Clinical Cancer Research Bridging the Lab and the Clinic in Cancer Medicine Infection and Cancer: Biology, Therapeutics, and Prevention
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fischer, M.
Right arrow Articles by Berthold, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fischer, M.
Right arrow Articles by Berthold, F.
Clinical Cancer Research Vol. 12, 5118-5128, September 1, 2006
© 2006 American Association for Cancer Research

Imaging, Diagnosis, Prognosis

Differential Expression of Neuronal Genes Defines Subtypes of Disseminated Neuroblastoma with Favorable and Unfavorable Outcome

Matthias Fischer1, André Oberthuer1, Benedikt Brors3, Yvonne Kahlert1, Matthias Skowron1, Harald Voth1, Patrick Warnat3, Karen Ernestus1,2, Barbara Hero1 and Frank Berthold1

Authors' Affiliations: 1 Department of Pediatric Oncology and Hematology and Center of Molecular Medicine Cologne, University Children's Hospital; 2 Department of Pathology, University Hospital, Cologne, Germany; and 3 Department of Theoretical Bioinformatics, German Cancer Research Center, Heidelberg, Germany

Requests for reprints: Matthias Fischer, Department of Pediatric Oncology and Hematology, University Children's Hospital, Kerpener Str. 62, 50924 Cologne, Germany. Phone: 49-221-478-6816; Fax: 49-221-478-4689; E-mail: matthias.fischer{at}uk-koeln.de.


   Abstract
Top
 Abstract
Materials and Methods
 Results
 Discussion
 References
 
Purpose: Identification of molecular characteristics of spontaneously regressing stage IVS and progressing stage IV neuroblastoma to improve discrimination of patients with metastatic disease following favorable and unfavorable clinical courses.

Experimental Design: Serial analysis of gene expression profiles were generated from five stage IVS and three stage IV neuroblastoma. Differential expression of candidate genes was evaluated by real-time quantitative reverse transcription-PCR in 76 pretreatment tumor samples (stage IVS n = 27 and stage IV n = 49). Gene expression-based outcome prediction was determined by Prediction Analysis for Microarrays using 38 tumors as a training set and 38 tumors as a test set.

Results: Comparison of serial analysis of gene expression profiles from stage IV and IVS neuroblastoma revealed ~500 differentially expressed transcripts. Genes related to neuronal differentiation were observed more frequently in stage IVS tumors as determined by associating transcripts to Gene Ontology annotations. Forty-one candidate genes were evaluated by quantitative reverse transcription-PCR and 18 were confirmed to be differentially expressed (P ≤ 0.001). Classification of patients according to expression patterns of these 18 genes using Prediction Analysis for Microarrays discriminated two subgroups with significantly differing event-free survival (96 ± 6% versus 40 ± 8% at 3 years; P < 0.0001) and overall survival (100% versus 72 ± 7% at 3 years; P = 0.0003). This classifier was the only independent covariate marker in a multivariate analysis considering the variables stage, age, MYCN amplification, and gene signature.

Conclusions: Spontaneously regressing and progressing metastatic neuroblastoma differ by specific gene expression patterns, indicating distinct levels of neuronal differentiation and allowing for an improved risk estimation of children with disseminated disease.

Neuroblastoma is a malignant embryonal tumor of the sympathetic nervous system accounting for 7% to 8% of childhood cancers. The biological and clinical behavior of the tumor is remarkably variable ranging from spontaneous regression to fatal tumor progression. These contrasting courses of disease may even occur in patients with disseminated neuroblastoma: Whereas most metastatic tumors of children at stage IV are characterized by aggressive growth and many patients succumb to their disease despite intensive treatment, those defined as stage IVS (age ≤12 months and dissemination restricted to liver, bone marrow, and/or skin) regularly show spontaneous regression resulting in an excellent patient's outcome (1, 2). Although survival rates of stage IV and IVS patients differ markedly, the precise delineation of a regressive phenotype remains challenging, and markers, such as age ≤12 months and metastatic pattern, are still a matter of debate (3, 4). However, because treatment strategies of these patients may vary from a "watch and wait" approach to myeloablative megatherapy with autologous stem-cell rescue, precise risk group assignment is critical for therapeutic decisions.

The molecular mechanisms underlying the process of spontaneous regression are still unknown. Whereas some authors postulated that natural immunologic tumor defense mechanisms could account for this phenomenon (5, 6), others suggested that spontaneous involution of the tumor might be a result of developmentally regulated programmed cell death of neuroblasts (1, 7–10). Current models of neuroblastoma tumorigenesis propose that there are at least two distinct subtypes of neuroblastoma differing in their tumor cell biology (9), one with the ability to regress or differentiate spontaneously and the other showing an aggressive phenotype. These two subtypes have been shown to differ from one another by several cytogenetic aberrations, such as MYCN amplification and deletions of the chromosomal regions 1p, 3p, and 11q (del1p, del3p, and del11q), all of which are associated with poor outcome (1, 2), and more recently by distinctive gene expression signatures with strong prognostic effect (11, 12). However, tumors of stage IVS disease were either absent or represented in minor numbers in the latter studies, and efforts of other studies to identify characteristic gene expression patterns of stage IVS tumors in comparison with stage IV neuroblastoma failed thus far (13, 14).

In this study, gene expression patterns of stage IV and IVS tumors were investigated using serial analysis of gene expression (SAGE) and quantitative reverse transcription-PCR (QPCR) to uncover transcripts related to spontaneous regression and progression of disseminated neuroblastoma. Differentially expressed transcripts were associated to Gene Ontology (GO) annotations to identify potential biological processes occurring within the transcriptome of spontaneously regressing metastatic neuroblastoma. Finally, a supervised class prediction analysis was done to evaluate the prognostic effect of the identified candidate genes.


   Materials and Methods
Top
 Abstract
 Materials and Methods
Results
 Discussion
 References
 
Characteristics of patients and tumor samples. A total of 76 pretreatment tumor samples from patients with disseminated neuroblastoma were obtained from the German tumor tissue bank (stage IV n = 49 and stage IVS n = 27), 8 of which were used for gene expression profiling by SAGE (see Table 1 ). Tumor samples were collected from patients registered in the German multicenter neuroblastoma study between 1992 and 2003 and treated according to the German NB90-NB97 trial protocols. Informed consent was obtained before analyses of tumor samples. Median age at diagnosis was 1,029 and 118 days for patients of stage IV and IVS disease, respectively, and median follow-up of patients was 3.3 years.


View this table:
[in this window]
[in a new window]
  		Table 1. Characteristics of patients and tumors analyzed in this study (n = 76)

 
Sample preparation and RNA extraction. Snap-frozen tissue (20-50 mg) was cryosliced into sections and homogenized by the FastPrep FP120 cell disrupter (Qbiogene, Inc., Carlsbad, CA). Tumor content of each sample was assessed by a pathologist (K.E.) and only samples with ≥60% tumor cells were included. Total RNA was extracted from homogenized samples using the TRIzol reagent (Invitrogen, Karlsruhe, Germany) according to the supplier's instructions. Polyadenylate-RNA was purified using the MessageMaker messenger isolation system (Invitrogen). Integrity of total RNA samples was examined by RNA Nano Chip assays on the 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany).

Serial analysis of gene expression. SAGE libraries were constructed from 100 to 200 ng polyadenylate-RNA or 20 µg total RNA using the I-SAGE kit (Invitrogen) following the manufacturer's protocol. Concatemers were cloned into a plasmid vector, and clones containing inserts were automatically sequenced (MWG-Biotech, Ebersberg, Germany). Extraction of SAGE tags from sequence files and exclusion of duplicate ditags and linker tags was done using the SAGE2000 software version 4.12 (http://www.sagenet.org/sage_protocol.htm). Genes corresponding to the tags were identified by comparison with the human UniGene reference database, Build 166 (http://www.ncbi.nlm.nih.gov/SAGE).

Quantitative reverse transcription-PCR. QPCR was done using the SYBR Green I reagent on the ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Weiterstadt, Germany) as described elsewhere (15). In brief, 5 µg total RNA was converted into first-strand cDNA in a volume of 52.5 µL. PCR amplification was done using standard conditions in a volume of 30 µL containing 0.4 µL of 1:10 diluted first-strand cDNA, 26.8 µL of 1x SYBR Green PCR Master Mix (Applied Biosystems), and 1.4 µL of 2.5 µmol/L forward and reverse primer each (Eurogentec, Seraing, Belgium). Primer sequences are available from the authors on request. Serial cDNA dilutions of the neuroblastoma cell line IMR-32 were used for standard curve calculation. Target gene expression levels were normalized to the geometric mean of transcript levels of the control genes SDHA and HPRT1, which have been shown to be consistently expressed in stage IV and IVS neuroblastoma (15), and calibrated to the minimal expression value within the total set of tumors.

Data analysis and statistics. SAGE profiles of stage IV and IVS neuroblastoma were compared using t test statistics, ANOVA, and Fisher's exact test with Bonferroni correction. Calculations were carried out in R (version 2.1.0, The R Foundation for Statistical Computing; http://www.r-project.org). To identify GO categories (http://www.geneontology.org) overrepresented in the tumor subtypes, the UniGene IDs were mapped to the Locus Link/Entrez entries. Associated GO categories were obtained from the Bioconductor (16) library humanLLmappings (version 1.10.0). Only GO categories in the "biological process" section of GO were used. Overrepresentation of GO categories was tested by using the hypergeometric test implemented in the Bioconductor package GOstats (version 1.4.0). Distributions of normalized gene expression levels determined by QPCR in stage IV and IVS tumors were compared using the Mann-Whitney U test. For supervised class prediction analysis, the nearest shrunken centroids method [Prediction Analysis for Microarrays (PAM)] was applied (17) after expression data had been transformed by calculating the log ratio of expression values to the median expression value of each gene. The implementation in the Bioconductor package pamr (version 1.25) was used. To consider all the information from the QPCR data, the PAM 10-fold cross-validation of the training set was done with a fixed threshold of zero for no shrinkage, thereby estimating the classification accuracy of the gene signature in the training set. The threshold of zero was also used for class prediction of patients in the test set. Unsupervised cluster analysis was done using the Genesis Microarray Software Suite version 1.4.0 (http://genome.tugraz.at/Software/GenesisCenter.html). Kaplan-Meier estimates for event-free survival (EFS) and overall survival (OS) were calculated and compared by log-rank test. Recurrence, progression, and death of disease were considered as events. Cox's proportional hazards regression model built on EFS was used for multivariate analysis.


   Results
Top
 Abstract
 Materials and Methods
 Results
Discussion
 References
 
Identification of genes differentially expressed in stage IV and IVS neuroblastoma. Gene expression profiles were generated from five samples of spontaneously regressing stage IVS disease and from three samples of fatal stage IV disease using SAGE (18). A total of 213,235 SAGE tags (211,889 tags after exclusion of linker tags) were extracted from 7,684 sequence files with single profiles varying between 19,601 and 29,407 tags. SAGE data of this study have been deposited in National Center for Biotechnology Information Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) and are accessible through Gene Expression Omnibus Series accession no. GSE4991. SAGE tags that were detected only once in the profiles were excluded from the study, leaving a total of 177,992 tags (19,547 unique tags) for statistical analyses. t test statistics revealed 429 tags differentially represented in stage IV and IVS tumors (P < 0.05), 277 of which were up-regulated and 152 were down-regulated in the latter subtype (Supplementary Table S1). Similarly, 517 and 508 tags with a distinct representation were detected by ANOVA and Fisher's exact test with Bonferroni correction after pooling profiles of each subgroup, respectively (P < 0.05; data not shown). The largest overlap identified by two tests was observed between t test and ANOVA (276 tags), whereas it was 143 tags between ANOVA and Fisher's exact test and 96 tags between t test and Fisher's exact test. A total of 65 tags were identified by all three tests. Because Fisher's exact test disregards information about the variance of tag counts within each subgroup but calculates differences based on averaged counts, and because t test and ANOVA results largely overlapped, we considered t test results for further analysis.

Genes corresponding to tags identified by the t test were associated to GO annotations to examine whether these transcripts can be categorized into functional classes that may provide information about the molecular basis of the distinct phenotypes. Fifty-two and 27 GO categories were overrepresented in stage IVS and IV tumors, respectively (Supplementary Table S2). Nine of the 52 GO categories of stage IVS tumors were related to neuronal differentiation or neuronal functions (e.g., "positive regulation of neuronal differentiation," "dendrite morphogenesis," "neuron migration," "neurotransmitter receptor biosynthesis," and "synaptic vesicle transport"; Supplementary Table S2A). In contrast, only two such categories were found among the GO classes of stage IV neuroblastoma ("vesicle organization" and "biogenesis and regulation of neurotransmitter secretion"; Supplementary Table S2B), which suggests that these tumors differ from stage IVS by gene expression patterns, reflecting distinct levels of neuronal differentiation. A less pronounced difference was observed regarding GO categories related to programmed cell death or growth control, with three of them being overrepresented in stage IVS tumors ("induction of apoptosis by intracellular signals," "cell cycle arrest," and "positive regulation of tumor necrosis factor-{alpha} biosynthesis," Supplementary Table S2A). The GO class "apoptosis" was overrepresented in stage IV tumors; however, subcategories of this class comprise genes with proapoptotic and antiapoptotic functions, which hampers interpretation of this finding.

Although SAGE results strongly suggested that neuroblastoma of stage IV and IVS disease harbor distinct gene expression patterns, the small sample numbers of this analysis might have generated an overestimation of differentially expressed transcripts. We therefore evaluated 41 genes with differential expression according to SAGE that are related to functional categories of neuronal differentiation and/or apoptosis or growth control using QPCR in a cohort of 38 neuroblastoma (stage IV n = 19 and stage IVS n = 19; Supplementary Table S3). In this set, 18 (44%) genes showed significantly distinct transcript abundances between the two subtypes (P < 0.05). Expression levels of these 18 genes were then evaluated in an independent series of 38 tumors (stage IV n = 30 and stage IVS n = 8). Differential expression was confirmed for 16 of 18 genes in this second set (P < 0.05), and a tendency was observed for the two remaining genes CADPS and PRAME (P = 0.10 and 0.14, respectively). In the combined set of 76 samples, expression levels of all 18 genes were significantly different in stage IV and IVS tumors (P ≤ 0.001 for each transcript; Fig. 1 ). Functional characteristics of these 18 genes are summarized in Table 2 . Expression levels of 17 genes were increased in stage IVS tumors, whereas one gene (PRAME) was down-regulated in this subset as reported previously (19). Taken together, SAGE and QPCR analyses showed different gene expression patterns in stage IVS and IV neuroblastoma that indicate distinct levels of neuronal differentiation of the tumor cells.


Figure 1
View larger version (29K):
[in this window]
[in a new window]
  		Fig. 1. Relative expression levels of 18 transcripts in stage IV (4) and IVS (4S) neuroblastoma as determined by QPCR. Transcript abundances of each gene were calibrated to the minimal expression value of the total tumor set. Boxes, median expression values and 25% and 75% quartiles; bars, SD; circles and asterisks, outlying and extreme values, respectively.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Genes differentially expressed in stage IVS and IV neuroblastoma as determined by QPCR

 
Definition of neuroblastoma subtypes according to gene expression signatures. Given the fact that each single gene delineated in Fig. 1 was differentially expressed with strong statistical significance between stage IVS and IV neuroblastoma that usually correspond to regressing and progressing phenotypes, respectively, we asked whether the combined analysis of these transcripts might be able to accurately predict the biological behavior of the tumors and consequently the clinical courses of the patients. A supervised class prediction analysis was therefore done using the PAM algorithm (17) to test the prognostic value of these 18 genes. Classification accuracy was evaluated in the first set of 38 patients by cross-validation. Whereas PAM voting for most children was analogous to staging of disease, three patients with stage IV and two patients with stage IVS tumors were classified into the opposite category (Fig. 2A ). PAM prediction was then used to classify tumors of the second series of patients (n = 38) based on expression levels of the selected 18 genes. Again, three stage IV and two stage IVS tumors were categorized into the opposite subgroup (Fig. 2B). Hierarchical clustering of the patients using these 18 genes visualized comparable gene expression patterns of tumors showing a similar clinical behavior (Fig. 2C). Kaplan-Meier estimates for EFS and OS of patients with a favorable (n = 29) and an unfavorable PAM voting (n = 47) differed with a higher level of significance (3-year EFS 96 ± 6% versus 40 ± 8%; P < 0.0001; 3-year OS 100% versus 72 ± 7%; P = 0.0003; Fig. 3 ) than those of subgroups defined by stage (3-year EFS stage IVS 88 ± 6% versus stage IV 46 ± 8%; P = 0.0027; 3-year OS 96 ± 4% versus 76 ± 7%, respectively; P = 0.0044; Fig. 3A and B).


Figure 2
View larger version (41K):
[in this window]
[in a new window]
  		Fig. 2. Cross-validation probability plot of the first set of tumors (stage IV n = 19 and stage IVS n = 19; A) and prediction probability plot of the second set of tumors (stage IV n = 30 and stage IVS n = 8; B) using PAM. P < 0.5 was counted as voting for the opposite tumor class. As a fixed threshold of zero for the PAM algorithm was set, the cross-validation results can be compared with a supervised tumor class prediction. Blue dots, stage IV patients; orange dots, stage IVS patients. Top, patients' outcome (red, event; green, no event) and PAM voting (red, unfavorable; green, favorable). C, hierarchical cluster analysis of the total set of tumors according to the 18 signature genes (variables: Manhattan distance and average linkage). Red, high expression; blue, low expression. Top, patients' stage (red, stage IV; green, stage IVS), outcome (red, event; green, no event), and PAM voting (red, unfavorable; green, favorable).

 

Figure 3
View larger version (12K):
[in this window]
[in a new window]
  		Fig. 3. Kaplan-Meier curves for EFS and OS of neuroblastoma patients with favorable (F) and unfavorable (UF) gene expression signatures. EFS (A) and OS (B) of patients classified according to favorable or unfavorable PAM voting (gray) in comparison with patients classified according to stage IV or IVS (black). EFS (C) and OS (D) of infants (≤12 months) with favorable and unfavorable gene expression signatures. EFS of stage IV (E) and IVS (F) patients with favorable and unfavorable gene expression signatures.

 
Outcome prediction by PAM was then evaluated in clinical subgroups defined by age at diagnosis and stage of disease. Infants <1 year old with an unfavorable signature (n = 6) displayed a significantly poorer outcome than those with a favorable signature (n = 25; 3-year EFS 42 ± 22% versus 96 ± 4%; P = 0.0008; 3-year OS 60 ± 22% versus 100%; P = 0.001; Fig. 3C and D). Patients with stage IV disease and a favorable gene signature (n = 6) had a significantly better EFS and tended to have a better OS than stage IV patients with an unfavorable signature (n = 43; 3-year EFS 100% versus 40 ± 8%; P = 0.04; Fig. 3E; 3-year OS 100% versus 73 ± 7%; P = 0.1). Examination of cytogenetic aberrations of these six favorable tumors revealed that four of five patients harbored a del11q, whereas only one patient had a del1p and none of them a MYCN amplification. Three children of this subgroup were ~1 year old at diagnosis (321, 359, and 371 days), but the remaining three patients were substantially older (619, 1,468, and 2,003 days). Similarly, infants with stage IVS disease and favorable PAM voting (n = 23) showed a superior EFS (96 ± 4% at 3 years) compared with those with an unfavorable gene signature (n = 4; 3-year EFS 38 ± 29%; P = 0.002; Fig. 3F). The latter subgroup comprised two stage IVS patients that experienced progression into stage IV neuroblastoma, one of them with fatal outcome. One of these two patients (NB87) lacked any of the prognostic cytogenetic aberrations MYCN amplification, del1p, del3p, and del11q, whereas the other one had a del11q (NB91; Table 1). Of the remaining two patients, one (NB72) harbored a MYCN amplification and a del1p and is currently in complete remission after intensive treatment according to the German high-risk protocol, whereas the other patient (NB80) lacked cytogenetic aberrations and is currently in complete remission without cytotoxic treatment. Within the group of stage IVS patients with a favorable signature, one child had relapsed from disease but reached complete remission without chemotherapy thereafter (3-year OS of this group is 100%).

Finally, the prognostic variables gene signature, stage, age, and MYCN status were evaluated in a multivariate Cox regression model built on EFS for the whole set of 76 patients. Outcome prediction by the gene signature turned out to be a superior, independent prognostic marker in this series of patients (P < 0.001; hazard ratio, 21.7; 95% confidence interval, 3-160).


   Discussion
Top
 Abstract
 Materials and Methods
 Results
 Discussion
References
 
Disseminated neuroblastoma exhibits a remarkable heterogeneity in its clinical and biological behavior. The contrasting phenotypes have been suggested to be due to distinct genetic programs of the tumor cells, which are supposed to be mirrored by characteristic differences on the transcriptome level (1, 7–10). Thus, identification of genes differentially expressed in stage IV and IVS tumors may contribute to the elucidation of the molecular mechanisms of spontaneous regression and provide novel prognostic markers that are able to accurately predict outcome of patients suffering from metastatic neuroblastoma.

In the present study, gene expression profiles of untreated neuroblastoma of stage IV and IVS disease that followed characteristic clinical courses were generated by SAGE. MYCN-amplified tumors were excluded from the SAGE analysis, because MYCN amplification is observed in only a minor fraction (~30-35%; refs. 1, 9) of unfavorable neuroblastoma and results in distinctive effects on the tumor's transcriptome on its own (20, 21), which may mask general mRNA alterations of aggressive neuroblastoma. Comparison of the SAGE profiles of stage IV and IVS tumors using various statistical tests revealed ~500 differentially expressed genes. Because tumor numbers analyzed by SAGE were small and thus might have resulted in an overestimation of transcriptomic differences, 41 genes that were supposed to be differentially expressed were selected and expression levels were determined by QPCR in a series of 76 tumors of stage IV and IVS disease. Eighteen genes showed significantly diverging transcript abundances, suggesting that ~50% of the ~500 transcripts identified by various statistical tests are differentially expressed in these two subtypes. This finding is in contrast to results from previous microarray analyses (13, 14) that failed to reliably discover discriminating gene expression patterns of stage IVS and IV neuroblastoma, which may in part result from intrinsic methodologic differences of SAGE and microarrays (22). However, it seems more likely that the discrepant findings are mainly due to the fact that previous studies either analyzed small sample numbers of stage IVS and IV tumors (n = 9 each; ref. 14) or used a microarray that covers only a minor fraction (4,608 cDNAs) of the human transcriptome (13), whereas in the present study a combined approach of genome-wide expression profiling followed by an independent evaluation of single candidate transcripts in a large set of tumors was applied.

Association of transcripts identified by t test statistics to GO annotations revealed a predominance of genes related to neuronal differentiation and neuronal functions in stage IVS compared with stage IV tumors, and differential expression levels of candidate genes were confirmed by QPCR analysis. Thirteen of the 17 transcripts found to be up-regulated in stage IVS neuroblastoma by QPCR have been shown to contribute to specific processes of neuronal differentiation, to be expressed at specific stages of neuronal development, to represent markers of mature neural tissues, or to be involved in synaptic functions or transmitter release (Table 2). These findings may either indicate that processes of neuronal differentiation are ongoing in cells of stage IVS tumors leading to a more differentiated molecular phenotype compared with cells of stage IV tumors or that dedifferentiation occurs in neoplastic cells of the latter subtype going along with their malignant transformation. Alternatively, these two subtypes may originate from precursor cells of distinct developmental stages. Similarly, a less differentiated molecular phenotype of stage IV neuroblastoma has been observed in comparison with favorable tumors of stage I disease recently (10, 23). However, this result might have been expected, because localized neuroblastoma of stage I or II often present a more differentiated histologic phenotype (24, 25), and development into mature ganglioneuroma has been described to be one of the potential clinical courses in these stages of disease (26). In contrast, the majority of both tumor cells of stage IVS and IV disease exhibit a histologically poorly differentiated stage of development (24, 25), which applied for the set of neuroblastoma of the present study as well (Table 1). Moreover, neuroblastoma of stage IVS disease rarely differentiate into benign ganglioneuroma but regress completely in the majority of cases (27, 28). High expression levels of genes that are induced by the differentiating agent retinoic acid in stage IVS tumors (SYN3, MAP7, CNR1, and MEIS1; Table 2) may further support the notion that molecular development has been proceeded in cells of this subtype compared with those of unfavorable metastatic neuroblastoma. In addition, 12 of the genes up-regulated in stage IVS tumors have been suggested to be involved in cell growth control or to represent tumor suppressors, and 5 of these are located at chromosomal regions 3p and 11q that are frequently deleted in unfavorable neuroblastoma (PCBP4, RBM5, ROBO1, IGSF4, and scotin; Table 2). This observation may indicate that several mechanisms of growth control are disrupted in stage IV tumors and may add evidence to the hypothesis that spontaneous regression in neuroblastoma is due to a delayed activation of developmentally regulated programmed cell death (1, 9, 10). It remains to be determined, however, whether down-regulation of the genes described here is causally involved in the malignant phenotype of aggressive neuroblastoma.

The classification accuracy of clinical outcome by this gene signature was examined using the PAM algorithm in a training and test set of 38 tumors each. Analysis of EFS and OS of patients with a favorable and an unfavorable PAM prediction showed that the gene signature reliably distinguished patients with beneficial and adverse outcome (Fig. 3). As expected, classifications by PAM and by stage (stage IV versus IVS) were highly concordant but with some remarkable exceptions. Four stage IVS tumors were classified into the unfavorable subset (Figs. 2 and 3F), two of which were characterized by progression into stage IV disease and another by a MYCN amplification. A second stage IVS tumor with a MYCN copy number of five revealed a favorable gene expression pattern. However, the prognostic influence of a few additional MYCN copies has been reported to be questionable (29, 30); indeed, this patient reached complete remission and is free of events since 3.5 years without any cytotoxic treatment. On the other hand, six children with stage IV neuroblastoma were categorized into the favorable subgroup. Although three of them were substantially older than 1 year at diagnosis and most tumors harbored cytogenetic aberrations of chromosome 11q, this subgroup was characterized by exceptionally benign clinical courses (Fig. 3E). The strong association of favorable outcome and gene expression patterns resembling that of regressing stage IVS tumors indicates that spontaneous involution may also occur in some stage IV tumors and probably even in patients >2 years old. Although this hypothesis has to be verified carefully in prospective studies, confirmation of this finding could result in a substantial reduction of treatment intensity for these patients. One has to consider, however, that all stage IV patients classified into the favorable subgroup by PAM voting had received cytotoxic treatment, which might have influenced the outcome of these children. Finally, the prognostic value of the gene signature was evaluated in a multivariate analysis, including age, stage, and MYCN status. In this set of tumors, the PAM classifier turned out to be a superior, independent predictor of patient's outcome, thereby supporting findings of recent reports, suggesting that gene expression signatures may be highly accurate prognostic markers for neuroblastoma patients (11, 12).

In conclusion, this is the first study that clearly shows that stage IVS and IV neuroblastoma differ by specific gene expression patterns, and classification of differentially expressed transcripts indicate distinct levels of neuronal differentiation of these two subtypes. In addition, we provide strong evidence that gene expression signatures are suitable to distinguish spontaneously regressing and progressing tumors at the time of diagnosis and thus may contribute to an improved risk estimation of patients with disseminated neuroblastoma.


   Acknowledgments
 
We thank Julia Ollenschläger and Caroline Kallwass for technical support.


   Footnotes
 
Grant support: Deutsche Krebshilfe grant 50-2719, Bundesministerium für Bildung und Forschung through the National Genome Research Network 2 grants 01GS0456 and 01GR0450, Competence Network Pediatric Oncology and Hematology, and Fördergesellschaft Kinderkrebs-Neuroblastom-Forschung e.V.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

Received 4/21/06; revised 6/13/06; accepted 6/23/06.


   References
Top
 Abstract
 Materials and Methods
 Results
 Discussion
 References
 

  1. Schwab M, Westermann F, Hero B, Berthold F. Neuroblastoma: biology and molecular and chromosomal pathology. Lancet Oncol 2003;4:472–80.[CrossRef][Medline]
  2. Maris JM. The biologic basis for neuroblastoma heterogeneity and risk stratification. Curr Opin Pediatr 2005;17:7–13.[CrossRef][Medline]
  3. Hero B, Simon T, Horz S, Berthold F. Metastatic neuroblastoma in infancy: what does the pattern of metastases contribute to prognosis? Med Pediatr Oncol 2000;35:683–7.[CrossRef][Medline]
  4. London WB, Castleberry RP, Matthay KK, et al. Evidence for an age cutoff greater than 365 days for neuroblastoma risk group stratification in the Children's Oncology Group. J Clin Oncol 2005;23:6459–65.[Abstract/Free Full Text]
  5. Bolande RP. The spontaneous regression of neuroblastoma. Experimental evidence for a natural host immunity. Pathol Annu 1991;26 Pt 2:187–99.
  6. Ollert MW, David K, Schmitt C, et al. Normal human serum contains a natural IgM antibody cytotoxic for human neuroblastoma cells. Proc Natl Acad Sci U S A 1996;93:4498–503.[Abstract/Free Full Text]
  7. Teitz T, Lahti JM, Kidd VJ. Aggressive childhood neuroblastomas do not express caspase-8: an important component of programmed cell death. J Mol Med 2001;79:428–36.[CrossRef][Medline]
  8. Kitanaka C, Kato K, Ijiri R, et al. Increased Ras expression and caspase-independent neuroblastoma cell death: possible mechanism of spontaneous neuroblastoma regression. J Natl Cancer Inst 2002;94:358–68.[Abstract/Free Full Text]
  9. Brodeur GM. Neuroblastoma: biological insights into a clinical enigma. Nat Rev Cancer 2003;3:203–16.[CrossRef][Medline]
  10. Nakagawara A. Neural crest development and neuroblastoma: the genetic and biological link. Prog Brain Res 2004;146:233–42.[Medline]
  11. Wei JS, Greer BT, Westermann F, et al. Prediction of clinical outcome using gene expression profiling and artificial neural networks for patients with neuroblastoma. Cancer Res 2004;64:6883–91.[Abstract/Free Full Text]
  12. Ohira M, Oba S, Nakamura Y, et al. Expression profiling using a tumor-specific cDNA microarray predicts the prognosis of intermediate risk neuroblastomas. Cancer Cell 2005;7:337–50.[CrossRef][Medline]
  13. Berwanger B, Hartmann O, Bergmann E, et al. Loss of a FYN-regulated differentiation and growth arrest pathway in advanced stage neuroblastoma. Cancer Cell 2002;2:377–86.[CrossRef][Medline]
  14. Schramm A, Schulte JH, Klein-Hitpass L, et al. Prediction of clinical outcome and biological characterization of neuroblastoma by expression profiling. Oncogene 2005;24:7902–12.[CrossRef][Medline]
  15. Fischer M, Skowron M, Berthold F. Reliable transcript quantification by real-time reverse transcriptase-polymerase chain reaction in primary neuroblastoma using normalization to averaged expression levels of the control genes HPRT1 and SDHA. J Mol Diagn 2005;7:89–96.[Abstract/Free Full Text]
  16. Gentleman RC, Carey VJ, Bates DM, et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 2004;5:R80.[CrossRef][Medline]
  17. Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression. Proc Natl Acad Sci U S A 2002;99:6567–72.[Abstract/Free Full Text]
  18. Velculescu VE, Zhang L, Vogelstein B, Kinzler KW. Serial analysis of gene expression. Science 1995;270:484–7.[Abstract/Free Full Text]
  19. Oberthuer A, Hero B, Spitz R, Berthold F, Fischer M. The tumor-associated antigen PRAME is universally expressed in high-stage neuroblastoma and associated with poor outcome. Clin Cancer Res 2004;10:4307–13.[Abstract/Free Full Text]
  20. Alaminos M, Mora J, Cheung NK, et al. Genome-wide analysis of gene expression associated with MYCN in human neuroblastoma. Cancer Res 2003;63:4538–46.[Abstract/Free Full Text]
  21. Boon K, Caron HN, van Asperen R, et al. N-myc enhances the expression of a large set of genes functioning in ribosome biogenesis and protein synthesis. EMBO J 2001;20:1383–93.[CrossRef][Medline]
  22. van Ruissen F, Ruijter JM, Schaaf GJ, et al. Evaluation of the similarity of gene expression data estimated with SAGE and Affymetrix GeneChips. BMC Genomics 2005;6:91.[Medline]
  23. Ohira M, Morohashi A, Inuzuka H, et al. Expression profiling and characterization of 4200 genes cloned from primary neuroblastomas: identification of 305 genes differentially expressed between favorable and unfavorable subsets. Oncogene 2003;22:5525–36.[CrossRef][Medline]
  24. Shimada H, Chatten J, Newton WA, Jr., et al. Histopathologic prognostic factors in neuroblastic tumors: definition of subtypes of ganglioneuroblastoma and an age-linked classification of neuroblastomas. J Natl Cancer Inst 1984;73:405–16.[Medline]
  25. Shimada H, Umehara S, Monobe Y, et al. International neuroblastoma pathology classification for prognostic evaluation of patients with peripheral neuroblastic tumors: a report from the Children's Cancer Group. Cancer 2001;92:2451–61.[CrossRef][Medline]
  26. Iwanaka T, Yamamoto K, Ogawa Y, et al. Maturation of mass-screened localized adrenal neuroblastoma. J Pediatr Surg 2001;36:1633–6.[CrossRef][Medline]
  27. Haas D, Ablin AR, Miller C, Zoger S, Matthay KK. Complete pathologic maturation and regression of stage IVS neuroblastoma without treatment. Cancer 1988;62:818–25.[CrossRef][Medline]
  28. Reynolds CP. Ras and Seppuku in neuroblastoma. J Natl Cancer Inst 2002;94:319–21.[Free Full Text]
  29. Spitz R, Hero B, Skowron M, Ernestus K, Berthold F. MYCN-status in neuroblastoma: characteristics of tumours showing amplification, gain, and non-amplification. Eur J Cancer 2004;40:2753–9.[CrossRef][Medline]
  30. Seeger RC, Brodeur GM, Sather H, et al. Association of multiple copies of the N-myc oncogene with rapid progression of neuroblastomas. N Engl J Med 1985;313:1111–6.[Abstract]
  31. Shibata H, Huynh DP, Pulst SM. A novel protein with RNA-binding motifs interacts with ataxin-2. Hum Mol Genet 2000;9:1303–13.[Abstract/Free Full Text]
  32. Wiedemeyer R, Westermann F, Wittke I, Nowock J, Schwab M. Ataxin-2 promotes apoptosis of human neuroblastoma cells. Oncogene 2003;22:401–11.[CrossRef][Medline]
  33. Drago J, McColl CD, Horne MK, Finkelstein DI, Ross SA. Neuronal nicotinic receptors: insights gained from gene knockout and knockin mutant mice. Cell Mol Life Sci 2003;60:1267–80.[CrossRef][Medline]
  34. Falk L, Nordberg A, Seiger A, Kjaeldgaard A, Hellstrom-Lindahl E. The {alpha}7 nicotinic receptors in human fetal brain and spinal cord. J Neurochem 2002;80:457–65.[CrossRef][Medline]
  35. Berg DK, Conroy WG. Nicotinic {alpha}7 receptors: synaptic options and downstream signaling in neurons. J Neurobiol 2002;53:512–23.[CrossRef][Medline]
  36. Sharma G, Vijayaraghavan S. Nicotinic receptor signaling in nonexcitable cells. J Neurobiol 2002;53:524–34.[CrossRef][Medline]
  37. Begbie J, Doherty P, Graham A. Cannabinoid receptor, CB1, expression follows neuronal differentiation in the early chick embryo. J Anat 2004;205:213–8.[CrossRef][Medline]
  38. Mato S, Del Olmo E, Pazos A. Ontogenetic development of cannabinoid receptor expression and signal transduction functionality in the human brain. Eur J Neurosci 2003;17:1747–54.[CrossRef][Medline]
  39. Freemantle SJ, Kerley JS, Olsen SL, Gross RH, Spinella MJ. Developmentally-related candidate retinoic acid target genes regulated early during neuronal differentiation of human embryonal carcinoma. Oncogene 2002;21:2880–9.[CrossRef][Medline]
  40. Guzman M, Sanchez C, Galve-Roperh I. Cannabinoids and cell fate. Pharmacol Ther 2002;95:175–84.[CrossRef][Medline]
  41. Jones S, Howl J. Cannabinoid receptor systems: therapeutic targets for tumour intervention. Expert Opin Ther Targets 2003;7:749–58.[CrossRef][Medline]
  42. Bernier G, Brown A, Dalpe G, De Repentigny Y, Mathieu M, Kothary R. Dystonin expression in the developing nervous system predominates in the neurons that degenerate in dystonia musculorum mutant mice. Mol Cell Neurosci 1995;6:509–20.[CrossRef][Medline]
  43. Dowling J, Yang Y, Wollmann R, Reichardt LF, Fuchs E. Developmental expression of BPAG1-n: insights into the spastic ataxia and gross neurologic degeneration in dystonia musculorum mice. Dev Biol 1997;187:131–42.[CrossRef][Medline]
  44. Korkola JE, DeVries S, Fridlyand J, et al. Differentiation of lobular versus ductal breast carcinomas by expression microarray analysis. Cancer Res 2003;63:7167–75.[Abstract/Free Full Text]
  45. Vanaja DK, Cheville JC, Iturria SJ, Young CY. Transcriptional silencing of zinc finger protein 185 identified by expression profiling is associated with prostate cancer progression. Cancer Res 2003;63:3877–82.[Abstract/Free Full Text]
  46. Biederer T, Sara Y, Mozhayeva M, et al. SynCAM, a synaptic adhesion molecule that drives synapse assembly. Science 2002;297:1525–31.[Abstract/Free Full Text]
  47. Mao X, Seidlitz E, Truant R, Hitt M, Ghosh HP. Re-expression of TSLC1 in a non-small-cell lung cancer cell line induces apoptosis and inhibits tumor growth. Oncogene 2004;23:5632–42.[CrossRef][Medline]
  48. Kuramochi M, Fukuhara H, Nobukuni T, et al. TSLC1 is a tumor-suppressor gene in human non-small-cell lung cancer. Nat Genet 2001;27:427–30.[CrossRef][Medline]
  49. Maeda R, Mood K, Jones TL, Aruga J, Buchberg AM, Daar IO. Xmeis1, a protooncogene involved in specifying neural crest cell fate in Xenopus embryos. Oncogene 2001;20:1329–42.[CrossRef][Medline]
  50. Zhuang Y, Faria TN, Chambon P, Gudas LJ. Identification and characterization of retinoic acid receptor ß2 target genes in F9 teratocarcinoma cells. Mol Cancer Res 2003;1:619–30.[Abstract/Free Full Text]
  51. Wermuth PJ, Buchberg AM. Meis1-mediated apoptosis is caspase dependent and can be suppressed by coexpression of HoxA9 in murine and human cell lines. Blood 2005;105:1222–30.[Abstract/Free Full Text]
  52. Marillat V, Cases O, Nguyen-Ba-Charvet KT, Tessier-Lavigne M, Sotelo C, Chedotal A. Spatiotemporal expression patterns of slit and robo genes in the rat brain. J Comp Neurol 2002;442:130–55.[CrossRef][Medline]
  53. Sundaresan V, Roberts I, Bateman A, et al. The DUTT1 gene, a novel NCAM family member is expressed in developing murine neural tissues and has an unusually broad pattern of expression. Mol Cell Neurosci 1998;11:29–35.[CrossRef][Medline]
  54. Wong K, Park HT, Wu JY, Rao Y. Slit proteins: molecular guidance cues for cells ranging from neurons to leukocytes. Curr Opin Genet Dev 2002;12:583–91.[CrossRef][Medline]
  55. Xian J, Aitchison A, Bobrow L, et al. Targeted disruption of the 3p12 gene, Dutt1/Robo1, predisposes mice to lung adenocarcinomas and lymphomas with methylation of the gene promoter. Cancer Res 2004;64:6432–7.[Abstract/Free Full Text]
  56. Moser M, Ruschoff J, Buettner R. Comparative analysis of AP-2{alpha} and AP-2ß gene expression during murine embryogenesis. Dev Dyn 1997;208:115–24.[CrossRef][Medline]
  57. Cho YS, Kim MK, Cheadle C, et al. A genomic-scale view of the cAMP response element-enhancer decoy: a tumor target-based genetic tool. Proc Natl Acad Sci U S A 2002;99:15626–31.[Abstract/Free Full Text]
  58. Harada H, Nagai H, Tsuneizumi M, Mikami I, Sugano S, Emi M. Identification of DMC1, a novel gene in the TOC region on 17q25.1 that shows loss of expression in multiple human cancers. J Hum Genet 2001;46:90–5.[CrossRef][Medline]
  59. Zhu J, Chen X. MCG10, a novel p53 target gene that encodes a KH domain RNA-binding protein, is capable of inducing apoptosis and cell cycle arrest in G(2)-M. Mol Cell Biol 2000;20:5602–18.[Abstract/Free Full Text]
  60. Pio R, Zudaire I, Pino I, et al. {alpha}CP-4, encoded by a putative tumor suppressor gene at 3p21, but not its alternative splice variant {alpha}CP-4a, is underexpressed in lung cancer. Cancer Res 2004;64:4171–9.[Abstract/Free Full Text]
  61. Mourtada-Maarabouni M, Sutherland LC, Meredith JM, Williams GT. Simultaneous acceleration of the cell cycle and suppression of apoptosis by splice variant {delta}-6 of the candidate tumour suppressor LUCA-15/RBM5. Genes Cells 2003;8:109–19.[Abstract]
  62. Oh JJ, West AR, Fishbein MC, Slamon DJ. A candidate tumor suppressor gene, H37, from the human lung cancer tumor suppressor locus 3p21.3. Cancer Res 2002;62:3207–13.[Abstract/Free Full Text]
  63. Bourdon JC, Renzing J, Robertson PL, Fernandes KN, Lane DP. Scotin, a novel p53-inducible proapoptotic protein located in the ER and the nuclear membrane. J Cell Biol 2002;158:235–46.[Abstract/Free Full Text]
  64. Elhamdani A, Martin TF, Kowalchyk JA, Artalejo CR. Ca(2+)-dependent activator protein for secretion is critical for the fusion of dense-core vesicles with the membrane in calf adrenal chromaffin cells. J Neurosci 1999;19:7375–83.[Abstract/Free Full Text]
  65. Wassenberg JJ, Martin TF. Role of CAPS in dense-core vesicle exocytosis. Ann N Y Acad Sci 2002;971:201–9.[Abstract/Free Full Text]
  66. Hintsch G, Zurlinden A, Meskenaite V, et al. The calsyntenins—a family of postsynaptic membrane proteins with distinct neuronal expression patterns. Mol Cell Neurosci 2002;21:393–409.[CrossRef][Medline]
  67. Fabre-Jonca N, Allaman JM, Radlgruber G, et al. The distribution of murine 115-kDa epithelial microtubule-associated protein (E-MAP-115) during embryogenesis and in adult organs suggests a role in epithelial polarization and differentiation. Differentiation 1998;63:169–80.[CrossRef][Medline]
  68. Komada M, McLean DJ, Griswold MD, Russell LD, Soriano P. E-MAP-115, encoding a microtubule-associated protein, is a retinoic acid-inducible gene required for spermatogenesis. Genes Dev 2000;14:1332–42.[Abstract/Free Full Text]
  69. Yao PJ, O'Herron TM, Coleman PD. Immunohistochemical characterization of clathrin assembly protein AP180 and synaptophysin in human brain. Neurobiol Aging 2003;24:173–8.[CrossRef][Medline]
  70. Zhou S, Sousa R, Tannery NH, Lafer EM. Characterization of a novel synapse-specific protein. II. cDNA cloning and sequence analysis of the F1-20 protein. J Neurosci 1992;12:2144–55.[Abstract]
  71. Morgan JR, Zhao X, Womack M, Prasad K, Augustine GJ, Lafer EM. A role for the clathrin assembly domain of AP180 in synaptic vesicle endocytosis. J Neurosci 1999;19:10201–12.[Abstract/Free Full Text]
  72. Zhang B, Koh YH, Beckstead RB, Budnik V, Ganetzky B, Bellen HJ. Synaptic vesicle size and number are regulated by a clathrin adaptor protein required for endocytosis. Neuron 1998;21:1465–75.[CrossRef][Medline]
  73. Ferreira A, Kao HT, Feng J, Rapoport M, Greengard P. Synapsin III: developmental expression, subcellular localization, and role in axon formation. J Neurosci 2000;20:3736–44.[Abstract/Free Full Text]
  74. Pieribone VA, Porton B, Rendon B, Feng J, Greengard P, Kao HT. Expression of synapsin III in nerve terminals and neurogenic regions of the adult brain. J Comp Neurol 2002;454:105–14.[CrossRef][Medline]
  75. Feng J, Chi P, Blanpied TA, et al. Regulation of neurotransmitter release by synapsin III. J Neurosci 2002;22:4372–80.[Abstract/Free Full Text]
  76. Leypoldt F, Flajolet M, Methner A. Neuronal differentiation of cultured human NTERA-2cl.D1 cells leads to increased expression of synapsins. Neurosci Lett 2002;324:37–40.[CrossRef][Medline]
  77. Ikeda H, Lethe B, Lehmann F, et al. Characterization of an antigen that is recognized on a melanoma showing partial HLA loss by CTL expressing an NK inhibitory receptor. Immunity 1997;6:199–208.[CrossRef][Medline]



This article has been cited by other articles:


Home page
 JCOHome page
B. Hero, T. Simon, R. Spitz, K. Ernestus, A. K. Gnekow, H.-G. Scheel-Walter, D. Schwabe, F. H. Schilling, G. Benz-Bohm, and F. Berthold
Localized Infant Neuroblastomas Often Show Spontaneous Regression: Results of the Prospective Trials NB95-S and NB97
J. Clin. Oncol., March 20, 2008; 26(9): 1504 - 1510.
[Abstract] [Full Text] [PDF]

Home page
 Mol Cancer ResHome page
H. Voth, A. Oberthuer, T. Simon, Y. Kahlert, F. Berthold, and M. Fischer
Identification of DEIN, a Novel Gene with High Expression Levels in Stage IVS Neuroblastoma
Mol. Cancer Res., December 1, 2007; 5(12): 1276 - 1284.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fischer, M.
Right arrow Articles by Berthold, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fischer, M.
Right arrow Articles by Berthold, F.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online