Fontan Fenestration Closure and Event-Free Survival

Abstract

Objectives

The purpose of this study was to evaluate the association of open and closed Fontan fenestration status with event-free survival.

Methods

All patients who underwent a fenestrated Fontan procedure at our institution from January 1994 through June 2007 were reviewed. Patient information was obtained from medical records. Patients were assigned to one of two study groups, Open vs Closed, based on their most recent fenestration status. Clinically relevant morbid events were tabulated and Kaplan-Meier event analysis used to create event-free probability curves with log-rank comparisons.

Results

161 patients were classified as Open and 51 as Closed. The median (inter-quartile range) time to event was 1.1 (0.1,3.3) years after the Fontan procedure; whereas, median time to closure was 1.2 (0.7,3.3) years. Median time to event was 1.5 (0.1,4.6) years in the Closed group and 1.1 (0.1,3.3) years in the Open group. Event-free probability analysis revealed no significant difference between groups (P=0.15). Follow-up median arterial oxygen saturation was higher in the Closed group – 96.0 (94.0,97.0)%, compared to the Open group – 91.0 (86.0,95.0)% (P<0.0001).

Conclusions

Fenestration closure was associated with higher arterial oxygen saturation, but not higher event-free survival. Time to event was slightly less than time to fenestration closure, suggesting potential merit in the evaluation of earlier fenestration closure. Adoption of specific fenestration management guidelines might help improve overall outcomes and enhance the quality of future studies.

INTRODUCTION

Fenestration of the Fontan circuit has been shown to improve early outcomes including decreased duration and quantity of chest tube drainage, shorter duration of mechanical ventilation, and shorter postoperative length of stay.^1–5 Even in the current era when fenestration is applied selectively by some groups, it continues to be used in high-risk patients such as those with hypoplastic left heart syndrome.

Although a benefit in the early postoperative period, Fontan fenestration has theoretical longterm risks including cyanosis and systemic thromboembolic events due to a persistent right to left shunt combined with an increased risk of thrombus formation due to venous stasis and hypercoagulability. Alternatively, a persistent fenestration may be a benefit because lower central venous pressure may decrease the risk of exercise intolerance, protein losing enteropathy, plastic bronchitis, and bradyarrhythmias. The question of if and when to intentionally close a fenestration remains unanswered. Management protocols tend to vary, ranging from active fenestration closure at predetermined time intervals to a hands-off approach in which the natural history of fenestration status is allowed to progress. A cross-sectional study of a large cohort of Fontan patients by the Pediatric Heart Network found that 20% of patients had patent fenestrations at a median of 8.6 years after the Fontan procedure.⁶ Of those that were closed at the time of follow-up, approximately half had been closed by active intervention (catheter-based or surgical).

Therefore, we carried out an analysis of a cohort of Fontan patients to determine the association between fenestration status and morbid clinical events.

PATIENTS & METHODS

Subjects

Records of all patients who underwent a fenestrated Fontan procedure at Children’s Hospital of Wisconsin from January 1994 through June 2007 were reviewed. Patients were assigned to one of two study groups, Open vs Closed, based on their most recent fenestration status as assessed by echocardiography. The presence of a detectable shunt by echocardiography, not intervention history, determined categorization as Open vs Closed. Of a total of 218 patients, 6 were excluded based on either undifferentiated ventricular morphology (1 patient) or intra-operative conversion to non-fenestrated status at the time of the original Fontan procedure (5 patients).

The Human Research Review Board at the Children’s Hospital of Wisconsin (CHW) authorized the collection of data from existing medical records and the waiver of the Health Insurance Portability and Accountability Act (HIPAA) for this retrospective study.

Data Collection

Study participants were identified from the Herma Heart Center Cardiology and Cardiothoracic Surgery Database. Patient information was obtained from medical records, echocardiography, cardiac catheterization, and operative reports. De-identified patient information was maintained using the REDCap web-based research data capture application.

Technique for Fenestrated Fontan

Fenestrations were routinely created in all patients undergoing the Fontan procedure. Earlier in the study interval, certain patients were selected for fenestration closure prior to leaving the operating room based on hemodynamics and echocardiographic findings. For the majority of patients in this study (and routinely in the current era), fenestrations were left open. Our technique for performing the lateral tunnel or extracardiac fenestrated Fontan procedure has been previously described.⁷ For all patients, a coronary punch was used to create a fenestration with a diameter of 3–5–4.0 mm. For the lateral tunnel Fontan procedure, the fenestration was placed at the center of the Gore-Tex polytetrafluoroethylene baffle (W.L. Gore & Associates, Inc, Flagstaff, AZ). For the extracardiac Fontan procedure, the atriotomy that results from detaching the inferior vena cava from the right atrium was sewn to the Gore-Tex tube as a large circle to centrally encompass the fenestration and prevent the adjacent atrial wall from impacting the size of the fenestration. For the extracardiac Fontan procedure, a fenestration purse-string suture and snare (left open) were placed in an accessible subcutaneous pocket below the sternotomy to permit subsequent potential closure of the snare and fenestration.

Technique for Fenestration Closure

Our general approach was to evaluate patients for fenestration closure at 1–3 years after the Fontan procedure. Patients were brought to the cardiac catheterization lab to first determine suitability for closure. The decision to close the fenestration was individualized for each patient and was based on multidisciplinary assessment of test-closure-induced changes in systemic venous pressure, cardiac output, and oxygen saturation. Closure was then completed during the same anesthetic session with echocardiographic and/or angiographic confirmation of closure. For the majority of patients undergoing closure, the fenestration snare was exposed through a limited skin incision, and the snare closed and secured with hemoclips. A minority of patients, including those with residual shunts after attempted snare closure, underwent closure with an Amplatzer septal occluder (AGA Medical Corporation, Golden Valley, MN).

Outcome Variables

Morbid events following Fontan palliation were defined as death, transplant, Fontan takedown, New York Heart Association (NYHA) functional classification III or IV, pacemaker placement, protein-losing enteropathy, stroke, thrombus, plastic bronchitis, subsequent thoracic operation(s), or post-Fontan mechanical circulatory support. The time interval during which a patient was free from an event determined event-free survival. Events (but not patients) were excluded from this analysis if they occurred within one calendar day of the Fontan procedure. Such events were regarded as being more related to the overall impact of the surgical procedure and less so to the fenestration. Duration of follow-up was calculated as the date from the Fontan procedure to the last known clinical visit if there were no events, or from the date of the Fontan procedure to the date of an event occurrence.

Statistical Analysis

Descriptive characteristics of the sample were summarized by median and inter-quartile range because variables were skewed. Wilcoxon-Mann-Whitney test was used to compare continuous variables, and the Chi-square test or Fisher exact test was used to compare fenestration groups (Open and Closed) for patient characteristics. Kaplan-Meier survival analysis was used to assess event-free survival with Wilcoxon log-rank comparisons between Open and Closed groups. All analyses were done in SAS version 9.2 (SAS Institute Inc., Cary, NC)

RESULTS

Group Characteristics

Of the 212 patients in this cohort, 161 patients were categorized as Open and 51 as Closed. Included in the Open group were 10 patients that underwent fenestration closure interventions with subsequent echocardiographically determined residual shunts; and 19 patients who had fenestration closure on the same day as their last known clinical follow-up. Included in the Closed group were 18 patients with spontaneously closed fenestrations. Patient characteristics are summarized in Table 1. There were no significant differences in patient characteristics between the Open and Closed groups.

Table 1.

Patient Characteristics

Patient Characteristic		Open # (%)	Closed # (%)	P-value
Gender
Male		111 (68.9%)	29 (56.9%)	0.11
Female		50 (31.1%)	22 (43.1%)
Diagnosis Category
Hypoplastic left heart syndrome		76 (47.2%	21 (41.2%)	0.84
Double inlet left ventricle		28 (17.4%)	12 (23.5%)
Tricuspid atresia		19 (11.8%)	8 (15.7%)
Unbalanced atrioventricular canal		10 (6.2%)	4 (7.8%)
Heterotaxy		11 (6.8%)	2 (3.9%)
Pulmonary atresia with intact ventricular septum		6 (3.7%)	2 (3.9%)
Other		11 (6.8%)	2 (3.9%)
Ventricular Morphology
Left		60 (37.3%)	24 (47.1%)	0.21
Right		101 (62.7%)	27 (52.9%)
Fontan Type
Lateral tunnel		37 (23.0%)	12 (23.5%)	0.94
Extracardiac		124 (77.0%)	39 (76.5%)
Age at Fontan (yrs)
Median (IQR)		2.92 (2.31,3.72)	3.05 (2.22,3.72)	0.80
Weight at Fontan (kg)
Median (IQR)		13.0 (11.7,14.7)	12.8 (11.0,14.9)	0.49

The mean and median (interquartile range IQR) duration of follow-up post-Fontan procedure was 3.8 years and 3.5 (0.9,6.1) years for the entire cohort; 3.1 years and 2.4 (0.7,4.9) years for the Open group; and 5.9 years and 6.0 (4.4,8.0) years for the Closed group. The mean and median interval from Fontan to fenestration closure was 2.1 years and 1.2 (0.7,3.3) years.

Events and Event-free Probability

Of a total of 92 events, 72 occurred in the Open group (in 50 of 161 patients, 31%), and 20 occurred in the Closed group (in 15 of 51 patients, 29%). In the Closed group, 7 patients had events prior to fenestration closure, and 9 patients had events after fenestration closure. Events are summarized according to closure status in Table 2. The most notable differences between the Open and Closed groups occurred with the event NYHA class > 2 – 16.8% in Open vs 2.0% in Closed; and the event protein-losing enteropathy – 4.3% in Open vs 0% in Closed.

Table 2.

Event Characteristics in the Open and Closed Fenestration Groups

Event Type	Open # (%)	Closed: Pre- Fenestration closure # (%)	Closed: Post- Fenestration Closure # (%)	Cumulative # (%)
Death	5 (3.1%)	0 (0%)	3 (5.9%)	8 (3.8%)
Heart Transplant	1 (0.6%)	0 (0%)	0 (0%)	1 (0.5%)
Fontan Takedown	1 (0.6%)	0 (0%)	0 (0%)	1 (0.5%)
NYHA >2 *	27 (16.8%)	0 (0%)	1 (2.0%)	28 (13.2%)
Pacemaker	7 (4.3%)	1 (2.0%)	1 (2.0%)	9 (4.2%)
PLE	7 (4.3%)	0 (0%)	0 (0%)	7 (3.3%)
Stroke	3 (1.9%)	0 (0%)	0 (0%)	3 (1.4%)
Thrombus	9 (5.6%)	1 (2.0%)	3 (5.9%)	13 (6.1%)
Plastic Bronchitis	0 (0%)	1 (2.0%)	0 (0%)	1 (0.5%)
Sub. Thorac. Ops.	11 (6.8%)	5 (9.8%)	3 (5.9%)	19 (9.0%)
ECMO	1 (0.6%)	0 (0%)	1 (2.0%)	2 (0.9%)
Total	72 (44.7%)	8 (15.7%)	12(23.5%)	92 (43.4%)

^*Ten patients had undocumented NYHA functional classification -9 in the open group, 1 in the closed group. For consistency, row percentages were calculated using a denominator of 161 for the Open group and 51 for the Closed group.

In the event-free probability (EFP) analysis, the median (IQR) time to event was 1.1 (0.1,3.3) years for the Open group and 1.5 (0.1,4.6) years for the Closed group. Figure 1 shows the EFP curves for the Closed and Open groups. The Closed group demonstrated marginally higher EFP; however, the difference was not statistically significant (P=0.08).

Figure 1.

Event-free probability curves comparing time to event in Open vs Closed groups using date of Fontan as time zero for calculating event intervals.

Event-free probability calculated with time intervals referenced to the time of Fontan has the benefit of a longer period of follow-up. As this analysis includes events occurring prior to fenestration closure in the group ultimately undergoing fenestration closure, it does not exclusively evaluate association of closed fenestration status with events. In an effort to address this issue, an additional analysis was completed, in which the median time to fenestration closure was selected as the reference point for measuring time to event. Therefore, patients who had no follow-up subsequent to 424 days after the Fontan procedure were excluded, and events (but not patients) in both the Open and Closed groups occurring prior to 424 days after the Fontan procedure were excluded. For this analysis, there were 47 patients in the Closed group and 102 patients in the Open group. Figure 2 shows the EFP curves for the Open and Closed groups, with no significant difference between groups (P=0.15). For this comparison, median length of follow-up was 2.8 (1.6,5.1) years for the Open group (increased due to skewness in the follow-up interval distribution curve) and 5.0 (3.6,6.9) years for the Closed group.

Figure 2.

Event-free probability curves comparing time to event in Open vs. Closed groups using median time of fenestration closure as time zero for calculating event intervals (see text for explanation).

The only significant difference in outcome measures at latest follow-up was arterial oxygen saturation. In non-parametric comparison, the Closed group had a median oxygen saturation of 96 (94.0,97.0)% compared to 91.0 (86.0,95.0)% for the Open group (P<0.0001).

Discussion

Our study contributes to the growing but incomplete knowledge pertaining to the management of Fontan fenestrations. The primary finding is that fenestration closure in our cohort was not associated with improved event-free probability, despite a probable influence of selection bias favoring higher event-free probability in the Closed group. Selection bias is inferred by referring to Table 2 - there were no cases of NYHA class > 2 prior to the time of fenestration closure in the group which ultimately underwent fenestration closure; whereas, there were 27 (16.8%) cases of this event in the Open group over a comparable follow-up interval. Furthermore, protein-losing enteropathy (PLE) was present in 7 patients in the Open group and in 0 patients in the Closed group. It would be highly provocative to suggest that a patent 4mm fenestration might cause PLE; whereas, it would be consistent with clinical practice to withhold patients with PLE from referral for fenestration closure.

In general, our results are consistent with a recent report of the Pediatric Heart Network in which there were no differences between groups in terms of thromboembolic events, exercise capacity, ventricular function, peak oxygen consumption, and other measures of performance.⁶ The only measure that differed was oxygen saturation – 8 points higher in patients with closed fenestrations. Other reports have also shown no difference in maximum oxygen consumption, with patients generally reaching 60–70% predicted values; and have highlighted the difficulties in predicting what degree of benefit active fenestration management might provide.^6,8–11 To date, other than arterial oxygen saturation, the only documented improvements (in single reports) associated with fenestration closure have been reduced ventilatory abnormalities with exercise and a lower incidence of late tachyarrhythmias.^10,12 As the majority of patients in the present report, as well as those in previous reports, had not reached mid-adulthood, the longer-term impact of fenestration status on various outcome measures remains unknown.

A second finding of our study is that fenestration closure was associated with higher follow-up arterial oxygen saturation. This is consistent with expectation and what has been reported by others.^6,9,10,13 Eliminating cyanosis and potentially at least one impetus for ongoing collateral formation and ineffectual cardiac output would seem to be appropriate. However, as pointed out by others, the long-term clinical relevance of increasing the oxygen saturation from the high 80’s to the mid 90’s remains to be verified.^6,13

A potentially important finding of our study is that median time to event was slightly less than median time to fenestration closure – events occurred at a median of 1.1 years in the overall cohort; whereas, fenestration closure occurred at a median of 1.2 years. This suggests that earlier closure might be necessary to realize any potential benefit. We are aware of one previous report by Goff et al in which fenestration closure occurred at a median of 9 months after the Fontan procedure.¹³ In their study, decompensation after fenestration closure correlated with shorter interval from the Fontan procedure. However, their study compared patients before and after fenestration closure and did not include a comparison group with patent fenestrations. In our opinion, the potential benefit of fenestration closure much earlier in the post-Fontan course remains an unresolved issue.

At present, the majority of available data (although limited) provides no consistent basis for active fenestration closure in all patients.^6,8–13 Since fenestration closure has yet to demonstrate efficacy for reducing stroke/systemic embolization, the only apparent justification for closure would seem to be clinically significant desaturation, based more so on secondary effects of cyanosis (not oxygen delivery).^6,14,15

The important question is whether the quality and magnitude of the available data are sufficient to guide patient management. The ongoing variation in management strategies suggests that it is not. From our perspective, there are three options that could offer better insight on the issue of optimizing fenestration management. Conducting a randomized controlled trial, although theoretically appealing, is limited by several practical constraints. A second and perhaps more practical option would be modeling behavior and performance of centers with superior long-term outcomes of single-ventricle management. A third option would be to adopt guidelines based on available evidence that provide a set of management options from which clinicians could agree to choose and consistently follow – this might at least reduce the marked variability in management strategies and enhance the quality of future clinical studies.

Limitations

This was a retrospective study susceptible to various forms of bias. In particular, clinicians may have avoided referral for fenestration closure those patients with a more concerning clinical status. An additional limitation is the different follow-up intervals of the two groups and the occurrence of several events prior to fenestration closure in the Closed group. Of the various methods for addressing this issue, we used the median time of fenestration closure as the zero time-point for determining time-to-event. For both EFP analyses, length of follow-up was relatively short and precluded evaluation of any potential differences in longer-term outcomes. An additional issue was categorizing patients. Given the Institutional Review Board governed time constraints, certain patients had no accessible follow-up after fenestration closure. We felt the best method for handling this was to categorize such patients in the Open group. Also, certain patients underwent attempted fenestration closure but had residual shunts, and certain patients experienced spontaneous closure. It was our best judgment to categorize these patients according to their physiological status. We ran additional analyses (not presented) with other methods of categorization, and the results were similar, but we cannot be certain that all possible methods of categorization would produce similar results.

Conclusions

In this cohort of fenestrated Fontan patients, fenestration closure was associated with higher arterial oxygen saturation, but not higher event-free survival. In the overall cohort, median time to event was somewhat shorter than median time to fenestration closure, suggesting potential merit in the evaluation of earlier fenestration closure. Collaborative learning from centers with superior long-term outcomes of single-ventricle management and the adoption of specific fenestration management guidelines might help improve overall outcomes and enhance the quality of future studies.