AI Method Shows Promise in Predicting Prostate Cancer Prognosis - EMJ

AI Method Shows Promise in Predicting Prostate Cancer Prognosis

A NEW AI-based method for measuring the volume of intraprostatic tumours through MRI imaging could improve the ability to predict the prognosis of patients with localised prostate cancer, according to a recent single-centre study.

The study, conducted between January 2021 and August 2023, evaluated 732 patients diagnosed with localised prostate cancer who underwent either radiation therapy (RT) or radical prostatectomy (RP). It aimed to determine if the tumour volume measured through AI-generated segmentations, referred to as VAI, could independently predict patient outcomes, specifically the time to metastasis.

Using advanced MRI imaging, an AI segmentation algorithm was developed and trained to delineate specific tumour lesions, classified as Prostate Imaging Reporting and Data System (PI-RADS) 3–5, within the prostate. Researchers randomly divided patients who received RT into cross-validation and test groups, allowing the AI model to learn from one set before predicting outcomes in the test group. The study also included patients undergoing RP, a common surgical treatment for prostate cancer.

According to the study results, an increase in the AI-determined tumour volume was strongly associated with a shorter time to metastasis. In the combined RT group, where patients had a median follow-up period of 6.9 years, the model revealed a 9% increased risk of metastasis for every additional millilitre of tumour volume. In the RP group, with a median follow-up of 5.5 years, each millilitre increase in tumour volume was associated with a 22% higher risk.

For predicting 7-year metastasis in the combined RT group, the AI model demonstrated an area under the receiver operating characteristic curve (AUC) of 0.84, outperforming the traditional National Comprehensive Cancer Network (NCCN) risk categorisation, which scored 0.74. The results were similar in the RP group for 5-year metastasis prediction, where the AI-based measurement achieved an AUC of 0.89 compared to NCCN’s 0.79.

The study concludes that AI-based tumour volume measurements provide an independent, prognostic marker for localised prostate cancer. This innovation could offer clinicians a powerful tool for more accurately assessing patient risk and personalising treatment strategies, enhancing the standard of care for prostate cancer patients.

 

Reference

Yang DD et al. AI-derived tumor volume from multiparametric MRI and outcomes in localized prostate cancer. Radiol. 2024;313(1):e240041.

 

Rate this content's potential impact on patient outcomes

Average rating / 5. Vote count:

No votes so far! Be the first to rate this content.