The PRIME (Predictive Biomarker Graphical Approach) method [1] can be used to identify a predictive biomarker and suggest a cut-off. The predicted lines from the model which includes treatment effect, biomarker effect, the interaction between treatment and biomarker, and prognostic factors showed whether a predictive biomarker exists by visualising the relationship between the endpoint and the biomarker by treatment [2]. The PRIME method uses the coefficients from the model with interaction to estimate the predicted risk. The predicted risk will be marginalised over the prognostic factors to estimate a predicted risk of the feature by treatment. When the treatment lines cross this will indicate a potential predictive feature and suggest a cut-off where the lines cross as the treatment benefit will change for the population at that feature value or cut-off. A p-value of the interaction term between the feature and treatment can also be used to confirm statistical significance of the feature being predictive.
Xiaowen Tian (xiaowen.tian1@astrazeneca.com)
Gina D’Angelo (gina.dangelo@astrazeneca.com)
- R >= 4.3.0
- R packages:
survival,ggplot2,stdReg(>= 3.4.2),grid,parallel,doFuture,foreach,future,magrittr,dplyr,rlang,stats,sandwich,utils,graphics,data.table
You can install the current development version of prime with:
if (!require("remotes")) install.packages("remotes")
remotes::install_github("AstraZeneca/r-package-prime")The tutorial can be found be navigating to the Tutorial pane of your RStudio IDE. All installed learnr tutorials in your R Library will be automatically indexed and displayed here. Please scroll down and find PRIME Tutorial then click Start Tutorial.
Alternatively, you can run your tutorial by using:
# install.packages("learnr")
learnr::run_tutorial("prime-tutorial", "prime")
[1] D’Angelo G, Tian X, Deng C, Zhou X: “Predictive biomarker graphical approach (PRIME) for Precision medicine”, 2025; arXiv:2504.08087.
[2] Janes H, Brown MD, Huang Y, Pepe MS. An approach to evaluating and comparing biomarkers for patient treatment selection. Int J Biostat. 2014;10(1):99-121.