| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Denver, CO
Dr. Lynch is Professor, Division of Radiology, National Jewish Medical and Research Center.
Correspondence to: David A. Lynch, MD, Professor, Division of Radiology, National Jewish Medical and Research Center, 1400 Jackson St, Denver, CO 80206; e-mail: lynchd{at}njc.org
In fibrotic lung diseases, there is increasing evidence that the extent of lung fibrosis on CT is an important predictor of prognosis. For example, in a multicenter study1 of patients with idiopathic pulmonary fibrosis enrolled in a treatment trial, the CT extent of lung disease, evaluated visually using a semiquantitative scoring system, was the strongest independent predictor of survival (p < 0.0001). Similarly, in the control arm of the recently published Scleroderma Lung Study,2 the CT extent of lung fibrosis at baseline was a strong predictor of FVC at 12 months (adjusted for baseline FVC) [p = 0.006]. In both of these studies, a semiquantitative visual scoring system was used to evaluate the extent of disease on CT. However, semiquantitative scoring systems are limited by the requirement for expert radiologists, and by moderate interobserver variation.345 CT has been used to accurately quantify emphysema for almost 20 years,67 but the development of quantitative CT-based measures for lung fibrosis has been more challenging. In the current exciting era of clinical trials for idiopathic pulmonary fibrosis and other fibrotic lung diseases, quantitative CT would potentially offer an objective, reproducible measure of disease extent in these conditions.
The CT attenuation histogram may be used to quantify lung fibrosis. In the lung parenchyma, CT attenuation, measured in Hounsfield units, is determined by the relative amounts of air, soft tissue, and blood in each volume element (voxel). The attenuation values for each pixel can be expressed as a histogram. The CT attenuation histogram of normal lung deviates from the Gaussian normal distribution in that it is sharply peaked at approximately – 800 Hounsfield units, and is markedly skewed to the left. Since lung fibrosis or inflammation cause an increase in the amount of soft tissue in the lung, it will increase mean lung attenuation, and will decrease the sharpness of the histogram peak (kurtosis), and the degree of leftward skewness of the curve. Mean lung attenuation, skewness, and kurtosis can therefore be used as measures of the extent of lung fibrosis. In a study of 24 subjects with idiopathic pulmonary fibrosis and 60 subjects with asbestosis, Hartley et al8 found that the mean and median lung attenuation derived from high-resolution CT images were independently associated with the presence of moderate-to-severe dyspnea, a higher profusion of chest radiograph abnormalities, a lower FVC, and abnormalities on BAL. More recently, a study9 of 144 patients with idiopathic pulmonary fibrosis showed that measures of mean lung attenuation, skewness, and kurtosis derived from the CT density histogram correlate with evidence of physiologic impairment.
The article by Camiciottoli et al5 in this issue of CHEST (see page 672) offers a further assessment of CT histogram-based measures (mean lung attenuation, skewness, and kurtosis) in 48 patients with systemic sclerosis (33 with limited disease, and 15 with diffuse disease), carefully characterized both by physiology and by quality of life scores. They found (not surprisingly) that the quantitative CT-based measures are much more reproducible than semiquantitative visual scores, with 
scores ranging from 0.89 to 0.97. On univariate analysis, the quantitative measures showed significant correlations with many components of the quality of life scores, and with 6-min walk test. On multivariate stepwise analysis, corrected for multiple comparisons, physiologic and quality of life variables accounted for 70% of the variance of mean lung attenuation, 78% of the variance of skewness, and 81% of the variance of kurtosis. However, physiologic values accounted for only 40% of the variance of the visual score. These results provide further evidence that quantitative CT measures offer a robust and valid measure of the severity of lung parenchymal abnormality in patients with lung fibrosis.
CT histogram-based measures have two potential limitations. First, they depend on the level of inspiration achieved for the scan. Ideally the inspired lung volume would be standardized by spirometry,101112 but spirometric control of CT is cumbersome, not widely used, and probably unnecessary,9 particularly since spiral CT can now be used to directly measure lung volumes.13 Secondly, these measures (like pulmonary physiologic tests) provide a global measure for normal and abnormal lung, and do not directly measure the type or extent of abnormality. Texture-based measures such as the adaptive multiple feature method may be able to discriminate the type of parenchymal abnormality,14 and may indeed identify parenchymal abnormality in areas of lung that appear visually normal.15 These texture-based methods may soon allow us to reproducibly separate abnormal from normal lung, providing an objective index of disease pattern and extent. Once validated, these techniques will likely be valuable for evaluating the effect of treatment in fibrotic lung diseases.
Footnotes
The author has no conflict of interest to disclose.
References
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
