serum samples from 151 patients with NB for MNA, using real-time quantitative PCR, and compared the results with MYCN status determined using paired tumor samples. We additionally investigated whether MNA status correlates with patient survival. When a cut-off value of 5 was used, serum-based MNA analysis was found to show good sensitivity (86%) and very high specificity (95%). The sensitivities for stage 1 and 2 might be acceptable, even though it is not as good as for stage 3 and 4 (67% for stage 1 and 2, 92% for stage 3, and 87% for stage 4). MNA status correlated with overall survival in our cohort of 82 patients, with survival data available (p < 0.01). The hazard ratio of MNA status was 4.98 in patients diagnosed at less than 18 months of age (95% confidence interval, 1.00–24.78), and 1.41 (95% confidence interval, 0.63–3.14) for those diagnosed at 18 months of age or older. Serum-based MNA analysis is rapid and non-invasive compared with tumor-based MNA analysis, and has potential to predict tumor MNA status. There is still a room to improve the sensitivity of the test for tumors of stages 1 and 2, nonetheless this assay might help to determine therapeutic strategies prior to tumor biopsy, especially for patients with a life-threatening condition, as well as for patients of less than 18 months of age whose risk-grouping and treatment allocation depends on their MNA status.

Transcription factor activating protein 2 beta (TFAP2B) mediates noradrenergic neuronal differentiation in neuroblastoma

Neuroblastoma is an embryonal pediatric tumor that originates from the developing sympathetic nervous system and shows a broad range of clinical behavior, ranging from fatal progression to differentiation into benign ganglioneuroma. In experimental neuroblastoma systems, retinoic acid (RA) effectively induces neuronal differentiation, and RA treatment has been therefore integrated in current therapies. However, the molecular mechanisms underlying differentiation are still poorly understood. We here investigated the role of transcription factor activating protein 2 beta (TFAP2B), a key factor in sympathetic nervous system development, in neuroblastoma pathogenesis and differentiation. Microarray analyses of primary neuroblastomas (n = 649) demonstrated that low TFAP2B expression was significantly associated with unfavorable prognostic markers as well as adverse patient outcome. We also found that low TFAP2B expression was strongly associated with CpG methylation of the TFAP2B locus in primary neuroblastomas (n = 105) and demethylation with 5-aza-2'-deoxycytidine resulted in induction of TFAP2B expression in vitro, suggesting that TFAP2B is silenced by genomic methylation. Tetracycline inducible re-expression of TFAP2B in IMR-32 and SH-EP neuroblastoma cells significantly impaired proliferation and cell cycle progression. In IMR-32 cells, TFAP2B induced neuronal differentiation, which was accompanied by up-regulation of the catecholamine biosynthesizing enzyme genes DBH and TH, and down-regulation of MYCN and REST, a master repressor of neuronal genes. By contrast, knockdown of TFAP2B by lentiviral transduction of shRNAs abrogated RA-induced neuronal differentiation of SH-SY5Y and SK-N-BE(2)c neuroblastoma cells almost completely. Taken together, our results suggest that TFAP2B is playing a vital role in retaining RA responsiveness and mediating noradrenergic neuronal differentiation in neuroblastoma.

Comparison of RNA-seq and microarray-based models for clinical endpoint prediction

Background

Gene expression profiling is being widely applied in cancer research to identify biomarkers for clinical endpoint prediction. Since RNA-seq provides a powerful tool for transcriptome-based applications beyond the limitations of microarrays, we sought to systematically evaluate the performance of RNA-seq-based and microarray-based classifiers in this MAQC-III/SEQC study for clinical endpoint prediction using neuroblastoma as a model.

Results

We generate gene expression profiles from 498 primary neuroblastomas using both RNA-seq and 44 k microarrays. Characterization of the neuroblastoma transcriptome by RNA-seq reveals that more than 48,000 genes and 200,000 transcripts are being expressed in this malignancy. We also find that RNA-seq provides much more detailed information on specific transcript expression patterns in clinico-genetic neuroblastoma subgroups than microarrays. To systematically compare the power of RNA-seq and microarray-based models in predicting clinical endpoints, we divide the cohort randomly into training and validation sets and develop 360 predictive models on six clinical endpoints of varying predictability. Evaluation of factors potentially affecting model performances reveals that prediction accuracies are most strongly influenced by the nature of the clinical endpoint, whereas technological platforms (RNA-seq vs. microarrays), RNA-seq data analysis pipelines, and feature levels (gene vs. transcript vs. exon-junction level) do not significantly affect performances of the models.

Conclusions

We demonstrate that RNA-seq outperforms microarrays in determining the transcriptomic characteristics of cancer, while RNA-seq and microarray-based models perform similarly in clinical endpoint prediction. Our findings may be valuable to guide future studies on the development of gene expression-based predictive models and their implementation in clinical practice.

Electronic supplementary material

The online version of this article (doi:10.1186/s13059-015-0694-1) contains supplementary material, which is available to authorized users.

Expression of FOXP3, CD14, and ARG1 in Neuroblastoma Tumor Tissue from High-Risk Patients Predicts Event-Free and Overall Survival

The prognosis of children with metastatic neuroblastoma (NB) > 18 months at diagnosis is dismal. Since the immune status of the tumor microenvironment could play a role in the history of disease, we evaluated the expression of CD45, CD14, ARG1, CD163, CD4, FOXP3, Perforin-1 (PRF1), Granzyme B (GRMB), and IL-10 mRNAs in primary tumors at diagnosis from children with metastatic NB and tested whether the transcript levels are significantly associated to event-free and overall survival (EFS and OS, resp.). Children with high expression of CD14, ARG1 and FOXP3 mRNA in their primary tumors had significantly better EFS. Elevated expression of CD14, and FOXP3 mRNA was significantly associated to better OS. CD14 mRNA expression levels significantly correlated to all markers, with the exception of CD4. Strong positive correlations were found between PRF1 and CD163, as well as between PFR1 and FOXP3. It is worth noting that the combination of high levels of CD14, FOXP3, and ARG1 mRNAs identified a small group of patients with excellent EFS and OS, whereas low levels of CD14 were sufficient to identify patients with dismal survival. Thus, the immune status of the primary tumors of high-risk NB patients may influence the natural history of this pediatric cancer.

Functional interplay between MYCN, NCYM, and OCT4 promotes aggressiveness of human neuroblastomas

Neuroblastoma is a pediatric solid tumor that originates from embryonic neural crest cells. The MYCN gene locus is frequently amplified in unfavorable neuroblastomas, and the gene product promotes the progression of neuroblastomas. However, the molecular mechanisms by which MYCN amplification contributes to stem cell-like states of neuroblastoma remain elusive. In this study, we show that MYCN and its cis-antisense gene, NCYM, form a positive feedback loop with OCT4, a core regulatory gene maintaining a multipotent state of neural stem cells. We previously reported that NCYM is co-amplified with the MYCN gene in primary human neuroblastomas and that the gene product promotes aggressiveness of neuroblastoma by stabilization of MYCN. In 36 MYCN-amplified primary human neuroblastomas, OCT4 mRNA expression was associated with unfavorable prognosis and was correlated with that of NCYM. The OCT4 protein induced both NCYM and MYCN in human neuroblastoma cells, whereas NCYM stabilized MYCN to induce OCT4 and stem cell-related genes, including NANOG, SOX2, and LIN28. In sharp contrast to MYCN, enforced expression of c-MYC did not enhance OCT4 expression in human neuroblastoma cells. All-trans retinoic acid treatment reduced MYCN, NCYM, and OCT4 expression, accompanied by the decreased amount of OCT4 recruited onto the intron 1 region of MYCN. Knockdown of NCYM or OCT4 inhibited formation of spheres of neuroblastoma cells and promoted asymmetric cell division in MYCN-amplified human neuroblastoma cells. These results suggest that the functional interplay between MYCN, NCYM, and OCT4 contributes to aggressiveness of MYCN-amplified human neuroblastomas.

HOXD-AS1 is a novel lncRNA encoded in HOXD cluster and a marker of neuroblastoma progression revealed via integrative analysis of noncoding transcriptome

Background

Long noncoding RNAs (lncRNAs) constitute a major, but poorly characterized part of human transcriptome. Recent evidence indicates that many lncRNAs are involved in cancer and can be used as predictive and prognostic biomarkers. Significant fraction of lncRNAs is represented on widely used microarray platforms, however they have usually been ignored in cancer studies.

Results

We developed a computational pipeline to annotate lncRNAs on popular Affymetrix U133 microarrays, creating a resource allowing measurement of expression of 1581 lncRNAs. This resource can be utilized to interrogate existing microarray datasets for various lncRNA studies. We found that these lncRNAs fall into three distinct classes according to their statistical distribution by length. Remarkably, these three classes of lncRNAs were co-localized with protein coding genes exhibiting distinct gene ontology groups. This annotation was applied to microarray analysis which identified a 159 lncRNA signature that discriminates between localized and metastatic stages of neuroblastoma. Analysis of an independent patient cohort revealed that this signature differentiates also relapsing from non-relapsing primary tumors. This is the first example of the signature developed via the analysis of expression of lncRNAs solely. One of these lncRNAs, termed HOXD-AS1, is encoded in HOXD cluster. HOXD-AS1 is evolutionary conserved among hominids and has all bona fide features of a gene. Studying retinoid acid (RA) response of SH-SY5Y cell line, a model of human metastatic neuroblastoma, we found that HOXD-AS1 is a subject to morphogenic regulation, is activated by PI3K/Akt pathway and itself is involved in control of RA-induced cell differentiation. Knock-down experiments revealed that HOXD-AS1 controls expression levels of clinically significant protein-coding genes involved in angiogenesis and inflammation, the hallmarks of metastatic cancer.

Conclusions

Our findings greatly extend the number of noncoding RNAs functionally implicated in tumor development and patient treatment and highlight their role as potential prognostic biomarkers of neuroblastomas.

Identifying rare events in rare diseases

Utilizing genomic signatures from diagnostic tumor samples to forecast clinical behavior and response to therapy has long been a goal, and we are now poised to further refine how we can identify the relatively rare patients with aggressive neuroblastoma masquerading as patients with a more benign form of the disease. Clin Cancer Res; 21(8); 1782-5. ©2014 AACR. See related article by Oberthuer et al., p. 1904.

Revised risk estimation and treatment stratification of low- and intermediate-risk neuroblastoma patients by integrating clinical and molecular prognostic markers

PURPOSE

To optimize neuroblastoma treatment stratification, we aimed at developing a novel risk estimation system by integrating gene expression-based classification and established prognostic markers.

EXPERIMENTAL DESIGN

Gene expression profiles were generated from 709 neuroblastoma specimens using customized 4 × 44 K microarrays. Classification models were built using 75 tumors with contrasting courses of disease. Validation was performed in an independent test set (n = 634) by Kaplan-Meier estimates and Cox regression analyses.

RESULTS

The best-performing classifier predicted patient outcome with an accuracy of 0.95 (sensitivity, 0.93; specificity, 0.97) in the validation cohort. The highest potential clinical value of this predictor was observed for current low-risk patients [5-year event-free survival (EFS), 0.84 ± 0.02 vs. 0.29 ± 0.10; 5-year overall survival (OS), 0.99 ± 0.01 vs. 0.76 ± 0.11; both P < 0.001] and intermediate-risk patients (5-year EFS, 0.88 ± 0.06 vs. 0.41 ± 0.10; 5-year OS, 1.0 vs. 0.70 ± 0.09; both P < 0.001). In multivariate Cox regression models for low-risk/intermediate-risk patients, the classifier outperformed risk assessment of the current German trial NB2004 [EFS: hazard ratio (HR), 5.07; 95% confidence interval (CI), 3.20-8.02; OS: HR, 25.54; 95% CI, 8.40-77.66; both P < 0.001]. On the basis of these findings, we propose to integrate the classifier into a revised risk stratification system for low-risk/intermediate-risk patients. According to this system, we identified novel subgroups with poor outcome (5-year EFS, 0.19 ± 0.08; 5-year OS, 0.59 ± 0.1), for whom we propose intensified treatment, and with beneficial outcome (5-year EFS, 0.87 ± 0.05; 5-year OS, 1.0), who may benefit from treatment de-escalation.

CONCLUSIONS

Combination of gene expression-based classification and established prognostic markers improves risk estimation of patients with low-risk/intermediate-risk neuroblastoma. We propose to implement our revised treatment stratification system in a prospective clinical trial.

Use of Attribute Driven Incremental Discretization and Logic Learning Machine to build a prognostic classifier for neuroblastoma patients

BACKGROUND

Cancer patient's outcome is written, in part, in the gene expression profile of the tumor. We previously identified a 62-probe sets signature (NB-hypo) to identify tissue hypoxia in neuroblastoma tumors and showed that NB-hypo stratified neuroblastoma patients in good and poor outcome 1. It was important to develop a prognostic classifier to cluster patients into risk groups benefiting of defined therapeutic approaches. Novel classification and data discretization approaches can be instrumental for the generation of accurate predictors and robust tools for clinical decision support. We explored the application to gene expression data of Rulex, a novel software suite including the Attribute Driven Incremental Discretization technique for transforming continuous variables into simplified discrete ones and the Logic Learning Machine model for intelligible rule generation.

RESULTS

We applied Rulex components to the problem of predicting the outcome of neuroblastoma patients on the bases of 62 probe sets NB-hypo gene expression signature. The resulting classifier consisted in 9 rules utilizing mainly two conditions of the relative expression of 11 probe sets. These rules were very effective predictors, as shown in an independent validation set, demonstrating the validity of the LLM algorithm applied to microarray data and patients' classification. The LLM performed as efficiently as Prediction Analysis of Microarray and Support Vector Machine, and outperformed other learning algorithms such as C4.5. Rulex carried out a feature selection by selecting a new signature (NB-hypo-II) of 11 probe sets that turned out to be the most relevant in predicting outcome among the 62 of the NB-hypo signature. Rules are easily interpretable as they involve only few conditions.

CONCLUSIONS

Our findings provided evidence that the application of Rulex to the expression values of NB-hypo signature created a set of accurate, high quality, consistent and interpretable rules for the prediction of neuroblastoma patients' outcome. We identified the Rulex weighted classification as a flexible tool that can support clinical decisions. For these reasons, we consider Rulex to be a useful tool for cancer classification from microarray gene expression data.

Chromosome 17/17q gain and unaltered profiles in high resolution array-CGH are prognostically informative in neuroblastoma

The prognostic relevance of chromosome 17 gain in neuroblastoma is still discussed. This investigation specifies the frequency, type, size, and transcriptional relevance in a large patient cohort. Primary tumor material of 202 patients was analyzed using high-resolution oligonucleotide array-based comparative genomic hybridization (aCGH) and correlated with clinical and survival data. A subset (n = 145) was correlated for differentially expressed genes (DEG) by microarray analysis. Chromosome 17 aCGH analysis showed numerical gain in 94/202 patients (47%), partial gain in 93/202 patients (46%), and no gain in 15/202 patients (7%). The frequency of partial gain was higher in stage 4 neuroblastoma (stage 1 15%; stage 2 12%; stage 3 16%; stage 4S 7%; and stage 4 50%). Overall survival (OS) was superior in patients with numerical gain compared with patients with partial gain or no gain (5-y-OS: 0.95 ± 0.02 vs. 0.63 ± 0.05 vs. 0.60 ± 0.13; P < 0.001). Gene expression analysis demonstrated 95/130 DEGs between tumors with numerical or partial chromosome/no gain. Only one DEG (CCKBR) was detected comparing tumors with partial gain and those with no gain. In patients with partial gain, the distribution of breakpoints did not correlate with stage and 11q status, but with MYCN amplification and 1p status. The "best" breakpoints in cases with partial 17q gain were at 42.5 Mb for event-free and 26.6 Mb for OS. Numerical gain of chromosome 17 is associated with a better prognosis than partial and no gain. The group of tumors with partial gain was similar to the group without gain with respect to stage distribution, outcome, and gene expression profile.

Aryl hydrocarbon receptor downregulates MYCN expression and promotes cell differentiation of neuroblastoma

Neuroblastoma (NB) is the most common malignant disease of infancy. MYCN amplification is a prognostic factor for NB and is a sign of highly malignant disease and poor patient prognosis. In this study, we aimed to investigate novel MYCN-related genes and assess how they affect NB cell behavior. The different gene expression found in 10 MYCN amplification NB tumors and 10 tumors with normal MYCN copy number were analyzed using tissue oligonucleotide microarrays. Ingenuity Pathway Analysis was subsequently performed to identify the potential genes involved in MYCN regulation pathways. Aryl hydrocarbon receptor (AHR), a receptor for dioxin-like compounds, was found to be inversely correlated with MYCN expression in NB tissues. This correlation was confirmed in a further 14 human NB samples. Moreover, AHR expression in NB tumors was found to correlate highly with histological grade of differentiation. In vitro studies revealed that AHR overexpression in NB cells induced spontaneous cell differentiation. In addition, it was found that ectopic expression of AHR suppressed MYCN promoter activity resulting in downregulation of MYCN expression. The suppression effect of AHR on the transcription of MYCN was compensated for by E2F1 overexpression, indicating that E2F1 is involved in the AHR-regulating MYCN pathway. Furthermore, AHR shRNA promotes the expression of E2F1 and MYCN in NB cells. These findings suggest that AHR is one of the upstream regulators of MYCN. Through the modulation of E2F1, AHR regulates MYCN gene expression, which may in turn affect NB differentiation.

NCYM, a Cis-antisense gene of MYCN, encodes a de novo evolved protein that inhibits GSK3β resulting in the stabilization of MYCN in human neuroblastomas

The rearrangement of pre-existing genes has long been thought of as the major mode of new gene generation. Recently, de novo gene birth from non-genic DNA was found to be an alternative mechanism to generate novel protein-coding genes. However, its functional role in human disease remains largely unknown. Here we show that NCYM, a cis-antisense gene of the MYCN oncogene, initially thought to be a large non-coding RNA, encodes a de novo evolved protein regulating the pathogenesis of human cancers, particularly neuroblastoma. The NCYM gene is evolutionally conserved only in the taxonomic group containing humans and chimpanzees. In primary human neuroblastomas, NCYM is 100% co-amplified and co-expressed with MYCN, and NCYM mRNA expression is associated with poor clinical outcome. MYCN directly transactivates both NCYM and MYCN mRNA, whereas NCYM stabilizes MYCN protein by inhibiting the activity of GSK3β, a kinase that promotes MYCN degradation. In contrast to MYCN transgenic mice, neuroblastomas in MYCN/NCYM double transgenic mice were frequently accompanied by distant metastases, behavior reminiscent of human neuroblastomas with MYCN amplification. The NCYM protein also interacts with GSK3β, thereby stabilizing the MYCN protein in the tumors of the MYCN/NCYM double transgenic mice. Thus, these results suggest that GSK3β inhibition by NCYM stabilizes the MYCN protein both in vitro and in vivo. Furthermore, the survival of MYCN transgenic mice bearing neuroblastoma was improved by treatment with NVP-BEZ235, a dual PI3K/mTOR inhibitor shown to destabilize MYCN via GSK3β activation. In contrast, tumors caused in MYCN/NCYM double transgenic mice showed chemo-resistance to the drug. Collectively, our results show that NCYM is the first de novo evolved protein known to act as an oncopromoting factor in human cancer, and suggest that de novo evolved proteins may functionally characterize human disease.

The MYCN oncogene has a central role in the genesis and progression of neuroblastomas, and its amplification is associated with an unfavorable prognosis. We have found that NCYM, a MYCN cis-antisense RNA, is translated in humans to a de novo evolved protein. NCYM inhibits GSK3β to stabilize MYCN, whereas MYCN induces NCYM transcription. The positive feedback regulation formed in the MYCN/NCYM-amplified tumors promotes the aggressive nature of human neuroblastoma. MYCN transgenic mice, which express human MYCN in sympathoadrenal tissues, spontaneously develop neuroblastomas. However, unlike human neuroblastoma, distant metastasis is infrequent. We established MYCN/NCYM double transgenic mice as a new animal model for studying neuroblastoma pathogenesis. We found that NCYM expression promoted both the metastasis and chemo-resistance of the neuroblastomas formed in the double transgenic mice. These results demonstrate that NCYM may be a potential target for developing novel therapeutic tools against high-risk neuroblastomas in humans, and that the MYCN/NCYM double transgenic mouse may be a suitable model for the screening of these new drugs.

Apoptotic cell death in neuroblastoma

Neuroblastoma (NB) is one of the most common malignant solid tumors in childhood, which derives from the sympathoadrenal lineage of the neural crest and exhibits extremely heterogeneous biological and clinical behaviors. The infant patients frequently undergo spontaneous regression even with metastatic disease, whereas the patients of more than one year of age who suffer from disseminated disease have a poor outcome despite intensive multimodal treatment. Spontaneous regression in favorable NBs has been proposed to be triggered by nerve growth factor (NGF) deficiency in the tumor with NGF dependency for survival, while aggressive NBs have defective apoptotic machinery which enables the tumor cells to evade apoptosis and confers the resistance to treatment. This paper reviews the molecules and pathways that have been recently identified to be involved in apoptotic cell death in NB and discusses their potential prospects for developing more effective therapeutic strategies against aggressive NB.

Logic Learning Machine creates explicit and stable rules stratifying neuroblastoma patients

BACKGROUND

Neuroblastoma is the most common pediatric solid tumor. About fifty percent of high risk patients die despite treatment making the exploration of new and more effective strategies for improving stratification mandatory. Hypoxia is a condition of low oxygen tension occurring in poorly vascularized areas of the tumor associated with poor prognosis. We had previously defined a robust gene expression signature measuring the hypoxic component of neuroblastoma tumors (NB-hypo) which is a molecular risk factor. We wanted to develop a prognostic classifier of neuroblastoma patients' outcome blending existing knowledge on clinical and molecular risk factors with the prognostic NB-hypo signature. Furthermore, we were interested in classifiers outputting explicit rules that could be easily translated into the clinical setting.

RESULTS

Shadow Clustering (SC) technique, which leads to final models called Logic Learning Machine (LLM), exhibits a good accuracy and promises to fulfill the aims of the work. We utilized this algorithm to classify NB-patients on the bases of the following risk factors: Age at diagnosis, INSS stage, MYCN amplification and NB-hypo. The algorithm generated explicit classification rules in good agreement with existing clinical knowledge. Through an iterative procedure we identified and removed from the dataset those examples which caused instability in the rules. This workflow generated a stable classifier very accurate in predicting good and poor outcome patients. The good performance of the classifier was validated in an independent dataset. NB-hypo was an important component of the rules with a strength similar to that of tumor staging.

CONCLUSIONS

The novelty of our work is to identify stability, explicit rules and blending of molecular and clinical risk factors as the key features to generate classification rules for NB patients to be conveyed to the clinic and to be used to design new therapies. We derived, through LLM, a set of four stable rules identifying a new class of poor outcome patients that could benefit from new therapies potentially targeting tumor hypoxia or its consequences.

The role of genetic and epigenetic alterations in neuroblastoma disease pathogenesis

Neuroblastoma is a highly heterogeneous tumor accounting for 15 % of all pediatric cancer deaths. Clinical behavior ranges from the spontaneous regression of localized, asymptomatic tumors, as well as metastasized tumors in infants, to rapid progression and resistance to therapy. Genomic amplification of the MYCN oncogene has been used to predict outcome in neuroblastoma for over 30 years, however, recent methodological advances including miRNA and mRNA profiling, comparative genomic hybridization (array-CGH), and whole-genome sequencing have enabled the detailed analysis of the neuroblastoma genome, leading to the identification of new prognostic markers and better patient stratification. In this review, we will describe the main genetic factors responsible for these diverse clinical phenotypes in neuroblastoma, the chronology of their discovery, and the impact on patient prognosis.

Temozolomide suppresses MYC via activation of TAp63 to inhibit progression of human glioblastoma

Glioblastoma multiforme (GBM) is a highly invasive and chemoradioresistant brain malignancy. Temozolomide (TMZ), a DNA-alkylating agent, is effective against GBM and has become the standard first-line drug. However, the mechanism by which TMZ regulates the progression of GBM remains elusive. Here, we demonstrate that TMZ targets TAp63, a p53 family member, inducing its expression to suppress the progression of human GBM. High levels of TAp63 expression in GBM tissues after TMZ treatment was an indicator of favourable prognosis. In human GBM cells, TMZ-induced TAp63 directly repressed MYC transcription. Activation of this TAp63-MYC pathway by TMZ inhibited human GBM progression both in vitro and in vivo. Furthermore, downregulation of MYC mRNA levels in recurrent GBMs after TMZ treatment correlated with better patient survival. Therefore, our results suggest that the TAp63-mediated transcriptional repression of MYC is a novel pathway regulating TMZ efficacy in GBM.

High genomic instability predicts survival in metastatic high-risk neuroblastoma

We aimed to identify novel molecular prognostic markers to better predict relapse risk estimate for children with high-risk (HR) metastatic neuroblastoma (NB). We performed genome- and/or transcriptome-wide analyses of 129 stage 4 HR NBs. Children older than 1 year of age were categorized as "short survivors" (dead of disease within 5 years from diagnosis) and "long survivors" (alive with an overall survival time ≥ 5 years). We reported that patients with less than three segmental copy number aberrations in their tumor represent a molecularly defined subgroup with a high survival probability within the current HR group of patients. The complex genomic pattern is a prognostic marker independent of NB-associated chromosomal aberrations, i.e., MYCN amplification, 1p and 11q losses, and 17q gain. Integrative analysis of genomic and expression signatures demonstrated that fatal outcome is mainly associated with loss of cell cycle control and deregulation of Rho guanosine triphosphates (GTPases) functioning in neuritogenesis. Tumors with MYCN amplification show a lower chromosome instability compared to MYCN single-copy NBs (P = .0008), dominated by 17q gain and 1p loss. Moreover, our results suggest that the MYCN amplification mainly drives disruption of neuronal differentiation and reduction of cell adhesion process involved in tumor invasion and metastasis. Further validation studies are warranted to establish this as a risk stratification for patients.

Kidins220/ARMS depletion is associated with the neural-to Schwann-like transition in a human neuroblastoma cell line model

Peripheral neuroblastic tumors exist as a heterogeneous mixture of neuroblastic (N-type) cells and Schwannian stromal (S-type) cells. These stromal cells not only represent a differentiated and less aggressive fraction of the tumor, but also have properties that can influence the further differentiation of nearby malignant cells. In vitro neuroblastoma cultures exhibit similar heterogeneity with N-type and S-type cells representing the neuroblastic and stromal portions of the tumor, respectively, in behavior, morphology, and molecular expression patterns. In this study, we deplete kinase D-interacting substrate of 220kD (Kidins220) with an shRNA construct and thereby cause morphologic transition of the human SH-SY5Y neuroblastoma cell line from N-type to S-type. The resulting cells have similar morphology and expression profile to SH-EP1 cells, a native S-type cell line from the same parent cell line, and to SH-SY5Y cells treated with BrdU, a treatment that induces S-type morphology. Specifically, both Kidins220-deficient SH-SY5Y cells and native SH-EP1 cells demonstrate down-regulation of the genes DCX and STMN2, markers for the neuronal lineage. We further show that Kidins220, DCX and STMN2 are co-down-regulated in cells of S-type morphology generated by methods other than Kidins220 depletion. Finally, we report that the association of low Kidins220 expression with S-type morphology and low DCX and STMN2 expression is demonstrated in spontaneously occurring human peripheral neuroblastic tumors. We propose that Kidins220 is critical in N- to S-type transition of neural crest tumor cells.

A model of binding on DNA microarrays: understanding the combined effect of probe synthesis failure, cross-hybridization, DNA fragmentation and other experimental details of affymetrix arrays

BACKGROUND

DNA microarrays are used both for research and for diagnostics. In research, Affymetrix arrays are commonly used for genome wide association studies, resequencing, and for gene expression analysis. These arrays provide large amounts of data. This data is analyzed using statistical methods that quite often discard a large portion of the information. Most of the information that is lost comes from probes that systematically fail across chips and from batch effects. The aim of this study was to develop a comprehensive model for hybridization that predicts probe intensities for Affymetrix arrays and that could provide a basis for improved microarray analysis and probe development. The first part of the model calculates probe binding affinities to all the possible targets in the hybridization solution using the Langmuir isotherm. In the second part of the model we integrate details that are specific to each experiment and contribute to the differences between hybridization in solution and on the microarray. These details include fragmentation, wash stringency, temperature, salt concentration, and scanner settings. Furthermore, the model fits probe synthesis efficiency and target concentration parameters directly to the data. All the parameters used in the model have a well-established physical origin.

RESULTS

For the 302 chips that were analyzed the mean correlation between expected and observed probe intensities was 0.701 with a range of 0.88 to 0.55. All available chips were included in the analysis regardless of the data quality. Our results show that batch effects arise from differences in probe synthesis, scanner settings, wash strength, and target fragmentation. We also show that probe synthesis efficiencies for different nucleotides are not uniform.

CONCLUSIONS

To date this is the most complete model for binding on microarrays. This is the first model that includes both probe synthesis efficiency and hybridization kinetics/cross-hybridization. These two factors are sequence dependent and have a large impact on probe intensity. The results presented here provide novel insight into the effect of probe synthesis errors on Affymetrix microarrays; furthermore, the algorithms developed in this work provide useful tools for the analysis of cross-hybridization, probe synthesis efficiency, fragmentation, wash stringency, temperature, and salt concentration on microarray intensities.

A MYCN-amplified cell line derived from a long-term event-free survivor among our sixteen established neuroblastoma cell lines

Although more than 110 neuroblastoma (NB) cell lines have been established, there have been neither reports on the rate of success to establish NB cell lines, nor well-documented NB cell lines from long-term-survivors. We attempted to establish NB cell lines from 114 patients. Sixteen NB cell lines were established from 12 patients. The success rates to establish cell lines were 1.4% (1/70) from patients in early stages, 25.0% (11/44) from those in advanced stages, and 10.5% (12/114) from those in all stages respectively. Eleven of these 12 patients eventually died. The surviving patient, who was in stage 4 with MYCN-amplification, has been event-free for 19 years after completing therapy. The serum MYCN DNA level in patient TK was very high before therapy, decreased after chemotherapy, and has remained at the normal levels until now. The gene expression profiling of the primary tumor and the K-N-TK cell line was analyzed with an NB-specific cDNA microarray, and indicated that the probability of 5-year survival was extremely low. Microarray-based comparative genomic hybridization (CGH) analysis indicated that genomic aberration profiles of the cell line were uncommon, with MYCN amplification, 17q gain and 11q loss. A unique KP-N-TK cell line, established from an event-free survivor, will be a useful tool for investigating how a patient can survive a tumor with an extremely poor prognosis.

Exon-level expression analyses identify MYCN and NTRK1 as major determinants of alternative exon usage and robustly predict primary neuroblastoma outcome

Background:

Using mRNA expression-derived signatures as predictors of individual patient outcome has been a goal ever since the introduction of microarrays. Here, we addressed whether analyses of tumour mRNA at the exon level can improve on the predictive power and classification accuracy of gene-based expression profiles using neuroblastoma as a model.

Methods:

In a patient cohort comprising 113 primary neuroblastoma specimens expression profiling using exon-level analyses was performed to define predictive signatures using various machine-learning techniques. Alternative transcript use was calculated from relative exon expression. Validation of alternative transcripts was achieved using qPCR- and cell-based approaches.

Results:

Both predictors derived from the gene or the exon levels resulted in prediction accuracies >80% for both event-free and overall survival and proved as independent prognostic markers in multivariate analyses. Alternative transcript use was most prominently linked to the amplification status of the MYCN oncogene, expression of the TrkA/NTRK1 neurotrophin receptor and survival.

Conclusion:

As exon level-based prediction yields comparable, but not significantly better, prediction accuracy than gene expression-based predictors, gene-based assays seem to be sufficiently precise for predicting outcome of neuroblastoma patients. However, exon-level analyses provide added knowledge by identifying alternative transcript use, which should deepen the understanding of neuroblastoma biology.

Integrative analysis of neuroblastoma and pheochromocytoma genomics data

Background

Pheochromocytoma and neuroblastoma are the most common neural crest-derived tumors in adults and children, respectively. We have performed a large-scale in silico analysis of altogether 1784 neuroblastoma and 531 pheochromocytoma samples to establish similarities and differences using analysis of mRNA and microRNA expression, chromosome aberrations and a novel bioinformatics analysis based on cooperative game theory.

Methods

Datasets obtained from Gene Expression Omnibus and ArrayExpress have been subjected to a complex bioinformatics analysis using GeneSpring, Gene Set Enrichment Analysis, Ingenuity Pathway Analysis and own software.

Results

Comparison of neuroblastoma and pheochromocytoma with other tumors revealed the overexpression of genes involved in development of noradrenergic cells. Among these, the significance of paired-like homeobox 2b in pheochromocytoma has not been reported previously. The analysis of similar expression patterns in neuroblastoma and pheochromocytoma revealed the same anti-apoptotic strategies in these tumors. Cancer regulation by stathmin turned out to be the major difference between pheochromocytoma and neuroblastoma. Underexpression of genes involved in neuronal cell-cell interactions was observed in unfavorable neuroblastoma. By the comparison of hypoxia- and Ras-associated pheochromocytoma, we have found that enhanced insulin like growth factor 1 signaling may be responsible for the activation of Src homology 2 domain containing transforming protein 1, the main co-factor of RET. Hypoxia induced factor 1α and vascular endothelial growth factor signaling included the most prominent gene expression changes between von Hippel-Lindau- and multiple endocrine neoplasia type 2A-associated pheochromocytoma.

Conclusions

These pathways include previously undescribed pathomechanisms of neuroblastoma and pheochromocytoma and associated gene products may serve as diagnostic markers and therapeutic targets.

Functional MYCN signature predicts outcome of neuroblastoma irrespective of MYCN amplification

Neuroblastoma is a pediatric tumor of the sympathetic nervous system. MYCN (V-myc myelocytomatosis viral-related oncogene, neuroblastoma derived [avian]) is amplified in 20% of neuroblastomas, and these tumors carry a poor prognosis. However, tumors without MYCN amplification also may have a poor outcome. Here, we identified downstream targets of MYCN by shRNA-mediated silencing MYCN in neuroblastoma cells. From these targets, 157 genes showed an expression profile correlating with MYCN mRNA levels in NB88, a series of 88 neuroblastoma tumors, and therefore represent in vivo relevant MYCN pathway genes. This 157-gene signature identified very poor prognosis tumors in NB88 and independent neuroblastoma cohorts and was more powerful than MYCN amplification or MYCN expression alone. Remarkably, this signature also identified poor outcome of a group of tumors without MYCN amplification. Most of these tumors have low MYCN mRNA levels but high nuclear MYCN protein levels, suggesting stabilization of MYCN at the protein level. One tumor has an MYC amplification and high MYC expression. Chip-on-chip analyses showed that most genes in this signature are directly regulated by MYCN. MYCN induces genes functioning in cell cycle and DNA repair while repressing neuronal differentiation genes. The functional MYCN-157 signature recognizes classical neuroblastoma with MYCN amplification, as well as a newly identified group marked by MYCN protein stabilization.

Krüppel-like factor 4 (KLF4) suppresses neuroblastoma cell growth and determines non-tumorigenic lineage differentiation

Neuroblastoma (NB) is an embryonal tumor and possesses a unique propensity to exhibit either a spontaneous regression or an unrestrained growth. However, the underlying mechanism for this paradoxical clinical outcome remains largely unclear. Quantitative RT-PCR analysis on 102 primary NB tumors revealed that lower Krüppel-like factor 4 (KLF4) expression is frequently found in the unfavorable NB (Mann-Whitney test, P=0.027). In particular with the high-risk factors such as age of patient >1 year, MYCN amplification and low TRKA expression, the decreased expression of KLF4 was significantly associated with an unfavorable NB outcome. Despite knockdown of KLF4 alone is not sufficient to increase tumorigenicity of NB cells in vivo, stable expression of KLF4 short hairpin RNA in Be(2)-C cells significantly promoted growth of NB cells and inhibited cell differentiation toward fibromuscular lineage. In concordant with these observations, overexpression of KLF4 in SH-SY-5Y cells profoundly suppressed cell proliferation by direct upregulation of cell-cycle inhibitor protein p21(WAF1/CIP1), and knocking down p21(WAF1/CIP1) could partially rescue the suppressive effect of KLF4. Importantly, KLF4 overexpressing cells have lost their neuroblastic phenotypes, they were epithelial-like, strongly substrate-adherent, expressing smooth muscle marker and became non-tumorigenic, suggesting that KLF4 expression is crucial for lineage determination of NB cells, probably, favoring spontaneous tumor regression. Subsequent global gene expression profiling further revealed that transforming growth factor beta (TGFβ) and cell-cycle pathways are highly dysregulated upon KLF4 overexpression, and myogenic modulators, MEF2A and MYOD1 were found significantly upregulated. Taken together, we have demonstrated that KLF4 contributes to the favorable disease outcome by directly mediating the growth and lineage determination of NB cells.

Identification and characterisation of STMN4 and ROBO2 gene involvement in neuroblastoma cell differentiation

To better understand neuroblastoma differentiation, we used microarray analysis to identify common gene expression changes from three differentiation models. This revealed STMN4 and ROBO2 to be consistently up-regulated in differentiated neuroblastoma cells induced by chromosome 1 transfer, MYCN knockdown, and 9-cis retinoic acid (9cRA). Furthermore, stable expression of transfected STMN4 or ROBO2 induced differentiation in IMR-32 cells. STMN4 and ROBO2 expression also increased in other 9cRA-induced differentiated neuroblastoma cell lines. Of clinical importance is that neuroblastoma patients with higher tumour mRNA expression of STMN4 and ROBO2 had better progression-free survival. This study highlights the importance of STMN4 and ROBO2 during neuroblastoma differentiation.

New aspects of neuroblastoma treatment: ASPHO 2011 symposium review

Neuroblastoma is the most common extracranial solid tumor of childhood, and the outcomes for children with high-risk and relapsed disease remain poor. However, new international strategies for risk stratification and for treatment based on novel tumor targets and including immunotherapy are being employed in attempts to improve the outcomes of children with neuroblastoma. A new international neuroblastoma risk classification system has been developed which is being incorporated into cooperative group clinical trials in North America, Japan, and Europe, resulting in standardized approaches for the initial evaluation and treatment stratification of neuroblastoma patients. Furthermore, novel treatment regimens are being developed based on improved understanding of neuroblastoma biology and on the recruitment of the immune system to specifically target neuroblastoma tumors. These approaches will lead to new therapeutic strategies that likely will improve the outcomes for children with neuroblastoma worldwide.

Design of a multi-signature ensemble classifier predicting neuroblastoma patients' outcome

Background

Neuroblastoma is the most common pediatric solid tumor of the sympathetic nervous system. Development of improved predictive tools for patients stratification is a crucial requirement for neuroblastoma therapy. Several studies utilized gene expression-based signatures to stratify neuroblastoma patients and demonstrated a clear advantage of adding genomic analysis to risk assessment. There is little overlapping among signatures and merging their prognostic potential would be advantageous. Here, we describe a new strategy to merge published neuroblastoma related gene signatures into a single, highly accurate, Multi-Signature Ensemble (MuSE)-classifier of neuroblastoma (NB) patients outcome.

Methods

Gene expression profiles of 182 neuroblastoma tumors, subdivided into three independent datasets, were used in the various phases of development and validation of neuroblastoma NB-MuSE-classifier. Thirty three signatures were evaluated for patients' outcome prediction using 22 classification algorithms each and generating 726 classifiers and prediction results. The best-performing algorithm for each signature was selected, validated on an independent dataset and the 20 signatures performing with an accuracy > = 80% were retained.

Results

We combined the 20 predictions associated to the corresponding signatures through the selection of the best performing algorithm into a single outcome predictor. The best performance was obtained by the Decision Table algorithm that produced the NB-MuSE-classifier characterized by an external validation accuracy of 94%. Kaplan-Meier curves and log-rank test demonstrated that patients with good and poor outcome prediction by the NB-MuSE-classifier have a significantly different survival (p < 0.0001). Survival curves constructed on subgroups of patients divided on the bases of known prognostic marker suggested an excellent stratification of localized and stage 4s tumors but more data are needed to prove this point.

Conclusions

The NB-MuSE-classifier is based on an ensemble approach that merges twenty heterogeneous, neuroblastoma-related gene signatures to blend their discriminating power, rather than numeric values, into a single, highly accurate patients' outcome predictor. The novelty of our approach derives from the way to integrate the gene expression signatures, by optimally associating them with a single paradigm ultimately integrated into a single classifier. This model can be exported to other types of cancer and to diseases for which dedicated databases exist.