Pulmonary Embolism in Pregnancy: Comparison of Pulmonary CT Angiography and Lung Scintigraphy

Transcription

1 Cardiopulmonary Imaging Original Research Ridge et al. CT and Scintigraphy of Pulmonary Embolism in Pregnancy Cardiopulmonary Imaging Original Research Carole A. Ridge 1 Shaunagh McDermott 1 Bridget J. Freyne 2 Donal J. Brennan 3 Conor D. Collins 1 Stephen J. Skehan 1 Ridge CA, McDermott S, Freyne BJ, Brennan DJ, Collins CD, Skehan SJ Keywords: CT, pregnancy, pulmonary embolism, radionuclide imaging DOI: /AJR Received January 7, 2009; accepted after revision May 10, Department of Radiology, St. Vincent s University Hospital, Elm Park, Dublin 4, Ireland. Address correspondence to C. A. Ridge (caroleridge@hotmail.com). 2 Curtin University of Technology, Perth, Western Australia. 3 Department of Obstetrics, National Maternity Hospital, Dublin, Ireland. AJR 2009; 193: X/09/ American Roentgen Ray Society Pulmonary Embolism in Pregnancy: Comparison of Pulmonary CT Angiography and Lung Scintigraphy OBJECTIVE. The purpose of this study was to retrospectively compare the diagnostic adequacy of lung scintigraphy with that of pulmonary CT angiography (CTA) in the care of pregnant patients with suspected pulmonary embolism. MATERIALS AND METHODS. Patient characteristics, radiology report content, additional imaging performed, final diagnosis, and diagnostic adequacy were recorded for pregnant patients consecutively referred for lung scintigraphy or pulmonary CTA according to physician preference. Measurements of pulmonary arterial enhancement were performed on all pulmonary CTA images of pregnant patients. Lung scintigraphy and pulmonary CTA studies deemed inadequate for diagnosis at the time of image acquisition were further assessed, and the cause of diagnostic inadequacy was determined. The relative contribution of the inferior vena cava to the right side of the heart was measured on nondiagnostic CTA images and compared with that on CTA images of age-matched nonpregnant women, who were the controls. RESULTS. Twenty-eight pulmonary CTA examinations were performed on 25 pregnant patients, and 25 lung scintigraphic studies were performed on 25 pregnant patients. Lung scintigraphy was more frequently adequate for diagnosis than was pulmonary CTA (4% vs 35.7%) (p = ). Pulmonary CTA had a higher diagnostic inadequacy rate among pregnant than nonpregnant women (35.7% vs 2.1%) (p < 0.001). Transient interruption of contrast material by unopacified blood from the inferior vena cava was identified in eight of 10 nondiagnostic pulmonary CTA studies. CONCLUSION. We found that lung scintigraphy was more reliable than pulmonary CTA in pregnant patients. Transient blood from the inferior vena cava is a common finding at pulmonary CTA of pregnant patients. P ulmonary embolism (PE) is the leading cause of maternal death in pregnancy [1]. Accurate and prompt diagnosis of PE during pregnancy is therefore essential. Lung scintigraphy and pulmonary CT angiography (CTA) are the imaging studies most commonly used in the diagnosis of PE during pregnancy. Neither the Fleischner Society nor the British Thoracic Society guidelines indicate which technique is preferred in pregnancy, although both organizations agree that pulmonary CTA is the first imaging test of choice in the general population because of its high sensitivity [2, 3]. Pulmonary CTA can be performed quickly and accurately to depict vessels to the level of sixth-order arterial branches [4]. Clinicians prefer this technique, particularly because it can be used to formulate an alternative diagnosis when PE has been excluded [5]. In our department, we have noticed with concern that a considerable number of pulmonary CTA studies performed on pregnant patients have poor pulmonary arterial opacification, making the images inadequate for diagnosis and causing artifacts that mimic thrombus (Fig. 1). This problem had led to repetition of CTA examinations and referrals for lung scintigraphy. Reports on imaging of PE during pregnancy have suggested that CT is less reliable for pregnant than nonpregnant patients [6, 7]. Factors such as the hyperdynamic state of pregnancy, hemodilution in pregnancy, and blood from the inferior vena cava (IVC) have been hypothesized as possible causes [7]. Contrast interruption leads to mixing of opacified blood from the superior vena cava (SVC) with unopacified blood from the IVC, resulting in poor opacification of the pulmonary arteries (Fig. 2). The causal mechanism AJR:193, November

2 Ridge et al. Fig year-old pregnant woman with dyspnea at 30 weeks gestation. Coronal reformatted pulmonary CT angiogram shows poor opacification of right pulmonary artery and ill-defined filling defect mimicking thrombus (arrow) with good contrast opacification of superior vena cava (SVC) and aorta (Ao). of inadequacy of pulmonary CTA studies for diagnosis during pregnancy has not, to our knowledge, been conclusively identified in the literature. The objectives of our study were to compare the diagnostic adequacy of lung scintigraphy with that of pulmonary CTA performed on pregnant patients with suspected PE and to identify, if possible, a cause of diagnostic inadequacy at image analysis. Materials and Methods Because this clinical study was retrospective, the institutional review board of our hospital exempted it from the requirement of approval and of informed consent from patients. Patient Groups Between July 1, 2006, and April 1, 2008, patients with suspected PE who had been consecutively referred for either pulmonary CTA or lung scintigraphy were included in the study. Patients were referred according to the preference of the referring clinician. However, if there was any lung parenchymal abnormality on the initial chest radiograph, pulmonary CTA was performed. For selection of pregnant patients, electronic records were searched for all referrals of women of childbearing age (15 45 years) for lung scintigraphy or pulmonary CTA. The pregnant patients within this group were selected on the basis of referral information on the computerized request form. Length of gestation, age, comments in the radiology report regarding nondiagnostic image quality, performance of further imaging studies, and final diagnoses were recorded. A computerized search of all pulmonary CTA and lung scintigraphic studies performed in the same time period was performed to establish the Fig year-old pregnant woman with pleuritic chest pain at 36 weeks gestation. Coronal reformatted pulmonary CT angiogram shows highattenuation contrast material in superior vena cava (SVC) and aorta (Ao), low-attenuation unopacified blood in inferior vena cava (IVC), and poor opacification of main pulmonary artery (MPA). overall institutional rate of diagnostic inadequacy among nonpregnant patients. This search yielded 1,420 pulmonary CTA and 96 lung scintigraphy reports, and a keyword search was performed with the terms inadequate, indeterminate, poor opacification, suboptimal, and nondiagnostic. This search yielded 30 pulmonary CTA examinations reported as inadequate for diagnosis. All lung scintigraphic studies performed on nonpregnant patients were considered diagnostic. Lung Scintigraphy Lung scintigraphy was performed with a double-head gamma camera equipped with low-energy, high-resolution parallel-hole collimators (Infinia, GE Healthcare). Perfusion scintigraphy was performed initially with low-dose (90 MBq) 99m Tclabeled macroaggregated albumin. If necessary, the perfusion scan was followed by a ventilation scan performed after inhalation of ultrafine 485 ± 72 MBq 99m Tc-carbon particles (Technegas, Cyclopharm) in a single long breath. All patients remained supine throughout the examination. Planar images in six standard views of pulmonary perfusion and ventilation were acquired with a matrix. Image acquisition was terminated in each plane when a satisfactory number of counts was reached (anterior and posterior acquisitions, 600 kilocounts; oblique perfusion acquisitions, 300 kilocounts; oblique ventilation acquisitions, 400 kilocounts). The ratio between count rates on the perfusion and ventilation scans was considered adequate if greater than 3.5. Pulmonary CTA Pulmonary CTA was performed with a 64- MDCT scanner (Somatom Sensation 64, Siemens Healthcare). Patients were examined in the supine position with both arms extended above the head. Images were acquired in the caudocranial direction during a single inspiratory breath-hold. The scan volume ranged from the costophrenic angles to the lung apices. A standard collimation of 0.6 mm was used with a gantry rotation speed of 0.33 seconds and a pitch factor of 0.9. Scanning was performed at 120 kv and 200 mas. Vessel opacification was achieved by IV injection of 75 ml of iopamidol (Niopam 370, Bracco) through a peripheral vein. The flow rate was kept constant at 4 ml/s throughout the procedure. The contrast agent was followed by a 50-mL saline flush. Individual contrast opacification was optimized with bolus tracking (CARE bolus, Siemens Healthcare) in the main pulmonary artery (MPA) at a trigger level of 100 HU. During bolus tracking the patient breathed quietly and was instructed to take a deep inspiration as soon as the threshold of attenuation in the MPA was reached. Image acquisition began at this point. Manual triggering was performed if this threshold was not reached. Standard departmental acquisition factors and breathing instructions were not altered because of a patient s pregnancy. Image Analysis Image analysis was performed with a standard workstation (Leonardo, Siemens Healthcare). The pulmonary CTA images were reconstructed at a slice thickness of 1 mm and reviewed at mediastinal windows (center, 50 HU; width, 350 HU). All pulmonary CTA measurements and calculations were performed by consensus of two radiologists with 17 and 3 years of experience in chest CT. All lung scintigraphic images were read by one of two staff radiologists with 12 and 10 years of experience in radionuclide imaging. The nondiagnostic lung scintigram was reviewed by the radiologist with 10 years of experience. Pulmonary CTA and lung scintigraphy of pregnant patients were classified as adequate or inadequate for the diagnosis of PE. The classification was based on comments in the electronic radiology reports regarding whether the studies were subjectively considered of diagnostic quality at the time of image acquisition. Objective image analysis was performed whereby all pulmonary CTA studies performed on pregnant patients were analyzed for MPA attenuation. Adequate opacification of the MPA was determined present if pulmonary artery opacification was greater than 250 HU [8]. Attenuation measurements were made by drawing a region of interest with an area equal to one-half the cross-sectional area of the vessel. To ensure reproducibility, MPA attenuation was measured in a standardized anatomic location 2 cm proximal to the MPA bifurcation. CTA images were assessed for the presence of artifact due to transient interruption of contrast 1224 AJR:193, November 2009

3 CT and Scintigraphy of Pulmonary Embolism in Pregnancy material by unopacified blood from the IVC. The following equation described in the literature [9] for calculation of the fraction of blood flow contributed by the IVC to the right side of the heart (K IVC ) was applied to the nondiagnostic CTA studies: K IVC = (C SVC C RA or RV )/(C SVC C IVC ). In this equation, the relative IVC contributions to the right atrium (RA) and right ventricle (RV) were calculated by equating attenuation in HU (C) in these chambers to a weighted average of the attenuations of the SVC and IVC. It was assumed that the SVC and IVC were the sole contributors of blood flow to the right side of the heart. To ensure reproducibility, attenuation measurements were made in standardized anatomic locations (distal SVC, proximal IVC, and the widest axial image of the right atrium and right ventricle). CTA studies confirmed to have transient interruption of contrast material by the IVC according to the equation were compared with K IVC values of a randomly selected age- and sex-matched control group of eight nonpregnant patients who had undergone diagnostic pulmonary CTA without artifact during the study period. Only one lung scintigraphic study was deemed inadequate for diagnosis and was subjectively assessed to determine the cause of diagnostic inadequacy. Statistical Analysis Age, gestation, and K IVC values were expressed as mean values. Homogeneity of variance was assessed with the independent Student s t test with application of Levene s coefficient. Statistical analyses were performed with the Student s t test for independent and paired variables and Pearson s chi-square test. Statistical significance was set at p < Calculations were performed on a standard PC with statistical software (SPSS version 14.0 for Windows, SPSS). Results Twenty-eight pulmonary CTA examinations were performed on 25 pregnant patients, and 25 lung scintigraphic studies were performed on 25 pregnant patients. Ten nondiagnostic pulmonary CTA examinations were performed on eight pregnant patients, and one nondiagnostic lung scintigraphic study was performed on a pregnant patient. All 25 patients referred for lung scintigraphy had normal findings on initial chest radiographs. Three of 25 patients referred for pulmonary CTA had abnormal findings on chest radiographs (two, lobar consolidation; one, pleural effusion). One patient referred for pulmonary CTA had a small pneumothorax, which was not detected on the initial chest radiograph. TABLE 1: Mean Age and Gestation According to Imaging Technique Pulmonary CT angiography Technique Age (y) Gestation (wk) All 32.6 ± ± 6.4 Diagnostic 30.8 ± ± 6 Nondiagnostic 37 ± ± 6.7 Lung scintigraphy All 31.8 ± ± 8.9 Diagnostic 31.7 ± ± 8.7 Nondiagnostic Note Values are mean ± SD. Rate of Diagnostic Inadequacy The rate of diagnostic inadequacy was significantly lower for lung scintigraphy than for CTA performed on pregnant patients (4% vs 35.7%) (p < ). The rate of diagnostic inadequacy of CTA also was lower among nonpregnant than pregnant patients (2.1% vs 35.7%) (p < 0.001). Patient Age and Gestation Application of Levene s coefficient showed that variances were equal (p > 0.05), and thus homogeneity was assumed. Table 1 displays the mean ages and gestation times of the groups according to imaging technique. Results of comparison of the mean ages of the pulmonary CTA and lung scintigraphy groups were not statistically significant (32.6 ± 5.6 [SD]) years vs 31.8 ± 5.4 years) (p = 0.579). Results of comparison of mean gestation times in the pulmonary CTA and lung scintigraphy groups also were not statistically significant (30.6 ± 6.4 weeks vs 26.5 ± 8.9 weeks) (p = 0.108). A comparison of the mean ages of the diagnostic and nondiagnostic pregnant CTA groups showed a statistically significant relation between age and diagnostic inadequacy (30.8 ± 5.6 years vs 37 ± 3.5 years) (p = 0.005). The relation between gestation and diagnostic inadequacy in both pulmonary CTA groups was not significant (p = 0.99). Pulmonary CTA Image Evaluation The difference in average attenuation of the MPA in the nondiagnostic image pregnant group compared with the diagnostic image pregnant group was statistically significant (111.4 ± 44.8 HU vs ± 50.4 HU) (p < 0.001). Only four of 18 diagnostic studies had an MPA attenuation greater than 250 HU; that is, 14 studies (77%) did not meet departmental criteria for satisfactory opacification of the MPA despite being subjectively considered adequate for diagnosis at the time of image acquisition. In the control group, mean attenuation of the MPA was ± 43 HU, and all images were considered diagnostic at the time of acquisition. Nondiagnostic pulmonary CTA studies of pregnant patients were further analyzed. Reasons for diagnostic inadequacy identified were transient interruption of contrast material by unopacified blood from the IVC (n = 8), missed bolus (n = 1), and poor subsegmental opacification (n = 1) despite adequate MPA opacification of 213 HU. Contribution of IVC to Right Side of the Heart Table 2 summarizes the relative IVC contributions to the right atrium and ventricle calculated for the nondiagnostic pulmonary CTA studies in which transient interruption of contrast material was identified (n = 8) compared with age- and sex-matched controls. group, the average relative IVC contributions to the right atrium and right ventricle were 91 ± 8% and 87 ± 6%, indicating a larger relative contribution from the IVC to the right heart than from the SVC. Among the controls, the average relative IVC contributions to the right atrium and right ventricle were 42 ± 14% and 62 ± 4%, indicating approximately equal contributions of both the IVC and SVC. The differences in K IVC values between case and control patients were statistically significant (right atrium, p < ; right ventricle, p < ). Evaluation of Lung Scintigrams Among 25 lung scintigraphic studies performed on pregnant patients, one was inadequate for diagnosis (4%). The patient had a history of smoking, and the study was nondiagnostic because of multiple small predominantly matched abnormalities of ventilation and perfusion. Twenty-one of 23 patients AJR:193, November

4 Ridge et al. TABLE 2: Mean Density Measurements and Calculated Relative Inferior Vena Caval Contributions to the Right Side of the Heart in Women Who Underwent Pulmonary CT Angiography Value (91.3%) underwent half-dose perfusion-only studies. Two patients underwent ventilation perfusion studies, and one of those patients whose initial CTA study was nondiagnostic was found to have PE. Additional Imaging Studies group (eight patients, 10 CTA studies), three patients (37.5%) underwent repeated pulmonary CTA. One of these patients had a diagnostic study that excluded PE, and the other two patients had nondiagnostic repeated studies. One of these patients was thought to be at high risk of PE and underwent lung scintigraphy, which excluded PE. Two patients underwent lung scintigraphy after the initial nondiagnostic CTA, and PE was diagnosed in one patient. Three patients did not undergo further imaging, and PE was excluded at a 6-month clinical follow-up examination. The patient who had a nondiagnostic lung scintigraphic study underwent pulmonary CTA. Final Diagnosis group (eight patients, 10 CTA studies), one patient had lobar pneumonia and one had PE diagnosed with lung scintigraphy. Among the other six patients, PE was excluded at further imaging (n = 3) or 6-month clinical follow-up examination (n = 3). In the diagnostic pulmonary CTA group (n = 18), spontaneous pneumothorax (n = 1), pleural effusion (n = 1), and lobar pneumonia (n = 1) were diagnosed. No cases of PE were diagnosed with CTA. In the diagnostic lung scintigraphy group (n = 24), PE was diagnosed in two patients. The other 22 had normal studies. The patient who had a nondiagnostic lung scintigraphic study was referred for pulmonary CTA, the findings of which were normal. Pregnant With Nondiagnostic Images Nonpregnant With Diagnostic Images Attenuation (HU) Superior vena cava 441 ± ± 80 Inferior vena cava 48 ± ± 17 Right atrium 83 ± ± 107 Right ventricle 99 ± ± 40 K IVC right atrium 91 ± 8 42 ± 14 K IVC right ventricle 87 ± 6 62 ± 4 Note Values are mean ± SD. K IVC = fraction of blood flow contributed by inferior vena cava to the right side of heart. Nonpregnant Patients Thirty of 1,420 pulmonary CTA studies (2.1%) performed on nonpregnant patients were considered inadequate for diagnosis. The causes of diagnostic inadequacy were poor subsegmental opacification (n = 9), poor enhancement of the MPA (n = 8), missed bolus due to scan timing (n = 7), interruption of contrast material by unopacified blood from the IVC (n = 5), and extravasation of contrast material (n = 1). Eighteen of these patients (60%) had concomitant pleural parenchymal disease. All lung scintigraphic studies of nonpregnant patients were considered adequate for diagnosis at the time of image acquisition (n = 96). Discussion Appropriate care of a pregnant patient with suspected PE demands prompt diagnosis with a reliable imaging study. The British Thoracic Society and Fleischner Society Guidelines call for pulmonary CTA for the general population [2, 3]. We performed an audit of pregnant patients with suspected PE that showed lung scintigraphy produces diagnostic images more frequently than does pulmonary CTA, most nondiagnostic CTA studies showing an artifact due to transient blood from the IVC. Artifact due to contrast interruption by unopacified blood from the IVC has been studied in nonpregnant patients [9]. Pulmonary CTA images are usually acquired after deep inspiration that begins soon after contrast injection has begun. Venous return to the right side of the heart increases nearly 50% during deep inspiration, and most of this venous return originates from the IVC [10]. Deep inspiration causes a decrease in intrathoracic pressure and a pressure gradient known as the thoracoabdominal pump. Wittram and Yoo [9] reported a 3% incidence of this artifact among nonpregnant patients. Our patient group was given the same prescan instructions detailed by those investigators. A possible explanation for the difference in incidence of this artifact between nonpregnant and pregnant patients is that in pregnancy, the resting pressure in the IVC increases substantially in the supine position, particularly in the third trimester, during which a sixfold increase in IVC pressure occurs [11]. Increased IVC pressure combined with the expected decrease in intrathoracic pressure during deep inspiration may increase venous return owing to the thoracoabdominal gradient. Another potential explanation is that a collateral system involving the vertebral and azygos systems develops and alleviates the mechanical effect of the gravid uterus in late pregnancy [11], which may contribute to increased venous return to the right side of the heart. Two of the 10 nondiagnostic pulmonary CTA studies did not show the artifact due to blood from the IVC. The difference in these two sets of images may be accounted for by the known hemodynamic alterations of pregnancy, including increases in cardiac output, total vascular resistance, heart rate, and plasma volume [12], which lead to dilution of contrast material [7]. Results of experimental studies [13] have shown that cardiac output is inversely related to peak arterial enhancement and time to arrival of contrast material in the aorta. Therefore, the increase in cardiac output in pregnancy may lead to poor peak arterial enhancement and a shorter contrast material arrival time [13]. This phenomenon also may explain why only four of the 18 pulmonary CTA studies considered adequate for diagnosis at image acquisition showed greater than 250 HU attenuation in the MPA. A review of PE imaging in pregnancy [7] discussed methods of overcoming the hemodynamic effects of pregnancy, including bolus triggering with shorter scan delay, high flow rate of contrast material flow, high concentration of contrast medium, and use of low-kilovoltage techniques. These alterations have been made to the pulmonary CTA protocol in our department. Solutions to the problem of interruption of contrast material suggested in the literature include image acquisition during shallow inspiration [9] or held expiration [14]. Pulmonary CTA studies in our department now are performed during shallow-held inspiration after adequate coaching by a technologist AJR:193, November 2009

5 CT and Scintigraphy of Pulmonary Embolism in Pregnancy The findings of our study are similar to the results of two studies [6, 15] of the quality of pulmonary CTA in pregnancy in which it was concluded that poor pulmonary arterial opacification is more common in pregnant than in nonpregnant women. U-King-Im et al. [6] found a 27.5% rate of nondiagnostic pulmonary CTA among pregnant patients, which resembled our rate of 35.7%. Andreou et al. [15] found mean MPA enhancement in pregnant patients was significantly lower than that in a nonpregnant control group, which was also a frequent finding in our study. The diagnostic yield of lung scintigraphy in pregnancy in our group was 96%. This issue also has been investigated by Scarsbrook et al. [16], who found a high diagnostic yield with lung scintigraphy (93%), and by Ezwawah et al. [17], who reported a 100% diagnostic yield using perfusion-only scintigraphy. Our direct comparison of the performance of lung scintigraphy with that of pulmonary CTA during pregnancy was the first published report of a direct comparison to our knowledge. Despite that, the first limitation of the study was that the number of subjects was small enough to limit the statistical power. The number is a reflection of the low referral rate among a typically young pregnant population with few comorbid medical conditions. A second limitation was the absence of echocardiographic data to exclude a patent foramen ovale or atrial septal defect. Calculation of K IVC was based on the assumption that the SVC and IVC were the sole contributors of blood flow to the right side of the heart. We did not have echocardiographic data because there were no clinical indications for performing echocardiography. None of the study subjects, however, had a history of cardiac disease, and the degree of contrast dilution was so great in the nondiagnostic studies that it is unlikely to have been due to unopacified blood passing through a small asymptomatic patent foramen ovale or atrial septal defect. A final limitation is that it can be argued that the high diagnostic adequacy rate in the lung scintigraphy group was a result of selection bias because all of these patients had normal chest radiographs and those referred for pulmonary CTA were therefore more likely to have pleural parenchymal disease contributing to nondiagnostic pulmonary CTA. However, the incidence of pleural parenchymal disease was similar in both the diagnostic and the nondiagnostic CTA groups, so selection bias in favor of lung scintigraphy was not likely. For pregnant patients, lung scintigraphy proved to be a more reliable imaging technique for the diagnosis or exclusion of PE than did pulmonary CTA, predominantly because of interruption of contrast material by unopacified blood from the IVC during CTA. Lung scintigraphy should therefore be considered the technique of choice for imaging of pregnant patients with suspected PE unless the image quality of pulmonary CTA can be optimized with adapted breathing maneuvers and contrast administration. Acknowledgments We thank Conrad Wittram of Massachusetts General Hospital and, for statistical advice, Leslie Daly of the Health Research Board Funded Center for Support and Training in Analysis and Research, University College, Dublin. References 1. Pabinger I, Grafenhofer H. Thrombosis during pregnancy: risk factors, diagnosis and treatment. Pathophysiol Haemost Thromb 2002; 32: Remy-Jardin M, Pistolesi M, Goodman LR, et al. Management of suspected acute pulmonary embolism in the era of CT angiography: a statement from the Fleischner Society. Radiology 2007; 245: British Thoracic Society Standards of Care Committee Pulmonary Embolism Guideline Development Group. British Thoracic Society guidelines for the management of suspected acute pulmonary embolism. Thorax 2003; 58: Ghaye B, Szapiro D, Mastora I, et al. Peripheral pulmonary arteries: how far in the lung does multi-detector row spiral CT allow analysis? Radiology 2001; 219: van Strijen MJ, Bloem JL, de Monyé W, et al.; Antelope-Study Group. Helical computed tomography and alternative diagnosis in patients with excluded pulmonary embolism. J Thromb Haemost 2005; 3: U-King-Im JM, Freeman SJ, Boylan T, Cheow HK. Quality of CT pulmonary angiography for suspected pulmonary embolus in pregnancy. Eur Radiol 2008; 18: Schaefer-Prokop C, Prokop M. CTPA for the diagnosis of acute pulmonary embolism during pregnancy. Eur Radiol 2008; 18: Jones SE, Wittram C. The indeterminate CT pulmonary angiogram: imaging characteristics and patient clinical outcome. Radiology 2005; 237: Wittram C, Yoo AJ. Transient interruption of contrast on CT pulmonary angiography: proof of mechanism. J Thorac Imaging 2007; 22: Kuzo RS, Pooley RA, Crook JE, Heckman MG, Gerber TC. Measurement of caval blood flow with MRI during respiratory maneuvers: implications for vascular contrast opacification on pulmonary CT angiographic studies. AJR 2007; 188: Scott DB, Kerr MG. Inferior vena caval pressure in late pregnancy. J Obstet Gynaecol Br Commonw 1963; 70: Valensise H, Vasapollo B, Novelli GP. Maternal cardiovascular haemodynamics in normal and complicated pregnancies. Fetal Matern Med Rev 2003; 14: Bae KT, Heiken JP, Brink JA. Aortic and hepatic contrast medium enhancement at CT. Part 2. Effect of reduced cardiac output in a porcine model. Radiology 1998; 207: Chen YH, Velayudhan V, Weltman DI, et al. Waiting to exhale: salvaging the nondiagnostic CT pulmonary angiogram by using expiratory imaging to improve contrast dynamics. Emerg Radiol 2008; 15: Andreou AK, Curtin JJ, Wilde S, Clark A. Does pregnancy affect vascular enhancement in patients undergoing CT pulmonary angiography? Eur Radiol 2008; 18: Scarsbrook AF, Evans AL, Owen AR, Gleeson FV. Diagnosis of suspected venous thromboembolic disease in pregnancy. Clin Radiol 2006; 61: Ezwawah O, Alkoteesh J, Barry JE, Ryan M. Pulmonary embolism in pregnancy: is nuclear medicine still a valid option? Ir Med J 2008; 101: AJR:193, November