Secondary Logo

Journal Logo

Original Articles: Hepatology

Partial Splenic Embolization Is a Safe and Effective Alternative in the Management of Portal Hypertension in Children

Vittorio, Jennifer; Orellana, Katherine∗,†; Martinez, Mercedes; Ovchinsky, Nadia∗,‡; Schlossberg, Peter§; Griesemer, Adam||; Lobritto, Steven

Author Information
Journal of Pediatric Gastroenterology and Nutrition: June 2019 - Volume 68 - Issue 6 - p 793-798
doi: 10.1097/MPG.0000000000002332


What Is Known

  • There are multiple approaches to manage the clinical complications of portal hypertension to treat/prevent variceal bleeding and spontaneous hemorrhage by mitigating thrombocytopenia.
  • No single approach is ideal for all patients given the heterogeneity of this population.

What Is New

  • Partial splenic embolization is a safe, effective, semi-invasive alternative to splenectomy in patients with portal hypertension in select situations.
  • Partial splenic embolization can be successfully performed in pre-and post-liver transplant patients, and select cirrhotic and noncirrhotic pediatric patients.
  • Partial splenic embolization may be a favorable alternative to splenectomy and portal systemic shunting because it preserves functional spleen mass and avoids postprocedure accelerated liver disease or encephalopathy.

Portal hypertension (PHTN) may stem from a wide variety of conditions. It frequently presents with splenomegaly resulting in leukopenia, thrombocytopenia, and the formation of collateral circulation including esophageal varices. PHTN develops when there is resistance to portal blood flow and is aggravated by increased portal collateral inflow. Resistance occurs at different levels including prehepatic (ie, portal vein thrombosis), intrahepatic (eg, cirrhosis), or posthepatic (eg, Budd-Chiari syndrome). The combination of thrombocytopenia and significant esophageal varices predisposes to potential life-threatening hemorrhage (1,2).

Traditionally, complications of PHTN have been managed with serial endoscopic variceal ligation (EVL) or invasive open surgical procedures including splenectomy, portosystemic shunting or orthotopic liver transplantation (OLT) (3,4). EVL directly treats varices, but does not address the underlying etiology of PHTN and often requires repeated procedures with recurrent anesthesia exposure. Portosystemic shunting reduces the risk of hemorrhage by decreasing portal pressure. Surgical intervention, however, may exacerbate underlying liver disease resulting in accelerated hepatic dysfunction and increased morbidity and mortality. It may also increase the risk of subsequent liver transplant surgery. Mesoportal shunting (Rex shunt) has emerged as the preferred treatment for PHTN from extrahepatic portal vein thrombosis, because it restores trophic blood flow back to the liver. This procedure is not an option for many patients with chronic liver disease.

Partial splenic embolization (PSE) was first described by Maddison (5) in 1973 for the treatment of thrombocytopenia and variceal bleeding in patients with cirrhosis. It selectively disrupts the arterial blood supply at the level of the end arterioles of the spleen resulting in partial splenic infarction and decreased spleen size, decreased portal venous inflow, increased circulating platelet count, and decreased size of gastroesophageal varices. PSE preserves residual functional spleen mass while avoiding postprocedure acceleration of underlying liver disease or encephalopathy. PSE has been proposed as a safe and effective alternative in the management of pediatric PHTN (6,7). We present our experience with the use of this modality in different pediatric populations with PHTN to better understand the rationale for safe and appropriate application of this treatment.


Patient Selection

We conducted a retrospective chart review of all pediatric patients ages 0 to 21 that underwent PSE for the management of PHTN between January 2010 and August 2017 at NewYork-Presbyterian Columbia University Medical Center. This study was approved by the Columbia University Internal Review Board.

The decision to proceed with PSE was determined by the primary hepatologist and surgeon after careful discussion with both patient and family members regarding alternative approaches. We recommended PSE to patients with bleeding tendencies that were not optimal candidates for an open surgical shunt or repeated endoscopies. This includes the following patient populations:

  • 1. Patients with baseline compromised respiratory function (ie, cystic fibrosis [CF]).
  • 2. Patients with significant synthetic dysfunction that may progress to liver failure after shunting.
  • 3. Patients with early signs of encephalopathy that could potentially worsen with increased portal systemic shunting.
  • 4. Patients with abnormal right heart function that could develop cardiac decompensation or shunt failure.
  • 5. Patients deemed to be marginal candidates for repeated anesthesia.
  • 6. Patients with splenic vein thrombosis.
  • 7. Patients with characteristics predicting poor wound healing or other surgical complications.
  • 8. Patients with unfavorable anatomy, increased transhepatic resistance, or elevated right heart pressures to permit mesoportal shunting (Rex shunt).

We favor the Rex shunt whenever feasible to address PHTN from extrahepatic portal vein thrombosis given the theoretical advantage of restoring trophic blood flow to the liver.

Indications for PSE in patients with varices that never had endoscopic intervention included symptomatic thrombocytopenia with significant bruising/bleeding (epistaxis); ascites; gut edema leading to failure to thrive; discomfort secondary to massive splenomegaly; and CF or other conditions predisposing to high risk for surveillance endoscopies with recurrent anesthesia exposure.

Technical Approach to Partial Splenic Embolization

PSE was performed by interventional radiology under general anesthesia via percutaneous femoral artery approach under sterile conditions utilizing guidelines based on the Spigos technique to minimize complications (8). Diagnostic angiographies of the celiac axis and splenic arteries were obtained to determine the target splenic arterial branches, to identify accessory branches to other tissues from the splenic artery and to determine baseline splenic volume. Significant accessory branches to other tissues were coiled to prevent inadvertent tissue injury related to aberrant particles. The goal of an embolization session was to reduce the splenic volume by 60% to 70% from baseline. PSE was accomplished using Embospher particles (South Jordan, UT/USA) delivered into the distal splenic artery beyond the pancreatic branches. Selective embolization with Embosphere particles was performed until absence of flow was documented in the targeted branches by sequential angiogram (Fig. 1A and B).

Arteriograms of partial splenic embolization (PSE) (A) pre-embolization with capillary blush demonstrating splenomegaly and (B) postembolization with capillary blush revealing multiple peripheral defects; esophageal varices before (C) and after (D) PSE; magnetic resonance imaging (MRI) showing splenomegaly (E) and postembolization with necrotic areas surrounding the spleen.

Postprocedure Management

All patients were admitted to the hospital after PSE for further observation and management. Since significant splenic infarction was achieved we anticipated and observed post-procedure pain, fever, and perisplenic collections. Patients received patient-controlled analgesia for pain control and to prevent splinting and associated pulmonary complications such as atelectasis and pneumonia. All patients received broad-spectrum intravenous antibiotic prophylaxis, typically with piperacillin/tazobactam, before the procedure to prevent infection of infarcted splenic tissue. This was continued for 5 to 7 days. Antibiotics were converted to an oral regimen before hospital discharge. Some patients required prolonged use of antibiotics due to persistent fever and abdominal pain. To reduce the risk of future infection by encapsulated organisms, should the procedure lead to a complication necessitating total splenectomy, we administered vaccines for pneumococcus (PCV13 followed by PPSV23), Haemophilus influenzae type B and meningococcus at least 2 weeks before the procedure. Abdominal sonograms with Doppler analysis of hepatic and splenic vessels were performed postprocedure to assess vessel patency and the degree of intra-abdominal-free fluid. Criteria for hospital discharge included adequate control of abdominal pain with oral medications, adequate oral intake/hydration, and absence of postprocedure infection. The patients were followed-up in the outpatient hepatology clinic. Hematologic indices including white blood cell (WBC) and platelet counts were monitored. Imaging was repeated as indicated. Surveillance endoscopy was also performed 3 months after PSE to evaluate for persistent varices.


Mean values of continuous data were compared using Student t tests or nonparametrics as appropriate. Categorical variables were compared using the Chi-squared test. A P value <0.05 was considered statistically significant.


Patient Demographics

Twenty-six patients with PHTN underwent PSE due to thrombocytopenia and/or concern for variceal bleeding. The patients ranged in age from 18 months to 20 years (mean age 13.1 years) and approximately half of the patients were female. Etiologies of underlying PHTN were varied as detailed in Table 1. Fifteen patients had at least 1 prior episode of variceal hemorrhage. All patients with upper gastrointestinal bleed were admitted to the ICU and treated with volume resuscitation, octreotide infusion, antibiotics, and urgent endoscopy. Information regarding blood transfusions was not available for all patients. Four patients had documented epistaxis. CF patients with recurrent hemoptysis and thrombocytopenia underwent PSE for this indication. Twenty-two patients, including those with CF, were undergoing serial EVL under general anesthesia.

Patient demographics and etiology

Before PSE median WBC was 2.85 (×109/L) and median platelet count was 53.0 (×109/L). PSE was performed in 4 patients >3 years after OLT and 3 patients underwent PSE before eventual OLT (range 10–55 months). There was no difference in management, adverse events, or complications in patients who had previously undergone OLT as compared to nontransplanted patients. Of the 4 patients who had PSE after OLT, 3 were on low-dose tacrolimus and 1 had a history of post-transplant lymphoproliferative disorder and was off immunosuppression. The follow-up period after PSE ranged from 21 to 93 months.

Monitoring Indices

Median WBC increased from 2.85 to 5.68 × 109/L (P < 0.001) and median platelet counts rose from 53.0 to 142.0 × 109/L (P = 0.04) following PSE. Since platelet counts rise in the acute phase, we continued to monitor values longitudinally. There was a significant increase in long-term platelet counts with values >100 × 109/L following PSE in all but 5 patients. All patients had significant splenomegaly before PSE. Because the changes to the spleen occurred in 3 dimensions, repeat imaging was not used as a marker of successful PSE. CF patients all reported improved respiratory mechanics post-PSE presumably secondary to reduction in spleen size and avoiding the negative impact of additional anesthesia (Table 2).

Monitoring indices following partial splenic embolization

Three patients failed to achieve 70% reduction in splenic perfusion volume leading to an insignificant change in spleen size and long-term platelet count after embolization. The fourth patient with CF was the only patient who had an Amplatzer device (St. Paul, MN) used initially. This patient underwent repeated PSE using Embosphere that resulted in a favorable increase in platelets. The fifth patient also underwent a second PSE but rebled from varices despite having a good initial platelet response. There was no difference in spleen size or laboratory values noted before PSE in these patients to predict their outcome.

Resolution of Varices

All 26 patients were noted to have evidence of varices before PSE, 22 of which required endoscopic intervention (84.6%). Fifteen of these patients had documented episodes of variceal hemorrhage. Fifteen patients underwent repeat endoscopy at our institution after PSE (Fig. 1C and D). Three patients demonstrated complete resolution of varices. An additional 7 patients were noted to have a decrease in variceal grade. Two patients were noted to have recurrent variceal hemorrhage after PSE and subsequently underwent further intervention. The first did not have a sustained platelet response and required subsequent splenectomy. The second was the youngest patient in our cohort (18 months) and the only patient to have undergone sclerotherapy rather than EVL for management of variceal hemorrhage before PSE.

Adverse Events

Following PSE we commonly observed fever, atelectasis, abdominal pain, abdominal distention, nausea, and perisplenic fluid collections. Seven patients required readmission after initial discharge as detailed in Table 3(9). Two of these patients were discharged only 1 day after PSE and returned the following day with fevers and abdominal pain with an unremarkable diagnostic evaluation. Three patients developed pulmonary infiltrates, 2 with underlying CF. All 3 were treated successfully with antibiotics. There were no serious adverse events from splenic abscess, splenic rupture, pancreatic infarction, sepsis, or death.

Partial splenic embolization–related adverse events

In regards to infection, 1 patient was readmitted with bacteremia 1 day after discharge secondary to Staphylococcus warneri treated with a change in oral antibiotic from amoxicillin/clavulanate to levofloxacin after sensitivities were reported. Splenic function appeared to be preserved during years of follow-up in all patients without increased incidence of infection. No patients developed infection with encapsulated organisms and all patients were vaccinated before PSE.

One patient in our study developed extensive thrombosis of her splenic vein, portal vein, and IVC at the level of the renal veins necessitating thrombectomy and urgent splenectomy 9 days after PSE. This patient had normal transhepatic pressures with massive debilitating splenomegaly. Laboratories before PSE showed a WBC of 1.4 (×109/L) and platelets of 153 (×109/L). The procedure was performed for compassionate reasons and an estimated 80% of the spleen was infarcted in a single procedure. This patient was found to have a genetic predisposition to myelodysplastic-like syndrome and thrombocytosis. Myelodysplastic-like syndrome has been associated with both thrombotic and hemorrhagic complications (10). Platelet levels peaked at an astonishing 2484 (×109/L) in this patient after PSE necessitating chronic antiplatelet therapy.

Further Intervention

Following PSE, 3 patients underwent successful liver transplantation. The mean time from PSE to OLT was 29.5 months (range 10.1–55.4 months). Indications for OLT included worsening liver synthetic function and concern for hepatocellular carcinoma in a patient with underlying cirrhosis. The indications for OLT were not related to hypersplenism. Four additional patients required a surgical shunt procedure at mean 19.4 months after PSE (range 7.56–27.72 months). The indication for surgical shunt in 3 patients was persistent thrombocytopenia and recurrent variceal hemorrhage in the fourth patient. Splenectomy was performed for recurrent variceal hemorrhage in 1 patient after PSE.


Complications of PHTN such as hypersplenism, variceal hemorrhage, and ascites can lead to significant morbidity and mortality in children. Surgical interventions include shunting or splenectomy; however, overwhelming sepsis postsplenectomy is a pressing concern (11,12). Total splenic embolization has been abandoned due to the high incidence of complications such as splenic abscess, splenic rupture, sepsis, splenic vein thrombosis, and pneumonia. PSE may offer a less invasive, safe, and effective alternative to surgical splenectomy. Immunologic integrity is preserved since a portion of splenic parenchyma remains. PSE addresses portal hypertension regardless of etiology as definitive therapy. Patients with poor synthetic function and significant PHTN should be transplanted. In patients with adequate synthetic function PSE can extend the time to transplant, obviate the need for transplant, and avoid splenectomy or open surgical shunting. Since there is no increase in portosystemic shunting following PSE, the incidence of postprocedure-accelerated liver disease and encephalopathy are reduced.

Most adverse effects following PSE are expected and can be managed conservatively. In our cohort we commonly observed fever, abdominal pain, and leukocytosis without evidence of infection. Pain was effectively controlled with intravenous and oral analgesics. Nonsteroidal anti-inflammatory drugs were avoided to minimize hemorrhage. Seven patients required readmission; however, 2 patients were discharged soon after PSE (1–2 days) and returned for evaluation of fevers and abdominal pain not present at time of discharge. Potential complications after PSE include splenic abscess, splenic rupture, and pancreatic infarction. There have been reports of death in adults after PSE from complications of severe hepatic insufficiency, complete splenic infarction with abscess, total portal vein thrombosis, and cardiac insufficiency (13–15). In our experience, prophylactic antibiotics and limiting splenic reduction to no >70% have avoided serious morbidity and mortality.

One patient in our study developed a severe adverse event including extensive thrombosis after PSE necessitating thrombectomy and urgent splenectomy. Since then any patient with splenomegaly and normal platelet count is excluded from PSE and referred for hematologic work-up. In addition, all patients with a prior history of portal vein thrombosis should have a hypercoagulable evaluation before PSE. In patients with massive splenomegaly we favor a staged approach to the embolization goal to assess response and ensure safety. Embolization is performed sequentially when indicated to achieve a target reduction in splenic volume.

Splenic regeneration has been reported in children after PSE (7). Prior case series of PSE in children limiting splenic embolization to 70% of initial volume showed that 75% of patients maintained platelet counts >100,000 for more than five years (6). Recurrence of thrombocytopenia appeared to be dependent on splenic regenerative rather than the volume of spleen infarcted. Sangro et al (16) found that when <50% of the spleen was embolized, there was lower morbidity but a higher incidence of recurrent hypersplenism. Similarly, 3 of our patients with recurrent thrombocytopenia (<100,000) had <50% of initial splenic volume infarcted. Repeat embolization in one that achieved the 70% reduction goal still progressed to thrombocytopenia and eventual splenectomy suggesting a role of splenic regeneration.

Other studies support the use of PSE in particular subsets of patients. An adult study applied PSE in 26 critically ill patients with thrombocytopenia and bleeding esophageal varices and showed a significant improvement in leukocyte counts, hemoglobin values and platelet counts with a corresponding decrease in bleeding (17). None of our patients were actively bleeding at the time of PSE. We selected patients with bleeding tendencies that were not necessarily optimal candidates for an open surgical shunt as outlined above.

After identifying select populations likely to benefit from the PSE approach we tried to examine if there were variables that would predict outcomes after PSE. We could not identify pre-procedure parameters that predicted outcome. In particular, there was no significant difference in the preprocedure platelet count between patients with and without recurrent thrombocytopenia. After PSE we noted a significant increase in platelets but to a lesser degree in patients who had recurrent hypersplenism. This would suggest that elective, repeat PSE may be needed in patients with limited initial results. A more extensive look at larger populations of patients may yield the power to identify factors favoring one intervention over another.

One question that frequently arises regarding PSE is the ideal volume of spleen to be targeted for infarction. Most studies seem to favor 60% to 70% reduction in perfusion volume. The desired outcome from decreased splenic inflow and subsequent outflow would be reduction in splenic sequestration and reduced portal pressures to prevent or mitigate variceal bleeding and ascites formation. Both improvement of hematologic indices and severity of complications appear to correlate with the amount of tissue infarcted. Previous studies have demonstrated that 60% to 70% embolization preserves immunologic function of the spleen and antegrade flow in the splenic vein (18). In addition, several studies have documented a higher incidence of complications when >70% of the spleen had been embolized in cirrhotics (16). One adult study observed that abdominal pain and fever lasted longer and required more pain medication in patients who had > 50% splenic perfusion volume embolized (19). We noted similar findings in our patient population. Severity of symptoms including fever, abdominal pain, and leukocytosis appeared to directly correlate with improvement in platelet count. Care should be taken, however, to avoid embolizing >70% of the spleen at 1 time. In the case of massive splenomegaly, perhaps performing sequential embolizations would avoid potential serious complications.

Another factor to consider is whether more than 1 approach to the treatment of PHTN could yield better outcomes than single interventions. Indeed, Xu et al (20) demonstrated that the combination of banding and PSE in portohypertensive patients decreased portal vein flow rate and maximal velocity and improved esophageal varices and hypersplenism. In our patients with esophageal varices, the combination of endoscopic ligation and PSE has been successful in controlling the progression of varices and decreasing the incidence of variceal hemorrhage. Inflow to the portal vein from the intestine is not affected by PSE and the physiology would mimic that of a distal splenorenal shunt with selective decompression of the splenic axis but preserved liver perfusion with lower overall portal pressures.

There are no current reports regarding the immunologic consequences of PSE and the role of vaccination. Splenectomy impairs the body's ability to produce antibodies against encapsulated microorganisms and predisposes patients to sepsis. The Centers for Disease Control and Prevention recommends immunoprophylaxis against Streptococcus pneumoniae, H influenzae and Neisseria meningitidis in all splenectomized patients (21). There are no standard guidelines as to when the vaccines should be administered, but common practice has been to administer immunoprophylaxis at a minimum of 2 weeks before elective splenectomy and 2 weeks after the procedure for emergency splenectomy. We chose to give our patients pneumococcal (PCV13 followed in 1 month by PPSV23 when feasible), H influenzae type B and meningococcal vaccines at least 2 weeks before PSE. To date we have not observed any significant infections in our cohort.

Limitations of our study include the small number of cases in our series and the retrospective design. Our patient population was heterogeneous compared with most pediatric publications that used PSE in a predominate biliary atresia, portohypertensive cohort. Although limited by small numbers we suggest that particular subpopulations can achieve better outcomes supporting a personalized approach in most cases. In addition, we have not performed any elegant studies of splenic function to substantiate our clinical observation of continued immunologic integrity post-PSE. Perhaps with advances in immunologic assays we can demonstrate this more scientifically going forward.

In conclusion, our study supports that PSE is a safe, effective, semi-invasive alternative to splenectomy in pediatric patients with PHTN. The combination of EVL and PSE is effective treatment for esophageal and gastric varices (not amenable to simple band ligation). Reduced splenic volume and splenic venous return effectively decreased portal flow and pressure. PSE can be successfully performed in pre- and post-transplant transplant patients, and select cirrhotic and noncirrhotic pediatric patients. PSE appears to preserve functional spleen mass and avoids postprocedure accelerated liver disease or encephalopathy. In contrast to shunting, PSE can be effective even in the setting of elevated right-sided heart pressures. Further multicenter collaborative studies with larger and varied patient populations will need to be performed to substantiate our approach and better define patient populations that would benefit most from PSE, sequential PSE or combined interventions. Liver transplantation remains the ideal intervention for anyone with significant decompensated PHTN or advanced hepatic dysfunction (Supplemental Table 1, Supplemental Digital Content,


1. Ling SC. Portal hypertension in children. Clin Liver Dis 2012; 1:139–142.
2. Feldman AG, Sokol RJ. Noncirrhotic portal hypertension in the pediatric population. Clin Liver Dis 2015; 5:116–119.
3. D’Amico G, Pagliaro L, Bosch J. The treatment of portal hypertension: a meta-analytic review. Hepatology 1995; 22:332–354.
4. Mileti E, Rosenthal P. Management of portal hypertension in children. Curr Gastroenterol Rep 2011; 13:10–16.
5. Maddison F. Embolic therapy of hypersplenism. Invest Radiol 1973; 8:280–281.
6. Nio M, Hayashi Y, Sano N, et al. Long-term efficacy of partial splenic embolization in children. J Pediatr Surg 2003; 38:1760–1762.
7. Kumpe DA, Rumack CM, Pretorius DH, et al. Partial splenic embolization in children with hypersplenism. Radiology 1985; 155:357–362.
8. Spigos DG, Jonasson O, Mozes M, et al. Partial splenic embolization in the treatment of hypersplenism. AJR Am J Roentgenol 1979; 132:777–782.
9. Basch E, Reeve BB, Mitchell SA, et al. Development of the National Cancer Institute's patient-reported outcomes version of the common terminology criteria for adverse events (PRO-CTCAE). J Natl Cancer Inst 2014; 106:pii: dju244.
10. Landolfi R, Di Gennaro L. Thrombosis in myeloproliferative and myelodysplastic syndromes. Hematology 2012; 17 (suppl 1):S174–S176.
11. Eraklis AJ, Filler RM. Splenectomy in childhood: a review of 1413 cases. J Pediatr Surg 1972; 7:382–388.
12. Cooper MJ, Williamson RC. Splenectomy: indications, hazards and alternatives. Br J Surg 1984; 71:173–180.
13. N’Kontchou G, Seror O, Bourcier V, et al. Partial splenic embolization in patients with cirrhosis: efficacy, tolerance and long-term outcome in 32 patients. Eur J Gastroenterol Hepatol 2005; 17:179–184.
14. Wholey MH, Chamorro HA, Rao G, et al. Splenic infarction and spontaneous rupture of the spleen after therapeutic embolization. Cardiovasc Radiol 1978; 1:249–253.
15. Hayashi H, Beppu T, Okabe K, et al. Risk factors for complications after partial splenic embolization for liver cirrhosis. Br J Surg 2008; 95:744–750.
16. Sangro B, Bilbao I, Herrero I, et al. Partial splenic embolization for the treatment of hypersplenism in cirrhosis. Hepatology 1993; 18:309–314.
17. Palsson B, Hallén M, Forsberg AM, et al. Partial splenic embolization: long-term outcome. Langenbecks Arch Surg 2003; 387:421–426.
18. Madoff DC, Denys A, Wallace MJ, et al. Splenic arterial interventions: anatomy, indications, technical considerations, and potential complications. Radiographics 2005; 25 (suppl 1):S191–S211.
19. Elmonem SA, Tantawy HI, Ragheb AS, et al. The outcome of partial splenic embolization for hypersplenism in the cirrhotic patients. Egypt J Radiol Nucl Med 2011; 42:35–42.
20. Xu RY, Liu B, Lin N. Therapeutic effects of endoscopic variceal ligation combined with partial splenic embolization for portal hypertension. World J Gastroenterol 2004; 10:1072–1074.
21. Recommendations of the Advisory Committee on Immunization Practices (ACIP): use of vaccines and immune globulins in persons with altered immunocompetence. MMWR Recomm Rep 1993; 42:1–18.

gastroesophageal varices; pediatrics; splenomegaly; thrombocytopenia

Supplemental Digital Content

Copyright © 2019 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition