Predicting neuroblastoma outcome

A gene-expression profile study, combined with analysis by artificial neural networks, has been developed to predict outcome of patients with neuroblastoma.

Fifty-six primary neuroblastoma tumour samples taken from 49 patients before treatment were retrospectively analysed by gene-expression profiling using cDNA microarrays that contained over 25,000 genes. The patients were divided into either good (event-free survival for greater than 3 years) or poor (death due to disease) outcome groups. Neural networks — specialized pattern algorithms modelled after the human brain — were then trained to recognize or predict which samples were associated with good or poor prognosis.

The algorithm identified 19 genes, including 2 prognostic markers that were previously reported (MYCN and CD44), that could be used to predict the outcome for 98% of these patients. Several other genes known to be involved in neuronal development were also associated with outcome, revealing new therapeutic targets. Using these predictor genes, the researchers were also able to partition the subset of patients classified as high-risk into good- and poor-outcome groups after therapy, allowing them to distinguish groups of ultra-high-risk patients that are not likely to respond to conventional therapy and therefore require alternative treatment strategies.

Because only 19 genes were associated with outcome, it is possible to develop simple prognostic assays for clinical use. Three of the genes associated with poor outcome encode proteins that are secreted into the blood, so they might also be used as serum prognosis factors. The authors have begun a larger validation study that includes over 300 neuroblastoma tumour samples. ORIGINAL RESEARCH PAPER Wei, J. S. et al. Prediction of clinical outcome using gene expression profiling and artificial neural networks for patients with neuroblastoma. Cancer Res. 64, 6883–6891 (2004)

Test for bladder cancer recurrence underway

A 3-year Phase III study to validate microsatellite DNA analysis as a method for detecting recurrent bladder cancer has been initiated by the National Cancer Institute's Early Detection Research Network (EDRN) at 12 centres across the United States and Canada.

The goal of the study is to associate microsatellite DNA sequences isolated from urine samples of 300 patients diagnosed with bladder cancer with cancer recurrence. DNA will be extracted from cells that are normally present in urine and compared with DNA sequences of unaffected cells, such as lymphocytes, from the same patients. Healthy individuals and those with non-cancerous bladder problems will be used as controls.

Early studies showed that this non-invasive procedure could identify patients with bladder cancer with over 90% accuracy. Bladder cancer has a high recurrance rate, so frequent surveillance is important. Current cytological or cytoscopic tests are often inaccurate, but this assay, if successfully validated, will provide a more sensitive and non-invasive approach. Final results are expected in September 2007, and the EDRN is also developing tests for protein biomarkers in blood serum that can be used to detect early tumours of the prostate and liver. FURTHER INFORMATION The Early Detection Research Network: http://www3.cancer.gov/prevention/cbrg/edrn.