Skip to main content
Premium Trial:

Request an Annual Quote

UK Researchers Developing Tool to Predict Prostate Cancer Progression


NEW YORK – A group of researchers led by the University of East Anglia in the UK have developed a urine-derived extracellular vesicle RNA (UEV-RNA) tool – called Prostate Urine Risk (PUR) – that provides diagnostic data about disease status and prognostic information for patients on active surveillance (AS) before tissue biopsy.

The researchers believe that the PUR tool can identify patients who may need treatment up to five years in advance, thereby potentially minimizing the need for repeated checkups for low-risk patients on AS. 

Clinicians typically rank patients' prostate cancer progression stage using different methods such as CAPRA scores or D' Amico stratification, which classifies patients as being low-, intermediate-, or high-risk for prostate-specific antigen failure post-radical therapy. D'Amico stratification uses a patient's Gleason score, PSA level in their bloodstream, and their prostate cancer clinical stage. 

Because prostate cancer is often multifaceted and errors in sampling collection can occur, research has focused on the development of noninvasive methods that predict risk groups and disease progression over time. Academic groups have used methods such as whole genome sequencing and next generation sequencing to detect a range of tumor-associated genomic alterations to predict prostate cancer growth.  

In a study published in BJU International recently, first author and UEA postgraduate researcher Shea Connell and his team collected 535 urine samples from a multicenter Movember cohort of US and UK patients ranging between the ages of 50 and 70. Several samples were provided by 20 of 87 men enrolled on an AS program from 2009 to 2014. 

Movember is an annual event occurring in the fall to highlight men's health issues, including prostate cancer.

Collecting 1,200 grams of urine per sample, the researchers then centrifuged the solution. Afterward, they harvested extracellular vesicles from each sample using microfiltration. The team then extracted the RNA and amplified about 5 to 20 ng as complementary DNA. 

Noting that prostate cancer cell growth biomarkers kallikrein 2 and kallikrein 3 appeared twenty-eightfold greater in the cohort's urine’s extracellular vesicle fraction, the researchers chose to further study the patients’ extracellular vesicles. 

Connell and his team began developing the PUR signatures by randomly dividing the samples in the D' Amico categories into a training set of 358 samples and a test dataset of 177 samples, stratified by their risk category.  

The team selected 167 gene probes — based on controls and previous prostate cancer diagnostic and prostate biomarkers — as a starting point to analyze potential biomarkers. Connell also noted that 30 probes had also been overexpressed in prostate cancer samples. Using a NanoString Prosigna expression analysis , the researchers ran the data through a customized LASSO algorithm to constrain the data. 

Examining the information from both datasets, the team established an optimal PUR model incorporating 36 genetic probes with four PUR predication signatures: normal tissue (PUR-1), low-risk (PUR-2), intermediate-risk (PUR-3), and high-risk (PUR-4).  For each sample, the four-signature model defined how likely the sample contained material within the four categories. 

The group placed urine samples that either had no evidence of cancer, or NEC (92 samples), or localized prostate cancer (443 samples), into the three risk categories: low- (143), intermediate- (208), and high-risk (101).

The researchers saw primary PUR-signatures (PUR-1 to 4) that were significantly linked to the patient’s clinical category (NEC, L, I, and H) in both training and test sets and, therefore,  initially tested the PUR-model’s ability to predict the presence of H or I disease from L or NEC with the initial urine biopsy. 

Connell explained that his team identified a combination of 35 different genes linked to prostate cancer that it could use to produce the PUR risk signatures, including PCA3, TMPRSS2-ERG, and HOXC6. They found that each PUR signature was significantly linked with its corresponding clinical category, successfully predicting "the presence of significant disease" with better performance "than a random predictor."

According to Connell, the assay could predict biopsy outcome and which risk group a patient would likely to be in, based on a liquid biopsy performed before a tissue biopsy. 

Examining the cohort of 87 men who provided longitudinal samples over a period of five years, the researchers used the PUR profiles to investigate the prognostic utility of PUR beyond categorizing D’ Amico Risk. They noted that the profiles of 23 men who progressed within five years of urine sample collection different significantly from the 49 men who did not progress over time. 

According to the study authors, the two groups had a large difference in time to progression: the poor group had a 60 percent progression within five years of urine sample collection, compared to 10 percent in the good prognosis population. 

Because the researchers had collected multiple samples from 20 men in the AS trial, they could examine the stability of urine profiles over time. In urine from patients whose prostates had not progressed to cancerous states, the team noted that the samples maintained stability compared to randomly selected samples from the entire cohort. 

As a result, the team believes that UEV-RNA could provide diagnostic information about aggressive prostate cancers prior to invasive biopsies, as well as prognostic information for "low-risk" men on AS.

One limitation that Connell noted he and his team encountered was that they were not able to test PUR stratification in a broader and independent active surveillance cohort. In addition, the team dealt with changes in clinical standards when monitoring prostate cancer patients over long periods of time. 

"Patients were treated by standards that have changed over time, which might now be considered aggressive," Connell explained. "We'd like to develop a bigger cohort that is completely independent from current models and follow them over time and see what happens." 

Connell and his team envision use of PUR in doctor’s offices, where clinicians could track patients undergoing active surveillance for prostate cancer, and he believes measurements of a patient’s PUR status could help correctly and noninvasively determine his risk over time. 

Connell said his team was also surprised that the PUR signature could predict prostate cancer progression in some patients up to five years before standard clinical methods could. He, therefore, believes the prognostic info could eventually help minimize the use of patient-elected radical intervention in certain AS populations. 

However, he acknowledged that future studies will need to validate the utility of PUR within active surveillance using several longitudinal cohorts of prostate cancer patients. He said that his team is currently looking at a larger validation study to "generate the evidence required to get the tool into the clinic faster." 

However, the study authors highlighted that "the dramatic differences in RNA expression profiles across the spectrum from high-risk cancer to patients with no evidence of cancer, confirmed in a test cohort, can leave no doubt that the presence of a cancer is substantially influencing the RNA transcripts found in urine EVs."  

While Connell's group has filed an IP for the prostate cancer classifier and are awaiting approval, plans to commercialize the assay are unclear. However, Connell believes that his team will most likely work with the university through a diagnostic laboratory and its technology transfer office.