Balancing prostate cancer treatment efficacy with the metabolic and cardiovascular (CV) risks of hormone therapy (HT) is crucial, particularly in the treatment of localized disease. Clinicians are beginning to use artificial intelligence (AI) models to identify intermediate- and high-risk patients who may or may not benefit from HT, potentially reducing off-target–related morbidity and mortality.

It is important to weigh the metabolic and CV risks of HT for patients with prostate cancer. We developed the Cancer of the Prostate Risk Assessment (CAPRA) score at the University of California, San Francisco, which estimates a patient’s risk of biochemical recurrence. We look at baseline prostate-specific antigen levels, cancer volume, tumor grade, and patient age, and then we come up with a score from 0 to 10, with 10 being a very strong predictor of prostate cancer recurrence and mortality. Interestingly enough, when we developed the CAPRA score, we found that it predicted overall survival rates, not just prostate cancer–specific survival rates. This suggests that there is something about how we treat men with advanced disease that may contribute to off-target–related mortality.

Much work has been done to better understand the metabolic and CV risks of HT for patients with prostate cancer. Such side effects include CV effects, weight gain, loss of muscle mass, bone marrow density changes, and cognitive issues. These are important issues to patients. Knowing this, we should strive to limit the use of HT to only those patients who will truly benefit from it.

HT is a mainstay for men with very advanced metastatic disease. So, the question is: What about patients with localized disease? We see a lot of men with intermediate- and high-risk nonmetastatic disease who undergo radiation therapy (RT) often combined with short- (4-6 months) and long-term (12-36 months) HT. However, we have given some pause to this approach because some patients may not benefit from it. In addition, some men do not recover normal testosterone levels after being on ADT. The factors predicting this are patient age, time on ADT, and baseline testosterone level.

A recent informative meta-analysis of the use of ADT with radiation showed nonlinear relative benefits of ADT, with reduced benefits beyond 9 to 12 months and, like we observed, increasing rates of overall mortality with increasing use. The investigators showed results in graphical form, allowing the reader to appreciate the impact for specific patient groups.

In these patients, AI is increasingly being used as a tool to assist in predicting which patients may benefit from HT. A study examining the use of Artera’s AI predictive model looked at digital pathology images from men treated with RT—with or without HT—to determine which patients benefit from HT and which do not. Results indicate that approximately one-half to two-thirds of men with intermediate-risk disease, and up to one-third of men with high-risk disease, may not benefit from HT. Many are using the ArteraAI Prostate Test (Artera) as a way to consider HT use in patients who are getting RT, and I see the rates of use for this test going up in general.