Hemodynamic consequences of spontaneous splenorenal shunts in deceased donor liver transplantation


  • This article was presented at the 16th Annual International Congress of the International Liver Transplantation Society (Hong Kong, 2010).


The presence of large spontaneous splenorenal shunts (SSRSs) is a risk factor for poor portal vein flow and liver dysfunction. The disconnection of splenorenal shunts by left renal vein (LRV) ligation has been suggested as a potential solution for improving portal flow. We reviewed the hemodynamic consequences of splenorenal shunts in deceased donor liver transplantation and investigated the role of LRV ligation. In 10 patients who underwent liver transplantation at our institution between January 2006 and April 2010, an SSRS was diagnosed preoperatively. Intraoperative portal and hepatic artery flows were measured with a transit time flowmeter. The shunt was disconnected in 6 patients for whom the portal flow after reperfusion was less than or equal to 1200 mL/minute. LRV ligation resulted in significant increases in the portal flow. There were no differences in renal function for the patients who underwent renal vein ligation and the patients who did not undergo ligation. In conclusion, LRV ligation disconnects splenorenal shunts and modulates the portal inflow without any detrimental effects on renal function. Liver Transpl 17:891–895, 2011. © 2011 AASLD.

Portal hypertension leads to the development of an extensive network of portosystemic collateral vessels. In 5% to 12% of patients, these collaterals include spontaneous splenorenal shunts (SSRSs).1-4 In advanced stages of chronic liver disease, the portal vein (PV) flow may become hepatofugal, and this may lead to a portal steal phenomenon with a dramatic decrease in hepatic perfusion.3, 5-9 Not infrequently, this is associated with PV thrombosis.10-12

However, large portosystemic shunts such as SSRSs may jeopardize the portal flow and lead to graft dysfunction due to portal hypoperfusion.13

The optimal surgical approach to SSRSs has not been well defined.14 In order to determine when these shunts should be closed to prevent portal hypoperfusion, portal flow and portal pressure measurements should be obtained before and after reperfusion.15 Because of the buffer response, this should be correlated with measurements of the hepatic artery (HA) flow16 to provide a full assessment of the consequences of SSRS disconnection by left renal vein (LRV) ligation.

We present a series of SSRS patients undergoing orthotopic liver transplantation for whom intraoperative hemodynamic measurements were obtained.


ALD, alcoholic liver disease; CT, computed tomography; DBD, donation after brain death; DCD, donation after cardiac death; HA, hepatic artery; HCV, hepatitis C virus; IVC, inferior vena cava; LRV, left renal vein; MELD, Model for End-Stage Liver Disease; PBC, primary biliary cirrhosis; PV, portal vein; SSRS, spontaneous splenorenal shunt.


From January 2006 to April 2010, 263 adult liver transplants were performed in our unit. Ten patients (3.8%) were found to have an SSRS during the preoperative assessment. All patients who were assessed for liver transplantation in our program underwent routine Doppler ultrasound and triple-phase abdominal computed tomography (CT) examinations. CT scanning was performed with a Toshiba Aquilion 16 CT scanner (a 1-mm slice thickness was used for reconstruction of the arterial and venous phases). The contrast agent Omnipaque (300 mL) was used. The degree of collateralization and the presence of SSRSs were determined during the portal venous phase of the scan (60 seconds after the contrast injection; Fig. 1); an SSRS was defined as a communication with a diameter greater than 1 cm. The main demographic data are shown in Table 1. The latest follow-up data were recorded by August 1, 2010. Institutional Review Board approval is not required.

Figure 1.

CT image of an SSRS.

Table 1. Preoperative Demographics
 Patient 1Patient 2Patient 3Patient 4Patient 5Patient 6Patient 7Patient 8Patient 9Patient 10
Recipient  age (years)55675557516166654565
Type of graftWholeWholeSplitWholeWholeWholeWholeWholeWholeWhole
Pretransplant  PV thrombosisNoCompletePartialNoNoNoNoNoCompleteNo
MELD score1091923141312291618
Hepatopulmonary  syndrome+++

Nine patients received a full graft, and 1 patient received a split graft [an extended right lobe with segments I and IV-VIII, the main PV, the right HA, and the entire inferior vena cava (IVC), including the middle and right hepatic veins, with a graft-to-recipient weight ratio of 1.6]. One graft was from a donation after cardiac death (DCD) donor. We measured native and postreperfusion flows in the HA, PV, and portocaval shunt with a transit time flowmeter (MediStim, Oslo, Norway). PV and HA measurements were obtained and recorded for a non-SSRS cohort of 100 adult liver transplant recipients. The mean PV flow before the abdomen was closed was 1740 mL/minute, with only 10% of the cases having a flow less than or equal to 1200 mL/minute. The mean HA flow in these patients was 262 mL/minute. On the basis of these data and presumptive physiological data,17-19 the LRV was ligated when the PV flow after reperfusion was less than or equal to 1200 mL/minute.


The flow measurements are shown in Table 2. The PV flow increased significantly from a mean of 203 mL/minute (range = 50-525 mL/minute) in the native PV (data not shown) to a mean of 1173 mL/minute (range = 300-1670 mL/minute) after graft reperfusion. When the LRV was occluded, the PV flow increased further in all cases but one; the mean flow was 2077 mL/minute (range = 1245-3460 mL/minute).

Table 2. Flow Measurements
 Patient 1Patient 2Patient 3Patient 4Patient 5Patient 6Patient 7Patient 8Patient 9Patient 10Mean
  • *

    LRV ligation before PV reperfusion.

Pretransplant PV thrombosisNoCompletePartialNoNoNoNoNoCompleteNo
PV flow after  reperfusion (mL/minute)300*16701024156013708371200140012001173
PV flow after reperfusion with  LRV clamping (mL/minute)25002000199023313460211012451400139023502077
LRV ligationYesYesNoYesNoNoYesYesNoYes6/10

Initially, the decision to ligate the LRV was made shortly after portal reperfusion and before the arterial anastomosis. For this reason, HA flows (Table 3) after LRV clamping are missing for patients 1 and 2. Patient 2 had almost complete PV thrombosis, and after hepatectomy and thrombectomy, the LRV was disconnected before liver implantation to ensure an optimal flow at the time of reperfusion. In those patients for whom the arterial anastomosis had been completed before the occlusion of the LRV, the arterial flow decreased in response to the clamping from 289 (open LRV) to 175 mL/minute (clamped LRV; Table 3).

Table 3. Changes in the HA Flow in Response to LRV Clamping
 Patient 3Patient 4*Patient 5Patient 6Patient 7*Patient 8*Patient 9Patient 10*Mean
  • *

    LRV ligation.

  • Not obtained.

HA flow after reperfusion (mL/minute)120480254180440255300289
HA flow after reperfusion with LRV  clamping (mL/minute)98160120175120314243175

Overall, 6 of the 10 patients underwent LRV ligation. Patient 2 underwent retransplantation for HA thrombosis. This was related not to the LRV ligation but instead to an on-table thrombosis due to a hilar injury during organ procurement that was not amenable to reconstruction.

Currently, 9 of the 10 patients are alive with no evidence of further PV complications. LRV ligation had no impact on renal function; the postoperative serum creatinine levels of the patients with LRV ligation and the patients without LRV ligation were comparable, as shown in Fig. 2.

Figure 2.

Comparison of the mean serum creatinine levels of patients with shunt ligation and patients without shunt ligation.


Patients with chronic liver disease develop portal hypertension, which opens portocaval communications such as SSRSs. The incidence of radiologically diagnosed SRSSs in our group was 3.8%, although a much higher incidence has been reported in patients with cirrhosis.20

In this series, the presence of an SSRS was almost invariably associated with a reduction in the PV flow. Furthermore, 3 of the 10 patients had PV thrombosis (which was complete for 2 patients).

An almost 2-fold increase was noted when the shunt was occluded by LRV clamping. This effect was reversible. When HA measurements were carried out, the increase in the PV flow was followed by a reduction in the HA flow. The presence and degree of the arterial buffer response16 are important considerations when one is deciding whether an SRSS needs to be disconnected. A significant reduction in the HA flow may lead to an increased risk of biliary complications,21 although meaningful conclusions cannot be drawn from this small series.

Although there is not strong evidence to suggest that these shunts should be routinely closed after reperfusion, it seems reasonable to assume that the PV flow should be at least 1000 mL/minute.22-24 We decided to occlude the LRV when the flow was less than or equal to 1200 mL/minute. Only 2 studies using a transit time flowmeter (MediStim) and addressing the same topic in whole liver transplantation have been reported previously.8, 22 The authors reported an increase in the PV flow after the disconnection of the portosystemic shunts, and this was consistent with our observations.


We acknowledge that this is a small, uncontrolled case series and that a further evaluation of LRV ligation, ideally in a randomized study, should be undertaken. This study does suggest, however, that LRV ligation is safe and produces an increase in the PV flow. We were able to restore physiological PV flows and avoid potential complications due to poor portal flows (particularly PV thrombosis). Although this does not translate into a different transplant outcome, further studies of hemodynamics in deceased liver transplantation and the implications for long-term outcomes are warranted.