Objectively Measured Chronic Lung Injury on Chest CT

Abstract

Background: Tobacco smoke exposure is associated with emphysema and pulmonary fibrosis, both of which are irreversible. We have developed a new objective CT analysis tool that combines densitometry with machine learning to detect high attenuation changes in visually normal appearing lung (NormHA) that may precede these diseases.

Methods: We trained the classification tool by placing 34,528 training points in chest CT scans from 297 COPDGene participants. The tool was then used to classify lung tissue in 9,038 participants as normal, emphysema, fibrotic/interstitial, or NormHA. Associations between the quartile of NormHA and plasma-based biomarkers, clinical severity, and mortality were evaluated using Jonckheere-Terpstra, pairwise Wilcoxon rank-sum tests, and multivariable linear and Cox regression.

Results: A higher percentage of lung occupied by NormHA was associated with higher C-reactive protein and intercellular adhesion molecule 1 (P for trend for both < .001). In analyses adjusted for multiple covariates, including high and low attenuation area, compared with those in the lowest quartile of NormHA, those in the highest quartile had a 6.50 absolute percent lower percent predicted lower FEV1 (P < .001), an 8.48 absolute percent lower percent predicted forced expiratory volume, a 10.78-meter shorter 6-min walk distance (P = .011), and a 56% higher risk of death (P = .003). These findings were present even in those individuals without visually defined interstitial lung abnormalities.

Conclusions: A new class of NormHA on CT may represent a unique tissue class associated with adverse outcomes, independent of emphysema and fibrosis.

Trial registration: ClinicalTrials.gov NCT00608764.

Keywords: computed tomography; emphysema; lung injury; pulmonary fibrosis.

Copyright © 2019 American College of Chest Physicians. Published by Elsevier Inc. All rights reserved.

Figures

Figure 1

Distribution of mean density of training points placed in visually normal parenchyma by smoking group. Each of the three distributions represents the probability that a “visually normal” training point has a given attenuation as measured in Hounsfield units.

Figure 2

Local density histogram distributions for all subtypes and tissue subtypes (classes). Note that the noise evident in the panlobular emphysema histogram and its histogram of distribution are due to the paucity of training data and the very low attenuation of those training points for that subtype.

Figure 3

CONSORT diagram detailing the available data. (1) Gray shading represents the subgroups used for the analyses performed at the level of the subject. (2) The mortality and biomarker subgroups overlap. (3) The italic/bold typeface indicates the number of patients in which training points were placed (these individuals were not excluded from the clinical analyses). (4) The classifier performance shown in Table 3 is the performance of the patch level classification relative to the training point label in these (italic/bold) individuals.

Figure 4

Sample classification results. Representative computed tomography images with parenchymal labels applied by the algorithm from individuals not among the training cases, including (A) a never-smoking normal with normal parenchyma, (B) a current smoker with visually normal but high attenuation parenchyma, and (C) a former smoker who has developed visual and objective evidence of interstitial changes.

Figure 5

Biomarkers by injured tissue quartile. Differences in C-reactive protein and intercellular adhesion molecule 1 by quartile of injured tissue. P value for trend analysis based on Jonkheere-Terpstra test. Pairwise comparisons based on Wilcoxon rank sum test using Bonferroni correction. ***P

Figure 6

Schema of proposed evolution of response to chronic injury. Proposed schema of the evolution of parenchymal changes in response to chronic injury such as cigarettes smoking and the sensitivity of imaging and clinical tools for their detection., ,