Association between prostate specific antigen change over time and prostate cancer recurrence risk: a joint model

Background: Prostate specific antigen (PSA) is an important biomarker to monitor patients after treated with radiation therapy (RT). The aim of this study is to evaluate the relationship between the PSA data and prostate cancer recurrence using the joint modeling. Methods: This historical cohort study was performed on 422 prostate cancer patients. Inclusion criteria included: patients with localized prostate cancer referring to Cancer Institute in Tehran (Iran) through 2007 to 2012, and under radiation therapy. Joint model has two components or sub-model. We showed the results by parameter estimating the longitudinal sub-model and survival sub-model. EM algorithm, Newton-Gauss and Gauss-Hermit law were used for final model parameters. R software version 3.2 was used for statistical analysis. Results: In this study, considering the inclusion and exclusion criteria, out of 422 patients, the data on 314 cases were selected for analysis and the main result of joint model was obtained. PSA directly and significantly was associated with recurrence risk, therefore  increasing 2.6 ml/lit PSA (one unit in transformed PSA) increases 39% recurrence risk (95% CI for RR: 1.09-1.77). Also slope of PSA trend has significant association with prostate cancer recurrence risk (95% CI for RR: 1.05-1.41). Conclusion: This study showed a significant relationship between PSA, and its slope with the recurrence risk by joint model, with regard to the pathological, demographic and clinical features in the Iranian population.