Comparison of Echocardiography and Computed Tomography/Magnetic Resonance Imaging for Assessing of Pulmonary Vein Stenosis in Children

Pulmonary vein (PV) stenosis is a rare disease with high recurrence and mortality rates. Objective. This study was aimed to evaluate the diagnostic accuracy of echocardiography versus cross-sectional modalities. Material and methods. Our study includes 11 consecutive patients who underwent a comprehensive PV anatomy assessment between April 2018 and June 2019 at Ukrainian Children’s Cardiac Center, Kyiv. Echocardiographic studies of each pulmonary vein followed by computed tomography (CT) or magnetic resonance imaging (MRI) were performed in all patients. A total of 27 CT/MRI studies were performed. Results. There were four patients whose echocardiography findings were completely confirmed by CT/MRI method. In some patients significant stenosis was underestimated by echocardiography. Sensitivity and specificity were 86.4 and 88.9 for echo, as well as 100 and 94.4 for CT/MRI, respectively. Conclusion. This study recognizes the limitations of echocardiography as an imaging tool for pulmonary veins morphology assessment. Cross-sectional studies overcome these limitations and provide excellent morphological as well as functional (MRI) evaluation of pulmonary veins.


Introduction
Pulmonary vein stenosis (PVS) in children is a rare but serious condition [1,2] which may occur in 10-17% of patients after total anomalous pulmonary venous connection (TAPVC) repair [3,4], or presents as a congenital/primary disease bearing high recurrence and mortality rates of 46-80% [5]. Pulmonary veins imaging is important for preoperative planning, outcomes assessment and followup. Different imaging modalities can provide valuable information on severity and location of the lesions. However, each method has its own benefits and limitations.
In this study we aimed to compare the diagnostic accuracy of echocardiography versus cross-sectional modalities (computed tomography or magnetic resonance imaging) for assessment of pulmonary veins characteristics and to develop an institutional PVS detection and surveillance algorithm.

Materials and methods
Our study includes 11 consecutive patients (Table 1) who underwent a comprehensive PV anatomy assessment between April 2018 and June 2019 at Ukrainian Children's Cardiac Center, Kyiv. Two-dimensional and Doppler echocardiographic studies of each pulmonary vein with following computed tomography (CT) or magnetic resonance imaging (MRI) were performed in all patients. A total of 27 CT/MRI studies were performed, and each was compared to the respective echocardiographic study.
Echocardiography. Each PV was assessed by echocardiography according to a standardized imaging protocol described by Kalfa et al. [6], consisting of three components: (1) severity of disease measured by mean echocardiographic pressure gradient and (2) extent of stenosis in each individual pulmonary vein (PV) and (3) uni-or bilateral extent of the disease. ECHSA Score was used to summarize the findings. A mean echo gradient of <2 mmHg, biphasic blood flow and the absence of PV narrowing confirmed the absence of stenosis. A mean echo gradient of 2 to 4 mmHg, 5 to 7 mmHg and >7 mmHg indicated mild, moderate and severe stenosis, respectively. Involvement of only a short length of the PV and presence of upstream dilatation was defined as a focal stenosis. Stenosis was considered diffuse if a significant length of the PV was af-Український журнал серцево-судинної хірургії 2020 2 (39) Загальні питання серцево-судинної хірургії fected with no upstream dilation of the PV. Each of PV has been assigned a score of 0-4: 0 for no stenosis; 1, 2, 3 for discrete mild, moderate and severe stenosis; 1.5, 2.5, and 3.5 for diffuse stenosis; 4 for atresia or occlusion.
All individual pulmonary veins scores were summarized to calculate the total score. If bilateral disease was present, 2 points have been added. Possible score range was 0-18 [14].
Cross-sectional imaging modalities. For CT/MRI assessment of pulmonary veins, we used the method proposed by the Toronto group [7,10]. Short and long cross-sectional diameters of each PV at the left atrial junction (downstream) and the bifurcation (upstream), the individual PV cross-sectional area indexed to body surface area (PVCSAi) [8] and total PV cross-sectional area indexed to body surface area (TCSAi) were measured. In addition, in patients undergoing MRI pulmonary artery flow redistribution was assessed.
Statistical analysis. Predictive accuracy for echocardiography and cross-sectional modalities was assessed by calculating the area under the receiver operating characteristics (ROC) curve.

Results
The results of echocardiography findings and crosssectional studies compared with intraoperative findings are presented in Table 1.
Echocardiography findings in patients 2, 6, 7 and 10 were completely confirmed by CT/MRI method.
In patient 1 CT scan revealed left upper PV atresia and left lower PV severe stenosis, whereas echocardiography showed only moderate stenosis of the left lower PV.
In patient 11 echocardiography ( Fig. 1) which was performed in a month after primary echo study revealed decreased mean echo gradients on left pulmonary veins.
Meanwhile, CT (Fig. 2) showed worsening of the stenosis compared to CT imaging performed month earlier.
CT scans in the first and third months following sutureless repair on the left pulmonary veins revealed unobstructed flow throw these veins.
Eight months after primary surgery, this patient was readmitted with clinical signs of infective endocarditis. Echocardiography revealed vegetations on tricuspid valve and suspected left pulmonary veins restenosis. CT imaging which was performed to confirm the diagnosis revealed new right upper PV stenosis and left pulmonary veins restenosis.
In patient 8, in addition to detected secundum atrial septal defect (ASD), severe right upper pulmonary vein (RUPV) stenosis was suspected during echocardiography. Preoperative CT revealed normal dimensions of RUPV. Thus, this patient underwent minimally invasive right midaxillary thoracotomy for ASD closure without unnecessary RUPV intervention. Patient 9 was admitted to our clinic with signs of pulmonary edema. Echocardiography revealed severe pulmonary hypertension with unclear cause and without relevant pressure gradients over pulmonary veins. Subsequent CT scan showed severe hypoplasia of all pulmonary veins.
All aforementioned patients had primary PVS. Patient 10 was the only one in this series with postsurgical PVS whose CT scan identified stenosis of pulmonary venous collector.
Statistical analysis. Both methods were compared using ROC analysis. ROC curves showed that cross-sectional study had higher sensitivity and specificity (Fig. 3). Thus, cross-sectional modalities were determined as those providing higher diagnostic accuracy.

Discussion
We studied diagnostic accuracy of echocardiography for assessment of pulmonary veins characteristics and compared it with CT/MRI method. We found that only in four patients CT/MRI methods completely confirmed echo findings. In some cases echocardiography could not pro-vide visualization of individual pulmonary veins. Subsequently, CT method defined these veins as suboccluded or atretic. Also, in these cases echo was not able to reliably detect the severity of the stenosis: CT/MRI method characterized individual veins as severely stenosed, whereas echocardiography only suspected possible stenosis of the same veins. Likewise, we observed a case where echocardiography showed high mean pressure gradients on an in- Table 1 Pulmonary veins characteristics

Figure 3. ROC curves of sensitivity and specificity of echo (a) and cross-sectional study (b) with their comparison (c)
Український журнал серцево-судинної хірургії 2020 2 (39) Загальні питання серцево-судинної хірургії dividual vein. CT method excluded this diagnosis. In a patient with infective endocarditis CT contributed to inform the surgical team of the possible scenario and to plan the surgical procedure, which subsequently was performed with deep hypothermic circulatory arrest. In addition, in case when echocardiography showed severe pulmonary hypertension with normal values of blood flow velocity in pulmonary veins, CT/MRI imaging allowed to determine the cause of detected findings.
Current scoring systems developed to assess the characteristics of pulmonary veins are based mainly on echocardiographic measurements [5]. Echocardiography is mostly accurate in measurement of pulmonary veins at downstream level and when there is no flow redistribution from lung segments with obstructed pulmonary veins to unobstructed segments. In case of unbalanced circulation between both lungs, assessment of pulmonary veins chara cteristics via echocardiography can lead to underestimation of the stenosis severity [9]. Tremblay et al. demonstrated that echocardiography couldn't provide complete visualization of individual pulmonary veins in the majority of cases [10]. Compared to echo, CT scan can reveal affected pulmonary veins segments that are remote from the heart (upstream).Thus, CT/MRI imaging provides a more comprehensive assessment of anatomy of pulmonary veins compared with traditional echocardiographic imaging. Currently, CT/MRI imaging could be considered as the gold standard for the evaluation of PV flow and anatomy. Vanderlaan et al. [11] adopted a protocol incorporating computed tomography or magnetic resonance imaging at 1, 6, and 12 months after PVS repair. Primary PVS is mostly characterized by abnormal neointimal proliferation, whereas in postsurgical patients heavily collagenized scarring is the most prominent histological finding [12]. These underlying morphology predisposes high recurrence rates and need for rigorous surveillance. Due to its zero radiation load and flow-measurement capabilities MRI may be more beneficial surveillance tool in PVS cohort.
According to our previous approach, CT/MRI imaging was performed only in complex cases [13] mostly to clarify the anatomy. Since April 2018 all patients in our institution with PVS undergo cross-sectional imaging for the veri fication of diagnosis and surveillance after repair.

Conclusion
This study recognizes the limitations of echocardiography as an imaging tool for pulmonary veins morphology assessment. Cross-sectional studies overcome these limitations and provide excellent morphological as well as functional (MRI) evaluation of pulmonary veins. We have implemented CT/MRI scans for diagnosis and surveillance in all patients with PVS.