Background: There are nearly 40,000 patients with nmCRPC in the US. This study aims to describe the characteristics of Veterans with nmCRPC and assess the impact of PSA trajectory, baseline PSA and development of metastases on OS.
Methods: Veterans Health Administration electronic health record data (01/2007-08/2017) were used to conduct a retrospective longitudinal study in Veterans with nmCRPC. Patients with castrate testosterone levels ( < 50 ng/dL) or continuous androgen deprivation therapy, a 25% rise in PSA from nadir and absolute increase of 2 ng/mL (index), and a PC diagnosis 12 months before index were selected. Patients were excluded if they had < 12 months of enrollment before or < 6 months after the index date, evidence of metastatic disease 12 months before index (defined based on diagnoses for metastatic disease, use of chemotherapy, immunotherapy, radiopharmaceuticals, or oral mCRPC therapy), or evidence of participation in a clinical trial (defined based on diagnosis codes or prescription for experimental therapies). PSA was measured in the 12 months before index, at index, and 6 months after index. Patients with similar PSA trajectories (PSA patterns over time) were grouped using group-based trajectory analysis (GBTA). OS was assessed with a Cox proportional hazards model adjusted for baseline characteristics, PSA trajectory group, and a timevarying indicator for metastases.
Results: 13,552 Veterans formed the study population. Mean (SD) age was 76.4 (9.4) years and most were white (66%). During follow up (mean [SD]=36  months) there were 6,552 (48%) deaths; median OS was 47 months. GBTA resulted in 4 PSA trajectory groups. The Cox model showed that PSA trajectory group was a strong predictor of OS; compared to Group 1 (n = 213), patients in the other groups were more likely to die (Group 2 [n = 6,558]: HR=1.602, P < .001; Group 3 [n = 5,323]: HR = 1.716, P < .001; Group 4 [n = 1,458]: HR = 2.005, P < .001). Higher log PSA at baseline increased risk of death (HR = 1.092, P < .001). Moreover, compared to patients without metastases, patients who eventually developed metastases were more likely to die (HR: 3.661, P < .001).
Conclusions: In this study of Veterans with nmCRPC, higher baseline PSA, PSA trajectory group, and development of metastases were associated with increased risk of death.