[Original names, locations, and numbers have been redacted for client privacy]

Naheed Ali, MD, PhD

Abstract / Summary

Lung cancer screening (LCS) with low-dose computed tomography (LDCT) may improve outcomes by detecting the disease early. One major unanswered question is whether the benefit-to-harm ratio found in clinical trials will apply to older and sicker populations as LCS is implemented in real-world settings and populations. Smoking history, the strongest risk factor for lung cancer, is also strongly linked to morbidity and death from non-lung cancer causes (e.g., chronic obstructive pulmonary disease emphysema), which limit life expectancy and increase the risk of complications from diagnostic or therapeutic procedures.

Our proposed study aims to precisely characterize this vulnerable subpopulation with a high burden of comorbidities and quantify the benefits and harms of LCS to help patients considering LCS make better decisions. Our study will use real-world data to characterize this subpopulation of "marginal" LCS candidates, reducing uncertainty for patients and providers. From 2016 to 2022, we will use electronic health records and claims data for 34,039 patients aged 55 to 80 undergoing annual LDCT screening in geographically diverse real-world settings. We will then simulate LCS outcomes in the US population using observational data and validated Cancer Intervention Simulation Network models. This proposal generates real-world data for validated simulation models to improve patient-centered decision-making in LCS candidates whose net benefits of screening are uncertain.

Public Health Significance

Presently, little is known about a possible threshold where the benefits of early lung cancer detection no longer outweigh the risk of death from a competing cause. By examining real-world, up-to-date data on the effects of LCS, our proposed study has direct value for advancing public health.

Detailed Objectives

Lung cancer is the leading cause of cancer deaths in the US and worldwide [1] because most patients are diagnosed with advanced, incurable diseases [2]. Through early detection, lung cancer screening (LCS) with low-dose computed tomography (LDCT) has the potential to alter lung cancer outcomes. In 2011, the National Lung Screening Trial (NLST) discovered that three annual LDCT screenings reduced lung cancer mortality by 20% among high-risk current and former smokers compared to chest radiography [3]. Thus, risk-based eligibility criteria now recommend LCS with LDCT.

The RCT is a powerful tool for assessing the benefits and harms of an intervention, but observational studies are needed to evaluate its performance in real-world settings and populations [4]. LCS's biggest question is whether the NLST's benefits/harms ratio will apply to an older, sicker population. US adults eligible for LCS are nearly twice as likely to be over 70 and smoke than NLST participants [5]. Approximately 3 million of the 8.6 million US LCS-eligible adults have chronic co-existing conditions that may reduce the net benefit of screening for early-stage disease [6]. We previously found that elderly stage IA lung cancer patients with two comorbidities were twice as likely to die within 90 days of surgery as those with one [7].

The main problem for LCS candidates is that smoking history, the strongest risk factor for lung cancer, is also strongly linked to morbidity and death from non-lung cancer causes like COPD and emphysema [8]. These diseases increase lung cancer risk and should benefit most from screening, but they also reduce life expectancy and increase the risk of complications from downstream diagnostic or therapeutic procedures [9]. LCS appears "marginal" for some patients, and the tipping point where potential harms outweigh benefits is unknown. We don't know what health factors likely cause this tipping point or how many patients may reach it. Our proposed study will precisely characterize this vulnerable subpopulation with a high comorbidity burden and quantify the benefits and harms of LCS to help patients considering LCS make informed decisions.

To fully characterize LCS with LDCT outcomes, we propose to collect and analyze data from real-world populations and settings, focusing on "marginal" LCS candidates. We propose to use electronic health records and claims data for patients aged 55–80 undergoing annual LDCT screening in geographically diverse real-world settings from 2016–2019 (retrospective cohort) and 2020–2022. A common data standard based on the PCORI-funded Watch the Spot (WTS) trial infrastructure will integrate the data into a single repository. Using these real-world data and validated CISNET models, we will simulate LCS outcomes across the entire US population eligible for screening.

Our exemplar hypotheses goals are:

Aim 1: Assess multimorbidity (chronic co-existing conditions, functional limitations, and/or impaired pulmonary function) in real-world LCS patients, focusing on marginal patients whose net benefit of LCS is uncertain. Race, socioeconomic status, and age affect this burden.

Aim 2: Quantify LCS's potential harms (e.g., false-positive results, procedure-related complications) and benefits (e.g., early-stage disease at diagnosis) in people with diverse multimorbidity.

Hypothesis: Chronic co-existing conditions, functional limitations, and impaired pulmonary function increase LCS harm.

Aim 3: Use validated CISNET simulation models and real-world data to compare LCS's long-term benefits and harms across subpopulations with diverse multimorbidity.

Hypothesis: Patients with moderate-to-severe comorbidity, functional limitations, and severe COPD will not benefit from LCS.

This proposal addresses calls to improve patient-centered decision-making in LCS candidates whose net benefits of screening are uncertain by using previously unavailable real-world data and validated simulation models. Our findings will aid LCS discussions between patients and providers. Drs. Mendez and Laylo, MPIs from the American Cancer Society's National Lung Cancer Roundtable, will present our study findings, continuing our team's history of influencing LCS guidelines [10].

Research Methodology