Alison Van Dyke

MD, PhD

National Cancer Institute

Postdoctoral Fellow

NLST (Learn more about this study)

NLST-234

Aug 15, 2016

Lung Cancer and Lung Scar Carcinoma following Identification of Lung Scarring on Screening Low Dose Helical Computed Tomography (CT) Scan in the National Lung Screening Trial (NLST)

Initially described in 1939, lung scar carcinomas are thought to be the result of inflammation secondary to chronic infection, injury, and/or interstitial lung disease (Bobba et al., 2011). Cancer can also elicit a desmoplastic scarring response. Consequently, when a lung cancer is diagnosed in the context of pulmonary scarring, the relative timing of scar formation and cancer development cannot be ascertained. Prior work in the Prostate, Lung, Colorectal, and Ovarian Screening Trial indicates that the prevalence of lung scarring on baseline chest x-ray ranges 7.5%-17% (Pinsky et al., 2007; Pinsky et al., 2006; Yu et al., 2008). Furthermore, risk of lung cancer was increased among participants with scarring present on baseline chest x-ray [hazard ratio (95% CI): 1.5 (1.2-1.8)] (Yu et al., 2008). Follow-up studies using low-dose spiral computed tomography (LDCT) vs. chest x-ray revealed 28.5% of participants in the Lung Screening Study had scarring/pulmonary fibrosis at baseline screening on LDCT vs. 12.5% identified by chest x-ray (Pinsky et al., 2007), suggesting that LDCT is more sensitive in detecting lung scarring than chest x-rays. Whether associations between lung scarring and lung cancer risk can be replicated using LDCT data remains unknown.

We hypothesize that, among participants in the LDCT arm of the National Lung Screening Trial (NLST), lung scarring identified on baseline will be associated with a greater risk of lung cancer and of lung scar carcinoma. We also hypothesize that these relationships between lung cancer risk and lung scarring vary by histologic subtype and tumor size.

Objectives/Methods:
1. To assess the association of lung scarring at baseline (present/absent on the T0 CT scan) and lung cancer in different periods after baseline using logistic regression for the first 2 years (considered as prevalent cancers) and during the subsequent follow-up period using Cox regression analysis. So doing will allow us to examine temporality and possibility of reverse causality. For these analyses, we will request already available data through the LDCT arm of NLST from both the LSS and ACRIN.

2. To examine the relationships between lung scarring at baseline (present/absent) and the presence of incident lung scar carcinomas on follow up in the LDCT arm of NLST, we will request all of the digital H&E slide images from diagnostic biopsy and resection specimens available from participants in the LDCT arm. A thoracic pathologist on this project (Dr. Van Dyke) will evaluate these slides for the presence and amount of scarring (none, sparse, dense with size ≤10 mm, dense with diameter >10 mm) and for the histologic pattern using a modification of the Modified Scar Grade by Maeshima et al. (2002). During the histologic evaluation, the pathologist will be blinded to patient characteristics and LDCT imaging findings. Using this classification, we will assess whether baseline scarring is associated with each of the lung cancer histologic patterns (NSCLC vs. SCLC and adenocarcinoma vs. squamous cell carcinoma vs. neuroendocrine carcinoma) and by tumor size, using similar methods as described in #1.

The objectives of this project are three-fold:

1. To examine associations between lung scarring at baseline in the low dose computed tomography (LDCT) scan arm of NLST.

2. To evaluate the relationships between lung scarring at baseline and the presence of incident lung scar carcinomas on follow up in the LDCT arm of NLST.

3. To investigate whether these relationships between lung scarring and lung cancer vary by histologic type or tumor size.