Extrahepatic Portal Vein Obstruction in Children: Role of Preoperative Imaging.

AIM: Extrahepatic portal vein obstruction (EHPVO) is characterized by features of recent thrombosis or portal hypertension with portal cavernoma as a sequel of portal vein obstruction. Imaging of spleno-portal axis is the mainstay for the diagnosis of EHPVO. The aim of this study is to analyze the role of imaging in the preoperative assessment of the portal venous system in children with EHPVO.
MATERIALS AND METHODS: A hospital-based cross-sectional study was conducted on twenty children with EHPVO aged between 1 and 18 years over a period of 1 year. The children were evaluated clinically, followed by upper gastrointestinal endoscopy. Radiological assessment included imaging of the main portal vein, its right and left branches, splenic vein, and superior mesenteric vein using color Doppler ultrasonography (CDUSG) and magnetic resonance portovenogram (MRP). Evidence of portal biliopathy, status of collaterals, and possible sites for portosystemic shunt surgery were also examined.
RESULTS: All the patients presented in chronic stage with portal cavernoma and only one patient (5%) had bland thrombus associated with cavernoma. The CDUSG and MRPs had a sensitivity of 66.6-90% and 96.7% and specificity of 91.5% and 98.3% respectively with regard to the assessment of the extent of thrombus formation and flow in the portal venous system. Both the modalities were found to be complementary to each other in preoperative assessment of EHPVO. However, the sensitivity of MRP was slightly superior to CDUSG in detecting occlusion and identifying portosystemic collaterals and dilated intrahepatic biliary radicals.
CONCLUSION: Results of the present study indicate that MRP is well suited and superior to CDUSG in the preoperative imaging of patients with EHPVO.

INTRODUCTION

Extrahepatic portal vein obstruction (EHPVO) is defined by obstruction of the extrahepatic portal vein with or without the involvement of the intrahepatic portal veins. It is characterized by features of recent thrombosis or portal hypertension with portal cavernoma as a sequel of portal vein obstruction.[1] Imaging of the spleno-portal axis is the mainstay for the diagnosis of EHPVO. Imaging modalities such as Doppler ultrasound (US), computed tomography, or magnetic resonance imaging (MRI), demonstrate portal vein obstruction, presence of intraluminal material, or portal vein cavernoma.[1] Usually, color Doppler ultrasonography (CDUSG) is the first imaging modality used and is accurate in the assessment of the portal venous system. However, in patients who are potential candidates for portosystemic shunt procedures, a more detailed diagnostic method that covers the whole portal venous system is required to choose the type of shunt surgery to be performed. Recently, dynamic contrast enhanced MR angiography with gadobenate dimeglumine have been shown to be very promising noninvasive methods for assessment of the portal venous system.[2]

Materials and Methods

Study group

The study comprised twenty patients, of both sex up to 18 years of age having symptoms related to EHPVO. We excluded patients who had evidence of liver cirrhosis and myeloproliferative disorders. The imaging findings were correlated with clinical features, laboratory investigations, upper gastrointestinal (GI) endoscopy, and intraoperative assessment of the portal venous system in children who underwent surgical management.

Patient preparation

Patients were kept nil per oral except plain water for 6–8 h preceding the examination. Risks of contrast administration were explained to the patient, and informed consent was obtained before administration of intravenous contrast medium. Imaging findings were verified by a senior consultant of the Radiodiagnosis Department.

Color doppler ultrasonography evaluation

With CDUSG, spectral, and color flow Doppler images were obtained in all cases. The examination was performed in supine position and included assessment of the upper abdominal viscera as well as assessment of diameter, direction, and character of flow within the portal vein, splenic vein (SV), and superior mesenteric vein (SMV). Normal phasicity with respiration and change of luminal diameter with deep inspiration and expiration was assessed.

Magnetic resonance portovenogram

Magnetic resonance portovenogram (MRP) imaging was performed with a 1.5-T superconducting magnet system. Three-dimensional fast imaging was performed with a fast field echo technique and a quadrature body coil in the coronal plane.

Before imaging, patients were instructed to hold their breath in the same manner 5 times consecutively with short intervals between the breath holds; they rehearsed to help reduce respiratory motion artifact. Debilitated patients who were unable to hold their breath for 12–24 s were instructed to perform shallow breathing during imaging.

The results of CDUSG and MRP were compared with each other by calculating sensitivity, specificity, positive predictive value, and negative predictive value and diagnostic accuracy of both modalities.

To observe the intermodality agreement, Cohen's kappa coefficient value was calculated for various observations of CDUSG and MRP. Kappa index values >0.80 were considered excellent agreement; between 0.60 and 0.80, good; between 0.41 and 0.60, moderate; and <0.40, fair to poor. Finally, to know the difference between the two modalities in identifying the findings, the results were compared by two-tailed Fisher's exact test calculated by 2 × 2 contingency table.

RESULTS

The youngest patient in this study was 1 year, and the oldest patient was 18 years old (median age 10 years). Females slightly outnumbered males by ratio of 3:2.

Among the twenty cases of EHPVO studied, the main portal vein was the most frequent site of involvement. CDUSG detected involvement in 18 cases and MRP in all twenty cases. Right portal vein (PV) involvement was shown in 12 cases on CDUSG and in 15 cases on MRP. The left PV was involved in ten cases on CDUSG and in 14 cases on MRP. Involvement of SV and SMV was noted in three and two cases respectively on CDUSG and MRP detected involvement of both veins in six cases.

A total of nine patients underwent various types of shunt surgery (Meso caval in 5, Meso-Rex and Spleno-renal in two patients each) and 11 patients had endoscopic variceal ligation. The patency/cavernous transformation of PV, the patency, and size of the left branch of portal vein and SMV were confirmed at surgery in patients who underwent Meso-Rex shunt. Same way patency and size of PV and SMV were confirmed at surgery in patients who underwent Mesocaval shunt [Table 1]. Splenomegaly and status of collaterals were also confirmed at surgery in patients who underwent shunt surgery.

Table 1

Upper gastrointestinal endoscopy and intra-operative findings

It was observed that MRP was more sensitive than CDUSG in the detection of most of the collateral veins particularly gastroesophageal (GE) junction (94.74% vs. 55.56%) and rectal collaterals (100% vs. 0%) [Figure 1a]. However, both the modalities were concordant in detecting many collaterals and there was moderate to very good agreement (kappa = 0.459–0.828) with no statistically significant difference between the two imaging modalities (P > 0.05) except for GE junction and rectal collaterals where there was poor agreement (kappa = 0.215 and 0) and statistically significant difference (P = 0.0310 and 0.0202 respectively) between the CDUSG and MRP.

Figure 1

Magnetic resonance portovenogram showing. (a) Multiple splenogastric, lienorenal, mesenteric, and rectal collaterals; (b) portal cavernoma with multiple splenogastric, lienorenal, mesenteric, and retroperitoneal collaterals; (c) portal biliopathy showing extrinsic compression of common hepatic duct with proximal biliary dilatation; and (d) portal cavernoma with narrow left portal vein (1.5 mm)

Although minimally raised level of bilirubin was noted in only four patients in the series, evidence of portal biliopathy (PB) was noted in seven patients on MRI, but only one on CDUSG. Type I biliopathy was seen in six patients and Type IIIb was seen in one patient with dilatation of the entire biliary tree including the common bile duct (CBD) [Figure 1b]. MRI was highly sensitive than USG (100% vs. 14.29%) for detecting PB because magnetic resonance cholangiopancreatography sequences could be acquired in the same setting of dynamic MRP. There was poor agreement (kappa = 0.178) and statistically significant difference (P = 0.0436) between the two imaging modalities.

Another significant observation noted in the present study was the detection of periportal fibrosis in 6 (30%) cases on MRI and in 3 (15%) cases on USG. Splenic infarct was noted in 2 (10%) patients on MRI but was not detected in any case on USG. Volume redistribution with caudate lobe hypertrophy was appreciated in only seven patients on MRI and in five patients on USG.

DISCUSSION

EHPVO is the most common cause of portal hypertension and upper GI bleeding in children in India. Although it is seen in all age groups among children, few studies reported a bimodal age of presentation – those secondary to umbilical vein catheterization or umbilical sepsis usually manifest early (3 years), whereas those following intra-abdominal infections or idiopathic ones manifest late (8 years) or sometimes into early adulthood.[34] In our series also bimodal age presentation was noted, however, none of the patients had a history of umbilical vein catheterization. Our study also noted a slight female preponderance.

Several studies have observed that EHPVO in childhood is most often chronic and presents with features of variceal bleeding and splenomegaly.[4567] Schettino et al. stated that as a general rule, initial thrombus formation is asymptomatic and the only sign may be the formation of new vessels which on Doppler ultrasonography, is known as portal cavernoma.[5] Recent studies suggest that children with EHPVO develop the progressive liver disease as adults as a consequence of PB.[6] Hence, adequate preoperative assessment of the portal venous system by a proper imaging modality is important to learn about the extent of the thrombotic process and patency of the left branch of the portal vein (if a Rex shunt is planned for example), presence of biliopathy and collaterals in porta, GE, and retroperitoneal areas, and also to plan postoperative management [Figures 1c, d, and 2a].[7]

Figure 2

(a) Peroperative image shows completed Rex shunt; (b) Doppler scan shows good flow across shunt

As a general rule, initial thrombus formation is asymptomatic, and the only sign may be the formation of new vessels which on Doppler ultrasonography, is known as portal cavernoma.[5] Similar observation was noted in the present study as well as by few other studies.[467] Our study revealed CDUSG sensitivity of 66.7% to 90% in detecting involvement of the portal vein and its two main branches. Shinde et al. and Luna and Vargas found the CDUSG sensitivity in their series of around 70%–90%.[89] In the present study, CDUSG was less sensitive for splenic and SMVs (50% and 40% respectively). This might be because of excessive bowel gas, improper bowel preparation, less cooperation by children during the procedure and also because CDUSG is an operator-dependent modality. Similar findings were observed by other authors as well.[101112] On the other hand, the present study revealed MRP to be highly sensitive for the entire portal venous system including the splenic and SMVs, which had 100% sensitivity except for left branch of the portal vein (93.3%). These results were similar to studies conducted by Sacerdoti et al. and Kreft et al. who reported MRP to be 100% sensitive for detecting thrombus/cavernoma in the entire portal venous system.[132] MRP provides additional information over CDUSG in assessing intrahepatic portal branches and is recommended for all surgical candidates.[14]

In regard to the site of spleno-portal axis involvements, the present study found good to moderate agreement between the imaging findings of CDUSG and MRP [Table 2]. Moreover, there was also no statistically significant difference between the imaging modalities. However as mentioned above, MRP was more sensitive and accurate particularly with respect to the evaluation of intrahepatic portal branches, splenic, and SMVs. In addition, MRP was superior to CDUSG in identifying portosystemic collaterals more adequately, and clearly demonstrated portal vessels that could not be visualized with CDUSG [Table 3]. This is helpful in making the choice of surgery and the type of shunt to be performed.

Table 2

Diagnostic results of CDUSG and MRP for the Site of Spleno-Portal Axis Involvement

Table 3

Diagnostic results of color Doppler ultrasonography and magnetic resonance portovenogram for detection of collateral veins

Recanalized paraumbilical vein was the only collateral where CDUSG was found to be more sensitive than MRP (100% vs. 50%). It may be because of use of coronal imaging plane for our MRP studies and that these collateral vessels are frequently anterior to the imaging volume and often go undetected on maximum intensity projection reconstructions but can be easily localized on transabdominal USG. Leyendecker et al. also had similar observations in their study.[15]

Although PB incidence is very common in EHPVO, only one-fifth of them are symptomatic.[161718] Increased prevalence of biliopathy in EHPVO is related to long-standing portal cavernoma in the biliary and peribiliary region, causing compressive, and ischemic changes on the biliary tree, the later ones may remain irreversible even after shunt surgery. Our study revealed PB in only 35% of patients [Table 4]. This might be because of smaller sample size and wide difference in the age group of the patients.

Table 4

Results of color Doppler ultrasonography and magnetic resonance portovenogram for detection of portal biliopathy

Classically, in EHPVO cases liver function is normal, and the architectural pattern is preserved. However, in some patients, the histopathology may reveal concentric condensation of reticulin fibers around portal tracts, sometimes extending into the parenchyma.[8] This may appear as bright echogenic focus on USG and hyperintense signal on T2-weighted image MRI around portal radicals. Volume redistribution in cavernous transformation may result in hypertrophy of the caudate lobe, left lateral segment atrophy and a normal or enlarged segment IV, which could be easily diagnosed with CDUSG and MRI scan. CDUSG is also more helpful in assessing the shunt functional status in postoperative patients [Figure 2b]. Splenic infarction due to occlusion of the splenic vascular supply may be segmental or may involve the entire organ. It is the result of arterial or venous compromise and is associated with a heterogeneous group of diseases. This was observed in two patients in our series; however, the exact cause was not established. MRI has the added advantage that any associated pathology of the abdominal organs can be identified, which helps in planning treatment strategies.

CONCLUSION

For preoperative assessment of the portal venous system, the role of USG and MRI are complementary to each other. In the present study, MRP was found to be more sensitive and accurate in diagnosis and superior to CDUSG in detecting patency of portal venous system and its branches, collaterals and had the added advantage of delineating changes in intrahepatic biliary ducts and differentiating choledochal varices from CBD stones. These findings were helpful in making the choice of surgery and the type of shunt to be performed.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.