[4Xin1] Poster session 2

[4Xin1-19] Prognostic Prediction for Patients with Castration-Resistant Prostate Cancer Using Medical Claims Database

〇Keita Takahashi¹, Toshiki Iwai¹, Kyoko Katsumata¹, Yasuo Sugitani¹ (1.CHUGAI PHARMACEUTICAL CO., LTD.)

Keywords:RWD, Medical, Prostate cancer, Gradient boosting, Medical Claims Database

The number of patients with prostate cancer in Japan was approximately about 95,000 in 2019. Prostate cancer ranks first in the tabulation of the number of patients by site of cancer in Japanese men. Hormonal therapy is the standard treatment; however, some can progress to castration-resistant prostate cancer (CRPC) if resistant to treatment was shown. In that case, less treatment options cause rarity of successful treatment and cure. Therefore, it is of great medical significance to predict the prognosis of CRPC. The aim of this study was to investigate the models for prognostic prediction of CRPC. Medical claims database was used (Medical Data Vision (MDV) (2008 to 2022)). The objective of this study was to predict prognostic by gradient boosting model with variables which were based on the data obtained in the actual clinical setting. The time from the diagnosis of prostate cancer to the diagnosis of CRPC, patient characteristics such as laboratory test values, and drugs were used as explanatory variables, and PSA response (known as tumor marker for prostate cancer) decrease by half after treatment was used as outcome.