�����鶹��Ƶ

Biomarker signatures of prostate cancer

Researchers identified biomarkers in urine that may help doctors diagnose prostate cancer in the future.Medioimages/Photodisc/Thinkstock

Prostate cancer is the second most common cancer in men in the United States. This buildup of abnormal cells in a man’s prostate — a gland below the bladder and in front of the rectum — often has no early symptoms and usually grows very slowly.

Treatment choices depend on many factors. More than half of prostate cancers stay within the gland and don’t become life-threatening. But doctors don’t have a way to reliably predict which tumors will progress and which are unlikely to cause problems.

A research team led by Dr. Thomas Kislinger at the University Health Network in Toronto, Canada, launched a search for noninvasive biomarkers to diagnose prostate cancer and predict whether it’s likely to spread. The team built on previous work, in which they identified more than 130 proteins that differed between fluid samples collected from patients with prostate-confined tumors and those with tumors that had spread beyond the gland. The new study was funded in part by NIH’s �����鶹��Ƶ Cancer Institute (NCI). Results appeared online in Nature Communications on June 28, 2016.

The team analyzed urine samples from 50 patients with prostate cancer — 37 with prostate-confined tumors and 13 with tumors that spread — and 24 healthy controls. A targeted protein screen revealed 34 potentially useful biomarkers. Twenty-four of these showed differences between patients with cancer and healthy controls, suggesting these markers could be useful for diagnosis. Fourteen were different between those with prostate-confined tumors and those with tumors that spread, suggesting these markers may be useful for predicting prognosis, or the cancer’s aggressiveness.

The team next analyzed urine samples collected from a second, independent group of 117 healthy controls and 90 patients with prostate cancer (61 with stage T2, prostate-confined cancer and 29 with stage T3, cancer that’s spread to nearby tissues called seminal vesicles). They used computational biology techniques to determine the importance of each biomarker for distinguishing between patient groups.

Using a machine learning technique, the team found a combination of biomarkers that predicted the diagnosis correctly in 70% of cases and the stage with 69% accuracy. These “biomarker signatures” outperformed the predictive accuracy of the prostate-specific antigen (PSA) protein, which is currently used for making early diagnoses of prostate cancer.

“The next step will be further studies with urine samples from 1,000 international patients to validate if the biomarkers identified have broader clinical utilities in prostate cancer,” Kislinger says.

—by Tianna Hicklin, Ph.D.