Improved neurocognitive function after radiologic closure of congenital portosystemic shunts.

Abstract

Congenital portosystemic shunts are rare vascular anomalies causing direct shunting of blood from the portal circulation into the central venous circulation (1,2). They are often a consequence of persistence of elements from the fetal circulation, in particular the ductus venosus. Patients with congenital portosystemic shunts often have no signs or symptoms of disease until adulthood, at which time encephalopathy is the leading presenting symptom (3–6). Shunts may be discovered during surgery for other reasons and as incidental findings during sonography or computed tomography. Recent advances in interventional radiology have provided an opportunity to change management strategies from surgical to less invasive interventions (7–9). We report two children with symptomatic portosystemic shunts in whom interventional radiology was used to obliterate the shunts. Clinical symptoms and neurocognitive function improved in both children after the shunts were obliterated and hepatic portal venous perfusion restored. CASE REPORTS Case 1 A 16-year-old girl presented with nonpitting edema and telangiectasias of the lower extremities. Her medical history included a small ventricular septal defect that had spontaneously closed, Caffey disease, and mild nonconjugated hyperbilirubinemia consistent with Gilbert disease. She used no oral contraceptives. A cardiac murmur on physical examination led to an echocardiogram that showed no structural cardiac disease. A liver mass was found incidentally. The liver was reported to be normal on physical examination. Her serum alanine aminotransferase (ALT) level was 50 IU/L (normal, <40 IU/L); aspartate aminotransferase (AST), 70 IU/L (normal, <45 IU/L); gamma-glutamyltransferase (GGT), 126 IU/L (normal, <40 IU/L); total protein 6.5 g/dL; albumin 3.3 g/dL; and total and conjugated bilirubin 2.5 and 0.1 mg/dL (normal, <1.2 and <0.4 mg/dL), respectively. Serum cholylglycine by radioimmunoassay was 827 pg/dL (normal, <64 pg/dL). Abdominal sonography and computed tomography revealed an isolated mass in the right lobe of liver. At surgical exploration, a 10- × 7- × 3.8-cm nodule was completely resected from the inferior right lobe. Three other nodules, two in the right upper lobe and one in the quadrate lobe, were identified but not resected. The histologic findings of the resected nodule were typical of focal nodular hyperplasia (FNH) (Fig. 1). She developed low-grade encephalopathy during the next 4 months, with deteriorating school performance and episodes of combative behavior. Her blood ammonia was 110 pmol/L (normal, 11–35 pmol/L). Her symptoms and blood ammonia improved when lactulose syrup was prescribed.FIG. 1.: Fibrous satellite scar within the hyperplastic nodule. Thick-walled arteries (A), portal venules (V), and bile ductule (Bd) proliferation can be detected.The patient was referred to our service 1 year after surgical exploration. Her school performance had again deteriorated. The former “B” student was failing. She was apathetic and had a flat affect. Her liver was palpable 10 cm below the xiphoid and was not palpable below the right costal margin in the anterior axillary line. There was no splenomegaly or ascites. Her lower legs showed extensive cutaneous telangiectasias and nonpitting edema. Numerous spider angiomas were present on her arms and back. Her only laboratory abnormalities were a mild leukopenia (3.7 K/pL) and hypoalbuminemia (3.3 g/dL). Low-grade encephalopathy, hyperammonemia, and elevated serum cholylglycine levels suggested portosystemic shunting. Sonography with Doppler failed to demonstrate a site of shunting. A review of computed tomography images suggested the presence of a patent ductus venosus (PDV). PDV was confirmed by superior mesenteric artery angiogram. The study also demonstrated multiple hypervascular nodules within the liver, especially in the enlarged left lobe. Using a transjugular approach, the portal vein was catheterized via the PDV. A venogram demonstrated another large shunt between the right portal and hepatic veins and numerous small intra-hepatic venovenous shunts (Fig. 2A). There was complete absence of parenchymal perfusion by portal blood.FIG. 2.: (A) Case 1. Two main shunts can be seen, PDV and shunt between right portal vein and right hepatic vein. PDV, Patent ductus venosus; s, shunt; PV, portal vein; HV: hepatic vein; RA, right atrium. (B) Case 1. No flow can be seen through shunts after multiple staged coil embolizations. PV, portal vein.Occlusion of the two main shunts briefly with a balloon resulted in increased perfusion of the hepatic parenchyma but a 20 mmHg rise in portal vein pressure. We assumed that the portal system would not tolerate simultaneous occlusion of these shunts, and a staged embolization was planned. Initially, a spider obstructing device was placed in the PDV, and multiple coils were packed upstream to completely obliterate it. One month later, a near complete occlusion of the right portal vein to the hepatic vein shunt was achieved using the same technique. After this procedure, there was hepatopetal flow in the portal vein and good perfusion of the hepatic parenchyma (Fig. 2B). The patient’s mood and school performance improved. Four months later, she returned with recurrence of apathy and poor school performance. Neurocognitive testing was performed, which showed findings consistent with organic brain dysfunction (results described below). The blood ammonia level was normal. A functional magnetic resonance scan of her brain showed only subtle nonspecific changes. A retrograde hepatic venogram showed excellent portal perfusion of the liver, but catheterization of the remaining shunts was not possible. Seven months later, a residual large shunt between the portal vein and right hepatic vein was successfully embolized. In the following year, she experienced a few brief episodes of lethargy that improved with lactulose administration. Her liver function remained normal. Mild hypersplenism persisted with a leukocyte count of 4.2 K/μL and platelets of 87 K/μL. Repeat neurocognitive testing showed considerable improvement in several areas of function. One other prominent venous shunt was obliterated angiographically, and final vascular study performed 1 year later demonstrated normal hepatic portal vein perfusion. A few small intrahepatic venovenous shunts were still present but had not increased in size and were too small to catheterize for embolization. In the 2 years since obliteration of the shunts, she has had improved mood, strength, and stamina. Her school performance returned to its pretreatment level. She attended junior college with average academic performance and was recently married. She is currently employed full time. Her leg edema and telangiectasias have mostly resolved. Her unconjugated hyperbilirubinemia has not resolved, suggesting that it was more likely a result of Gilbert disease than of portosystemic shunting. Case 2 A 3-month-old infant with multiple cutaneous hemangiomas underwent sonography to investigate an eventration of the right hemidiaphragm discovered on chest x-ray. A hemangioma was identified within the dome of the right lobe of the liver, which appeared to be partially obstructing the hepatic venous outflow. Blood vessels with large flow detected by Doppler ultrasound were thought to be associated with the hemangioma. The liver was soft and of normal size, and there was no splenomegaly. Liver function test results were normal. One year after presentation, an ultrasound done to evaluate hepatic outflow showed a regression of the hemangioma but a worsening of the appearance of the high flow vessels. In the venous phase of the superior mesenteric arteriogram, a hemangioma comprising 25% to 30% of the right lobe of liver and a large PDV rapidly shunting most portal flow into the right atrium were found (Fig. 3A). A large atrial septal defect also was demonstrated. The hepatic hemangioma disappeared spontaneously by 2.5 years of age, and the septal defect was surgically closed shortly thereafter.FIG. 3.: (A) Case 2. PDV and filling of hemangioma with contrast material at the bottom of figure. PDV, patent ductus venosus; PV, portal vein. (B) Case 2. No flow through PDV after two stents and coil placements. PV, portal vein.The child’s parents became concerned about delayed development at 3 years of age. Their concerns were confirmed by Stanford-Binet Intelligence Scale 4th edition (10) during evaluations at preschool. We performed a comprehensive neurocognitive evaluation at 4 years, which indicated he had deficits indicative of organic brain dysfunction (results reported below). His liver size remained normal, but the spleen became palpable 2 cm below the left costal margin. An endoscopy demonstrated no varices. Because the child’s neurodevelopmental assessment suggested there was low-grade hepatic encephalopathy, embolization of the PDV was planned. The transjugular route was used to enter the PDV. Test balloon occlusion of the PDV produced a pressure gradient of 20 mmHg between the portal vein and the right atrium, indicating that staged closure was necessary. A treatment strategy was used that depended on the natural tendency for wire mesh stents placed intraluminally to endothelialize and develop intimal hyperplasia. Two stents were deployed in the PDV, one inside the other at an interval of 2 months. Ultrasound was used to determine flow in the PDV and the need for additional intervention. After 1 month, shunt flow was reduced to a level allowing transvenously placed coils to complete the obliteration of the PDV (Fig. 3B). Portal perfusion of the hepatic parenchyma appeared normal. The patient’s parents reported some improvement in his school performance after PDV closure. Neurocognitive testing after shunt closure showed improvement in several areas of function. However, he continued to display disruptive behavior, repetitive motor movements, and mild social skills deficits, and received a diagnosis of Asperger syndrome. Two years after shunt closure, he has shown additional gradual improvement in all aspects of his behavior and performance. His strength and stamina have improved significantly, although he has persistent weakness on his left side. He currently attends special classes for children with attention deficit disorders, in which he functions acceptably. Neurocognitive Function Before and After Closure of Portosystemic Shunts In both cases, neuropsychologic evaluation was performed around the time of treatment and then well after closure of the shunts. Tests used were predominantly standardized measures with norms derived from extensive samples based on age and other factors. Specific information regarding each test can be found in individual test manuals (see Tables 1 and 2 for references). Although an attempt was made when possible to use the same measures before and after treatment, this was less feasible in patient 2. Tables 1 and 2 present the data as standard scores, which have a mean of 100 and a standard deviation of 15, and represent the tested child’s function relative to expectations for age. Comparisons of results between the two evaluations were made based on the change in scores and the standard error of measurement (SEM) of the test, which is an estimation of the amount of error attributable to unreliability in an individual’s observed test score (11). Using this method, 95% confidence intervals were constructed for each test when SEMs were available. A difference in score (Z score) before and after treatment exceeding the 95% confidence interval was considered significant. However, this statistic may be questioned because our patients are single cases, and the comparisons made are between single repeated tests and not those of large groups. Given the limitations of current quantitative data, our interpretations are also based on available qualitative data.TABLE 1: Neurocognitive functioning for case 1TABLE 2: Neurocognitive functioning for case 2Patient 1 had neuropsychologic evaluation at age 17 years and again at 18 7/12 years. At the time of her initial evaluation, she reported difficulty with attention, memory, and fatigue. Testing showed intellectual ability in the low average range. She had specific difficulties in the areas of working memory (mental juggling of complex material, mental multitasking), attention, learning and memory (likely related to attention), fine motor speed, and energy/stamina. She had a flat affect, which she later acknowledged was associated with other symptoms of depressed mood. Re-evaluation 19 months later indicated stable or improved function. Performance improved most in the areas of fine motor speed, mood, energy and stamina, learning, and memory (likely related to attentional factors). Stable, mild to moderate deficits remained in the areas of working memory and sustained attention. Her mood had normalized. See Table 1 for details. Patient 2 underwent neuropsychologic evaluation at age 41/12 years and at age 511/12 years. At the initial evaluation, his parents were concerned about attention and anxiety. Testing revealed global mild cognitive delays and delayed adaptive (self-help) skills. Re-evaluation after shunt closure revealed striking improvement in intellectual ability. However, concerns about attention and anxiety remained. Furthermore, he displayed mild to moderate difficulty with learning and memory (likely related to attention) and persistent deficits in adaptive skills. Fine motor speed and dexterity also remained below average for age. See Table 2 for details. In both cases, there was evidence of improved neurocognitive functioning after closure of the portosystemic shunts and restoration of hepatic portal venous perfusion. In patient 1, an adolescent at the time of intervention, improvement was seen in mood, energy, and fine motor speed. Stability in attention allowed for more effective learning of new information. It is possible that her improved mood also contributed to her higher functioning. However, stable deficits remained in other areas of “fluid” functioning, including sustained attention and working memory, which often are sensitive to mood changes. It is also possible that her depressed mood at the time of initial testing was a symptom of low-grade hepatic encephalopathy, which improved along with neurocognitive function after closure of her shunts. In patient 2, a preschooler at the time of his surgery, more global developmental and intellectual delays were observed initially, which improved after obliteration of his portosystemic shunt. This patient’s pattern of global delays with early disease onset in contrast to more circumscribed deficits primarily in fluid functioning with later disease onset is observed with many types of brain insult occurring during neural development (12). The relationship of his Asperger syndrome to the portosystemic shunting is unclear, especially in light of his neurocognitive improvement at the time that diagnosis was made. DISCUSSION These cases illustrate some important aspects of an emerging understanding of congenital portosystemic shunts. First, we used a treatment paradigm employing interventional radiology, rather than surgery. Second, our evaluations of these two patients suggest that shunting of portal blood from the liver may have major consequences on neurologic development and learning. Finally, we observed a novel association between congenital portosystemic shunts and FNH. Experience with these patients suggests that congenital portosystemic shunts should not be considered to be benign. Rather, they may have deleterious effects on neurologic development and function. The literature suggests that patients with congenital portosystemic venous shunts often have no symptoms until adulthood, when alterations in neurocognitive function develop (3,5,6,13–15). However, the question remains whether early detection and intervention might improve performance in adult patients long before clinical symptoms develop. Recent reports suggest that children with congenital portosystemic shunts are likely to have neurologic symptoms and that their intensity correlates with the magnitude of shunting (4,16). It is not clear why or how portosystemic shunting affects neurologic function. There is substantial evidence that poor neurocognitive outcome is associated with earlier age of onset in a number of medical conditions, such as closed head injury, cranial irradiation, hydrocephalus, heart disease, and liver disease (12,17–19). Subclinical hepatic encephalopathy has been reported in an infant with congenital absence of the portal vein (20), and we have observed defects in neurocognitive function in children with idiopathic extrahepatic portal vein thrombosis (21) that resolve after restoration of portal flow by mesenteric to left portal vein bypass surgery (22) (unreported data). These observations and the findings in the current cases suggest that the developing nervous system might be sensitive to the effect of portosystemic shunting and/or deprivation of portal blood flow to the liver. Damage incurred during development could reduce the functional capacity of the adult. These speculations suggest that early closure of shunts and restoration of portal flow to the liver is indicated if it can be done without harm to the patient. Both of our patients had impairment of neurocognitive function to a degree that treatment was clearly indicated. Initially, neither had any demonstrable flow in the intrahepatic portal venous system, and all mesenteric flow passed immediately through the shunts. Coincident with obliteration of shunts, both had dramatic increases in portal blood flow to the hepatic parenchyma. The fact that portal pressure increased to unacceptably high levels on test occlusion of the shunts indicated that the portal system might not tolerate an increased flow after shunt obliteration. Dye studies during test occlusion showed small, underdeveloped portal venous systems in both patients. Staged occlusion resulted in increased size and resumption of normal flow in the portal system. We cannot say with certainty whether elimination of shunting or restoration of hepatic parenchymal perfusion was the cause of the improved neurologic function we observed. In cirrhosis, poor hepatic perfusion with portal blood seems to have a greater propensity to cause encephalopathy than does the presence of major portosystemic shunts (23). We think that restoration of portal flow was a major benefit to the treatment approach we used in these patients. Liver tumors consistent with FNH were detected during the evaluation of lower limb edema in our first patient. At the time of her initial evaluation, she had no signs or symptoms suggesting either liver disease or a congenital portosystemic shunt. Indeed, the shunt was not detected during comprehensive imaging and surgical removal of the largest of the tumors. FNH has been reported in patients with absent or atretic portal veins, portal hypertension, intrahepatic portosystemic venous shunts, patent ductus venosus, hepatic hemangiomas, and surgical splenorenal shunts (14,24–38). All of these conditions produce or are associated with reduced portal blood supply, which may play a role in the development of FNH. We suggest that the discovery of FNH in a child should raise suspicion that a portosystemic vascular imbalance might be present. The pathogenesis of FNH is unknown. It appears to be a reactive process, rather than an autonomous proliferation of hepatocytes (such as hepatic adenoma or adeno-carcinoma). Pre-existing arterial malformations and irregular distribution of hepatic blood supply have been hypothesized to produce hyperplastic responses in the liver parenchyma, leading to nodule formation (39,40). Why an imbalance in blood supply may cause the liver to develop nodules remains to be determined. Uneven distribution of nutrients and factors regulating hepatocyte regeneration may play a role (41). The reported association between FNH and telangiectasis of the limbs, as observed in our patient, and with central nervous system and endocrine tumors, vascular malformations of other organs, and hemihypertrophy all suggest there may also be systemic maldistribution of growth factors in this disease (28,31,37,38). Our experience has led us to consider a pre-emptive approach to treating children with congenital portosystemic shunts. We suspect that these shunts are not, in fact, benign in childhood. We suggest that neurocognitive function in young children with shunts should be evaluated before deficits are clinically evident. Definitive treatment can then be undertaken as soon as abnormalities are detected. The effectiveness of conservative management of low-grade hepatic encephalopathy by protein-restricted diet and lactulose administration has not been established in children with congenital portosystemic shunts. Given the rarity of the condition, response to these measures may never be adequately tested. We think that conservative therapy may be used as an adjunct or a prelude to definitive treatment but should not constitute the main therapeutic strategy. It appears that neurocognitive dysfunction is likely when there is a markedly diminished portal blood flow to the hepatic parenchyma. Therefore, we suggest that obliterative treatment should be considered as preventative therapy in patients with congenital portosystemic shunts and severely reduced portal blood flow, even before neurocognitive dysfunction is detected. Children with unexplained neurocognitive dysfunction consistent with mild hepatic encephalopathy (e.g., disturbed memory, reduced attention span) should be evaluated for the possibility of portosystemic shunting (6,20,42,43). The yield for such an evaluation would be low because congenital portosystemic shunts are rare, but the absence of signs and symptoms pointing to liver disease in these children makes routine clinical evaluation unreliable. Finding a shunt presents a potentially treatable cause for brain dysfunction. Although hyperammonemia is not a consistent finding, measuring blood ammonia after protein loading might detect patients with this condition. Surgery has been the mainstay of treatment for this disease in the past. Recent innovations in interventional radiology now provide an alternative to surgical intervention. We think that angiography is necessary before any invasive intervention because the size and number of shunts and the existence of other vascular anomalies determine the appropriate approach. The presence of numerous shunts, as observed in our first patient, requires planned repetitive embolization procedures. Portal hypertension may result from abrupt changes in portal hemodynamics, which may also determine the need for a staged procedure. During test balloon occlusions in our patients, hypoplastic intrahepatic portal systems did not appear to tolerate the newly established portal blood load. Therefore, staged occlusive procedures were performed to permit the portal system to adjust to its new hemodynamic status. Dislodgment of coils into the systemic circulation is a risk during these procedures. The association of congenital portosystemic shunts with cardiac anomalies heightens the risk for stroke. Both of our patients had cardiac septal defects, one of which required surgical closure. All patients should have a careful cardiac evaluation before treatment is undertaken. Embolization of shunts has been successfully performed via transhepatic, mesenteric, umbilical venous, and transcaval routes (7–9,16,44,45). We completed obliteration of the shunts in our patients via the transjugular route. Modern pediatric interventional radiology permits this treatment even in very small patients, as illustrated by its use in a newborn with hepatic failure and a patent ductus venosus, which was obliterated during preparation for hepatocyte transplantation (46). In summary, these two cases emphasize the presenting symptoms and a novel treatment to congenital portosystemic shunts. Although sonography and computed tomography have increased the detection rate of these lesions, angiography is valuable for visualization of complicated lesions. Interventional radiology was successfully used to obliterate the shunts in our patients and resulted in improved neurocognitive function. We hypothesize that deprivation of portal blood to the liver was a key determinant in the development of neurocognitive abnormalities.