AI and Cancer Research ── Prostate ── 2026-11-02
Prostate cancer: treatment, trials and AI
- Abstract
Background: Whole-gland habitat analysis may characterize intraprostatic heterogeneity without requiring index-lesion delineation. We developed and externally evaluated a lesion-independent biparametric MRI (bpMRI)-based whole-gland habitat framework for pretreatment identification of pathologically high-risk prostate cancer (PCa) and assessed a prespecified capsule-proximal periprostatic adipose tissue (PPAT) extension for incremental predictive value.
Methods: This retrospective two-center study included 301 patients with pathologically confirmed PCa: 241 for model development and 60 for independent external validation. Pathologically high-risk PCa was defined as Gleason score ≥ 4+3 or International Society of Urological Pathology grade group ≥ 3. Patient-specific K = 2 whole-gland habitats were generated by joint k-means clustering of T2-weighted images and apparent diffusion coefficient maps, with the lower-ADC cluster designated H1. Habitat-derived signatures and a PPAT 0–3 mm radiomics signature were combined with log-transformed prostate-specific antigen density (log-PSAD) in score-level fusion models. Development performance was estimated by stratified five-fold cross-validation using out-of-fold predictions; fitted models were applied unchanged to the external validation cohort.
Results: The external validation cohort included 47 patients with pathologically high-risk and 13 with lower-risk PCa. Conventional whole-gland radiomics combined with log-PSAD achieved an external area under the receiver operating characteristic curve (AUC) of 0.791 (95% CI, 0.664–0.899), and the standalone PPAT signature achieved an AUC of 0.702 (95% CI, 0.540–0.845). H1-derived gland fusion achieved an AUC of 0.836 (95% CI, 0.715–0.936), compared with 0.840 (95% CI, 0.710–0.941) for the H1-derived PPAT-extended fusion model. Adding PPAT changed the AUC by 0.003 (95% CI, −0.046 to 0.048; bootstrap p = 0.881). H1-derived gland fusion also had a lower Brier score (0.146 vs. 0.175), lower expected calibration error (0.153 vs. 0.219), and higher average net benefit over both prespecified threshold-probability ranges.
Conclusions: In a small independent external validation cohort ( n = 60), the lesion-independent whole-gland habitat framework showed discrimination between pathologically high-risk and lower-risk PCa. Adding capsule-proximal PPAT produced only a minimal increase in AUC and did not reduce prediction or calibration error or increase net benefit. Larger multicenter studies are needed to confirm the external findings and determine whether PPAT provides incremental predictive value beyond H1-derived gland fusion.
Journal IF-equivalent: 5.3 (OpenAlex 2-year mean citedness, value as of 2026-10-02, retrieved 2026-10-03; not the official Clarivate IF)Reference: Huang J, Gong J, Zhang Y, Yang O, Li J, Chen J, et al. Biparametric MRI–based whole-gland habitat analysis for pretreatment identification of pathologically high-risk prostate cancer: a two-center study with external validation. Cancer Imaging. 2026 Oct 31 [Epub ahead of print]. doi:10.1186/s40644-026-01139-0.Checked: Abstract only - Abstract
We established and validated perineural invasion density (PNID) as a superior prognostic marker and developed an MRI-based AI model for its preoperative prediction. In this multicenter study of 1091 PCa patients undergoing radical prostatectomy, we evaluated PNID's association with biochemical recurrence (BCR) and radiographic progression-free survival (rPFS). We also analyzed concurrent β-blocker or antimuscarinic use with adjuvant ADT. A deep learning model was developed in a primary cohort (n = 523) and validated in two external cohorts (n = 317, 251). High-PNID was an independent prognostic factor for BCR (HR 4.06, 95% CI: 2.75-5.97, P < 0.001) and rPFS (HR 4.61, 95% CI: 2.97-7.16, P < 0.001). Notably, concurrent β-blocker use was significantly associated with favorable BCR outcomes exclusively in the High-PNID subgroup (HR 0.32, 95% CI: 0.12-0.84, P = 0.021), with no benefit in the Non-High PNID population or with antimuscarinics. The AI model showed excellent performance predicting High-PNID, with AUCs of 0.917 (training), 0.842 (internal validation), 0.889 and 0.915 (external validations). Model-identified High-PNID patients had significantly worse outcomes. PNID surpasses traditional PNI in prognostic accuracy and identifies a high-risk phenotype. High-PNID may define a subgroup benefiting from adjuvant β-blocker therapy, providing an exploratory clinical rationale for targeting the nerve-tumor microenvironment in stratified subpopulations. Hence, this model may be a reliable, non-invasive tool for preoperative identification of this high-risk subgroup, aiding clinicians in personalized risk stratification and treatment planning.
Journal IF-equivalent: 9.5 (OpenAlex 2-year mean citedness, value as of 2026-10-18, retrieved 2026-10-19; not the official Clarivate IF)Reference: Liu B, Zhang L, Zhu W, Yang M, Liu X, Li X, et al. Non‐Invasive Assessment of Perineural Invasion Density: A Novel Marker for Prognosis and Treatment Stratification in Prostate Cancer. Advanced Science. 2026 Oct 31 [Epub ahead of print]. doi:10.1002/advs.78460.Checked: Abstract only