Cardiopulmonary Imaging Original Research

Transcription

1 Cardiopulmonary Imaging Original Research Iwasawa et al. CT in Pulmonary Hypertension Cardiopulmonary Imaging Original Research Tae Iwasawa 1 Shingo Kato 2 Takashi Ogura 3 Yuka Kusakawa 2 Shinichiro Iso 1,4 Tomohisa Baba 3 Kazuki Fukui 2 Mari S. Oba 5 Iwasawa T, Kato S, Ogura T, et al. Keywords: computer-aided design, hypertension, idiopathic pulmonary fibrosis, interstitial, lung disease, MDCT, pulmonary DOI: /AJR Received June 13, 2013; accepted after revision October 19, Supported by the Kanagawa Cancer Research Fund. 1 Department of Radiology, Kanagawa Cardiovascular and Respiratory Center, , Tomioka-higashi, Kanazawa-ku, Yokohama, Kanagawa , Japan. Address correspondence to T. Iwasawa (tae_i_md@wb3.so-net.ne.jp). 2 Department of Cardiology, Kanagawa Cardiovascular and Respiratory Center, Kanagawa, Japan. 3 Department of Respiratory Medicine, Kanagawa Cardiovascular and Respiratory Center, Kanagawa, Japan. 4 Department of Radiology, Yokohama Rousai Hospital for Labour Welfare Corporation, Yokohama, Japan. 5 Department of Biostatistics and Epidemiology, Yokohama City University School of Medicine, Yokohama, Japan. WEB This is a web exclusive article. AJR 2014; 203:W166 W X/14/2032 W166 American Roentgen Ray Society Low-Normal Lung Volume Correlates With Pulmonary Hypertension in Fibrotic Idiopathic Interstitial Pneumonia: Computer-Aided 3D Quantitative Analysis of Chest CT OBJECTIVE. We investigated whether the lung volume determined on CT, especially the volume of the normal lung, is correlated with mean pulmonary artery pressure (PAP) in patients with chronic fibrosing idiopathic interstitial pneumonia (IIP). MATERIALS AND METHODS. The subjects were 40 patients with IIP who underwent right heart catheterization (RHC) and chest CT. Thirty-three patients (82.5%) were smokers or former smokers. Using a computer-aided system, the lungs in the 3D CT images were automatically categorized pixel-by-pixel with gaussian histogram-normalized correlations, and the relative volume of each lesion to the CT lung volume was calculated as normal(%), ground-glass opacities(%), consolidation(%), emphysema(%), and fibrosis(%). The relationship between each volume(%) and pulmonary hypertension was evaluated using logistic regression analysis. ROC curves were constructed to assess the predictive value of these CT-based volumes in the identification of pulmonary hypertension. RESULTS. Sixteen patients had pulmonary hypertension at rest (mean PAP > 25 mm Hg on RHC). Emphysema constituted more than 10% of the CT lung volume in 13 patients. On multivariate analysis of each volume(%), normal(%) was significant for detecting pulmonary hypertension (odds ratio, 0.92; 95% CI, ; p = 0.02). On ROC analysis, the AUC of normal(%) was ( ). CONCLUSION. The relative CT volume of any single lesion was of limited value in predicting pulmonary hypertension in patients with pulmonary fibrosis and emphysema. In these patients, normal(%), measured by a 3D computer-aided system, was correlated with pulmonary hypertension measured by RHC. P ulmonary arterial hypertension is an unfavorable complication in idiopathic interstitial pneumonia (IIP) [1]. Idiopathic pulmonary fibrosis (IPF) is the most common form of chronic fibrosing interstitial pneumonia in the American Thoracic Society/European Respiratory Society (ATS/ERS) classification of IIP [2]. The reported prevalence of pulmonary hypertension in patients with IPF ranges from 32% to 85% [3 7]. The development of pulmonary hypertension in IIP patients has important prognostic implications [4]. Lettieri et al. [4] reported that 1-year mortality rates of patients with IPF were higher in those with pulmonary hypertension compared with those without pulmonary hypertension (28.0% vs 5.5%, respectively; p = 0.002). Although early diagnosis of pulmonary hypertension is important, the overlapping main symptoms shortness of breath and exercise limitation make it difficult to detect pulmonary hypertension in patients with IIP [8]. Right heart catheterization (RHC) is the most accurate diagnostic tool for pulmonary hypertension. However, because RHC is invasive and expensive, a reliable and noninvasive surrogate marker is desirable. CT is an essential imaging modality and widely performed in patients with IIP [2, 9]. The overall extent of pulmonary fibrosis on CT (the combination of reticulation and honeycomb changes) correlates with disease severity, and the extent of honeycombing and reticulation on CT increases with disease progression [10, 11]. Recent studies have suggested that the biologic process underlying progressive fibrosis also contributes to vascular remodeling and pulmonary hypertension [3]. If fibrogenesis correlates directly with the biologic mechanism involved in the induction of pulmonary hypertension, the se- W166 AJR:203, August 2014

2 CT in Pulmonary Hypertension verity of lung fibrosis should correlate with the prevalence and extent of pulmonary hypertension. However, previous studies have concluded that the CT score, an index of fibrosis using 25% steps, does not correlate with pulmonary hypertension [12, 13]. These somewhat disappointing results are likely due to emphysema. In smokers, some emphysema may be associated with interstitial pneumonia [2, 14]. Cottin et al. [15, 16] reported a high prevalence of pulmonary hypertension in patients with pulmonary fibrosis combined with emphysema. Emphysema is also a well-known etiologic factor of pulmonary hypertension [17]. In patients with lung damage caused by both fibrosis and emphysema, it is difficult to isolate the main contributor to pulmonary hypertension. We hypothesized that normal lung volume correlates negatively with the severity of pulmonary hypertension. In many of the previous studies on this topic, the main target of CT evaluation in IIP was the extent of fibrosis; normal lung volume was not measured directly [14, 18]. In this study, we examined whether the CT-determined lesion volume data, especially normal lung, correlated with pulmonary hypertension in patients with IIP. Materials and Methods Subjects This retrospective cross-sectional study using CT data was approved by the institutional review board, and informed consent was waived. To avoid selection bias, all patients with pulmonary fibrosis who underwent RHC between January 2011 and A Fig year-old man with idiopathic pulmonary fibrosis combined with emphysema. Nonclassifiable fibrosis was confirmed on surgical biopsy. Patient was former smoker (30 pack-years). Mean pulmonary artery pressure was 20 mm Hg measured by right heart catheterization. A, Three-dimensional CT image analysis shows emphysema volume (%LAA-950) 29.9%, fibrosis 11.6%, and normal 52.5%. B and C, Sagittal reformatted CT images show computer-aided segmentation result on sagittal images corresponding to A. Pink = normal, dark blue = emphysema, light green = ground-glass opacity, light blue and yellow = fibrosis, dark green = trachea and bronchi, orange = vessels. April 2013 were candidates for this study. Seventy-eight patients with pulmonary fibrosis (or suspected pulmonary fibrosis) underwent RHC during this period. Patients with fibrotic IIP who underwent CT within 6 months of RHC were included. Patients with heart disease or pulmonary vascular disease (heart valve insufficiency [n = 4], left heart disease [pulmonary arterial wedge pressure > 15 mm Hg, n = 4], and chronic pulmonary arterial thromboembolism [n = 3]) were excluded. Patients with lung cancer (including postoperative state, n = 8) and secondary pulmonary fibrosis (due to collagen disease or vasculitis, n = 14) as well as four patients who did not undergo CT within 6 months of RHC were also excluded. One patient with pneumothorax was excluded because the lesion volume on CT could not be accurately estimated due to collapse of the lung. Thus, 40 consecutive patients with chronic fibrosing interstitial pneumonia were included. Of these 40 patients, 33 (82.5%) were smokers or former smokers (mean smoking history, 36.8 pack-years; median, 35.8 pack-years; range, pack-years and median interval between stopping smoking and RHC, 10.5 years; range, 0 58 years). All 40 patients had demonstrable interstitial abnormalities suggestive of pulmonary fibrosis in the lower lung zones on CT images. Nineteen patients (47.5%) had undergone surgical lung biopsy and were diagnosed histopathologically with pulmonary fibrosis (usual interstitial pneumonia [UIP] pattern [n = 12], fibrosing nonspecific interstitial pneumonia [NSIP] pattern [n = 5], and nonclassifiable fibrosis [n = 2]). According to the criteria developed by the ATS [2, 9, 19], five patients with pathologically fibrotic NSIP were diagnosed B as having idiopathic fibrosing NSIP, and 35 patients were clinically diagnosed as having IPF. RHC was performed in resting patients using a standard technique. Pulmonary hypertension was defined as mean pulmonary artery pressure (PAP) > 25 mm Hg at rest in the setting of a normal pulmonary arterial wedge pressure of 15 mm Hg [1]. Pulmonary function tests (PFTs) were conducted in only 32 of 40 patients within 6 months of RHC. The median interval between PFT and RHC was 42 days (range, days). CHESTAC-8800, CHESTAC-33 (Chest MI), and Fudac-77 (Fukuda Denshi) were used to measure vital capacity (VC), forced expiratory volume in 1 second (FEV 1 ), total lung capacity (TLC), and diffusing capacity using standard measurement techniques [20]. The results are expressed as percentage of predicted performance using standard values [20]. CT and Analysis by Radiologists Thin-section CT images were obtained during inspiration in the supine position using a 64- MDCT scanner (Aquilion-64, Toshiba). The median interval between CT and RHC was 31 days (mean, 45 days; range, days). All CT images were obtained without contrast administration. Slice thickness was 0.5 mm (n = 31) or 1 mm (n = 9). Images were reconstructed with standard and high-resolution algorithms. The diameters of the pulmonary artery and ascending aorta were measured at the widest section as described by Wells et al. [21]. One author (a board-certified radiologist with more than 20 years of experience) measured the diameter of the main pulmonary artery at the level of its bifurcation and the diameter of the ascending aorta at its C AJR:203, August 2014 W167

3 Iwasawa et al. maximum dimension using a CT viewer and an electrical caliper (Synapse, Fujifilm). The extent of parenchymal abnormality was determined for each lung using a 5-point scale as described previously with some modifications (0, no involvement; 1, 1 25% involvement; 2, 26 50% involvement; 3, 51 75% involvement, and 4, % involvement) [12, 13, 22]. Images were evaluated independently by two board-certified chest radiologists with 20 and 10 years of experience who were blinded to the clinical information. Each lung was scored in three zones (upper zone, lung apex to aortic arch; middle zone, aortic arch to a position inferior to the pulmonary veins; and lower zone, from the inferior pulmonary veins to the diaphragm). The mean score for each of the six zones was calculated for each parenchymal pattern. Fibrosis (reticulation, honeycombing, and architectural distortion), emphysema (low-attenuation area without a thick wall), and ground-glass opacity (GGO) (hazy parenchymal opacity in the absence of reticular opacity or architectural distortion) were evaluated [12]. The sum of these scores was called the total CT score. Interobserver variation of these scores was analyzed with the kappa statistic. Interobserver agreement was classified as slight (κ = ), fair (κ = ), moderate (κ = ), substantial (κ = ), or almost perfect (κ = ). The kappa value was 0.65 for the emphysema score, 0.49 for the fibrosis score, 0.56 for the GGO score, and 0.52 for the total score. There was moderate to substantial agreement for each score. A final score was obtained by taking the average of the scores of the individual observers. A Fig year-old man with idiopathic pulmonary fibrosis combined with emphysema. Patient was former smoker (45 pack-years). Mean pulmonary artery pressure was 29 mm Hg measured by right heart catheterization. A, Image from computer-aided analysis shows emphysema volume (%LAA-950) 32.5%, fibrosis 16.0%, and normal 35.7%. B and C, Sagittal reformatted CT images show computer-aided segmentation results corresponding to A. Sagittal images show decreased normal lung compared with case shown in Figure 3. Pink = normal, dark blue = emphysema, light green = ground-glass opacity, light blue and yellow = fibrosis, dark green = trachea and bronchi, orange = vessels. Volume Analysis on CT All CT images were transferred to a PC (Build- To-Order PC with Intel Core i7 CPU 950 at 3.07 GHz using Microsoft Windows 7 64 bit). Before the segmentation process, the CT values were corrected by the mean attenuation value in tracheal gas of 1000 HU [23]. The lung on thin-section CT was extracted using a semiautomated threshold technique that selects all pixels between 200 and 1024 HU. The mathematic morphologic closing and opening techniques were then applied to include the areas of subpleural consolidation or thick reticulation with more than 200 HU. In some cases, residual subpleural lesions needed to be included in the lung manually using a paint tool. After that, the bronchial trees and blood vessels were excluded using the failure-recovery algorithm reported by Iwao et al. [24]. The algorithm consists of region growing, failure detection, history storage, and backtracking modules. It is possible to backtrack to a previous stable state using history storage and restart the processing from the state when a failure is detected. Consequently, tracking with high accuracy proceeds at a low computational cost. After each lesion was extracted from the lung, it was segmented using two types of procedures: one was the simple threshold technique designed to measure emphysema volume as %LAA-950 [25], and the other was a gaussian histogramnormalized correlation system for measuring the volume of each lesion [26, 27]. Briefly, this system divides lung pixels into five categories along predesigned samples using CT attenuation values and their local histograms. The five categories were as follows: normal, emphysema, GGO, consolidation, and fibrosis. The volume of each B lesion and total CT lung volume were computed automatically by the computer-aided diagnosis (CADx) system. CT lung volume was expressed as both an absolute value and relative to the predicted total lung capacity ( %predtlc ). The volume of each lesion was expressed relative to CT lung volume (expressed as a percentage). Statistical Analysis The subjects were divided into two groups (with and without pulmonary hypertension). Differences in patients characteristics and CT results were tested for significance using the Mann- Whitney U test. The relationship between mean PAP and CT findings was examined by Spearman correlation analysis. Univariate and multivariate regression analyses were also performed to assess the independent predictive values of CT measurements. ROC curves were used to compare the diagnostic capability of CT results. The cutoff value that yielded the highest sensitivity and specificity was identified. The highest negative predictive values were also considered on the basis of the use of CT as a screening test in the diagnosis of pulmonary hypertension. This was followed by statistical comparisons of sensitivity, specificity, and accuracy using the McNemar test. All statistical analyses were performed using SPSS software, version 21. A p value less than 0.05 was considered significant. Results High mean PAP (> 25 mm Hg) was identified on RHC in 16 of 40 patients (the pulmonary hypertension group). The mean PAP in the remaining 24 subjects was 25 mm Hg, C W168 AJR:203, August 2014

4 CT in Pulmonary Hypertension and they were categorized as the nonpulmonary hypertension group [1]. Tables 1 and 2 compare the PFT and CT findings in the two groups. There were no significant differences between the two groups with regard to age, smoking habit, and time between CT and RHC. The median diffusing capacity was significantly worse in patients with pulmonary hypertension. Table 2 and Figures 1 and 2 show the results of CT analysis. The median diameter of the pulmonary artery was larger in patients with pulmonary hypertension (34.7 mm; range, mm) than in those without pulmonary hypertension (30.4 mm; range, mm) (p = 0.002). There was no significant difference between the two groups in fibrosis and GGO scores. However, the median total score was significantly greater in patients with pulmonary hypertension (4.1; range, ) than in the nonpulmonary hypertension group (3.1; range, ) (p = 0.001). Of the 40 patients, 13 had emphysema of more than 10% on %LAA-950 analyses, but only eight of these 13 patients had pulmonary hypertension. TABLE 1: Comparison of Physiologic Data Between Patients With and Without Pulmonary Hypertension Characteristic Total (n = 40) Without Pulmonary Hypertension Mean PAP 25 mm Hg (n = 24) With Pulmonary Hypertension Mean PAP > 25 mm Hg (n = 16) Mean PAP (mm Hg) 21 (7 48) 16.5 (7 25) 31 (26 48) < Age (y) 70.5 (53 82) 71.5 (54 81) 70 (53 82) 0.76 Sex (M/F) 29/11 16/8 13/ Smoking (pack-years) 35.8 (0 94.5) 33.8 (0 94.5) 37.3 (0 78) 0.99 Never-smokers/former smokers 7/33 5/19 2/ Surgical biopsy done (UIP/fibrosing NSIP/nonclassifiable) 12/5/2 9/3/2 3/2/0 No. of patients with PFT Time between PFT and RHC (d) 42 (1 130) 43 (1 130) 40.5 (3 129) 0.58 Predicted vital capacity(%) 76.6 ( ) 78.0 ( ) 72.2 ( ) 0.19 Predicted FEV 1 (%) 78.3 (36 120) 80.9 ( ) 68.6 (36 120) 0.13 Predicted TLC(%) 73.8 (42 128) 73.8 (42 128) 58.9 ( ) 0.25 Diffusion capacity of lung for carbon monoxide (%) 45.4 ( ) 55.2 ( ) 36.9 ( ) 0.04 Note Data in parentheses are range. PAP = pulmonary artery pressure, UIP = usual interstitial pneumonia, NSIP = nonspecific interstitial pneumonia, PFT = pulmonary function test, RHC = right heart catheterization, FEV 1 = forced expiratory volume in 1 second, TLC = total lung capacity TABLE 2: Comparison of CT Results Between Patients With and Without Pulmonary Hypertension Characteristic Total (n = 40) Without Pulmonary Hypertension Mean PAP 25 mm Hg (n = 24) With Pulmonary Hypertension Mean PAP > 25 mm Hg (n = 16) Time between CT and RHC (d) 31 (0 149) 48.5 (3 135) 24 (0 149) 0.32 Pulmonary artery diameter (mm) 32.6 ( ) 30.4 ( ) 34.7 ( ) Aorta diameter (mm) 34.9 ( ) 34.9 ( ) 34.9 ( ) 0.57 Emphysema score 1.3 (0 3.7) 1.0 (0 2.5) 1.9 (0 3.7) 0.07 Fibrosis score 1.2 (0.4 2) 1.1 (0.4 2) 1.3 (0.4 2) 0.14 GGO score 0.8 (0 3.5) 0.9 ( ) 0.7 (0 3.5) 0.92 Total score 3.5 ( ) 3.1 ( ) 4.1 ( ) Percentage emphysema volume (LAA-950) 6.8 (0 43.4) 4.7 (0 29.9) 9.8 (0 43.4) 0.09 Percentage LAA-950 > 10% CT lung volume 3.6 ( ) 3.9 ( ) 3.2 ( ) 0.46 CT lung volume (%predtlc) 70.3 ( ) 70.6 ( ) 66.6 ( ) 0.41 Normal(%) 56.8 ( ) 60.7 ( ) 46.9 ( ) < Emphysema(%) 7.7 (0 57.3) 5.2 (0 31.6) 10.8 (0 57.3) 0.10 GGO(%) 10.2 (4 39.2) 9.4 ( ) 12.8 (4 39.2) 0.05 Consolidation(%) 0.2 (0 2.7) 0.2 (0 2.7) 0.3 (0 2.6) 0.14 Fibrosis(%) 18.8 ( ) 16.2 ( ) 23.8 (9 48.6) 0.03 Note Data in parentheses are range. PAP = pulmonary artery pressure, RHC = right heart catheterization, %predtlc = percentage to predicted total lung capacity, GGO = ground-glass opacity. p p AJR:203, August 2014 W169

5 Iwasawa et al. TABLE 3: Results of Univariate and Multivariate Logistic Analyses Characteristic On gaussian histogram-normalized correlation analysis, the median normal(%) was smaller in patients with pulmonary hypertension than in those without pulmonary hypertension (46.9%; range, % vs 60.7%; range, %; p < 0.001). The median fibrosis(%) was larger in patients with pulmonary hypertension than in those without pulmonary hypertension (23.8%; range, % vs 16.2%; range, %; p = 0.03). The other gaussian histogram-normalized correlation results were not significantly different between the two groups (Table 2). Figure 3 shows the relationship between mean PAP and CT findings. Normal(%) (r = 0.708) was negatively correlated with mean PAP (p < 0.001). Fibrosis(%) (r = 0.401, p = 0.001) and emphysema(%) (r = 0.403, p = 0.01) were positively correlated with mean PAP, whereas CT lung volume and CT lung volume ( %predtlc ) were not correlated with mean PAP. The %LAA-950 (r = 0.410, p = 0.009) was positively correlated with mean PAP. Diffusing capacity (%pred) (r = 0.427, p = 0.015) was negatively correlated with mean PAP. The CT total score was correlated with mean PAP (r = 0.630, p < 0.001), and diameter of the pulmonary artery (r = 0.444, p = 0.004) was positively correlated with mean PAP. Table 3 shows the results of the univariate and multivariate logistic regression analyses. Univariate analysis showed that diameter of the pulmonary artery, emphysema score, total score, normal(%), and fibrosis(%) were significant variables for detecting pulmonary hypertension. Variables Univariate Analysis Multivariate Analysis Odds Ratio 95% CI p Odds Ratio 95% CI p Predicted diffusing capacity (%) (n = 32) Diameter of pulmonary artery Emphysema score Fibrosis score GGO score Total score Percentage emphysema volume (LAA-950) CT-based volume Normal(%) Emphysema(%) GGO(%) Consolidation(%) Fibrosis(%) Note GGO = ground-glass opacity. were considered for multivariate models if their p values were less than The diameter of the pulmonary artery and total score showed no significant correlation, whereas diameter of the pulmonary artery and normal(%) showed a significant weak correlation (r = 0.41, p = 0.008) on Spearman correlation analysis. However, total score was highly significantly correlated with normal(%) (r = 0.84, p < 0.001). The number of patients in our study was not high enough to enable us to distinguish between total score and normal(%). The two variables of diameter of the pulmonary artery and normal(%) were then investigated, and normal(%) was a significant variable for predicting pulmonary hypertension (odds ratio, 0.92; range, ; p = 0.02). Figure 4 shows the ROC curves for the presence of pulmonary hypertension and CT volume analysis and diameter of the pulmonary artery. The AUC was (95% CI, ) for normal(%), ( ) for total score, and ( ) for diameter of the pulmonary artery. Table 4 lists the sensitivity, specificity, and accuracy for CT evaluation. When a cutoff value of 51% was used for normal(%), the accuracy was 77.5%. When a cutoff value of 3.4 was used for the CT total score, the accuracy was 70.0%. Discussion In this study, normal lung volume, represented by normal(%) on 3D CT images, correlated negatively and significantly with pulmonary hypertension in patients with chronic fibrosing interstitial pneumonia. On logistic regression analysis, decreased normal(%) accurately predicted pulmonary hypertension in our subjects. On the other hand, the results showed a weak correlation between mean PAP and fibrosis(%). Total lung volume measured on CT and expressed as CT lung volume (%predtlc) did not correlate with mean PAP. These results are explainable by the coexistence of emphysema. Cottin et al. [16] proposed pulmonary fibrosis combined with emphysema as an im- TABLE 4: Diagnostic Accuracy of CT Parameters for Predicting Pulmonary Hypertension Parameter Cutoff Value Sensitivity (%) Specificity (%) Negative Predictive Value (%) Accuracy (%) Normal(%) < (10/16) 87.5 (21/24) 77.8 (21/27) 77.5 (31/40) Total score > (13/16) 62.5 (15/24) 83.3 (15/18) 70.0 (28/40) Diameter of pulmonary artery (mm) > (15/16) 54.2 (13/24) 92.9 (13/14) 70.0 (28/40) Note Data in parentheses are number/total. W170 AJR:203, August 2014

6 CT in Pulmonary Hypertension Normal (%) %LAA Mean PAP (mm Hg) portant phenotype of pulmonary fibrosis. Hyperinflation and high compliance of the emphysematous areas of the lungs probably compensate for the volume loss due to fibrosis [28], whereas pulmonary emphysema and fibrosis may have additive or synergistic effects on the capillary beds of the lung, resulting in low diffusing capacity [16, 18]. In the Sensitivity Mean PAP (mm Hg) Specificity A C Fibrosis (%) CT Lung Volume (%) Mean PAP (mm Hg) Mean PAP (mm Hg) Fig. 4 Graph shows ROC curve analysis for data derived from thin-section CT to predict pulmonary hypertension. Thick line indicates normal lung volume on CT (normal(%)) (AUC, 0.849; 95% CI, ). Dotted line indicates total score (AUC, 0.827; ). Dashed line indicates diameter of pulmonary artery (AUC, 0.789; ). B D Fig. 3 Correlation between mean pulmonary artery pressure (PAP) and CT findings. A D, Graphs show that CT-based volume (normal(%)) (A) correlates negatively with mean PAP (r = 0.708, p < 0.001), fibrosis (fibrosis(%)) (B) correlates with mean PAP (r = 0.401, p = 0.001), emphysema volume (%LAA-950) (C) correlates with mean PAP (r = 0.410, p = 0.009), and CT lung volume (D) does not correlate with mean PAP. Solid circles represent patients with emphysema of more than 10% in %LAA and open circles, patients without or with small volume of emphysema. current study, 82.5% of subjects were smokers or former smokers. Thirteen patients had emphysema with %LAA-950 of more than 10% and met the criteria of pulmonary fibrosis combined with emphysema defined by Ryerson et al. [18] and Mejía et al. [14]. These patients had large CT lung volume and low fibrosis(%) compared with patients without emphysema (Fig. 1). The results indicate that the score of a single lesion (emphysema only or fibrosis only) in pulmonary fibrosis combined with emphysema patients is not sufficient for the expression of disease severity. We hypothesized that normal lung volume is an important index in these patients. Previous studies on the relationship between IIP and pulmonary hypertension did not investigate normal lung volume, although the volume of functional lung (i.e., lung volume minus emphysema volume, tumor volume, or atelectatic lung volume) has been applied to patients with lung cancer and chronic obstructive pulmonary disease [29, 30]. Recent studies have proposed several methods of segmenting 3D CT images of the lungs in patients with IIP [24, 31 35], and some studies have measured normal lung volume in patients with pulmonary fibrosis using various computer-aided systems [27, 34]. In the current study, the sagittal color images (Figs. 1 and 2) clearly showed the craniocaudal extent of each le- AJR:203, August 2014 W171

7 Iwasawa et al. sion, and the system used for analysis was helpful for the understanding of residual normal lung volume. In previous studies, the fibrosis score on CT images did not correlate with pulmonary hypertension [12, 13]. Unfortunately, the study by Zisman et al. [12] did not include emphysema. The study by Alhamad et al. [13] included only a small proportion of smokers, and they analyzed patients with various diseases, including collagen vascular diseases. In patients with collagen vascular diseases, increased pulmonary vascular resistance is probably multifactorial, including chronic pulmonary arterial thromboemboli and pulmonary venoocclusive disease [36]. Although our study included only patients with IIP and most subjects were former smokers, we believe that the results are consistent with these previous studies. The results of the CADx system and radiologist scores in our study are highly correlated with each other [27, 37]. The fibrosis score correlated significantly with fibrosis(%) (r = 0.716, p < 0.001). Fibrosis(%) was significant for predicting pulmonary hypertension on univariate analysis, although the p value was low. The fibrosis score was not a significant variable. The small number of patients involved in the study may have affected the results. In binary analysis, any single data value plays a key role in a small study. This study has certain limitations. The study design was retrospective, and the study was based on a small number of patients from a single institution. Patients were continuously enrolled during the study, but selection bias cannot be ruled out because not all patients with IIP in our hospital underwent RHC. Further studies of larger numbers of patients are needed to confirm the utility of 3D CT analysis in patients with IIP. In conclusion, the results showed that normal lung volume measured on 3D CT, represented by normal(%), correlated with mean PAP measured by RHC. The noninvasive and accessible nature of CT makes it an attractive tool for predicting pulmonary hypertension. The pathobiology of pulmonary hypertension in pulmonary fibrosis is not completely understood [38]. However, our results suggest that quantitative measurement of normal lung volume can be useful for predicting pulmonary hypertension in patients with chronic fibrosing interstitial pneumonia, especially those with coexisting emphysema. References 1. McLaughlin VV, Archer SL, Badesch DB, et al. ACCF/AHA 2009 expert consensus document on pulmonary hypertension: a report of the American College of Cardiology Foundation Task Force on Expert Consensus Documents and the American Heart Association developed in collaboration with the American College of Chest Physicians; American Thoracic Society, Inc.; and the Pulmonary Hypertension Association. J Am Coll Cardiol 2009; 53: Travis WD, Costabel U, Hansell DM. An official American Thoracic Society/European Respiratory Society Statement: update of the International Multidisciplinary Classification of the Idiopathic Interstitial Pneumonias. Am J Respir Crit Care Med 2013; 188: Farkas L, Gauldie J, Voelkel NF, Kolb M. Pulmonary hypertension and idiopathic pulmonary fibrosis: a tale of angiogenesis, apoptosis, and growth factors. Am J Respir Cell Mol Biol 2011; 45: Lettieri CJ, Nathan SD, Barnett SD, Ahmad S, Shorr AF. Prevalence and outcomes of pulmonary arterial hypertension in advanced idiopathic pulmonary fibrosis. Chest 2006; 129: Nathan SD, Noble PW, Tuder RM. Idiopathic pulmonary fibrosis and pulmonary hypertension: connecting the dots. Am J Respir Crit Care Med 2007; 175: Nathan SD, Shlobin OA, Ahmad S, et al. Serial development of pulmonary hypertension in patients with idiopathic pulmonary fibrosis. Respiration 2008; 76: Shorr AF, Wainright JL, Cors CS, Lettieri CJ, Nathan SD. Pulmonary hypertension in patients with pulmonary fibrosis awaiting lung transplant. Eur Respir J 2007; 30: Behr J, Ryu JH. Pulmonary hypertension in interstitial lung disease. Eur Respir J 2008; 31: Raghu G, Collard HR, Egan JJ, et al. An official ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fibrosis evidence-based guidelines for diagnosis and management. Am J Respir Crit Care Med 2011; 183: Best AC, Meng J, Lynch AM, et al. Idiopathic pulmonary fibrosis: physiologic tests, quantitative CT indexes, and CT visual scores as predictors of mortality. Radiology 2008; 246: Lee HY, Lee KS, Jeong YJ, et al. High-resolution CT findings in fibrotic idiopathic interstitial pneumonias with little honeycombing: serial changes and prognostic implications. AJR 2012; 199: Zisman DA, Karlamangla AS, Ross DJ, et al. High-resolution chest CT findings do not predict the presence of pulmonary hypertension in advanced idiopathic pulmonary fibrosis. Chest 2007; 132: Alhamad EH, Al-Boukai AA, Al-Kassimi FA, et al. Prediction of pulmonary hypertension in patients with or without interstitial lung disease: reliability of CT findings. Radiology 2011; 260: Mejía M, Carrillo G, Rojas-Serrano J, et al. Idiopathic pulmonary fibrosis and emphysema: decreased survival associated with severe pulmonary arterial hypertension. Chest 2009; 136: Cottin V, Le Pavec J, Prevot G, et al. Pulmonary hypertension in patients with combined pulmonary fibrosis and emphysema syndrome. Eur Respir J 2010; 35: Cottin V, Nunes H, Brillet PY, et al. Combined pulmonary fibrosis and emphysema: a distinct underrecognised entity. Eur Respir J 2005; 26: Adir Y, Shachner R, Amir O, Humbert M. Severe pulmonary hypertension associated with emphysema: a new phenotype? Chest 2012; 142: Ryerson CJ, Hartman T, Elicker BM, et al. Clinical features and outcomes in combined pulmonary fibrosis and emphysema in idiopathic pulmonary fibrosis. Chest 2013; 144: Travis WD, Hunninghake G, King TE Jr, et al. Idiopathic nonspecific interstitial pneumonia: report of an American Thoracic Society project. Am J Respir Crit Care Med 2008; 177: The Japanese Respiratory Society. Guideline for pulmonary function test. Tokyo, Japan: Medical View, Wells JM, Washko GR, Han MK, et al. Pulmonary arterial enlargement and acute exacerbations of COPD. N Engl J Med 2012; 367: Kazerooni EA, Martinez FJ, Flint A, et al. Thinsection CT obtained at 10-mm increments versus limited three-level thin-section CT for idiopathic pulmonary fibrosis: correlation with pathologic scoring. AJR 1997; 169: Best AC, Lynch AM, Bozic CM, Miller D, Grunwald GK, Lynch DA. Quantitative CT indexes in idiopathic pulmonary fibrosis: relationship with physiologic impairment. Radiology 2003; 228: Iwao Y, Gotoh T, Kagei S, Iwasawa T. Tsuzuki SG. Integrated lung field segmentation of injured regions and anatomical structures from chest CT images biological and medical systems. (abstr) Proceedings of 8th IFAC Symposium on Biological and Medical Systems. Budapest, Hungary: Elsevier, 2012: Han MK, Kazerooni EA, Lynch DA, et al. Chronic obstructive pulmonary disease exacerbations in the COPDGene study: associated radiologic phenotypes. Radiology 2011; 261: Asakura A, Gotoh T, Iwasawa T, Saito K, Kasaka H. Classification system of the CT images with nonspecific interstitial pneumonia. J Inst Image Electron Eng Jpn 2004; 33: Iwasawa T, Asakura A, Sakai F, et al. Assessment of prognosis of patients with idiopathic pulmonary fibrosis by computer-aided analysis of CT images. J Thorac Imaging 2009; 24: W172 AJR:203, August 2014

8 CT in Pulmonary Hypertension 28. Doherty MJ, Pearson MG, O Grady EA, Pellegri- 31. Rosas IO, Yao J, Avila NA, Chow CK, Gahl WA, 35. Yoon RG, Seo JB, Kim N, et al. Quantitative as- ni V, Calverley PM. Cryptogenic fibrosing alveo- Gochuico BR. Automated quantification of high- sessment of change in regional disease patterns on litis with preserved lung volumes. Thorax 1997; resolution CT scan findings in individuals at risk for serial HRCT of fibrotic interstitial pneumonia 52: pulmonary fibrosis. Chest 2011; 140: with texture-based automated quantification sys- 29. Wu MT, Pan HB, Chiang AA, et al. Prediction of 32. Zavaletta VA, Bartholmai BJ, Robb RA. High tem. Eur Radiol 2013; 23: postoperative lung function in patients with lung cancer: comparison of quantitative CT with perfusion scintigraphy. AJR 2002; 178: Ohno Y, Iwasawa T, Seo JB, et al. Oxygen-enhanced magnetic resonance imaging versus computed tomography: multicenter study for clinical stage classification of smoking-related chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2008; 177: resolution multidetector CT-aided tissue analysis and quantification of lung fibrosis. Acad Radiol 2007; 14: Wang J, Li F, Li Q. Automated segmentation of lungs with severe interstitial lung disease in CT. Med Phys 2009; 36: Park SC, Tan J, Wang X, et al. Computer-aided detection of early interstitial lung diseases using lowdose CT images. Phys Med Biol 2011; 56: Johnson SR, Granton JT. Pulmonary hypertension in systemic sclerosis and systemic lupus erythematosus. Eur Respir Rev 2011; 20: Iwasawa T, Ogura T, Sakai F, et al. CT analysis of the effect of pirfenidone in patients with idiopathic pulmonary fibrosis. Eur J Radiol 2014; 83: Hanumegowda C, Farkas L, Kolb M. Angiogenesis in pulmonary fibrosis: too much or not enough? Chest 2012; 142: AJR:203, August 2014 W173