Introduction

The Fontan operation is considered definitive palliation in patients with single-ventricle physiology. Since its first description in 1971 [1], several modifications have been introduced, including the lateral tunnel or fenestrated Fontan operation [2, 3]. Lacking a subpulmonic ventricle, the Fontan circulation has to maintain high systemic venous pressure and low pulmonary artery pressure and the preload is limited [4, 5]. The limitation of preload is more prominent during exercise [6]. In extracardiac conduit type Fontan circulation, conduit size can be an important factor in determining preload [6, 7].

In an extracardiac Fontan operation, a larger conduit for body size is used to allow for the patient growth. However, a large conduit causes inefficient flow due to turbulence or stagnation and might cause later problems such as thrombosis or stenosis [8, 9]. Despite improvements in surgical strategies and techniques, the optimal conduit size in the extracardiac Fontan operation has not yet been determined. We aimed to determine optimal extracardiac conduit size based on exercise capacity in patients with Fontan circulation.

Subjects and Methods

We retrospectively reviewed the medical records of 677 patients over 10 years old who underwent a Fontan operation in Seoul National University and Sejong General Hospital. Of these, 46 were mortality or heart transplantation cases, and 228 did not have extracardiac conduit type Fontan circulation (atriopulmonary, lateral tunnel type, and Kawashima type). A total of 527 patients were excluded for various causes that could limit exercise capacity (Table 1). We compared hemodynamic data from cardiac catheterization and results from cardiopulmonary exercise test. The correlation between results from cardiopulmonary exercise test and conduit diameter for body surface area was analyzed.

Table 1 Characteristics of excluded patients
Full size table

Patients underwent progressive exercise testing with expiratory gas analysis. Cardiopulmonary exercise testing (Quinton Q-stress®, Cardiac Science, WI, USA/Vmax®, CareFusion, CA, USA/GE Medical systems, T2100 CASE, Mexico) was performed with a modified Bruce protocol. Expiratory gas analysis was performed with a Medical Graphics metabolic cart (TrueOne 2400®, ParvoMedics, Salt Lake City, UT, US). Oxygen consumption, carbon dioxide production (VCO2), and minute ventilation (VE) were measured on a breath by breath basis and analyzed in 15-s intervals. Peak VO2 was defined as the highest VO2 achieved by the subject during the test. Ventilatory anaerobic threshold (VAT) was measured using the V-slope method when it could be accurately determined [10]. Values for VO2 were indexed to body surface and expressed as percentage of predicted values for healthy age- and sex-matched subjects as reported by Cooper and Weiler-Ravell [11]. The ventilatory equivalent of carbon dioxide (VE/VCO2) was measured at the ventilatory anaerobic threshold (VAT) determined with the V-slope method [10]. The respiratory exchange ratio (RER) (VCO2/VO2) was measured continuously. The pulse O2 (VO2/heart rate [HR]) was measured at peak VO2 and indexed to body surface area. The pulse O2 was equal to the product of stroke volume and the arterial-venous O2 content difference. Because the arterial-venous O2 content difference at peak exercise varies little among subjects, the pulse O2 was used as a surrogate for stroke volume at peak exercise [12].

Resting 12-lead electrocardiograms were performed in sitting and standing position and during brief hyperventilation. Heart rate was monitored continuously. The chronotropic index, a measure of response that is independent of resting HR and stroke volume, was defined as: chronotropic index = (maximal achieved HR − resting HR)/(predicted maximal HR − resting HR) [13, 14].

Cardiac catheterization was performed with the patient conscious and the pressure and oxygen saturation were measured in a stable state. Calculation of systemic and pulmonary flow and their ratio (Qp/Qs) was made using the principles described by Fick, with assumed values for O2 consumption according to data published by LeFarge and Miettinen [15]. The superior and inferior vena cava saturations were taken as the mixed venous and systemic arterial saturation in the descending thoracic aorta. Pulmonary venous saturations were obtained by pulmonary artery wedge or left ventricle indirectly. Total pulmonary vascular resistance was expressed in indexed Woods Units (WU). An oxygen delivery was calculated as the product of the cardiac index and arterial O2 content. Because all initial hemodynamic evaluations were performed with the patient breathing room air, the contribution of dissolved O2 was excluded from these calculations. Cardiac catheterization was performed by a pediatric cardiologist with more than 5 years of experience.

Statistical analysis was performed with SPSS 22.0 (SPSS Inc., Chicago, IL, USA) for Windows and R. Descriptive data are presented as numbers with percentage, mean with standard deviation (SD) as appropriate. Generalized additive model (GAM) is a method for finding the non-linear relationship between independent variables and response variables using nonparametric function assuming that the relationship between independent variables and response variables is not linear. Generalized additive model (GAM) is applied to identify non-linear association between conduit diameter per BSA and predicted peak VO2. We also performed GAM between conduit diameter per BSA and VE/VCO2 with adjustment for gender, Fontan operation age, ventricular Morphology, isomerism, and Fenestration. Statistical test were two sided, and a p value ≤ 0.05 was considered statistically significant. All analyses were performed by using SPSS 25.0 (IBM Corp., Armonk, NY, USA) and R software (version 3.4.1; The Comprehensive R Archive Network: https://cran.r-project.org). This study was approved by the institutional review board in Seoul National University and Sejong General Hospital.

Results

The medical records of 677 patients were reviewed and 527 were excluded for various causes limiting exercise capacity (Table 1). The baseline patient characteristics are shown in Table 2. In correlation study between predictive peak VO2 and hemodynamic data from cardiac catheterization such as CVP, VEDP, PVR, and Qs, only CVP had significant linear negative correlation with peak VO2 (p value 0.026). In correlation study between VE/VCO2 and hemodynamic data from cardiac catheterization such as CVP, VEDP, PVR, and Qs, only VEDP had significant linear negative correlation with peak VO2 (p value 0.041).

Table 2 Baseline characteristics (Mean ± SD)
Full size table

For adjusting the impact of other factors on exercise capacity, we analyzed by quadratic model (Tables 3, 4). Analysis of the correlation between predicted peak VO2 and conduit diameter per body surface area (BSA) showed a significant convex curved correlation (p = 0.0387) and a maximum peak VO2 at about 12.7 mm/m2 conduit diameter per BSA. (Fig. 1a) We analyzed the correlation between VE/VCO2 and conduit diameter per BSA. These patients showed a significant concave curved correlation pattern (p = 0.0211) (Fig. 1b), with a lowest value at 12.4 mm/m2 conduit diameter per BSA. Patients had maximum exercise capacity at about 12.5 mm/m2. Even though we analyzed the significance between exercise capacity and other factors such as age at Fontan operation, ventricular morphology, isomerism, and fenestration in Fontan pathway, they did not showed the significance statistically (Tables 3, 4).

Table 3 The quadratic model between predicted peak VO2 and conduit diameter per BSA
Full size table
Table 4 The quadratic model between VE/VCO2 and conduit diameter per BSA
Full size table
Fig. 1
figure 1

Correlation analysis between predicted peak VO2 and conduit diameter per body surface area (a) showed a significant convex curved pattern and a maximum peak VO2 at about 12.7 mm/m2 conduit diameter per BSA. Correlation analysis between VE/VCO2 and conduit diameter per BSA (b) showed a significant concave curved correlation

Full size image

Discussion

Our results showed that patients with extracardiac Fontan circulation showed maximum exercise capacity at about 12.5 mm/m2 conduit diameter per BSA. More importantly, the group with smaller conduits showed better exercise capacity than those with larger conduits. Most surgeons have determined conduit size based on the size of the inferior vena cava [9]. As an extracardiac conduit using a Gore-Tex conduit does not have growth potential, there may be concern that a smaller conduit can cause hemodynamic problems as the patient grows. Particularly because of the small size of the inferior vena cava with use of a small conduit (especially a 16-mm Gore-Tex conduit), this limitation has been a concern and larger conduits have been used [16]. However, our results showed that patients with larger conduits showed a greater decrease in exercise capacity. These results suggest that flow stagnation with use of a larger conduit rather than flow disturbance with use of a smaller conduit negatively affects exercise capacity. Similar results have been reported in prior studies using computational models [8]. In this study, 16- and 18-mm conduits were found to be optimal and larger conduits had redundant space. In our study, the size of the optimal conduit was about 20 mm when considering the body surface area of the patient (BSA 1.57 m2 × 12.5 mm), and results of the previous study are almost the same considering the luminal stenosis by intimal endothelization.

The other theoretical background of this phenomenon is based on the conduit area change rate. (Fig. 2a) Based on a 20-mm conduit, although 16-mm and 24-mm conduits have the same diameter difference, the area change rate is not same. The area change rate of the 16-mm conduit is − 35.9% and area change rate of the 24-mm conduit is + 43.9%. As the size of the conduit increases, the rate of change in the conduit area is greater than the rate when it increases. In addition, luminal narrowing caused by endothelization of the Gore-Tex conduit further exacerbates the difference in the area change rate between large and small conduit (Fig. 2b). Therefore, it is thought that when the conduit size is increased based on an optimal size, the change rate of the area is larger, causing a further decrease in exercise capacity.

Fig. 2
figure 2

The area change rate based on a 20-mm conduit (a) and the area change rate assuming luminal narrowing with intimal growth (b)

Full size image

The most difficult aspect of this study was the selection of appropriate patients. In fact, this study was not easy because there were too many factors that could affect exercise capacity [7]. Therefore, the authors analyzed patients in detail and excluded all patients with factors that could affect their exercise capacity. Only 150 of the 677 patients in both institutions were included in the study. Patients with suspected bronchial, pulmonary arterial or pulmonary parenchymal abnormalities on chest CT were excluded. Because patients with a fenestration or veno-veno collateral vessels have increased VE/VCO2 [17], those with a fenestration on imaging or a lung/whole body ratio < 0.9 on a radioisotope scan were excluded from analysis of VE/VCO2.

Statistical analysis was also a problem. As shown in previous results, the relationship between the predicted peak VO2 and conduit size and the relationship between the VE/VCO2 and conduit size showed curved correlation patterns (Fig. 1a, b). A curved correlation pattern cannot exclude the influence of other factors. Therefore, to rule out such interference, we used the quadratic model and could exclude influence by gender, age at Fontan age, ventricular morphology, isomerism, and fenestration.

As a retrospective and double-center study, this research had several limitations. Moreover, there can be a selection bias because this study included a small number of patients with heterogeneous diseases. Pulmonary function testing was not performed in many patients and airway or pulmonary parenchymal diseases were excluded by chest CT angiography. We did not directly measure preload, but estimated the value based on cardiopulmonary exercise test. However, it was difficult to measure the preload in a static state with current technology, and was even more difficult to measure during exercise. The curved correlation pattern did not completely eliminate the influence of other factors; however, we overcame this problem by quadratic model.

Despite the above limitations, our study was significant for several reasons. The optimal conduit size in Fontan circulation was found to be about 12.5 mm/m2 conduit diameter per BSA (18 or 20 mm Gore-Tex conduit); these patients showed the best exercise capacity. Patients with larger conduits were less able to exercise. These results suggest that larger-sized conduit from optimal size may be more attenuating factor rather than smaller for supplying preload during exercise. Therefore, we need not try to put in too large sized conduit in operation room. We discussed the theoretical background using the area change rate concept.