Background: Darolutamide (Daro) and enzamutamide (Enza) are both next generation androgen receptor inhibitors with demonstrated activity in men with CRPC. Although both agents are associated with survival improvement, their toxicity profiles are different. To help decipher whether this may impact on patient preference, we designed the ODENZA trial. Methods: ODENZA is a prospective, randomized, open-label, multicenter, cross-over, phase II trial of preference between Daro and Enza in men with asymptomatic or mildly symptomatic metastatic CRPC. Patients were randomized 1/1 to receive Daro 1200 mg/d for 12 weeks followed by Enza 160 mg/d for 12 weeks (Daro-Enza arm) or the reverse sequence (Enza-Daro arm). In both arms, the second treatment was given in absence of evidence of cancer progression at week 12. The primary endpoint was patient preference between the two drugs, as assessed by a questionnaire at week 24. The Prescott's test was used to determine treatment preference in patients fullfilling pre planned criteria (exposure to both treatments, no progression at week 12, and completion of the preference questionnaire). A p-value greater than 0.05 indicates that there is no difference in preference between treatments. Stratification factors were performance status and prior taxane for mCSPC. After week 24, patients went on to an extension period during which they received the chosen treatment until progression or toxicity. The main secondary objectives included reasons for preference, response at week 12, cognitive assessment, and toxicity. Results: Overall 249 pts were randomized, median age 72y (68; 79), ECOG PS 0 (56%), prior taxanes (22%). Two hundred pts fulfilled the pre-planned criteria for evaluation of the preference primary endpoint : 97 (48.5% [41.3;55.7]), 80 (40.0% [33.0;47.0]), and 23 (11.5% [6.8;16.2]) chose Daro, Enza, and had no preference, respectively (unilateral p-value of 0.92). After preference assessment, 186 patients entered the extension period: 103 (55.4%) and 83 (44.6%) received Daro and Enza respectively. The most common factors influencing patient preference all numerically favored Daro over Enza, without significant differences were: less fatigue (44% vs 29%), ease of taking the medication (37% vs 31%), better quality of life (36% vs 28%), ability to be more active (26% vs 15%), ability to concentrate (22% vs 15%) and less falls (6% vs 3%). A PSA50 response was achieved in 76.2% and 83.9% at week 12 with Daro and Enza respectively (p = 0.13). Fatigue was the most frequently reported all grade adverse event at week 12, in 21% and 36% with Daro and Enza, respectively. Conclusions: More patients with early mCRPC preferred Daro over Enza, although the difference did not reach significance, with fatigue as the key influencing factor. Clinical trial information: NCT03314324