Pericarditis is classified based on duration of the acute episode—acute for those with pericarditis for less than 4 weeks; incessant for ongoing pericarditis between at least 4–6 weeks but less than 3 months without remission, chronic for pericarditis lasting for over 3 months, and recurrent pericarditis refers to 2 or more episodes of acute pericarditis with at least a 4–6 week interval without symptoms [1]. Recurrent pericarditis events occur in 15–30% after the first episode of pericarditis, and in 50% after 2 or more episodes [7, 8, 9, 10]. The diagnostic criteria for pericarditis based on guidelines include at least 2 out of the 4 criteria of (1) pericarditis chest pain, (2) pericardial rub on physical examination, (3) new ECG changes (widespread ST-elevation or PR depression) and new or worsening pericardial effusion; while other supporting features include elevated inflammatory markers (such as C-reactive protein, sedimentation rate and white cell count), and imaging evidence of pericardial inflammation (especially by MRI, Fig. 1) [1]. The main etiology categories of pericarditis are idiopathic, infective (most commonly viral, but also bacterial including tuberculosis, fungal and parasitic), autoimmune, iatrogenic (such as post-cardiac surgery and interventions), neoplastic, metabolic and drug-related. Tuberculosis pericarditis is mainly prevalent in developing countries and rare in developed countries.

Fig. 1.

Multi-modality imaging features of acute pericarditis case. (A) Small pericardial effusion (arrow) subcostal view of echocardiography. (B) Pericardial thickening (arrow) on black-blood spin echo sequence of magnetic resonance imaging (MRI). (C) Severe circumferential increased pericardial signal indicating edema (arrow) on T2-weighted short tau inversion recovery imaging of MRI. (D) Severe circumferential pericardial enhancement indicating inflammation/fibrosis (arrow) on delayed gadolinium enhancement sequence of MRI.

Based on this criteria, echocardiography is the first-line imaging modality recommended for all patients undergoing pericarditis evaluation, although it is often normal [1, 2]. Apart from its main role in identifying pericardial effusion, echocardiography can examine the presence of tamponade physiology present in approximately 3% of acute pericarditis, identify pericardial thickening, evaluate for the presence of regional wall motion abnormalities that may indicate concurrent myocardial involvement (myo-pericarditis) in approximately 5%, or look for alternative diagnoses like acute coronary syndrome or aortic dissection [2]. Echocardiography is also important during follow-up after initial pericarditis event for resolution of pericardial effusion if it had been present, along with signs of constrictive physiology, discussed in the later section.

In contrast to echocardiography and MRI, CT has a limited role for evaluating pericardial inflammation. Pericardial thickening which may enhance of iodinated contrast and the presence of pericardial effusion may be observed on chest CT when there is pericarditis [11]. CT is however usually ordered for evaluating alternative causes of chest symptoms such as coronary artery disease, acute aortic syndrome, pulmonary thromboembolism and other lung pathologies [12]. In chronic pericarditis, CT can identify calcifications that may indicate the presence of constriction that should be confirmed by other imaging modalities [2].

Perhaps the most important application of MRI in pericardial diseases is its ability to identify pericardial inflammation [2, 10]. The key features include pericardial thickening, best assessed on black-blood spin echo sequences; pericardial edema, assessed using T2-short tau inversion recovery (STIR) sequences as high signal intensity; and inflammation or fibrosis on late gadolinium enhancement sequences again as high signal intensity (Fig. 1) [2, 13, 14]. Histologically, pericardial late gadolinium enhancement correlates with fibroblastic proliferation, neovascularization and chronic inflammation and granulation tissue [15]. Some studies have reported moderate sensitivity (63–68%) and high specificity (up to 100%) of the T2-STIR sequence for acute pericarditis, however this is significantly lower in practice, as elevated signal can also be seen with pericardial effusion or MRI artefact [16, 17]. The delayed enhancement sequence has been reported to have moderate to high 65–100% sensitivity and high specificity 99–100% for pericarditis, however again in practice this is lower with pericardial fat, pleuritis and artefact potentially interfering with scan interpretation, and fat saturated pulses added to delayed enhancement sequences are recommended to improve the positive predictive value of pericardial enhancement [14, 16, 17, 18]. MRI can also evaluate for concomitant myocardial involvement and inflammation (myocarditis) where they be left ventricular dysfunction and regional wall motion abnormalities on cine sequences; increased myocardial signal intensity on T2-STIR or elevated T2-mapping values implying myocardial edema; increased myocardial signal intensity on delayed gadolinium enhancement sequences or elevated T1-mapping values consistent with myocardial inflammation and fibrosis; and along with early gadolinium enhancement suggesting hyperemia [19]. As such MRI is strongly recommended in the initial diagnosis of pericarditis with a complimentary role to clinical, inflammatory biomarkers and echocardiography assessment, especially if the diagnosis remains uncertain after the other tests [1, 2, 10].

Based on MRI and clinical findings, the staging of pericarditis have been proposed which can guide its management [20]. In acute pericarditis, both T2-STIR and delayed gadolinium enhancement are positive for increased signal intensity. The next stage of recurrent pericarditis, delayed enhancement remains positive, while T2-STIR may be positive or negative. In the chronic pericarditis phase, T2-STIR becomes negative while delayed enhancement remains positive. Finally in the burnt out phase, both T2-STIR and delayed enhancement of the pericardium are typically negative. Whereas in the first 2 to 3 phases anti-inflammatory therapies are the cornerstones to therapy, in the refractory or burnt out phase, diuresis for symptoms is often required and pericardiectomy surgery may be considered [10]. Each of these 4 phases may correspond to the transient, subacute or effusive constrictive, chronic and calcific stages of constrictive physiology [20]. Techniques for pericardial enhancement quantitation and semi-quantitation have been proposed with reasonably high intra and inter-observer agreement, and some studies have demonstrated their prognostic value, for example higher grade corresponding to higher rate of treatment failure with anti-inflammatory therapies, higher risk of pericarditis recurrence, but also higher likelihood of constrictive physiology improvement as discussed in a later section [21, 22]. Based on all of these techniques, MRI is a valuable tool in the surveillance of pericarditis to assess treatment response to help guide weaning or identifying the need for additional anti-inflammatory agents.

Pericardial effusion arises from increased production and/or decreased resorption of pericardial fluid leading to abnormal accumulation in the space between the visceral and parietal pericardium [1]. Echocardiography remains the first-line and preferred imaging modality to evaluate for pericardial effusion, and the size is generally classified as trivial (effusion only seen in systole), small (1 cm), moderate (1.0–1.9 cm), large ($>$2 cm) and very large ($>$2.5 cm), irrespective of the presence of tamponade [1, 2]. This should be measured when the adjacent chamber is at its largest during the cardiac cycle (end-ventricular diastole for ventricles, end-ventricular systole for atria), and size is graded as trivial if the pericardial effusion can be seen at some points but not others during the cardiac cycle and 1 cm, whereas small means effusion is seen throughout the cardiac cycle [2, 23]. The effusion can be circumferential or localized, therefore careful interrogation is necessary of parasternal, apical and subcostal windows are required, and sometimes transesophageal echocardiography may be required, such as post-cardiac surgery to identify posterior effusion on transgastric short axis views [2]. It is also important to distinguish pericardial effusion from pleural effusion (for example the latter lies posterior while the formal lies anterior to the descending aorta on parasternal long axis images) as well as epicardial fat pad. Depending on the cause or chronicity of the pericardial effusion, the fluid may contain stranding, loculations or even hematoma, which may affect the appropriate therapy and approach to pericardiocentesis [24].

It is critical for cardiology clinicians to be familiar with echocardiographic features that may be seen in pericardial tamponade, despite it ultimately being a clinical diagnosis incorporating signs of hemodynamic compromise [2]. The main echocardiographic criteria according to guidelines (Fig. 2) include dilated inferior vena cava with minimal (50%) collapse; diastolic collapse with the right ventricle (this sign may be absent in severe pulmonary hypertension with high right ventricular pressure; left heart collapse increases specificity); right atrial collapse ($>$34% of the duration of cardiac cycle increases specificity); and large pericardial effusion with swinging heart [1, 2, 25, 26, 27]. Other supporting echocardiographic findings for tamponade include significant respirophasic variation of inflows (mitral E wave $>$30%, tricuspid E wave $>$60%); diastolic filling blunting and increased expiratory flow reversals of hepatic vein, E/A ratio often 1.0, and large or rapidly enlarged effusion. Tamponade may also be present for loculated effusions or less than large effusion sizes, especially with rapid effusion accumulation [2].

Fig. 2.

Echocardiography evaluation of pericardial tamponade. (A) Dilated inferior vena cava (2.7 cm) with minimal 50% collapse (arrow). (B) Large circumferential pericardial effusion ($>$3 cm in diastole) on parasternal long axis view (white arrows), with left pleural effusion behind it and the descending aorta (yellow arrow). (C) Right atrial systolic collapse on apical 4-chamber view (arrow). (D) Right and left ventricular diastolic collapse on apical 4-chamber view (white and yellow arrows respectively).

Effusive-constrictive pericarditis is an increasingly recognized entity, where both pathological pericardial features (including tamponade) are concurrently present [2]. There is usually a significant pericardial effusion to start with, and the fluid can be organized, echogenic or loculated by echocardiography [2, 28]. The hallmark feature is that after pericardiocentesis is performed to remove the fluid, there is persistently elevated right atrial pressure, right and left ventricular end-diastolic pressures (with dip and plateau waveform) and prior echocardiography characteristics of constriction such as respirophasic septal shift emerge [2, 29].

CT is not commonly used to assess pericardial effusion, usually when incompletely seen on echocardiography or for pre- or post-procedural evaluation [2]. More commonly, the CT chest is performed for another reason, and the pericardial effusion is incidentally found [30]. The main advantages of CT would be to in some cases better visualized the full extent of the pericardial effusion, to characterize the presence of loculations, distinguishing effusion from fat, cyst, hematomas, and masses [2, 31]. By Hounsfield units, fat is usually below 0 (negative), transudative fluid 0–10, exudative fluid 20–60 and blood $>$60, although there can be some overlap [32].

Cardiac MRI is also rarely used to evaluate pericardial effusion and should not be used in unstable patients with suspected tamponade, however both might be incidentally identified when MRI is performed for other indications [33]. As part of pericardial inflammation or constriction assessment, the presence and extent of effusion is also examined. Just like CT, MRI can separate pericardial effusion from cysts, fat and other masses [2, 31]. Pericardial effusion typically has elevated T2-signal but low signal on delayed enhancement sequences, although exudates and blood with higher protein and cell contents increase T1 and reduces T2 signal relative to transudates [34]. MRI findings are tamponade physiology features can are similar to echocardiography such as right ventricular diastolic and/or right atrial systolic collapse, plethoric inferior vena cava and large effusion with swinging heart, generally seen in cine bright-blood sequences imaging [33].

Pericardial constriction is a feared chronic complication of pericardial disease, occurring in approximately 1–2% of pericarditis patients, sharing the same etiology but often challenging to diagnose [1, 35]. Echocardiography is also the first-line and often self-sufficient modality for evaluation of constrictive physiology [2, 36]. The Mayo Clinic Criteria being the recommended algorithm for diagnosing this entity that is frequently challenging to distinguish from restriction or diastolic dysfunction, with several key findings shown in (Fig. 3) [36, 37, 38]. The first part of the algorithm has the dual criteria of mitral inflow E/A ratio $>$0.8 and dilated inferior vena cava, with the absence of either meaning that constriction or restriction is unlikely. Secondly, respirophasic ventricular septal motion abnormality shift is expected to be present, and cardiac catheterization or further imaging may be considered if this was absent but there are ongoing concerns for constriction. The next key parameter is the mitral medial e’ velocity by tissue Doppler, where $>$8 cm/s supports constriction diagnosis, 6 cm/s supports restrictive cardiomyopathy, and 6–8 cm/s suggests mixed constriction and restriction heart disease. Furthermore, the presence of annulus reversus (mitral annular lateral e’ being less than medial e’) and/or hepatic vein end-diastolic and expiratory reversal divided by forward flow velocity $\ge$0.8 are both strongly supportive of constriction.

Fig. 3.

Multi-modality imaging evaluation of constrictive pericarditis. Echocardiography (A) dilated inferior vena cava (2.3 cm) with minimal 50% collapse (arrow); (B) respirophasic interventricular septal shift (arrow, insp, inspiration; exp, expiration) on parasternal short axis view; (C) conical deformity of the ventricles (arrow) on apical 4-chamber view, (D) annulus reversus with higher mitral annular medial than lateral e’ (18 versus 15 cm/s, arrows); and (E) increased expiratory end-diastolic reverse velocity/forward velocity at 0.9 of hepatic vein (arrow). Computed tomography (F) extensive pericardial calcifications (arrow). Magnetic resonance imaging (G) free-breathing cine sequence respirophasic interventricular septal shift on short axis view (arrow).

Of note, in the Mayo Clinic Criteria derivation study, the presence of respirophasic ventricular septal shift and either medial e’ velocity $\ge$9 cm/s or hepatic vein ratio in expiration $\ge$0.79 provided the highest diagnostic accuracy for constrictive physiology, with sensitivity 85%, specificity 91%, positive predictive value 97% and negative predictive value 65% [37]. The criteria was externally validated in a cohort study by Cleveland Clinic, with medial e’ velocity being the most important parameter, along with respirophasic septal shift and hepatic vein reversal ratio [39]. Other supportive echocardiography findings for constriction include abnormal pericardial thickening and calcification, tethering of the left or right ventricular free wall or apex to the pericardium or conical deformity, significant respirophasic variation in mitral or tricuspid inflows at least 25 and 40% respectively or that of pulmonary vein diastolic flow, and lower ratio of lateral to septal wall strain [2, 40, 41]. On the other hand, findings supporting restrictive cardiomyopathy include those associated with severe diastolic dysfunction, such as deceleration time 150 ms, isovolumetric relaxation time 50 ms, E/e’ ratio $>$15 and left atrial volume indexed $>$48 mL/m${}^2$ [36].

Transient constriction is increasingly recognized in a subset of ~17% of patients with constrictive physiology, where rather than being chronic it typically resolves within 3–6 months [35, 42]. The underlying pathophysiology is mainly persistent pericardial inflammation rather than fibrosis and calcification seen [43]. Identification of this condition allows timely initiation or continuation of ant-inflammatories to potentially prevent progression to constrictive pericarditis. Not only will there be echocardiography and MRI signs of constrictive physiology, but concomitant pericardial inflammation on MRI (as discussed before, with 86% sensitivity and 80% specificity for this entity) would be observed, whereas pericardial calcification is less often seen on CT [44]. In this sense, inflammation actually portends higher chance of transient constriction and better prognosis.

Cardiac CT may assist with diagnosis of pericardial constriction if inconclusive by echocardiography alone [2]. Supportive findings include abnormal thickening 4 mm or more, pericardial calcifications (where CT is the best modality to assess this, although only 25–50% of constriction patients with calcifications), dilated inferior vena cava and atria, conical deformity of the ventricle(s), and extracardiac findings such as pleural effusion, ascites and hepatosplenomegaly [2, 45, 46, 47]. Pericardial calcification often has an irregular distribution, such as preferentially affecting the basal anterolateral left ventricle, right ventricular outflow tract, and adjacent to the mitral and tricuspid annulus [48]. If cine images of the cardiac chambers over one cardiac cycle are performed using retrospective ECG-gating, then respirophasic septal shift and wall tethering may be observed [2]. Perhaps just as valuable is CT’s role in the evaluation of the thoracic anatomy or extracardiac pathologies as part of pre-operative evaluation for pericardiectomy or other cardiothoracic surgeries, such as location of pericardial and aortic calcifications, and cardiovascular structures relative to the sternum which are especially important in redo cardiac surgery [49].

MRI is actually a valuable second-line but under-utilized imaging tool for evaluating constrictive pericarditis [2]. Standard cine imaging with steady state free precession or gradient echo sequences not only assess chamber size and function, but also typical constriction findings such as abnormal interventricular septal motion, wall tethering, conical ventricular deformities, and dilated inferior vena cava, while free breathing sequences allows assessment for respirophasic septal shift (Fig. 3) [2, 50]. Pericardial thickness often increased in constrictive pericarditis can be assessed by these bright-blood sequences or black-blood spin echo sequences, as well as dilated inferior vena cava. Quantitative measures include lower short-axis cardiac area at end-inspiration/end-expiration, and higher relative atrial volume index ratio (left versus right) to be present in constrictive pericarditis [51, 52]. Acquiring phase-contrast sequences real-time over 10 seconds with free breathing an detect mitral and tricuspid inflow, with $>$25% and $>$45% respiratory variation respectively suggesting constriction [53]. Concurrent MRI findings of pericardial effusion and/or inflammation described in earlier sections must be evaluated [15]. The presence of pericardial inflammation may suggest the disease being earlier in the disease course and higher chance of transient constriction, and is associated with response and improvement to anti-inflammatory therapy, while its absence may push patient towards needing diuresis and potentially pericardiectomy surgery [10, 13]. How well MRI identifies pericardial constriction was evaluated in a study at Cleveland Clinic, which found the combination of relative interventricular septal excursion and pericardial thickness on MRI having the highest discriminative value (c-statistic 0.98, sensitivity 100% and specificity 90%) [54]. Another controlled study found pericardial thickness $\ge$2 mm, right ventricular end-diastolic volume $\le$133 mL and septal flattening by MRI to best discriminate constrictive physiology, all with c-statistics exceeding 0.90 [55].

Although echocardiography can identify the presence of pericardial masses, the information it provides can be limited in terms of tissue characterization as well as the extent and spread of the mass. However, some typical tumor features are as follows: haemangioma are usually hyperechoic with septae; lymphangiomas appear heterogenous and hypoechoic with septae, lipomas are usually circumscribed and echogenic; teratomas are usually heterogeneous with both cystic and calcific echogenic areas; lymphoma are usually hypoechoic with effusion; while mesothelioma is usually associated with pericardial thickening and effusion [31, 56]. For pericardial cysts, a hypoechoic space next to the heart chamber border most commonly the right atrium is seen on echocardiography, with lack of blood flow within demonstrated by Doppler or intravenous contrast [31, 57]. Pericardial cyst and diverticula differ where the former has a constant size with altering shape while the other changes in both size and shape with body posture and breathing. Echocardiography may provide guidance in aspiration and drainage of pericardial cysts as well [58].

CT has greater ability for tissue characterization than echocardiography, and is able to assess the extent, local or distant spread of masses, lymph node involvement and many pericardial tumors have associated effusions that are hemorrhagic or exudative [31]. Hemangiomas appear heterogenous with contrast enhancement; lymphangiomas are heterogenous with low attenuation and septae; lipoma have low fat-level attenuation that is circumscribed, and sometimes can surround coronary arteries; teratomas usually have contain areas of calcification and fat; lymphoma are hypoattenuating with contrast enhancement; fibromas are homogeneous with no or minimal enhancement given lack of vascularity; sarcomas are broad-based masses which invade adjacent structures; and mesothelioma is seen as diffuse irregular pericardial thickening with effusion [31, 56]. Pericardial cysts are seen as a well-circumscribed homogeneous mass with thin wall on CT, with fluid density, unaffected by intravenous contrast (Fig. 4) [59, 60].

Fig. 4.

Multi-modality imaging tissue characterization of pericardial cyst (arrows in all panels) adjacent to the right atrium. (A) Computed tomography axial slice, cyst was 10 Hounsfield units. (B) Magnetic resonance imaging (MRI) steady-state free precession bright blood sequence axial slice, cyst has increased signal. (C) MRI T2-short tau inversion recovery sequence, cyst has high signal. (D) MRI late gadolinium enhancement sequence axial slice, cyst has low signal.

MRI’s main advantage amongst imaging modalities is its ability in tissue characterization, and this is no different when applied to pericardial masses. Depending on tumor extension, the pericardium or myocardium may show thickening, or pericardial effusions, the latter often exudative or hemorrhagic with high signal intensity on T1-weighted sequences [31]. On T1-weighted, T2-weighted and gadolinium enhanced sequences, many tumors have low, high and high signal intensities [14, 31, 61, 62, 63]. Hemangiomas generally appear heterogeneous on all sequences, while lipomas have high signal intensity on all sequences, however its signal can be uniquely suppressed on fat-saturation pulse sequences. Fibroma have low vascularity and therefore have low signal intensity on T2-weighted sequence and none to minimal enhancement on gadolinium enhanced sequences. Mesotheliomas appear homogeneous on T1-weighted but have heterogenenous elevated signal on T2-weighte and gadolinium enhanced sequences. Of note, some studies have suggested heterogenous gadolinium uptake to indicate areas of increased lesion nodularity, growth and/or necrosis [64]. Pericardial cysts also appear as a well-circumscribed homogeneous mass with thin wall on MRI, displaying hypointense signal on T1-weighted sequence unless there is an exudative or hemorrhagic component, with hyperintense signal on T2-weighted sequence and no signal on LGE sequence (Fig. 4) [31, 65]. Lastly, pericardial hematomas show hyperintense, heterogeneous and hypointense signal on T1 and T2 weighted sequences in the acute, subacute and chronic stages, and no signal on LGE sequences regardless of timeframe [31].