This prospective single-center study was conducted from May 2017 to June 2019 after approval was obtained from the institute ethics committee; informed consent was obtained from the guardians of all patients. All pediatric patients (younger than 18 years) with chronic BCS (defined as symptoms for longer than 6 months) for whom image-guided intervention was planned during the study period were recruited for participation. All patients had been diagnosed with BCS by means of ultrasound and had been evaluated for possible image-guided intervention as part of their clinical workup. Demographic variables, clinical symptoms, and baseline laboratory values (obtained within 48 hours before intervention) were recorded for included patients. Included patients underwent additional SWE evaluation for investigational purposes as an adjunct to clinical CDUS examinations performed before intervention and at serial time points after intervention to monitor for disease recurrence. No included patient had uncorrectable coagulopathy or secondary BCS (i.e., from hepatic tumor or abscess compressing hepatic venous outflow).

All patients underwent an ultrasound examination on the day of the image-guided intervention, performed before the procedure. The evaluation included both CDUS and investigational 2D SWE. Patients fasted for 4 hours before the ultrasound examination [22]. The examinations, including both CDUS and SWE, were performed by a single radiologist with 3 years of experience in diagnostic ultrasound (N.D.), who used an Aixplorer ultrasound unit (Supersonic Imagine).

The CDUS examination was conducted to evaluate the status of the hepatic veins, IVC, portal vein, collaterals, and ascites with focus on the number, degree, and length of narrowing of the hepatic veins and IVC. The CDUS assessment was qualitative; no Doppler indexes were evaluated.

A standard protocol was used to perform 2D SWE [22]. A single crystal 1–6 MHz (SC6-1) convex transducer was placed over the right intercostal space with the child in the supine position and the right arm abducted. The system default SWE settings were used, including high-penetrance SWE acoustic power, spatial smoothing of 5, and high persistence. Initially, the SWE box was placed on the right lobe of the liver approximately 2 cm deep to the capsule and avoiding major vessels and focal lesions [21, 23, 24]. The image was frozen when there was complete color filling in the SWE box after a 3-second breath-hold for children who could perform it and with free breathing otherwise. A circular ROI (1-cm diameter; Q-Box, Supersonic Imagine) was then placed on a homogeneous color area within the SWE box to record LSMs in kilopascals (Young modulus) (Fig. 1). The stabilization index was maintained above 90%. Three values were measured from the same area, and the median of these values was recorded [25].

After the baseline ultrasound examination that included LSM, the image-guided intervention was performed with general anesthesia. All interventions were performed jointly by two interventional radiologists (S.S. with 22 and K.S.M. with 12 years of experience in pediatric interventions). Liver congestion was relieved by recanalization of the native hepatic veins and IVC or by direct intrahepatic portosystemic shunt (DIPS). The former was preferred because the recanalization restores normal physiology; after DIPS insertion, venous blood flow bypasses the liver. Venous angioplasty was considered indicated when stenosis was smooth and discrete. Stent deployment was considered indicated when stenosis was irregular, long segment, or recurrent or when in-stent thrombosis was present. DIPS was considered indicated only when no hepatic veins that could be recanalized were visualized on CDUS. A combination of transjugular, transhepatic, and transfemoral approaches were used, depending on the morphology of the occluded vein, presence of ascites, and technical ease. Results of the earlier CDUS examination were used to inform the choice of therapeutic intervention.

Liver biopsy was performed at intervention for BCS through the same access used for the intervention (Rosch-Uchida transjugular liver access set, LABS100, Cook Medical; 18-gauge semiautomatic biopsy gun, Bard Mission, BD). Liver biopsy was not performed in patients in whom the procedure became prolonged owing to technical complexity. Three 15- to 20-mm-long cores were obtained with an 18-gauge semiautomatic biopsy needle. The biopsy sample was assessed by a pathologist (P.D.) with 12 years of experience in staging of fibrosis using Metavir score (F0 to F4) [26]. The sample was stained with Sirius red, and fibrosis was quantified with computerized image analysis software ((Image-Pro Plus version 6.1, Media Cybernetics) to determine the collagen proportionate area (CPA).

All children underwent clinical and ultrasound follow-up for disease recurrence (restenosis or reocclusion). All follow-up ultrasound examinations were performed by the same radiologist who performed the baseline examination (N.D.) and included both CDUS and investigational SWE by the previously described approach (including recording of LSM based on the median of three values). CDUS and SWE examinations were performed 24 hours, 1 month, and 3 months after the procedure and every 3 months thereafter or when recurrence was clinically suspected. Recurrence was diagnosed on B-mode ultrasound on the basis of the presence of luminal narrowing or echogenic material within the recanalized vein or stent and/or on CDUS when a vein or stent showed no, partial, or reversal of flow within the vein or stent; reappearance and dilation of collaterals; or reappearance or worsening of ascites. SWE was not used clinically for diagnosing recurrence. If a recurrence was detected on ultrasound, the patient was referred for image-guided reintervention. SWE was performed 24 hours after reintervention.

LSMs were summarized at baseline and 24 hours, 1 month, and 3 months after intervention. Summary measures of LSMs were expressed as median, except for summary of relative decrease in LSM between time points, which was expressed as mean. Freidman test was used to compare the mean ranks of LSMs among the time points. Paired Wilcoxon signed-rank tests were used to compare LSMs between baseline and each postintervention time point. Mann-Whitney tests were used to compare patients with and without recurrence in terms of LSM at each time point and in terms of absolute and relative decreases in LSMs from baseline to 24 hours. In patients with recurrence, LSMs were summarized at recurrence and at most recent measurement before recurrence. The absolute and relative decreases between these two LSMs at each time point were summarized, stratified by Metavir score. Spearman correlation coefficients with 95% CIs were computed between LSM and both Metavir score and CPA at each time point and between Metavir score and CPA. Mann-Whitney tests were used to compare patients grouped as having F0–F2 versus F3–F4 scores in terms of LSM at each time point and in terms of absolute and relative decrease in LSM from baseline to 24 hours. A spaghetti plot was constructed to depict LSMs at all time points in all patients; trend lines were generated for patients with and without recurrence. Values of p < .05 were considered statistically significant. Statistical analysis was performed with commercial SPSS software for Microsoft Windows (version 18, IBM).

During the study period, image-guided intervention was planned for 35 children with chronic BCS; consent was obtained for all 35 of these children. One was lost to follow-up before the intervention, and two died before the intervention, resulting in a final study sample of 32 patients (28 boys, four girls; mean age, 9 years; range, 3–14 years) (Fig. 2). Baseline demographic, clinical, and laboratory characteristics of the 32 patients are shown in Table 1. The image-guided interventions included hepatic vein angioplasty (n = 9), IVC angioplasty (n = 5), hepatic vein stenting (n = 7), combined hepatic vein stenting and IVC angioplasty (n = 5), combined hepatic vein and IVC angioplasty (n = 2), and DIPS (n = 4).

All 32 patients underwent 24-hour follow-up evaluation, at which time one recurrence was diagnosed. Among the other 31 patients, one was lost to follow-up, and 30 underwent 1-month follow-up evaluation, at which time three recurrences were diagnosed. Among the other 27 patients, two were lost to follow-up, and 25 underwent the 3-month follow-up evaluation, at which time two recurrences were diagnosed. Among the other 23 patients, 12 were lost to follow-up, and 11 underwent at least one additional follow-up imaging evaluation with CDUS and SWE after the 3-month time point. Three additional recurrences were diagnosed during this further imaging evaluation, giving a total of nine recurrences among the 32 patients (28.1%). The mean follow-up duration for the 32 patients was 4.4 months (range, 24 hours–19 months). Among the nine patients with recurrence, the mean interval between intervention and recurrence was 4 months (range, 24 hours–18 months). Among the 23 patients without recurrence, the mean follow-up duration was 4.3 months (range, 24 hours–19 months).

LSMs varied significantly among the time points (p < .001). The median LSM at baseline was 43.7 (interquartile range [IQR], 33.0–65.4) kPa, at 24 hours was 22.5 (IQR, 16.8–32.0) kPa (p = .001 vs baseline), at 1 month was 18.7 (IQR, 14.2–32.0) kPa (p = .001), and at 3 months was 16.7 (IQR, 11.5–22.5) kPa (p = .001), corresponding to reductions from baseline of 48.5%, 57.2%, and 61.8% (Figs. 3 and 4). Table 2 shows the median LSM at each time point, stratified by presence or absence of recurrence. Median LSM was not significantly different between the two groups at baseline or at any time point after intervention (p = .34–.93). In addition, patients without recurrence and patients with recurrence were not significantly different in terms of mean relative decrease in LSM from baseline to 24 hours after intervention (47.1% ± 14.1% [SD] vs 36.7% ± 27.6%; p = .31) or in terms of median absolute decrease between these two time points (23.4 vs 13.6 kPa; p = .36). However, among patients without recurrence, median LSM decreased from 24.6 kPa at 24 hours to 16.1 kPa at 3 months, whereas in patients with recurrence, median LSM increased from 19.1 kPa at 24 hours to 21.4 kPa at 3 months. Figure 5 depicts LSMs of all patients over time, including trend lines for patients with and without recurrence.

Table 3 summarizes characteristics of the nine patients with recurrence. Three of the nine patients with recurrence had clinical evidence of recurrence (ascites in all three). The other six had no clinical signs, and the diagnosis was based solely on imaging surveillance. All nine recurrences were diagnosed on the basis of B-mode ultrasound and/or CDUS findings. Two of nine patients had in-stent hepatic vein thrombosis with reversal of flow; five (all whom underwent angioplasty) had significant restenosis (> 30% narrowing of diameter) at the hepatic vein ostium and the suprahepatic IVC with distal vein dilatation; and one patient each had complete thrombosis of the IVC and of the hepatic veins.

Recurrence in one patient was diagnosed at the first postintervention evaluation (LSM, 44.2 kPa at baseline vs 52.3 kPa at 24 hours). In an additional patient with recurrence, the LSM at recurrence was similar to the most recently recorded prior value (17.0 kPa at 3 months vs 17.5 kPa at 4 months). This patient had undergone hepatic vein stenting and IVC angioplasty. Although in-stent thrombosis developed in the hepatic veins during follow-up, the IVC remained patent, and a dilated caudate vein drained into the IVC. In the other seven patients with recurrence, LSM at recurrence was at least 10% greater than the most recently recorded value. In the eight patients with recurrence diagnosed after the 24-hour measurement, the median absolute increase in LSM at recurrence compared with the most recently recorded value was 11.0 (IQR, 6.1–24.4) kPa, and the mean relative increase was 61.6% ± 37.9%.

Among the nine patients with recurrence, six underwent reintervention, which included angioplasty of the hepatic vein stent in two patients with in-stent thrombosis, stent placement in three patients (hepatic vein in two, IVC in one) with recurrent stenosis after balloon angioplasty, and DIPS placement in one patient. The median absolute decrease and mean relative decrease in LSM 24 hours after reintervention compared with LSM at recurrence diagnosis were 28.3 (IQR, 16.2–35.2) kPa and 41.0% ± 35.6%. In the previously noted patient for whom LSM was similar at diagnosis and for the most recent prior value, LSM increased 28.6% from the time of recurrence to 24 hours after reintervention. In the other five patients who underwent reintervention, LSM decreased at least 34.8% 24 hours after reintervention.

Liver biopsy was performed at intervention in 18 patients (all boys; mean age, 8 years), yielding a Metavir score of F0 in 0, F1 in one patient (5.6%), F2 in five (27.8%), F3 in five (27.8%), and F4 in six (33.3%) patients. The mean CPA was 16.2% ± 12.2%. Median LSMs stratified by Metavir score are shown in Table 4. LSM was not significantly correlated with Metavir score at baseline (n = 18; r = 0.11 [95% CI, –0.37 to 0.54]; p = .51), 24 hours (n = 17; r = 0.39 [95% CI, –0.11 to 0.73]; p = .11), 1 month (n = 14; r = 0.31 [95% CI, –0.21 to 0.73]; p = .31), or 3 months (n = 10; r = 0.52 [95% CI, –0.16 to 0.86]; p = .12). Similarly, LSM was not significantly correlated with CPA at baseline (r = 0.09 [95% CI, –0.39 to 0.53]; p = .76), 24 hours (r = 0.33 [95% CI, –0.33 to 0.70]; p = .26), 1 month (r = 0.17 [95% CI, –0.39 to 0.64]; p = .54), or 3 months (r = 0.69 [95% CI, 0.11 to 0.92]; p = .02) (Fig. 6). CPA and Metavir score at baseline were moderately correlated (r = 0.71 [95% CI, 0.36 to 0.88]; p = .001). When Metavir score was grouped as F0–F2 versus F3–F4 (Table 5), the median LSM at 24 hours was higher for the F3–F4 group than for the F0–F2 group (29.7 vs 18.8 kPa; p = .03), though the groups were not different at other time points (all p > .05). Neither the absolute nor the relative decrease in LSM from baseline to 24 hours was significantly different between the two groups (both p > .05).