New research at Flinders aims to reduce the margin for error in the grading of prostate tumours.

Prostate tumours are graded according to their pattern of growth. When a patient is suspected to have prostate cancer, they will undergo a biopsy where a small amount of tissue is extracted from the prostate. The biopsy sample is then reviewed to determine if there are any cancer cells.

The Gleason system is used for grading prostate cancer. The system grades prostate tissues into five histological patterns from the least aggressive (grade 1) to the most aggressive (grade 5). Most cancers have a grade of 3, or above.

As prostate tumours often contain a mix of growth patterns, the grades of the primary (most predominant) and secondary (second-most predominant) patterns are determined. By adding these two numbers together, the Gleason ‘score’ is assigned.

The decision on the appropriate prostate cancer treatment takes into consideration many factors, but the grading of the tumour forms the cornerstone for treatment and management.

Chief Investigator Dr Gobert Lee explained, “Gleason score 7 is a medium-grade cancer. It suggests an intermediate risk for aggressive cancer, or not, and the prognosis and treatment plan can vary largely,” Dr Lee said.

“A Gleason score 7 can mean (3+4) or (4+3). A Gleason 7 (4+3) would probably require surgery and/or radiotherapy. But a Gleason 7 (3+4) is a slower growing cancer and may need only active surveillance.

“Active surveillance eliminates side effects often associated with surgery or radiotherapy such as urinary incontinence and erectile dysfunction, and leads to a better overall quality of life for the patient.”

Currently, tumour grading is performed by pathologists examining tumour cells.

This process can be subjective, challenging and have a large margin of error. A quantitative method could assist in more accurate grading of Gleason 7 (3+4) versus (4+3).

Flinders researchers will adopt a deep learning approach to analyse prostate biopsy images.

Dr Lee’s team will use a computer algorithm to distinguish Gleason 3+4 and 4+3, with the aim of supporting better prediction and clinical decision making in prostate cancer treatment.

“With deep learning, it’s like training a pathologist. We will present thousands of images to the deep learning algorithm and, based on these images, it will learn to distinguish the difference in grades of prostate cancer,” Dr Lee said.

“This approach provides consistent grading and can improve objectivity in tumour grading.”

This is one of nine projects being undertaken by researchers in the new Freemasons Centre for Male Health and Wellbeing at Flinders University. The Centre’s research is supported through funding from the Masonic Charities Trust and Flinders Foundation.

The Freemasons Centre for Male Health and Wellbeing is an SA and NT research alliance involving Flinders University and Flinders Foundation, Masonic Charities Trust, the South Australian Health and Medical Research Institute, Menzies School of Health Research and the University of Adelaide.

Research category: Men and Boys’ Mental Health

Project title: Deep decision-making for prostate tumour Gleason Score 7 – active surveillance or radical treatment?

Chief and Associate Investigators: Dr Gobert Lee (Flinders University) (Pictured), A/Prof Sonja Klebe (Flinders University), Dr Mariusz Bajger (Flinders University)