For decades, it has been known that prostate specific antigen ‒ or PSA ‒ tests are a flawed way to diagnose prostate cancer.

Many men have a high PSA without having cancer. Others have low PSA that might lead to aggressive tumors being missed in screenings. This has led to overtreatment of men who didn't need biopsies or whose cancers would never have become dangerous and undertreatment of those whose tumors were missed.

Now, researchers from Stanford University have developed a method of genetically adjusting PSA levels to screen out the "noise" and better identify those truly at risk.

"There was this really great opportunity to take a test that has flaws but has a lot of very unique advantages and try to improve it," said study co-author Linda Kachuri, a cancer epidemiologist and assistant professor.

Right now, PSA thresholds may incorporate age, family history and race before a man is believed to need a biopsy to confirm or dismiss prostate cancer. "Adding genetics isn't that big of a next step," said senior author John Witte, a professor and genetic epidemiologist.

In the study of nearly 100,000 men published in the journal Nature Medicine, the team found 128 locations across the genome associated with PSA. They developed a risk score that indicates whether each man's combination of genetic variants in those locations is likely to give him a naturally high, low or normal PSA.

Added to other factors already used to determine appropriate PSA levels, the genetically adjusted PSA better predicted aggressive prostate cancers. Among the study group of men with European ancestry, it would have prevented up to 31% of unnecessary biopsies, though it also would have waved off biopsies in 12% of men who had cancer and would have benefited from a biopsy, the study showed.

What is a PSA test? What does it look for?

"The PSA is really the best blood cancer marker in all of medicine, but it's not like a pregnancy test where everybody who's positive is pregnant," said Dr. William Catalona, a prostate cancer surgeon and professor of urology at Northwestern University’s Feinberg School of Medicine.

Instead, he said, "just like some men are tall and some are short, some are high PSA secreters and some are low PSA producers."

Prostate-specific antigen is a protein produced by normal as well as cancerous cells in the prostate gland. The PSA test measures the level of this protein in the blood.

Protein levels are often high among men with prostate cancer but can also be elevated because of inflammation or enlargement of the prostate.

Other research has shown that fewer than one-third of men with elevated PSA levels actually had prostate cancer and 15% of men with normal PSA levels did have prostate cancer.

Someone who happens to have a naturally high PSA level is more likely to receive a biopsy and "go down the pathway for overdiagnosis and overtreatment," and someone with a naturally low level might not pass the threshold for a biopsy until his cancer is more advanced and therefore harder to treat, said Catalona, who was not involved in the new research but has collaborated with Witte.

What the genetic score showed

The genetic analysis did not look at a man's risk for prostate cancer, just his likelihood of having high or low PSA, Witte said.

"The focus of this is not to understand the genetics of prostate cancer but the genetics of PSA," he said. "We're trying to remove the noise in the PSA test that's not due to cancer."

Currently, the genetically adjusted PSA was developed based on men of European ancestry, but the team is working on updating its algorithm with a larger, more diverse group of men.

Witte said he hopes to develop an integrated predictive model that includes a man's age, family history of prostate cancer, genetic predisposition to prostate cancer, and genetically adjusted PSA using this score.

"The idea is to develop an integrative risk score model that includes all these different factors," he said.

The hardest part is yet to come, Witte said: "putting the model into practice and communicating it back to physicians and patients."

Kachuri said she expects that similar risk scores eventually will be developed for a wide range of cancer types.

"Ultimately this will be part of a much larger system predicting risk of cancer."

Contact Karen Weintraub at kweintraub@usatoday.com.

Health and patient safety coverage at USA TODAY is made possible in part by a grant from the Masimo Foundation for Ethics, Innovation and Competition in Healthcare. The Masimo Foundation does not provide editorial input.