Skip to Main Content

An official website of the United States government

Blood-based biomarkers as a complement to CT screening for lung cancer

Principal Investigator
Name
Edwin Ostrin
Degrees
MD, PhD
Institution
University of Texas MD Anderson Cancer Center
Position Title
Assistant Professor
Email
About this CDAS Project
Study
NLST (Learn more about this study)
Project ID
NLST-698
Initial CDAS Request Approval
Jul 29, 2020
Title
Blood-based biomarkers as a complement to CT screening for lung cancer
Summary
Only 24% of lung cancer is diagnosed as localized disease, when it is most amenable to treatment. The frequency of diagnosis at late stage is a major reason why lung cancer is the leading cause of cancer death in the United States. Screening for lung cancer by annual low-dose CT scan (LDCT), as detailed in the 2011 National Lung Screening Trial (NLST), reduces mortality by 20% but is limited by high rates of false positivity and to date, relatively low uptake. Approximately 24% of patients enrolled in screening had a nodule detected by LDCT, of which 96% proved not to be cancer. Much of the improved survival came through the detection of incident cancers, detected at initial or follow-up LDCT at years 1 and 2, which tended to be earlier stage. However, a significant fraction of these 649 incident cancers (18%) presented with regional or distant spread. Moreover, 70% of 44 cancers that presented after a negative LDCT, termed interval cancers, presented at late stage. In previous work, we have demonstrated that a 4-protein biomarker panel (4MP), measured by immunoassay on plasma, improved a smoking-based risk model of cancer. The same panel also improves a nodule-sized based risk model in indeterminate nodules. In this proposal, we will test whether this panel is of utility in a LDCT-based cancer screening setting to identify subjects at increased risk and thus worthy of more intensive follow-up. This proposal will assess whether this 4MP performs in a LDCT-based screening setting.

We hypothesize that the 4MP will help to refine risk in LDCT-based lung cancer screening. We will test if it can identify subjects at higher risk for development of incident and interval cancers, as well as evaluate if it improves clinical models of risk for indeterminate pulmonary nodules detected. We also hypothesize that the 4MP will identify subjects with interval and incident small cell lung cancers. Screening subjects diagnosed with this cancer have not shown mortality benefit through LDCT-based screening, and thus performance of the 4MP in these patients may lead to improved survival. We propose the following two aims: 1) To test the positive and negative predictive value of the 4MP among subjects with incident or interval lung cancer and controls. 2) To test the positive and negative predictive value of the 4MP among subjects with indeterminate nodules.

Peripheral plasmas collected during the NLST trial will be obtained from the ACRIN biorepository. Prediagnostic and diagnostic plasmas for interval and incident cancers will be assessed using our validated laboratory protocols and compared to NLST controls in a 1:4 case to control cohort model. Diagnostic likelihood ratios will be calculated to evaluate whether the 4MP can identify patients who would benefit from more frequent imaging or to more intensive work-up. For the second Aim, these scores will be incorporated into clinical risk scores, for instance the Lung-RADS score or the Brock University model for nodule risk.
Aims

Aim 1: To test the positive and negative predictive value of the 4MP among subjects with incident or interval lung cancer and controls. Rationale: Over the 3-year period of the NLST trial, there were a total of 44 interval cancer cases and 649 screen-detected incidence cancer cases. Of the 649 screen-detected incidence cancer cases, 168 and 211 (n=379 total) were detected at the 12- and 24-month follow-up screening intervals, respectively. Of the NLST incident cancers, 18% presented with advanced stage disease whereas 70% of interval cancers presented with stage III or IV.

Study Design: Plasmas are available for approximately 60 T1-incident cancers and 80 T2-incident cancers. These include interval cancers. The 4MP will be measured using our validated multiplex immunobead assay on one-year prediagnostic plasmas and cancer-free controls using a 4:1 control-to-case design. Cancer-free controls are defined as those individuals that were negative by CT and that did not develop lung cancer during the follow-up of the study; matching criteria will be based on age, race, smoking, and gender. We will evaluate the classifier performance (AUC) as well as the positive and negative predictive value of the 4MP for distinguishing lung cancer cases from controls. We will additionally evaluate the combination of our biomarkers with a the PLCOm2012 risk score based on subject characteristics.

Aim 2: To test the positive and negative predictive value of the 4MP among subjects with indeterminate nodules. Rationale: Over the three years of CT screening, approximately 24% of patients had a nodule detected by screening. The likelihood that a nodule is cancerous largely revolves around size. Current recommendation for smaller nodules is surveillance with additional imaging, such as PET/CT, or repeat CT, which may delay diagnosis. Additionally, it is more frequent for incident lung cancers to occur after a positive screen in the previous 12 months. Thus, a key area of interest is whether the 4MP can identify subjects with low risk based on CT findings (for instance Lung-RADS category 2) that may be at increased risk based on their marker findings.

Study Design: The 4MP will be measured in available plasma samples taken at the time of a positive CT for cancer-free controls and at all positive CTs for those diagnosed with cancer (including pre-diagnostic and diagnostic time points). As above, we will use a 4:1 control-to-case design. Approximately 96 plasmas are available for T0-detected cancers, 60 for T1-detected cancers, and 71 for T2-detected cancers. We will evaluate the classifier performance (AUC) as well as the positive and negative predictive value of the 4MP for distinguishing lung cancer cases from controls. We will evaluate the combination of our biomarkers with the reported Lung-RADS score and a clinical risk score aimed at estimating risk of cancer for indeterminate pulmonary nodules, for instance the Brock University calculator, taking into account nodule size, location, speculation, and texture. The combined risk scores will be used to construct a model to identify high-risk individuals who may benefit from more closer follow-up.

Collaborators

Sam Hanash -- University of Texas MD Anderson Department of Clinical Cancer Prevention
Johannes Fahrmann -- University of Texas MD Anderson Department of Clinical Cancer Prevention
James Long -- University of Texas MD Anderson Department of Biostatistics