Predictive gene signatures and transcriptional regulatory networks of spontaneous regression in neuroblastoma

Coordinator:    PD Dr. Matthias Fischer
Institution: Klinik und Poliklinik für Allgemeine Kinderheilkunde der Universität zu Köln Abteilung Kinderonkologie und -hämatologie
Neuroblastoma is a tumor arising from cells of the developing sympathetic nervous system. It is one of the most frequent solid tumors in childhood with about 130 newly diagnosed cases in Germany per year. The natural courses of neuroblastoma are remarkably heterogeneous: Relentless progression occurs in about half of the affected children, which frequently results in death of disease despite the most intensive multimodal therapies. In contrast, spontaneous regression of the disease without any cytotoxic treatment is observed in other children. The molecular basis of the different natural courses is still not known. In addition, discrimination of the divergent subtypes at the time of diagnosis has remained a challenge to physicians. However, an accurate prediction of the course of the disease is an indispensable prerequisite for the selection of the most adequate treatment of each individual patient.

To improve risk estimation and treatment stratification of neuroblastoma patients, we generated a gene expression-based classifier based on microarray experiments of primary neuroblastoma samples. We observed accurate outcome prediction by the classifier in both retrospective and prospective studies (Oberthuer et al., 2010). In the current low and intermediate patient risk groups, the classifier was the only independent prognostic marker in multivariate analyses. Based on these results, we have established a novel treatment stratification system for low and intermediate neuroblastoma patients, in which therapy intensity is tailored according to the molecular properties of the individual tumor. We are now planning to implement the prognostic classifier in the upcoming German neuroblastoma trial as a therapy stratifying marker.

In addition, we aimed at establishing a predictive gene signature that may identify ultra-high-risk patients in whom current standard therapies will fail. The identification of an ultra-high-risk group may allow for the development of alternative treatment strategies for patients with a poor chance of cure. In an international collaboration project, we defined a classifier that accurately predicts the outcome of patients with metastasized neuroblastoma (Asgharzadeh et al., 2012).

To elucidate the mechanisms of spontaneous tumor regression and differentiation in neuroblastoma, we investigated the functional role of two transcription factors, both of which have been implicated in early neuronal developmental processes. We found that expression of these two genes was associated with exceptionally favorable patient outcome. Inducible re-expression of the transcription factor Hox-C9 in neuroblastoma cells abrogated tumor growth almost completely in vitro and in vivo by activating the intrinsic apoptotic pathway, pointing towards a role of Hox-C9 in spontaneous regression (Kocak et al., 2013). Inducible re-expression of the second transcription factor led to neuronal differentiation and senescence in neuroblastoma cells, while lentiviral knock-down strongly impaired retinoic-acid induced neuronal differentiation, suggesting that this factor may be involved in mediating neuronal differentiation. Notably, high expression of this gene in primary neuroblastoma was significantly associated with favourable patient outcome upon retinoic-acid treatment, and may thus be used as a predictive biomarker for differentiation therapies in neuroblastoma.

Fig. 1 (left):
Hierarchical cluster analysis of 174 neuroblastomas using gene expression data of the 144 classifier genes. Lines represent tumors, columns represent genes. Gene expression levels are visualized as log-values ranging from blue (+1.0) to red (-1.0). The column on the right hand side indicates the results of the classifier prediction (green, favorable; red, unfavorable).

Fig. 2 (right):
Kaplan-Meier estimates for event-free survival for 171 patients that were subdivided into (A) low-, (B) intermediate-, and (C) high-risk groups according to the NB2004 stratification system after classification by the 144-gene classifier (F, favorable classifier prediction; UF, unfavorable classifier prediction).
INTRANET (Members login)