Can AI Help Men Avoid ADT? The ASTuTE Trial is Testing That Question

A prospective clinical trial is testing whether artificial intelligence can help determine which men with intermediate-risk prostate cancer really need short-term androgen deprivation therapy alongside radiotherapy. The ASTuTE study uses the ArteraAI Prostate Test, which analyses digitised biopsy images together with clinical information to estimate both the risk of future metastasis and the likelihood that a patient will benefit from adding ADT.

The first interim analysis includes 200 patients, 150 with unfavourable intermediate-risk and 50 with favourable intermediate-risk disease. Seeing the AI biomarker result changed the final treatment decision in 27.5% of patients. Among men who initially planned to receive ADT, 70.3% ultimately decided against it after the test results were discussed, while only 2.4% moved in the opposite direction.
Overall, the proportion of patients whose final shared decision was to receive short-term ADT fell from 37% before testing to 12.5% afterwards. Clinicians themselves recommended ADT in 46% of cases before seeing the test and only 15% afterwards. The effect was particularly pronounced in unfavourable intermediate-risk disease, where treatment decisions changed in 35% of patients, compared with only 4% in the favourable-risk group.
Interestingly, only 15% of patients in this contemporary cohort were classified as biomarker-positive for ADT benefit. The AI also exposed considerable biological overlap between conventional risk groups: some favourable intermediate-risk patients had metastasis risks similar to the typical unfavourable group, while some conventionally unfavourable patients had very low predicted long-term risk.
The test is not starting from scratch. An earlier validation using the randomised RTOG 9408 trial found that men classified as biomarker-positive gained a significant reduction in distant metastases from short-term ADT, while the much larger biomarker-negative group did not show the same benefit. That retrospective evidence is what provides the biological and clinical basis for ASTuTE.
There is an important limitation: ASTuTE has so far shown that AI changes treatment decisions, not that omitting ADT is prospectively proven to be equally safe. Long-term cancer outcomes will be needed before that question is fully answered.

Source.

Clinical trial.