ASCO GU 2026: PAM50 Intrinsic Subtyping and Decipher Genomic Classifier in BCR After Prostatectomy: Implications for ADT Selection

(UroToday.com) The 2026 American Society of Clinical Oncology Genitourinary (ASCO GU) cancers symposium held in San Francisco, CA, between February 26^th and 28^th, 2026, was host to the Poster Session A: Prostate Cancer. Dr. Jonathan D. Tward presented Poster 390: PAM50 intrinsic subtyping and Decipher genomic classifier in BCR after prostatectomy: Implications for androgen deprivation therapy (ADT) selection.

Dr. Tward examined the relationship between PAM50 molecular subtypes and Decipher genomic classifier scores in men with biochemical recurrence after radical prostatectomy, and whether the association of ADT with metastasis following salvage radiotherapy varies by subtype.

He noted that PAM50 classifies prostate cancer lineage into Luminal B versus Non–Luminal B (Luminal A and Basal), and may inform response to androgen pathway therapy. Decipher, in contrast, is commonly used to guide the addition of ADT to salvage radiotherapy. This analysis sought to understand how these two genomic tools track together and whether ADT benefit differs by subtype.

This was a single-institution, prospectively collected post-prostatectomy cohort of men with biochemical recurrence. Radical prostatectomy tissue underwent both PAM50 and Decipher testing. The primary endpoint was metastasis, with death prior to metastasis treated as a competing risk. A Fine–Gray competing risks model included subtype (Luminal B vs Non–Luminal B), ADT with first-course salvage radiotherapy, and a subtype-by-ADT interaction term. Time origin was radical prostatectomy.

Among 123 patients (Luminal A n=30, Luminal B n=53, Basal n=40), Decipher scores differed significantly by subtype. Median genomic classifier scores were highest in Luminal B (0.80), followed by Basal (0.70) and Luminal A (0.60), with a statistically significant difference across groups.

High-risk Decipher scores (0.85–1.00) were observed in 41.5% of Luminal B tumors compared with 22.5% of Basal and 13.3% of Luminal A tumors.

Importantly, many Non–Luminal B tumors crossed Decipher thresholds commonly used to support ADT with salvage radiotherapy: approximately 63% had GC >0.45, 53% had GC ≥0.60, and 19% had GC ≥0.85. Basal tumors demonstrated higher rates of positive surgical margins compared with Luminal B and Luminal A.

In the competing risks analysis (19 metastatic events and 3 competing deaths), ADT was strongly associated with a lower hazard of metastasis in Luminal B tumors (subdistribution hazard ratio 0.10, p<0.001, reference = Luminal B without ADT). Non–Luminal B tumors without ADT trended toward lower hazard compared with Luminal B without ADT, though this did not reach statistical significance. The interaction term between subtype and ADT was not statistically significant, suggesting no clear evidence that the association between ADT and metastasis differed by PAM50 subtype. Descriptively, the highest metastasis incidence was observed in Luminal B patients who did not receive ADT.

Dr. Tward emphasized several key points:

• PAM50 correlates with genomic risk, with Luminal B tumors demonstrating the highest Decipher scores.
• A substantial proportion of Non–Luminal B tumors still meet Decipher thresholds commonly used to justify ADT with salvage radiotherapy.
• ADT was strongly associated with reduced metastasis risk in this cohort.
• No statistically significant subtype-specific interaction was detected, though the analysis was limited by event number and observational treatment assignment.

Clinically, these findings suggest that ADT should not be withheld solely on the basis of a Non–Luminal B subtype, particularly when Decipher scores are high. Luminal B tumors appear to derive the greatest benefit from ADT. Basal tumors, given their higher margin positivity and features suggestive of locoregional failure, may warrant further study of earlier or intensified salvage strategies.

Presented by: Jonathan Tward, MD, PhD, FASTRO, Professor, Department of Radiation Oncology, University of Utah, Salt Lake City, UT