Skip to main content
Skip table of contents

PSA and survival data set

Download data set: PSA_survival_dataset.csv

This data set has been originally published in:

Desmée, S, Mentré, F, Veyrat-Follet, C, Sébastien, B, Guedj, J (2017). Using the SAEM algorithm for mechanistic joint models characterizing the relationship between nonlinear PSA kinetics and survival in prostate cancer patients. Biometrics, 73, 1:305-312.

It contains PSA kinetics and survival data for 400 men with metastatic Castration-Resistant Prostate Cancer (mCRPC) treated with docetaxel and prednisone, the first-line reference chemotherapy, which constituted the control arm of a phase 3 clinical trial. In this context of advanced disease, the incidence of death is high and the PSA kinetics is closely monitored after treatment initiation to rapidly detect a breakthrough in PSA and propose rescue strategies. In more details:

  • PSA kinetics: 6627 PSA measurements were collected, among which roughly 20% were pretreatment, 60% on treatment, and 20%posttreatment. 2.5% are below the limit of quantification (LoQ),set at 0.1 ng.ml−1.

  • survival: 286 patients deceased (71.5%), leading to a median survival of 656 days.

PSA measurements are displayed on the figure below. Since the data density is too high to see clearly, the next figure is a selection of six individuals, that allows to identify the shape of PSA measurements followed by most individuals: a decrease of PSA concentration after the treatment initiation at time zero, followed by an increase some time later due to resistance.

image-20240704-135404.png

image-20240704-135349.png

The following figure shows the Kaplan-Meier curves for the survival data split in two groups by the median time of end of treatment

image-20240704-135334.png

Below is an extract of the data set file:

image-20240704-135318.png

The columns are:

  • TIME: the time of the measurement or of the dose, column-type TIME.

  • ID: the subject ID, column-type ID.

  • Y: measurements of PSA concentration in ng.ml−1, or death events, column-type OBSERVATION.

  • CENS: CENS=1 indicates values below the limit of quantification (LoQ), column-type CENSORING.

  • YTYPE: the type of measurement. There are two measurement types in this data set: the PSA concentration measurement (corresponding to the PD), and death events, column-type OBSERVATION ID.

  • TIME_ENDtr: day of end of treatment, column-type CONTINUOUS COVARIATE.

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.