Functional Impairment in Emphysema: Contribution of Airway Abnormalities and Distribution of Parenchymal Disease

Functional Impairment in Emphysema Chest Imaging Original Research Zelena A. Aziz 1 Athol U. Wells 2 Sujal R. Desai 3 Stephen M. Ellis 4 Amanda E. Walker 5 Sharyn MacDonald 6 David M. Hansell 1 Aziz ZA, Wells AU, Desai SR, et al. DOI:10.2214/AJR.04.1578 Received October 10, 2004; accepted after revision December 8, 2004. 1 Department of Radiology, Royal Brompton Hospital, Sydney St., London SW3 6NP, England. Address correspondence to Z. A. Aziz (z.aziz@rbh.nthames.nhs.uk). 2 Interstitial Lung Disease Unit, Royal Brompton Hospital, London, England. 3 Department of Radiology, King s College Hospital, London, England. 4 Department of Radiology, The London Chest Hospital, London, England. 5 Department of Radiology, St. Mary s Hospital, London, England. 6 Department of Radiology, Christchurch Hospital, Auckland, New Zealand. AJR 2005; 185:1509 1515 0361 803X/05/1856 1509 American Roentgen Ray Society Functional Impairment in Emphysema: Contribution of Airway Abnormalities and Distribution of Parenchymal Disease OBJECTIVE. The aim of this study was to identify ancillary morphologic features on high-resolution CT that modify airflow obstruction and gas transfer levels in individuals with emphysema. MATERIALS AND METHODS. The extent of emphysema on high-resolution CT was quantified by density masking in 101 patients. CT scans were evaluated for airway abnormalities (bronchial wall thickness, extent of bronchiectasis, bronchial dilatation, and evidence of small airways disease) and disease heterogeneity (uniformity, core rind distribution, craniocaudal distribution, and lung texture). Stepwise regression analysis was used to determine CT features that influenced forced expiratory volume in 1 sec (FEV 1 ) and the single-breath diffusing capacity for carbon monoxide (DLCO) for a given extent of emphysema. RESULTS. The extent of emphysema using automated estimation was 28.4% ± 12.3% (mean ± SD). On univariate analysis the extent of emphysema correlated strongly with FEV 1 (R = 0.63, p < 0.0005) and DLCO (R = 0.63, p < 0.0005) levels. Stepwise regression analysis revealed that bronchial wall thickness and the extent of emphysema were the strongest independent determinants of FEV 1 (model R 2 = 0.49; p = 0.002 and < 0.001, respectively); the extent of bronchiectasis and degree of bronchial dilation did not separately influence FEV 1 levels. The only morphologic features linked to DLCO levels on multivariate analysis were increasingly extensive emphysema and a higher proportion of emphysema in the core region (model R 2 = 0.45; p < 0.001 and 0.002, respectively). CONCLUSION. The important additional CT abnormalities in individuals with emphysema that influence FEV 1 and DLCO levels irrespective of disease extent are bronchial wall thickness and core rind heterogeneity, respectively. These observations have implications for the accurate functional assessment of patients considered for lung volume reduction surgery. n patients with emphysema, reductions in gas transfer and forced ex- I piratory volume in 1 sec (FEV 1 ) correlate best with disease extent on CT [1 3] and most accurately predict survival [4, 5]. However, only 10 67% of the reduction in FEV 1 and single-breath diffusing capacity for carbon monoxide (DLCO) is accounted for by disease extent on CT [6 10]. Therefore, other morphologic features are likely to influence both FEV 1 and gas transfer levels. Identification of functionally important CT features is of particular relevance in patient selection for lung volume reduction surgery. In cases of severe disease, FEV 1 and gas transfer levels are the most frequently used functional selection criteria [11]. It is therefore important to know whether functional impairment is a true reflection of morphologic disease extent in individual patients. In the present study, we explored relationships between specific lung function parameters (FEV 1 and DLCO) and morphologic features on CT. Both large- and small-airways abnormalities and the distribution of emphysema were evaluated. The specific aim was to determine whether these ancillary CT features influence airflow obstruction and gas transfer levels once the extent of emphysema is taken into account. Materials and Methods Patients presenting to our institution between July 1994 and July 2000 with evidence of emphysema on high-resolution CT were identified retrospectively from computerized reports using a word-search facility (McDonnell Douglas Hospital Information System). This approach was adopted to ensure that a wide range of patients with emphysema of varying severity, from mild to extensive, was studied. On re- AJR:185, December 2005 1509

Fig. 1 High-resolution CT image in 76-year-old man with emphysema and bronchial wall thickening within right lower lobe that was scored as grade 1 by both radiologists (bronchial dilatation score, grade 1; bronchiectasis score, grade 2). view of the case notes, patients with concurrent interstitial lung disease on high-resolution CT (n = 3), lobectomy or pneumonectomy (n = 2), pneumothorax or pleural effusion (n = 2), or α 1 -antitrypsin deficiency (n = 4) were excluded. Our ethics committee has given approval for retrospective analyses of clinical and imaging data. Pulmonary Function Tests FEV 1 levels were measured using a rolling seal spirometer (P.K. Morgan). DLCO levels were measured using a carbon monoxide single-breath technique: Gas transfer results were adjusted for hemoglobin (model B, P. K. Morgan; and 6200 Autobox DL, Sensormedics). FEV 1 and DLCO values were expressed as percentages of values predicted for age, sex, and height. CT Technique CT scans were obtained on an electron beam CT scanner (Imatron, GE Healthcare). Thin sections (1 3 mm) at 10-mm intervals were obtained in all patients. Patients were scanned in the supine position with images obtained at full inspiration. Images were reconstructed with a high-spatial-resolution algorithm and photographed at appropriate window settings (level, 500 H; width, 1,500 H). None of the patients received IV contrast material. CT Estimation of Emphysema Extent: Automated Densitometric Quantification The extent of emphysema was quantified on each CT section using an automated density mask technique in which voxels with attenuation values below a specific threshold are highlighted [12, 13]. The technique involved the segmentation of anatomic structures, the use of gradient correction to compensate for the density differences in the lung parenchyma due to gravity, and the application of a classification algorithm that identified and measured the areas of decreased attenuation corresponding to diseased lung [13]. The cutoff level between normal lung density and abnormal low-attenuation lung was defined as 950 H, because this value accurately predicts the macroscopic and microscopic extent of emphysema [9, 14]. The percentage of low-attenuation emphysematous destruction was calculated for the whole lung. Analysis of Airway Abnormalities Two experienced chest radiologists independently scored the six lobes (the lingula was regarded as a separate lobe). The extent of bronchiectasis was graded as follows: grade 1, localized bronchiectasis affecting one or part of one bronchopulmonary segment; grade 2, bronchiectasis in more than one bronchopulmonary segment; and grade 3, generalized cystic bronchiectasis. The severity of bronchial dilatation was quantified by comparing the internal diameter of the bronchus to its adjacent pulmonary artery: grade 0, no bronchiectasis; grade 0.5, trivial dilatation; grade 1, 100 200% arterial diameter; grade 2, > 200 300% arterial diameter; and grade 3, > 300% arterial diameter. Bronchial wall thickness was similarly quantified relative to the adjacent pulmonary artery: grade 0, none; grade 0.5, trivial bronchial wall thickening; grade 1, < 50% of the arterial diameter (Fig. 1); grade 2, 50 100% of the arterial diameter; and grade 3, > 100% of the arterial diameter. If a mosaic attenuation pattern was present, the decreased attenuation component that was attributable to small airways disease in each lobe was also quantified using the following scale: grade 0, none; grade 1, 25%; grade 2, 26 50%; grade 3, 51 75%; grade 4, > 75 100%. The total lung score for each CT feature was derived by summing the scores for each lobe; the final scores used in analysis were the sum of the total lung scores of both observers. Analysis of Heterogeneity of Emphysema On the basis of previous studies, four aspects of disease heterogeneity were visually scored. All images for analysis and the division of lung into regions for method 1 and into core rind areas for method 2 were preselected by a third chest radiologist, who was not one of the two observers. Method 1: overall uniformity (adapted from Wisser et al. [15]) The purpose of this scoring method was to quantify the overall uniformity of emphysematous destruction within the lungs. Three anatomic levels were defined: the upper border of the aortic arch, 2 cm below the level of the carina, and 2 cm above the highest diaphragmatic dome. At each level, each lung was divided into two regions by a horizontal line, its position ensuring that each lung was separated according to the most prominent difference in the extent of emphysema. Each region had to be at least 30% of the whole lung area. The amount of destroyed lung parenchyma (severity of emphysema) in all of the resulting 12 segments was graded: grade 0, normal lung; grade 1 (mild), 25% air space compared with lung; grade 2 (moderate), 25 50% air space; grade 3 (marked), 50% air space; or grade 4 (severe), no normal lung [16]. The difference between the median of the three highest scores and the median of the three lowest scores was calculated and used to express the overall degree of heterogeneity from 0 (homogeneous) to 4 (marked heterogeneity) (Fig. 2). Method 2: core rind heterogeneity (adapted from Nakano et al. [17]) This method determined the differential distribution of emphysema between the inner (core) and outer (rind) regions of the lung (Fig. 3). The same three sections used for the first method were analyzed. At each level, each lung was divided into core and rind regions: The area of each lung approximates to that of a semicircle. The boundaries of the core region were obtained by multiplying the radius of each lung by 0.7; this ensured that both peripheral and central regions comprised 50% of the total lung area. The percentage of emphysematous lung within the inner and outer segments was estimated to the nearest 5%. For analysis, the extent of emphysema within the core region was expressed as a ratio of the total area of emphysema. Method 3: craniocaudal heterogeneity (adapted from Cederlund et al. [18]) The distribution of emphysema between upper and lower lung regions was assessed. Four levels in the cranial part 1510 AJR:185, December 2005

Functional Impairment in Emphysema and four levels in the caudal part of the lung were selected; the most cranial slice was at the level of the great vessels. The next three sections interspaced by 20 mm were then selected. The most caudal section was that which included only a small part of the diaphragm. Three levels interspaced by 20 mm cranial to this were selected. The uppermost image from the cranial lung and the uppermost image from the caudal lung were regarded as an image pair. The next selected image from the cranial lung and from the caudal lung was regarded as another image pair. This resulted in four image pairs. Image pairs were classified into five grades of heterogeneity: 1, obviously more emphysema in the cranial image; 2, somewhat more emphysema in the cranial image; 3, equal extent of emphysema; 4, somewhat more emphysema in the caudal image; or 5, obviously more emphysema in the caudal image. Method 4: texture of lung The aim was to identify individuals with a pattern of emphysema that was not diffuse, but clustered, with admixed islands of normal lung. Five levels were analyzed: The three levels used for the other methods and two further levels were chosen so that the five levels were equally spaced throughout the lung. At each level the right and left lungs were assessed separately and graded as follows: grade 1, predominantly centrilobular emphysema; grade 2, predominantly panacinar emphysema; or grade 3, A C contiguous areas of normal or near-normal lung that occupy and total between 25% and 65% of the lung being assessed. If grade 3 was scored, then the percentage of this area was approximated to the nearest 5%. The area of normal or near-normal lung was termed an island (Figs. 4 and 5). For analysis, the percentage of islands for each patient was calculated (number of sections with islands / total number of sections with emphysema). Scores for both observers were averaged for each aspect of heterogeneity evaluated. Statistical Analysis Population clinical characteristics, CT extent of emphysema, and pulmonary function indexes are Fig. 2 45-year-old man with emphysema. A C, High-resolution CT images show examples of scoring uniformity of emphysematous destruction at level of aortic arch (A), carina (B), and 2 cm above level of aortic arch (C). Overall heterogeneity score given by both observers was 1.5 (difference between medians of three best and worst sections). B AJR:185, December 2005 1511

Fig. 3 High-resolution CT image in 68-year-old woman illustrates division of each lung into core and rind regions. Extent of emphysema is greater in core region of lung when compared with rind, or peripheral, region. Fig. 4 High-resolution CT image of 42-year-old man shows islands of normal or near-normal lung adjacent to areas of emphysematous destruction. This case was therefore scored as grade 3 with contiguous area of normal or near-normal lung estimated at 35% and 45% by two observers, respectively. expressed as means or medians depending on data distribution. Univariate correlations were examined using Spearman s rank correlation coefficient (R). Independent relationships between CT features and pulmonary function indexes were identified using stepwise regression models, with individual functional indexes evaluated as the dependent variables in separate models. The variance of the physiologic variables accounted for by the CT features is expressed as the square of the correlation coefficient (R 2 ). The assumptions of multiple linear regression were met in all analyses as judged by testing for heteroscedasticity and omitted variables. A p value of less than 0.05 was taken to indicate statistical significance. Interobserver agreement in categoric variables was quantified using the weighted kappa coefficient (κ w ) [19]. All statistical analyses were performed using STATA software (version 4.0, StataCorp). Results The study group consisted of 101 patients. There were 35 women and 66 men with a mean age of 61 years (range, 26 86 years). A full smoking history was available in 99 of 101 patients (median smoking history, 35 pack-years; range, 7 175 pack-years). Pulmonary function data, CT emphysema scores, and univariate correlations between the extent of emphysema and FEV 1 and the extent of emphysema and DLCO are shown in Table 1. Interobserver agreement for global CT scores was good for bronchial wall thickness (κ w = 0.67), extent of bronchiectasis (κ w = 0.69), and bronchial dilatation (κ w = 0.69). The prevalence of high-resolution CT features of obliterative small airways disease (n = 7) was too low to warrant analysis. Interobserver agreement for heterogeneity scores on CT was good for craniocaudal distribution of disease (method 3) (κ w = 0.79), the assessment of overall uniformity of emphysematous destruction (method 1) (κ w = 0.70), and core rind heterogeneity (method 2) (κ w = 0.65) and were moderate for the assessment of lung texture (method 4) (κ w = 0.45). As shown in Table 2, there were significant correlations between FEV 1 and the overall uniformity of emphysematous destruction, the severity of bronchial wall thickening, and the extent of bronchiectasis. Similar relationships were observed between the first two CT variables and DLCO. Significant independent relationships, as identified by regression models between FEV 1 and DLCO, and CT features are shown in Table 3. Stepwise regression analysis revealed that bronchial wall thickness and the extent of emphysema were the strongest independent determinants of FEV 1 (R 2 = 0.49; regression coefficient [RC] = 3.42 and 1.27, p = 0.002 and < 0.001, respectively). In the regression model, the extent of bronchiectasis and bronchial dilatation were not significant independent predictors of FEV 1. In multivariate stepwise regression analysis, only core rind heterogeneity independently predicted DLCO; a higher percentage of emphysema in the core was associated with a reduction in DLCO. The combination of the extent of emphysema and the percentage of emphysema in the core region accounted for 45% of the variability of DLCO (R 2 = 0.45, RC = 1.01 and 0.52, p < 0.001 and 0.002, respectively). In the regression model, craniocaudal heterogeneity, overall uniformity, or lung texture were not retained as a significant determinant of DLCO. Discussion Airway obstruction as a defining feature of emphysema was first reported by Laennec [20], although the exact site and the pathophysiology of the obstruction have been continuing sources of controversy. Early pathophysiologic studies showed that coexisting small airways [7, 21] and large airway [22] diseases independently contribute to airflow limitation. The second and important functional consequence of emphysema is a reduction of gas diffusing capacity, although the extent of emphysema is not the only factor in predicting the variability of DLCO [3, 10]. Recently, there has been an interest in the distribution of emphysema, prompted by the results of several studies that concluded that heterogeneous disease was associated with a better outcome in patients who underwent lung volume reduction surgery [15, 23 25]. Thus, the aim of this study was to determine whether particular CT morphologic features (airway abnormalities and the distribution of emphysema) independently predicted airflow obstruction and gas transfer. Our study has shown that bronchial wall thickness and the proportion of emphysema within the core, or central part of the lung, as judged by high-resolution CT modify the fundamental relationship between the overall extent of emphysema and FEV 1 and DLCO levels, respectively. The correlation between the extent of emphysema and the severity of airflow obstruction is well recognized [2, 7, 26]. However, outliers are common, with many patients having severe airway obstruction (as measured by FEV 1 and residual volume) but apparently limited emphysema on CT [7, 27]. Other studies have failed to show a direct relationship between the extent of alveolar wall destruction and the severity of airflow obstruction [28 30], suggesting that the extent of emphysema is not the only morphologic abnormality causing airflow obstruction. 1512 AJR:185, December 2005

Functional Impairment in Emphysema In our study, significant correlations were observed between FEV 1 and the uniformity of emphysema and between the extent of bronchiectasis and the degree of bronchial wall thickening, with regression analysis showing that bronchial wall thickness was the important determinant of FEV 1. This feature and the extent of emphysema accounted for 49% of the variability of FEV 1. The negative relationship between bronchial wall thickness and FEV 1 is in agreement with a study by Nakano et al. [31]. In that study, the wall thickness of a single segmental airway (superior segment of right upper lobe) was measured using an airway analysis software program [31]. In our study, a global estimation was made of airway wall thickness, bronchial dilatation, and extent of bronchiectasis and small airways disease, which would seem appropriate when attempting to evaluate the contribution of airway abnormalities to parameters of airflow limitation. Although pathologic studies have suggested a role for small airway disease in airflow limitation in emphysema [21, 32], we were unable to assess the contribution of small airway disease to airflow limitation in emphysema because of the small number of cases in which the presence of small airway disease was reported. A mosaic attenuation pattern on high-resolution CT representing small airway disease is difficult, if not impossible, to identify on an inhomogeneous background of widespread emphysema [33]. Consequently, it remains uncertain whether the A Fig. 5 51-year-old man with emphysema who was outlier in our study. A and B, High-resolution CT images show degree of outflow obstruction is not mirrored by extent of emphysema. Note bronchial wall thickening in upper (A) and lower (B) lobes. low scores of mosaic attenuation pattern reflect a true or spuriously low prevalence of this pattern. The difficulty in distinguishing between the two conditions highlights a limitation of our study; the word-search method of selecting patients with emphysema may have resulted in some cases of obliterative small airway disease being included as emphysema cases. In addition, very subtle emphysema tends to be underreported, and these individuals would not have been identified by our word-search method. Our results also showed that the severity of bronchial dilatation was not an independent predictor of FEV 1. Nakano et al. [31] found that in patients with emphysema, those with a larger luminal area of the superior segmental bronchus of the right upper lobe had less severe airflow obstruction. Although this may be true for univariate analysis, in the regression model, bronchial dilatation was not a significant factor in determining the variability of FEV 1. Our results also indicate that the distribution of emphysema does not independently influence FEV 1. The core rind distribution of emphysema emerged as a determinant of gas transfer in the present study. We found that an increased percentage of emphysema within the core region was associated with a lower DLCO and this finding supports those of two previous studies [17, 34]. A potential explanation for this observation is that pulmonary blood flow is significantly greater in the central region of the lung compared with the periphery [35]; therefore, destruction of the lung in this region has a greater effect on gas transfer than similar changes in the periphery of the lung. Strikingly, no other aspect of disease heterogeneity was found to influence DLCO. The finding that the craniocaudal distribution of disease did not independently predict the variability of gas transfer may at first appear surprising. Several previous studies have shown a stronger correlation between the percentage of emphysema within the lower zone and DLCO than with the upper zone [26, 34]; however, while the zonal distribution of emphysema may differentially affect DLCO, this effect vanishes when the overall extent of emphysema is taken into account in multivariate analysis. Similarly, the hypothesis that large islands of spared lung among emphysematous areas would result in relative preservation of gas transfer, compared with a uniform distribution of emphysema, was not supported by our findings. The study design necessitated the inclusion of several methods of defining disease heterogeneity, which were modeled on methods used by previous investigators. Undoubtedly, there is an inherent subjectivity involved in the semiquantitative methods used; however, some of the variables assessed (particularly the analysis of lung texture) do not lend themselves to quantification using automated methods. In support of our methods, agreement between our two observers ranged from moderate to good for all analyses of disease heterogeneity. B AJR:185, December 2005 1513

TABLE 1: Pulmonary Function Data, the Extent of Emphysema on CT as Judged by Density Mask Quantification, and Univariate Correlations Between Pulmonary Functional Indexes and the Extent of Emphysema Mean % of the CT Extent of Emphysema Pulmonary Function Data Predicted Value ± SD R p FEV 1 42.4 ± 26.7 0.63 < 0.0005 DLCO 48.1 ± 20.8 0.63 < 0.0005 CT extent of emphysema (%) 28.4 ± 12.3 Note FEV 1 = forced expiratory volume in 1 sec, DLCO = single-breath diffusing capacity for carbon monoxide. TABLE 2: Univariate (Spearman s) Correlations Between CT Morphologic Features and Percentage Predicted FEV 1, DLCO, and CT Extent of Emphysema FEV 1 DLCO CT Extent of Emphysema CT Morphologic Features R p R p R p Overall uniformity 0.31 < 0.001 0.23 < 0.005 0.42 NS Core rind heterogeneity 0 0.09 NS 0.15 NS Craniocaudal heterogeneity 0 0.09 NS 0.05 NS Lung texture 0.14 NS 0.13 NS 0.18 NS Extent of bronchiectasis 0.28 < 0.005 0.16 NS 0.14 NS Bronchial wall thickening 0.42 0 0.39 0.0001 0.30 < 0.005 Severity of bronchial dilatation 0.24 NS 0.17 NS 0.19 NS Note FEV 1 = forced expiratory volume in 1 sec, DLCO = single-breath diffusing capacity for carbon monoxide, NS = not significant. TABLE 3: Significant Independent Relationships Between FEV 1 and DLCO and CT Morphologic Features Pulmonary Function Data % Emphysema Core Rind Heterogeneity Bronchial Wall Thickness FEV 1 (R 2 = 0.49) Regression coefficient 1.27 3.42 95% CI 1.59 to 0.94 5.52 to 1.31 p 0.000 0.002 DLCO (R 2 = 0.45) Regression coefficient 1.01 0.52 95% CI 1.35 to 0.83 0.85 to 0.20 p 0.000 0.002 Note FEV 1 = forced expiratory volume in 1 sec, DLCO = single-breath diffusing capacity for carbon monoxide, CI = confidence interval. In conclusion, we have shown that in patients with emphysema, bronchial wall thickness and core rind heterogeneity are important high-resolution CT features that influence FEV 1 and DLCO levels, respectively. These observations may explain the disparity that is sometimes encountered between the extent of emphysema at CT and measurements of airflow obstruction and gas transfer. Our observations should enable a more accurate functional assessment of patients being considered for lung volume reduction surgery. References 1. Klein JS, Gamsu G, Webb WR, Golden JA, Muller NL. High-resolution CT diagnosis of emphysema in symptomatic patients with normal chest radiographs and isolated low diffusing capacity. Radiology 1992; 182:817 821 2. Heremans A, Verschakelen JA, Van fraeyenhoven L, et al. Measurement of lung density by means of quantitative CT scanning: a study of correlations with pulmonary function tests. Chest 1992; 102:805 811 3. Gould GA, Redpath AT, Ryan M, et al. Lung CT density correlates with measurements of airflow limitation and the diffusing capacity. Eur Respir J 1991; 4:141 146 4. Kanner RE, Renzetti AD Jr, Stanish WM, et al. Predictors of survival in subjects with chronic airflow limitation. Am J Med 1983; 74:249 255 5. Traver GA, Cline MG, Burrows B. Predictors of mortality in chronic obstructive pulmonary disease: a 15-year follow-up study. Am Rev Respir Dis 1979; 119:895 902 6. Sakai F, Gamsu G, Im J, Ray CS. Pulmonary function abnormalities in patients with CT-determined emphysema. J Comput Assist Tomogr 1987; 11:963 968 7. Gelb AF, Schein M, Kuei J, et al. Limited contribution of emphysema in advanced chronic obstructive pulmonary disease. Am Rev Respir Dis 1993; 147:1157 1161 8. Nishimura K, Murata K, Yamagishi M, et al. Comparison of different computed tomography scanning methods for quantifying emphysema. J Thorac Imaging 1998; 13:193 198 9. Gevenois PA, De Vuyst P, de Maertelaer V, et al. Comparison of computed density and microscopic morphometry in pulmonary emphysema. Am J Respir Crit Care Med 1996; 154:187 192 10. Bae KT, Slone RM, Gierada DS, Yusen RD, Cooper JD. Patients with emphysema: quantitative CT analysis before and after lung volume reduction surgery work in progress. Radiology 1997; 203:705 714 11. Meyers BF, Yusen RD, Guthrie TJ, et al. Results of lung volume reduction surgery in patients meeting a national emphysema treatment trial high-risk criterion. J Thorac Cardiovasc Surg 2004; 127:829 835 12. Kinsella M, Muller NL, Abboud RT, et al. Quantitation of emphysema by computed tomography using a density mask program and correlation with pulmonary function tests. Chest 1990; 97:315 321 13. Chabat F, Desai SR, Hansell DM, Yang GZ. Gradient correction and classification of CT lung images for the automated quantification of mosaic attenuation pattern. J Comput Assist Tomogr 2000; 24:437 447 14. Gevenois PA, de Maertelaer V, De Vuyst P, et al. Comparison of computed density and macroscopic morphometry in pulmonary emphysema. Am J Respir Crit Care Med 1995; 152:653 657 15. Wisser W, Klepetko W, Kontrus M, et al. Morphologic grading of the emphysematous lung and its relation to improvement after lung volume reduction surgery. Ann Thorac Surg 1998; 65:793 799 16. Slone RM, Gierada DS. Radiology of pulmonary emphysema and lung volume reduction surgery. Semin Thorac Cardiovasc Surg 1996; 8:61 82 17. Nakano Y, Sakai H, Muro S, et al. Comparison of low attenuation areas on computed tomographic scans between inner and outer segments of the lung 1514 AJR:185, December 2005

Functional Impairment in Emphysema in patients with chronic obstructive pulmonary disease: incidence and contribution to lung function. Thorax 1999; 54:384 389 18. Cederlund K, Bergstrand L, Hogberg S, et al. Visual classification of emphysema heterogeneity compared with objective measurements: HRCT vs spiral CT in candidates for lung volume reduction surgery. Eur Radiol 2002; 12:1045 1051 19. Brennan P, Silman A. Statistical methods for assessing observer variability in clinical measures. BMJ 1992; 304:1491 1494 20. Laennec RTH. Treatise on the diseases of the chest and on mediate auscultation, 4th ed. London, England: DeSilver Thomas and Company, 1835 21. Hogg JC, Macklem PT, Thurlbeck WM. Site and nature of airway obstruction in chronic obstructive lung disease. N Engl J Med 1968; 268:1355 1360 22. Macklem PT, Fraser RG, Brown WG. Bronchial pressure measurements in emphysema and bronchitis. J Clin Invest 1965; 44:897 905 23. Nakano Y, Coxson HO, Bosan S, et al. Core to rind distribution of severe emphysema predicts outcome of lung volume reduction surgery. Am J Respir Crit Care Med 2001; 164:2195 2199 24. Weder W, Thurnheer R, Stammberger U, et al. Radiologic emphysema morphology is associated with outcome after surgical lung volume reduction. Ann Thorac Surg 1997; 64:313 319 25. Gierada DS, Yusen RD, Villanueva IA, et al. Patient selection for lung volume reduction surgery: an objective model based on prior clinical decisions and quantitative CT analysis. Chest 2000; 117:991 998 26. Gurney JW, Jones KK, Robbins RA, et al. Regional distribution of emphysema: correlation of high-resolution CT with pulmonary function tests in unselected smokers. Radiology 1992; 183:457 463 27. Sandek K, Bratel T, Lagerstrand L, Rosell H. Relationship between lung function, ventilation perfusion inequality and extent of emphysema as assessed by high-resolution computed tomography. Respir Med 2002; 96:934 943 28. Silvers GW, Maisel JC, Petty TL, Filley GF, Mitchell RS. Flow limitation during forced expiration in excised human lungs. J Appl Physiol 1974; 36:737 744 29. Wright JL, Lawson LM, Pare PD, Hogg JC. The detection of small airways disease. Am Rev Respir Dis 1984; 129:989 994 30. West WW, Nagai A, Hodgkin JE, Thurlbeck WM. The National Institutes of Health Intermittent Positive Pressure Breathing trial: pathology studies. III. The diagnosis of emphysema. Am Rev Respir Dis 1987; 135:123 129 31. Nakano Y, Muro S, Sakai H, et al. Computed tomographic measurements of airway dimensions and emphysema in smokers: correlation with lung function. Am J Respir Crit Care Med 2000; 162:1102 1108 32. Gelb AF, Hogg JC, Muller NL, et al. Contribution of emphysema and small airways in COPD. Chest 1996; 109:353 359 33. Copley SJ, Wells AU, Muller NL, et al. Thin-section CT in obstructive pulmonary disease: discriminatory value. Radiology 2002; 223:812 819 34. Haraguchi M, Shimura S, Hida W, Shirato K. Pulmonary function and regional distribution of emphysema as determined by high-resolution computed tomography. Respiration 1998; 65:125 129 35. Hakim TS, Lisbona R, Dean GW. Gravity-independent inequality in pulmonary blood flow in humans. J Appl Physiol 1987; 63:1114 1121 AJR:185, December 2005 1515