The National Lung Screening Trial demonstrated that those randomized to annual CT scanning experienced a 20% relative reduction in mortality from lung cancer. A challenge to CT screening is, however, the scope of such an undertaking. Determination of which smokers are at greatest risk may allow selected allocation of limited health care resources. A detailed objective analysis of that first screening CT scan may facilite this process.

Several investigations suggest that comorbid conditions may mitigate or exacerbate the risk of developing lung cancer. For example, the presence of emphysema or cardiovascular disease is associated with increased odds of having lung cancer while higher BMIs may be protective for such disease. Our lab has developed and refined automated techniques to objectively assess emphysematous destruction of the lung parenchyma, skeletal muscle, subcutaneous fat, visceral fat, fatty infiltration of the liver (which in aggregate may describe body composition), and atherosclerotic calcification of the thoracic aorta (a metric of cardiovascular risk). We hypothesize that the application of these techniques to the NLST scans will increase the specificity of CT screening for lung cancer as detailed in the following specific aims.

Hypothesis Aim 1: Objective assessments of emphsyema are associated with an increased risk of developing lung cancer.
Aim 1: To determine if emphysema and its radiologic subtypes are associated with prevalent lung cancer and the subsequent risk of developing lung cancer.
Approach Aim 1: We will objectively assess continuous measures of parenchymal remodeling in the whole lung and by lobe. Cox proportional hazard models will be used to determine time to event (lung cancer diagnosis). Secondary analysis will examine the relationship of parenchymal remodeling with prevalent lung cancers on the baseline CT scan.

Hypothesis Aim 2: Objective CT based assessments of body composition can identify smokers at increased risk for developing lung cancer.
Aim 2: To determine if skeletal muscle, subcutaneous fat, visceral fat, and fatty liver are associated with prevalent lung cancer and the risk of developing lung cancer.
Approach Aim 2: We will objectively assess pectoralis muscle area, subcutaneous fat adjacent to the pectoralis muscle, visceral fat, and fatty liver. Continuous measures will be utilized in Cox proportional hazard models to determine time to event. The association of dichotomous data (about median) with the development of lung cancer will be determined by ROC analysis. Iterative categorization of the body composition data will be tested to maximize the c-statistic.

Hypothesis Aim 3: Objective assessments of the presence and amount of thoracic aortic calcification can identify smokers at increased risk for developing lung cancer.
Aim 3: To determine if thoracic aortic calcification are associated with prevalent lung cancer and the risk of developing lung cancer.
Approach Aim 3: We will objectively assess thoracic aortic calcification (TAC). Continuous measures will be utilized in Cox proportional hazard models to determine time to event. Secondary analysis will examine the relationship of TAC with prevalent lung cancers on the baseline scan.

Raul San Jose Estepar, Brigham and Women's Hospital
James C. Ross, Brigham and Women's Hospital
David S. Gierada, Mallinckrodt Institute of Radiology, Washington University School of Medicine