Friday, April 1, 2011 (Last Updated: 04/04/2011)
FRIDAY, April 1 (HealthDay News) -- A stroma-specific classifier constructed on the basis of potential stroma-specific genes can detect a nearby prostate tumor from genetic expression profiles from tumor-free samples, according to a study published in the April 1 issue of Cancer Research.
Zhenyu Jia, Ph.D., from the University of California in Irvine, and colleagues compared gene expression profiles from 13 biopsies containing stroma near tumor, and 15 biopsies from volunteers without prostate cancer, to identify whether gene expression changes in stroma can detect nearby tumors. Significant gene expression changes in approximately 3,800 samples were filtered to identify 114 candidate stroma-specific genes. On the basis of these, a stroma-specific classifier was constructed, and then tested on 364 independent samples, including 243 tumor bearing samples and 121 nontumor samples (normal biopsies or autopsies, remote stroma, or stroma within a few millimeters of tumor). The accuracy of prediction was compared with classifiers trained with sets of 100 randomly generated genes.
The investigators found that, in tumor-free samples, the stroma-specific classifier had an average accuracy of 97 percent, sensitivity of 98 percent, and specificity of 88 percent in predicting tumor status of patients. In tumor-free samples, classifiers trained with randomly generated genes could not predict tumor status of the patients.
"These results indicate that the prostate cancer microenvironment exhibits reproducible changes useful for categorizing the presence of tumor in patients when a prostate sample is derived from near the tumor but does not contain any recognizable tumor," the authors write.
Several authors disclosed financial relationships with Proveri Inc.
Hematology & Oncology
Copyright © 2011 HealthDay. All rights reserved.