ORIGINAL ARTICLE

Chronic Pulmonary Thromboembolism

Air Trapping on Computed Tomography and Correlation with Pulmonary Function Tests
Hiroaki Arakawa, MD, Eric J. Stern, MD, Takaaki Nakamoto, MD, Mutsuhisa Fujioka, MD, Noboru Kaneko, MD, and Hiroshi Harasawa, MD

Objective: We evaluated lung attenuation on inspiratory/expiratory

computed tomography (CT) and spiral CT angiography (CTA) from patients with chronic pulmonary embolism and correlated the CT findings with pulmonary function test (PFT) results.

Methods: We retrospectively reviewed 9 patients with chronic embolism (mean age, 62.3 years; 5 women and 4 men). Paired inspiratory, expiratory CT and matching CTA images were reviewed, and lung attenuation was evaluated in each segment. Lower attenuation on inspiratory images and air trapping on expiratory images were subjectively assessed and correlated with PFT results. The relationship between the presence of clot and lung attenuation was evaluated. Nine age-matched normal subjects served as controls. Results: Lower attenuation with mosaic perfusion and air trapping
were identified in 6 and 9 patients, respectively (mean scores, 8.1 and 11.3, respectively). Air trapping was identified in 19 (42.2%) of 45 segments with lower attenuation on inspiratory images, but was also noted in 31 segments with normal inspiratory attenuation. Air trapping was associated with the presence of proximal arterial stenosis (P < 0.01), and the area showed less contrast enhancement than the adjacent lung (P < 0.05). Extent of air trapping correlated inversely with PFT parameters of peripheral airway obstruction such as maximum mid-expiratory flow rate (r = 0.86, P = 0.003). On the other hand, extent of mosaic perfusion did not correlate with PFT.

n patients with chronic pulmonary thromboembolic disease (CPTE), a mosaic pattern of lung attenuation is the most common lung parenchymal abnormality seen with thin-section computed tomography (CT).1 The lower attenuation areas have previously been thought to represent regions of hypoperfusion due to chronically obstructed pulmonary arteries whereas the higher attenuation areas were considered to represent compensatory hyperperfusion; the so-called mosaic perfusion pattern.2,3 This mosaic perfusion pattern is not seen in patients with acute pulmonary thromboembolism.4,5 However, several recent reports have described air trapping demonstrated with expiratory CT scanning in patients with CPTE.6 Furthermore, it has been occasionally reported that patients with chronic pulmonary thromboembolism show airway obstruction on pulmonary function tests.79 Due to the lack of any published reports of the correlation between CT findings and pulmonary function test (PFT) results in chronic thromboembolism, we evaluated lung attenuation using inspiratory and expiratory thin-section CT scans and correlated the CT findings with those of PFT test results.

MATERIALS AND METHODS Patients

We retrospectively reviewed 9 patients with chronic pulmonary embolism that were referred for CT angiography between September 2001 and March 2002 (age range, 43 to 81 years; mean, 62.3 years; 5 women and 4 men). Informed consent for the CT study was obtained from all patients. Six patients had deep vein thrombosis and the other 3 had no identifiable cause. All patients were clinically stable during the period of CT and pulmonary function test. Only 1 patient was a current smoker (30 pack-years) and the others were never smokers. None of the patients had any other known preexisting lung disease. A final diagnosis of chronic pulmonary thromboembolism was made based on the combination of clinical manifestations, findings on spiral CT angiography (n = 9), pulmonary angiography (n = 7), right-heart catheterization (n = 7), ventilation and perfusion scans (n = 7), and cardiac ultrasonography

Conclusions: Air trapping is commonly seen in chronic embolism

and is found in areas of relative hypoperfusion. The extent of air trapping correlates with parameters of peripheral airway obstruction. Key Words: chronic pulmonary embolism, computed tomography, small airway disease (J Comput Assist Tomogr 2003;27:735742)
From the Department of Radiology (Drs Arakawa and Fujioka) and the Department of Cardiology and Pneumology (Drs Nakamoto, Kaneko, and Harasawa), Dokkyo University School of Medicine, Tochigi, Japan; and the Department of Radiology (Dr Stern), Harborview Medical Center, University of Washington School of Medicine, Seattle, Washington. Reprints: Hiroaki Arakawa, MD, Department of Radiology, Dokkyo University School of Medicine, 880 Kita-Kobayashi, Mibu, Tochigi 321-0293, Japan (e-mail: arakawa@dokkyomed.ac.jp). Copyright 2003 by Lippincott Williams & Wilkins

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

735

Arakawa et al

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

(n = 2). Findings of spiral CT angiography of the pulmonary arteries consistent with chronic thromboembolism included emboli eccentric and contiguous with the vessel wall, evidence of recanalization, arterial stenosis or web, and reduction of more than 50% of the overall arterial diameter.10 Only segmental-sized arteries and larger were considered for evaluation. Angiographic criteria for the diagnosis of chronic pulmonary embolism were pouching defects, webs or bands, intimal irregularities, abrupt vascular narrowing, complete vascular obstruction, or any combination of these.11 Seven patients underwent pulmonary angiography, and all of these had angiographic evidence of chronic embolism. The other 2 patients showed ventilation-perfusion mismatch and evidence of chronic embolism on CT angiography. Nine age-matched non-smokers that had normal pulmonary function test results and no known active or prior pulmonary disease underwent inspiratory and expiratory thin-section CT scans and served as controls. These subjects were initially suspected of having diffuse or focal lung disease and were referred for CT examination. Based on the clinical, radiographic, and pulmonary function test results, these subjects were regarded as normal.

CT Protocol
CT scanning was performed in all patients using a Somatom Volume Zoom CT scanner (Siemens, Erlangen, Germany). Paired inspiratory and expiratory thin-section (1 mm) CT scans were performed without contrast material administration followed by spiral CT angiography of pulmonary arteries. Paired inspiratory and expiratory scans were obtained with the patient in the supine position. Scans were performed from above the aortic arch to below the right diaphragm with 2-cm interspacing. Spiral CT angiography was performed after obtaining paired thin-section scans. After intravenous injection of low osmolar iodinated contrast material (Iopamiron 300, Nihon Shering, Osaka, Japan), scanning was initiated after a 15-second delay and performed at 1-mm collimation, spiral pitch of 1:6, and 0.75 seconds/gantry rotation from the lung base to the apex. Scan duration was 30 to 40 seconds. The spiral data was reconstructed with 1-mm collimation and both mediastinal and lung window images were produced. Lung window images were reconstructed using a high-frequency algorithm and were displayed with a window width of 1200 H and window level of 650 H while the mediastinal window images were displayed with a window width of 280 H and window level of 70 H.

Image Analysis
Mediastinal window images were reviewed in the cine mode at the CT console. One of the authors and the attending radiologist at our department together reviewed the mediastinal window images, and the presence of clots or arterial stenosis was evaluated by consensus. Pulmonary artery branches

were evaluated from the pulmonary trunk to the segmental arteries for the presence of arterial clots or stenosis. To validate the comparison of lung attenuation, the percentage changes in cross-sectional lung area between inspiratory and expiratory scans and between inspiratory and spiral scans were measured for each lung and in each lung zone. Three levels were evaluated including the level of the aortic arch, tracheal carina, and inferior pulmonary vein. Crosssectional lung areas were determined using the built-in CT software that automatically measures pixels with predetermined lung attenuation. We used lung attenuation between 500 H and 1024 H for extracting lung parenchyma and airways at each of the 3 levels. The presence of mosaic perfusion and air trapping was visually determined in each segment (a total of possible 162 segments; 10 in right lung, 8 in left lung, in 9 patients), and the extent was scored independently by 2 dedicated chest radiologists. The extent of lower attenuation areas on the inspiratory CT images (mosaic perfusion) and that of air trapping on the expiratory CT images were estimated at each of the 3 lung zones and for each lung using a 4-point scale: 0 (no such area visible), 1 (1% to 25% of the cross-sectional area of lung affected), 2 (25% to 50% of lung affected), 3 (50% to 75% of lung affected), and 4 (75% to 100% of lung affected).12 The maximum possible score for each lung was 12 (3 zones 4 points at each level), and for both lungs was 24. The border between the upper and middle zone was set at the mid-aortic arch, and that between the middle and lower zone was set at the level of the inferior pulmonary vein. Mosaic perfusion was defined as present if there were areas with mixed high and low attenuation areas of the lung on inspiratory scans considered to result from regional differences in lung perfusion.13 Air trapping was considered to be present if the area on the inspiratory scan showed little or no increase in lung attenuation after exhalation. To evaluate the pulmonary blood flow in areas with different lung attenuation, we calculated lung densities in the areas with differing lung attenuation on inspiratory or expiratory images in a separate session by one of the authors. To minimize the time difference for contrast material to reach each area, calculation was performed only in adjacent areas with differing lung attenuation in the same segment at 1 scan level. Two separate lung attenuation measurements using a small region of interest (ROI) were obtained and the average determined. Each ROI was 1.0 to 2.0 cm2 and was placed in each area in question to avoid large vessels. In each area of interest, lung attenuation at full inspiration, expiration, and postcontrast enhancement were calculated, and the differences in lung attenuation between inspiratory and expiratory scans and inspiratory and postcontrast-enhanced scans then determined. Because lung attenuation is closely associated with lung volume, the degree of intravenous contrast enhancement in the region of interest needs to be determined at nearly identical lung vol 2003 Lippincott Williams & Wilkins

736

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

Chronic Pulmonary Thromboembolism

umes. Therefore, we did not calculate the postcontrastenhanced lung attenuation when the cross-sectional lung area differed by more than 5% from that of the inspiratory highresolution computed tomography (HRCT) images.

TABLE 1. Pulmonary Function Test Results in Patients with Chronic Pulmonary Embolism and in Controls VC FVC FEV1 FEF50 FEF25 MMEF DLCO 74.1* 88.0* 56.8* 75.6* 31.2 47.7 56.2 81.6 84.6 94.8

Pulmonary Function Tests

Pulmonary function tests were performed in all patients within 1 month of the CT scans using computerized pulmonary function instruments (FUDAC 50 or 70, Fukuda Denshi, Tokyo, Japan). The vital capacity (VC), forced vital capacity (FVC), forced expiratory volume in 1 second (FEV1), forced expiratory flow rate at 50% of FVC (FEF50), forced expiratory flow rate at 25% of FVC (FEF 25 ), and maximum midexpiratory flow rate between 25% to 75% of FVC (MMEF) were determined by spirometry while the diffusing capacity of carbon monoxide (DLCO) was determined by the single-breath method. Each result was expressed as a percentage of the predicted value.

Patients 106.8 104.1 Controls 103.6 102.6

Note: Values are expressed by percentage. VC, vital capacity; FVC, forced vital capacity/ FEV1, forced expiratory volume in 1 second; FEF50 forced expiratory flow rate at 50% of FVC; FEF25, forced expiratory flow rate at 25% of FVC; MMEF, maximum mid-expiratory flow rate between 25% to 75% of FVC; DLCO, diffusing capacity of carbon monoxide. *P < 0.05 P < 0.005

Statistical Analysis
Statistical analysis was performed using SPSS software (SPSS Japan Inc., Tokyo, Japan). The intraclass correlation coefficient was calculated for the agreement of scores for air trapping and those for mosaic perfusion. The intraclass correlation coefficient is used to appreciate the interobserver agreement of a quantitative variable.14,15 The differences in pulmonary function parameters and CT scores between the controls and the patients were assessed using Student t test. The relationship between average CT scores and pulmonary function parameters were correlated with Spearman rank correlation coefficient. The correlations between the presence of clots or arterial stenosis and air trapping or mosaic perfusion in each segment were assessed using the 2 test. The differences in attenuation increase after exhalation and those after administration of contrast material were assessed between higher and lower attenuation areas and also between areas with and without air trapping using Student t test.

ages ranged from 0.19 to 0.57 with a mean of 0.31 (SD, 0.09) in the patient group and from 0.16 to 0.53 with a mean of 0.37 (SD, 0.11) in the control group. No statistically significant differences were observed in cross-sectional area change between the patients and control group, indicating comparable effort between the 2 groups. All images qualified as adequate for expiratory image analysis in both groups. A mosaic perfusion pattern was noted in 6 patients and air trapping was noted in all 9 patients. Interobserver agreement for the presence or absence of air trapping and mosaic perfusion was at 100%. Decreased lung attenuation attributed to mosaic perfusion was seen in a total of 45 (36.6%) segments. Of these, 34 (75.6%) segments had clot or arterial stenosis in the subtending artery. Conversely, in 41 segments without proximal arterial clots or stenosis, 11 (24.4%) segments showed decreased lung attenuation (Table 2). Statistical analysis did not show significant correlation between the presence of clots or arterial stenosis and that of lower attenuation on inspiratory CT (P > 0.05).

RESULTS Pulmonary Function Test

The pulmonary function test results of the 9 patients and age-matched control subjects are shown in Table 1. The FEV1, FEF50, FEF25, and MMEF were significantly lower in patients with chronic embolism than in controls whereas VC, FVC, and DLCO were comparable between the 2 groups, although DLCO was lower in the patients than in controls.

TABLE 2. Number of Segments Showing Mosaic Perfusion, Air Trapping, and Arterial Stenosis or Clot Stenosis/Clot in the Proximal Artery Absent Inspiratory image Normal attenuation Lower attenuation Expiratory image No air trapping Air trapping 30 11 31 10 Present 48 34 42* 40*

Image Analysis
We excluded 39 segments where either the same exact inspiratory-expiratory level was not obtained or a motion artifact was present. In the remaining 123 segments, the crosssectional area change between inspiratory and expiratory im 2003 Lippincott Williams & Wilkins

Note: Figures indicate the number of segment. *The presence of air trapping was associated with the presence of clot in the artery leading to the segment (P < 0.01).

737

Arakawa et al

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

In 13 segments with mosaic perfusion on inspiratory image, lung densities were calculated in adjacent areas with differing attenuation in each segment and were compared (Table 3). The mean lung attenuation of higher attenuation areas on inspiratory images was 815.8 H (SD, 43.2 H) while that of the expiratory images was 636.1 H (SD, 90.1 H). The mean lung attenuation of adjacent lower attenuation areas was 896.2 H (SD, 37.7 H) on inspiratory images and 802.8 H (SD, 102.1 H) on expiratory images. The increase in mean attenuation after exhalation was significantly higher (198.6 H; SD, 77.7H) in the higher attenuation areas than the adjacent areas (89.5 H; SD, 76.4 H) (P < 0.005). The increase in mean lung attenuation after contrast administration was significantly higher for higher attenuation areas (45.6 H; SD, 27.4 H) than for lower attenuation areas (24.8 H; SD, 23.3 H) in the same segments (P < 0.05). Air trapping was noted to be present, at least partially, in 50 segments (40.7%) and was identified in upper lobes in 25, in the right middle lobe in 5, and in lower lobes in 20 segments. In 19 segments, air trapping was identified in areas of lower attenuation on inspiratory images (Fig. 1), but was also noted in 31 areas with normal inspiratory attenuation (Fig. 2). Of these, clot or arterial stenosis was present in 40 segments whereas in 10 segments, no clots or arterial stenosis were identified (Table 2). On the other hand, in the 73 segments without air trapping, clots or arterial stenosis were present on CT in 42 segments and were absent in 31 segments. The presence of air trapping was significantly associated with the presence of clots or stenosis in the artery leading to the segment (P < 0.01). In 15 segments with air trapping on expiratory image, lung densities were calculated in adjacent areas with differing attenuation in each segment and were compared (Table 4) (Fig. 3). In the air trapping areas, mean lung attenuation was 892.7 H (SD, 19.5 H) on inspiratory images while the corresponding area on expiratory image was 827.7 H (SD, 24.9 H). In the adjacent area without air trapping, the mean attenuation was 879.7 H (SD, 18.9 H) on inspiration and 683.6 H (SD, 75.4 H) at expiration. The mean increase in lung attenuation after exhalation was 65.0 H (SD, 25.0 H) and 196.1 H (SD, 61.6 H) in areas with and without air trapping, respectively (P < 0.001).

The increase in mean lung attenuation after contrast administration was 16.2 H (SD, 12.0 H) and 35.0 H (SD, 18.1 H) in areas with air trapping and areas without air trapping, respectively (P < 0.05).

Scoring and Correlation with Pulmonary Function Tests

The air trapping score ranged from 3 to 19 with a mean of 11.3 (SD, 5.2) while the score for the extent of mosaic perfusion ranged from 0 to 24 with a mean of 8.1 (SD, 9.8). The intraclass correlation coefficient for air trapping score was 0.69 while that for lower attenuation area was 0.97. Positive correlation was observed between the air trapping score and that for lower attenuation area (r = 0.565, P < 0.05). Correlation with pulmonary function test results showed significant inverse correlation with FEF25 (r = 0.828, P = 0.006) and MMEF (r = 0.866, P = 0.003), but not with FEV1/FVC (P = 0.116) or FEF50 (P = 0.295). On the other hand, the extent of inspiratory lower attenuation area did not show any significant correlation with pulmonary function parameters. The air trapping score for the control subjects ranged from 0 to 5 (mean, 2.2) and was significantly smaller than in patients with chronic embolism (P < 0.0001). No control subject showed mosaic perfusion.

DISCUSSION
In patients with chronic pulmonary embolic disease, regional hyperventilation and low alveolar carbon dioxide tension have been suggested to cause regional bronchoconstriction and air trapping.16 However, recent research on human subjects suggests that the possible mechanisms are more complex. In pulmonary hypertension, due either to primary or secondary forms, increased expression of endothelin-1 in vascular endothelial cells17 and decreased expression of nitric oxide synthase have both been reported.18 Because endothelin-1 is a potent bronchoconstrictor19 and nitric oxide is a potent bronchodilator,20 both increased expression of endothlin-1 and decreased expression of nitric oxide synthase can cause peripheral airway obstruction.21 Mechanical obstruction of the bron-

TABLE 3. Lung Attenuation of Higher Attenuation Areas and their Adjacent Areas at Inspiration, Expiration, and Contrast Enhancement (n = 13) Inspiration Higher attenuation area Lower attenuation area 815.8 (43.2) 896.2 (37.7) Expiration 636.1 (90.1) 802.8 (102.1) Increase of Lung Attenuation after Expiration 179.7 (76.4)* 93.4 (85.1)* Contrast Enhancement 45.6 (27.4) 24.8 (23.3)

The numbers are expressed by Hounsfield unit. Numbers in parentheses indicated standard deviation. *P < 0.005. P < 0.05.

738

2003 Lippincott Williams & Wilkins

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

Chronic Pulmonary Thromboembolism

chi and bronchioles was also suggested by some authors,9,21 but was not confirmed by pathological studies.22,23 In the mechanical theory, tortuous and irregularly shaped arterial walls can cause mechanical obstruction of the accompanying bronchi running parallel to the occluded and cicatrized artery.9,21 The bronchial walls may become weakened and collapsible

because the bronchial artery delivers the blood flow to the pulmonary parenchyma instead of the bronchial wall.9 Clinical evidence of airway obstruction in chronic embolism has occasionally been reported.79 Remy-Jardin et al,9 using thin-section CT, reported progressive cylindrical dilatation of bronchi distal to occluded pulmonary artery in 21 of 33 patients who were followed for chronic embolism and suggested possible airway changes in this disease.9 Pulmonary function test results in their series showed patients had an FEV1 within the normal range, but a mean value of MMEF was reduced, indicating the presence of small airway obstruction. In another report of 87 patients with chronic embolism, FEV1/FVC and MMEF showed significant decreases with the increasing size of the postembolic perfusion defect assessed by scintigraphy or angiography whereas lung volume was normal.8 These authors suggested a role for airway obstruction in chronic embolism, but also suggested the possible contribution of low physical capacity and pleural adhesion in those patients. Evidence of peripheral airway obstruction has also been reported in 171 patients with primary pulmonary hypertension.21 In that study, FEV1/VC was decreased and airway resistance was mildly increased compared with a control group. However, the reduction of those parameters representative of peripheral airways such as FEF50 and FEF25 was more pronounced. Because air trapping not exceeding one-quarter of the cross-sectional area in each lung can be seen in normal subjects, an air trapping score of up to 6 in our scoring system may not necessarily represent an abnormality.24 Two patients with pulmonary hypertension had an air trapping score <7. On the other hand, 2 patients without pulmonary hypertension had an air trapping score >7. As such, airway obstruction and air trapping in patients with chronic pulmonary embolism cannot simply be explained by coexisting pulmonary hypertension. Multiple factors are likely at play. Areas of air trapping were often found in the areas with lower attenuation on inspiratory images, but were also identified in areas with normal inspiratory lung attenuation. The presence of air trapping was associated with the presence of arterial stenosis or clotting and was significantly more common in the segment distal to the occluded artery than a segment

FIGURE 1. Mosaic perfusion and air trapping in an 81-year-old woman who had deep vein thrombosis and chronic pulmonary embolism. A, CT angiography showed a clot in the right descending artery (arrow). A large clot was identified in the main pulmonary artery as well (not shown). B, Inspiratory thinsection CT image showed mosaic perfusion with higher attenuating area mainly in the central zone of both lower lungs (arrows). C, Expiratory thin-section CT image at the same level as in (B) showed normal increase in lung attenuation in higher attenuation areas. Areas with air trapping were identified in the peripheral zones of both lungs (arrows). 2003 Lippincott Williams & Wilkins

739

Arakawa et al

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

FIGURE 2. Air trapping in a 67-year-old woman who had chronic pulmonary embolism due to repeated thrombosis from deep vein thrombosis. A and B, CT angiography showed a small chronic clot in the posterior segmental branch of the right upper lobe (A) and medial segmental branch of the right middle lobar artery (B) (arrows). A clot was also noted in the descending artery of the right lower lobe. C, Inspiratory thin-section CT image showed linear opacities in the right lung representing parenchymal scarring due to repeated pulmonary embolism. However, lung attenuation was homogeneous (range, 860 H to 905 H). D, Expiratory thin-section CT image showed little increase in lung attenuation in the right upper and middle lobes as well as in the lateral subsegment of the left superior segment (range, 17 H to 44 H increases). The other areas showed normal increase in lung attenuation (range, 187 H to 229 H increases). Although the cross-sectional area change between inspiratory and expiratory images was smaller in the right lung (0.22) than in the left lung (0.35) due to pleural disease, the expiratory image was regarded as adequate for interpretation.

without clots (P < 0.01). Furthermore, CT demonstrated strong negative correlations between the extent of air trapping and pulmonary function parameters of small airways such as FEF25 and MMEF. Correlation with FEV1/FVC and FEF50 did not reach significance, most likely because FEV1/FVC reflects obstruction in the larger airways and the number of cases examined was small. The extent of lower attenuation lung parenchyma on inspiratory CT did not correlate with any pulmonary function parameters, although the extent of lower attenuation weakly correlated with that of air trapping. This result might be explained by the previously reported poor correlation between areas with lower attenuation on CT and decreased pulmonary

perfusion.2,3 As such, we regard this sign as a poor indicator of proximal arterial stenosis. Mosaic perfusion was noted in 6 (66.7%) patients in our series. The increase in mean lung attenuation after contrast administration was significantly higher in the higher attenuation areas than in the lower attenuation areas in the same segments (P < 0.05). Thus, the current study has confirmed the finding of a previous report that those areas that show higher attenuation are more perfused than adjacent lower attenuation areas.3 It should be noted that lung attenuation increase after exhalation was only 83.4 H (SD, 85.1 H) in lower attenuation areas as compared with 179.7 H (SD, 76.4 H) in higher attenuation areas in our study. The result indicates that a significant number
2003 Lippincott Williams & Wilkins

740

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

Chronic Pulmonary Thromboembolism

TABLE 4. Lung Attenuation of Air Trapping Areas and their Adjacent Areas at Inspiration, Expiration, and Contrast Enhancement (n = 15) Inspiration Air trapping area No air trapping area 892.7 (19.5) 879.7 (18.9) Expiration 827.7 (24.9) 683.6 (75.4) Increase of Lung Attenuation after Expiration 65.0 (25.0)* 196.1 (61.6)* Contrast Enhancement 16.2 (12.0) 35.0 (18.1)

The numbers are expressed by Hounsfield unit. Numbers in parentheses indicated standard deviation. *P < 0.001. P < 0.05.

FIGURE 3. Mosaic perfusion due to air trapping in a 67-year-old woman with chronic pulmonary embolism. A, CT angiography showed a large clot in the posterior wall of the right main pulmonary artery. B, CT angiography at the origin of segmental arteries to the left upper lobe showed a chronic arterial clot that was eccentric and contiguous with the vessel wall (arrow). C, Inspiratory thin-section CT image showed mosaic perfusion in both upper lung zones. CT values in the higher and lower attenuation areas in the left upper lobe were 828 H and 914 H, respectively (asterisks). D, Expiratory thin-section CT image at the same level as in (C) showed normal increase in lung attenuation in the higher attenuation areas (around 103 H increase) whereas little increase was noted in the lower attenuation areas (45 H increase), thus confirming air trapping as a cause of lower attenuation on the inspiratory image. After administration of contrast material, the increase in lung attenuation was 34 H in the higher attenuation area in the left upper lobe whereas it was 5 H in the lower attenuation area (air trapping area) in the same lobe. 2003 Lippincott Williams & Wilkins

741

Arakawa et al

J Comput Assist Tomogr Volume 27, Number 5, September/October 2003

of lower attenuation areas with mosaic perfusion were actually the areas with air trapping. One of the limitations of the current study is that the postcontrast enhancement was measured from the data of spiral CT angiography. Because a time-density curve could not be obtained from this data, the measurement of postcontrast enhancement was not reliable at different areas and the maximal contrast enhancement in each area was not obtained. To exclude the possible contribution of time difference to each area, we compared 2 adjacent areas in the same lobe at the same scan level. Using this method, we believe we were successful in determining contrast perfusion in the same segment; however, we were unable to compare the absolute value of contrast enhancement among different categories of lung attenuation as a whole. Another limitation is that we could not and did not detect pulmonary arterial abnormalities beyond the segmental levels, at least potentially explaining the finding of air trapping in areas without intra-arterial clot or stenosis. Finally, the number of patients included in this study was small, and selection bias of the patient cannot be excluded. In summary, we found evidence of airway obstruction in patients with chronic pulmonary embolism by directly showing air trapping on expiratory CT images. Air trapping is not specific to airway diseases and can be seen in chronic pulmonary embolism. The presence of air trapping was significantly associated with the presence of proximal arterial stenosis or clots, and the extent of air trapping was significantly associated with the impairment of the pulmonary function parameters of the small airways in these patients. Air trapping can be regarded as the better indicator of the presence of chronic pulmonary arterial stenosis or clot than the inspiratory CT finding of hypoperfusion. REFERENCES
1. Bergin CJ, Rios G, King MA, et al. Accuracy of high-resolution CT in identifying chronic pulmonary thromboembolic disease. AJR Am J Roentgenol. 1996;166:13711377. 2. King MA, Bergin CJ, Yeung DWC, et al. Chronic pulmonary thromboembolism: detection of regional hypoperfusion with CT. Radiology. 1994;191:359363. 3. Schwickert HC, Schweden F, Schild HH, et al. Pulmonary arteries and lung parenchyma in chronic pulmonary embolism: preoperative and postoperative CT findings. Radiology. 1994;191:351357. 4. Worthy SA, Muller NL, Hartman TE, et al. Mosaic attenuation pattern on

5.

6. 7. 8.

9.

10.

11.

12.

13.

14. 15.

16.

17.

18.

19. 20.

21. 22. 23. 24.

thin-section CT scans of the lung: differentiation among infiltrative lung, airway, and vascular diseases as a cause. Radiology. 1997;205:465470. Im JG, Choi YW, Kim HD, et al. Thin-section CT findings of the lungs: experimentally induced bronchial and pulmonary artery obstruction in pigs. AJR Am J Roentgenol. 1996;167:631636. Arakawa H, Kurihara Y, Sasaka K, et al. Air trapping on CT of patients with pulmonary embolism. AJR Am J Roentgenol. 2002;178:12011207. Colp CR, Williams MHJ. Pulmonary function following pulmonary embolization. Am Rev Respir Dis. 1962;85:799807. Riedel M, Stanek V, Widimsky J. Spirometry and gas exchange in chronic pulmonary thromboembolism. Bull Eur Physiopathol Respir. 1981;17: 209221. Remy-Jardin M, Remy J, Louvegny S, et al. Airway changes in chronic pulmonary embolism: CT findings in 33 patients. Radiology. 1997;203: 355360. Remy-Jardin M, Remy J, Deschildre F, et al. Diagnosis of pulmonary embolism with spiral CT: comparison with pulmonary angiography and scintigraphy. Radiology. 1996;200:699706. Auger W, Fedullo P, Moser K, et al. Chronic major-vessel thromboembolic pulmonary artery obstruction: appearance at angiography. Radiology. 1992;182:393398. Stern EJ, Webb WR, Gamsu G. Dynamic quantitative computed tomography. A predictor of pulmonary function in obstructive lung diseases. Invest Radiol. 1994;29:564569. Webb WR, Muller NL, Naidich DP. Standardized terms for highresolution computed tomography of the lung: a proposed glossary. J Thorac Imaging. 1993;8:167185. Fermanian J. The degree of concordance between observers: quantitative case. Rev Epidemiol Sante Publique. 1984;32:408413. Lucidarme O, Coche E, Cluzel P, et al. Expiratory CT scans for chronic airway disease: correlation with pulmonary function test results. AJR Am J Roentgenol. 1998;170:301307. Wilhelmsen L, Bjure J, Korsgren M, et al. Pathophysiology of pulmonary embolism with special reference to recurrent embolism. Bull Physiopathol Respir. 1970;6:99121. Giaid A, Yanagisawa M, Langleben D, et al. Expression of endothelin-1 in the lungs of patients with pulmonary hypertension. N Engl J Med. 1993; 328:17321739. Giaid A, Saleh D. Reduced expression of endothelial nitric oxide synthase in the lungs of patients with pulmonary hypertension. N Engl J Med. 1995; 333:214221. Fagan KA, McMurtry IF, Rodman DM. Role of endothelin-1 in lung disease. Respir Res. 2001;2:90101. Belvisi MG, Stretton CD, Yacoub M, et al. Nitric oxide is the endogenous neurotransmitter of bronchodilator nerves in humans. Eur J Pharmacol. 1992;210:221222. Meyer FJ, Ewert R, Hoeper MM, et al. Peripheral airway obstruction in primary pulmonary hypertension. Thorax. 2002;57:473476. Wagenvoort CA. Pathology of pulmonary thromboembolism. Chest. 1995;107:10S17S. Presti B, Berthrong M, Sherwin RM. Chronic thrombosis of major pulmonary arteries. Hum Pathol. 1990;21:601606. Lee KW, Chung SY, Yang I, et al. Correlation of aging and smoking with air trapping at thin-section CT of the lung in asymptomatic subjects. Radiology. 2000;214:831836.

742

2003 Lippincott Williams & Wilkins