A systematic search was conducted on Cochrane, Embase and MEDLINE. Search terms are provided in Figure 1. Moreover, to avoid losing any related publications, an Open-Gray search was conducted. Related journals and reference lists of identified articles were cross-checked for other relevant studies of interest. Retrospective and prospective observational or randomized controlled trials from the year 200, written in English, reporting the pre-determined outcomes including children or adults undergoing catheter-based intervention on a Fontan fenestration in human subjects were included. Exclusion criteria were case reports, non-original articles, systematic reviews, meta-analyses, non-published works, studies not describing any of the pre-determined outcome measures, or studies that included > 20% of spontaneous fenestration closures. The search was conducted with the assistance of two experienced librarians.

Two independent reviewers (Z.N.L. and A.F.C.) screened all identified studies. In case of multiple publications with sample overlap, the most recent report was included. In each article, the criteria for inclusion and exclusion were independently evaluated by the two reviewers to verify the correctness of selection. In the case of disagreement between reviewers, a consensus was agreed upon. In multiple studies with overlapping study populations, the study with the greatest overall follow-up was included. The first author and/or the corresponding author of three of the included studies were contacted to clarify reported data, particularly regarding the size of the fenestration at the time of surgery.

Study design, year of Fontan surgery, surgical type of Fontan, fenestration use, and clinical follow-up were documented. The baseline demographics were extracted from the individual studies. The outcomes extracted included early or late mortality, Fontan takedown, heart transplantation, stroke, thromboembolism, or peri-interventional complications. Per-interventional changes in vital signs and changes in hemodynamic parameters including cardiac index, exercise duration, minute ventilation, maximal oxygen consumption, peak exercise oxygen pulse, and ventilatory anaerobic threshold were extracted. Additional clinical outcomes of interest, including protein-losing enteropathy, plastic bronchitis, and arrhythmias were similarly recorded. Complications were deemed major if the grade was > 2 or minor if the grade was ≤ 2 on the Clavien-Dindo classification system (34).

A meta-analysis was performed to compare the outcomes before and after catheter-based Fontan fenestration closure. Outcomes were compared before and after fenestration enlargement or creation for Fontan failure. The mean differences and the corresponding 95% confidence intervals (CI) were estimated using a random effects meta-analysis model, which accounts for variability induced by between-study heterogeneity. Cochran’s Q statistic and I² index was used to quantify and test heterogeneity between studies (35). All statistical analyses were performed using R software version 3.6.3 (36) and the meta-package (37). Probabilities with P-value < 0.05 were considered statistically significant, and all statistical tests were two-sided.

There is no concrete guidance on what should be done for the fenestration over the medium or the long-term. Because of this, there is dramatic heterogeneity fenestration management amongst centers. Because of the theoretic risk of thromboembolism and detrimental effects of prolonged cyanosis, some centers routinely close all fenestrations in the catheterization lab at about 12 months postoperatively (39). Patients who had spontaneous closure of their fenestration probably did so because of low trans-pulmonary pressures and resultant minimal relative flow across their fenestrations. This patient population, who can make up to about 40% of fenestrated Fontan patients, represent a low risk for Fontan failure (40). The long-term answer surrounding the question of what to do for the other 60% of patients was not clear. A persistent fenestration may be a surrogate for physiologic intolerance of the Fontan circulation, and this difference may not be readily apparent by pre-Fontan hemodynamic parameters (41–75). This has led some centers to either shift from routine closure to not closing any fenestrations (76) or to be very selective in which patients are referred for transcatheter closure (54, 62). For instance, McCrossan and Walsh (62). Will refer patients for trans-catheter Fontan fenestration closure if there is persistent hypoxia, satisfactory ventricular function, and absence of significant Fontan circuit obstruction.

After the patient is referred to the catheterization lab, the interventional cardiologist may find that the patient is not appropriate for closure. Many centers use the parameters set forth by the group at Boston Children’s Hospital in 1995 (77). At the time of catheterization and after measuring baseline features, the fenestration is then test occluded for 10 min. If the right atrial pressure exceeded 18 mmHg, or the arteriovenous difference in oxygen saturation increased by > 33%, or the right atrial saturation was < 40%, then the patient was considered to have an unfavorable response to test occlusion ad the fenestration was left open. Similarly, Goreczny et al. will do the 10-min test occlusion and will avoid permanent fenestration closure if the systemic venous pressure is above 18 mmHg or if there is an increase of 5 mmHg after balloon occlusion (53). At the Children’s Hospital of Michigan closure is deferred after balloon occlusion if the Fontan pressure is ≥ 20 mmHg, the Fontan pressure increases by 4 mmHg or more, if there is a decrease in cardiac output by ≥ 50%, or a decrease in systemic oxygen transport of ≥ 46% (78).

These strict criteria during test occlusion may identify patients that would not tolerate fenestration closure, but it may be that this provocative measure is sensitive but not specific for those patients that may not have the proper hemodynamics but would tolerate closure anyway. Ozawa et al. studied the long-term outcomes after fenestration closure in patients at risk for Fontan failure. They compared high-risk Fontan patients, as defined by pre-Fontan mean pulmonary arterial pressure ≥ 15 mmHg or systemic atrio-ventricular valve regurgitation ≥ moderate, compared to a standard risk group and a group whose fenestration had closed spontaneously. Protein-losing enteropathy-free survival rates did not differ between groups (p = 0.72). This was at the expense, in the high-risk group, of persistent cyanosis from veno-venous collaterals and lower peak oxygen consumption (p = 0.019) and lower anaerobic threshold (p = 0.023) compared to the standard risk group (79).

Exercise capacity as described by maximal oxygen consumption (VO₂ max) has found diametrically opposite conclusions in certain reports. Mays et al. investigated 20 patients before and after Fontan fenestration closure. In their analysis, the VO₂ max increased to 1.24 ± 0.35 L/min from 1.18 ± 0.46 L/min, p < 0.005 (61). Conversely, Meadows and colleagues also investigated 20 patients before and after fenestration closure. In this series, the percent predicted VO₂ max increased to 74 ± 18.6% from 70.9 ± 18.6%, P = NS. In both studies, there was a wide variation in patient age and size (63). To take into account growth-related changes between pre- and post-fenestration closure exercise tests, different measurement endpoints were considered. Such differences may explain why exercise capacity was statistically significant in one study but not the other. Another potential explanation may be a combination of competing interests. One would surmise that increasing systemic arterial saturation would improve VO₂ max and exercise duration, but this increase in systemic arterial saturation was at the expense of cardiac index and mixed venous oxygen saturation. Furthermore, neither study was able to define the amount of right to left shunting under exercise conditions. In his series Meadows stated that, at rest in the catheterization lab, all patients had no more than trivial shunting at the end of the catheterization, but that does not exclude the possibility that more significant shunting could have occurred with effort (63). The cardiopulmonary response to exercise in the Fontan circulation is complicated by an inability to regulate heart rate or stroke volume in response to exercise, systemic venous flow dynamics, and the compound contribution of ventilation on the Fontan circuit.

As stated earlier, the Fontan fenestration is associated with well documented early postoperative benefits. Studies identifying long-term benefits are sparse. Atz et al. showed a significant increase in the rate of fenestration across multiple centers between 1987 and 2002. After adjusting for era, patient age, and year of Fontan, this large, multi-center cross-sectional study found few associations between a persistent fenestration and negative long-term outcomes (40). There was a greater number of non-fenestrations associated catheter re-interventions, most commonly for coiling of systemic venous and aorticopulmonary collaterals. Unsurprisingly, the resting oxygen saturation was lower with a fenestration in the long-term, but they found no difference in the number of long-term re-interventions, incidence of protein-losing enteropathy or arrhythmias. The Australian and New Zealand Fontan registry has been a wealth of knowledge in understanding the Fontan outcomes in the modern era (80). In a propensity score-matched analysis they demonstrated that there was no difference in long-term survival (87% vs. 90% at 20 years; p = 0.16) or freedom from failure (73% vs. 80% at 20 years; p = 0.10) between patients with and without fenestration, respectively. There was more freedom from thromboembolism in the non-fenestrated vs. the fenestrated group (89% vs. 84%; p = 0.03) (80). On the other hand, in the study from Children’s Hospital of Michigan, the incidence of a composite outcome of death, transplant, deteriorated heart failure, plastic bronchitis, or protein-losing enteropathy was significantly higher when there was an open fenestration (60% vs. 6%; p < 0.01) (78). As mentioned earlier the patients with the open fenestration in this study were a select group of patients who failed to meet proper hemodynamics with test occlusion which may be the difference in the findings between these studies (78).

The present review and meta-analysis have select limitations. First, a large number of published articles reported incomplete data. For instance, no suitable data were reported to analyze the impact of complications after closure of the fenestration, such as stroke and/or systemic thrombo-embolism, as well as the occurrence of liver failure due to the increased systemic venous pressure. We therefore limited our presentation of the results to the most relevant information in regard to the long-term outcome of fenestration. Second, data extracted from observational study designs should be interpreted with caution due to the inherent limitations of confounding and a high degree of selection bias. Most importantly, select studies on the management of Fontan fenestration in patients with failed Fontan circulation had inadequate statistical analyses, which limited the ability to rigorously assess this fundamental question in the present analysis. Third, the first two limits are the consequence of the considerable controversy and widely varying institutional practices concerning Fontan fenestrations in respect to indication for establishment as well as management during long-term follow-up.