ASCO 2026: Classification Tree Phenotypes for Early Treatment-Limiting Toxicity After ARPI Initiation in mHSPC from the IRONMAN Registry

(UroToday.com) The 2026 American Society of Clinical Oncology Genitourinary (ASCO) Annual Meeting, held in Chicago, IL, will host the Prostate, Testicular, and Penile Cancer – Posters Session. Alexandra Larkin will present Abstract 5095: Classification tree phenotypes for early treatment-limiting toxicity after ARPI initiation in mHSPC from the IRONMAN registry.

Although androgen receptor pathway inhibitors have significantly improved survival outcomes in metastatic hormone-sensitive prostate cancer (mHSPC), early treatment-limiting toxicities remain clinically relevant and may impact treatment persistence and quality of life. Using data from the IRONMAN registry, the investigators sought to develop a simple and clinically interpretable classification tree capable of stratifying patients according to risk of early adverse outcomes following ARPI initiation.

The analysis included 583 men with mHSPC initiating first-line ARPI therapy who had at least 12 months of follow-up or experienced an adverse outcome within the first year of treatment. The composite adverse outcome included either a serious adverse event or ARPI discontinuation due to toxicity or other non-progression-related causes.

Dr. Larkin notes that 122 of 583 patients (20.9%) experienced an adverse outcome within the first year, including 85 serious adverse events and 37 ARPI discontinuations, of which 28 were attributed to toxicity. Adverse outcome-free survival did not significantly differ according to the specific ARPI agent utilized, allowing all ARPIs to be pooled for analysis.

The final recursive-partitioning classification tree retained nine clinically relevant predictors, including EORTC global health rating, hemoglobin, pain score, lactate dehydrogenase (LDH), age, docetaxel exposure, opioid agonist use, and systemic steroid use.

The model demonstrated an in-sample AUC of 0.76 (95% CI 0.71–0.81), while cross-validated AUC was 0.60 (95% CI 0.55–0.65). Patients were ultimately grouped into low-, intermediate-, and high-risk phenotypes with clearly distinct risks of adverse outcomes at 1 year.

Among the 583 patients analyzed, 72% were categorized as low risk, with a 1-year adverse outcome risk of 10.7% (95% CI 7.7–13.6). Intermediate-risk patients comprised 14% of the cohort and demonstrated a 1-year adverse outcome risk of 26.3% (95% CI 16.6–35.9). The remaining 14% were classified as high risk, with a markedly elevated 1-year adverse outcome risk of 69.1% (95% CI 59.1–79.2).

High-risk phenotypes were characterized by poorer baseline health status, symptomatic disease burden, anemia and/or elevated LDH, as well as a treatment-intensity phenotype involving triplet therapy combined with opioid and steroid use.

Take home messages:

  • A simple classification tree identified clinically interpretable phenotypes associated with early adverse outcomes following ARPI initiation in mHSPC
  • Approximately 21% of patients experienced a serious adverse event or treatment discontinuation within the first year of therapy
  • The model stratified patients into low-, intermediate-, and high-risk groups with markedly different 1-year adverse outcome risks
  • High-risk phenotypes were associated with poor baseline health status, symptomatic disease, anemia, elevated LDH, and more intensive systemic treatment
  • These decision rules may help support patient counseling, toxicity monitoring, and shared decision-making in routine clinical practice
  • External validation of the model remains warranted

Presented by: Alexandra Larkin, MPH, Clinical Research Coordinator III at The University of Chicago, Chicago, IL.

Written by: Julian Chavarriaga, MD, Clinical Assistant Professor, Urologic Oncologist, Department of Urology at Penn State Health @chavarriagaj on X during tthe American Society of Clinical Oncology Genitourinary (ASCO) Annual Meeting held in Chicago, IL between May 29th and June 1st, 2026