Nomogram aims to enable informed decision-making and personalized treatment
Studies have found that prostate cancer is overdiagnosed in up to 42 percent of cases, prompting men to receive unnecessary treatment that can cause devastating side effects, including impotence and incontinence.
Now, researchers at Fred Hutchinson Cancer Research Center and the University of Washington have developed a personalized tool that can predict the likelihood of prostate cancer overdiagnosis. They announced their findings this week in the online issue of the Journal of the National Cancer Institute.
An overdiagnosed cancer is defined as one that would never cause symptoms or pose a risk to the patient, therefore not require treatment. Treatment of such cancers provides no benefit and can only cause harm.
The researchers created a nomogram, a graphical calculating device, that incorporates a patient's age, prostate-specific antigen (PSA) level and Gleason score - which grades prostate cancer tissue based on how it looks under a microscope - to determine the likelihood that screening-detected prostate cancer has been overdiagnosed.
The goal, said Roman Gulati, the study's lead author, is to provide patients and clinicians with a tool that can help them better determine personalized treatment options.
"Men with screen-detected prostate cancer are making decisions about treatment based on limited information about the chances that their cancer has been overdiagnosed," said Gulati, a statistical research associate in Fred Hutch's Public Health Sciences Division.
"We think this is a useful tool for patients and their providers because it helps to tailor knowledge of the risks and benefits of different treatment choices to their individual situations."
To develop the nomogram, the researchers created a virtual population model representing U.S. men aged 50 to 84 years from 1975 to 2005 and applied existing data on PSA levels, biopsy practices and cancer diagnosis patterns to learn about cancer progression in patients with and without screening. A virtual population was required since it was impossible to know how cancer may have progressed in real patients who were screened and subsequently treated.
The researchers, who also included University of Washington urologist John Gore, M.D., and biostatistician Lurdes Inoue, Ph.D., then overlaid screening and biopsy patterns on the model to determine when the men would have been diagnosed with and without screening, and which would have died of other causes. They came up with a prediction model which estimates that the likelihood of overdiagnosis ranges from 2.9 to 88.1 percent depending on patient age, PSA level and Gleason score at diagnosis.