Laurent Briollais

PhD

Lunenfeld-Tanenbaum Research Institute - Mount Sinai Hospital

Scientist

PLCO (Learn more about this study)

PLCO-194

Mar 1, 2016

Optimizing prostate cancer PSA screening through genetic profiling: Towards a personalized screening approach

The widespread use of PSA testing has enhanced prostate cancer (PCa) detection at an earlier, more curable stage but on the other hand, its perverse effects have included the (over-)diagnosis of a considerable proportion of indolent cancers unlikely to progress to clinically significant, lethal PCa. In this project, we plan to investigate whether information on genetic profiles can help stratifying PCa risks in the general population, leading to optimal PSA screening strategies, tailored at each individual’s personal risk. As a proof of principle, we will use genetic variants from the PLCO GWAS to evaluate a new personalized screening approach based on genetic profiling. We will use data collected prospectively during the trial on screening outcomes, baseline variables and genotypes from the GWAS SNPs.

Our methodology will be based on multistate models (MSMs) that we will develop specifically for this project. In diseases where multiple outcomes or events can occur over time (i.e. repeated PSA measures, PCa), MSMs are often used to analyse the study data. The outcomes are defined as states. Individuals can move between these states over their lifetimes and a change of state, which occurs when an event happens, is called a transition. Transitions are specified through hazard functions, such as Cox proportional hazards regression approaches. Despite its suitability to model screening outcomes over time, MSMs have seldom been used in observational studies to assess screening efficiency but have been used more extensively in simulation studies of screening cost-efficiency, e.g. with the MISCAN approach. We have recently developed a specific MSM that can assess the effect of the gap time since the last screening visit on the time to cancer or time to death and that we are planning to apply and extent for this project.

We anticipate that this proposal could help defining and promoting a new perspective for PSA screening through personalized genetic profiling.

Our main objective is to better understand the implication of the genetic variants in serum PSA levels and PCa development and assess how genetic profiles can help improving PSA screening. We will take advantage of the PLCO study design that accrued individuals longitudinally to pursue the following objectives:

- In Aim 1, we will develop a general statistical framework based on multistate models to estimate the transition probabilities between the different states of the model (i.e. “healthy”, “elevated PSA”, PCa).
- In Aim 2, we will assess the impact of genetic variants from the KLK region and clinical variables on the transition probabilities estimated in Aim 1. We will identify risk variants increasing the probability to transition to the PCa state, with or without a prior elevated PSA and classify individuals in our sample based on these genetic profiles.
- In Aim 3, we will perform simulations using the MISCAN program to determine screening optimality.