Although the trend in operative mortality for modified Fontan procedures has been one of steady diminution from mortality rates approaching 30% in the earliest series to mortality rates less than 5% in contemporary reports, some factors contributing to morbidity and mortality seem to persist. One of the major factors contributing to morbidity and failure of the Fontan circulation is the group of events referred to as thromboembolic complications.

Thromboembolic complications in Fontan patients fall into two major categories [44]. The first involves thrombosis within the surgically created pathways between the vena cavae and the pulmonary arteries. These pathways are frequently referred to as the Fontan pathway or Fontan circuit. Thrombus formation within these pathways can cause local obstruction to flow with adverse hemodynamic consequences and may also embolize or extend into the pulmonary arteries [2, 20, 25, 34, 47, 57]. Also, in the presence of either intentional (surgically created) or incidental right-to-left intracardiac communications, thrombus in the systemic venous pathway or compartment may embolize through “fenestrations” to the systemic arterial circulation [28, 49, 51]. The second major category of thromboembolic events in Fontan patients involves thrombus originating in the pulmonary venous pathway or compartment (e.g., on the pulmonary venous side of a baffle) or in the systemic ventricle or ligated main pulmonary artery stump [37, 40, 48, 53]. In these cases, morbidity occurs principally in the form of embolism to the central nervous system [42, 62], the coronary circulation [49, 62], or, less commonly, elsewhere in the systemic circulation.

Numerous hypotheses have been invoked to explain the troubling frequency of thromboembolic events in Fontan patients. Thromboembolism after Fontan operations has been variously attributed to low flow states, stasis in the venous pathways, right-to-left shunts, blind cul-de-sacs, prosthetic materials, atrial arrhythmias, and hypercoagulable states. Numerous investigations, most retrospective, have been undertaken to characterize thromboembolic events; describe the frequency and circumstances of these occurrences; and relate the risk of thromboembolic events to patient, surgical, hemodynamic, and hematologic factors. A wide variety of practices have evolved with respect to various strategies of prophylactic anticoagulation in the hopes of minimizing the occurrence and morbidity of thromboembolism after Fontan operations. Although numerous authors tout their hypotheses, and the preferred anticoagulation strategies of institutions or individual practitioners, efforts toward evidence-based practice in this area are very limited.

In 2002, Monagle and Karl [45] published an article that included the work product of a comprehensive MEDLINE literature search of the English language literature from 1971 to 2000 using the following key words: Fontan, univentricular heart, children, thrombosis, congenital heart disease, cavopulmonary, and palliation. They identified and analyzed 51 studies and gleaned information concerning incidence, potential morbidity and mortality, risk factors, prophylactic options, and risk/benefit ratio of prophylactic anticoagulation as related to thromboembolic events after Fontan surgery. They concluded that, at the time, there was insufficient data to make clear recommendations about optimal anticoagulant prophylaxis.

Not included among the 51 reports analyzed in Monagle and Karl’s review was our cohort study [30], also published in 2002, in which low-dose aspirin prophylaxis (81 mg/day) was used consistently and exclusively as a prophylactic anticoagulation regimen in 72 consecutive patients undergoing Fontan operations during a 5-year period. The patients had been monitored prospectively, with thromboembolism as the primary outcome measure. There were no early or late deaths. Follow-up was complete with 2882 patient-months at a mean of 40 months. There were no documented thromboembolic events: All suspicious events were investigated by transesophageal echocardiography and brain imaging. There were no hemorrhagic events or aspirin-related complications. This study was unique at the time because of the consistent use of a single anticoagulation strategy in a cohort of patients followed prospectively with evaluation for thromboembolic events as a primary end point. With the goal of shedding more light on this complex subject, we updated the literature search originally performed by Monagle and Karl using identical methodology and key words. A survey of the literature from 2000 through 2003 identified 21 additional studies to supplement the original list of 51. These publications included 8 case reports, 1 prospective cohort study that included some details about thromboembolism among other reported outcomes, 1 prospective cohort study (with case controls) in which thromboembolism was the primary outcome measure, 7 retrospective cohort studies that included some details about thromboembolism among the reported outcomes, and 3 retrospective cohort studies in which thromboembolism was the primary outcome measure, in addition to our study, which was the only one to evaluate the consistent use of a single strategy of prophylactic anticoagulation. Herein are summarized the findings of this now current review of the literature.

Timing and Incidence of Thromboembolic Complications

Monagle and Karl [45] reviewed eight retrospective studies that had thromboembolic events as the primary outcome measure [16, 17, 23, 32, 36, 42, 54, 59]. The number of patients in these eight separate series ranged from 25 to 654 and totaled 1585. The percentage of patients experiencing thromboses ranged from 3 to 16%, and the percentage experiencing stroke or arterial emboli ranged from 3 to 19%. Among reports published subsequent to that review, Coon and associates [12] investigated the frequency of thrombus in patients followed at the Children’s Hospital of Philadelphia after a Fontan operation. Between 1987 and 1999, 592 patients underwent echocardiography after Fontan operations, and 52 patients (8.8%) had intracardiac thrombus. Freedom from thrombus was 92, 90, 84, and 82% at 1, 3, 8, and 10 years after the Fontan operation, respectively. There was no difference in frequency of thrombus based on type of operation (atriopulmonary connection vs lateral tunnel) or the presence of fenestrations. Thrombus was detected in the systemic venous atrium in 26 patients (48%), in the pulmonary venous atrium in 22 patients (44%), in both atria in one patient (2%), in the hypoplastic ventricular cavity in 2 patients (8%), and in the ligated pulmonary artery stump in 1 patient (2%). A cerebral vascular accident was documented at approximately the time of thrombus detection on echocardiography in 8 of the 52 patients (15%). Of these 8, 4 were in atrial fibrillation/flutter and 3 had protein-losing enteropathy. Of the 52 patients with thrombus, 24 (46%) were on low-dose aspirin, 6 (12%) were on warfarin, and 1 (2%) was on heparin (for protein-losing enteropathy) at the time of detection of thrombus. The detection of thrombus on echocardiography was within the first year after Fontan operation in 34 of the 52 cases (65%). The median time interval between Fontan operation and detection of thrombus was 2.3 months (range, 1 day to 163 months). The curve describing the frequency of thrombus occurrence over time closely resembled previously published curves for the development of arrhythmia and protein-losing enteropathy, leading the authors to suggest a lifelong risk of thrombus formation in Fontan patients and the possibility of a codependent relationship between these late complications.

In 2002, Seipelt and associates [56] reported a retrospective series of 101 Fontan operations between 1986 and 1998 with analysis of thromboembolic events as the primary outcome measure. Of 85 survivors available for evaluation, 13 patients (15.3%) experienced thromboembolic events. Type of operation had no influence on the rate of thromboembolism. Patients were further analyzed by medical regimen within three groups: no anticoagulation, aspirin therapy, and Coumadin. There were complex interactions between date of operation, type of operation (modified Fontan vs total cavopulmonary connection), and prophylactic anticoagulant medical regimen (virtually none of the earlier Fontan patients received Coumadin, whereas half of the more recent total cavopulmonary connection patients did). Thromboembolic events occurred in patients within each of the three anticoagulant regimens, but there was a lower incidence in the more recent cohort managed with Coumadin.

Chun and associates [11] reported the incidence of stroke after Fontan procedures in 139 patients as 3.6% (seven strokes in 5 patients). Events occurred between 2 weeks and 9 years postoperatively. Two strokes occurred while on aspirin and Warfarin, two while on aspirin alone, and three while on no anticoagulant medications. Of the 5 patients, 3 had unfenestrated Fontans and 2 had fenestrated Fontans. Curiously, no intracardiac thrombus was detected by transthoracic echo at the time of the strokes. Transesophageal echos were done within a few days of stroke in 2 patients and did not demonstrate intracardiac thrombus. To our knowledge, this is the only study to have identified previous pulmonary artery banding as a risk factor for stroke. The authors invoked a potential mechanism similar to that in patients with a ligated main pulmonary artery stump.

An interesting cross-sectional study evaluating the prevalence of clinically silent pulmonary emboli in adults after Fontan operations was reported in 2003 by Varma et al. [61]. All consecutive adult Fontan patients attending the clinic at the University of Toronto Congenital Cardiac Center for Adults underwent ventilation–perfusion scanning and blood testing for thrombophilic tendency. Five adult patients (17%) had an intermediate or high probability for pulmonary embolism on ventilation perfusion scan, all of which were confirmed by computed tomography (CT) pulmonary angiography. No patient had a thrombophilic tendency (deficiency of protein C, protein S, antithrombin III, or antiphospholipid syndrome), although complete hematologic surveys were not done on patients receiving warfarin. Thirty percent of the patients were taking warfarin because of atrial flutter or atrial fibrillation, either chronic or paroxysmal. None of them had pulmonary emboli. Later age at the time of Fontan operation was associated with increased risk of silent pulmonary embolism, as was the type of Fontan anatomy (lateral tunnel > right atrium–right ventricle connection > atrial pulmonary connection). Not associated with silent pulmonary embolism were atrial arrhythmias, right atrial thrombus, or previous systemic thromboembolic events.

Three studies have compared transthoracic echocardiography (TTE) with transesophageal echocardiography (TEE) in the diagnosis of thromboembolic events after Fontan surgery. Stumper et al. [60], in a cross-sectional survey of 18 patients, found three intracardiac thrombi using TEE, only one of which was detected by TTE. The three positive cases were confirmed by angiography. Fyfe et al. [27] found six thrombi in 4 patients by TEE, only one of which was detected by TTE. The cases defined as positive by TEE were confirmed by angiography, surgery, or resolution of findings after treatment. Balling et al. [7] performed a cross-sectional study of 52 patients after Fontan operations. Seventeen patients (33%) had thrombus seen on TEE, only one of which was identified on TTE. Frequency of thromboembolic events reported is increased in recent studies compared with earlier studies, and the significant incidence of clinically silent thromboembolic events is noteworthy. Improved survival rates, longer duration of follow-up, improved diagnostic studies, and increased awareness of the potential for thromboembolic events must all contribute to the apparent increase in prevalence.

Mortality Associated with Thromboembolic Complications

In general, information on the management and outcome of thromboembolic events in Fontan patients is scarce and poorly documented. Monagle and Karl [45] summarized the management approaches described in the literature and subsequent outcomes (Table 1).

Table 1 Outcomes of thromboembolic events after Fontan procedures according to antithrombotic treatmenta
Full size table

Complete resolution of thrombosis was obtained in 48% of cases and death occurred in that compendium 25% of cases. Follow-up ranged from 1 month to 5 years [6, 8, 15, 17, 20, 21, 23, 25–27, 29, 32, 35, 36, 38, 39, 41, 50, 54, 58, 60, 62].

Risk Factors for Thromboembolic Complications

The influence of patient age at operation on subsequent risk of thromboembolic events is uncertain. Whereas at least one cohort study identified older age at Fontan operation as a risk factor [61], other studies showed no correlation between age at surgery and risk of thromboembolic events. In retrospective cohort studies, type of Fontan operation performed, type of material used for the conduit, and the use of valved versus nonvalved conduits did not influence the incidence of venous thrombosis [32, 54]. Potentially important exceptions to these general observations are highlighted in a report by Schoof et al. [55] of thrombus development in all three of their patients who underwent extracardiac total cavopulmonary connection using bovine jugular vein (Contegra, Medtronic, Minneapolis, MN, USA) as the extracardiac connection between the inferior vena cava and the pulmonary artery. Although the incidence of thromboembolic events appears to be independent of the individual type of Fontan pathway connections, Konstantinov and associates [47] from the Mayo Clinic have described thrombosis of both intracardiac and extracardiac conduits after modified Fontan operations in patients with azygous continuation of the inferior vena cava. The inference is that conduits carrying only hepatic venous blood may have a higher incidence of thrombosis.

Among potential hemodynamic risk factors for thromboembolic phenomena, low cardiac output, polycythemia, arrhythmias, and right-to-left shunts have all been discussed more completely than they have been analyzed. Although there is a general consensus that right-to-left shunts in congenital heart disease are associated with an increased risk of cerebral vascular embolization and stroke, du Plessis et al. [23] and Day et al. [17] assessed the role of fenestration in causing strokes and neither group found a significantly increased incidence of stroke in patients with fenestrations.

Danielson [16] reported that 16 of 18 patients who had strokes after Fontan operations at the Mayo Clinic had low cardiac output. The number of patients considered to have low cardiac output who did not experience strokes or thromboembolic events is unknown. du Plessis et al. [23] reported that polycythemia had no relationship to the risk of stroke in Fontan patients at Children’s Hospital Boston. Rosenthal et al. [54] found arrhythmias in 71% of patients who subsequently developed thrombosis compared to 43% of those who did not. However, 70% of patients diagnosed with thromboembolic events were in sinus rhythm. Already noted was the study by Varma et al. [61], in which patients who received warfarin because of atrial arrhythmias had a zero incidence of clinically silent pulmonary embolism. du Plessis et al. found no relationship between arrhythmia and stroke after Fontan operation. Day et al. [17], in assessing potential risk factors for stroke after the Fontan procedure, reported the presence of residual right-to-left shunts (nonsurgical fenestrations) in 6 of 7 patients with neurologic symptoms and in only 3 of 21 asymptomatic patients who also underwent post-Fontan catheterization. However, the denominator of asymptomatic patients with residual shunts is unknown.

In 1997, Kaulitz et al. [36] from Hannover, Germany, analyzed sequelae of the Fontan operation in the 80 survivors among 90 patients who underwent modified Fontan procedures between 1986 and 1994. Of 5 patients (6.2%) in whom intra-atrial thrombus was detected by transthoracic echocardiography, 3 had early postoperative thrombus formation despite heparin therapy. Each was believed to have mild obstruction of the cavopulmonary connection and preoperatively had raised pulmonary arteriolar resistance. Late postoperative atrial thrombosis was diagnosed on routine echocardiogram in 2 patients; both had previously developed signs of protein-losing enteropathy.

Hematologic Factors in Fontan Patients

Cromme-Dijkhuis et al. [13] were among the first to measure coagulation factors and describe quantitative abnormalities involving both procoagulant and anticoagulant proteins in patients following Fontan operations. In their first study of 37 patients [13], they described 63 coagulation abnormalities in 24 patients. These included subnormal levels of protein C, antithrombin III, and factors II and X. In their second study [14], which evaluated an additional 66 patients, 62% were reported to have protein C deficiency. Deficiencies of protein S (6%), antithrombin III (4%), factor II (36%), factors VII and IX (43%), factor X (36%), and plasminogen (15%) were also detected. Although these findings have been questioned because the authors did not use age-appropriate reference ranges for normal values [43], they show alterations that suggest the possibility of a procoagulant state after Fontan operations.

Jahangiri et al. [33], in a cross-sectional study of 20 children who had undergone modified Fontan procedures, reported similar coagulation factor abnormalities. However, their findings also included levels of factor VII that were significantly less than the normal range. Factor VII deficiency, if moderate in degree, should predispose to bleeding, not coagulation. This, together with the fact that protein C is a natural anticoagulant synthesized in the liver as a vitamin K-dependent protein, suggests a complex interaction between procoagulant factors and anticoagulant factors in Fontan patients. Furthermore, Jahangiri and other investigators [31] ruled out localized differences in coagulation abnormalities within the heart in patients late after the Fontan operation. Thus, the available information concerning the state of the coagulation system in Fontan patients suggests a complex field of physiologic variables, some potentially predisposing to thrombotic events and some potentially predisposing to a hypocoagulable state. Until recently, no authors had demonstrated any clear relation between the presence of coagulation factor abnormalities and the clinical and hemodynamic condition of the patients. To shed further light on this question, Odegard and colleagues [46] at Harvard designed a prospective case–control study to evaluate coagulation factor abnormalities and hemodynamic variables in children undergoing the Fontan operation. Coagulation factors were assayed in 20 children (age, 6.4 ± 2.9 years) at a mean of 3.7 ± 2.3 years after the Fontan procedure and in 24 age-matched healthy control subjects. Normal reference intervals were based on the control group. Concentrations of protein C, factors II, V, VII, and X, plasminogen, and antithrombin III were significantly lower in Fontan patients compared with age-matched controls. Factor VIII was significantly elevated in 6 patients (35%), 2 of whom had protein-losing enteropathy and thromboembolic events. A higher superior vena cava pressure was predictive of an elevated factor VIII level (p = 0.003). No other significant hemodynamic variables were predictive of a procoagulant or anticoagulant abnormality.

Whereas considerable emphasis has been placed on quantitation of coagulation factor proteins in post-Fontan patients, less attention has been focused on the issue of platelet reactivity. Ravn et al. [52] investigated platelet reactivity and quantified coagulation markers in a cross-sectional survey of 24 patients (median age, 11 years) at 2 years (range, 0.5–6) after a total cavopulmonary connection (n = 14) or a bidirectional Glenn anastomosis (n = 10). The reduction in serum proteins and clotting factors was generally similar to that described by other authors. None of the patients had clinically apparent thromboembolic events. However, increased platelet reactivity was observed ex vivo both after collagen-induced platelet aggregation [median, 73% (range, 61–84) in patients vs 61% (range, 47–69) in controls; p < 0.01] and after ADP-induced platelet aggregation [median, 69% (range, 52–77) in patients vs 56% (range, 40–66) in controls; p < 0.05). Among the many investigators who have examined the influence of connection geometry on the hemodynamic efficiency of various types of Fontan pathways, Monagle and Karl [45] evaluated the role of shear stress, among other major flow parameters, and hypothesized that changes in local flow structure produced changes in maximum shear stress values that may have consequences for platelet activation and thrombus formation in the clinical situation.

Discussion

Ultimate reduction of the morbidity and mortality associated with Fontan’s operation requires a strategy to minimize thromboembolic events. Three decades after the popular acceptance of the modified Fontan procedure as primary therapy, first for tricuspid atresia and later for a wide variety of malformations, there remain more controversies than hard facts concerning the causes and nature of these complications [1, 3, 4, 5, 9, 10, 18, 19, 22, 24]. Although there is general agreement that ligation of the main pulmonary artery, leaving a blind pouch or cul-de-sac distal to the pulmonary valve, is a worrisome substrate for the occurrence of thromboembolism [37, 40, 48, 53], almost any other assertion with respect to thromboembolic events after Fontan operations is the subject of controversy. The extent to which surgical factors and patient factors contribute to the overall risk remains poorly defined. Importantly, there is little agreement regarding the efficacy of various forms of prophylactic anticoagulant therapy in reducing the morbidity and mortality from thromboembolic events after Fontan operations.

What do we know? We know that (1) thromboembolic events occur more frequently, both early and late after modified Fontan operations, than they do after any other form of cardiac reconstruction other than prosthetic valve replacement; (2) thromboembolic events contribute to failure of the Fontan circulation and may occur with increased frequency in the “failing” Fontan circulation; (3) thromboembolic events occur in patients receiving heparin, aspirin, or Coumadin, as well as combinations or none of these; and (4) the factors predisposing to thromboembolic events after Fontan operations likely represent a complex field of biologic factors with multiple interactions. As such, it is very unlikely that a single agent will represent the solution to this complex problem. In our institution, we completely avoid direct caval cannulation and the use of central venous lines. Single atrial cannulation and a very brief period of hypothermic circulatory arrest to create either a lateral atrial tunnel or extracardiac inferior vena cava-to-pulmonary artery connection avoids dissection around the cavae with the attendant risks of bleeding, caval injury or distortion, and phrenic nerve dysfunction. Obviating the need for repairs at caval cannulation sites, and avoiding the use of central venous lines, minimizes the likelihood of developing a nidus of thrombus within the venous pathway in the perioperative period. The use of only transthoracic atrial lines allows continuous monitoring of cardiac filling pressures during the early postoperative period, without the additional presence of foreign bodies in the cavopulmonary pathway. Inotropic support is routinely administered for 48–72 hours postoperatively, even in patients with ideal hemodynamics. We believe that maximizing cardiac output without significant elevation of venous pressure may contribute to the reduction of early postoperative thromboembolic events.

In our practice, we are encouraged with the results of our initial 5-year trial [30] and subsequent experience over 3 additional years with the use of aspirin (81 mg/day) beginning on the first postoperative day and continuing indefinitely during long-term follow-up. Despite routine surveillance, in addition to focused investigation of all clinically suspicious events, we have detected no thromboembolic events in our patient group. We are aware of the multiple reports of thromboembolic events occurring in Fontan patients while on aspirin, but we have so far not experienced this morbidity in our series.

We emphasize the importance of careful ongoing evaluation of post-Fontan patients for thromboembolic events. Clinically suspicious occurrences must be investigated in a timely fashion. This includes transesophageal echocardiography in circumstances in which there is an alteration from baseline hemodynamics and also transesophageal echocardiography and brain imaging (CT scan and/or magnetic resonance imaging) when there is suspicion of a cerebrovascular event. Investigations described previously have highlighted the occurrence of clinically silent thromboembolic events in patients in other series. Our routine follow-up includes clinical evaluation with transthoracic echocardiography at 6-month intervals and cardiac catheterization with angiography 1 year after the Fontan procedure. A transesophageal echocardiogram is performed in patients who demonstrate new onset of atrial arrhythmias, and both transthoracic echocardiography and cardiac catheterization with angiography are undertaken to investigate any hemodynamic deterioration. Although subclinical thrombi may have been missed in our patients, they have not progressed to clinically relevant events. In addition, although we remain satisfied with the results of this regimen, we concur with those who recommend Coumadin in the setting of poor hemodynamics and chronic venous hypertension and for patients with uncontrolled atrial tachyarrhythmias. We also concur with the use of Coumadin in adult patients who undergo Fontan revisions or conversions for failure of an initial form of Fontan connection.

The problem of reducing thromboembolic complications in Fontan patients is, of course, not as simple as finding a single optimal regimen of anticoagulant therapy. In patients with chronic effusions or protein-losing enteropathy, it is important to measure procoagulant and anticoagulant factor levels. In replacing the protein losses associated with these pathologic conditions, it is important to periodically administer fresh frozen plasma in addition to albumin in order to replete stores of protein C, protein S, and antithrombin III and to avoid a prothrombotic state.