UniSci - Daily University Science News
Home Search

clear.gif (52 bytes)

New Algorithm For Detecting Prostate Cancer Earlier

A new computerized method for detecting prostate cancer may lead to earlier detection and eliminate unnecessary biopsies. The findings will be presented Thursday at the 52nd Annual Meeting of the American Association for Clinical Chemistry.

In a prospective study of 151 men conducted by Drs. Richard Babaian and Herbert Fritsche, the new technique for detecting prostate cancer has shown promising results and will be investigated further in a multi-site national study funded by the Dade Behring Company, expected to begin before the end of the year.

Herbert A. Fritsche, Ph.D., professor and chief of clinical chemistry at the University of Texas M.D. Anderson Cancer Center in Houston, TX, will present data to show the value of this computerized method on Thursday, July 27.

Typically, men are recommended for biopsy when their PSA (prostate specific antigen) values are greater than 4.0. Recent studies have suggested, however, that cancer is just as prevalent -- on the order of 20 percent -- in men whose PSA levels are in the 2.5-4.0 range as in those in the 4.0-10 range.

As a result, more men will now qualify for prostate biopsies. Biopsy of all men in this category, however, would result in an unacceptable number of false positive cases.

In order to determine who should be biopsied, Dr. Fritsche and his collaborators evaluated a single-valued index (known as the Prostate Cancer Diagnostic Index, or PCDI), calculated by a neural network-derived algorithm (a mathematically-based pattern recognition system) designed to detect prostate cancer, and then compared the PCDI to the free-to-total PSA ratio to identify men for biopsy.

A pre-biopsy blood sample was tested for total and free serum PSA, and the neural network components creatine kinase and prostatic acid phosphatase. This data, and the subject's age, were the data inputs for the neural network calculation.

Prostate cancer was detected in 24.5 percent (37 out of 151) of the men (all of whom had consented to an 11-core multi-site biopsy of the prostate).

The PCDI correctly identified 93 percent of the 37 men, while the free to total PSA ratio correctly identified 67 percent of the cases, at a cut-off value of 15 percent. At 92 percent sensitivity, the specificity of the neural network was 62 percent, which was significantly better than that of the free/total PSA ratio (11 percent).

This assessment of a new neural network-derived algorithm, says Dr. Fritsche, suggests that this test can effectively predict prostate cancer and also help to avoid many unnecessary biopsies.

AACC, founded in 1948, is the world's most prestigious professional association for clinical laboratorians, clinical and molecular pathologists, and others in related fields. Clinical laboratorians are specialists trained in all areas of human laboratory testing, including genetic and infectious diseases, the presence of tumor markers and DNA.

The primary professional commitment of clinical laboratorians is the understanding of these tests and how to use them accurately to detect and monitor the treatment of human disease.

Related website:

AACC website

[Contact: Donna Krupa ]






clear.gif (52 bytes)

Add the UniSci Daily Java News Ticker to Your Site or Desktop.
Click for a demo and more information.



Please direct website technical problems or questions to webmaster@.

Copyright 1995-2001 UniSci. All rights reserved.