The Lung EArly Proteins project: A LEAP toward implementing biomarkers in lung cancer screening

The Lung EArly Proteins (LEAP) Project Lung cancer is the most common cause of cancer death worldwide. Screening by low-dose CT can reduce lung cancer mortality among current and former smokers, but the balance of benefits and harms could be improved by leveraging novel tools to optimize decision making. Our objective is to translate a novel panel of protein biomarkers to optimize the decision (1) to initiate screening and (2) to biopsy a nodule. We developed the INTEGRAL panel within the Integrative Analysis of Lung Cancer Risk and Etiology U19 project. We identified proteins to include on the panel by analyzing pre-diagnostic samples collected up to 3 years before diagnosis among over 700 lung cancer case-control pairs in prospective cohort studies. The proteins improved the AUC of a smoking-based model by 0.15, with improvements across histological types. The Lung EArly Proteins (LEAP) project will carry out two late-stage translational studies to evaluate the potential to implement this panel in two key contexts. Aim 1 will determine whether repeated measurements of protein markers over time can better predict development of lung cancer than a single measurement. A repeat-measures model for lung cancer will be developed using blood samples (up to 5 per participant) from 546 lung cancer cases and 546 controls in the Prostate, Lung, and Ovarian Cancer Screening Trial, and independently validated using samples from 539 cases and 539 controls from the Carotene and Retinol Efficacy Trial. We hypothesize that the AUC of the model with repeat measurements will be higher than for a model with a single measurement. Aim 2 will determine whether the protein markers can identify cancers among high-risk screen-detected lung nodules. A biomarker-based prediction model for nodule malignancy will be developed using blood samples from 685 participants (131 cancers) in the Pittsburgh Lung Screening Study, and then validated among 830 participants (213 cancers) in the Multicentric Italian Lung Detection (bioMILD) trial and the St Elizabeth Lung Cancer Screening program. We hypothesize that the AUC of the biomarker model will be higher than for an established, validated nodule malignancy prediction model (the Brock model). The LEAP project will clearly define the contexts in which the INTEGRAL panel is clinical useful. Our mission is to reduce lung cancer mortality by optimizing a proven and effective screening intervention, using novel tools to extend its benefits to all high-risk individuals and reduce screening harms.