Prediction rules for aggressive prostate cancer in AS population

Sanda-PHI PCA3
Sanda, MartinEmory University
Feng, ZidingFred Hutchinson Cancer Research Center
No design specified.
Prostate and Urologic Cancers Research Group

We plan develop and validate prediction rules, based on PCA3 and phi along with clinical variables, for aggressive prostate cancer, i.e., Gleason score >= 7, in active surveillance population using the PASS cohort. This project pertains to the development phase of the two-phase study, to formulate prediction rules and provide initial evaluation.

In the development phase, we plan to formulate a number of rules: The first one is to linearly combine PCA3, phi, and clinical variables via logistic regression. The clinical variables include age, BMI, race, DRE result, initial biopsy indicator, number of previous negative biopsies after diagnosis, cores ratio (ratio of biopsy cores containing cancer to total cores), months since diagnosis, and prostate volume. Appropriate variable selection will be implemented in the logistic regression to screen out non-informative clinical variables. The second rule is to linearly combine PCA3, phi, and initial biopsy indicator only, without other clinical variables. Obviously, this rule is expected to have less predictive power than the first one, but can be more practical if the predictive power reduction is limited. The third rule is a “or” logic combination of PCA3, phi, and a composite score of clinical variables. The composite score is obtained via logistic regression. The fourth rule is a “or” logic combination of PCA3 and phi only.

No datasets are currently associated with this protocol.

Version 5.1.0