Hee-Soon Juon

PhD

Thomas Jefferson University

Professor

NLST (Learn more about this study)

NLST-361

Oct 16, 2017

Assessing Racial Disparities of Lung Cancer Risk Prediction Model

Lung Cancer is the number one cause of cancer-related death in the U.S. and many other countries. In 2015, there were an estimated 221,200 cases of lung cancer and 158,040 deaths from the disease in the U.S. Over the past 20 years, lung cancer incidence and mortality rates have decreased for all races; however, African Americans have not experienced dramatic decline compared to White Americans. Although detecting lung cancer at early stage gives additional 10-year survival rate, African Americans were diagnosed with distant stage of lung cancer. Yet, there have been few studies of racial differences in lung cancer screening and risk factor assessments of lung cancer.
Lung cancer detecting is achieved from lung cancer screening recommended by U.S. Preventative Service Task Force (USPSTF) using low-dose computed tomography (LDCT) in adults aged 55-74 years who have 30 pack-year smoking history and currently smoking or quit less than 15 years. In the National Lung Screening Trial (NLST), 53,454 current or former heavy smokers were randomized to receive LDCT or chest X-ray for three annual screens. Nevertheless, lung cancer screening rates were low in African Americans due to lack of knowledge, stigma, and access to care.
We want to apply NLST to Tammemagi risk prediction tools to calculate for each individual who were screened using LDCT. Then risk prediction scores of African American and White Americans will be compared to see racial disparities based on risk prediction. In addition, we will assign Lung-RADS, a grading system of the nodules developed by American College of Radiology, to individuals who screened using LDCT to examine racial disparities.

Last, we will use our cohort and compare it to calculate values of NLST data to see difference between Philadelphia confined data and the regional data on NLST to see if the distribution of the cohort is similar or different. Thomas Jefferson University Hospital has built a cohort of 733 individuals referred for the lung cancer screening during the period of May 2013 to July 2017. About 71% (n=521) participated in lung cancer screening. Lung-RADS are available in our cohort data. While NLST data had only 4% of African Americans, our cohort was mainly confined to Philadelphia with ~40% participants being African Americans.

The primary objective of the study is to examine racial disparities of lung cancer risk prediction model and lung nodule malignancy (e.g., Lung-RADS). Using participant- and nodule-level data, the study is designed to compare Tammemagi lung cancer risk prediction and nodule malignancy prediction between Whites and African Americans. Tammemagi risk prediction tools include a comprehensive set of risk factors including age, socioeconomic status, BMI, family history of lung cancer, COPD, personal history of cancer, race, smoking status, pack-years smoked, and smoking duration. Lung-RADS is a grading system of the nodules to have standardized reports of positive criteria for Lung-RADS which categories 1 to 2 constitute negative screening results, and categories 3 to 4 constitute positive results.

Aim 1: Examine the racial disparities of lung cancer prediction models between African Americans and Whites after constructing Tammermagi risk production tool
Aim 2: Examine the racial disparities of lung nodule malignancy between African Americans and Whites after constructing Lung-RADS
Aim 3: Assess if the findings of NLST are comparable to the population of Philadelphia