Predicting Prostate Futures:

It seems that most science fiction movies involving the imagined future uses of genetics or artificial intelligence lean heavily toward the dystopian. In the noir “Gattaca” mentioned earlier in this issue by Scott Bratman, MD, PhD, genetic predeterminism casts a dreary pall over society. Likewise, films like “The Terminator” and “The Matrix” offer terrifyingly dark visions of artificial intelligence gone wrong in the worst ways.

The original “Blade Runner,” with its evil androids running amok, was set in 2019. And yet here we are, in the year 2025, where at least in the world of prostate cancer, I am personally optimistic that the AI and genomic tools available to guide treatment decisions and predict the future for patients are a good thing, with a lot more of a utopian than dystopian vibe.

We commonly use National Comprehensive Cancer Network (NCCN) risk groups for localized prostate cancer to aid in treatment decisions. However, these groups depend on subjective assessments, such as a pathologist’s interpretation of the Gleason score, a radiologist’s MRI analysis, or the rectal examiner’s feel. Consequently, treatment decisions regarding active surveillance (AS), radiation with or without hormone therapy, or duration of hormone therapy may vary based on the contributing specialists’ expertise, level of alertness at the time, or digital technique. Additionally, contemporary considerations include complex data not accounted for in NCCN risk groups. PSMA scans, germline and somatic genomic profiles, histologic findings such as cribriform patterns, PSA density, and PSA velocity are just some of the extra inputs available.

The past decade has given us several new commercially available biomarkers that improve upon NCCN risk groups in predicting important oncologic outcomes like metastasis after therapy. These tests aid clinicians in deciding between treatment intensification and deintensification. While randomized trials provide population-level guidance, there can still be over- or under-treatment within certain cohorts of patients. The new advanced risk classifiers address this issue by enabling more personalized treatment considerations.¹

Some of the major new tools currently available are listed in the following table:

The NCCN guidelines list these biomarkers as advanced tools demonstrating superior prognostic performance to standard risk stratification methods. GPS and Prolaris have been established as pre-treatment guideline considerations since 2015-16, with Decipher and Artera appearing for pre-treatment consideration in 2018 and 2023. Although the current NCCN guideline does not suggest a preferred test, it may seem confusing that they only highlight those (Decipher and Artera) tested in tissues banked by the RTOG. Although testing biomarkers in trial patients is ideal, the limited access of these tissues to biomarker researchers remains a challenge. High-quality observational studies, real-world data, and robust statistical methodologies have generated strong, reproducible evidence in Prolaris and GPS as well.

As a radiation oncologist in Utah, known for its exceptional snow, I understand the importance of not using wide powder skis on icy groomers or thin groomer skis in deep powder. You need the right tool for the right job. Similarly, selecting the appropriate risk classifier depends on the clinical scenario one would like to address. When weighing which biomarker to use, choose the one whose score report tells you explicitly how outcomes would change between the interventions you are considering.

Before we dive into examples, I first want to acknowledge the passing of a giant in this biomarker space. The original intent of this article was to include additional comments from the renowned Felix Feng, MD, FASTRO. Sadly, Dr. Feng passed away in December. Among his many scientific achievements were foundational studies that led to the development of the ArteraAI Prostate Test, and he was a co-founder of the company that commercialized the assay. His significant contributions led to many advancements in this space, and his legacy will certainly live on in the many trainees who benefited from his tutelage over the years, for which he was honored with the ASTRO Mentorship Award during the 2024 Annual Meeting.

Let’s consider a few representative case examples, and I will offer my own thoughts on how best to select the right test in these settings.

Case 1: A 66-year-old overall healthy male has a PSA of 4.2. A biopsy revealed Gleason pattern 3+4 in one of 12 core biopsies. He is considering active surveillance (AS) or active intervention.
Prolaris is a great option here, as it was developed and tested in surveillance patients for meaningful oncologic endpoints like death from prostate cancer. Among Favorable Intermediate risk patients, about 70% could safely consider AS based on Prolaris testing results. The other tests have thresholds that would suggest that about half could consider AS, but those estimates are not derived from conservatively managed patients and are therefore less certain.

Case 2: The same patient elected active surveillance, and he is now 70 years old and still very healthy. He has a repeat biopsy, and that reveals Gleason 4+3 in one core and 3+4 in three other cores. He desires to be treated now with radiotherapy.
The NCCN guidelines recommend using androgen deprivation therapy (ADT) in men with Unfavorable Intermediate risk disease. However, both Artera and Prolaris have something unique to offer from the other tests in their reporting of how ADT may work. Artera has a predictive model that tells you if using ADT is futile, regardless of the risk of metastasis. In contrast, Prolaris will report the absolute risk reduction of using or omitting ADT, so that a patient can consider the therapeutic ratio specific to their case.

Case 3: A 71-year-old patient with bad BPH and Gleason 3+3 prostate cancer is considering a TURP versus a Radical Prostatectomy.
The GPS test will inform the patient about the probability of discovering Grade group 3 or worse prostate cancer and the presence of extracapsular extension at prostatectomy. If these risks are high, he may want to pursue a radical prostatectomy instead of surveillance and a more conservative BPH procedure.

Case 4: The patient in case 3 had a radical prostatectomy showing Gleason 4+3 and a positive margin. His PSA was undetectable after surgery but is now 0.3 13 months later. He is referred to you for salvage radiation therapy.
Although RTOG 9601 and 0534 demonstrated a benefit of adding ADT to salvage radiation, post hoc subgroup analyses of RTOG 9601 suggested using ADT may not be effective at PSA values <0.7. The Decipher test can be used to determine if ADT is likely to be helpful versus futile in this patient population.

Hollywood knows many people enjoy a good scare with their popcorn, and so it’s no surprise that for every whimsical “The Hitchhiker’s Guide to the Galaxy” we have a chilling “Resident Evil” series. Fortunately, though, the present and future outlook for prostate cancer patients is more “Buckaroo Banzai” than “Inception.” The previous case examples do not by any means cover the full spectrum of indications for the new wave biomarkers, and each must be used judiciously while accounting for the total clinical context. However, taken together, they represent legitimate progress toward refining and personalizing care for prostate cancer patients.