While androgen deprivation therapy (ADT) in combination with radiotherapy has been shown to improve overall survival (OS), ADT can cause adverse events, including vasomotor symptoms, sexual changes, and fatigue, that may negatively affect patient quality of life.
A study published in NEJM Evidence detailed the development of an artificial intelligence (AI)-based model to predict which patients with localized prostate cancer would benefit from the use of ADT.
Researchers used digital pathology images of pretreatment prostate tissue and clinical data from 5727 patients taken from 5 phase 3, randomized trials that utilized radiotherapy with or without ADT as treatment. The AI model used baseline data to create a binary output to determine if a given patient would benefit from ADT or not.
Validation was performed using data from the RTOG 9408 randomized, clinical trial, which randomly assigned patients to radiotherapy with or without 4 months of ADT. Fine-Gray regression analysis and restricted mean survival times were used to determine the interaction between treatment and the predictive AI model, and within predictive model-positive (patients who benefited from ADT) and model-negative (those who didn’t) subgroup treatment effects.
The validation cohort, which had a median follow-up of 14.9 years, demonstrated that ADT significantly improved time to distant metastasis. A total of 543 (34%) patients were model-positive, with ADT reducing the risk of distant metastasis compared with radiotherapy alone. In the 1051 patients who were model-negative, ADT did not provide benefit.
The AI-based predictive model successfully identified patients with an intermediate risk of prostate cancer who were likely to benefit from treatment with short-term ADT.