Real-time individual predictions of prostate cancer recurrence using joint models, "Beyond the Abstract," by Jeremy M. G. Taylor, PhD

BERKELEY, CA ( - This paper describes the methodology and results associated with a new online calculator for predicting prostate cancer recurrence. The calculator is designed to be an aid to patients and their health care providers in making decisions about whether to start salvage hormone therapy. The calculator is applicable for men who were previously treated with external beam radiation therapy for localized prostate cancer, and since the time of treatment, have been monitoring their PSA values. What the calculator provides is an evidence-based quantitative assessment of the risk of clinical recurrence (positive biopsy or metastatic disease) within the next 3 years. The website for this calculator is There are many calculators available in prostate cancer, which are sometimes called nomograms, but what is unique about the calculator described in our paper is that it can take as input ,and utilize, a whole sequence of PSA values, whereas other calculators use just a single snapshot of a PSA value. The underlying model is based on some fairly sophisticated statistical ideas that include joint models of longitudinal and event time data, and Markov chain Monte Carlo methods. The paper is characterized by the very careful, principled, and fully-documented approach to building the statistical model and evaluating it. The statistical model is built from three large datasets from the University of Michigan, Beaumont Hospital, and RTOG, and then validated on a dataset from British Columbia.

The paper is published in Biometrics, a top biostatistical journal, and the technical level of the paper will make it quite challenging for most prostate cancer clinicians. But this should not deter someone from using the online calculator. It is not necessary to understand all the details of a calculator if it has been built in a correct way, using good data, and has been demonstrated to be valid on an external dataset. The calculator can be viewed as a decision aid. Currently the health care provider would take information derived from the patient’s history (sequence of PSA values as well as maybe his pretreatment T-stage, biopsy Gleason grade, and PSA value.) perform some sort of “calculation” in their mind which combines this patient-specific information with their experience, their knowledge of the literature and clinical guidelines, come up with an assessment of the level of risk of recurrence, and then make a decision about whether to initiate salvage hormone therapy. What the online calculator provides is a data-based version of the same calculation, it provides the probability of clinical recurrence within the next 3 years, derived in a defined and principled way from the same information.

The calculator can be used for monitoring patients, and an updated estimate of risk obtained after the next PSA is measured. It could also be used for defining entry criteria for enrollment into a clinical study, when the goal is to include those who are at high risk for clinical recurrence.

A drawback of the calculator is that it only applies to patients treated with external beam radiation. A similar calculator for patients who received brachytherapy as their initial treatment, or surgery, or planned hormone therapy along with radiation therapy would certainly be useful as well.

Written by:
Jeremy M. G. Taylor, PhD as part of Beyond the Abstract on This initiative offers a method of publishing for the professional urology community. Authors are given an opportunity to expand on the circumstances, limitations etc... of their research by referencing the published abstract.

Department of Biostatistics, Department of Radiation Oncology and Comprehensive Cancer Center, University of Michigan, Ann Arbor, MI USA

Real-time individual predictions of prostate cancer recurrence using joint models - Abstract

More Information about Beyond the Abstract