ESTRO 2025 - Abstract Book

S2090

Clinical - Urology

ESTRO 2025

4012

Digital Poster Characterisation of MMAI risk and treatment benefit in a contemporary UK localised Prostate Cancer cohort diagnosed with pre-biopsy biparametric MRI Suneil Jain 1 , Yi Ren 2 , Danielle Croucher 2 , Jack Murphy 1 , Ross Murphy 3 , Xiaoxuan Yang 2 , Dibya Mukherjee 2 , Anna Jurek Loughery 4 , Erin Stewart 2 , Trevor Royce 2 , Timothy N Showalter 2 , Andre Esteva 5 , Felix Feng 1 , Emma Allott 1 1 PGJCCR, Queen's University Belfast, Belfast, United Kingdom. 2 Clinical Development, Artera Inc, Los Altos, USA. 3 Centre for Genomic Medicine, Ulster University, Coleraine, United Kingdom. 4 EEECS, Queen's University Belfast, Belfast, United Kingdom. 5 Office of Executives, Artera Inc, Los Altos, USA Purpose/Objective: Current prognostic tools like the National Comprehensive Cancer Network (NCCN) and Cambridge Prognostic Groups (CPG), defined by clinical variables, provide essential frameworks for prostate cancer (PCa) management but suffer from sub-optimal prognostic accuracy, leading to over- or under-treatment. Advanced prognostic biomarkers incorporate tumour biology, enabling more personalised treatment strategies. Leveraging digital pathology-based artificial intelligence (AI) methods, multi-modal AI (MMAI) PCa biomarkers are commercially available that outperform NCCN categories [1] and can predict benefit from short-term ADT (ST-ADT) addition to radiation in NCCN-intermediate patients [2]. The MMAI prognostic risk model has been validated in non-US populations [3,4] demonstrating generalisability, but neither the prognostic risk or ST-ADT biomarkers have been characterised in contemporary patient cohorts. Material/Methods: This study examined the distribution of ArteraAI’s MMAI prognostic risk and ST -ADT benefit groups within clinically defined risk categories of CPG2-5 sequentially-selected localised PCa cases from the Northern Ireland Biobank (NIB) and compared this to commercial data from the US (Table 1). All NIB patients underwent pre-biopsy diagnostic bpMRI and transperineal biopsies. Using diagnostic digitised H&E biopsy images, tumor stage, age, and PSA, the MMAI algorithm stratified patients into low/intermediate/high prognostic risk groups and positive/negative ST-ADT predictive benefit groups. Results: Among the 250 NIB cases analysed, 15 were excluded for insufficient clinical data or non-localised disease. In the remaining 235 patients, 2.6% were CPG2, 60% CPG3, 28% CPG4, 8.9% CPG5. MMAI risk stratification categorised NIB patients as MMAI-low (8.1%), MMAI-intermediate (73%), and MMAI-high (19%). The majority of intermediate-risk NIB patients (CPG3) were classified as MMAI-intermediate (82.4%), while 10.6% were down-classified (MMAI-low) and 7.0% up-classified (MMAI-high) (Table 2). Within CPG3 patients, 57.0% were ST-ADT positive. The results above contrast contemporary US intermediate-risk distributions (CPG3), with 61.7% down-classified (MMAI-low), and only 19.2% ST-ADT positive. The latter finding remained true within Gleason 4+3 patients only (US: 21.1% positive, NIB: 65.7% positive).