See “When Should We Perform TIPS in Children?” by Rosenthal on page 577.
The management of PH in adults is primarily directed at preventing and treating gastrointestinal bleeding and ascites. The therapeutic choices are represented mainly by medical therapy, endoscopic treatment of varices, shunting procedures, and liver transplantation (LTX) (1). Among the shunting options, transjugular intrahepatic portosystemic shunt (TIPS) has become a useful and effective tool (2,3).
According to the most recent consensus on PH, persistent bleeding despite combined pharmacological and endoscopic therapy is best managed by placing an expanded polytetrafluoroethylene (e-PTFE)-covered stent to create a TIPS, often serving as a bridge to transplantation (4). There is a lack of these data in children (1,5,6).
In children, this procedure has been considered technically demanding and to offer short-term benefits, so its feasibility and efficacy have not been adequately challenged (7). Studies on TIPS in paediatric subjects are scant and only a few cases with a short follow-up have been published (8–10). Our aim was to challenge the idea that TIPS is not valuable in children; we therefore report a retrospective analysis of technical and clinical results of TIPS carried out in children studied in our centre to review its feasibility and efficacy achievable in young subjects, following skills improvement and newly developed tools in this field.
Patients were selected from those referred to the Paediatric Liver, Gastrointestinal, and Transplantation Unit, Ospedali Riuniti, Bergamo (Italy), which is the largest referral centre for paediatric hepatology and transplantation in Italy, between 2005 and 2010. In this period, 191 paediatric liver transplants (78% with split livers) have been performed at our institution and 62 children underwent a Kasai portoenterostomy for biliary atresia. All of the available medical records, including laboratory investigations, radiological imaging and procedures, surgical interventions, transplantation reports, and discharge letters, were reviewed.
Patient Selection and Features
Children were considered eligible candidates for TIPS placement if they developed one or more significant complications of PH, including gastrointestinal bleeding or ascites, unresponsive to medical and endoscopic management. The protocol followed at our centre for children with PH is the following:
1. Children with PH underwent a first visit diagnostic workup comprising blood tests (full blood cell count, liver and renal function, coagulation, blood ammonia), imaging (Doppler ultrasound, contrast-enhanced computerized tomography angiogram if indicated), and oesophagogastroduodenoscopy.
2. Follow-up visits were scheduled every 4, 6, or 12 months according to the clinical status, the severity of PH, and the course of the disease.
3. Patients who bled or developed large varices underwent endoscopic eradication (sclerotherapy in the first 12 months of age and banding ligation in children older than 1 year) and were given propranolol.
4. Consideration for shunting was given to patients on β-blockers with relapsing bleeding or relapsing large varices following endoscopic eradication.
TIPS Placement and Follow-up
In all of the eligible patients, the procedure included the transjugular approach; puncture of the portal vein (PV); portal venogram; measurement of a portosystemic gradient (PSG) before TIPS; shunt placement; and measurement of PSG after TIPS (Table 1).
Absolute contraindications for the TIPS procedure included clinical cholestasis or coagulopathy with an international normalised ratio (INR) >2 because we thought that the long-term benefits of TIPS were minimal if there was already severe coagulopathy. The patient characteristics are listed in Table 2.
All of the procedures were performed under general anaesthesia by the same interventional radiologist in an angiographic unit. Platelets were routinely administered if the blood count was <50 × 10−9 L. The procedure was carried out with ultrasound guidance via 5-F sheaths inserted into the right internal jugular vein.
A 5-F vertebral angiographic catheter and an angled wire with hydrophilic coating (Radiofocus-Glidecath, Terumo Europe N.V. 3001, Leuven, Belgium) were advanced into the hepatic vein. The guidewire was then replaced with the Amplatz Ultra Stiff Wire Guide (THSF-35-180-AUS2, Cook Europe, Bjaeverskov, Denmark). To perform the transhepatic PV puncture, a 19-gauge standard Colapinto needle was used in 2 patients. Because of its inadequate stiffness, it was replaced with a 5.2-F needle catheter of RUPS-100 Rosch-Uchida Transjugular Liver Access Set (Cook Inc, Bloomington, IN). During the PV puncture, a second operator performed a continuous ultrasound monitoring using a 3.5-MHz probe in a semisagittal plane via a lateral intercostal approach. The aim was to visualise the needle tip and the PV branches simultaneously to ease PV puncture. After PV access was gained, a steerable guidewire (0.035-inch diameter, Terumo, Tokyo, Japan) was advanced followed by a 5-F catheter within the PV.
A portal venogram in an anterior and oblique projection was carried out to localise the insertion into the PV to exclude leakage. The presence of competitive shunts, such as a spontaneous splenorenal shunt, was evaluated.
The PSG was measured and the shunt was not placed if it was <12 mmHg. The hydrophilic guidewire was subsequently replaced with a 180-cm superstiff wire (0.035 inch, Amplatz wire, Boston Scientific, Maple Grove, MN) and the parenchymal tract was predilated with 5- to 8 mm/4 cm PTA balloons (Fox Abbott Vascular Devices Amsterdam, the Netherlands).
To create a portosystemic shunt between a hepatic vein and the PV, a self-expandable nitinol Memotherm stent (Bard Angiomed Ltd, Crawley, UK) and a Viatorr e-PTFE covered stent graft (W.L. Gore, Flagstaff, AZ) were used.
When a hepatic vein remnant was not noted, the PV was punctured directly from the retrohepatic inferior vena cava (IVC) under ultrasound guidance and the shunt was placed between the PV and IVC. In patients with portal cavernoma (PVC), the shunt was implanted between a large collateral vessel and a hepatic vein.
Initial stent dilatation was carried out with a 6-mm diameter balloon followed by gradual overdilation to lower PSG. Finally, the stent was dilated to 7 to 8 mm according to patient size, anatomical structures, and PSG pressure.
Venography was carried out after stent placement and pressures were measured to confirm the correct stent positioning, the normal flow through the shunt, and the degree of PSG reduction.
If residual varices were identified, then embolization of gastric and oesophageal varices was not performed. After the procedure, the patients were transferred to the intensive care unit for a 6-hour observation period. Oral aspirin at 3 mg · kg−1 · day−1 was given to prevent shunt thrombosis and maintained for 6 months unless the platelet count was <50 × 10−9 L. A Doppler ultrasound was carried out within 48 hours to evaluate shunt patency.
After TIPS placement, all of the patients were treated and looked after at our centre. Follow-up assessments were performed by means of physical examination, laboratory investigations (including serum ammonia, transaminases, serum bilirubin and albumin, platelet count, and INR), and abdominal Doppler ultrasound.
After discharge, patients were reviewed every 3 months for 1 year, then every 6 months, or whenever new symptoms arose. Ultrasound Doppler evaluation included the patency of the device, the measurement of maximal peak flow velocity, and the evaluation of the direction of flow within the shunt and in the PV. Shunt stenosis or thrombosis was suspected if one or more alterations were noted, including absent flow inside the shunt, low peak shunt velocity (<50 cm/second), high peak shunt velocity (>180 cm/second), low mean PV velocity (<30 cm/second), and significant change in shunt velocity (>50 cm/second) compared with the immediate postprocedural result. Transjugular portal venography was carried out if the patient developed new complications of PH or an ultrasound evaluation highlighted shunt dysfunction. We defined the follow-up period as the time span from the first day of TIPS insertion to the last follow-up visit or transplantation.
Data are reported as medians and ranges unless specified differently. We used the Student t test to evaluate significant changes in the mean portal pressure gradients and the laboratory data before and after TIPS. All of the analyses were performed in Microsoft Excel and the statistical significance was established at P ≤ 0.05.
Thirteen children (9 girls, 4 boys), median age 9.8 years (range 2.2–18.4 years), were selected for the procedure, of whom 4 had received an LTX because of nonsyndromic bile duct paucity, congenital hepatic fibrosis, Alagille syndrome, and biliary atresia (1 each).
The underlying condition causing PH was noncirrhotic PV thrombosis (PVT) in 3 patients (of whom 1 after LTX), ductal plate malformation in 2 (congenital hepatic fibrosis and cerebellar vermis hypoplasia/aplasia, oligophrenia, ataxia, coloboma, and hepatic fibrosis syndrome), Budd-Chiari syndrome (BCS) in 3 (caused by polycythemia vera, antiphospholipid syndrome, and graft outflow obstruction), veno-occlusive disease after LTX, sclerosing cholangitis, cystic fibrosis, intestinal failure–associated liver disease in chronic intestinal pseudo-obstruction syndrome, and chronic graft dysfunction (1 each). No patient had decompensated liver disease at the time of TIPS insertion (Table 2).
Indications for TIPS placement were recurrent gastrointestinal bleeding in 8 of 13 patients (61.5%) and refractory ascites in 5 of 13 (38.5%), all of which were unresponsive to medical and endoscopic treatment. The source of gastrointestinal bleeding was oesophageal in 5 of 8 patients (62.5%) and occult in 3 of 8 patients (37.5%). Eight patients had splenomegaly and hypersplenism with a platelet count <100 × 10−9 L in 6 of 8 (75%), and <50 × 10−9 L in 2 of 8 (25%) patients (Table 2).
TIPS placement was attempted in 13 patients, but in 1 with BCS (patient 12), the shunt was not placed because PSG was low (5 mmHg) because of intrahepatic collateral veins. Shunt placement was successfully performed in 11 of 12 (92%) and failed in 1 patient with PVT and PVC after LTX (patient 13).
Eleven patients (7 girls, 4 boys) with a median age of 9.8 years (range 2.2–18.4 years) and a median weight of 30.0 kg (11.5–96 kg) had a TIPS insertion. Two patients (patients 3 and 4) were on the waiting list for transplantation (liver-lung and multivisceral transplantation, respectively); 3 (patients 5, 7, 11) had undergone an LTX with 3 split-livers (2 left lateral segments and 1 right lobe), and the remainders were not active candidates for transplantation (Table 2). Five children (patients 1, 6, 7, 11, 12) were receiving anticoagulant therapy (heparin in 3 patients and aspirin in 2) before the procedure.
Access to the liver was achieved through the right internal jugular vein in 10 patients and left internal jugular vein in 1 (patient 1). In 9 of 11 patients, PV puncture was performed 1 to 2 cm distal to the PV bifurcation. In 10 of 11 patients (91%), 11 e-PTFE-covered stents were used to create a portosystemic shunt. All of the patients had 1 stent placed apart from patient 6 who had BCS and required 2 overlapping stents to create a shunt from PV to IVC. Only 1 patient received an uncovered Memotherm Bard stent. A total of 12 stents were used in 11 patients.
In 2 patients (patients 8 and 10), the shunt was placed from a hepatic vein to a large collateral vessel of a PVC. Stents ranged between 7 and 8 mm in diameter in all of the patients. The median duration of the procedure was 2.07 hours (1.2–2.2 hours) (Table 3).
Efficacy of TIPS
PSG after shunt placement dropped from a median of 23 (16–35) to 10 mmHg (5–15 mmHg) (P < 0.01) (Fig. 1). No patient had complications related to the procedure. Soon after shunt placement, 3 children had elevated blood ammonia levels that decreased after administration of a course of oral antibiotics and lactulose. They had no clinical signs of hepatic encephalopathy; on closer inspection, they were perfectly oriented in time and space.
Complications of PH disappeared completely in 8 of 11 patients (73%), partially in 2 (18%), and persisted in 1 (9%) (Table 4). All of the 7 children with a history of gastrointestinal bleeding had no recurrence of bleeding after TIPS placement. TIPS placement improved ascites in 3 of 4 patients (75%). In 1 case (patient 6), ascites disappeared completely and in 2 cases (patients 1 and 7), it decreased significantly. In 1 patient (patient 11), ascites did not change, although the shunt was perfectly inserted and PSG reduced down to 8 mmHg.
Splenomegaly and hypersplenism did not improve after TIPS insertion. The median platelet count before and after TIPS was 90 × 10−9 and 118 × 10−9 L, respectively (P = 0.93).
During the follow-up period, blood tests (including bilirubin, transaminases, albumin, INR) revealed stable liver function in 8 patients and worsening in 3 (patients 1, 2, 11) who eventually underwent LTX because of progressive coagulopathy.
Median ammonia levels before and after the procedure were 42 (28–96) and 75 μmol/L (35–100), respectively (P < 0.05) (Fig. 2). Nevertheless, none of the patients developed overt encephalopathy despite free diet and no prophylactic treatment.
Shunt dysfunction was suspected by ultrasound follow-up examination in 3 of 11 patients (27%). One of them had progression of oesophageal varices (patient 3), 1 had worsening ascites (patient 7), and 1 no symptoms (patient 6). A total of 3 reinterventions, 1 in each patient, were performed under general anaesthesia 18 months (patients 3 and 6) and 3 months (patient 7) after TIPS insertion, respectively. Two patients (patients 3 and 6) had a shunt thrombosis resolved by balloon angioplasty and placement of a new covered stent inside the recanalised shunt. One patient (patient 7) had shunt stenosis treated by balloon angioplasty, but PSG remained high (18 mmHg) and an additional covered stent was placed to reduce the PSG to 12 mmHg. Ascites resolved after shunt revision.
Patients who underwent TIPS placement have a median follow-up period of 1.2 years (0.2–5.7). Seven of them (patients 3, 5, 6, 7, 8, 9, 10) have still a patent shunt after a median of 20.4 months (range 7–67 months); 4 (patients 1, 2, 4, 11) eventually underwent transplantation (liver in 3, multivisceral in 1) after a median of 6 months (1.5–33 months), with a patent shunt at transplant surgery. No patient had technical complications related to the procedure in the long term nor clinical events after TIPS placement (Table 4).
In paediatrics, noncirrhotic PH is the first cause of long-term PH and gastrointestinal bleeding (6). Some children with compensated cirrhosis may be sent to transplantation before the development of end-stage liver disease because of severe PH that is unresponsive to medical and endoscopic treatment (11). Thus, many children with liver disease have complicated PH but no strict indication to transplantation either in the short term or ever. Nevertheless, TIPS is often not considered in this setting, probably because of the conviction of short-term benefits and patency; the relatively larger organs pool available for children as compared with adults probably leads to lower the threshold to direct paediatric patients with severe PH towards transplantation. Newly developed stents have been shown to maintain patency in the long term; concomitantly, the number of available organs for paediatric recipients is decreasing. These arguments should lead one to consider TIPS indicated in children as it is in adults.
To date, >1000 adults have been enrolled in multiple controlled trials on the use of TIPS, and the overall success rate is >90% in many series (2,3). The Society of Interventional Radiology developed guidelines for the creation of TIPS in which they reported that the technical success (including both creation of the shunt and a decrease in PSG to <12 mmHg) should be achieved in 95%, whereas clinical success (resolution of PH complication) in 90% of cases (3). We found that it is possible to reach the same technical and clinical success rate in children as it is in adults.
Another group of children who can take advantage of TIPS are those who had an LTX and developed PH because of vascular complications or chronic rejection; in our own experience, these patients often bleed from nonoesophageal occult varices and are difficult to manage. In our series of 13 children evaluated and 11 treated with TIPS, 4 patients had already received LTX, a condition that probably makes TIPS insertion more challenging; 1 patient had been put on the waiting list for multivisceral transplantation and 7 had no indications for transplantation. TIPS cured ascites and gastrointestinal bleeding in all but 1 patient, with a median follow-up of those not requiring LTX of 20 months. We therefore used this procedure successfully both as a bridge to transplantation and as a tool to control PH in the relative long term.
Despite good control of ascites and bleeding, splenomegaly and hypersplenism did not change after TIPS. TIPS is not an efficient treatment option to improve splenomegaly and hypersplenism, at least in the short term (10,12).
A large number of technical difficulties and complications can occur during and after TIPS insertion, and this can be another reason to discourage the performance of TIPS in young subjects. Puncture of the PV is the most time-consuming portion of the procedure and often, many attempts are necessary (7,9). To reduce complications related to puncture site, we used an ultrasonographic guidance in all of the patients. Suitable miniaturized puncture sets for TIPS in children are not available, making the procedure more difficult and time consuming. In our experience, we considered a 19-gauge standard Colapinto insufficiently stiff to perform transhepatic PV puncture, and after 2 procedures, it was replaced by a 5.2-F needle catheter of RUPS-100 Rosch-Uchida puncture set, which is widely used in adults. TIPS after LTX can be more challenging than in a native liver, especially in children who commonly receive a left lateral segment graft as a result of anatomical differences, including the absence of the PV bifurcation, which is a landmark during portal venography, and the presence of a single hepatic vein (13,14). Nevertheless, the strategy we adopted allowed us to carry out the procedure in a 25-month-old patient weighing 12.5 kg who had a left lateral segment graft (patient 11) (Fig. 3), being the youngest transplated patient who had a TIPS reported so far (9–12). We also carried out the procedure successfully in 2 of 3 patients believed to have PV thrombosis and cavernoma who were not amenable to be cured by a mesoportal bypass. PVC is not a contraindication for TIPS placement, but it often reduces the success rate because sometimes a large collateral vessel in which to insert the shunt is not present (15). TIPS can be a good alternative to shunt surgery in children with severe complications of PV thrombosis and an anatomy unsuitable for mesoportal bypass.
Soon after the shunt was inserted, 3 patients had transient hyperammonemia, without overt encephalopathy, which was treated with a short course of medical therapy (antibiotics and lactulose). Development of encephalopathy is a major concern after a portosystemic shunt is carried out. The reported incidence of encephalopathy in adult patients who underwent TIPS is 30% to 40% (16). In our patients, median ammonia levels increased after TIPS placement, but none of them developed overt encephalopathy during the follow-up period, and the parents did not report signs or symptoms suggesting clinical encephalopathy. Encephalopathy is related mainly to the final shunt diameter; the larger the shunt, the higher the risk of hepatic encephalopathy, and therefore further adjustments to the diameter of the TIPS may be used to balance the significant reduction of PSG with the risk of bouts of encephalopathy (17–19).
In our study, shunt patency was maintained in all of the patients, of whom 7 had a patent shunt at last check-up and 4 up to LTX with a median follow-up of 20 months and 6 months, respectively, and the longest patency time of 67 months. This remarkable long-term patency is probably related to the use of an e-PTFE-covered stent graft that reduces bile and mucin permeation and clearly improves shunt preservation as compared with bare stents. In a large series of 157 adults treated with Viatorr covered stent grafts, the long-term patency was vastly improved compared with 316 patients who had uncovered stents (20). In other studies, it has been demonstrated clearly that TIPS carried out with an e-PTFE-covered stent has better long-term results (21). Few studies have been reported on the use of a covered stent graft in children. Four children with PH complications were successfully treated with such devices and showed clinical and biochemical improvement in all and recurrent stenosis in 1 (22).
Usually, this kind of stent is considered inappropriate for infants and young children because there are few available sizes and the shortest Viatorr-Gore device is presently 4 cm in length (covered segment), which is considered too long for young subjects (7). We successfully placed 11 Viatorr e-PTFE-covered stent grafts in 10 patients; 2 of them were younger than 5 years and weighed <15 kg.
A drawback of TIPS is the high rate of shunt dysfunction, reported in many studies, which can be caused by either thrombosis or hyperplasia of the intima (23). We observed shunt dysfunction in 3 of 11 patients (27%), which is a lower rate than that reported in previous studies (9–12). Two patients with shunt thrombosis and 1 had shunt stenosis were successfully treated with angioplasty and restenting, for a total of 3 reinterventions.
In conclusion, our results suggest that TIPS is feasible, safe, and effective in children with PH unresponsive to medical and endoscopic treatment. We believe that, after careful and cautious consideration of alternatives, TIPS should be considered as a valuable option for the treatment of PH complications in children, both in native livers and in transplanted grafts, as a bridge to transplantation and for long-term management. There are probably several paediatric candidates who can benefit from a procedure that should become part of the armamentarium available to manage children with PH.
We thank Prof Piero Amodio for reviewing this article and providing advice on discussion organisation.