e23022 Background: Overall survival (OS) is commonly used as a primary endpoint in phase III oncology trials in the United States, serving as direct evidence of clinical benefit and as an objective outcome measure. Other endpoints such as progression-free survival (PFS) and objective response rate (ORR) are also utilized. Factors such as a cancer type and treatment line are considered important for primary endpoint selection, however, the fundamental factor influencing primary endpoint selection across various cancer types or treatment lines remains unclear. I hypothesize that the prognosis of the target patient population is one of the key fundamental factors across any groups in primary endpoint selection. This study aims to investigate the associations between the prognosis and the type of primary endpoint. Methods: All phase 3 clinical trials conducted in the US, focusing on drug therapies for cancer, between January 1, 2015 and Dec 31, 2022 were included. Multinomial logistic regression was utilized to assess the relationships between prognosis, indicated by the median overall survival of the control arm, and the type primary endpoint among OS, non-OS time-to-event endpoints (Other TTE), and response or remission rate (RR). Models were adjusted for cancer type, treatment line, the primary endpoint of the previous clinical study for the same target patient population, monotherapy, mechanism of action (MOA), biomarker for the drug, orphan drug designation (ODD), and accelerated approval (AA) before regulatory approval. Results: Out of 5,877 phase 3 clinical trials, 151 studies and 167 evaluation sets were included in the analysis. When the prognosis and the type of primary endpoint were tested, this factor demonstrated a statistically significant difference with a positive coefficient for Other TTE, RR, and a co-primary endpoint of OS and Other TTE compared to OS. The prognosis, the median OS of the control arm, was 10.5 months (95% CI 5.00-16.6) for OS, 18.5 months (95% CI 7.50-34.3) for Other TTE, 32.9 months for RR, and 13.6 months (95% CI 8.7-46.6) for the co-primary endpoint of OS and Other TTE respectively. No effect modification was observed for Other TTE and the co-primary endpoint of OS and Other TTE. Conclusions: A better prognosis is associated with the selection of surrogate endpoints in phase 3 oncology trials among many factors. The associations between the prognosis and the type of primary endpoint were observed for Other TTE and a co-primary endpoint of OS and Other TTE. These findings elucidate one of the fundamental factors for primary endpoint selection and emphasize the importance of prognosis in selecting a primary endpoint in phase 3 clinical trial in oncology in the US. However, further study is required to comprehensively understand the impact of RR on primary endpoint selection, as this analysis was limited by the lack of median OS data for RR.