Author + information
- Received January 7, 2017
- Revision received March 31, 2017
- Accepted April 4, 2017
- Published online June 12, 2017.
- Maxim Itkin, MDa,∗ (, )
- David A. Piccoli, MDb,
- Gregory Nadolski, MDa,
- Jack Rychik, MDc,
- Aaron DeWitt, MDc,
- Erin Pinto, MSN, RNa,
- Jonathan Rome, MDc and
- Yoav Dori, MD, PhDa
- aCenter for Lymphatic Imaging and Interventions, Children’s Hospital of Philadelphia/Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania
- bDivision of Gastroenterology, Hepatology and Nutrition, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania
- cDivision of Cardiology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania
- ↵∗Address for correspondence:
Dr. Maxim Itkin, Center for Lymphatic Imaging and Interventions, Hospital of University of Pennsylvania, 3400 Spruce Street, Philadelphia, Pennsylvania 19104.
Background Protein-losing enteropathy (PLE), characterized by loss of proteins in the intestine, is a devastating complication in patients with congenital heart disease. The cause of PLE is unknown, but lymphatic involvement has been suspected.
Objectives The authors evaluated the use of lymphangiographic imaging and liver lymphatic embolization as a treatment for PLE.
Methods This was a single-center, retrospective review of imaging and interventions used in 8 consecutive patients with liver lymphatic embolization and congenital heart disease with elevated central venous pressure complicated by PLE.
Results Liver lymphangiography was performed in 8 patients (5 males, 3 females; median age, 21 years), 7 of whom demonstrated leakage of liver lymph into the duodenum through abnormal hepatoduodenal lymphatic communications. This was confirmed by duodenoscopy with simultaneous injection of isosulfan blue dye into the liver lymphatics in 6 of 7 patients. Liver lymphatic embolization with ethiodized oil in 2 patients resulted in a temporary increase in albumin blood level and symptom improvement in 1 patient, but was complicated by duodenal bleeding in both patients. Of the remaining 6 patients, liver lymphatic embolization with n-butyl cyanoacrylate glue resulted in sustained improvement of the serum albumin level and symptoms in 3 patients, temporary improvement in 2 patients, and no change in 1 patient with median follow-up of 135 days (range, 84 to 1,005 days).
Conclusions The authors demonstrated liver lymph leakage as a cause of PLE in patients with congenital heart disease and elevated central venous pressure. Lymphatic embolization led to improved albumin levels and relief of symptoms. Further experience with the technique is needed to determine long-term outcome of this procedure.
Protein-losing enteropathy (PLE) is characterized by a severe loss of proteins, primarily albumin, into the intestinal tract. The diagnosis of PLE is suspected by history, physical examination, and hypoalbuminemia, confirmed by an elevated level of α1-antitrypsin in stool (1). Clinical symptoms of PLE include soft tissue swelling, diarrhea, and ascites. PLE following a total caval pulmonary connection surgical procedure in patients with congenital heart disease (CHD) was first described by Crupi et al. (2). The prevalence of PLE in this patient population is 3.7% to 24% (3,4). Five-year survival after the initial diagnosis of PLE in patients post-surgery was reported to be as low as 50%; however, with recent advances in medical therapy the outcome has improved with increased survival to 88% and 72% at 5 and 10 years, respectively (4–6). Intestinal lymphangiectasia identified by endoscopic biopsy in patients with PLE has implicated the lymphatic system in the pathogenesis of this disorder (6,7). The production of liver lymph in patients with congestive heart failure is increased and results in dilation of hepatoduodenal lymphatic ducts and duodenal lacteals (8). It has been hypothesized that lymphatic overdistention of the intestinal lacteals can result in their rupture with the subsequent diffuse leakage of protein-rich liver lymph into the small intestine (7,9–11).
Liver lymphangiography technique has been described for visualization of the hepatic lymphatic ducts (12). Embolization of the liver lymphatic ducts with glue has been shown to resolve a lymphatic leak in 1 patient with hepatic lymphorrhea (12). In this study, we report our initial experience in adopting these techniques to treat PLE.
We conducted a retrospective review of imaging and medical records of 8 consecutive patients with CHD and elevated central venous pressures (CVP) complicated by PLE who underwent lymphatic imaging and interventions in our institution. Data collection included patient demographics, clinical presentation, biochemical assessments, lymphatic imaging results, procedural techniques, and outcomes. Our institutional review board approved the study and a Health Insurance Portability and Accountability Act waiver was granted before its initiation.
Documenting lymphatic leakage
Liver lymphangiography was performed as described by Guez et al. (12). In brief, access to liver parenchyma was performed by placing a 10-cm, 21-gauge Chiba needle near the branch of the portal vein, using ultrasound guidance. Through this needle, a 15-cm, 27-gauge needle was advanced coaxially until the tip was positioned adjacent to the main portal vein or 1 of its branches. Opacification of the liver lymphatic vessels was performed by gently injecting water-soluble iodinated contrast medium while retracting the needle until the liver lymphatic ducts were visualized.
Before injection of iodinated contrast, an endoscopic evaluation of the stomach and duodenum was performed. After initial hepatic lymphangiography, the endoscope was retracted to the periampullary region, and in 7 of 8 patients, duodenoscopy with concurrent injection of isosulfan blue 1% was performed. The goal of duodenoscopy was to identify and confirm the location of the liver lymphatic leakage into the duodenum before embolization of the liver lymphatic ducts.
Liver lymphatic embolization
After identification of the dilated liver lymphatic ducts and radiographic and endoscopic evidence of contrast leakage in the duodenum, liver lymphatic embolization was performed. In the first 2 patients, liver lymphatic ducts embolization procedures were performed by injection of 1 to 2 ml of ethiodized oil. In the next 6 patients, embolization was performed by injection of 1 to 2 ml n-butyl cyanoacrylate (n-BCA) glue diluted with ethiodized oil. All embolizations were performed through the same access needle that was positioned in the liver lymphatic ducts during the lymphatic imaging phase described previously. In patients P1 to P4, the embolization targeted proximal intrahepatic lymphatic ducts. In patients P5 to P8, the goal of the embolization was to deliver n-BCA glue distally close to the leakage site within the duodenum, as identified by duodenoscopy.
Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, North Carolina). The baseline characteristics of each group were reported as median values with ranges or number values with percentages. To assess the statistical significance of the pre- and post-change in albumin levels, a paired Student t test was conducted and confirmed with a nonparametric Wilcoxon signed rank test.
A total of 8 patients were included in this study. Patient demographics, cardiac diagnosis, and surgical history are shown in Table 1. All patients presented with profoundly low blood albumin and total protein levels, which were monitored serially throughout the course of care. All patients had abnormally elevated spot stool α1-antitrypsin level (>100 mg/dl) or abnormally elevated α1-antitrypsin clearance (>27 ml/24 h) at initial diagnosis of PLE; however, stool samples were not serially followed. CVP was elevated in all patients with median of 14 mm Hg (range, 10 to 18 mm Hg). In patient P8, with a superior cava–pulmonary connection, pressures in the inferior and superior vena cava were reported separately. All patients had evidence of significant peripheral edema, particularly pre-tibial edema and variable degrees of ascites. In all patients, cardiac catheterization data were reviewed and it was concluded that no interventional or surgical options were available to further optimize their cardiac circulation. All patients failed conservative medical therapies and 5 of 8 patients were receiving regular albumin transfusions (Table 1).
Lymphangiography and lymphatic embolization
Liver lymphangiography was successful in all patients and demonstrated dilated liver lymphatic ducts with the presence of hepatoduodenal connections in all patients (Figure 1). Duodenoscopy before liver lymphangiogram demonstrated no focal lesions. Most patients had diffusely thick duodenal folds with numerous dilated lacteals. Occasional minor, nonbleeding erosions were seen. The ampulla was identified in all patients. Duodenoscopy with concurrent injection of isosulfan blue 1% during the procedure was performed in 7 of 8 patients, demonstrating leakage of the blue dye into the duodenal lumen in 6 of them (Figure 2). In Patient P8, isosulfan blue 1% injection demonstrated staining of a small region in the first portion of the duodenum and the antrum of the stomach without evidence of leakage into duodenum (Table 2).
In total, 15 embolization procedures were performed in 8 patients. Patients P1, P3, and P7 had 2 embolization procedures and Patient P6 had 4 embolization procedures. Patients P1 and P2 underwent embolization of the liver lymphatics with oil-based contrast; n-BCA glue was used for liver lymphatic embolization in Patients P3 to P8 and for the second embolization in Patient P1.
Outcome of liver lymphatic embolization
Embolization with ethiodized oil resulted in temporary normalization of the albumin blood level in Patient P1 but there was no change in the albumin level in Patient P2 (Figure 3). Within days of the procedure, both patients developed duodenal bleeding. Endoscopy in Patient P1 demonstrated a Dieulafoy lesion that was successfully treated by endoscopic clipping. In Patient P2, diffuse duodenal erosions were observed and successfully managed conservatively.
In 4 patients (P3, P4, P5, and P7), n-BCA glue embolization (Figure 4) resulted in an increase in albumin blood level and significant symptomatic improvement that commenced in most of the patients 1 week post-procedure. In all these patients, duodenoscopy demonstrated leakage of the isosulfan blue dye into duodenum. In Patient P3, PLE recurred 6 months after the procedure; in Patient P6 the improvement lasted for more than 1 month after each embolization procedure. Patients P4, P5, and P7, who underwent distal embolization with n-BCA glue close to the duodenal wall, had improvement of the serum albumin level and symptoms that have persisted at the time of data collection, 131 days, 116 days, and 140 days, respectively (Central Illustration). In Patient P8 without documented leakage into the duodenum on duodenoscopy, glue embolization did not change the albumin blood levels or symptoms.
In Patient P1, n-BCA glue embolization did not result in an increase in blood level of albumin, in spite of an increase following the first liver lymphatic embolization with oil-based contrast. This second procedure was complicated by a thromboembolic stroke caused by venous clot during concurrent cardiac catheterization.
We defined the patients as having “sustained” improvement if serum albumin level and symptoms persisted at the time of data collection and “temporary” improvement if the symptoms recurred by the data collection point. Overall 3 patients had sustained improvement (P4, P5, and P7) and 3 patients (P1, P3, and P6) temporary improvement (Central Illustration).
The increase in blood albumin levels from pre-procedure to the maximum post-procedure level was statistically significant (mean change, 1.50 ± 1.03; p = 0.008), based on a paired Student t test with a median follow-up of 119.5 days (range, 95 to 893 days). Consistent results were obtained using the nonparametric Wilcoxon signed rank test (p = 0.016) (Table 3).
PLE in patients with CHD and elevated CVP is a devastating condition characterized by profound enteric protein loss. Although the etiology has been poorly understood, several mechanisms have been proposed including bowel inflammation and abnormal elevation in mesenteric vascular resistance (13). Meadows et al. (14) observed a higher prevalence of PLE in patients with thoracic duct ligation and suggested that lymphatic obstruction of the thoracic duct contributes to development of PLE. Observations of intestinal lymphangiectasia on biopsy suggested that the lymphatic system plays a primary role in PLE (6,7).
In this study, we demonstrated that the mechanism of albumin loss in patients with CHD with elevated CVP and PLE is leakage of this albumin-rich lymph into the duodenal lumen, predominantly through focal hepatoduodenal lymphatic channels. This lymphatic leakage was documented by injection of iodinated contrast during fluoroscopy and was confirmed using endoscopy with injection of isosulfan blue dye directly into liver lymphatic ducts with subsequent visualization of focal concentrated egress of dye into the duodenal lumen.
A hypothesis of PLE in CHD
The concentration of albumin in the liver lymph is high and has been shown to be 83% to 91% of the plasma concentration. We hypothesized that these hepatoduodenal lymphatic connections potentially represent an anatomical variant that has no clinical significance in patients without elevated CVP; however, they become clinically evident in patients with lymphatic congestion because of competition between the liver and intestinal lymphatic drainage.
Ernest Starling in 1894 first noted that elevated CVP results in liver congestion and subsequent increase in liver lymphatic flow (15). The dilation of hepatoduodenal lymphatic ducts and dilated duodenal lacteals in these patients was demonstrated in a post-mortem study by Ludwig et al. (8). The increase of liver lymphatic flow results in distention of the hepatoduodenal lymphatic connections and intestinal lacteals, which subsequently leads to leakage of the protein-rich lymph into intestine lumen when the barrier for leak is broken down. We hypothesized that PLE develops when the combination of this lymphatic anatomical variant and elevated CVP is present. This hypothesis could explain the lack of correlation between development of PLE and severity of elevated CVP.
Liver lymphangiography is a largely unknown technique that was described first in 1963 by Moreno et al. (16) as a part of the hemodynamic evaluation of the liver in patients with liver cirrhosis. Liver lymphangiography has been used to understand the contribution of liver lymph to the symptoms of portal hypertension, to demonstrate the role of liver lymphatic flow in liver metastasis, and for the diagnosis of liver lymphorrhea (12). In this study, we used a coaxial technique previously described by Cope (17) to access the liver lymphatic system and were able to opacify the liver lymphatic ducts in all 8 patients (12). Normal liver lymphatics are very small and, in most cases, cannot be opacified during liver lymphangiography (18). However, in patients with liver cirrhosis and congestive heart failure, liver lymphatic ducts are significantly dilated because of venous congestion (8). For that reason, even visualization of the liver lymphatic ducts during lymphangiography is considered a sign of dilation.
Liver lymphatic embolization
Two patients who underwent liver lymphatic embolization with ethiodized oil experienced an unexpected severe upper gastrointestinal bleed. Ethiodized oil has been described as an efficient and safe lymphatic embolization agent but has not been studied in this setting (19,20). Although the exact mechanism that led to bleeding is not entirely clear, we believe the oil-based contrast caused erosion of the intestinal wall while traversing a friable and edematous intestinal mucosa. Consequently, we modified the embolization technique to use n-BCA glue, which occluded hepatoduodenal lymphatics without crossing into the duodenal lumen, as determined by direct observation. This technique did not result in any complications. However, further studies need to be conducted to determine the long-term safety of this approach.
Glue embolization of hepatoduodenal connections to prevent leakage of the hepatic lymph into duodenum resulted in a rapid increase in serum albumin levels and improvement in symptoms in most of the patients. In the first few patients, however, the increase of the albumin level was temporary. In these patients, embolization of intrahepatic lymphatics was performed. As we learned more about the anatomy and the hepatoduodenal connections, we modified our technique. In the last 4 patients, a strategy was used to force the n-BCA glue toward the distal hepatoduodenal connections, close to the leakage points near the duodenal mucosa that were identified with endoscopy. This resulted in a significant increase in the blood level of the albumin and dramatic improvement of the symptoms that were sustained at 4-month follow-up. Knowledge of the liver lymphatic anatomy and physiology is somewhat lacking and limited to only a handful of publications (21). Fortunately, there is a renewed interest in the liver lymphatic system and new studies should provide more details of liver lymphatic anatomy, providing further guidance for liver lymphatic interventions (22,23).
This was a retrospective review of treatment in a small number of patients and, as such, subject to all the limitations inherent in such a report. In particular, we attributed clinical outcomes (symptom resolution and increased serum albumin) to our interventions, using each patient’s baseline data and state as their own control. It is possible that the changes we observed were caused by other confounding factors, such as disease variability or concurrent treatments. As noted, only 1 patient had a major change in medical therapy immediately before or after embolization therapy (P3 underwent initiation of steroids close to the second embolization).
In this report, we demonstrated leakage of liver lymph into the duodenum, mostly through abnormal focal hepatoduodenal lymphatic connections in 8 patients. Our findings strongly suggested that this pathophysiology is the proximate cause of PLE in this cohort. Similar to our findings in the condition of plastic bronchitis after Fontan operation, in which we have proposed spillage of lymph into the airway under conditions of venous hypertension and lymphatic congestion as the etiology, we believe the mechanism of lymphatic congestion within the liver with spillage into the duodenum is likely the dominant cause of PLE in our CHD population (24,25).
Embolization of these hepatoduodenal lymphatic connections with n-BCA glue can provide at least temporary normalization of serum albumin levels and significant symptomatic improvement. Further studies are needed to determine the long-term outcome and potential for sustained efficacy of these procedures, as well as refinement of the optimal techniques for closure of abnormal hepatoduodenal connections that will provide longer-term resolution of PLE.
COMPETENCY IN PATIENT CARE AND PROCEDURAL SKILLS: In patients with congenital heart disease and protein-losing enteropathy, albumin-rich lymph leaks from the liver into the intestinal lumen. Percutaneous embolization of hepatoduodenal channels identified by hepatic lymphangiography and contrast-guided duodenoscopy can raise serum albumin levels and improve symptoms.
TRANSLATIONAL OUTLOOK: Further research is needed to understand the optimum timing, method safety, and long-term outcomes of hepatoduodenal lymphatic embolization in patients with congenital heart disease and other forms of right heart failure associated with protein-losing enteropathy.
Dr. Itkin has received grant funding from Guerbet, LLC. Dr. Nadolski has received grant funding from Teleflex Medical and Guerbet, LLC. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- congenital heart disease
- central venous pressure
- n-butyl cyanoacrylate
- protein-losing enteropathy
- Received January 7, 2017.
- Revision received March 31, 2017.
- Accepted April 4, 2017.
- 2017 American College of Cardiology Foundation
- John A.S.,
- Johnson J.A.,
- Khan M.,
- Driscoll D.J.,
- Warnes C.A.,
- Cetta F.
- Connor F.L.,
- Angelides S.,
- Gibson M.,
- et al.
- Altman P.L.,
- Katz D.D.
- Yoffey J.M.,
- Courtice F.C.
- Guez D.,
- Nadolski G.J.,
- Pukenas B.A.,
- Itkin M.
- Clain D.,
- McNulty J.
- Matsumoto T.,
- Yamagami T.,
- Kato T.,
- et al.
- Rouviere H.,
- Tobias M.J.
- Pupulim L.F.,
- Vilgrain V.,
- Ronot M.,
- Becker C.D.,
- Breguet R.,
- Terraz S.
- Iwakiri Y.
- Dori Y.,
- Keller M.S.,
- Rome J.J.,
- et al.
- Dori Y.,
- Keller M.S.,
- Rychik J.,
- Itkin M.