- Design
- Prespecified biomarker analysis of a phase 3 randomised trial
- Population
- 492 men with metastatic hormone-sensitive prostate cancer in ENZAMET
- Primary outcome
- Treatment-by-classifier interaction for overall survival
- Effect
- APIC-negative HR 0.50 (95% CI 0.35 to 0.73); APIC-positive HR 1.04 (0.67 to 1.62)
This biomarker analysis applied an artificial intelligence pathology image classifier (APIC), developed and locked in the earlier CHAARTED trial, to digitised H&E slides from ENZAMET, which randomised men with metastatic hormone-sensitive prostate cancer to androgen deprivation plus enzalutamide or a nonsteroidal antiandrogen. Of 1125 participants, 492 had suitable tissue, with median follow-up of 68.8 months.
The classifier called 64% APIC-negative and 36% APIC-positive, and the treatment effect differed by group (interaction p=0.014). In APIC-negative disease, enzalutamide improved survival (HR 0.50, 95% CI 0.35 to 0.73; five-year survival 76.5% vs 58.3%). In APIC-positive disease there was little evidence of benefit (HR 1.04, 0.67 to 1.62). The pattern was similar in low-volume disease and in men not given docetaxel.
A test that reads routine slides and predicts benefit from an androgen receptor inhibitor would be valuable, especially where drug cost decides access, as in India. But this is a secondary analysis of under half the trial, and the classifier is not commercially available. It should not be used to withhold treatment that current guidelines recommend.
- Continue to offer treatment intensification with an androgen receptor pathway inhibitor in metastatic hormone-sensitive prostate cancer as guidelines recommend.
- Ensure diagnostic biopsy slides are archived and retrievable, as slide-based biomarkers may become relevant.
- Do not withhold enzalutamide on the basis of this classifier, which is not validated for routine use.
- Watch for prospective validation before considering AI-based treatment selection.
Why it matters
It shows routine H&E slides may carry information about drug response that clinical factors miss.
Don't overread it
A retrospective biomarker analysis of part of one trial does not justify withholding enzalutamide from APIC-positive men.
The statistics, in plain English
The interaction p-value of 0.014 suggests the benefit genuinely differed between groups, which is stronger than simply comparing two subgroups. Still, only 44% of the trial had usable tissue, and the APIC-positive interval (0.67 to 1.62) is wide enough to include a modest benefit.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for urology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free