ASTRO 2025: Discussant: Biomarker Breakthroughs in Prostate Cancer

(UroToday.com) The 2025 ASTRO annual meeting featured a biomarker breakthroughs prostate cancer session and a discussant presentation by Dr. Alan Dal Pra. Dr. Dal Pra noted that amongst the six abstracts to discuss, four focused on localized prostate cancer and two on metastatic prostate cancer.

Biomarkers and risk stratification tools support personalized treatment, and are either prognostic (define the risk of developing an outcome, and ideally are independent of standard tools to help guide therapeutic decisions), or predictive (define the likelihood of benefit from a specific therapy, and are ideally proven in randomized trials). The current NCCN guideline-approved biomarkers for prostate cancer (version 2.2026) are highlighted in the following table:
The first abstract discussed by Dr. Dal Pra was “Testosterone Recovery post Androgen Depletion Therapy in Intermediate Risk Prostate Cancer Patients. Does a Higher Dose of Radiotherapy Impact on Survival? Long-Term Data from a Phase III Trial” by Dr. Abdenour Nabid. This study included two phase III trials of 400 intermediate-risk patients undergoing 6 months of ADT + radiotherapy, either 70 or 76 Gy. The median follow-up was 15 years, during which time >4,700 testosterone measurements were obtained. Approximately 76% of patients recovered their testosterone, with a median time to recovery of ~1.6 years. For radiotherapy dose (70 Gy versus 76 Gy), there was no overall survival difference (HR 1.06, 95% CI 0.80-1.39, p = 0.70); however, testosterone recovery was associated with improved overall survival (HR 0.53, 95% CI 0.39-0.73, p < 0.001). Dr. Dal Pra noted several points of caution for this study:

• This was a post hoc analysis
• Radiotherapy doses were below the current standard, but are unlikely to change the outcomes
• This is association, not causation: testosterone recovery may be a surrogate of general health. It may be a biomarker of “host resilience”. Of note, testosterone had no impact on prostate cancer mortality; as such, the overall survival benefit is likely from non-cancer causes such as cardiovascular disease and frailty

Future directions include:

• The importance of survivorship endpoints (cardiovascular disease, metabolic health, and quality of life)
• Tailoring ADT duration decisions, weighing the prostate cancer control versus systemic risks
• Refining predictors of testosterone recovery (age, comorbidity, baseline testosterone, BMI, and molecular genomics)
• Interventions to promote recovery (lifestyle, medical) – is there a role for testosterone replacement therapy?

The second abstract discussed by Dr. Dal Pra was “Hallmark Interferon-Alpha Response Analysis in Localized Prostate Cancer: Analysis of the NRG Oncology/RTOG 9202, 9413, 9902, 0126, and 0521 Clinical Trials” By Dr. Adam Olson. This study included 663 biopsy samples from RTOG 9202, 9413, 9902, 0126, and 0521. High IFN-alpha signature is associated with increased risk of biochemical failure and progression (RTOG 0126), and increased risk of metastasis-free survival and overall survival (RTOG 0521). The predictive signal showed that high INF-alpha has a greater benefit from long-term ADT versus short-term ADT (HR for biochemical failure 0.25 versus 0.71, p-interaction = 0.005 for biochemical failure). Thus, INF-alpha is prognostic (worse outcomes overall) and may predict benefit from long-term ADT. Dr. Dal Pra noted several points of caution for this study:

• This was a post hoc, exploratory analysis across 5 RTOG trials with heterogeneous designs and populations
• The 97 gene IFN-alpha set had no pre-specified cutpoint
• The study had a limited number of events for overall survival and prostate cancer-specific mortality, with wide confidence intervals
• The biology is complex: INF-alpha may reflect immune activation or inflammation

Future directions include:

• Prospective validation in modern radiotherapy + ADT/ARPI cohorts
• Test integration with other immune signatures, such as INF-gamma, T-cell inflamed
• Evaluate the predictive utility for systemic intensification (ADT duration, ARPIs, IO?) 

The third abstract discussed by Dr. Dal Pra was “Genomic Classifier and Multimodal AI Biomarker in Localized Prostate Cancer: Two Sides of the Same Coin” by Drs. Khatri and Yamoah. This analysis included 247 radical prostatectomy patients and 59 radiotherapy patients. The correlation was weak overall (R² = 0.29), and none in the metastasis cohort. Concordance of the risk categories was good, but not great: 15% of multimodal artificial intelligence are high genomic classifier, and 13% of high multimodal artificial intelligence are low genomic classifier in the radical prostatectomy cohort.

Overall, both biomarkers are prognostic for metastasis: in the radical prostatectomy cohort, the genomic classifier had an AUC of 0.92, and a multimodal artificial intelligence AUC of 0.97. In the radiotherapy cohort, the genomic classifier had an AUC of 0.83, and a multimodal artificial intelligence AUC of 0.88. This study also found that the genomic classifier and multimodal artificial intelligence identify distinct biology, with minimal overlap that is complementary and not redundant:

• Genomic classifier: immune related (LAG3, CD4, HLA-B) and DNA repair pathways
• Multimodal artificial intelligence: immune signatures (T-cell exhaustion, immune suppression), p53 pathway

Dr. Dal Pra noted several points of caution for this study:

• This was a retrospective design, with selection bias, no randomization and limited follow-up: radical prostatectomy patients had a median follow-up of 4.2 years, and radiotherapy patients had a follow-up of 1.9 years
• There were wide confidence intervals in correlation estimates
• There is no comparison of predictive value

Future directions include:

• Prospective validation in larger cohorts (integration of both randomized clinical trials)
• Can artificial intelligence pathology approximate or enhance molecular/genomic testing?
• Integrated models may improve risk stratification and predictive value:
  • Multimodal artificial intelligence (intermediate risk prostate cancer) and genomic classifier-derived PAM50 (post-op) can improve decision-making for ADT use
  • Genomic classifier-derived PORTOS can improve decision-making for radiotherapy dose

The fourth study discussed by Dr. Dal Pra was “Influence of HSD3B1 Genotype on Prostate Cancer Outcomes Following Definitive Radiation Therapy with Androgen Deprivation Therapy: A Retrospective Analysis of the Million Veterans Program” by Kylie Morgan. This study included 3,170 localized prostate cancer patients treated with radiotherapy + ADT from the Million Veterans Program. Of the 3,170 participants, 209 (6.6 %) had the homozygous adrenal-permissive genotype, 1,105 (34.9 %) had the heterozygous adrenal-permissive genotype, and 1,856 (58.5 %) had the homozygous adrenal-restrictive genotype. This study found that HSD3B1 homozygous adrenal-permissive genotype was associated with worse outcomes, including an increased risk of biochemical recurrence (HR 1.48, 95% CI 1.08-2.05; p = 0.016) time to metastases (HR 1.43, 95% CI 1.00-2.03; p = 0.05), and time to mCRPC (HR 1.98, 95% CI 1.12-3.41, p = 0.018). Dr. Dal Pra notes that these results in localized prostate cancer are consistent with advanced prostate cancer in which the HSD3B1 homozygous adrenal-permissive genotype is associated with resistance to ADT. The novelty is that this study shows that this association extends to curative intent radiotherapy + ADT in localized prostate cancer. Dr. Dal Pra noted several points of caution for this study:

• This was a retrospective registry, with risk of coding/selection bias, and with a small sample size for the HSD3B1 homozygous adrenal-permissive genotype
• This is a veteran cohort of patients, who are older, have more comorbidities, and so there may be concerns for generalizability
• There are unclear treatment details, such as the duration/type of ADT and radiotherapy details

Future directions include:

• Validation in modern cohorts
• Exploring combination with genomic classifiers (Decipher, Multimodal Artificial Intelligence)
• Is there predictive value? This may be a potential biomarker for ADT duration and intensification strategies 

The fifth abstract discussed by Dr. Dal Pra was “Validation of Six Androgen Production, Uptake and Conversion Genes (APUC-6) in the ECOG-ACRIN E3805 CHAARTED Prostate Cancer Trial” by Dr. Phuoc Tran. The APUC-6 genes are HSD3B1, HSD3B2, CYP3A43, CYP11A1, CYP11B1, CYP17A1, and this study utilized a cohort of 160 metastatic hormone-sensitive prostate cancer patients from the CHAARTED trial. The key findings were that APUC-6 high/androgen receptor low patients had the best outcomes, with a progression-free survival of 47.3 months, and overall survival of 58.1 months. Additionally, APUC-6 low/androgen receptor low patients benefited the most from docetaxel + ADT versus ADT alone (HRs 0.37-0.39). Thus, APUC-6/androgen receptor expression may be both prognostic and predictive for docetaxel benefit. Dr. Dal Pra noted several points of caution for this study:

• This study had a small sample size of 160 patients, and some subgroups of patients were really small
• This was an exploratory analysis, without pre-planned stratification
• CHAARTED predates the widespread use of ARPIs in metastatic hormone-sensitive prostate cancer

Future directions include:

• Validation: replicating findings in larger, modern trials (including ARPI trials)
• APUC-6/androgen receptor expression combined with the Decipher genomic classifier could improve the selection of patients for docetaxel
• Localized disease: testing APUC-6 utility in high-risk localized prostate cancer
• This could inform biology-driven systemic intensification strategies

The final abstract discussed by Dr. Dal Pra was “Predictive Role of Tumor Fraction and Copy Number Alteration Burden in mCRPC Patients Receiving Tandem Actinium-Lutetium Radionuclide Therapy” by Dr. Mariam Amghar. This study included 78 mCRPC patients with a median age of 75.5 years treated with tandem ²²⁵Ac-PSMA-617 and ¹⁷⁷Lu-PSMA-617. Tumor fraction was strongly correlated with PSA but superior in distinguishing metastatic stage (p = 0.027). Additionally, tumor fraction had a ~5 fold increased risk of relapse in the next cycle (p = 0.026). Copy number variant burden had two distinct clusters: low tumor fraction/low copy number variant versus high tumor fraction/high copy number variant (p = 8.09e-08). Furthermore, low copy number variant burden was associated with longer survival (13.8 versus 8.3 months, p = 0.112). Thus, ctDNA (tumor fraction and copy number variant) may track treatment response and resistance better than PSA. Dr. Dal Pra noted several points of caution for this study:

• This was a retrospective, single-center study design, and exploratory in nature
• The cohort was small (n = 78) with short follow-up
• Survival signals were not statistically significant (ie. copy number variant, p = 0.112)
• Biomarker cut-offs and reproducibility need validation 

Future directions include:

• Validate tumor fraction and copy number variant burden in larger, multi-institutional trials: ctDNA versus conventional markers (ie. PSA, imaging) for real time monitoring
• Test predictive utility (tandem therapy versus monotherapy)
• Integration with genomic classifiers and molecular subtyping

Dr. Dal Pra concluded this discussant presentation with the following take-home points:

• Dose alone is not enough; blind radiation dose escalation shows limited survival impact
• Tumor biology matters in genomics, artificial intelligence, pathology, and immune signatures
• Host biology matters, germline HSD3B1 predicts ADT resistance, and T recovery impacts survival
• Integrated biology guides therapy APUC-6 stratifies the benefit of docetaxel, and ctDNA tumor fraction/copy number variant burden tracks resistance to novel radionuclide therapy

Presented by: Alan Dal Pra, MD, University of Miami - Sylvester Comprehensive Cancer Center, Miami, FL

Written by: Zachary Klaassen, MD, MSc – Urologic Oncologist, Associate Professor of Urology, Georgia Cancer Center, Wellstar MCG Health, @zklaassen_md on Twitter during the 2025 American Society for Radiation Oncology (ASTRO) Annual Meeting, San Francisco, CA, Sat, Sept 27 – Wed, Oct 1, 2025.