A pathology test that uses artificial intelligence could be used to examine prostate cancer samples and predict which patients are likely to relapse after surgery, a study shows.
The test, called Precise MD post-op, is for patients whose disease has not yet metastasized, or spread to distant regions of the body. It automates a commonly used grading system for prostate cancer prognosis, called the Gleason score, by combining morphological and biomarker data into complex algorithms.
“By refining diagnoses, we can guide patients toward the best treatment option and optimize care,” senior author Carlos Cordon-Cardo, MD, PhD, chair of the Department of Pathology at the Mount Sinai Health System and professor of pathology, genetics and genomic sciences, and oncological sciences at the Icahn School of Medicine, said in a press release.
The study, “Development and validation of a novel automated Gleason grade and molecular profile that define a highly predictive prostate cancer progression algorithm-based test,” was published in the journal Prostate Cancer and Prostatic Diseases.
Surgery is often the selected treatment for patients with prostate cancer whose disease is still confined to the prostate gland. But despite its good prognosis, 25-30% of men will eventually see their disease return.
Identifying which patients are likely to recur is of paramount importance in improving their outcomes, as these patients could be monitored more closely or receive additional therapy, such as radiation or chemotherapy.
While clinical features, like PSA levels, the Gleason score — a measure of aggressiveness based on a tumor’s microscopic appearance — and tumor extension are often predictive of high-risk disease, doctors still require more accurate predictors of tumor recurrence.
Researchers at the Icahn School of Medicine at Mount Sinai in New York City have now developed a new approach to determine a patient’s Gleason score without requiring a pathologist.
The approach uses cutting-edge microscopic technology that examines tissue morphology and several prostate cancer biomarkers — androgen receptor, Ki67, cytokeratin 18, cytokeratin 5/6, and alpha-methylacylCoA racemase. Then, using artificial intelligence, researchers are able to build algorithms that predict a patient’s chances of disease recurrence on a scale from zero to 100.
Using prostate cancer samples from 590 patients who had surgery, the team found that the Precise MD post-op test was better than the traditional Gleason score or PSA levels at predicting which patients were at a high or low risk of progressing eight years after their surgery.
In addition, the tool reclassified 58% of patients deemed at intermediate risk as having a low risk of progression. An additional 42% were reclassified as high risk, giving them the opportunity to potentially receive additional measures to prevent their cancer from returning.
“In sum, this study introduces an innovative platform to assess prostate cancer risk, revealing that patients with high Precise Post-op scores have a higher likelihood of having clinical failure within 8 years,” the researchers wrote. “The Precise Post-op test guided by machine learning competes with Gleason grading utilizing novel image features that combine morphometry with biological attributes that appear to more accurately reflect disease potential.”
“The introduction of machine learning systems in traditional prostate cancer grading represents an important step towards a more objective and biological reflection of personalized risk assignment,” said Howard Soule, PhD, executive vice president and chief scientific officer of the Prostate Cancer Foundation.