Multimodal indirect imaging signs of pulmonary embolism

Abstract

The clinical diagnosis of pulmonary embolism is often difficult, as symptoms range from syncope and chest pain to shock and sudden death. Adding complexity to this picture, some patients with non-diagnosed pulmonary embolism may undergo unenhanced imaging examinations for a number of reasons, including the prevention of contrast medium-related nephrotoxicity, anaphylactic/anaphylactoid reactions and nephrogenic systemic fibrosis, as well as due to patients’ refusal or lack of venous access. In this context, radiologists’ awareness and recognition of indirect signs are cornerstones in the diagnosis of pulmonary embolism. This article describes the indirect signs of pulmonary embolism on chest X-ray, unenhanced CT, and MRI.

Introduction

Despite advances in imaging and laboratory tests in recent decades, the diagnosis of pulmonary embolism (PE) continues to be a medical challenge, primarily because the clinical scenarios of this disease are diverse, ranging from asymptomatic disease, dyspnea, and chest pain to hemodynamic instability, hypotension, and shock. On the other hand, some patients with clinically suspected PE may receive alternative diagnoses upon the completion of investigation. The modern diagnostic approach for PE based on clinical pretest probability scores, such as the modified Wells and Geneva scores is definitive for decision management. However, even in patients with low pretest clinical probability, a prevalence of up to 10% may be acceptable according to these scores.¹

Given the potentially severe consequences of PE and the risks of therapeutic anticoagulation, this disease should be promptly suspected and confirmed or excluded with reasonable certainty. In this context, imaging is a cornerstone of investigation. The definitive diagnosis of acute and chronic PE is based on the detection of direct arterial findings with various appearances. Angiotomography (angio-CT) is very accurate for the diagnosis of PE, as it shows direct signs of filling defects, but its indication depends on clinical suspicion and the absence of restrictions regarding venous contrast medium injection. Planar ventilation/perfusion lung scintigraphy (V/Q) is an established diagnostic test for suspected PE, and along with V/Q single photon emission CT (SPECT) and MRI may be alternative modalities for PE diagnosis in specific clinical contexts.¹

The identification of indirect ancillary findings may be important in several situations, such as with the use of technically inadequate examinations providing poor visualization of vascular opacification, in cases of distal arterial embolism, and with the use of unenhanced examinations due to concerns regarding renal function or to clinically unsuspected PE. In this context, the detection of indirect signs of PE using other modalities may be of paramount importance, avoiding late diagnosis and complications from the disease. In the following sections, we describe indirect findings of PE on chest X-ray (CXR), unenhanced CT, and MRI.

Chest X-ray

CXR has traditionally been the initial thoracic imaging modality used for the evaluation of patients with acute chest symptoms. Unfortunately, radiographic signs are neither sensitive nor specific for PE, and the main contribution of CXR in this context is to establish alternative diagnoses in patients with respiratory symptoms. Focal lung oligemia and pulmonary infarction were the first CXR signs of PE to be recognized. Focal oligemia, or the Westermark sign, is an area of increased lung transparency distal to an occluded vessel; pulmonary infarction appears on CXR as a pleural-based wedge shaped consolidation, or Hampton’s hump (Figure 1). Two other signs reflecting vascular patterns may be observed: Fleischner’s sign, related to pulmonary artery enlargement; and Palla’s sign, indicating vascular proeminence before an arterial occlusion (Figure 2). Also unilateral diaphragm elevation may be observed.²

Figure 1.

Chest X-ray showing a pleural-based wedge-shaped consolidation in the right lower lobe (Hamptons’ hump; arrow), which was confirmed by angio-CT to be a pulmonary infarction, in a patient with acute pulmonary embolism.

Figure 2.

Chest X-ray showing enlargement of the main pulmonary artery (Fleischner’s sign; white arrow) and prominence of the right descending pulmonary artery (Palla’s sign; black arrows) in a patient with chronic pulmonary embolism.

However, as shown in the Prospective Investigation of Pulmonary Embolism Diagnosis study, CXR does not provide adequate information for PE diagnosis, and its main value was to exclude alternative entities such as pneumonia.^3,4 In this study, the most common findings were atelectasis, areas of increased density and pleural effusion. Still of note, a normal radiograph had a predictive negative value of only 74%.² Values of sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for Westermark sign (14%, 92%, 38%,76%) and Hampton hump (22%, 82%, 29%, 76%) were also reported in Prospective Investigation of Pulmonary Embolism Diagnosis.²

Computed tomography

Widely available, angio-CT is a safe and rapid imaging modality that can be used to confirm or exclude the diagnosis of PE, and to point to alternative diagnoses. Direct signs of PE in angio-CT studies include: (a) complete arterial occlusion, (b) partial filling defects surround by contrast medium, and (c) peripheral partial filling defects forming acute angles with the arterial wall. Given the complexity of clinical diagnosis, some patients with unsuspected PE may undergo unenhanced chest CT for the evaluation of several other cardiopulmonary conditions; in some cases, the symptoms of PE mimic abdominal pathologies, and a partial view of the lung bases on abdominal CT may provide clues leading to the diagnosis of PE. Problems related to the lack of venous access, renal failure, and the risk of allergic/allergoid reactions may also influence radiologists’ decision to perform unenhanced imaging studies in patients with cardiopulmonary symptoms and without a clear clinical picture of PE. All of these points justify the need for attention to indirect signs on unenhanced CT.^5,6

Lung parenchymal signs

Mosaic perfusion

Mosaic perfusion is defined as a patchwork of differing attenuation that may represent: (a) patchy interstitial disease, (b) obliterative airway disease or (c) occlusive vascular disease. In occlusive vascular disease, lung hypoperfusion appears as low-attenuation areas in which the vascular caliber is reduced due to narrowing or occlusion of affected arterial territories (Figure 3). Conversely, ground-glass opacities may appear interritories with patent pulmonary arterial vasculature, presumably due to pulmonary edema. Mosaic perfusion may be recognized more easily in cases of chronic pulmonary thromboembolism, but it may also be seen in acute PE.⁷ Since airway diseases may also be associated with mosaic perfusion, careful examination of bronchial morphology and eventually additional expiratory acquisition searching for air trapping should be made before suggesting occlusive arterial disease.⁷

Figure 3.

(A) Axial CT image in the lung window setting with maximal intensity projection reformatting showing extensive mosaic perfusion and, of note, multiplanar reformatting highlights the differences in vessel caliber in hypoattenuating and hyperattenuating lung zones. Axial angio-CT imaging confirms chronic pulmonary embolism (arrows in B).

Pulmonary infarction

The lung parenchyma has a dual blood supply from the pulmonary and bronchial arteries, both connected by capillary anastomoses. Although this system protects the lung against ischemic events, pulmonary infarction may occur, and it is more common in peripheral than in central PE. The classical appearance of pulmonary infarction on CT is wedge-shaped, pleural-based consolidation. The ancillary finding of central lucencies is a good predictor of pulmonary infarction in a peripheral consolidation (98% specificity, 46% sensitivity), conversely, the finding of air bronchograms inside a consolidation makes it unlikely related to lung infarction.⁸

The reversed halo sign (RHS) has also been described as a morphological appearance of infarction, and was found in 18% of patients with pulmonary embolism⁹ (Figure 4). Although the reversed halo sign is a non-specific finding, some morphological characteristics of the lesion may narrow the differential diagnosis toward pulmonary infarction. Thus, low-attenuation areas inside the halo, with or without reticulation, are highly suggestive of PI. The importance of this finding in diagnosing pulmonary infarctions in immunocompetent patients has been stressed in recent literature, with frequencies ranging from 88 to 95% and specificity and sensitivity values of 98 and 46%.^8–10 Lower-lung predominance, subpleural involvement and pleural effusion also suggest PI.^8–10 Of note, the finding of air bronchograms inside a consolidation makes it unlikely related to lung infarction.⁸ Some attention may be payed to peripheral, faint ground-glass opacities, which may represent an early sign of pulmonary infarction¹¹ (Figures 5 and 6). Over weeks, a lung infarction decreases in size, becoming a flat peripheral scar in the lung, the so-called “melting sign”; such scarring lesions may also be observed in cases of chronic PE.

Figure 4.

(A) Axial CT image in the lung window setting and (B) axial angio-CT image. (A) shows a wedge-shaped pleural-based consolidation with the reversed halo sign and internal reticulation in the right lower lobe, compatible with pulmonary infarction (arrow). (B) Axial angio-CT confirmed the presence of a filling defect in the same arterial territory (arrow).

Figure 5.

Axial CT images in lung window settings (A, B). In (A), a faint ground-glass opacity is visible on the right inferior lobe (arrow); on an examination performed 2.5 months later (B), it had become a fan-shaped consolidation compatible with pulmonary infarction (arrow), and an arterial filling defect was visible on the right inferior interlobar artery (not shown).

Figure 6.

(A) Axial CT image with the lung window setting and (B) angio-CT image. Ground-glass opacity is seen on the right inferior lobe (arrow in A), and angio-CT confirmed the presence of a peripheral thrombus in the same vascular territory (arrow in B).

Vascular findings

Spontaneous thrombus visualization

The spontaneous visualization of thrombus inside pulmonary arteries in an unenhanced study has a variable reported frequency, ranging from 6 to 41.2%, and is associated mainly with central PE, presenting as hyperattenuating, hypoattenuating, and even calcified thrombi.^12–14 Such phenomena may be explained by the reduction of hydric content as the thrombus retracts, which increases its hemoglobin concentration and elevates its attenuation value above that of the regional blood pool, leading to a hyperattenuating presentation (Figure 7A). On the other side, thrombus attenuation decreases over time, generating attenuation values inferior to those of the blood pool, and hypoattenuating thrombi may be detected spontaneously on unenhanced CT images (Figure 7B). Calcified thrombi are observed in a small number of patients with chronic PE (Figure 7C).^12,13,15 Detecting hyperdense material in pulmonary arteries lumen had an over all sensitivity of 36%, specificity of 99%, PPV of 90% and NPV of 85.6% for pulmonary embolism, and when considering central pulmonary embolism it was found a sensitivity of 66.7%, specificity of 99.1%, PPV of 88.9% and NPV of 96.4%. Although it seems to be an important indirect sign for central pulmonary embolism, its value is limited in peripheral disease.¹⁶

Figure 7.

Axial unenhanced CT images with mediastinal settings showing spontaneous characterization of pulmonary thrombi. (A) Saddle-shaped highattenuation thrombus (arrow); (B) low-attenuation clot (arrow); (C) a chronic calcifiedperipheral thrombus (arrow).

Magnetic resonance imaging

Recent technical advances have reduced the examination time and improved the spatial resolution of MRI, resulting in fewer motion artifacts and increased acceptance of the method. Unenhanced imaging with steady-state free precession sequences (SSFPS) allows luminal visualization without the need for breath-holding. Conventional T1 andT2 sequences may provide additional information on the lung morphology and vascular bed. Indirect findings may provide support for the diagnosis of PE in cases of technically limited contrast-enhanced angiographic examination, or with the use of unenhanced MRI.

Lung parenchymal signs—pulmonary infarction

Wedge-shaped subpleural consolidations compatible with pulmonary infarcts have variable patterns on MRI, depending on the age of the underlying blood in the lung parenchyma. Acute infarcts with <24 h evolution present with low T1 and high T2 signals, whereas subacute infarcts up to 1 week of age show T1 hyperintensity (Figure 8).

Figure 8.

(A) Axial angio-CT image, (B) axial CT image in the lung window setting, (C) unenhanced axial T1 magnetic resonance image, and (D) diffusion-weighted image. Images show a pulmonary thrombus in the artery of the right lower lobe (arrow in A) and a pleural-based wedge-shaped consolidation in the same arterial territory (arrows in B, C, and D). In (C), note the hyperintensity inside the consolidation, suggestive of a hemorrhagic component. In (D), note the high signal intensity inside the consolidation (reflecting restriction), which was accompanied by a low apparent diffusion coefficient (not shown).

A recent diffusion-weighted MRI series revealed a significant difference in apparent diffusion coefficient values between pulmonary infarcts and lung atelectasis, showing the value of this technique for the differentiation of these conditions.^17,18

Vascular signs—spontaneous characterization of thrombus

Direct thrombus identification in the subacute phase is based on the detection of methemoglobin, known to reduce T1 and thus cause a high signal on heavily weighted T1 sequences (Figure 9). Of note, this sign is observed in subacute thrombosis, helping to differentiate between old and new clots.¹⁹ SSFPS images allow luminal vascular evaluation without the use of gadolinium enhancement. These sequences are included in most conventional MRI protocols, and have the potential advantage of resistance to motion degradation, allowing satisfactory imaging in patients incapable of full apnea. In SSFPS sequences, intravascular clots are characterized as signal voids within the high-intensity blood pool (Figure 10). SSFPS has shown good overall performance and inter reader concordance for PE detection, with overall sensitivity, specificity, PPV and NPV values in per-embolus analysis of as much as 85%, 100%, 100% and 95,8% respectively.^17,20 The general performance of SSFPS may be lower for distal arterial branches.^17,20,21 The radiologist must be aware about off-resonance artifacts, generally attributable to unwanted magnetic field variations inside the imaging volume, which may simulate thrombi in vessels lumen.²² Repeating SSFPS in other planes is a useful strategy for surpassing this artifact and avoiding false-positive embolism.¹⁷ Although finding a thrombus in a proximal pulmonary arterial branch in a non-enhanced MRI sequence may suffice for a PE diagnosis, associating a complimentary method may be warranted for accessing subsegmental or distal arterial ramifications.

Figure 9.

Unenhanced axial T1 fat-saturated magnetic resonance image showing high-intensity clots are visible on the right pulmonary artery and left descending inferior artery (arrows).

Figure 10.

(A) Angio-CT image and (B) image from a true fast imaging with steady-state precession sequence, similarly showing central filling defects in the right pulmonary artery.

Conclusion

Imaging plays a critical role in the diagnosis of PE, and indirect signs on unenhanced examinations may be the only clues generating the suspicion of thromboembolism in patients with clinically unsuspected PE. Given the serious repercussions of undiagnosed PE, radiologists should pay attention to indirect signs that may indicate the need for additional enhanced imaging studies.