An AI-based pathology classifier to predict benefit from enzalutamide in metastatic hormone-sensitive prostate cancer (mHSPC) from ENZAMET (ANZUP 1304).

Abstract

108 Background: The ENZAMET trial established that adding enzalutamide (ENZ) to androgen deprivation therapy (ADT) improves overall survival (OS) in mHSPC. However, heterogeneity in treatment response and toxicity complicate clinical decision-making. We evaluated a previously developed Artificial Intelligence Pathology Image Classifier (APIC) to determine whether it could identify ENZAMET participants (pts) more likely to benefit from adding ENZ rather than NSAA to ADT. Methods: This AI biomarker study analyzed digitized H&E tumor specimens from ENZAMET (ANZUP 1304, NCT02446405), a phase 3 trial randomizing pts with mHSPC to ADT plus ENZ or a non-steroidal antiandrogen (NSAA), with early docetaxel (EDx) permitted. APIC, a model quantifying nuclear morphology and tumor-immune architecture validated in CHAARTED (Medina et al, Clin Can Res 2025), was applied without modification. The primary multivariable analysis evaluated the treatment-APIC interaction for OS using Cox models adjusted for disease volume (CHAARTED criteria), EDx, and age. Sensitivity analyses excluded pts receiving EDx. APIC associations with 18 circulating immune markers were explored. All tests were two-sided with p<0.05 considered significant. Results: Among 393 evaluable pts (median follow-up 70 months), 248 (63%) were APIC-negative and 145 (37%) APIC-positive. APIC significantly modified ENZ benefit (interaction p=0.010). APIC-negative was associated with improved OS with ENZ versus NSAA (HR 0.42, p<0.001; 5-year OS 82% vs 59%), while APIC-positive showed no benefit (HR 0.98, p=0.92; 5-year OS 57% vs 57%). APIC-treatment interaction was significant (p=0.02) in the multivariable model adjusted for clinical covariates. In low-volume disease, ENZ improved OS in APIC-negative (HR 0.19, p=0.0001) but not APIC-positive (HR 1.12, p=0.8; interaction p=0.002). Excluding EDx use (n=227), APIC-negative was associated with ENZ benefit (HR 0.29, p<0.001; 5-year OS 88% vs 60%, interaction p=0.043), while no significant benefit was observed for APIC-positive (HR 0.75, p=0.39; 5-year OS 61% vs 54%) (Table). Analysis of circulating immune markers identified elevated plasma myeloid progenitor inhibitory factor 1 (MPIF1) in APIC-positive pts (1.32-fold, 95% CI 1.11–1.58, p=0.002). Conclusions: APIC status was associated with benefit of ENZ for pts with mHSPC. APIC might help guide treatment selection for pts with mHSPC considered for androgen receptor pathway inhibitors and/or docetaxel. Clinical trial information: NCT02446405 . APIC Status HR (95% CI) P Value Interaction P Biomarker cohort (n=393) Negative 0.42 (0.27–0.64) <0.001 0.01 Positive 0.98 (0.61–1.56) 0.9 Low-Volume Disease Subgroup (n=209) Negative 0.19 (0.08–0.44) 0.0001 0.002 Positive 1.12 (0.55–2.29) 0.8 No docetaxel cohort (n=227) Negative 0.29 (0.15–0.56) <0.001 0.04 Positive 0.75 (0.40–1.44) 0.4