2010 RSNA Abstracts

Quantitative CT of the lungs and airways in healthy non-smoking adults: relationship to age, gender, and body mass index: for the COPDGene® study.

Zach, Murphy, Wilson, Newell, Schroeder, Hoffman, Lynch

Purpose
The purpose of this study is to evaluate whether quantitative CT (QCT) measures of emphysema, gas trapping and airway wall thickening vary with age, gender, or body mass index (BMI) in healthy non-smoking adults, aged between 45 and 80.
Methods
As part of the COPDGene^TM study, 99 healthy non-smokers (32 male, 67 female, mean age 62.7 ± s.d. 8.9, mean BMI 28 ± s.d. 5.0) underwent volumetric CT at full inspiration and at the end of a normal expiration. On QCT analysis (Pulmonary Workstation, VIDA Diagnostics), emphysema-like changes were defined as lung tissue with attenuation values ≤ -950 Hounsfield Units (HU) on inspiratory CT. Gas trapping was defined as lung tissue ≤ -856 HU on expiratory CT. Multivariate analysis was used to determine the relationships between age, gender and body mass index (BMI), and QCT parameters.
Results
Mean ( ± s.d.) % emphysema-like changes was 2.0 ± 2.6, and mean % gas trapping was 10.8±9.8. Mean % emphysema-like changes was 3.3 ± 3.3 in men, compared with 1.4 ± 2.1 in women (p=0.002). The % emphysema-like changes decreased slightly with age (p=0.04) and did not change significantly with higher BMI. % gas trapping did not show any relationship with age, gender or BMI.
Segmental airway luminal diameter was greater in men than in women, 5.95 ± 0.79mm vs 5.09 ± 0.49mm (p<0.0001). Segmental airway wall thickness was also significantly greater in men than in women, 1.62 ± 0.17mm vs. 1.46 ± 0.18mm respectively (p<0.0001). Segmental airway measurements did not show a relationship with age or BMI. On multivariate analysis, the relationships between % emphysema-like changes and age and gender, and between segmental wall parameters and gender remained statistically significant.
Conclusions
In normal older adults, men have higher % emphysema-like changes and have larger luminal airway diameters and airway wall thickness as compared with women. In contrast to previous studies, greater age was associated with slightly lower emphysema extent, and % gas trapping appears to be relatively independent of age, gender and BMI.

Visual CT subtypes of COPD. Preliminary observations from the COPDGene Trial.

Presented on behalf of the COPDGene Qualitative CT Workshop Participants

Authors: Lynch, Jacobson, Murphy, Wilson, Newell, Grenier, Kauczor, Crapo.

Purpose: The COPDGene^® research program convened a qualitative scoring workshop with the goal of defining key visual CT characteristics that may define specific COPD subtypes.
Material and Methods: 395 CT scans were reviewed by 58 radiologists and pulmonologists, to identify qualitative features that can be used to define CT phenosubtypes of COPD. Consensus standards to be used for evaluation were developed using expert presentations and group discussion. Each observer was asked to study 80 cases, resulting in 9-11 readings for each case. Data recorded included emphysema presence, type (paraseptal, bulla, centrilobular, panlobular), zonal distribution and severity, bronchial wall thickening, cylindrical bronchial dilatation, bronchiectasis, centrilobular nodules, mosaic attenuation, gas trapping, tracheobronchial disease, ground-glass, honeycombing and dominant CT phenotype (normal, large airway, small airway, emphysema, and mixed).
Results: No significant differences in agreement scores were found between radiologists and pulmonologists. Emphysema globally and by specific type had the highest inter-observer agreement (Kappa .43-.67). A 2-way, unsupervised cluster analysis was run which identified 4 disease clusters. The clusters and the variables associated with them were (1) tracheal disease with saber sheath, tracheobronchomalacia, airway outpouching and presence of mucoid material in central airways; (2) small airway disease with mosaic attenuation, ground-glass, honeycombing and centrilobular opacities (3) Centrilobular and panlobular emphysema with gas trapping, and airway wall thickening and (4) bronchial disease with cylindrical bronchial dilatation and bronchiectasis. Visual assessment of airway wall thickening appeared to be a better logistic discriminator among disease groups than similar data obtained from the automated methods.
Conclusion: Qualitative evaluation of CT features appears to identify specific COPD subtypes.

Clinical relevance: Visual subtypes derived from CT images may facilitate genetic analysis and individualized therapy

Quantitative CT measurements of emphysema and air trapping correlate with physiologic airway obstruction in smokers with chronic obstructive pulmonary disease (COPD): the COPDGene® Study

J D Schroeder, MD, Boulder, CO; J D Newell, MD; J Zach; J R Murphy, PHD; C G Wilson; E A Hoffman, PhD; et al. 

PURPOSE
In individuals with COPD, quantitative CT (QCT) has the potential to quantify emphysema and air trapping as distinct components of COPD which may be important for phenotyping of the disease and for subsequently devising individualized treatment. The purpose of this study is to evaluate the relationship between QCT measurements of emphysema and gas trapping and physiologic measures of disease severity.

METHOD AND MATERIALS
QCT analysis (Pulmonary Workstation, VIDA Diagnostics) and spirometric evaluation were completed on 1759 subjects in the COPDGene Study (771 smoking controls without evidence of airway obstruction, 140 with GOLD Stage 1, 427 with GOLD Stage 2, 272 with GOLD Stage 3, and 149 with GOLD Stage 4). Subjects underwent volumetric CT at full inspiration (N=1759) and at the end of a normal expiration (N=1630). QCT analysis was performed using VIDA software. On QCT, % emphysema is defined as % lung voxels <= -950 Hounsfield units (HU) on inspiratory CT and % air trapping is defined as % lung voxels <= -856 HU on expiratory CT. Multivariate analysis was performed to evaluate the relationship between demographic, clinical and QCT variables and FEV1 % predicted and FEV1/FVC ratio.

RESULTS
Univariate R-squared regression values for the relationship between physiologic parameters and QCT for FEV1 % predicted are: 0.43 (% emphysema) and 0.61 (% air trapping). R-squared values for FEV1/FVC ratio are: 0.53 (% emphysema) and 0.71 (% air trapping.) P<0.0001 for all regression slopes. On multivariate analysis, significant predictors of lower FEV1 % predicted and FEV1/FVC ratio are: % air trapping on QCT (t ratios -42, -50), BMI (t ratios -8, -2), male gender (t ratios 2, 0), age at enrollment (t ratios 4, -0.8), and hours of supplemental oxygen use per day (t ratios -10, -8). These models account for 66% and 73% of variability in FEV1 % predicted and FEV1/FVC ratio, respectively. Of these ratios, % air trapping is the most powerful predictor based on its proportion of the total variance.

CONCLUSION
QCT is strongly associated with spirometric impairment in cigarette smokers. In particular, % air trapping, combined with clinical parameters, strongly correlates with physiologic measurements of airway obstruction.

CLINICAL RELEVANCE/APPLICATION
Quantitative CT measurements of gas trapping and emphysema describe distinct components of COPD which may be important for phenotyping the disease and devising individualized treatment.

Phantom for the Cross Characterization of Scanners in a Multi-center Quantitative CT Lung Study

 JP Sieren1, EA Hoffman1, KR Gunderson1, DA Lynch3, JD Newell3 & PF Judy2 

University of Iowa, Iowa City1, Brigham & Womens Hospital, Boston2, National Jewish Health, Denver3 

Purpose: Multi-center studies using quantitative CT to assess presence of emphysema have observed that tracheal air measurements are often greater than the expected value of -1000HU: this measurement varies not only between individuals but also between scanner models. Phantoms were developed to study this bias.

Methods: The COPDGene Study has developed two phantoms. The COPDGene 1 phantom contains air holes within a lung-equivalent foam. We have created a "COPDGene 2" phantom in which the central 3 cm air hole is surrounded by a 9 cm thick acrylic cylinder. In an alternative phantom (Univ. of Iowa modified Catphan500) an air hole was bored off center in a homogeneous water equivalent cylindrical slab. Outside air was obtained from slices at one end of these scans where only air inside the scanner gantry was imaged. The performance of these three phantoms was compared across four models of scanners: Siemens Definition Flash (DF), AS+, AS and the Definition (D) Scan protocol was: 0.75mm thick, pitch 1, CTDIvol 15mGy, B35 kernel.

Results: The data for all four scanners using the B35 kernel are summarized in the table. The newly modified COPDGene 2 phantom with the thicker ring around the inside air showed a positive shift in values that were similar to the modified Catphan500, although COPDGene2 showed increased noise. COPDGene1 Phantom is least effective in regards to simulating in vivo tracheal air measurements. COPDGene2 and Catphan500 show air to be more negative than seen in vivo human tracheal regions. Human tracheal air (mean) ranged from -985 to -955HU in 761 COPDGene human subjects randomly sampled from seven of the most used scanners in the study.

Conclusion: Measurements of air attenuation within the COPDGene2 and modified Catphan500 phantoms show evidence of positive shift. However, these phantoms do not fully reproduce the degree of attenuation shift found within the trachea. The full explanation for why tracheal air is more positive than seen in any of these phantoms remains a topic of investigation.