Enzalutamide associated with greater cognitive decline than darolutamide in advanced prostate cancer

Patients with advanced prostate cancer treated with darolutamide (Nubeqa) experienced significantly less objectively measured cognitive decline over 24 weeks than those receiving enzalutamide (Xtandi), according to results from the randomized phase 2 ARACOG (AFT-47) trial presented at the 2026 American Society of Clinical Oncology (ASCO) Annual Meeting in Chicago, Illinois.¹

The findings were presented by Alicia K. Morgans, MD, MPH, of Dana-Farber Cancer Institute in Boston, Massachusetts.

“Maintaining cognitive function during treatment is a primary consideration for patients undergoing therapy for advanced prostate cancer,” Morgans explained during the presentation. “Unfortunately, data previously describing the effects of hormonal therapy on cognitive function have been limited, with many studies of poor quality and rigor. Prospective comparisons defining cognitive change in the setting of distinct androgen receptor pathway inhibitors, or ARPIs, which are ubiquitously used in many states of prostate cancer, have been limited, particularly in US populations.”

Thus, the ARACOG trial sought to assess cognitive function among patients receiving darolutamide vs patients receiving enzalutamide.

“We specifically chose these agents because there are structural differences that cause differential blood brain penetration, at least in preclinical models, suggesting that blood brain penetration is greater with enzalutamide and that CNS levels are greater with enzalutamide than darolutamide,” Morgans added.

Trial overview

ARACOG was a prospective, randomized, open-label phase 2 study that enrolled 111 patients with advanced prostate cancer between August 2021 and March 2025. Eligible participants had metastatic hormone-sensitive prostate cancer (mHSPC; 54.5%), metastatic castration-resistant prostate cancer (mCRPC; 34.5%), or non-mCRPC (10.9%) and were randomly assigned to receive either darolutamide (n = 55) or enzalutamide (n = 56). Cognitive assessments were performed at baseline and at 12, 24, and 48 weeks.

The primary end point was the maximally changed cognitive domain (MCCD), defined as the percentage change from baseline to 24 weeks in the most affected domain among validated Cambridge Neuropsychological Test Automated Battery (CANTAB) assessments. The cognitive domains evaluated included executive function, visual memory, attention, and working memory. Morgans noted that these cognitive outcomes were measured using research-based neuropsychological assessments rather than clinical diagnoses of dementia or neurocognitive disorders.

At 24 weeks, patients receiving enzalutamide demonstrated a substantially greater decline in objective cognitive performance than those treated with darolutamide. The MCCD decreased by 36.1% in the enzalutamide arm (working memory/executive function domain) compared with 15.8% in the darolutamide arm (visual memory/executive function domain), meeting the study’s primary end point (P = .009). Investigators reported statistically significant differences favoring darolutamide across 4 of 5 cognitive testing modules at 24 weeks.

Morgans also emphasized that an important interpretive consideration is the learning effect inherent to serial cognitive testing: When individuals repeat analogous cognitive tests, observed test scores are expected to rise over time even if underlying cognitive function is stable, due to repeated testing. Against this backdrop, patients treated with darolutamide showed increasing median CANTAB scores across modules between baseline and 24 weeks—consistent with preserved cognitive function and a measurable learning effect. In contrast, patients treated with enzalutamide showed flat or slightly declining median scores across the same modules, suggesting that the expected learning-effect gains were being offset by treatment-related cognitive decline. Morgans again noted that this interpretation should be approached cautiously, as CANTAB modules are research instruments, not clinical diagnostic tools, and cannot be used to diagnose dementia.

An exploratory crossover component allowed patients meeting predefined cognitive or neurologic deterioration criteria to switch therapies. Although similar numbers of patients in each treatment arm met crossover eligibility requirements, only patients assigned to enzalutamide elected to cross over to darolutamide. By week 24, 30 patients had switched from enzalutamide to darolutamide.

Interpreting the findings

Morgans noted several limitations of the ARACOG study. The study was conducted primarily through US academic centers and enrolled a predominantly White patient population, potentially limiting generalizability. Investigators also noted baseline imbalances, including a greater proportion of patients with mCRPC and a higher proportion of Black patients in the enzalutamide arm.

Additionally, darolutamide was supplied through the clinical trial, whereas enzalutamide was obtained through standard-of-care channels. Morgans acknowledged that financial considerations and treatment access may have influenced crossover decisions.

Nevertheless, the findings provide prospective evidence that cognitive outcomes may differ among ARPIs and highlight the importance of incorporating quality-of-life considerations into treatment selection.

REFERENCE