- Email: [email protected]
Relationship between pulmonary artery volumes at computed tomography and pulmonary artery pressures in patients with- and without pulmonary hypertension
European Journal of Radiology 67 (2008) 466–471
Relationship between pulmonary artery volumes at computed tomography and pulmonary artery pressures in patients with- and without pulmonary hypertension Jens J. Froelich a,∗ , Helmut Koenig a , Lennard Knaak b , Stefan Krass c , Klaus J. Klose a a b
Department of Radiology, Philipps-University Hospital, Baldingerstrasse, 35043 Marburg, Germany Department of Medicine, Philipps-University Hospital, Baldingerstrasse, 35043 Marburg, Germany c MeVis Research, Universit¨ atsallee 29, 28359 Bremen, Germany Received 17 April 2007; received in revised form 10 August 2007; accepted 22 August 2007
Abstract Objectives: This study was designed to determine the relationship between pulmonary artery (PA) volumes at computed tomography (CT) and PA pressures at right-sided heart catheterization in patients with and without pulmonary hypertension (PAH) to develop a noninvasive CT method of PA pressure quantification. Materials and methods: Sixteen patients with chronic sleep apnea syndrome underwent contrast enhanced helical CT (slice thickness 3 mm; pitch 2; increment 2 mm) at inspiration. Eight patients had PAH while cardiopulmonary disease has been excluded in eight other patients. Vascular volumes were determined using a 3D technique (threshold seeded vascular tracing algorithm; thresholds −600 H [lower] and 3000 H [upper]). Right-sided heart catheterization measurements were available for linear regression analysis of PA volumes and pressures. Results: Correlation between PA pressures and volumes (normalized for BMI), was high in both groups (without PAH: r = .85; with PAH .90, Pearson). Compared to elevated PA pressures in patients with pulmonary hypertension (p < .005), PA volumes also were significantly increased (p < .05) among the groups. Conclusions: High correlation was found between PA volumes and mean PA pressures in patients with- and without PAH. Significant differences in PA volumes at CT-volumetry may admit non-invasive determination of pulmonary hypertension. © 2007 Published by Elsevier Ireland Ltd. Keywords: Computed tomography (CT); Hypertension; Pulmonary; Pulmonary arteries; Lung; Lung diseases
1. Introduction Generally pulmonary hypertension is assessed quantitatively with traditional parameters of vascular reactivity such as pressure, flow and resistance by means of right-sided heart catheterization as the diagnostic gold standard. Many efforts have been made to establish reliable and reproducible diagnostic imaging methods for non-invasive assessment of pulmonary artery (PA) pressure. Various attempts have been made to cor-
∗
Corresponding author at: Department of Radiology and Nuclear Medicine, Klinikum Bad Hersfeld, Seilerweg 29, 36251 Bad Hersfeld, Germany. Tel.: +49 6621 88 1326; fax: + 49 6621 88 1324. E-mail addresses: [email protected] (J.J. Froelich), [email protected] (H. Koenig), [email protected] (L. Knaak), [email protected] (S. Krass), [email protected] (K.J. Klose). 0720-048X/$ – see front matter © 2007 Published by Elsevier Ireland Ltd. doi:10.1016/j.ejrad.2007.08.022
relate PA size, measured with different imaging techniques, with PA pressure measured at right heart catheterization. Earlier results with plain radiography, echocardiography, MRI and CT suggested that such correlation exists, but measurements of pulmonary artery diameters at chest radiography were unreliable and also CT has not demonstrated sufficient accuracy to be used clinically yet [1–6]. Current CT techniques with three-dimensional postprocessing allow quantitative volumetry of various anatomic subsets. Determining lung volumes and comparing results with pulmonary function tests, recently good correlation has been found for FEV1, total lung volume, vital capacity and residual volume [7–9]. The objective of this study was to evaluate CT-volumetry of pulmonary arteries, prospectively correlating PA volumes with PA pressures in patients with- and without pulmonary hypertension. It was analyzed whether pulmonary artery volumes may be
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
quantified from spiral CT-examinations for non-invasive determination of pulmonary hypertension. Regression analysis with PA pressures was conducted in both groups.
467
2. Methods and materials
under rest conditions [10]. Informed consent was obtained after the nature of the procedures had been fully explained. The investigation had been approved by the institutional review board and was conducted in concordance with guidelines for good clinical practice [11].
2.1. Patients
2.2. Instruments
Sixteen patients (14 male, 2 female) with a median age of 50 years (range 30–69 years) were included in this prospective evaluation. Body size, weight and body mass index (BMI) were similar among the groups (p > .05, t-test; Table 1a and b). Pulmonary hypertension was present in eight patients and could be excluded in eight other patients. All patients received spiral CT of the chest and right-sided heart catheterization for basic investigation of chronic sleep apnea within the same week. Presence of fibrotic pulmonary disease was primarily excluded in all patients after analysis of HR-CT images by a board-approved radiologist. Reduction of forced expiratory volumes at 1 s (FEV1 ) was present in six patients, indicating presence of pulmonary emphysema, however, no significant difference was found among the groups (p > .05, Table 1a and b). No modification of therapy was obtained between the period of right heart catheterization and the CT-study. Pulmonary artery hypertension was defined as an increase of mean PA pressure above 25 mmHg with pulmonary capillary wedge pressure (PCWP) of less or equal than 15 mmHg and a pulmonary vascular resistance (PVR) exceeding 240 dyn × s × cm−5 during right heart catheter measurements
2.2.1. Right-sided heart catheterization A 5.5 F (French) Swan-Ganz balloon-tipped indwell thermodilution catheter was inserted through a jugular or antecubital vein and advanced into a main pulmonary artery using pressure monitoring guidance. Measurements were performed in triplicate by the thermodilution technique (10 ml chilled saline solution for each injection). Pulmonary arterial pressure (PAP), cardiac output (CO), PCWP and PVR (80 × (mean PAP − PCWP)/cardiac output) were recorded. Right heart catheterizations were carried out by the same operator. No complications were observed during the procedures. 2.2.2. Helical CT acquisition Spiral CT of the chest was performed using a Somatom Plus4 scanner (Siemens Medical Solutions, Forchheim, Germany) with the patient in supine position and arms elevated above the head. After a scout view obtained at deep inspiration, a cross-sectional scan plane was defined covering all four cardiac chambers. Using an antecubital venous access, 15 ml non-ionic contrast material test bolus (Solutrast 300, Altana, Constance, Germany) was injected at a flow rate of 3 ml/s, using
Table 1 Individual right heart catheter-, CT-volumetric and spirometric measurements in patients with pulmonary hypertension (a) and in patients without cardiopulmonary disease (b) Patient no.
BMI
PAPmean
CO
PCWP
PVR
VC (%)
FEV1
PAV (ml)
PAVrel.
(a) P1 P2 P3 P4 P5 P6 P7 P8
27 34 33 41 35 26 34 31
28 25 27 71 26 54 26 31
4.1 6.2 4.4 4.3 7.7 4.2 5.5 4.3
11 5 13 13 15 14 7 13
332 258 255 1235 387 471 276 335
97 81 80 85 108 108 52 66
91 84 65 80 92 106 38 46
349 220 278 419 266 393 237 319
9.423 7.480 9.174 17.179 9.310 10.218 8.058 9.889
33 4.4
36 16
5.1 1.2
11.4 3.3
444 307
85 18
75 22
310 68
10.091 2.810
31 30 27 26 25 40 29 33
12 13 12 15 13 17 18 19
5.9 7.6 4.7 6.2 5.5 6.3 8.8 8.7
5 6 7 10 11 7 8 7
95 74 85 65 29 127 91 110
84 100 88 96 92 87 94 97
73 105 72 100 70 88 83 92
200 229 212 233 224 242 269 288
6.200 6.870 5.724 6.058 5.600 9.680 7.801 9.504
30 4.5
14.9 2.6
6.7 1.4
85 28
92 5
85 12
237 27
7.180 1.540
Mean ±S.D. (b) N1 N2 N3 N4 N5 N6 N7 N8 Mean ±S.D.
7.6 1.9
BMI: body mass index (kg/m2 ), PAPmean : mean pulmonary artery pressure (mmHg), CO: cardiac output (l/min), PCWP: pulmonary capillary wedge pressure (mmHg), PVR: pulmonary vascular resistance (dyn s cm−5 ), VC: vital capacity (%), FEV1 : forced expitratory volume after 1 s, PAV (ml): pulmonary artery volume (ml), PAVrel : relative pulmonary artery volume (normalized to BMI).
468
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
an automatic power injector (Medrad, Indianola, PA, USA). Five seconds after initiation of contrast injection, 15 subsequent static single scans (120 kV, 90 mA, 3 mm slice thickness, pitch 0), were obtained at 2 s intervals to determine maximum contrast enhancement within the right atrium. According to the individual circulation time, a subsequent spiral CT scan was initiated to achieve maximum contrast enhancement within the pulmonary arteries. The predefined scan volume included pulmonary base and apex. Slice thickness was 3 mm at a rotation speed of 0.75 s with a tube current of 206 mA and voltage of 140 kV. Table feed was 6 mm to include the predefined scan volume during a single inspiratory breathold sequence with caudocranial slice-acquisition at a maximum scan time of 40 s. Images were reconstructed at 2 mm intervals using soft and sharp reconstruction algorithms. According to predetermined delay times, 80 ml non-ionic contrast material (Solutrast 300, Altana, Constance, Germany) were injected at a flow rate of 3 ml/s. Uncompressed CT-images (Digital Imaging and Communications in Medicine, DICOM Format) were transferred to a postprocessing workstation (Silicon Graphics, Indigo2, Mountain View, CA; OS IRIX, Version 6.2) for subsequent pulmonary segmentation, vascular reconstruction and volumetric assessment. 2.2.3. Pulmonary segmentation After transfer to the workstation, individual voxel dimensions were recorded and consequently images were transformed into a losslessly compressed Tagged Image File Format (TIFF-format) to reduce the data volume. When images have been loaded into a dedicated postprocessing application (ImageLabMed, MeVis, Bremen, Germany), both halves of the lung were segmented with a threshold-based algorithm at −700 HU (lower boundary). The segmentation process was started interactively marking a single pulmonary seeding point. Subsequently, the algorithm automatically propagated to neighboring voxels which were added to the segmented volume if the boundary criteria were met. Contour irregularities at the segmented margins were closed by applying morphologic operations. This procedure was applied for complete elimination of anatomic structures (e.g. chest wall, mediastinum) irrelevant for further analysis. 2.2.4. Vascular reconstruction After completion of pulmonary segmentation, a seed-voxel was placed within any pulmonary artery branch and a vascular segmentation algorithm (threshold seeded vascular tracing algorithm; thresholds, −600 HU [lower] and 3000 HU [upper]) traced and reconstructed the pulmonary artery tree. Beginning at the seed-voxel, this algorithm evaluated the density values of the neighborhood and developed a wavefront, expanding through the vascular tree. Branches of the wavefront defined and described vascular sections. Based on a neighborhood and connectivity analysis, surrounding tissue was excluded from reconstruction [12,13]. Segmentation proceeded automatically without further manual interference and was not limited to certain branching levels. The segmentation process included all distal, peripheral pulmonary artery segments. Manually a sagital cut plane had to be positioned at the pulmonary hilum to
Fig. 1. Three-dimensional image after vascular reconstruction of the right half of the lung displayed in volume rendering technique (VRT).
avoid mediastinal transgression of the pulmonary segmentation algorithm (e.g. due to cardiac pulsation in the central pulmonary arteries. Therefore, the pulmonary trunk could not be included in the volumetric analysis, although previous studies have suggested that this may be particularly valuable for assessment of pulmonary hypertension [10]. A starting point for the vascular trace operator could arbitrarily be defined within any hilar pulmonary artery segment. After initiation of vascular tracing, progression of the reconstruction algorithm was displayed continuously. Following reconstruction, the pulmonary artery tree was displayed in volume rendering technique (VRT; Fig. 1). VRT images were similar among patients with- and without pulmonary hypertension without visually evident differences regarding the depth of vascular reconstruction or caliber of pulmonary arteries. Interstitial and vascular volumes were determined based on the number of segmented voxels and voxel size. This process was separately performed on both halves of the lung and required about 2 min to conclude. 2.3. Statistical analysis Mean pulmonary artery pressures and pulmonary artery volumes (PAV) were evaluated for statistical significance between both groups using Students t-test. PAV was normalized according to the individual body mass index (PAVrel. = PAV × BMI). Significance was accepted at a level p < .05. Among patients with- and without PAH Pearson correlation was determined between PAPmean and PAVrel. Additionally, regression analysis was performed for PAP and PAVrel. in patients with- and without PAH as well as pooled in all patients. 3. Results 3.1. Patients with pulmonary hypertension Eight patients (six male, two female) with pulmonary hypertension were evaluated. Median age was 49 years.
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
469
3.2. Patients without pulmonary hypertension Pulmonary hypertension was excluded in eight male patients with chronic sleep apnea. Median age was 56 years and mean BMI was 30 ± 4.4 kg/m2 . Mean pulmonary artery pressure was 14.9 ± 2.6 mmHg (range 12–19 mmHg). According to CTvolumetric measurements, mean PAV was 237 ± 27 ml (range 200–288 ml). Mean PAV normalized for BMI was 7.180 ± 1.540 (PAVrel. ). Again there was high correlation of mean PAP with PAVrel. (r = .85, Pearson; Fig. 2b). Individual measurements for right heart catheter measurements, CT-volumetry and spirometry are given in Table 1b. 3.3. Comparison between groups According to Students t-test, patients with pulmonary hypertension had highly significant increased mean pulmonary artery pressure compared to patients without pulmonary hypertension (36 ± 16 mmHg vs. 14.9 ± 2.6 mmHg; p < .005, t-test). Compared to patients without pulmonary hypertension, PAV and PAVrel. (normalized to BMI) also were significantly increased in patients with pulmonary hypertension (310 ± 68 ml vs. 237 ± 27 ml; p < .05; resp. 10.092 ± 2.820 vs. 7.180 ± 1.540; p < .05, t-test). Pooling data from both groups shows high correlation of mean PAP with PAVrel. as well (r = .89, Pearson; Fig. 2c). 4. Discussion
Fig. 2. (a) High correlation of mean pulmonary artery pressures with pulmonary artery volumes normalized to BMI (PAVrel. ) in patients with pulmonary hypertension (n = 8; r = .90, Pearson). (b) Correlation between mean pulmonary artery pressures and normalized pulmonary artery volumes (PAVrel. ) in patients without pulmonary hypertension is similar (n = 8; r = .85, Pearson) as in (a). (c) High correlation between PAPmean and PAVrel. also is found in pooled data including all patients with- and without pulmonary hypertension (n = 16; r = .89, Pearson).
Mean BMI was 33 ± 4.4 kg/m2 . According to right heart catheter examination, mean pulmonary artery pressure was 36 ± 16 mmHg (range 25–71 mmHg) and CT-volumetric measurements resulted in a mean PAV of 310 ± 68 ml (range 220–419 ml). Mean PAV normalized for BMI was 10.091 ± 2.810 (PAVrel. ) Correlation of mean PAP with PAVrel. was high (r = .90, Pearson; Fig. 2a). Results of right heart catheter-, CT-volumetric measurements and spirometry for patients with pulmonary hypertension are summarized in Table 1a.
To reduce invasiveness and the associated risk of catheterbased examinations, various investigations have been performed for correlation of pulmonary artery pressure with non invasive radiomorphologic parameters. According to some of these investigations the best indicator for pulmonary hypertension may be the diameter of the right pulmonary artery [10]. However, correlation between pulmonary artery diameters and pulmonary artery pressures is reported variably [5,14–17]. All former non-invasive investigations of pulmonary hypertension were performed based on two-dimensional measurements of vascular diameters from plain radiographs, MRI, computed tomography or have been derived echocardiographically [18]. In a correlative study in 1984, Kuriyama et al. [14] examined 32 patients who underwent right-sided heart catheterization and chest CT. This population had cardiopulmonary disease without reported interstitial lung disease. Using a multiple regression analysis, the authors found that the combination of main PA and right interlobar PA cross-sectional area normalized for body surface area showed the best correlation with mean PA pressure (r = .93, p < .005). The investigators however could not demonstrate utility of their multiple regression equation in patients outside their own cohort. Four years later, Moore et al. [6] demonstrated only a trend toward correlation of main PA diameter with mean PA pressure in their study of 24 patients with pulmonary vascular disease or chronic lung disease who had undergone both chest CT and right-sided heart catheterization. They attributed their disparate results to differences in PA wall compliance between the two
470
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
groups. The small sample size of patients with chronic lung disease and the narrow range of PA pressures in this study group may have contributed to the failure to demonstrate a statistically significant correlation between main PA diameter and mean PA pressure. Schmidt et al. examined 50 patients with pulmonary hypertension secondary to chronic thromboembolism before and after surgical thrombendarterectomy with CT and right-sided heart catheterization. They found no close correlation between the extent of any radiographic sign and the degree of pulmonary artery pressure. After surgery, reversibility was most significant for the pulmonary trunk on CT (p < .005) [5]. These authors concluded that conventional radiography and CT do not provide morphometric parameters that allow prediction of pulmonary artery pressure. In 1997, Haimovici et al. [4] measured PA vessel diameters in 55 patients prior to lung and heart–lung transplantation at CT and correlated with PA pressures. They found the best correlation combining main and left main PA cross-sectional area corrected for body surface area with mean PA pressure (r = .87). However, again it was found that these normalized PA measurements with CT do not predict PA pressure reliably enough to be a consistently useful clinical tool. These previous results are in some conflict to an evaluation of Tan et al. [17], retrospectively evaluating 45 patients who had undergone both chest CT and right-sided heart catheterization in 1998. The authors found that the most specific findings for the presence of pulmonary hypertension are CT-determined main pulmonary artery diameters ≥29 mm and a segmental arteryto bronchus ratio >1:1 in three or four lobes with a specifity of 100%. However, no linear correlation between the degree of pulmonary hypertension and CT-determined main pulmonary artery diameters was found in this investigation. This work concludes that standard chest CT scans may be used to screen for pulmonary hypertension in patients with parenchymal lung disease. Depending on the underlying disease, pulmonary hypertension may generate various morphological alterations within the lung [19]. Any chronic increase of pulmonary artery pressure induces dilatation and limited hypertrophy of the right ventricle. Additionally, increase of pulmonary artery pressure is associated with dilatation of the pulmonary trunk including both main pulmonary artery segments while peripheral pulmonary arteries remain rather uninvolved. Additionally, an increase in peripheral pulmonary artery resistance reduces cardiac ejection fraction [10]. Therefore, various studies determining normal diameters of right and left pulmonary arteries at CT conclude with differing results [14–16]. Most investigations have underlined the importance of PA compliance. Thus, the modification of the diameter of the pulmonary trunk during systolic and diastolic intervals may be of importance. Decrease of compliance in severe pulmonary hypertension could lead to a decrease in variation of diameter. This is reflecting that the coherence among intravascular pressure, luminal size, interstitial pulmonary alterations and physiological factors (e.g. level of inspiration) is complex and may not be assessed with simple morphological or volumetric analysis alone. In non-invasive quantification of pulmonary
hypertension the best correlation was reported between diameter of the pulmonary trunk and pulmonary artery pressure by some investigations [5,14], but even here, a large variation of correlation coefficients has been presented (r = .43–.83). Our method and findings differ in several aspects from those of previous studies [4–6,14–17,20]. Primarily, this is the first investigation utilizing a CT-volumetric method of pulmonary artery measurement. Additionally, all patients in our study have had hemodynamic validation of PA pressures and groups have prospectively been defined into patients withand without pulmonary hypertension. Patients with interstitial fibrotic lung disease or mediastinal lymphadenopathy were not found in this investigation due to primary exclusion after analysis of HR-CT images by a board-approved radiologist (Reviewer #2, comment 4). In our evaluation mean pulmonary artery pressure was 36 ± 16 mmHg in patients with pulmonary hypertension and 14.9 ± 2.6 mmHg in patients without pulmonary hypertension (p < .005, t-test). Mean pulmonary artery volumes were 310 ± 68 ml in patients with pulmonary artery hypertension and 237 ± 27 ml in patients without pulmonary hypertension. Normalized for BMI, PAVrel. was 10.091 ± 2.810 and 7.180 ± 1.540, respectively. These differences both were significant among the groups (p < .05, t-test). Pulmonary artery volumes, as derived in this investigation seem to correlate well with data from literature [21], stating that the pulmonary artery blood volume is between 175 and 250 ml under rest conditions. While PA volumes were significantly different between the groups, additionally high correlation was found between PA volumes (normalized for BMI) and mean PA pressures in patients with PA hypertension (r = .90, Pearson) and without (r = .85), as well as in pooled data of all patients (r = .89, Pearson). Normal mean pulmonary arterial pressure at sea level is considered to be about 12–16 mmHg. Definite pulmonary hypertension is present when mean pressures at rest exceed 25 mmHg. Diagnosis of PAH generally also requires the presence of pulmonary hypertension with two other conditions: pulmonary artery wedge pressure should be less than 15 mmHg and pulmonary vascular resistance should be greater than 240 dyn s cm−5 . According to Table 1a, these requirements have been fulfilled in our patient population primarily suffering from sleep apnea syndrome. However, generally pulmonary hypertension was not very excessive in our population thus possibly enhancing power of our findings. It may reflect that hypoxia related to sleep apnea is a secondary cause of pulmonary hypertension, while frequently its severity is only moderate. In addition the group of patients without pulmonary hypertension and sleep apnea also should have presented hypoxemia. Consequently, a modification of PA compliance cannot be excluded in this group either. Therefore, selection of patients suffering from comventional pulmonary hypertension such as WHO Group 1 (idiopathic, associated with collagen vascular disease, etc.) may have resulted in even improved correlation between PA pressures and PA volumes. In chronic pulmonary hypertension, volumes of the central PA segments become expanded while peripheral intrapulmonary PA artery volumes rather remain unchanged [10] or may even be decreased. Our volumetric analysis included all branches of
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
the pulmonary arteries except for the mediastinal portion of the pulmonary trunk without further detailed segmental analysis. Exclusion of central pulmonary artery segments and especially the pulmonary trunk was necessary to avoid mediastinal transgression of the vascular reconstruction algorithm due to close anatomic neighborhood and cardiac pulsations. Since volumetric enlargement due to pulmonary hypertension majorly affects central PA segments, further segmental differentiation of PA volumes with inclusion of the pulmonary trunk possibly may even enhance significance of CT-volumetric analysis. Presumably recent advancements in scanner technology with superior spatial and temporal resolution particularly in combination with ECG gating may further improve accuracy of CT-volumetry. Limitations of CT-volumetry of pulmonary arteries include associated radiation exposure and the need to administer iodinated contrast material, which is not required during right-sided heart catheter measurements. The postprocessing mechanism however is easily applicaple and may be standardized further for routine application. Results of CT-volumetry may also be transferrable to other forms of PAH or interstitial lung disease, wherein there maybe real loss of pulmonary vascular bed such as emphysema, thromboembolic PAH or pulmonary fibrosis. Additionally, computed tomography simultaneously delivers associated morphological information within the mediastinum, pleural space and pulmonary tissue. The study population was small in this investigation (n = 16) and has been divided into two subgroups. Therefore, the results need to be confirmed in larger studies, applying multidetector CT-technology with ECG gating and inclusion of central pulmonary artery segments. However, in complement to previous diameter-related measurements [5,14–17], determination of pulmonary artery volumes with CT may predict PA pressure reliably enough to accomplish a valuable clinical tool in patients with pulmonary hypertension. Even though presumably right-sided heart catheterization will remain the gold standard for diagnosis of pulmonary hypertension, CT could become an interesting non-invasive alternative for assessment of disease with promising further perspectives. Conflicts of interest None of the authors has to disclose any conflicts. Acknowledgements This work was performed at Philipps-University Hospital, Baldingerstrasse, 35043 Marburg, Germany. The study was part of the VICORA-Project (Virtual Institute for Computer Assistance in Clinical Radiology), granted by the German Federal Ministry for Education and Research.
471
References [1] Lupi E, Dumont C, Tejada VM, et al. A radiologic index of pulmonary arterial hypertension. Chest 1975;68:28–31. [2] Bush A, Gray H, Denison DM. Diagnosis of pulmonary hypertension from radiographic estimates of pulmonary arterial size. Thorax 1987;42:207–11. [3] Matthay RA, Schwarz MI, Ellis Jr JH, et al. Pulmonary artery hypertension in chronic obstructive pulmonary disease: determination by chest radiography. Invest Radiol 1981;16:95–100. [4] Haimovici JB, Trotman-Dickenson B, Halpern EF, et al. Relationship between pulmonary artery diameter at computed tomography and pulmonary artery pressures at right-sided heart catheterization. Acad Radiol 1997;4:327–34. [5] Schmidt HC, Kauczor HU, Schild HH, et al. Pulmonary hypertension in patients with chronic thromboembolism: chest radiograph and CT evaluation before and after surgery. Eur Radiol 1996;6:817–25. [6] Moore NR, Scott JP, Flower CD, et al. The relationship between pulmonary artery pressure and pulmonary artery diameter in pulmonary hypertension. Clin Radiol 1988;39:486–9. [7] Kauczor HU, Heussel CP, Fischer B, et al. Assessment of lung volumes using helical CT at inspiration and expiration: comparison with pulmonary function tests. AJR 1998;171:1091–5. [8] Arakawa A, Yamashita Y, Nakayama Y, et al. Assessment of lung volumes in pulmonary emphysema using multidetector helical CT: comparison with pulmonary function tests. Comput Med Imaging Graph 2001;25:399–404. [9] Zaporozhan J, Ley S, Eberhardt R, et al. Paired inspiratory/expiratory volumetric thin-slice CT scan for emphysema analysis: comparison of different quantitative evaluations and pulmonary function test. Chest 2005;128:3212–322. [10] Henk CB, Gabriel H, Fleischmann D, et al. Pulmonary hypertension and cor pulmonale. Radiologe 1997;37:388–401. [11] Dixon Jr JR. The International Conference on Harmonization Good Clinical Practice guideline. Qual Assur 1998;6:65–74. [12] Zahlten C, Juergens H, Everts CJG, et al. Portal vein reconstruction based on topology. Eur J Radiol 1995;19:96–100. [13] Zahlten C, Juergens H, Peitgen HO. Reconstruction of branching blood vessels from CT-data. In: Goebel M, Mueller H, Urban B, editors. Visualization in scientific computing. Vienna, A: Springer; 1995. p. 41–52. [14] Kuriyama K, Gamsu G, Stern RG, et al. CT-determined pulmonary artery diameters in predicting pulmonary hypertension. Invest Radiol 1984;19:16–22. [15] Ng CS, Wells AU, Padley SP. A CT sign of chronic pulmonary arterial hypertension: the ratio of main pulmonary artery to aortic diameter. J Thorac Imaging 1999;14:270–8. [16] Burakowska B, Pawlicka L, Oniszh K, et al. Value of spiral computed tomography in pulmonary hypertension. Pol Arch Med Wewn 2004;111:431–41. [17] Tan RT, Kuzo R, Goodman LR, et al. Utility of CT scan evaluation for predicting pulmonary hypertension in patients with parenchymal lung disease Medical college of Wisconsin lung transplant group. Chest 1998;113:1250–6. [18] Kauczor HU, Heussel CP, Thelen M. Update on diagnostic strategies of pulmonary embolism. Eur Radiol 1999;9:262–75. [19] Teichmann V, Jezek V, Herles F. Relevance of width of right descending branch of pulmonary artery as a radiologic sign of pulmonary hypertension. Thorax 1970;25:91–6. [20] O’Callagan JP, Heitzman ER, Somogyi JW, et al. CT-evaluation of pulmonary artery size. J Comput Assist Tomogr 1982;6:101–4. [21] Van Mierop LH, Kutsche LM. Embryology of the heart. In: Hurst JW, Logue RB, Rackley CE, Schlant RC, Sonnenblick EH, Wallace AG, Wenger NK, editors. The heart. New York: McGraw Hill; 1986. p. 1091.
Relationship between pulmonary artery volumes at computed tomography and pulmonary artery pressures in patients with- and without pulmonary hypertension Jens J. Froelich a,∗ , Helmut Koenig a , Lennard Knaak b , Stefan Krass c , Klaus J. Klose a a b
Department of Radiology, Philipps-University Hospital, Baldingerstrasse, 35043 Marburg, Germany Department of Medicine, Philipps-University Hospital, Baldingerstrasse, 35043 Marburg, Germany c MeVis Research, Universit¨ atsallee 29, 28359 Bremen, Germany Received 17 April 2007; received in revised form 10 August 2007; accepted 22 August 2007
Abstract Objectives: This study was designed to determine the relationship between pulmonary artery (PA) volumes at computed tomography (CT) and PA pressures at right-sided heart catheterization in patients with and without pulmonary hypertension (PAH) to develop a noninvasive CT method of PA pressure quantification. Materials and methods: Sixteen patients with chronic sleep apnea syndrome underwent contrast enhanced helical CT (slice thickness 3 mm; pitch 2; increment 2 mm) at inspiration. Eight patients had PAH while cardiopulmonary disease has been excluded in eight other patients. Vascular volumes were determined using a 3D technique (threshold seeded vascular tracing algorithm; thresholds −600 H [lower] and 3000 H [upper]). Right-sided heart catheterization measurements were available for linear regression analysis of PA volumes and pressures. Results: Correlation between PA pressures and volumes (normalized for BMI), was high in both groups (without PAH: r = .85; with PAH .90, Pearson). Compared to elevated PA pressures in patients with pulmonary hypertension (p < .005), PA volumes also were significantly increased (p < .05) among the groups. Conclusions: High correlation was found between PA volumes and mean PA pressures in patients with- and without PAH. Significant differences in PA volumes at CT-volumetry may admit non-invasive determination of pulmonary hypertension. © 2007 Published by Elsevier Ireland Ltd. Keywords: Computed tomography (CT); Hypertension; Pulmonary; Pulmonary arteries; Lung; Lung diseases
1. Introduction Generally pulmonary hypertension is assessed quantitatively with traditional parameters of vascular reactivity such as pressure, flow and resistance by means of right-sided heart catheterization as the diagnostic gold standard. Many efforts have been made to establish reliable and reproducible diagnostic imaging methods for non-invasive assessment of pulmonary artery (PA) pressure. Various attempts have been made to cor-
∗
Corresponding author at: Department of Radiology and Nuclear Medicine, Klinikum Bad Hersfeld, Seilerweg 29, 36251 Bad Hersfeld, Germany. Tel.: +49 6621 88 1326; fax: + 49 6621 88 1324. E-mail addresses: [email protected] (J.J. Froelich), [email protected] (H. Koenig), [email protected] (L. Knaak), [email protected] (S. Krass), [email protected] (K.J. Klose). 0720-048X/$ – see front matter © 2007 Published by Elsevier Ireland Ltd. doi:10.1016/j.ejrad.2007.08.022
relate PA size, measured with different imaging techniques, with PA pressure measured at right heart catheterization. Earlier results with plain radiography, echocardiography, MRI and CT suggested that such correlation exists, but measurements of pulmonary artery diameters at chest radiography were unreliable and also CT has not demonstrated sufficient accuracy to be used clinically yet [1–6]. Current CT techniques with three-dimensional postprocessing allow quantitative volumetry of various anatomic subsets. Determining lung volumes and comparing results with pulmonary function tests, recently good correlation has been found for FEV1, total lung volume, vital capacity and residual volume [7–9]. The objective of this study was to evaluate CT-volumetry of pulmonary arteries, prospectively correlating PA volumes with PA pressures in patients with- and without pulmonary hypertension. It was analyzed whether pulmonary artery volumes may be
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
quantified from spiral CT-examinations for non-invasive determination of pulmonary hypertension. Regression analysis with PA pressures was conducted in both groups.
467
2. Methods and materials
under rest conditions [10]. Informed consent was obtained after the nature of the procedures had been fully explained. The investigation had been approved by the institutional review board and was conducted in concordance with guidelines for good clinical practice [11].
2.1. Patients
2.2. Instruments
Sixteen patients (14 male, 2 female) with a median age of 50 years (range 30–69 years) were included in this prospective evaluation. Body size, weight and body mass index (BMI) were similar among the groups (p > .05, t-test; Table 1a and b). Pulmonary hypertension was present in eight patients and could be excluded in eight other patients. All patients received spiral CT of the chest and right-sided heart catheterization for basic investigation of chronic sleep apnea within the same week. Presence of fibrotic pulmonary disease was primarily excluded in all patients after analysis of HR-CT images by a board-approved radiologist. Reduction of forced expiratory volumes at 1 s (FEV1 ) was present in six patients, indicating presence of pulmonary emphysema, however, no significant difference was found among the groups (p > .05, Table 1a and b). No modification of therapy was obtained between the period of right heart catheterization and the CT-study. Pulmonary artery hypertension was defined as an increase of mean PA pressure above 25 mmHg with pulmonary capillary wedge pressure (PCWP) of less or equal than 15 mmHg and a pulmonary vascular resistance (PVR) exceeding 240 dyn × s × cm−5 during right heart catheter measurements
2.2.1. Right-sided heart catheterization A 5.5 F (French) Swan-Ganz balloon-tipped indwell thermodilution catheter was inserted through a jugular or antecubital vein and advanced into a main pulmonary artery using pressure monitoring guidance. Measurements were performed in triplicate by the thermodilution technique (10 ml chilled saline solution for each injection). Pulmonary arterial pressure (PAP), cardiac output (CO), PCWP and PVR (80 × (mean PAP − PCWP)/cardiac output) were recorded. Right heart catheterizations were carried out by the same operator. No complications were observed during the procedures. 2.2.2. Helical CT acquisition Spiral CT of the chest was performed using a Somatom Plus4 scanner (Siemens Medical Solutions, Forchheim, Germany) with the patient in supine position and arms elevated above the head. After a scout view obtained at deep inspiration, a cross-sectional scan plane was defined covering all four cardiac chambers. Using an antecubital venous access, 15 ml non-ionic contrast material test bolus (Solutrast 300, Altana, Constance, Germany) was injected at a flow rate of 3 ml/s, using
Table 1 Individual right heart catheter-, CT-volumetric and spirometric measurements in patients with pulmonary hypertension (a) and in patients without cardiopulmonary disease (b) Patient no.
BMI
PAPmean
CO
PCWP
PVR
VC (%)
FEV1
PAV (ml)
PAVrel.
(a) P1 P2 P3 P4 P5 P6 P7 P8
27 34 33 41 35 26 34 31
28 25 27 71 26 54 26 31
4.1 6.2 4.4 4.3 7.7 4.2 5.5 4.3
11 5 13 13 15 14 7 13
332 258 255 1235 387 471 276 335
97 81 80 85 108 108 52 66
91 84 65 80 92 106 38 46
349 220 278 419 266 393 237 319
9.423 7.480 9.174 17.179 9.310 10.218 8.058 9.889
33 4.4
36 16
5.1 1.2
11.4 3.3
444 307
85 18
75 22
310 68
10.091 2.810
31 30 27 26 25 40 29 33
12 13 12 15 13 17 18 19
5.9 7.6 4.7 6.2 5.5 6.3 8.8 8.7
5 6 7 10 11 7 8 7
95 74 85 65 29 127 91 110
84 100 88 96 92 87 94 97
73 105 72 100 70 88 83 92
200 229 212 233 224 242 269 288
6.200 6.870 5.724 6.058 5.600 9.680 7.801 9.504
30 4.5
14.9 2.6
6.7 1.4
85 28
92 5
85 12
237 27
7.180 1.540
Mean ±S.D. (b) N1 N2 N3 N4 N5 N6 N7 N8 Mean ±S.D.
7.6 1.9
BMI: body mass index (kg/m2 ), PAPmean : mean pulmonary artery pressure (mmHg), CO: cardiac output (l/min), PCWP: pulmonary capillary wedge pressure (mmHg), PVR: pulmonary vascular resistance (dyn s cm−5 ), VC: vital capacity (%), FEV1 : forced expitratory volume after 1 s, PAV (ml): pulmonary artery volume (ml), PAVrel : relative pulmonary artery volume (normalized to BMI).
468
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
an automatic power injector (Medrad, Indianola, PA, USA). Five seconds after initiation of contrast injection, 15 subsequent static single scans (120 kV, 90 mA, 3 mm slice thickness, pitch 0), were obtained at 2 s intervals to determine maximum contrast enhancement within the right atrium. According to the individual circulation time, a subsequent spiral CT scan was initiated to achieve maximum contrast enhancement within the pulmonary arteries. The predefined scan volume included pulmonary base and apex. Slice thickness was 3 mm at a rotation speed of 0.75 s with a tube current of 206 mA and voltage of 140 kV. Table feed was 6 mm to include the predefined scan volume during a single inspiratory breathold sequence with caudocranial slice-acquisition at a maximum scan time of 40 s. Images were reconstructed at 2 mm intervals using soft and sharp reconstruction algorithms. According to predetermined delay times, 80 ml non-ionic contrast material (Solutrast 300, Altana, Constance, Germany) were injected at a flow rate of 3 ml/s. Uncompressed CT-images (Digital Imaging and Communications in Medicine, DICOM Format) were transferred to a postprocessing workstation (Silicon Graphics, Indigo2, Mountain View, CA; OS IRIX, Version 6.2) for subsequent pulmonary segmentation, vascular reconstruction and volumetric assessment. 2.2.3. Pulmonary segmentation After transfer to the workstation, individual voxel dimensions were recorded and consequently images were transformed into a losslessly compressed Tagged Image File Format (TIFF-format) to reduce the data volume. When images have been loaded into a dedicated postprocessing application (ImageLabMed, MeVis, Bremen, Germany), both halves of the lung were segmented with a threshold-based algorithm at −700 HU (lower boundary). The segmentation process was started interactively marking a single pulmonary seeding point. Subsequently, the algorithm automatically propagated to neighboring voxels which were added to the segmented volume if the boundary criteria were met. Contour irregularities at the segmented margins were closed by applying morphologic operations. This procedure was applied for complete elimination of anatomic structures (e.g. chest wall, mediastinum) irrelevant for further analysis. 2.2.4. Vascular reconstruction After completion of pulmonary segmentation, a seed-voxel was placed within any pulmonary artery branch and a vascular segmentation algorithm (threshold seeded vascular tracing algorithm; thresholds, −600 HU [lower] and 3000 HU [upper]) traced and reconstructed the pulmonary artery tree. Beginning at the seed-voxel, this algorithm evaluated the density values of the neighborhood and developed a wavefront, expanding through the vascular tree. Branches of the wavefront defined and described vascular sections. Based on a neighborhood and connectivity analysis, surrounding tissue was excluded from reconstruction [12,13]. Segmentation proceeded automatically without further manual interference and was not limited to certain branching levels. The segmentation process included all distal, peripheral pulmonary artery segments. Manually a sagital cut plane had to be positioned at the pulmonary hilum to
Fig. 1. Three-dimensional image after vascular reconstruction of the right half of the lung displayed in volume rendering technique (VRT).
avoid mediastinal transgression of the pulmonary segmentation algorithm (e.g. due to cardiac pulsation in the central pulmonary arteries. Therefore, the pulmonary trunk could not be included in the volumetric analysis, although previous studies have suggested that this may be particularly valuable for assessment of pulmonary hypertension [10]. A starting point for the vascular trace operator could arbitrarily be defined within any hilar pulmonary artery segment. After initiation of vascular tracing, progression of the reconstruction algorithm was displayed continuously. Following reconstruction, the pulmonary artery tree was displayed in volume rendering technique (VRT; Fig. 1). VRT images were similar among patients with- and without pulmonary hypertension without visually evident differences regarding the depth of vascular reconstruction or caliber of pulmonary arteries. Interstitial and vascular volumes were determined based on the number of segmented voxels and voxel size. This process was separately performed on both halves of the lung and required about 2 min to conclude. 2.3. Statistical analysis Mean pulmonary artery pressures and pulmonary artery volumes (PAV) were evaluated for statistical significance between both groups using Students t-test. PAV was normalized according to the individual body mass index (PAVrel. = PAV × BMI). Significance was accepted at a level p < .05. Among patients with- and without PAH Pearson correlation was determined between PAPmean and PAVrel. Additionally, regression analysis was performed for PAP and PAVrel. in patients with- and without PAH as well as pooled in all patients. 3. Results 3.1. Patients with pulmonary hypertension Eight patients (six male, two female) with pulmonary hypertension were evaluated. Median age was 49 years.
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
469
3.2. Patients without pulmonary hypertension Pulmonary hypertension was excluded in eight male patients with chronic sleep apnea. Median age was 56 years and mean BMI was 30 ± 4.4 kg/m2 . Mean pulmonary artery pressure was 14.9 ± 2.6 mmHg (range 12–19 mmHg). According to CTvolumetric measurements, mean PAV was 237 ± 27 ml (range 200–288 ml). Mean PAV normalized for BMI was 7.180 ± 1.540 (PAVrel. ). Again there was high correlation of mean PAP with PAVrel. (r = .85, Pearson; Fig. 2b). Individual measurements for right heart catheter measurements, CT-volumetry and spirometry are given in Table 1b. 3.3. Comparison between groups According to Students t-test, patients with pulmonary hypertension had highly significant increased mean pulmonary artery pressure compared to patients without pulmonary hypertension (36 ± 16 mmHg vs. 14.9 ± 2.6 mmHg; p < .005, t-test). Compared to patients without pulmonary hypertension, PAV and PAVrel. (normalized to BMI) also were significantly increased in patients with pulmonary hypertension (310 ± 68 ml vs. 237 ± 27 ml; p < .05; resp. 10.092 ± 2.820 vs. 7.180 ± 1.540; p < .05, t-test). Pooling data from both groups shows high correlation of mean PAP with PAVrel. as well (r = .89, Pearson; Fig. 2c). 4. Discussion
Fig. 2. (a) High correlation of mean pulmonary artery pressures with pulmonary artery volumes normalized to BMI (PAVrel. ) in patients with pulmonary hypertension (n = 8; r = .90, Pearson). (b) Correlation between mean pulmonary artery pressures and normalized pulmonary artery volumes (PAVrel. ) in patients without pulmonary hypertension is similar (n = 8; r = .85, Pearson) as in (a). (c) High correlation between PAPmean and PAVrel. also is found in pooled data including all patients with- and without pulmonary hypertension (n = 16; r = .89, Pearson).
Mean BMI was 33 ± 4.4 kg/m2 . According to right heart catheter examination, mean pulmonary artery pressure was 36 ± 16 mmHg (range 25–71 mmHg) and CT-volumetric measurements resulted in a mean PAV of 310 ± 68 ml (range 220–419 ml). Mean PAV normalized for BMI was 10.091 ± 2.810 (PAVrel. ) Correlation of mean PAP with PAVrel. was high (r = .90, Pearson; Fig. 2a). Results of right heart catheter-, CT-volumetric measurements and spirometry for patients with pulmonary hypertension are summarized in Table 1a.
To reduce invasiveness and the associated risk of catheterbased examinations, various investigations have been performed for correlation of pulmonary artery pressure with non invasive radiomorphologic parameters. According to some of these investigations the best indicator for pulmonary hypertension may be the diameter of the right pulmonary artery [10]. However, correlation between pulmonary artery diameters and pulmonary artery pressures is reported variably [5,14–17]. All former non-invasive investigations of pulmonary hypertension were performed based on two-dimensional measurements of vascular diameters from plain radiographs, MRI, computed tomography or have been derived echocardiographically [18]. In a correlative study in 1984, Kuriyama et al. [14] examined 32 patients who underwent right-sided heart catheterization and chest CT. This population had cardiopulmonary disease without reported interstitial lung disease. Using a multiple regression analysis, the authors found that the combination of main PA and right interlobar PA cross-sectional area normalized for body surface area showed the best correlation with mean PA pressure (r = .93, p < .005). The investigators however could not demonstrate utility of their multiple regression equation in patients outside their own cohort. Four years later, Moore et al. [6] demonstrated only a trend toward correlation of main PA diameter with mean PA pressure in their study of 24 patients with pulmonary vascular disease or chronic lung disease who had undergone both chest CT and right-sided heart catheterization. They attributed their disparate results to differences in PA wall compliance between the two
470
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
groups. The small sample size of patients with chronic lung disease and the narrow range of PA pressures in this study group may have contributed to the failure to demonstrate a statistically significant correlation between main PA diameter and mean PA pressure. Schmidt et al. examined 50 patients with pulmonary hypertension secondary to chronic thromboembolism before and after surgical thrombendarterectomy with CT and right-sided heart catheterization. They found no close correlation between the extent of any radiographic sign and the degree of pulmonary artery pressure. After surgery, reversibility was most significant for the pulmonary trunk on CT (p < .005) [5]. These authors concluded that conventional radiography and CT do not provide morphometric parameters that allow prediction of pulmonary artery pressure. In 1997, Haimovici et al. [4] measured PA vessel diameters in 55 patients prior to lung and heart–lung transplantation at CT and correlated with PA pressures. They found the best correlation combining main and left main PA cross-sectional area corrected for body surface area with mean PA pressure (r = .87). However, again it was found that these normalized PA measurements with CT do not predict PA pressure reliably enough to be a consistently useful clinical tool. These previous results are in some conflict to an evaluation of Tan et al. [17], retrospectively evaluating 45 patients who had undergone both chest CT and right-sided heart catheterization in 1998. The authors found that the most specific findings for the presence of pulmonary hypertension are CT-determined main pulmonary artery diameters ≥29 mm and a segmental arteryto bronchus ratio >1:1 in three or four lobes with a specifity of 100%. However, no linear correlation between the degree of pulmonary hypertension and CT-determined main pulmonary artery diameters was found in this investigation. This work concludes that standard chest CT scans may be used to screen for pulmonary hypertension in patients with parenchymal lung disease. Depending on the underlying disease, pulmonary hypertension may generate various morphological alterations within the lung [19]. Any chronic increase of pulmonary artery pressure induces dilatation and limited hypertrophy of the right ventricle. Additionally, increase of pulmonary artery pressure is associated with dilatation of the pulmonary trunk including both main pulmonary artery segments while peripheral pulmonary arteries remain rather uninvolved. Additionally, an increase in peripheral pulmonary artery resistance reduces cardiac ejection fraction [10]. Therefore, various studies determining normal diameters of right and left pulmonary arteries at CT conclude with differing results [14–16]. Most investigations have underlined the importance of PA compliance. Thus, the modification of the diameter of the pulmonary trunk during systolic and diastolic intervals may be of importance. Decrease of compliance in severe pulmonary hypertension could lead to a decrease in variation of diameter. This is reflecting that the coherence among intravascular pressure, luminal size, interstitial pulmonary alterations and physiological factors (e.g. level of inspiration) is complex and may not be assessed with simple morphological or volumetric analysis alone. In non-invasive quantification of pulmonary
hypertension the best correlation was reported between diameter of the pulmonary trunk and pulmonary artery pressure by some investigations [5,14], but even here, a large variation of correlation coefficients has been presented (r = .43–.83). Our method and findings differ in several aspects from those of previous studies [4–6,14–17,20]. Primarily, this is the first investigation utilizing a CT-volumetric method of pulmonary artery measurement. Additionally, all patients in our study have had hemodynamic validation of PA pressures and groups have prospectively been defined into patients withand without pulmonary hypertension. Patients with interstitial fibrotic lung disease or mediastinal lymphadenopathy were not found in this investigation due to primary exclusion after analysis of HR-CT images by a board-approved radiologist (Reviewer #2, comment 4). In our evaluation mean pulmonary artery pressure was 36 ± 16 mmHg in patients with pulmonary hypertension and 14.9 ± 2.6 mmHg in patients without pulmonary hypertension (p < .005, t-test). Mean pulmonary artery volumes were 310 ± 68 ml in patients with pulmonary artery hypertension and 237 ± 27 ml in patients without pulmonary hypertension. Normalized for BMI, PAVrel. was 10.091 ± 2.810 and 7.180 ± 1.540, respectively. These differences both were significant among the groups (p < .05, t-test). Pulmonary artery volumes, as derived in this investigation seem to correlate well with data from literature [21], stating that the pulmonary artery blood volume is between 175 and 250 ml under rest conditions. While PA volumes were significantly different between the groups, additionally high correlation was found between PA volumes (normalized for BMI) and mean PA pressures in patients with PA hypertension (r = .90, Pearson) and without (r = .85), as well as in pooled data of all patients (r = .89, Pearson). Normal mean pulmonary arterial pressure at sea level is considered to be about 12–16 mmHg. Definite pulmonary hypertension is present when mean pressures at rest exceed 25 mmHg. Diagnosis of PAH generally also requires the presence of pulmonary hypertension with two other conditions: pulmonary artery wedge pressure should be less than 15 mmHg and pulmonary vascular resistance should be greater than 240 dyn s cm−5 . According to Table 1a, these requirements have been fulfilled in our patient population primarily suffering from sleep apnea syndrome. However, generally pulmonary hypertension was not very excessive in our population thus possibly enhancing power of our findings. It may reflect that hypoxia related to sleep apnea is a secondary cause of pulmonary hypertension, while frequently its severity is only moderate. In addition the group of patients without pulmonary hypertension and sleep apnea also should have presented hypoxemia. Consequently, a modification of PA compliance cannot be excluded in this group either. Therefore, selection of patients suffering from comventional pulmonary hypertension such as WHO Group 1 (idiopathic, associated with collagen vascular disease, etc.) may have resulted in even improved correlation between PA pressures and PA volumes. In chronic pulmonary hypertension, volumes of the central PA segments become expanded while peripheral intrapulmonary PA artery volumes rather remain unchanged [10] or may even be decreased. Our volumetric analysis included all branches of
J.J. Froelich et al. / European Journal of Radiology 67 (2008) 466–471
the pulmonary arteries except for the mediastinal portion of the pulmonary trunk without further detailed segmental analysis. Exclusion of central pulmonary artery segments and especially the pulmonary trunk was necessary to avoid mediastinal transgression of the vascular reconstruction algorithm due to close anatomic neighborhood and cardiac pulsations. Since volumetric enlargement due to pulmonary hypertension majorly affects central PA segments, further segmental differentiation of PA volumes with inclusion of the pulmonary trunk possibly may even enhance significance of CT-volumetric analysis. Presumably recent advancements in scanner technology with superior spatial and temporal resolution particularly in combination with ECG gating may further improve accuracy of CT-volumetry. Limitations of CT-volumetry of pulmonary arteries include associated radiation exposure and the need to administer iodinated contrast material, which is not required during right-sided heart catheter measurements. The postprocessing mechanism however is easily applicaple and may be standardized further for routine application. Results of CT-volumetry may also be transferrable to other forms of PAH or interstitial lung disease, wherein there maybe real loss of pulmonary vascular bed such as emphysema, thromboembolic PAH or pulmonary fibrosis. Additionally, computed tomography simultaneously delivers associated morphological information within the mediastinum, pleural space and pulmonary tissue. The study population was small in this investigation (n = 16) and has been divided into two subgroups. Therefore, the results need to be confirmed in larger studies, applying multidetector CT-technology with ECG gating and inclusion of central pulmonary artery segments. However, in complement to previous diameter-related measurements [5,14–17], determination of pulmonary artery volumes with CT may predict PA pressure reliably enough to accomplish a valuable clinical tool in patients with pulmonary hypertension. Even though presumably right-sided heart catheterization will remain the gold standard for diagnosis of pulmonary hypertension, CT could become an interesting non-invasive alternative for assessment of disease with promising further perspectives. Conflicts of interest None of the authors has to disclose any conflicts. Acknowledgements This work was performed at Philipps-University Hospital, Baldingerstrasse, 35043 Marburg, Germany. The study was part of the VICORA-Project (Virtual Institute for Computer Assistance in Clinical Radiology), granted by the German Federal Ministry for Education and Research.
471
References [1] Lupi E, Dumont C, Tejada VM, et al. A radiologic index of pulmonary arterial hypertension. Chest 1975;68:28–31. [2] Bush A, Gray H, Denison DM. Diagnosis of pulmonary hypertension from radiographic estimates of pulmonary arterial size. Thorax 1987;42:207–11. [3] Matthay RA, Schwarz MI, Ellis Jr JH, et al. Pulmonary artery hypertension in chronic obstructive pulmonary disease: determination by chest radiography. Invest Radiol 1981;16:95–100. [4] Haimovici JB, Trotman-Dickenson B, Halpern EF, et al. Relationship between pulmonary artery diameter at computed tomography and pulmonary artery pressures at right-sided heart catheterization. Acad Radiol 1997;4:327–34. [5] Schmidt HC, Kauczor HU, Schild HH, et al. Pulmonary hypertension in patients with chronic thromboembolism: chest radiograph and CT evaluation before and after surgery. Eur Radiol 1996;6:817–25. [6] Moore NR, Scott JP, Flower CD, et al. The relationship between pulmonary artery pressure and pulmonary artery diameter in pulmonary hypertension. Clin Radiol 1988;39:486–9. [7] Kauczor HU, Heussel CP, Fischer B, et al. Assessment of lung volumes using helical CT at inspiration and expiration: comparison with pulmonary function tests. AJR 1998;171:1091–5. [8] Arakawa A, Yamashita Y, Nakayama Y, et al. Assessment of lung volumes in pulmonary emphysema using multidetector helical CT: comparison with pulmonary function tests. Comput Med Imaging Graph 2001;25:399–404. [9] Zaporozhan J, Ley S, Eberhardt R, et al. Paired inspiratory/expiratory volumetric thin-slice CT scan for emphysema analysis: comparison of different quantitative evaluations and pulmonary function test. Chest 2005;128:3212–322. [10] Henk CB, Gabriel H, Fleischmann D, et al. Pulmonary hypertension and cor pulmonale. Radiologe 1997;37:388–401. [11] Dixon Jr JR. The International Conference on Harmonization Good Clinical Practice guideline. Qual Assur 1998;6:65–74. [12] Zahlten C, Juergens H, Everts CJG, et al. Portal vein reconstruction based on topology. Eur J Radiol 1995;19:96–100. [13] Zahlten C, Juergens H, Peitgen HO. Reconstruction of branching blood vessels from CT-data. In: Goebel M, Mueller H, Urban B, editors. Visualization in scientific computing. Vienna, A: Springer; 1995. p. 41–52. [14] Kuriyama K, Gamsu G, Stern RG, et al. CT-determined pulmonary artery diameters in predicting pulmonary hypertension. Invest Radiol 1984;19:16–22. [15] Ng CS, Wells AU, Padley SP. A CT sign of chronic pulmonary arterial hypertension: the ratio of main pulmonary artery to aortic diameter. J Thorac Imaging 1999;14:270–8. [16] Burakowska B, Pawlicka L, Oniszh K, et al. Value of spiral computed tomography in pulmonary hypertension. Pol Arch Med Wewn 2004;111:431–41. [17] Tan RT, Kuzo R, Goodman LR, et al. Utility of CT scan evaluation for predicting pulmonary hypertension in patients with parenchymal lung disease Medical college of Wisconsin lung transplant group. Chest 1998;113:1250–6. [18] Kauczor HU, Heussel CP, Thelen M. Update on diagnostic strategies of pulmonary embolism. Eur Radiol 1999;9:262–75. [19] Teichmann V, Jezek V, Herles F. Relevance of width of right descending branch of pulmonary artery as a radiologic sign of pulmonary hypertension. Thorax 1970;25:91–6. [20] O’Callagan JP, Heitzman ER, Somogyi JW, et al. CT-evaluation of pulmonary artery size. J Comput Assist Tomogr 1982;6:101–4. [21] Van Mierop LH, Kutsche LM. Embryology of the heart. In: Hurst JW, Logue RB, Rackley CE, Schlant RC, Sonnenblick EH, Wallace AG, Wenger NK, editors. The heart. New York: McGraw Hill; 1986. p. 1091.