Artificial Neural Networks Can Predict Clinical Outcomes of Neuroblastoma Patients
Researchers at the National Cancer Institute (NCI), part of the
National Institutes of Health (NIH), have used artificial neural
networks (ANNs) and DNA microarrays to successfully predict the
clinical outcome of patients diagnosed with neuroblastoma (NB).
The ANNs also identified a minimal set of 19 genes whose expression
levels were closely associated with this clinical outcome. Currently,
the Children’s Oncology Group (COG), sponsored by NCI, stratifies
patients with neuroblastoma into high-, intermediate- and low-risk
groups based on several factors. However, while stratification can
guide patient treatment, it is not a predictor of survival. Now,
the predictive power of microarray gene expression analysis coupled
with ANNs could assist physicians in the treatment of individual
patients.
Neural networks are specialized pattern recognition algorithms modeled
after the human brain; they learn by experience. ANNs are often
used in identification programs, such as fingerprint or voice recognition
software. Javed Khan, M.D., and his team at NCI’s Pediatric
Oncology Branch, adapted an ANN algorithm to identify patterns in
NB tumor gene expression. The study, which appears in the Oct. 1,
2004, Cancer Research*, was performed in collaboration with colleagues
from the NCI, Germany and Australia.
First, the researchers performed gene expression analysis using
cDNA microarrays containing over 25,000 genes to create global gene
expression profiles of primary tumors from 49 patients diagnosed
with NB whose clinical outcome was known. The patients were divided
into either good (event-free survival for greater than three years)
or poor (death due to disease) outcome groups. “Setting aside
independent test samples, neural networks were trained to recognize
or predict ‘alive’ or ‘dead’ expression
profiles from the remaining samples,” said Khan. “Then
we determined if we could predict the outcome for the test samples
using these trained ANNs.” They found that the ANNs could
predict the clinical outcome from any individual gene profile with
an accuracy of about 88 percent.
As these gene profiles consisted of over 25,000 genes, the researchers
tried to optimize the profiles and find the minimum number of genes
that could act as a predictor set. The ANNs identified 19 genes
whose expression levels could accurately predict clinical outcome.
When only looking at these 19 genes, ANN prediction accuracy increased
to 95 percent, and performed much better than the current Children’s
Oncology Group (COG) risk stratification. Two of the genes in this
group, MYCN and CD44, have previously been connected to NB prognosis
MYCN amplification is one of the strongest independent factors
of poor prognosis and several of the other genes are known to
be involved in neuronal development.
Using the 19 predictor genes, the ANNs were also able to partition
the subset of patients classified as high-risk into good and poor
outcome groups. “What was most exciting,” said Khan,
“was that we were able to predict which of the high-risk patients
would fail conventional therapy. This has major clinical implication
since we are now able to distinguish a group of ultra-high-risk
patients who will not respond to conventional therapy and therefore
require alternative treatment strategies. We may also be able to
reduce the intensity and thereby reduce the toxicity of treatment
regime to those predicted to survive based on their gene expression
profile.”
“And since we are using 19 genes instead of 25,000,”
Khan added, “we can translate our findings to the clinic because
simple prognostic assays can be developed based on this small number
of genes. In fact, three of the genes found to be over-expressed
in poor outcome tumors encode proteins secreted into the blood,
meaning they could be used as serum prognosis markers in a simple
blood test.” In collaboration with industry, Khan’s
lab is now developing clinical assays based on these 19 genes and
planning to test for the presence of these serum markers in other
patients with NB for the prognostic prediction.
Khan cautions that more validation studies are required. His lab
now has begun a larger validation study using 300 NB tumor samples
from national trials based in the United States (COG) and the United
Kingdom (UKCCSG: United Kingdom Childhood Cancer Study Group).
For more information about cancer, visit the NCI Web site at http://www.cancer.gov
or call NCI's Cancer Information Service at 1-800-4-CANCER (1-800-422-6237).
* Wei JS, Greer BT, Westermann F, Steinberg SM, Son CG, Chen QR,
Whiteford CC, Bilke S, Krasnoselsky AL, Cenacchi N, Catchpoole D,
Berthold F, Schwab M, and Khan J. Prediction of clinical outcome
using gene expression profiling and artificial neural networks for
patients with neuroblastoma. Cancer Research, Oct. 1, 2004; vol.
64(19).
|