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

Cancer Cell. 2005 Apr;7(4):337-50. doi: 10.1016/j.ccr.2005.03.019.

Abstract

To predict the prognosis of neuroblastoma patients and choose a better therapeutic protocol, we developed a cDNA microarray carrying 5340 genes obtained from primary neuroblastomas and examined 136 tumor samples. We made a probabilistic output statistical classifier that provided a high accuracy in prognosis prediction (89% at 5 years) and a highly reliable method to validate it. Kaplan-Meier analysis indicated that the patients in an intermediate group defined by existing markers are divided by microarray into two further groups with 5 year survivals for 36% and 89% of patients (p < 10(-4)), i.e., with unfavorably and favorably predicted neuroblastomas, respectively. According to these results, we developed a gene subset chip for a clinical tool, for which our classifier exhibited 88% prediction accuracy.

Publication types

  • Research Support, Non-U.S. Gov't
  • Validation Study

MeSH terms

  • Algorithms
  • Artificial Intelligence
  • Gene Expression Profiling / methods*
  • Gene Expression Profiling / statistics & numerical data
  • Humans
  • Models, Statistical
  • Neoplasm Staging
  • Neoplasms / diagnosis
  • Neoplasms / genetics
  • Neuroblastoma / classification
  • Neuroblastoma / diagnosis*
  • Neuroblastoma / genetics
  • Oligonucleotide Array Sequence Analysis / methods*
  • Oligonucleotide Array Sequence Analysis / statistics & numerical data
  • Prognosis
  • Reproducibility of Results
  • Sensitivity and Specificity
  • Survival Analysis