New Machine Learning Approach Aims to Enhance Prostate Cancer Treatment
A recent study presented at the Society of Nuclear Medicine and Molecular Imaging 2026 Annual Meeting introduces a novel machine-learning technique that could improve treatment planning for metastatic castration-resistant prostate cancer (mCRPC). This method leverages pre-existing data from pre-therapy PET/CT scans to estimate radiation doses to tumors and healthy organs before commencing therapy with prostate-specific membrane antigen (PSMA), thus allowing for more personalized treatment plans and potentially reducing toxicity risks.
Radiation dosimetry is a critical component of optimizing ⁷⁷Lu-PSMA radiopharmaceutical therapy. Traditionally, dosimetry calculations rely on post-therapy imaging, which can be cumbersome and resource-intensive. The new approach proposes using pre-therapy PET/CT to assess how effective a treatment might be, while also evaluating potential risks involved.
Amit Nautiyal, PhD, a scientist and NIHR fellow at University Hospital Southampton, explained the innovation: "Our study sought to determine if information already available from these scans could guide treatment planning before therapy begins and support more personalized care."
The proof-of-concept study included nine patients with mCRPC undergoing ⁷⁷Lu-PSMA radiopharmaceutical therapy. The research team developed a machine learning model to predict absorbed radiation doses, analyzing 57 tumors, 36 salivary glands, and 18 kidneys. This model used uptake-based PET metrics, radiomic features, and clinical biomarkers to make predictive assessments, which were then compared with dosimetry calculated post-therapy to assess prediction accuracy.
According to the findings, the 18F-PSMA PET/CT-based machine learning model demonstrated a promising capability to predict tumor and organ absorbed dose effectively. By incorporating varied data points and accounting for patient-level variability, there is potential to utilize pre-therapy information to predict post-therapy dosimetry reliably.
"If validated in larger studies, this approach may improve patient selection and support better decision-making during pre-treatment assessment, helping to optimize ⁷⁷Lu-PSMA therapy for individual patients," Nautiyal noted. This study underscores the burgeoning role of imaging not just in diagnosis, but also in actively guiding personalized treatment strategies.
This research forms part of a five-year plan dedicated to collecting more comprehensive data and developing a validated model. Future efforts will focus on larger, multi-center cohorts for enhanced pre-therapy absorbed dose predictions and independent validation to aid patient stratification in clinical settings.