Estimating the optimal positivity criterion for lung cancer screening with low-dose computed tomography: A decision modeling approach
Principal Investigator
Name
Rowan Iskandar
Degrees
Ph.D.
Institution
Brown University
Position Title
Assistant Professor
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-467
Initial CDAS Request Approval
Jan 7, 2019
Title
Estimating the optimal positivity criterion for lung cancer screening with low-dose computed tomography: A decision modeling approach
Summary
In 2014, the U.S. Preventive Services Task Force (USPSTF) issued a recommendation that high-risk adults receive annual low-dose computed tomography (LDCT) screening for lung cancer.
The USPSTF policy recommendations were largely informed by the results of the National Lung Screening Trial (NLST) which provided definitive evidence of a substantial mortality benefit: a 20% reduction in lung cancer mortality with three annual screens with LDCT compared to X-ray.
Despite the significant mortality reduction, LDCT screening is not without unintended harms: false positives.
In NLST, the rate of positive screening tests in the LDCT arm was 24.2%, more than three times as high as in the control arm.
False positives generate unnecessary resource-utilization costs of diagnostic follow-up procedures, risk of medical complications and radiation from these procedures, and also quality of life impact, such as anxiety.
The high rate of false-positive findings in NLST may be barriers to the widespread adoption of LDCT screening.
One strategy to reduce the false positive rate (FPR) would be to use a more stringent criteria for declaring a test result to be positive (positivity criterion), i.e. test results are more likely to be declared negative.
Recent effort for decreasing false positives has led to a development of the Lung Imaging Reporting and Data System (Lung-RADS).
Compared to the NLST positivity criterion, the Lung-RADS system increases the size cutoff for a positive baseline screening result from a 4-mm greatest transverse diameter to a 6-mm transverse bi-dimensional average (and to 20 mm for nonsolid nodules) and requires growth for preexisting nodules.
Pinsky et al. retrospectively applied Lung-RADS in NLST and found a substantial reduction in FPR.
However, such reductions are accompanied by an increase in false negative rate which represents a missed opportunity to prevent cancer deaths.
Furthermore, the Lung-RADS positivity criterion was determined based on balancing the true and false positive rates, which are the characteristics of the test, not the characteristics of the screening beneficiaries, i.e. individuals.
In principle, the optimal positivity criterion of a screening test represents the combination of false-negatives and false-positive that yield the greatest expected life expectancy when applied to a particular decision problem.
Balancing the risks and benefits to determine the optimal positivity criterion can be done explicitly using decision analytic models in which every possible positivity criterion is considered a choice option.
The relative value of each choice can then be explicitly compared by quantifying the lifetime consequences after the implementation of each choice, i.e., in this context, what happens to the health outcomes of the patients who fall into the false negative cases.
By using a mathematical model, we can extrapolate the NLST data beyond the trial follow-up period to quantify the long-term consequences of false negative cases.
This is particularly important since trial follow-up period is only up to 6 years and lung cancer is a lifetime disease.
Therefore, we plan to use a decision-theoretic approach to address the problem of identifying the optimal positivity criterion of LDCT screening for lung cancer.
The USPSTF policy recommendations were largely informed by the results of the National Lung Screening Trial (NLST) which provided definitive evidence of a substantial mortality benefit: a 20% reduction in lung cancer mortality with three annual screens with LDCT compared to X-ray.
Despite the significant mortality reduction, LDCT screening is not without unintended harms: false positives.
In NLST, the rate of positive screening tests in the LDCT arm was 24.2%, more than three times as high as in the control arm.
False positives generate unnecessary resource-utilization costs of diagnostic follow-up procedures, risk of medical complications and radiation from these procedures, and also quality of life impact, such as anxiety.
The high rate of false-positive findings in NLST may be barriers to the widespread adoption of LDCT screening.
One strategy to reduce the false positive rate (FPR) would be to use a more stringent criteria for declaring a test result to be positive (positivity criterion), i.e. test results are more likely to be declared negative.
Recent effort for decreasing false positives has led to a development of the Lung Imaging Reporting and Data System (Lung-RADS).
Compared to the NLST positivity criterion, the Lung-RADS system increases the size cutoff for a positive baseline screening result from a 4-mm greatest transverse diameter to a 6-mm transverse bi-dimensional average (and to 20 mm for nonsolid nodules) and requires growth for preexisting nodules.
Pinsky et al. retrospectively applied Lung-RADS in NLST and found a substantial reduction in FPR.
However, such reductions are accompanied by an increase in false negative rate which represents a missed opportunity to prevent cancer deaths.
Furthermore, the Lung-RADS positivity criterion was determined based on balancing the true and false positive rates, which are the characteristics of the test, not the characteristics of the screening beneficiaries, i.e. individuals.
In principle, the optimal positivity criterion of a screening test represents the combination of false-negatives and false-positive that yield the greatest expected life expectancy when applied to a particular decision problem.
Balancing the risks and benefits to determine the optimal positivity criterion can be done explicitly using decision analytic models in which every possible positivity criterion is considered a choice option.
The relative value of each choice can then be explicitly compared by quantifying the lifetime consequences after the implementation of each choice, i.e., in this context, what happens to the health outcomes of the patients who fall into the false negative cases.
By using a mathematical model, we can extrapolate the NLST data beyond the trial follow-up period to quantify the long-term consequences of false negative cases.
This is particularly important since trial follow-up period is only up to 6 years and lung cancer is a lifetime disease.
Therefore, we plan to use a decision-theoretic approach to address the problem of identifying the optimal positivity criterion of LDCT screening for lung cancer.
Aims
Our specific aims are:
(1) to develop a general decision-theoretic framework for calculating the optimal positivity criterion of a screening test, and
(2) to estimate the optimal positivity criterion based on NLST data and the extrapolation of the trial data via mathematical modeling.
Collaborators
Ilana Gareen, Brown University School of Public Health
Orestis Panagiotou, Brown University School of Public Health