Key result
MMAI test optimizes risk stratification beyond NCCN criteria by better predicting distant metastasis events.
Why the study?
ADT increases cardiovascular disease risk, especially in underserved populations, necessitating integrated AI risk classifiers to optimize outcomes and guide comorbidity-informed treatment selection.
Does the integration of the MMAI test with NCCN risk criteria improve risk stratification and guide ADT use in patients with localized prostate cancer?
Cohort (n=440)
Does the integration of the MMAI test with NCCN risk criteria improve risk stratification and guide ADT use in patients with localized prostate cancer?
Answers are grounded in this paper and cite their sources.
The MMAI test improves risk stratification beyond standard NCCN criteria in localized prostate cancer, potentially reducing unnecessary ADT use and its associated cardiovascular risks.
e17131 Background: Androgen deprivation therapy (ADT) is the backbone of systemic treatment for prostate cancer (PCa). However, ADT is also associated with increased cardiovascular disease (CVD) risk. This is particularly evident among members of the underserved population who have the highest incidence of PCa as well as treatment related mortality due to CVDs. The emergence of personalized biomarkers such as the Artera multimodal artificial intelligence (MMAI) has shown superior predictive ability for ADT treatment benefit as well as prognostic ability by aiding in the detection and prognostication of aggressive PCa. Here, we report the integration of the MMAI test with existing risk classifiers to optimize outcomes and guide comorbidity-informed treatment selection in PCa. Methods: We retrospectively analyzed 440 patients with localized PCa who underwent standard-of-care (SOC) treatment, defined as radical prostatectomy (RP) or radiotherapy (RT) ± ADT. MMAI scores were generated utilizing histopathological image features combined with clinical parameters. The primary objective assessed the benefit of adding MMAI test to existing National Comprehensive Cancer Network (NCCN) risk information to optimize risk stratification. Secondary objective evaluated the effect of MMAI risk-adaptive ADT use in men with intermediate-risk PCa. Results: Final analytical cohort included 436 evaluable cases with biopsy derived MMAI scores, including 83 cases with biochemical recurrence (BCR) and 20 distant metastasis (DM) events. Among men classified as low NCCN risk (n = 108), 11% (n = 12) had high MMAI scores suggesting high DM risk. Similarly, 17% (n = 14) of NCCN favorable intermediate risk (n = 84) patients had high MMAI scores, indicative of high DM potential. Conversely, 14% (n = 18) of NCCN unfavorable–risk (n = 131) and 3% (n = 3) of NCCN high-risk patients (n = 113) had low MMAI scores indicative of low risk of DM. We further examined cases with discordance between NCCN and MMAI risk categories to evaluate the risk of aggressive tumor biology as described by the BCR and DM events. Among discordant cases, a higher number of BCR (P = .03) and DM (P = .001) events were observed in the NCCN Low/MMAI high risk group (BCR = 9, DM = 5) compared with NCCN high/MMAI low risk group (BCR = 1, DM = 0). Furthermore, we evaluated 131 patients classified as unfavorable intermediate NCCN risk, for whom ADT would be the SOC. Of the 131 patients, 24 were classified as MMAI high-risk, with 20 of the 24 patients (83%) receiving ADT, whereas in patients classified as MMAI low/intermediate risk (n = 107), only 3 (3%) were treated with ADT. Conclusions: Our results indicate that MMAI optimizes risk stratification beyond NCCN criteria, limiting overtreatment with ADT and mitigating CVD risk.
No takes yet. Share an insight, caveat, or question.
Trivedi et al. (2026) conducted a cohort in Localized prostate cancer (n=440). Artera multimodal artificial intelligence (MMAI) test vs. NCCN risk classification alone was evaluated on Biochemical recurrence (BCR) and distant metastasis (DM) events (p=0.001).
Synapse has enriched 10 related papers. Consider them for comparative context: