Evaluating net benefit of individualized screening for lung cancer to identify those who benefits most from it
Principal Investigator
Name
Mohsen Sadatsafavi
Degrees
M.D., PhD.
Institution
The University of British Columbia
Position Title
Associate Professor
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-1486
Initial CDAS Request Approval
Dec 16, 2025
Title
Evaluating net benefit of individualized screening for lung cancer to identify those who benefits most from it
Summary
Following landmark studies like the NLST trial, the clinical significance of screening with low-dose computed tomography (LDCT) in reducing lung cancer mortality has been established. However, less is known about which groups of at-risk individuals will benefit the most from screening. Indeed, patients may have differing baseline risks for lung cancer, as well as differences in the likelihood of nodule detection via LDCT. Moreover, the willingness to opt for LDCT could also differ from patient to patient, based on personal values and circumstances (weighing harms and benefits).
Patient-specific characteristics can be considered to make individualized predictions of screening benefit for each patient using risk prediction algorithms. Multivariable risk prediction models, such as PLCOm2012, have been developed and validated to provide individualized estimates of lung cancer risk. These models have been implemented in clinical settings to determine LDCT screening eligibility. However, risk predictions cannot assess whether the benefit of screening by LDCT is heterogeneous among at-risk individuals over and beyond their baseline lung cancer risk. Therefore, in this project, we aim to investigate the heterogeneity of the screening effect and the clinical utility of LDCT, using the NLST data.
The utility of multivariable risk prediction and the net benefit of risk-based screening have previously been examined using NLST data (PLoS Med, 2017, doi: 10.1371/journal.pmed.1002277). Our contribution in this work will be the application of a screening benefit framework to determine whether PLCOm2012 is associated with clinical utility, in terms of absolute reduction of lung cancer risk, and if so, across what ranges of screening thresholds. To this aim, we will identify factors that contribute to screening effect heterogeneity and apply the net benefit calculation framework proposed by Vickers et al (Trials, 2007, doi: 10.1186/1745-6215-8-14) to estimate the individualized benefit of screening.
We will consider the net benefit of three strategies: 1) screening all NLST participants using LDCT, 2) screening no participants, and 3) a model-based approach for selecting individuals for screening based on their estimated expected personalized benefit. The latter strategy will involve modifying PLCOm2012, a risk prediction model, into an individualized screening benefit predictor for selecting patients for screening. This allows us to find important predictors of screening benefit and estimate the net benefit of a model-based approach across a range of screening thresholds, which reflects the reduction in lung cancer mortality weighted by the harms of screening. In practice, this represents the threshold at which the reduction in risk of lung cancer mortality outweighs the burden of unnecessary screening.
Aims
Specific Aims:
Aim 1: To determine whether individual-specific factors, particularly those used in established models such as PLCOm2012, could be modified and used to create prediction models for personalized benefit from screening (in terms of lung cancer risk reduction).
Aim 2: To identify and rank individual characteristics that best predict absolute risk reduction with LDCT screening.
Project goals:
1. To quantify and compare the net benefit of an individualized prediction model for guiding LDCT screening relative to default screening strategies (standard of care).
2. To identify and rank patient characteristics that best predict lung cancer risk reduction due to screening.
Collaborators
Mohsen Sadatsafavi The University of British Columbia
Sum Yin Crystal Leung The University of British Columbia
Ramin Rezaeianzadeh The University of British Columbia