Long-term outcome of primary percutaneous stent angioplasty for pediatric posttransplantation portal vein stenosis

PV reconstruction during LT

The type of graft (whole, reduced size, split) as well as the individual recipient anatomy defined how reconstruction of the PV was performed. In general, in grafts with regular hilar region for whole and reduced size organs, PV anastomosis was performed in an end-to-end fashion. For splits, the PV usually is reconstructed by an anastomosis between the left PV of the donor and a right-/left-branch patch of the recipient PV in order to overcome the usually observed difference in diameter between adult and pediatric vessels. In case of fibrotic transformation of the recipient PV (as often seen in biliary atresia), the donor vein inserted directly at the confluence of the vessel is long enough or the recipient PV is replaced by a size-adapted allograft. To optimize PV flow, any known or intraoperatively identified collaterals are ligated as reported by others.^[14] Left gastric vein was ligated in case of insufficient flow; likewise, interruption of the left renal vein was attempted to improve portal flow in selected cases.

Definition of PVS and intervention criteria

Angiographic evaluation confirmed the diagnosis of PVS and assessed the hemodynamic relevance, including pre- and postinterventional trans-stenotic pressure gradient and collateral flow in the extrahepatic PV circulation. If applicable, stenosis of >50% prestenotic PV diameter and/or a trans-stenotic pressure gradient of ≥3.0 mm Hg served as criteria for intervention.

Procedural details

Percutaneous interventions were performed under general anesthesia. We established transhepatic access to the intrahepatic PV in most patients via intercostal puncture using micropuncture sets (Cook Medical, Bloomington, NJ) with a standard 21G needle under US and fluoroscopic guidance. A 0.018″ soft-tip wire or a coronary wire was used to pass through the stenotic vascular segment followed by 4.0- or 5.0-F diagnostic catheters for pressure monitoring and selective angiography. Transhepatic (n = 32) or trans-splenic (n = 8) PV angioplasties were performed with premounted stents (Palmaz Blue; Cordis Corporation, Warren, NJ; n = 25) in infants and small children, with a balloon diameter of 5–7 mm and length of 12–18 mm. Palmaz Genesis (Cordis Corporation, Miami, FL; n = 11) and VALEO stent (Bard Inc., New Providence, NJ; n = 1) were reserved for children with larger PV diameter or reinterventions. The balloon size was chosen according to the prestenotic PV diameter of either the portal or mesenteric vein. Distinct stent oversizing was avoided for the prevention of endothelial damage with the risk of PVT. Portography and manometry were repeated to evaluate procedural effectiveness and exclude vessel injury (Figure 1).

The trans-splenic approach has been described in detail elsewhere.^[15] It was used for patients with complete PVT, failed transhepatic approach, or increased risk for transhepatic puncture. Neither intraoperative PV stenting during LT^[16] nor direct access to the PV via minilaparotomy^{[17, 18]} were performed during the study period in our center.

After successful sheath introduction, the patients received heparin 50–100 IE/kg as bolus injection followed by 400 IE/kg/day for 24–48 h after catheterization. A single dose of antibiotic was administered to every patient with stent angioplasty. Unless contraindicated by increased bleeding risks or thrombocyte count below 50,000/μl, a platelet inhibitor (acetylsalicylic acid 1–3 mg/kg/day or clopidogrel 0.5–2 mg/kg/day) was added for at least 3–6 months. US was performed at the end of the procedure and at least twice within the first 48 h after the intervention to rule out bleeding complications.

Technical success, clinical success, and complications

Clinical success was defined as improved portal hypertension (e.g., reduction of ascites, splenomegaly, gastrointestinal tract bleeding, and thrombocytopenia).

Adverse events were graded according to the guidelines of the Cardiovascular and Interventional Radiological Society of Europe^[19] and the Society of Vascular Surgery.^[20] The need for additional surgery, unplanned catheter intervention, adverse sequelae, and death were considered severe complications. Other events were considered mild or moderate events.

Follow-up, patency, and risk factors for postinterventional restenosis

Clinical, laboratory, and imaging parameters were monitored during standardized follow-up (f/u) visits according to the local pediatric LT protocol. For this study, postinterventional in-hospital examinations and f/u visits 2 weeks; 3–6 months; and 1, 3, 5, and 10 years after PTBA were analyzed. During f/u visits, PV stent diameter and flow velocity were routinely evaluated by Doppler US to assess stent patency. According to the criteria of the Society of Vascular Surgery,^[20] primary patency was defined as no need for additional endovascular intervention after placement. Assisted patency was considered as status after reintervention during f/u. PV status after total occlusion followed by reintervention was classified as secondary patency.^[18] Patients with primary PTBA and subsequent stent placement for restenosis were considered patients with PTSA during further f/u. Retransplantation was considered the end of f/u period.

There are risk factors for development of posttransplantation PVS reported in the literature: age, weight, underlying disease, and surgical details; see Table 1. We initially hypothesized that these parameters as well as early posttransplantation PVS would result in a higher risk of restenosis after stent angioplasty and potentially lead to a shorter primary patency. We used the Cox regression model to analyze the effect of those factors on the freedom of reintervention and divided patients into subgroups below or above (1) 8-kg bodyweight or (2) 12 months of age at LT, (3) 10-kg body weight or (4) 24 months of age at PTSA, and (5) time between LT and PTSA below or above 3 months. Next, we assumed that suffering from fixed portal hypertension (chronic end-stage liver disease) might also play a role in a higher restenosis rate. Therefore, we introduced an additional risk profile based on the following criteria: prior abdominal surgery (e.g., Kasai operation), liver retransplantation, PVT followed by surgical revision, PV collaterals, and retrograde PV flow. Each patient's risk for PV restenosis was assessed using a risk score, resulting in three groups with low, moderate, or high risk.

Statistical analysis

Statistical analysis and visual representation of data were performed using the statistical programming language R (https://www.r-project.org/) version 1.3.1093. Parametric and nonparametric univariate and multivariate statistical testing were conducted for the analysis of the data. Student t test was used for normal distributed data. Other data were analyzed using Wilcoxon signed rank test. Cox regression and Kaplan–Meier analysis were used to analyze freedom from reintervention survival, with log-rank testing to discriminate between subgroups. Statistical significance was defined at a p value of <0.01.

Cohort characterization

Between 2004 and 2020, 470 pediatric LTs (median 2.0 years) have been carried out at our center. PVS was found in 44 patients (9.3%) receiving a total of 46 PV interventions.

Forty patients received a PV stent and were analyzed retrospectively in this study. In most patients (n = 38), stent angioplasty was performed as primary intervention. In two out of six patients, it was performed as second percutaneous intervention after previous balloon angioplasty.

The median time interval from LT to PTSA was 5.6 months, resulting in median age and weight at the percutaneous intervention of 23.5 months and 10 kg, respectively. For a detailed cohort description, see Tables 2 and 3.

All patients with PTSA showed relevant laboratory pathology (thrombopenia, mild increase of hepatic transaminases without other etiology such as graft rejection) and aggravating US findings during f/u, such as smaller PV diameter, progressive splenomegaly, increasing collateral flow, and reduced antegrade PV flow with risk for subsequent thrombosis.

Ascites was the leading symptom of relevant PVS in the subacute phase after LT. Twelve patients required continuous drainage and permanent intravenous fluid substitution before catheter intervention. By contrast, ascites was documented in only two patients with PVS occurring later than 6 months after LT.

Percutaneous intervention (technical success, clinical success, complications)

In 34 patients, the intrahepatic PV could successfully be reached through transhepatic access (Figure 1). In a single patient, this was switched to trans-splenic puncture during the same intervention. Trans-splenic access, as described before,^[15] was the primary access route in five patients.

Technical success could be achieved in all patients. Stent angioplasty led to a significant increase in the mean stenotic diameter (1.7–6.4 mm) and a significant reduction in the mean trans-stenotic pressure gradient (9.3–0.9 mm Hg residual gradient; Figures 1 and 2). In 10 cases, data before intervention were not available due to (sub)total occlusion of the vessel.

Six patients (median 26 months) in our cohort received primary PTBA. In this small group, the mean PV diameter could be increased by 3 mm (4–7 mm), reducing the mean PV gradient from 5.0 to 1 mm Hg. According to PV pressure and minimal PV diameter, these patients may retrospectively be considered having moderate PVS at the time of first intervention, except one with PVT, as described elsewhere.^[15] One case turned out as kinking PV with only mild PVS (diameter 7 mm, gradient 2 mm Hg) with almost no change after intervention (diameter 8 mm, gradient 2 mm Hg).

All patients, except for one patient with subarachnoidal bleeding, received standard anticoagulation with heparin as described in procedural details. Antiplatelet therapy with acetylsalicylic acid, clopidogrel, or both were given for at least 3–6 months. The duration of treatment was decided on the individual risk assessment for PVT.

The standard immune suppression regimen in our LT center is tacrolimus. PVS intervention alone did not lead to a change in immunosuppression.

In two patients, in-stent thrombosis or perisplenic hematoma occurred independently as complications of PTSA. Both conditions resolved with conservative treatment alone and were therefore graded as moderate/grade 3.^{[19, 20]} Early in-stent thrombosis occurred in coincidence with peri-interventional infection and resulted in a residual gradient of 6.0 mm Hg. After successful treatment with high-dose intravenous heparin (600 IU/kg/d) over 96 h in combination with platelet inhibition and intravenous antibiotics, patency was achieved for currently more than 12 years. Puncture-related perisplenic hematoma was not accompanied by hemodynamic compromise or a significant drop in hemoglobin concentration, and no vessel damage was found. The patient received a single elective transfusion of red cell concentrate.

A reduction of the size of the spleen and an increase in platelet count are major components of postinterventional clinical success. Splenomegaly of 3.4 cm above 95th percentile (average, long axis) was found in 37/38 patients before intervention. Two patients were excluded from this statistic due to polysplenia. Within the first 6 months after PTSA, spleen size decreased by an average of 1.5 cm in 30/38 patients. Half of the patients (10/20) reached a normal diameter at the 5-year f/u. A reduction in diameter persisted in 18/20 patients for >5 years and 7/8 patients for >10 years. Concordantly, the mean platelet count increased significantly (+20%) in 26/39 patients within 1 year (187,000/μl vs. 236,000 tsd/μl; p < 0.01).

The mean PV maximum flow velocity (v_max) before intervention was 160 cm/s (median 180 cm/s; maximum 400 cm/s; data available from 36 patients). Successful PTSA led to a significant decrease of >50% in PV flow velocity 1 year after intervention.

Collaterals due to portal hypertension could be identified in 16 patients before catheter intervention. In this group, PTSA resulted in a decrease of the mean PV flow in nine patients, whereas the v_max increased (55 to 89 cm/s) in seven patients. Accompanying PVT was found in five patients during angiography (4/5) but did not affect interventional success.

Gastrointestinal bleeding was seen in five patients with coexisting portosystemic shunts before percutaneous angioplasty (PTA) and did not occur after treatment.

In all 12 patients, with free abdominal fluid due to early postoperative PVS, stent angioplasty resolved ascites, allowing subsequent removal of the drainage, recovery, and discharge.

Follow-up, patency, and risk factors for postinterventional restenosis

Thirty-eight patients with PTSA had a f/u of at least 1 year, 21 patients >5 years, and eight patients ≥10 years. The median clinical f/u period was 5.7 years (range 2–156 months). The maximum f/u period was 13 years.

One patient underwent retransplantation during f/u due to biliary complications, without evidence for restenosis at the time of reoperation. Intraoperatively, the stent could be removed without interference with the new graft. One patient died of pneumonia 1 year after intervention.

Because of the variance in f/u length, primary patency rates were calculated at standardized time points with patients' adequate f/u data. We report an overall primary patency rate of 75% with rates of 100%, 90%, 91%, 77%, and 75% after 1, 3, 5, 7, and 10 years, respectively. For detailed patency rates, see Figure 3A.

Two primary patients with PTBA received secondary stents for recurrent PVS 8 and 15 months after initial angioplasty, respectively.

Ten patients received diagnostic catheterization for suspected in-stent stenosis during f/u; for detailed information on patients with restenosis, see Tables S2 and S3. Time to reintervention ranged from 20 to 126 months (mean/median 78.4/74.5 months) after the first PTSA. Restenosis was only evident in 5 of these 10 patients, resulting in reangioplasty. In two patients, a second stent was telescoped into the first one, and three patients could be treated with balloon angioplasty alone (Figure 1). The other five patients showed neither a reduction in PV diameter nor a pressure gradient but received prophylactic balloon angioplasty to adapt the stent diameter to somatic growth. As interventional catheterization was performed, all 10 patients were classified as assisted patency, independent of objective restenosis before reintervention.

All reinterventions could be carried out successfully without any complication, resulting in an overall assisted patency rate of 100%.

Besides the risk factors of young age, low body weight, and early posttransplantation PVS, all patients were systematically analyzed for the preexisting additional risk factors linked to fixed portal hypertension. Our cohort showed an almost even distribution with 13, 15, and 12 patients at low, moderate, and high risk, respectively. Most patients at low risk had received LT for acute liver failure without symptoms of chronic portal hypertension such as reverse portal flow or extensive collaterals. However, neither in the univariate nor in the multivariate Cox regression analysis did any of these parameters show a significant effect on primary patency (Figures 3 and S1, Table S4).

Reinterventions were performed in 3/27 patients with end-to-end anastomosis and 7/13 patients with atypical anastomosis; see Table S2. Although indicating a higher restenosis rate in patients with atypical PV anastomosis, low case numbers prevented statistical comparison of different types of PV anastomosis.

Technical and clinical success, complications of primary PTSA

The most reliable parameters for assessing the immediate effectiveness of angioplasty include the trans-stenotic pressure gradient and the diameter of the stenotic segment. In our cohort, both parameters improved significantly after PTSA in every patient securing markedly improved intrahepatic PV perfusion after intervention (Figure 2A).

Long-term success was evaluated based on clinical criteria such as platelet count, splenomegaly, PV flow, and portosystemic shunts. Comparable with the excellent hemodynamic success achieved, we report a significant clinical improvement after percutaneous PV intervention. Within the first months after PTSA, we noted a highly significant (p < 0.01) increase in the mean platelet count, comparable to clinical success data reported by Cheng et al. in a similar cohort (n = 11; 79,000 vs. 135,000/μl).^[27] At the same time, the diameter of the spleen decreased significantly by an average of 1.5 cm (p < 0.01). In our cohort, ascites was present in approximately one-third of the patients (12/40, predominantly in the subacute phase after LT) and gastrointestinal bleeding occurred in five children before intervention (11%), with two patients receiving blood transfusions. Neither of these symptoms recur postinterventionally.

Cheng et al. reported promising results after primary PTSA in a small cohort with 4 adult and 12 pediatric patients (mean age/weight 5.8 years/21 kg). The reported technical success was 68.8%.^[28] Ko et al.^[29] showed even better results in a small pediatric cohort (n = 12). Other authors reported a comparable reduction of trans-stenotic pressure gradient.^{[8, 10-13, 30]} However, most of these studies reported combined balloon and stent angioplasty data and PV diameters are often not reported. In a recently published systemic review of the literature, Sare et al.^[31] concluded that primary PTSA had technical success of 100%, which was significantly higher compared with balloon angioplasty. They clearly favored primary stent implantation as treatment of choice.

In two patients, moderate complications occurred during percutaneous intervention and resolved with standard conservative care. This reflects the safety of primary stent placement in an experienced pediatric catheterization laboratory, even in infancy. Other groups report comparable results.^{[10, 14, 27]}

Patency rate, reintervention, and risk for postinterventional restenosis

All PV stents remained patent until the last f/u visit. With 10/10 successful reinterventions (for detailed information, see Tables S2 and S3), we report an overall primary patency rate of 75% and an assisted patency of 100%. Because of the variance in f/u length, primary patency was calculated at standardized time points of 1, 3, 5, 7, and 10 years resulting in 100%, 90%, 91%, 77%, and 75%, respectively. In five patients, PV stents did not need any reintervention for >10 years (maximum 13 years). In the literature, patency rates of primary PTSA have been reported to be 90.9% (f/u 12 months)^[27] or 100% (f/u 10–58 months).^[29] This is in marked contrast to PTBA cohorts, where comparable patency could only be achieved with recurrent reinterventions with up to five additional balloon dilatations^[11] or secondary stent placement,^[26] respectively.

Early data by Funaki et al. in 1997 showed a 64% recurrence rate of PVS in 11 pediatric patients within 1–13 months (mean 6.3 months) after PTBA. Stent placement could achieve longer freedom from reintervention.^{[13, 23, 24]} More recent reports by Yabuta et al.^[11] and Patel et al.^[8] favoring primary PTBA showed improved early results with lower reintervention rates: 11/43 patients (mean age at intervention 4.1 years)^[11] and 13/41 (median age 1.7 years),^[8] respectively. However, in the first pediatric cohort, eleven patients underwent two, six patients needed three, and two received five interventions until patency was achieved.^[11] In the second group of smaller patients, the authors stated that just over half of their patients eventually required stent placement during f/u.^[26]

Based on these results from the literature, balloon angioplasty of PVS is generally burdened by high recurrence rates with the need for one or more reinterventions in <2 years of f/u. Shim et al. reported no significant difference in long-term outcome between both techniques.^[32] However, although their PTSA primary patency rate was 100%, three patients with balloon angioplasty needed reintervention. In our opinion, this reflects one of the key points of percutaneous intervention for PVS. Although technical success may be achieved regardless of the technique, balloon angioplasty has higher reintervention rates, making stent angioplasty the superior technique. This is in line with a recent systematic review of 243 pediatric patients: primary PTSA has higher technical success and lower restenosis rates (2.5% vs. 29.7%) compared with PTBA alone.^[31]

Notably, in our cohort, almost half of the patients (19/40) successfully received PTSA within 6 months after LT and 6 within 45 days after LT (PTSA n = 4; PTBA n = 2). PVS early after LT is often more elastic, and PTBA alone might be ineffective and more prone to a possible vessel tear at the anastomotic site. This is why in these patients, primary PTSA is recommended even in the subacute phase after transplantation to avoid early reoperation.^[29] Although both balloon and stent angioplasty are considered safe and effective treatment options for PVS after pediatric LT, most authors agree that stent insertion should be preferred in cases with potential need for early reintervention.^{[22, 23, 32]}

For intraoperative and immediate postoperative PV complications, hybrid strategies combining a surgical approach with catheter-based angioplasty have been successfully used as well.^[16-18] However, considering potential hemodynamic instability as well as worse imaging quality compared with a pediatric catheterization laboratory, these strategies should be reserved for bailout situations.

No published data exist on risk factors for recurrent stenosis after portal angioplasty for posttransplantation PVS treatment. Considering that low body weight and age at LT have been identified as risk factors of PV complications (Table 3),^{[2, 4, 21]} we hypothesized that the same factors would result in a higher rate of restenosis rate after intervention. Surprisingly, neither one of those factors nor symptoms of fixed portal hypertension could predict if or when a patient would need reintervention sooner than others (Figure 3).

Stent migration of self-expanding stents have been described in the literature as potential long-term complications after angioplasty of PVs.^[33] However, to the best of our knowledge, there is no report of stent migration after stent angioplasty with balloon expandible stents to treat posttransplantation PVS. In 15 years of interventional stenting of the PV, we did not observe any case of stent migration during f/u imaging.

Postinterventional medication may also influence primary patency of PV stents. Our early patients 15 years ago received acetylsalicylic acid monotherapy like other PTA patients. Over the years, the benefits of dual antiplatelet therapy in adult cardiologic interventions were adapted for pediatric angioplasty. In our center, we used a combination of clopidogrel and acetylsalicylic acid for patients considered at risk for postinterventional in-stent stenosis or thrombosis. Although there are no published data for primary stent placement, Sanada et al. even suggested a strategy of triple anticoagulant therapy with low-molecular-weight heparin, warfarin, and acetylsalicylic acid^[34] in their study group of pediatric patients with PTBA (age 1.1–19.8 years). In their study, no recurrent PVS was seen over a f/u of 9 months to 3.9 years with triple therapy. However, larger balloons were used for angioplasty in this group, suggesting a bigger vessel size and, therefore, a better general prognosis.

Other authors identified jump graft or conduit reconstruction of the PV as risk factors for posttransplantation PVS.^{[5, 35]} However, the type of anastomosis is usually determined by the available vasculature, such as short pedicles or extensive collateral vessels. In our cohort, reintervention was necessary in more (7/13) patients with atypical PV anastomoses. However, the small number of affected patients prevented further statistical analysis.

Our study has some limitations. First, this study was retrospective, so risk factors for PV complications could not be adequately analyzed and identified. Second, the data set is incomplete in some patients. Third, the low number of patients receiving PTBA did not allow for statistical comparison of PTBA and PTSA.