Abstract
Relative area of emphysema below -910 Hounsfield units (RA-910) and 15th percentile density (PD15) are quantitative computed tomography (CT) parameters used in the diagnosis of emphysema. New concepts for noninvasive diagnosis of emphysema are aerosol-derived airway morphometry, which measures effective airspace dimensions (EAD) and aerosol bolus dispersion (ABD).
Quantitative CT, ABD and EAD were compared in 20 smokers with chronic obstructive pulmonary disease (COPD) and 22 patients with α1-antitrypsin deficiency (AAD) with a similar degree of airway obstruction and reduced diffusion capacity.
In both groups, there was a significant correlation between RA-910 and PD15 and pulmonary function tests (PFTs). A significant correlation was also found between EAD, RA-910 and PD15 in the study population as a whole. Upon separation into two groups, the significance disappeared for the smokers with COPD and strengthened for those with AAD, where EAD correlated significantly with RA-910 and PD15. ABD was similar in the two groups and did not correlate with PFT and quantitative CT in either group.
In conclusion, based on quantitative computed tomography and aerosol-derived airway morphometry, emphysema was significantly more severe in patients with α1-antitrypsin deficiency compared with patients with usual emphysema, despite similar measures of pulmonary function tests.
Pulmonary emphysema is defined in pathological terms 1, and lung biopsy is considered the gold standard to establish an accurate diagnosis. However, in clinical practice, lung biopsy is rarely used on this indication, because of the relatively high risk of the procedure and the availability of noninvasive methods of investigation that help to establish the diagnosis. A widely used method is the measurement of the forced expiratory volume in one second (FEV1), which has the advantage of being inexpensive and easy to obtain; however, FEV1 has low sensitivity 2, 3 and poor correlation to the degree of tissue destruction 4. Accurate assessment of the severity of emphysema and identification of possible subtypes are pivotal in understanding the natural history of this complex disease. This will provide an opportunity to monitor the therapeutic effects of available and future remedies. In addition, a precise noninvasive diagnosis of emphysema at an early stage may further motivate patients to quit smoking; nevertheless, counselling to assist smoking cessation should be given, even in the absence of clinical, physiological and radiological features of emphysema.
Computed tomography (CT), and particularly high-resolution CT, reveals areas of low attenuation, which are the radiological equivalent of tissue loss seen in pathology specimens 5, 6. CT is highly specific and sensitive in diagnosing emphysema 7, and in vivo studies have shown a correlation with pathology ranging 0.7–0.9 7–9. Due to their digital nature, CT images can be objectively evaluated by computer analysis to assess the presence and severity of emphysema. Commonly applied CT-derived measures of emphysema are the relative area of emphysema below -910 Hounsfield units (RA-910) 10 and the 15th percentile density (PD15) 11, which are derived from the histogram of pixel attenuation values obtained from image analysis.
Other novel concepts suggested for noninvasive diagnosis of emphysema are aerosol-derived airway morphometry (ADAM), which measures effective peripheral airspace dimensions (EAD) 12, and aerosol bolus dispersion (ABD), which is a measure for convective gas mixing in the lungs 13. Studies have shown that both parameters are significantly increased in patients with emphysema 14–16, reflecting enlargement of the distal airspaces and ventilation inhomogeneities. The purpose of this study was to compare quantitative CT, pulmonary function tests (PFTs), ABD and EAD in patients with chronic obstructive pulmonary disease (COPD), and assess the differences in these parameters between patients with usual COPD and patients with α1-antitrypsin deficiency (AAD).
MATERIALS AND METHODS
Patients
This was a prospective study in which 43 patients (>50 yrs old) with COPD were initially included. One patient was subsequently excluded because CT showed multiple pulmonary metastases from a primary colon cancer. Of the remaining 42 patients, 20 patients (six males and 14 females) had usual COPD, and 22 (10 males and 12 females) had AAD of PI*ZZ phenotype verified by isoelectric focusing. COPD in both groups was defined as FEV1/forced vital capacity (FVC) <70% and FEV1 <80% predicted post-bronchodilator 17. All patients with usual COPD were current smokers with a smoking history of >20 pack-yrs. No ex-smokers with usual COPD were included in the study. Two patients with AAD were life-long nonsmokers and 20 patients were ex-smokers for ≥6 months before the study start. No active smokers with AAD were included. All 42 patients were free of any concomitant severe pulmonary or extrapulmonary disease. The study was approved by the Ethics Committee of the County of Copenhagen, Denmark, and all participants provided informed written consent. Patient characteristics are shown in table 1⇓.
Characteristics of patients, and results of pulmonary function, quantitative computed tomography, aerosol-derived airway morphometry and aerosol bolus dispersion (ABD) tests
Pulmonary function tests
Each patient underwent PFTs prior to ABD studies and CT of the lungs. PFTs were performed according to European Respiratory Society recommendations 18, 19. Patients visited the respiratory laboratory in the morning (Dept of Respiratory Medicine, Gentofte University Hospital, Hellerup, Denmark). A pressure-compensated flow plethysmography (Vmax 229; SensorMedics, Bilthoven, The Netherlands) was applied. First, three maximal forced expirations were recorded, and patients who demonstrated baseline FEV1 values <80% predicted underwent a reversibility test 15 min after the inhalation of 1 mg of terbutaline. Reversibility ≥15% and ≥200 mL resulted in exclusion from the study. Patients with irreversible airflow obstruction then underwent measurement of static lung volumes (i.e. total lung capacity (TLC) and residual volume (RV)). Carbon monoxide transfer coefficient (KCO) was measured by the single-breath technique; alveolar volume (VA) was obtained by the dilution of methane during the single-breath manoeuvre. The carbon monoxide diffusion capacity (DL,CO) was calculated as the product of KCO and VA. Predicted values for PFT parameters were calculated according to the reference values proposed by the European Community for Steel and Coal 18, 19. Results are shown in table 1⇑.
Computed tomography
CT scans were performed as low-dose multi-slice CT using the LightSpeed QX/i (GE Medical Systems, Milwaukee, WI, USA). Volume scans of the entire lung were acquired and no contrast medium was used. Image acquisition was performed with 5-mm collimation, rotation time 0.8 s, 30-mm table feed per rotation (pitchx = 1.5), corresponding to a total scan time of 10–12 s. The voltage across the X-ray tube was 140 kV and the tube current was 40 mA. The field of view was 40 cm and the matrix size was 512×512. Patients were scanned at suspended full inspiration in the supine position in a caudal–cranial direction (z-axis) to avoid breathing artefacts at the level of the diaphragm. Images were reconstructed using low spatial-resolution algorithm (soft).
CT images were analysed using Pulmo-CMS (MEDIS Medical Imaging Systems, Leiden, The Netherlands). The program automatically detects the lung contours in contiguous (or overlapping) images using a region-growing algorithm to separate lung tissue from the thoracic wall and mediastinum at a threshold of -380 HU. The lung contours are then checked in each slice to ensure the correctness of the procedure and to edit the contours, if necessary. Afterwards, a frequency distribution of the pixel attenuation values of the total lung is generated. From this histogram, the total lung volume, the volume of air, RA-910 and PD15 are extracted. RA-910 (also called the emphysema index) relies on the “density mask” concept, first described by Muller et al. 10, and is defined as the percentage of voxels with attenuation values below -910 HU. The density mask provides an overall percentage of lung involvement by emphysema. PD15 is also extracted from the histogram of pixel attenuation values and is defined as the density value in HU at which 15% of the voxels have lower densities. By adding 1,000, the density values can be converted into g·L-1 units (e.g. PD15 of -950 HU corresponds to 50 g·L-1, i.e. 15% of the voxels have a density <50 g·L-1).
Aerosol morphometry
Monodisperse aerosol particles uniformly settle in calm air with a constant velocity (vs). The motion of these particles, suspended within intrapulmonary airways or airspaces during breath holding, leads to particle deposition on airway or airspace surfaces. The particle-loss rate is larger for particles suspended within small airspaces than for particles located within large airspaces. The decline in particle number with time can be used to evaluate peripheral airspace dimensions 12.
For this purpose, the lungs are filled by an inspiration of a monodisperse aerosol. The inspired aerosol volume can be considered as comprised of small volume elements, which penetrate longitudinally into different volumetric lung depths (Vp). During a breath-holding period (t), the particles settle onto airway surfaces. The reduction of the particle concentration in a defined volume element is measured in the expired air. The particle recovery (R) is defined as the ratio of particle concentration in an exhaled and inhaled volume element, and R decreases with increasing breath-holding time (t). For each lung depth, this relationship can be approximated by the exponential function:
EAD is calculated for each volumetric lung depth Vp by fitting an exponential function to the recovery values measured for each lung depth and the corresponding breath-holding times.
The evaluation of R(Vp) requires the measurement of the particle concentration in the volume element penetrating into lung depth Vp during inspiration and in the volume element exhaled after exhalation of the aerosol volume Vp. This is achieved by continuous monitoring of particle concentration with an on-line, open-flow system combining aerosol photometry and pneumotachography. In this study, EAD was calculated for a relative volumetric lung depth of 20% of the intrathoracic lung volume at breath hold (EAD20).
Aerosol bolus dispersion
Gas transport in the lungs is due to diffusion and convection. Since monodisperse aerosol particles with diameters 0.5–1 mm behave like a “nondiffusive gas”, they can be used as tracers for convective gas transport 13. Therefore, a small volume (bolus) of the inspired air is labelled with these particles. During inspiration, particles are convectively transported into air volumes, which are initially particle free. Therefore, in the exhaled air, the aerosol particles are distributed over a larger air volume than in the inhaled air and the bolus is dispersed.
The width of an exhaled aerosol bolus inhaled into a certain volumetric lung depth Vp can be quantified by its volumetric half width (H50), which is defined as the air volume in which the particle concentration exceeds half the maximum concentration. To account for the contribution of the width of the inhaled bolus (H50,i) to the width of the exhaled bolus (H50,e), a corrected half-width ABD is introduced, which is given by:
In this study, ABD was calculated for boluses inhaled into a volumetric lung depth of 600 cm3 (ABD600).
Particle generation and classification
Monodisperse di-2-ethylhexyl sebacate (DEHS) droplets were produced by heterogeneous nucleation of DEHS vapour on NaCl nuclei in nitrogen. The aerosol was then diluted with particle-free air to achieve a particle concentration of 2×104 cm-3. The terminal settling velocity (vs) of the droplets was measured in a convection-free sedimentation cell. Particle size throughout the study was ∼0.9 µm.
Statistical analysis
Pearson’s correlation was performed to evaluate the relationship between PFTs, CT parameters and aerosol parameters. The unpaired t-test was used to investigate the statistical significance of differences between groups. The requested level of significance was 0.05.
RESULTS
Patient characteristics, pulmonary function tests and quantitative computed tomography
The mean age of the patients was 63 yrs (range 51–78 yrs). Patients with AAD were generally younger than patients with usual COPD. Patient characteristics are shown in table 1⇑. The degree of airway obstruction (FEV1) and reduction in DL,CO in the two groups was similar, whether expressed in absolute values or % pred, calculated according to European equations 18, 19. TLC and RV were significantly larger in patients with AAD, both in absolute and % pred values. In addition, lung density, as measured by either PD15 or RA-910, was significantly different between the two groups (table 1⇑; fig. 1⇓).
Box-plots of the 15th percentile density (PD15) and relative area below -910 HU (RA-910) in patients with α1-antitrypsin deficiency (AAD) and patients with usual chronic obstructive pulmonary disease (COPD). The horizontal line represents the median. •: outliers, i.e. observations below the 10th percentile and above the 90th percentile.
Results of aerosol morphometry
Measurement of EAD at a relative lung depth (Vpr) of 20% was feasible in 16 patients with AAD and 15 patients with usual COPD. EAD20, which indicates the presence of emphysema, was significantly higher in patients with AAD (mean±sd 0.95±0.38 mm) than patients with usual COPD (0.64±0.31 mm; p = 0.017; fig. 2⇓). In addition, EAD20 in both groups was significantly higher than values reported in healthy individuals (0.34±0.05 mm; p<0.001) 12.
Box-plots of the effective airway diameter for a relative volumetric lung depth of 20% of the intrathoracic lung volume at breath hold (EAD20) and aerosol bolus dispersion for boluses inhaled into a volumetric lung depth of 600 cm3 (ABD600) in patients with α1-antitrypsin deficiency (AAD) and patients with usual chronic obstructive pulmonary disease (COPD). The horizontal line represents the median. •: outliers, i.e. observations below the 10th percentile and above the 90th percentile.
Measurement of ABD600 was obtained from all patients with AAD and 17 patients with usual COPD. ABD600 was elevated in all patients with AAD (634±135 cm3) and all patients with usual COPD (602±153 cm3). In both groups, ABD600 was significantly higher than previously reported normal values in healthy individuals (346±53 cm3; p<0.001) 13. However, there was no statistically significant difference in ABD600 between the two patient groups (p = 0.49; table 1⇑; fig. 2⇑).
Correlation between pulmonary function tests and computed tomography parameters
When the whole group was considered, there was highly significant correlation between PD15 and RA-910 on one side and measures of airway obstruction, hyperinflation and the diffusion capacity on the other. When each group was analysed separately, the same significant correlation was found, except for the correlation between RA-910 and DL,CO % pred in patients with usual COPD, which did not reach statistical significance (r = -0.44; p = 0.05). Correlation coefficients are shown in table 2⇓.
Correlation between pulmonary function test(PFT), computed tomography parameters and aerosol parameters for the whole patient group and for each subgroup
Correlation between pulmonary function test and aerosol morphometry
Results of the correlation analysis between EAD20 and ABD600 and PFTs in both groups are shown in table 2⇑. In brief, EAD20 showed significant negative correlation with the diffusion capacity, both in patients with AAD and usual COPD. In both groups, no correlation was found between EAD20 and measures of airway obstruction and hyperinflation; however, EAD20 correlated significantly with FEV1 and FEV1/FVC, when all patients were considered. ABD600 in both groups did not correlate with any measurement of PFT.
Correlation between computed tomography parameters and aerosol morphometry
When all 42 patients were considered, a significant correlation was found between EAD20 and PD15 (r = -0.67; p<0.001) and RA-910 (r = 0.64; p<0.001). Upon dividing the study population into two groups, the significance disappeared for the patients with usual COPD, and strengthened for the patients with AAD, where EAD20 showed a significant negative correlation with PD15 (r = -0.80; p<0.001) and a significant positive correlation with RA-910 (r = 0.71; p = 0.002). ABD600 was similar in both groups, and showed no significant correlation with PD15, RA-910 or EAD20 in either group. Results of the correlation analysis are shown in table 3⇓.
Correlation between computed tomography parameters and aerosol parameters for the whole patient group and for each subgroup
DISCUSSION
Expiratory airflow limitation is the hallmark physiological change of COPD. However, the pathological changes of the disease precede airflow limitation by a number of years, and measures of airflow limitation correlate relatively poorly with the severity of emphysema and are insensitive in mild emphysema 4. Studies have shown that mild-to-moderate emphysema can be present without reduction in FEV1 2, 3, 20. Conversely, diffusion capacity has better correlation to the pathological extent of emphysema 4, but the method has relatively large variability and is unspecific for emphysema. Gurney 21 put forward that measures of PFT give a global indication of lung function missing information about the morphological subtype, anatomical distribution and severity of tissue involvement. This is clearly supported by the current study, which shows that two distinct subgroups of patients with different degrees of emphysema may have similar FEV1 and DL,CO.
The severity of emphysema was assessed noninvasively by two completely different approaches: a radiological approach using quantitative CT, and a physiological approach using aerosol probing to measure the diameter of the distal airspaces. CT is more sensitive than PFT in the diagnosis of pulmonary emphysema, and investigators have suggested quantitative CT to assess the severity of emphysema 10, 22, 23. Both visual 6, 24 and computer assessment 8, 10, 22, 25 of CT in emphysema have shown a good correlation with the pathological extent of emphysema, thus reflecting the characteristic feature of the disease, namely permanent airspace enlargement without obvious fibrosis 1. The same aspect of the disease is studied by ADAM, which measures airspace dimensions. For the distal airspaces, EAD can be considered to be equivalent to the mean linear intercept 26.
The most common morphological subtype of emphysema is centrilobular emphysema (CLE), which is strongly associated with smoking. The morphological subtype associated with AAD is panlobular emphysema (PLE), characterised by uniform destruction of the pulmonary lobules with lower lobe predominance. In the current study, CT lung density using both RA-910 and PD15 revealed twice as much extensive emphysema in patients with AAD than smokers with CLE, despite the two groups being clinically matched and having the same degree of airflow limitation and reduction in diffusion capacity. The correlation between CT parameters and PFT parameters shows that CT performs equally well for parameters of airway obstruction (FEV1 and FEV1/FVC), hyperinflation (TLC and RV) and alveolar surface area (DL,CO) for the whole population, for patients with AAD and usual COPD. Regarding alveolar surface area, CT performance is better for AAD than for usual COPD; this is a clear reflection of the more extensive alveolar destruction associated with PLE in AAD.
The marked difference in quantitative CT between the two groups cannot simply be explained by the difference in hyperinflation. Gevenois et al. 27 studied the effect of acute and chronic hyperinflation on CT lung density in a group of asthmatics. They found that CT lung density, using CT mean lung density (MLD) and relative area below -950 HU (RA-950), did not change significantly after bronchial challenge inducing air trapping in a group of mild asthmatics. Furthermore, lung density in a second group of patients with chronic asthma and hyperinflation was not different from the control group, despite a significant difference in RV and TLC 27.
EAD was obtained at a Vpr of 20% of the intrathoracic gas volume at breath hold, because this lung depth certainly represents peripheral lung regions 16. A total of 15 patients with usual COPD (75%) and 16 with AAD (73%) were able to respire the air volume necessary for obtaining EAD20. The fact that the measurement was inconclusive in one-quarter of the study population is a clear limitation of the technique and the current study. EAD20 of all patients was greater than in normal individuals; nevertheless, the higher values in patients with AAD confirm the CT finding that distal airspaces in those patients are larger, again reflecting more severe lung destruction than in usual COPD. However, the potential of CT to separate the two subgroups seems to be much bigger than for EAD (figs 1⇑ and 2⇑).
Beinert et al. 14 reported EAD in 25 patients with COPD, 14 with usual COPD and nine with AAD admitted for i.v. substitution therapy. The mean EAD in usual COPD was 0.64±0.15 mm, which is similar to the results in the current study. In addition, they noticed that patients with AAD had the largest EAD values (1.14±0.31 mm). Their figures are significantly higher than those reported in the current study (p<0.05), probably as a result of more severe illness. Beinert et al. 14 measured MLD in 10 patients, five with AAD, and found the closest relationship to EAD (r = -0.82; p<0.05). However, MLD is more influenced by the partial volume effects and pixels with higher CT density than the parameters used in the current study (PD15 and RA-910).
EAD20 in patients with usual COPD was significantly higher than in healthy individuals, which suggests the presence of a certain degree of emphysema in most patients with COPD. This is also confirmed by the high RA-910 and low PD15. Pathological studies have shown a high incidence of CLE and paraseptal emphysema in heavy smokers with or without airway obstruction 28. Post mortem studies have shown that emphysema is common in autopsies, and mild degrees of emphysema have been reported in 50–70% of cases 29.
There is an important difference between the two study groups; all patients with usual COPD were current smokers, whereas 20 patients with AAD were ex-smokers and two were life-long nonsmokers. This factor may contribute to the significant difference in airspace size assessed by EAD and lung density assessed by quantitative CT. Some of the difference might be due to the ongoing inflammation and excessive mucus secretion in smokers with usual COPD 30; however, the inflammatory changes only partially subside after smoking cessation 31. This is evident in patients with AAD, where progressive airway obstruction and tissue destruction take place, despite smoking cessation. The probable contribution of small airway disease to the difference between the two groups is difficult to assess in the current design. To the current authors’ knowledge, there are no comparative or longitudinal studies of the influence of smoking cessation on distal airspace size or lung density changes. These are undoubtedly interesting areas for future research.
The current authors are not aware of any previous study of ABD600 in patients with AAD. ABD600 was similarly increased in the two patient groups, reflecting similar degrees of disturbances in lung ventilation. It has been shown previously that ABD is increased in patients with emphysema 15, 32. ABD is also slightly increased in smokers without airway obstruction 32, in children with mild asthma 33, and in patients with cystic fibrosis 34. Those and other studies have shown that ABD assesses different aspects of the pulmonary physiology than PFT. This is also clear from the current results, which show no significant correlation between ABD and any measure of PFT. However, one study in patients with emphysema has shown some correlation between ABD600 and FEV1 (r = -0.37; p<0.05), but no correlation to static lung volumes and diffusion capacity 35. ABD600 seems to be sensitive to the early changes in lung physiology that lead to ventilation inhomogeneity; however, the method is unspecific for emphysema and is incapable of assessing the severity of emphysema and differentiating between subtypes of emphysema. This is also obvious from the correlation analysis with CT lung density, which did not reveal any significant correlation, even in patients with AAD with more severe emphysema. In contrast to the current findings, Kohlhaufl et al. 16 reported a significant correlation between ABD600 and relative area below -900 HU, relative area below -925 HU, RA-950, MLD and visual scoring of emphysema. The explanation for this discrepancy is not clear.
In conclusion, the current study shows remarkable differences in quantitative computed tomography parameters and effective peripheral airspace dimension between smokers with usual chronic obstructive pulmonary disease and patients with α1-antitrypsin deficiency. These differences seem to be closely related to the pathological changes that are characteristic of the subtype of emphysema, and can only partly be explained by differences in pulmonary function tests.
Acknowledgments
The authors would like to acknowledge the support of members of the Software Performance Reproducibility in Emphysema Assessment: Demonstration (SPREAD) group.
- Received June 22, 2004.
- Accepted August 16, 2004.
- © ERS Journals Ltd
References
