Web-based PREDICT Prostate Tool Provides Personalized Cancer Prognosis

A new web-based tool, called PREDICT Prostate, incorporates patients’ clinical and demographic data to provide a personalized prognosis to men with prostate cancer, helping them choose between radical treatment or watchful waiting.

The online tool was launched the same day the results of the study, “Individual prognosis at diagnosis in nonmetastatic prostate cancer: Development and external validation of the PREDICT Prostate multivariable model,” were published in PLOS Medicine.

Prostate cancer is one of the leading causes of morbidity among men. In most cases, the disease manifests itself later in life, progresses slowly, and is usually not life-threatening. However, in some cases, prostate cancer may become aggressive, spread to other tissues and organs, and be fatal.

“Estimating prognosis within these contexts is therefore highly important. Despite this importance, there are no high-quality individualized prognostic models available for clinical counseling and decision-making,” the researchers said.

According to the National Institute for Health and Care Excellence (NICE), current prognostic models are only 60-70% accurate. After showing that the accuracy of these models can be improved to more than 80% by categorizing patients into five risk categories instead of the typical three, researchers from the University of Cambridge are now developing a personalized prediction tool that would be available to patients at no extra cost.

The new web-based prognosis tool, called PREDICT Prostate, was designed to incorporate patients’ clinical and demographic data, including age, prostate-specific antigen (PSA) test results, tissue biopsy results, cancer grade and stage, and the presence of other medical conditions.

According to the study findings, the tool is able to estimate patients’ survival anywhere between 10 and 15 years, with up to 84% accuracy.

“As far as we are aware, this is the first personalized tool to give an overall survival estimate for men following a prostate cancer diagnosis,” David Thurtle, Academic Clinical Fellow in Urology at the University of Cambridge and Addenbrooke’s Hospital (part of Cambridge University Hospital NHS Foundation Trust) and first author of the study, said in a press release.

Besides predicting overall survival, PREDICT Prostate is also capable of incorporating additional information, including whether a patient decides to undergo treatment or simply be monitored, to arrive at a more accurate prediction.

“This is the choice that faces nearly half of all men who are diagnosed with prostate cancer. We hope it will provide a more accurate and objective estimate to help men reach an informed decision in discussion with their consultant,” Thurtle said.

The web-based tool was developed based on long-term survival data from a large U.K. dataset of over 10,000 men who had been diagnosed with non-metastatic prostate cancer. An additional dataset from over 2,500 prostate cancer patients from Singapore was used for external validation.

“We believe this tool could significantly reduce the number of unnecessary — and potentially harmful — treatments that patients receive and save the NHS millions every year,” said Vincent Gnanapragasam, University Lecturer and Honorary Consultant at CUH and corresponding author of the study.

“This isn’t about rationing treatments — it’s about empowering patients and their clinicians to make decisions based on better evidence. In some cases, treatment will be the right option, but in many others, patients will want to weigh up the treatment benefits versus the risks of side effects. It will also show men who do need treatment a realistic estimate of their survival after treatment,” Gnanapragasam said.

Investigators recommended that the web-based tool should be used by patients who are being followed by a physician. They also noted that the tool may be inadequate for patients with aggressive forms of cancer that have already spread at the time of diagnosis.

“Further external validation of the model should be established to explore accuracy and generalizability across other contexts — particularly testing validity amongst non-Caucasians and those detected through screening,” the scientists said.