Valve thrombosis can occur in mechanical prosthetic valves and is increasingly recognised in transcatheter and surgically implanted bioprosthetic valves. The risk of thrombosis of mechanical valves is higher in the mitral position compared with aortic position and in older generation valves (ball and cage valves). There is a wide spectrum of presentation from the asymptomatic patient to those with embolic complications or cardiogenic shock. A combination of transthoracic and transesophageal echocardiography is required to assess the haemodynamic effect of thrombosis (valve gradients and area), leaflet motion and thrombus size. CT or cinefluoroscopy may be useful in selected cases to assess leaflet motion or help identify the aetiology of valve obstruction where echocardiography is inconclusive. Exclusion of pannus or vegetation is important. Management of non-obstructive thrombus is primarily optimisation of anti-coagulation. Treatment of obstructive thrombus requires a decision between slow, low-dose fibrinolysis or valve surgery. Factors which need to be considered include thrombus size, New York Heart Association Class, presence of concomitant coronary artery disease or other valve dysfunction, surgical risk and contraindication to fibrinolysis. This review examines the incidence, aetiology, clinical features, imaging algorithms and management of prosthetic valve thrombosis.
Statistics from Altmetric.com
Incidence and aetiology
The overall incidence rate of mechanical valve thrombosis is in the region of 0.4 per 100 patient-years. Mitral valve thrombosis (0.5 per 100 patient-years) is roughly five times that of aortic valve thrombosis (0.1 per 100 patient-years).1 There are limited data on the outcome of tricuspid prosthetic valves. However, prosthetic valve thrombosis is more frequent and occurs in up to 2 per 100 patient-years.2 3 The incidence of bioprosthetic valve thrombosis has been poorly studied. Meta-analysis of aortic bioprosthesis estimates the incidence at 0.03 per 100 patient-years.4
The formation of thrombosis can be explained by Virchow’s triad: stasis, endothelial injury and hypercoagulability.5 Prosthetic valve material is thrombogenic and activates the clotting system via factor XII as well as platelets. Platelet deposition has been shown to occur on valve dacron sewing rings within the first 24 hours of implantation.6 Non-physiological blood flow, characterised by areas of high shear stress or low velocity/stagnant blood, promotes platelet activation and thrombosis. Each individual mechanical prosthetic valve design has a characteristic blood flow pattern.7 Ball and cage valves have a characteristic wake downstream of the cage apex. Adjacent to this is an area of high velocity forward flow and shear stress leading to activation of platelets, which promotes thrombus formation at the apex of the cage. Single tilting disks have an area of stagnant blood flow adjacent to the aorta downstream of the minor orifice while bileaflet valves have symmetric, less turbulent central blood flow. Therefore, ball and cage valves have a higher risk of thrombosis compared with newer tilting disk or bileaflet valves.7 8 Individual patients will be susceptible to thrombus if they have a hypercoagulable state (pregnancy, lupus anticoagulant or anticardiolipin positive).9 In addition, lack of adherence with anticoagulation and subsequent subtherapeutic anticoagulation will increase risk of thrombosis.
Bioprosthetic valves manufactured from porcine origin or bovine pericardium are either stented or stentless and are thought to be less thrombogenic than mechanical valves. This is thought to be, in part, due to improved haemodynamics as there is less obstruction to blood flow, particularly in stentless valves.7 8 The greatest risk of thromboembolism is thought to be in the first 3 months postimplantation due to the lack of endothelialisation of the prosthetic material.6 Homografts are similar in haemodynamic profile to stentless valves. In transcatheter aortic valves, incomplete endothelialisation, prosthesis malpositioning and flow turbulence are likely to be factors.5 8
There is a wide spectrum of presentation. Obstructive valve thrombosis is associated with haemodynamic compromise (increased valve gradients and reduced valve area) and clinical symptoms consistent with reduced cardiac output and heart failure. Often, the presentation is insidious with progressive symptoms with the majority of patients noting dyspnoea on exertion.10 Patients with delayed diagnosis may present in fulminant cardiogenic shock. Non-obstructive valve thrombosis may be found in asymptomatic patient (identified on routine imaging) or identified as part of the investigation of the aetiology of thromboembolism. It is important to differentiate thrombus from alternative explanations for their clinical presentation (endocarditis or pannus formation). For patients with thrombus, a history of poor adherence/interruption or subtherapeutic anticoagulation therapy is common.11
The main finding on clinical examination is absence of prosthetic opening or closing sound (for mechanical valves), new heart murmur or signs compatible with pulmonary oedema and heart failure.
The purpose of imaging is to identify the aetiology, severity and haemodynamic effects of valve obstruction. Typical changes include increased transvalvular gradients, reduced valve leaflet mobility, abnormal prosthetic regurgitation and the presence of thrombus attached to the valve. Transthoracic echocardiography (TTE) and transesophageal echocardiography (TEE) are key diagnostic investigations. Where necessary, additional cinefluoroscopy or CT may provide additional information. The exact algorithm of modalities used will depend on the information gained at each stage. A proposed algorithm is displayed in figure 1.
Mechanical prosthetic valves
Echocardiography—transthoracic and transesophageal
A variety of colour Doppler signs may indicate valve thrombosis. Obstruction may cause flow acceleration, thereby aliasing of colour Doppler signal on the outflow surface of the valve. Restriction/fixation of a valve leaflet can cause regurgitation together with a lack of the usual washing jets. If an orifice is completely occluded, there may limited flow visualisation by colour Doppler in the valve orifice.12
Prosthetic valve gradients
Elevated gradients usually mean valve obstruction. However, interpretation of pressure gradients needs to take into account the normal range for the size and type of valve implanted and be compared with gradients measured after valve implantation. Elevated gradients may also occur in high flow states or patient prosthetic mismatch. Obstruction to valves results in delayed time to peak velocity (acceleration time). Ben Zekry et al 13 found that an acceleration time of ≥100 ms had a 86% sensitivity and specificity for detecting prosthetic aortic valve obstruction and thereby helps differentiate between normally functioning aortic valves or patients with patient prosthetic mismatch and stenotic/obstructed aortic valves. Calculation of Doppler velocity index, effective orifice area and valve resistance may also be helpful. Saad et al 14 compared a group of normal St Jude prosthetic valves with the ones which were regurgitant or stenotic. All stenotic valves had a Doppler velocity index ≤0.27 and an effective valve area ≤0.75 cm2 and resistance of ≥280 dynes/s/cm. For mitral valves, Fernandes et al 15 identified a peak E velocity ≥1.9 m/s, VTImv/VTIlvo ≥2.2 and pressure half time of ≥130 ms identified 95% of obstructed mitral valves.
Thrombus versus pannus
Acoustic shadowing from mechanical prostheses may mean it is difficult to visualise all parts of the valve and, therefore, small thrombi may be missed. TEE can visualise the atrial side of mitral prostheses better than TTE and usually this allows detection of the cause of obstruction.16 The size of thrombus on TEE is an important predictor of outcome in patients receiving thrombolysis. A small thrombus area (<0.8 cm2) is associated with low risk of haemorrhagic or embolic complications post thrombolysis.17 Differentiation of thrombus from pannus causing obstruction to the valve is important and may be difficult. Features suggestive of thrombus include a large soft tissue mass at the site of obstruction with leaflet restriction, while for pannus leaflet motion can be normal. Barbetseas et al 18 showed ultrasound video intensity is useful. This can be quantified as the video intensity ratio of the mass to the mechanical prosthesis. A ratio of <0.7 had a positive predictive value of 87% for thrombus and negative predictive value of 89%.
Cinefluoroscopy allows assessment of leaflet motion and measurement of mechanical leaflet opening and closure angles.19 This is particularly helpful where echocardiographic visualisation of valve motion is suboptimal. The advantage of this technique is that it is rapid and readily accessible; however, it may miss partially obstructed leaflets. Furthermore, it is not suitable for investigation of valve thrombosis in pregnancy due to its ionising radiation.
Improvement in CT now allows evaluation of prosthetic heart valve leaflet motion.19 20 In patients where echocardiography has not been able to identify the aetiology of prosthetic valve obstruction, CT can be useful to identify the substrate for obstruction. Symersky et al 21 found that CT provided additional information to identify the likely cause of valve obstruction in 8 out of 13 patients (thrombus or pannus) where echocardiography was uncertain. Figure 2 demonstrates the additive value of CT in prosthetic valve thrombosis. Furthermore, recent data have shown that the attenuation value (measured by Hounsfield unit) is able to help differentiate between pannus and thrombus.22 A high value (Hounsfield unit ≥145) is likely to represent pannus, with values below this representing thrombus. Despite improvement in CT, occasional artefacts from mechanical prostheses may mean some patients have suboptimal image quality for analysis.
Echocardiography remains the principle modality to diagnose obstruction.23–25 TTE is useful to identify haemodynamic features suggestive of valve thrombosis (elevated transvalvar gradients compared with baseline postoperative gradients). However, morphological features such as reduced leaflet motion and the presence of thrombus may not always be identified on transthoracic studies and, therefore, a low threshold for performing TEE should be adopted in patients with elevated transvalvar gradients and clinical symptoms.24
Differentiation of obstruction due to valve thrombosis from structural deterioration may be difficult. Recent data have shown that independent imaging predictors of bioprosthetic valve thrombosis include >50% increase in mean Doppler gradient compared with baseline, increased cusp thickness and abnormal cusp mobility.25 This combination had a 72% sensitivity and 90% specificity for bioprosthetic valve thrombosis. Subtherapeutic anticoagulation and paroxysmal atrial fibrillation were also predictors. In patients with structural valve deterioration, calcification of the leaflets and significant regurgitation were more prominent features. Figure 3 shows the typical features of bioprosthetic valve thrombosis.
Heras et al 26 identified a high rate of thromboembolism for the first 90 days after both aortic and mitral bioprosthetic valve replacement. In this study, patients with bioprosthetic mitral valves who were anticoagulated had a lower rate of thromboembolism (2.5% per year) compared with those who did not receive anticoagulation (3.9% per year), p=0.05. A lack of anticoagulation, history of thromboembolism, mitral valve replacement and age were predictors of thromboembolism. Furthermore, a recent analysis of Danish registry data in 4075 patients undergoing bioprosthetic aortic valve replacement confirmed the benefit of anticoagulation in reducing stroke for the first 3 months. This study also demonstrated that the use of anticoagulation reduced the risk of cardiovascular death for up to 180 days post replacement.27 The American College of Cardiology (ACC) guidelines suggest it is reasonable to anticoagulate between 3 and 6 months post bioprosthetic aortic or mitral valve replacement (grade IIa, level of evidence B-NR).28 The European Society of Cardiology (ESC) guidelines state that anticoagulation for 3 months should be considered for mitral bioprosthetic valve and may be considered for aortic bioprosthetic valves (grade IIa, level of evidence C).29 However, the ESC guidelines were published prior to the Danish registry being published.
There are limited data informing management of bioprosthetic thrombosis; however, there is emerging evidence supporting the use of a trial of anticoagulation.23 24 30 Egbe et al 30 performed a prospective trial of anticoagulation with warfarin (target international normalised ratio 2 to 3) in patients with suspected bioprosthetic valve thrombosis. This strategy reduced valve gradients in 83% of the cohort (response defined as 50% reduction in valve gradient) within a median time of 11 weeks. There were no thromboembolic events. The non-responders underwent valve replacement. The options for patients who are non-responders to anticoagulation or patients who are haemodynamically unstable at presentation are valve surgery or fibrinolysis. There are sparse data on the optimal strategy as most of the literature is limited to case reports/series. However, the decision-making process to decide between surgery and fibrinolysis is similar to those for patients with mechanical valve thrombosis.
Non-obstructive thrombus can be found early after mechanical mitral valve replacement in up to 10% of patients. Laplace et al 31 showed that optimising anticoagulation in this group was associated with a good outcome and few complications in patients with small thrombi (<5 mm). In patients with larger thrombi (>5 mm), the risk of embolic complication increased. The ESC guidelines29 recommend optimisation of anticoagulation coupled with interval repeat imaging to check for resolution of thrombus. Options for optimisation of anticoagulation include administering subcutaneous low molecular weight or intravenous heparin for those with a subtherapeutic INR until correct levels are achieved or increasing the target INR. However, there will be an increased risk of bleeding. For large thrombus (>10 mm) with evidence of thromboembolism, surgery is recommended.
The management plan should be individualised for each patient. Consideration of the imaging findings, feasibility of fibrinolysis and evaluation of surgical risk needs to be considered. The ESC valve guidelines recommend (class 1) surgery as treatment for obstructive valve thrombosis in critically ill patients without serious comorbidity. Thrombolysis should be considered when surgery is not available or patients are unlikely to survive surgery.29 Recently, updated ACC guidelines recommend either low-dose, slow infusion fibrinolysis or surgery, dependent on multiple factors including clinical experience and surgical expertise28 (table 1).
Studies examining the effect of fibrinolysis therapy on outcomes in patients with mechanical prosthetic thrombosis are summarised in table 2.17 32–38 Reports of fibrinolysis have shown a variable success rate (58%–90%). In addition, the rate of thromboembolism and haemorrhage in some series have been high. The PRO-TEE registry demonstrated that the risk of complications in fibrinolytic therapy was related to thrombus area and history of prior stroke.17 Each 1 cm2 increment in thrombus size increased the risk of complications by 2.4-fold. A thrombus area (on TEE) <0.8 cm2 identified a low-risk group with a complication rate of 6% versus 29% for thrombus area ≥0.8 cm2–1.59 cm2, and 47% for thrombus area ≥1.6 cm2. Subsequent studies have used a low-dose, slow infusion fibrinolysis protocol guided by repeat imaging to assess response and the need for further/repeat fibrinolysis. Ozkan et al 32 showed initial treatment with a 25 mg dose of tissue-type plasminogen activator over 25 hours (with repeat dosing if resolution of thrombus was incomplete) achieved a 90% success rate with low complications (<2% mortality, bleeding and major embolus). Furthermore, this strategy has been successful in treating valve thrombosis in pregnancy with successful resolution of thrombus in all cases of a recent published series.39 Imaging is critical to this approach as exclusion of pannus or endocarditis as a cause of obstruction is essential. Where these pathologies have been excluded, fibrinolysis avoids the need for cardiac surgery with a low complication rate in patients with small thrombus burden. Further investigation is required to identify the risk of treating larger thrombus burden with a careful TEE-guided low-dose, slow fibrinolysis approach.
In patients with right-sided valve mechanical thrombosis, there are limited data on efficacy of fibrinolysis. Once recent study of 16 patients who underwent fibrinolysis of mechanical tricuspid or pulmonary valves demonstrated high success rates (100% for pulmonary valves and 75% for tricuspid valves) with no haemorrhagic or embolic complications.40 The ACC guidelines state fibrinolysis is reasonable in this patient group.28
Table 3 summarises studies which identify the outcomes of cardiac surgery and studies which compare surgery with fibrinolysis in patients with mechanical prosthetic thrombosis.11 41–49 Mortality from cardiac surgery ranged from 5% to 36%. Direct comparison of surgery to fibrinolysis is difficult as the allocation to either group was not randomised but based on physician and patient decisions. One of the advantages of surgery is that it allows replacement of older generation valves which may be more thrombogenic than newer valves. Second, in cases where the aetiology of obstruction is uncertain or not correctly identified, it allows definitive diagnosis and treatment. In the PRO-TEE registry of the nine patients who underwent surgery after failure of fibrinolysis, four patients were found to have pannus with thrombus and one patient had entrapment of chordae within the valve.17 Patients with recurrent thrombosis or large thrombus burden are also likely to benefit from surgery rather than fibrinolysis. Several studies have shown functional class is a predictor of the risk of surgery.11 47 Deviri et al 11 showed perioperative mortality was substantially higher in patients in New York heart association class IV (17.5%) compared with those in classes I–III (4.7%). Therefore, functional class should be factored into the overall determination of risk.
Ultimately, the choice between cardiac surgery and fibrinolysis will depend on patient characteristics, including prosthesis type (older vs newer generation), surgical risk (age, comorbidities, haemodynamic status) and imaging (thrombus burden, certainty of pannus/endocarditis exclusion).
Valve thrombosis occurs in both mechanical and bioprosthetic (surgical and transcatheter) valves. Multimodality imaging (transthoracic and transesophageal echocardiography with CT/cinefluoroscopy in selected cases) is critical to identify aetiology and guide management strategy. Non-obstructive thrombosis is treated by optimisation/trial of anticoagulation. Treatment of obstructive thrombus is dependent on patient characteristics and imaging features to decide between fibrinolysis or cardiac surgery.
Contributors WL and SB conceived the idea for the review. WL and SB performed the literature review and drafted the manuscript. WL, GL and SB made critical revisions to the manuscript and approved the final manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.