Magnetic resonance three-dimensional steady-state free precession imaging of the thoracic duct in patients with Fontan circulation and its relationship to outcomes

This was a retrospective, single-center, study of patients with Fontan circulation who underwent CMR with 3D bSSFP imaging at Boston Children’s Hospital between January 1, 2011 and November 1, 2021. Typical imaging parameters for the 3D bSSFP whole-heart angiogram in an adult patient are as follows: field of view 410 × 240 mm, voxel size 1.6 × 1.6 × 1.6 mm³ reconstructed to 0.6 × 0.6 × 0.6 mm³, flip angle 110°, echo time 2.2 ms, repetition time 4.5 ms, and compressed SENSE reduction factor 3. Respiratory motion compensation was performed using a self-navigator to track and gate heart position [12]. In the cases of multiple CMR examinations per patient, only the most recent study was included. Frequency matching with respect to age at CMR was used to construct a comparison group of patients with repaired TOF. Other clinical data reviewed included clinical notes, exercise stress tests, cardiac catheterizations, and echocardiograms obtained within 1 year of CMR.

Thoracic duct visualization was classified as completely visualized, nearly completely visualized, incompletely visualized, or not visualized. A completely visualized thoracic duct is one in which the entire length through the thorax to its insertion at the venous angle can be identified. A nearly completely visualized thoracic duct is one in which only a short length (no greater than the length of two vertebrae) is not visualized. Larger portions of the thoracic duct are not well visualized in an incompletely visualized thoracic duct. Thoracic duct size was represented by the maximum thoracic duct diameter. To obtain the maximum thoracic duct diameter the largest portion of the thoracic duct was first identified in the coronal plane. Then, the largest diameter of that area was obtained in a short axis (double oblique) view. Thoracic duct tortuosity was qualitatively graded as none, mild, moderate, and severe based on predefined representative cases (Fig. 1) and the following guidance: no turns/deviations from course for no tortuosity, one turn or multiple very subtle turns for mild tortuosity, more than one large turn or multiple small turns for moderate tortuosity, and turns along the entire or nearly entire course for severe tortuosity. In patients in which a T2-weighted (T2W) sequence for lymphatic assessment was obtained, the T2W sequence was reviewed to assess whether the identified the thoracic duct on the 3D bSSFP sequence corresponded with the thoracic duct visualized on the T2W sequence. Patients were classified by New York Heart Association (NYHA) functional class based on review of the clinical notes. Aortopulmonary collateral burden was represented as a percent of systemic blood flow and was calculated using the systemic method [13]. Exercise test results were only included if peak respiratory exchange ratio was greater than 1.09 (a threshold for peak exertion in our lab). A composite clinical outcome was defined as the presence of PLE, plastic bronchitis, listing for heart transplantation, or death. The study was approved by the Institutional Review Board at Boston Children’s Hospital and the requirement for informed consent was waived.

Thoracic duct visualization was dichotomized as completely or nearly completely visualized versus incompletely or not visualized. For all analyses except for interobserver and intraobserver agreement, thoracic duct tortuosity was analyzed as a dichotomous variable: none or mild versus moderate or severe. Descriptive statistics are presented as mean ± standard deviation (SD) for continuous measures with approximately a normal distribution and median (interquartile range) for other continuous or ordinal measures. Categorical data are described as frequency and percentage. Where mean comparisons are presented, Student’s t-test or ANOVA was performed and where median comparisons are presented, a Wilcoxon rank sum test or Kruskal–Wallis test was performed. For comparisons of categorical variables, a Fisher exact or Chi-square test was performed.

Analysis of covariance was also used to estimate mean thoracic duct size adjusted for differences in age at CMR in patients with Fontan circulation with vs. without clinical outcome. Multiple linear regression was used to estimate the association between thoracic duct size and clinical variables while controlling for age. A second observer repeated measurements for thoracic duct visualization, tortuosity, and maximum diameter in 50 patients. The first observer repeated measurements for thoracic duct tortuosity and maximum diameter in 50 patients 1 year after initial data collection to assess intraobserver agreement. For thoracic duct visualization and tortuosity (analyzed as a categorical variable with 4 categories), agreement was assessed using percentage agreement and a kappa statistic. For thoracic duct diameter, agreement was assessed using the intraclass correlation coefficient.

A p-value < 0.05 was considered to be statistically significant. Analyses were performed with SAS 9.4 (SAS Institute, Inc., Cary, North Carolina) and R version 4.03 (R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://​www.​R-project.​org/​).

Between the groups with Fontan circulation and rTOF, there was no difference in sex, age, or body surface area (BSA) at CMR (Table 1). The thoracic duct was better seen (complete or nearly complete visualization in 65% vs. 22%, p < 0.001) in patients with Fontan circulation relative to patients with rTOF. There was no difference between the frequencies of image artifact between the two groups (9% vs. 11%, p = 0.75, Table 1). In patients with Fontan circulation, the thoracic duct was more often well seen in male patients (Additional file 1: Table S2). There was no statistically significant difference in visualization based on the primary cardiac diagnosis, type of Fontan, ventricular morphology, age, or BSA.

Contrast was administered to 84 (44%) patients with a Fontan circulation. Gadobutrol (Gadavist^®) was administered to 76 patients, gadobenate dimeglumine (Magnevist^®) to 7 patients, and gadofosveset trisodium (Ablavar^®) to one patient. At our institution, if a patient receives contrast it is routinely administered prior to acquisition of the 3D bSSFP sequence. There was no difference in thoracic duct visualization rates between those who received contrast and those who did not receive contrast (complete or nearly complete visualization of 69% in those that did receive contrast vs. 61% in those that did not, p = 0.29).

The thoracic duct diameter was measured in 171 patients with Fontan circulation and 20 patients with rTOF. In patients with rTOF, thoracic duct diameter did not correlate with BSA, with a Spearman’s rank-order correlation of r_s(18) = − 0.034, p = 0.89. Therefore, unadjusted mean or median diameters are reported when comparing these two groups of patients. The thoracic duct was larger (median diameter of 2.50 mm IQR 2.10, 3.10 mm vs. 1.95 mm IQR 1.70, 2.60 mm, p = 0.002) in patients with Fontan circulation relative to patients with rTOF (Table 1). Thoracic duct tortuosity was assigned to 169 patients with Fontan circulation and 20 patients with rTOF. There was no difference in thoracic duct tortuosity between patients with Fontan circulation and patients with rTOF (moderate or severe tortuosity in 27% vs. 8%, p = 0.17). While the primary cardiac diagnosis and type of Fontan pathway were different, thoracic duct tortuosity did not vary by sex, age at Fontan operation, need for Fontan revision, age at MRI, BSA, or the presence of heterotaxy syndrome (Table 2).

Clinical events in patients with Fontan circulation are demonstrated in Table 3. In the patients with Fontan circulation, 1% had plastic bronchitis (in each case it was diagnosed prior to CMR), 2% had PLE (half were diagnosed prior to CMR), 11% had NYHA class ≥ II, 9% had a major thrombotic event, and 28% had a history of arrhythmia (23% with an atrial arrhythmia, 2% with both an atrial and ventricular arrhythmia). Thoracic duct diameter increased mildly with age in patients with Fontan circulation, R = 0.19, p = 0.01, and age at CMR differs according to outcome status for several of the outcomes. Therefore, age-adjusted means with standard error (SE) were reported. The thoracic duct diameter was larger in patients with PLE (age-adjusted mean of 4.11 SE 0.49 mm vs. 2.72 SE 0.07 mm, p = 0.005) vs. patients without PLE, and was more tortuous in patients with NYHA class ≥ II vs. patients with NYHA class I (moderate or severe tortuosity in 55% vs. 26%, p = 0.02).

Of the 189 patients with Fontan circulation, 51 (27%) had a fenestration, 130 (69%) did not have a fenestration, and 8 (4%) had an unknown fenestration status. Age at CMR was significantly younger in the fenestration group (mean age of 14.4 ± 8.6 vs. 20.0 ± 10.0 years, p < 0.001). After adjusting for age, mean thoracic dimension was not significantly different based on fenestration status (age-adjusted mean of 2.58 SE 0.15 mm vs. 2.82 SE 0.09 mm in those with vs. without a fenestration, p = 0.18). There was no difference in the frequency of moderate or severe thoracic duct tortuosity between those with and without a fenestration (moderate or severe tortuosity in 31% with a fenestration vs. 27% without a fenestration, p = 0.70).

Patients with more tortuous thoracic ducts had a higher indexed ventricular end-systolic volume (mean of 70.0 ± 29.9 mL/m² vs. 57.3 ± 2.60 mL/m², p = 0.03), lower creatinine (mean of 0.61 ± 0.22 mg/dL vs. 0.70 ± 0.20 mg/dL, p = 0.04), and a higher absolute lymphocyte count (mean of 1.80 ± 0.78 K cells/µL vs. 0.76 ± 0.36 K cells/µL, p = 0.003, Table 4). Thoracic duct size was inversely correlated with ejection fraction independent of age (controlling for age), although the magnitude was small (partial correlation = − 0.22, p = 0.02, Table 5). Four patients were listed for heart transplantation (three were listed prior to CMR, one patient eventually underwent heart transplant) and four patients died. The median time from CMR to last date of clinical follow-up was 10.8 months (IQR 0.1, 27.0 months). Thoracic duct size (age-adjusted mean of 2.04 SE 0.50 mm vs. 2.77 SE 0.08 mm in those who died vs. survived, p = 0.149) and tortuosity (moderate or severe tortuosity in 25% vs. 30% in those who died vs. survived, p = 1.00) were not associated with death. The composite outcome was present in 11 (6%) of patients with Fontan circulation (mean age of 18.8 ± 12.8 years in those with the composite outcome and 18.3 ± 9.9 years in those without the composite outcome, p = 0.87) and was not associated with thoracic duct diameter (age-adjusted mean of 2.94, standard error (SE) 0.30 mm vs. 2.73 SE 0.08 mm in those with severe complication vs. without severe complication, p = 0.50) or tortuosity (moderate or severe tortuosity in 55% vs. 28% in those with severe complication vs. without severe complication, p = 0.09). Thoracic duct visualization and tortuosity showed substantial inter- and intraobserver agreement and maximum thoracic duct diameter showed moderate to good agreement (Table 6) [14, 15].

There are a number of limitations to our study. A portion of the clinical events occurred prior to the CMR; hence, the thoracic duct variables studied should be viewed as correlations and not predictors of clinical outcomes. Similarly, hazard of composite outcome from the time of CMR could not be performed. The use of CMR to visualize the thoracic duct also biases the sample as patients with pacemakers and ICDs will be excluded, and sicker patients are more likely to be referred for CMR. Additionally, given the T2 prep pulse of 3D bSSFP sequences, the presence of pleural or pericardial effusions may make it harder to distinguish lymphatic channels. Also, in patients with venous decompressing collaterals it may be difficult to distinguish lymphatic channels from venous channels. Thoracic duct visualization and size is affected by Fontan hemodynamics on the day of the CMR examination and patient diet preceding the examination. As only moderate agreement on qualitative assessment of thoracic duct tortuosity was noted, quantitative assessments of thoracic duct tortuosity may be useful for future assessments [11].