What’s the best way to determine if a high-risk prostate cancer (PCa) patient is likely to have metastasis? Researchers have been busy analyzing genetic variants or other biomarkers for their prognostic value. Since no one factor has emerged as a top predictor, most doctors have relied on up-to date guidelines from the National Comprehensive Cancer Network (NCCN), which are reviewed and revised annually.

NCCN guideless as generally predictive

The NCCN guidelines emphasize risk stratification, and the various risk levels are stratified in a tables and flow charts that include clinical factors (age, PSA, Gleason score, stage, number of high-risk features). Based on the risk level, the guidelines recommend next clinical steps, e.g., further testing, imaging, management strategy, etc.) Thus, doctors have tended to rely on the NCCN factors that determine which patients are high-risk, and what actions are needed. Such patients require a greater number of therapeutic decisions because their disease is more complex. Knowing how likely a person’s PCa might spread has a bearing on which choices are best.

Since the likelihood that an individual’s PCa will metastasize (spread) is linked with a high risk level, it’s understood that a patient who meets NCCN criteria for having high-risk disease is more likely to experience metastasis. But that is a generalization. Is there a way to obtain a more precise probability for any given individual?

The affirmative answer now seems to rest with Artificial Intelligence (AI). In June, 2022 NPJ Digital Medicine published the work of a multi-institutional collaboration among US and Canadian researchers. The team had trained and validated a patient-level, multi-modal artificial intelligence predictive program in high-risk prostate cancer. They named it Artera AI Prostate. Their goal was to accurately predict endpoints correlated with NCCN’s high-risk characteristics.^[i]

Artificial Intelligence prediction is more accurate than NCCN

The team had access to records of 5654 PCa patients with an average follow-up of 10.4 years. The records included images of their PCa slides plus clinical factors for each patient who had at least one high-risk factor. They found that their program had “superior discriminatory performance across all endpoints.” The authors wrote,

This artificial intelligence-based tool improves prognostication over standard tools and allows oncologists to computationally predict the likeliest outcomes of specific patients to determine optimal treatment. Outfitted with digital scanners and internet access, any clinic could offer such capabilities, enabling global access to therapy personalization.

Advantage for metastatic PCa patients

The speed and accuracy of the Artera AI Prostate model means better decision-making for doctors and their patients. There are many treatment choices that use combinations of surgical/radiological interventions, androgen deprivation, chemotherapies and immunotherapies. The modalities and doses must be tailored to the severity of each patient’s disease. Ideally, the aggressiveness of the integrated therapies will be matched to PCa’s likely course.

At Sperling Prostate Center we stay abreast of AI developments as they relate to PCa, and we have been leaders in incorporating AI in our MRI-based services to patients. We congratulate the research team on their work, and the many ways in which it will improve outcomes for those at risk for metastatic disease.

NOTE: This content is solely for purposes of information and does not substitute for diagnostic or medical advice. Talk to your doctor if you are experiencing pelvic pain, or have any other health concerns or questions of a personal medical nature.

References

[i] Esteva A, Feng J, van der Wal D, Huang SC et al. Prostate cancer therapy personalization via multi-modal deep learning on randomized phase III clinical trials. NPJ Digit Med. 2022 Jun 8;5(1):71.