Use of artificial intelligence to determine cancer recurrence

By Kalli Spencer

After a radical prostatectomy there is always a risk that the cancer may recur. Certain factors can increase this risk such as high grade features in the initial biopsy specimen, cancer that may have emerged from the prostate capsule, micrometastases (cancer that cannot be detected by any imaging), technical difficulties in surgery or no explanation at all. in addition to the genetic makeup of cancer in a particular individual that gives it a predisposition to grow and invade. In its quest for survival, cancer develops several mechanisms to evade the immune system and may even counter the effects of treatment. The genetics behind this are still largely unknown. But scientists are studying it and success has been seen in other tumors such as colon and stomach cancer. A team from the University of Wisconsin in the United States has developed a unique diagnostic test that utilizes artificial intelligence to detect genetic disorders and determine which patients’ prostate cancer is more likely to recur and progress and therefore guide earlier treatment options.

Features at the cellular level and at the molecular level (within the cell) in the formation and progression of prostate cancer in each patient are different and complex. This complicates the development of clinically relevant biomarkers (blood tests) and timing and sequencing for drug selection strategies to beat cancer at its own game. Recent studies have explained that there is a complex interaction between cancer cells and cells and the molecules between them, known as the tumor microenvironment. In this environment it is the immune cells that the cancer targets (the rate of which varies between individuals) preventing itself from destroying.

This complex process cannot be visualized by the human eye looking under a microscope on a glass slide (a thin rectangular piece of glass on which small pieces of prostate tissue are located). They need more advanced analytical techniques. Huang et al have created a digital whole slide image (WSI) of prostate tissue stored on a computer. Here comes the complexity: specialized artificial intelligence software is able to extract subvisual features of cancer cells using deep convolution neural networks (DCNNs). The task of predicting outcomes from WSI was particularly challenging due to the large size of these images (approximately 100,000 × 100,000 pixels) and the fact that unusual cell features associated with adverse outcomes, which are also unknown at this time, could be present in any part of the imaged tissue. .

To date, only a few biomarkers that can be used to guide prostate cancer treatment using molecular biology and bioinformatics have been discovered and investigated. These include molecular markers, such as Ki-67, p53, MYC, PTEN, Rb, AR, and ERG. Knowing that a patient’s cancer has these biomarkers can help stratify their cancer risk profile but also determine the likelihood of tumor recurrence. Today these molecular markers have been combined with the traditional Gleeson grading system to create a scoring system that can help determine the patient’s prognosis. DECIPHER, POLARIS, Oncotype Dx are currently being investigated in other clinical trials. Preliminary clinical trial findings suggest that most of these genome-based scores, however, only slightly increase the prognostic accuracy of the current Gleeson Grade assessment. Traditional tissue processing techniques in the laboratory have made identification of biomarkers difficult.

The results of trials looking at cancer treatments have shown that the adoption of new therapies more quickly in the course of prostate cancer has improved patient outcomes. Notable clinical trials such as PROSPER, SPARTAN, and ARAMIS in which drugs such as enzalutamide, apalutamide, or darolutamide, respectively, were added to standard androgen deprivation therapy have shown improvements in several clinical outcomes including overall survival; time to develop metastases; the time it takes for prostate-specific antigen (PSA) to rise in the blood; and time to start chemotherapy.

Moving to the earlier state of prostate cancer, the TITAN trial examining the addition of apalutamide for metastatic castration-sensitive cancers (usually reserved for castration-resistant cancers) demonstrated an advantage to adding this drug. The STAMPEDE trial demonstrated better results with the addition of radiation in patients with low-volume metastatic disease. Therefore, it is imperative to be able to stratify prostate cancer patients at risk of recurrence for targeted therapy in a timely manner to improve their survival.

The authors’ new technology utilizing artificial intelligence to detect specific features in prostate cancer cells will lead to the discovery of new biomarkers that will help identify those at risk of recurrence and improve outcomes.

Reference:

^{American Society of Clinical Oncology}

^{Huang W et al. New artificial intelligence powered method for prediction of early recurrence of prostate cancer after prostatectomy and cancer triggers. Journal of Clinical Oncology. Clinical Cancer Informatics 2022. 6(e2100131)}

---

About the Author

Kalli Spencer

MBCh, FC Urol (SA), MMed (Urol), Dip.Couns (AIPC)

Kalli is an internationally renowned Urological Surgeon, specializing in oncology and robotic surgery. He trained and worked in South Africa, before moving to Australia where he worked at Macquarie University Hospital and Westmead Hospital. His passion for what he does extends beyond the operating room, through public health advocacy, education and public awareness about men’s health, cancer and sexuality.

Kalli has been involved with the Australian Prostate Cancer Foundation for many years, advocating for better cancer care and facilitating community prostate cancer support groups.

/ Public Release. Material from this original organization/author may be timely, edited for clarity, style and length. The views and opinions expressed are those of the author. See more here.

#artificial #intelligence #determine #cancer #recurrence