Maldistribution of pulmonary blood flow in patients after the Fontan operation is associated with worse exercise capacity

A database search identified all post-operative Fontan patients who had a CMR study and cardiopulmonary exercise test (CPET) at Boston Children’s Hospital between January 1999 and July 2017. Patients were included if differential branch pulmonary artery (PA) flow could be calculated by CMR and if they had a maximal effort on exercise stress testing, defined as a respiratory exchange ratio of ≥1.09 or a heart rate of ≥75% predicted. Patients were excluded if there were any interventions or procedures between the CMR and CPET or if the period between CPET and CMR was more than 2 years. The Boston Children’s Hospital Committee on Clinical Investigation approved this retrospective study and waived the requirement for informed consent.

CMR studies were performed with 1.5 Tesla scanners (Philips Healthcare, Best, the Netherlands or GE Medical Systems, Milwaukee, Wisconsin). The details of the CMR protocols used in our laboratory for assessment of patients after the Fontan operation have been published [13‐15]. Briefly, ventricular assessment was performed by an electrocardiographically-gated, balanced steady-state free precession (bSSFP) cine CMR in vertical and horizontal ventricular long-axis planes, and a stack of slices in a ventricular short-axis plane encompassing the atrioventricular junction through the cardiac apex. Retrospectively cardiac gated, free-breathing, through-plane phase-contrast flow meaurements were obtained in the branch pulmonary arteries and vena cavae. Care was taken to align the imaging plane perpendicular to flow and to obtain slice positions and orientations that were proximal to the PA branching [4].

If a patient had multiple CMR studies, the most recent study with complete flow data was used for analysis. MPBF was calculated by direct measurement of the branch PA flow on phase contrast imaging. Branch PA flow was measured by manually tracing each branch PA on phase contrast imaging using QFlow (Medis Medical Imaging Systems, Leiden, The Netherlands) (Fig. 1a) [16]. Percentage flow to each PA was calculated. In patients without baffle leaks or patent fenestrations who had unilateral PA stents where direct PA flow could not be measured, flow in the stented PA was calculated using the following formula: superior vena cava flow+ inferior vena cava flow – the non-stented branch PA flow. Branch PA cross-sectional area was calculated by measuring two orthogonal dimensions at the narrowest segment and indexed to body surface area (Fig. 1b). A branch PA symmetry index (PASI) was calculated as the ratio of the area of the smaller pulmonary artery to the larger pulmonary artery. PASI is always ≤1 with values closer to 1 reflecting more symmetric branch PAs [17]. PA compression by the ascending aorta or aortic root was determined by review of CMR images by a provider who was blinded to the PA blood flow distribution. PA compression was defined as narrowing of the branch PA to < 75% of its original diameter as it crossed posterior to the ascending aorta or the aortic root (Fig. 1c) [18]. Lung volumes were calculated by Simpson’s method using manual tracing of the lung fields in each slice on an axial image bSSFP stack (Fig. 1d). Lung volume discrepancy was defined as the absolute value of the difference between the right and left lung volume percentage. Aortic root total area was measured by adding dominant to non-dominant aortic root area as measured on axial planes [19].

Ventricular volumes and function were measured by manual tracing of endocardial and epicardial borders on each short-axis bSSFP cine slice at end-diastole (maximal volume) and end-systole (minimal volume) as previously described [5, 15]. Analysis was performed using commercially available software (QMass, Medis Medical Imaging Systems, Leiden, The Netherlands) and (cmr⁴², Circle Cardiovascular Imaging Inc., Calgary, Canada).

Demographic and clinical data, including underlying diagnoses and type of single ventricle based on ventricular dominance, were abstracted from the medical records. The type of surgical palliation was classified as lateral tunnel, extracardiac conduit, right atrium-to-PA anastomosis, or right atrium-to-right ventricle connection. Additional parameters included age at Fontan, time from Fontan to CMR, and number and type of surgical and catheterization interventions before and after CMR. Arrhythmia history was compiled by review of Holter monitors, electrocardiograms, electrophysiology catheterizations, and clinic notes. Other relevant clinical variables included a history of heart failure (defined as New York Heart Association class II or greater), protein-losing enteropathy, stroke, thrombus, seizures, liver disease, or pacemaker or defibrillator placement.

A maximal CPET was performed using a calibrated cycle ergometer and ramp protocol (Corival Load Cycle 400, Lode BV, Groningen, The Netherlands). The test starts with setting an initial work rate based on the patient’s body surface area (BSA) with linear increases every minute reaching a peak exercise after 10 min. Gas exchange was analyzed at rest, during exercise, and during recovery to determine measures of oxygen uptake (VO₂) [20]. Since peak VO₂ is influenced by age, sex, and body weight, the percent of predicted peak VO₂ value (% predicted VO₂) was used due to the wide age range in this study [21].

The Student t-test or Mann-Whitney U test was used to compare two groups of continuous symmetric or non-symmetric variables, respectively, or Fisher exact for categorical variables, as appropriate. A normal approximation to binomial confidence interval was constructed for the percentage of patients with MPBF. Univariate association between normally distributed variables was estimated using the Pearson correlation coefficient. A stepwise multivariable linear regression modeling procedure with 0.1 as the significance level for entry and 0.05 as the significance level for retention in the model was constructed to determine independent predictors of % predicted VO₂. A stepwise multivariable logistic regression model with 0.1 as the significance level for entry and 0.05 as the significance level for retention in the model was constructed to identify the independent factors associated with the presence or absence of MPBF. Continuous predictor variables were also categorized into tertiles to assess potential nonlinearity, but no nonlinear significant associations were found (data not shown). All p-values were two-tailed (where applicable) and differences and associations were considered significant when p < 0.05. Statistical analyses were performed using SPSS Statistics for Windows, Version 24.0 (International Business Machines, Armonk, New York, USA) and JMP®, Version 12 (SAS Institute Inc., Cary, North Carolina, USA).

MPBF was present in 53 patients (36%, 95% CI 29–45%). A histogram of differential PA blood flow is shown in Fig. 2. Total PA blood flow was 2.46 ± 0.6 L/m². Average LPA blood flow was 1.07 ± 0.35 L/m² which represented 44 ± 12% of total PA flow. Average RPA blood flow was 1.38 ± 0.44 L/m² which represented 56 ± 11% of total PA flow. There was a weak correlation between branch PA flow percentage and ipsilateral PA cross-sectional area and less than 10% of the variability can be explained by cross sectional area (r² < 0.1) (Fig. 3). Similarly there was a weak correlation between branch PA flow percentage and lung volume percentage (Fig. 3).

The mean % predicted VO₂ was 63 ± 16%. Patients with MPBF had lower % predicted VO₂ compared to patients without MPBF (60 ± 14% versus 65 ± 16%; p = 0.04) (Fig. 4). Additional univariate associations of lower % predicted VO₂ are shown in Table 2 and included longer time since Fontan, older age at the Fontan operation, atriopulmonary connection Fontan, heart failure symptoms, presence of a fenestration, and a higher ventricular mass-to-volume ratio. On multivariable analysis, only MPBF, time since Fontan, and ventricular mass-to-volume ratio were associated with a lower % predicted VO₂ (Table 3).

During a median follow-up period of 4.2 years [IQR 2.1–8.4] after the CMR, 14 (10%) patients died or were listed for heart transplant. Of the 10 patients who died, deaths were attributed to arrhythmias (n = 4), heart failure (n = 4), renal failure (n = 1), and protein losing enteropaty (n = 1). The follow-up period was similar among those with and without MPBF (Table 4). MPBF was not associated with death or listing for transplant (p = 0.60). MPBF was not associated with other comorbidities including atrial flutter, heart failure, protein-losing enteropathy, major thrombotic events, liver disease, or stroke (Table 4).

On univariate analysis, a surgical history of a Damus-Kaye-Stansel anastomosis, having a cardiac diagnosis of HLHS, prior Glenn procedure, and history of prior PA intervention were associated with MPBF (Table 1). The CMR parameters and their associations with MPBF are shown in Table 5. On univariate analysis, MPBF was associated with larger ventricular volumes, increased aortopulmonary collateral flow, larger ascending aorta and aortic root cross-sectional areas, PA compression by the ascending aorta or aortic root, PASI, and larger lung volume discrepancies. On multivariable logistic regression analysis, only PA compression by the ascending aorta or aortic root was associated with MPBF (OR = 6.5, 95% confidence interval 5.6–7.4, p < 0.001). Branch PA compression was seen in 32 patients (21%), and involved the LPA in 30 patients and the RPA in 2 patients.

This is a single-center CMR study which may limit the generalizability of these results to all Fontan patients. In particular, patients with pacemakers and defibrillators could not be evaluated; these devices are used in 13% of patients with Fontan circulation [37]. Sicker patients and patients with symptoms could be over-represented in this study, as these patients are more likely to be evaluated by CMR [38]. Also, 78 patients were excluded due to the incomplete flow data. Many of these patients had branch PA stents and may have had severe MPBF. Finally, this study included only patients with a maximal CPET, which limited the sample size and may have caused selection bias toward patients with higher functional status.