Quantitative Imaging Analysis to Identify Chronic Respiratory Disease

Chronic respiratory diseases (CRDs), such as chronic obstructive pulmonary disease (COPD) and interstitial lung disease (ILD) are currently the 4th leading cause of death in the U.S., yet often remain undiagnosed and under-treated until the advanced stages. Current research suggests an increased prevalence and rising incidence of CRDs among Veterans relative to the general population. Yet, despite a

high prevalence and evidence supporting improved outcomes with early medical management, no screening programs currently exist for CRDs. Chest computed tomography (CT), a medical imaging modality employed for lung cancer screening (LCS), can detect structural changes in the lungs associated with CRDs, but their

use has been limited by (1) the labor-intensive nature and inter-person variability of visual interpretation of images, (2) clinical reports which are often focused solely on acute findings (lung nodules, pneumonia) with inconsistent reporting of chronic conditions. Quantitative imaging analysis (QIA) techniques have been

developed which can objectively detect and quantify a broad range of pathological changes directly from chest CT imaging data, often with increased sensitivity relative to visual methods. We assert the application of QIA to clinically obtained chest CT data within the auspices of well-organized LCS program represents an

opportunity to identify and characterize undiagnosed CRDs among a high-risk Veteran population. We propose to develop and validate a clinical tool, the Quantitative Imaging Analysis-based Risk Summary (QIA-RS), which will translate imaging information from LCS chest CTs into practicable evidence in three CRD domains: lung function impairment, symptoms and functional status, and future respiratory

healthcare utilization. QIA will be performed using TRM-approved software behind the VA firewall to assess features of CRD (e.g. emphysema, airway wall thickness, interstitial lung abnormalities, and total lung capacity) on archived and newly acquired chest CT data from patients enrolled in the VA Boston LCS program

(4,777 unique referrals between 2017-2019, with ~1400 new referrals/year). Clinically-ascertained spirometry available in approximately 2,400 subjects, will be used to train and validate models to predict lung function impairment using QIA features as predictors (QIA-RS lung function impairment domain – Aim 1). Because

individuals with undiagnosed CRDs (the target population for our QIA-RS tool) have been incompletely characterized in the literature, we propose to recruit individuals with no previous history of lung disease at the time of LCS (n=300) for an in-person study visit where lung function, respiratory symptoms, and functional

status (exercise capacity, health related quality of life) will be assessed and used to identify thresholds of QIA- assessed features associated with impairments (Aim 2 – QIA-RS respiratory symptom and functional status domain). We will follow individuals recruited in Aim 2 (n=300) via telephony and medical record review for 12

months to assess prospective (a) respiratory events (telephone, outpatient, urgent care / emergency, hospitalization encounters for respiratory symptoms) and (b) new respiratory medication use and will integrate data on lung function and respiratory symptoms (Aim 2) and common and low abundance inflammatory

markers to refine risk estimates for QIA-assessed features as predictors of respiratory outcomes (Aim 3 – QIA-RS respiratory healthcare utilization domain). The validated QIA-RS tool, which will provide succinct reports of risks associated with CRDs along with actionable recommendations for care, represents a scalable,

imaging-based solution to identify and risk stratify previously undiagnosed CRDs among Veterans. This application of QIA technology to clinically-ascertained imaging studies represents an innovative and efficient use of existing data to promote the delivery of personalized care for individual Veterans and will assist in

resource allocation for disease management at the organizational level.