Prediction of Clinical Outcome Using Gene Expression Profiling and Artificial Neural Networks for Patients with Neuroblastoma

Abstract

Currently, patients with neuroblastoma are classified into risk groups (e.g., according to the Children's Oncology Group risk-stratification) to guide physicians in the choice of the most appropriate therapy. Despite this careful stratification, the survival rate for patients with high-risk neuroblastoma remains <30%, and it is not possible to predict which of these high-risk patients will survive or succumb to the disease. Therefore, we have performed gene expression profiling using cDNA microarrays containing 42,578 clones and used artificial neural networks to develop an accurate predictor of survival for each individual patient with neuroblastoma. Using principal component analysis we found that neuroblastoma tumors exhibited inherent prognostic specific gene expression profiles. Subsequent artificial neural network-based prognosis prediction using expression levels of all 37,920 good-quality clones achieved 88% accuracy. Moreover, using an artificial neural network-based gene minimization strategy in a separate analysis we identified 19 genes, including 2 prognostic markers reported previously, MYCN and CD44, which correctly predicted outcome for 98% of these patients. In addition, these 19 predictor genes were able to additionally partition Children's Oncology Group-stratified high-risk patients into two subgroups according to their survival status (P ‫؍‬ 0.0005). Our findings provide evidence of a gene expression signature that can predict prognosis independent of currently known risk factors and could assist physicians in the individual management of patients with high-risk neuroblastoma.

Novel risk stratification of patients with neuroblastoma by genomic signature, which is independent of molecular signature

“…Recently advanced cytogenetic analyses revealed that given subsets of neuroblastomas with a favorable prognosis possess the hyperdiploid karyotype of chromosomes (Look et al, 1984;Tomioka et al, 2003) and that the other subsets with an unfavorable prognosis usually possess the diploid or tetraploid karyotype and often have MYCN amplification, gains of chromosome arms 1q, 2p and 17q, as well as allelic losses of chromosome arms 1p, 3p and 11q (Brodeur, 2003;Schwab et al, 2003). We and other investigators have previously reported the high accuracy of geneexpression profiling to predict the prognosis of neuroblastoma (Wei et al, 2004;Ohira et al, 2005). However, the prognostic significance of genomic signatures when using a high-resolution DNA microarray in primary neuroblastomas has never been reported.…”

Prediction of clinical outcome and biological characterization of neuroblastoma by expression profiling

“…Prediction of clinical outcome has only recently come into the focus of microarray studies (Pomeroy et al, 2002;van de Vijver et al, 2002;Ntzani and Ioannidis, 2003;Nutt et al, 2003). Most recently, a data set has been published, in which 19 genes were sufficient to predict outcome in high-risk neuroblastoma (Wei et al, 2004). Since these data were obtained in a limited number of tumor samples (28 training samples and 21 test samples), there is clearly a need for testing this predictor prospectively in a larger study cohort.…”

“…Of additional interest is the chromosomal localization of five predictor genes, which map to regions frequently deleted or amplified in neuroblastoma: NDUFAB1 (16p12), NME1 and NME2 (17q21), TKT (3p14) and HSPCA (14q32). Most recently, a 19-gene predictor for neuroblastoma was identified using artificial neural networks (Wei et al, 2004). Owing to the overall low numbers of neuroblastoma patients being treated each year, both expressionprofiling studies were conducted using relatively limited sample sizes (49 primary neuroblastomas in the study by Wei et al (2004) and 69 primary neuroblastomas in this study).…”

Altered expression of cell cycle genes distinguishes aggressive neuroblastoma

“…The seven-digit numbers on top of the heat map denote the corresponding GO annotation, which are described in Table 2 in the same order Wei and Khan (2002). The reference RNA used in these studies consisted of equal portions of total RNA obtained from the seven human cancer cell lines (Wei et al, 2004). Total RNA was amplified as described by Wang et al (2000); hybridization and washing of the microarrays were performed as described by Hegde et al (2000).…”