“Splenic artery steal syndrome” is a misnomer: The cause is portal hyperperfusion, not arterial siphon



Splenic artery embolization (SAE) improves hepatic artery (HA) flow in liver transplant (OLT) recipients with so-called splenic artery steal syndrome. We propose that SAE actually improves HA flow by reducing the HA buffer response (HABR). Patient 1: On postoperative day (POD) 1, Doppler ultrasonography (US) showed patent vasculature with HA resistive index (RI) of 0.8. On POD 4, aminotransferases rose dramatically; his RI was 1.0 with no diastolic flow. Octreotide was begun, but on POD 5 US showed reverse diastolic HA flow with no signal in distal HA branches. After SAE, US showed markedly improved flow, RI was 0.6, diastolic flow in the main artery, and complete visualization of all distal branches. By POD 6, liver function had normalized. RI in the main HA is 0.76 at 2 months postsurgery. Patient 2: On POD 1, RI was 1.0. US showed worsening intrahepatic signal, with no signal in the intrahepatic branches and reversed diastolic flow despite good graft function. On POD 7, SAE improved the intrahepatic waveform and RI (from 1.0 to 0.72). Patient 3: Intraoperative reverse diastolic arterial flow persisted on PODs 1, 2, and 3, with progressive loss of US signal in peripheral HA branches. SAE on POD 4 improved the RI (0.86) and peripheral arterial branch signals. Patient 4: US on POD 1 showed good HA flow with a normal RI (0.7). A sudden waveform change on POD 2 with increasing RI (0.83) prompted SAE, after which the wave form normalized, with reconstitution of a normal diastolic flow (RI 0.68). In conclusion, these reports confirm the usefulness of SAE for poor HA flow but suggest that inflow steal was not the problem. Rather than producing an increase in arterial inflow, SAE worked by reducing portal flow and HABR, thereby reducing end-organ outflow resistance. Evidence of this effect is the marked reduction of the RI after the SAE to 0.6, 0.72, 0.86, and 0.68, in patients 1-4, respectively. SAE reduces excessive portal vein flow and thereby ameliorates an overactive HABR that can cause graft dysfunction and ultimately HA thrombosis. Liver Transpl 14:374–379, 2008. © 2008 AASLD.

Hepatic artery (HA) complications can be a source of devastating morbidity and mortality following orthotopic liver transplantation (OLT). “Splenic artery steal syndrome” (SASS) is believed to lead to HA insufficiency, and in some cases, hepatic artery thrombosis (HAT).1 An important and often discounted determinant of HA flow is the degree of portal flow to the allograft. Portal hyperperfusion (PHP) is common posttransplantation, especially in patients with uncompensated severe portal hypertension. PHP is known to cause a decrease in HA flow due to the HA buffer response (HABR).2 We present here 4 whole-organ liver recipients with progressively worsening arterial perfusion and/or concomitant onset of reverse diastolic flow, 1 of whom also had early hepatic allograft dysfunction. Clinical characteristics and operative details appear in Table 1. All patients were treated with splenic artery embolization (SAE). Doppler ultrasonography (US) findings strongly implicate PHP and accentuated HABR as the cause of the clinical picture, which was heretofore known as SASS.

Table 1. Clinical Characteristics and Operative Details
PtAge/ SexMELDGraft weight (gr)Ischemia Cold/warm (min)BaselinePost-reperfusion FlowsPresence of Intraop/ Post Op reverse diastolic arterial flow
CO/CI (L/min)CVP (mmHg)PVHAHA Augmented*
368/F222302582/429.3/3.6101590132371YES Intra-op

HA, hepatic artery; HABR, HA buffer response; HAT, hepatic artery thrombosis; OLT, orthotopic liver transplantation; PHP, portal hyperperfusion; POD, postoperative day; PV, portal vein; RI, resistive index; SA, splenic artery; SAE, splenic artery embolization; SASS, splenic artery steal syndrome; US, ultrasound.


From February 2005, we routinely started to collect data on intraoperative liver hemodynamics. At 1 hour after reperfusion, the portal vein (PV) and the HA flows were measured with a transit time flowmeter (VeriQ System Medistim A/S, Oslo, Norway).

To evaluate the HABR, we also measured the HA flow after occluding the PV for 30 seconds. A complete set of intraoperative hemodynamic measurements was available on 143 patients and is reported in Table 2. Postoperative liver vascular Doppler US was also performed on postoperative day (POD) 1, and whenever clinically indicated thereafter. Patients with a resistive index (RI) of 1.0 were followed daily until normalization of the RI. A total of 15 patients out of 143 (10.4%) presented with an RI of 1.0 at the first Doppler US.

Table 2. Intraoperative Postreperfusion Hemodynamic Data on 143 Liver Transplant Patients
 Liver Weight (gr)CO (L/min)CIPV Flow (ml/min)PV Flow/100 g (ml/min)HA Flow (ml/min)HA Flow/100 g (ml/min)Augmented HA Flow (ml/min)Total Blood Flow (ml/min)
  1. Abbreviations: CO, cardiac output; CI, cardiac index; PVF, portal vein flow; HAF, hepatic artery flow; Augm, augmented; BF, blood flow; SD, standard deviation.

Standard deviation296.02.91.52740.259.1156.914.199.8728.1

We present here 4 of the 15 patients in whom the Doppler US detected a progressively worsening arterial perfusion and/or concomitant onset of reverse diastolic flow, and the interventions performed.

Patient 1

Patient 1 had Laennec's cirrhosis and a history of recurrent bleeding esophageal varices and hepatic encephalopathy. At 1 hour postreperfusion, PV and HA flows were 3.794 mL/minute (226 mL/minute/100 g) and 94 mL/minute (5.6 mL/minute/100 g), respectively. We measured HABR by clamping the PV for 30 seconds. This maneuver augmented HA flow by 25%, to 122 mL/minute. The HA waveform showed reversed flow during the diastolic phase when the PV was open; with PV clamping the waveform normalized (Fig. 1).

Figure 1.

Patient 1: Intraoperative post-reperfusion flow measurements with the portal vein open (left side) and clamped (right side). The arrow shows the reverse diastolic flow that disappears after clamping the portal vein.

US at 6 hours posttransplantation showed normal flow direction in all hepatic vessels. The main HA RI was 0.88, with PV velocity 115.6 cm/second (Fig. 2A). Over the first 48 hours post-OLT, aspartate aminotransferase/alanine aminotransferase, and bilirubin levels progressively decreased. On POD 3, the aspartate aminotransferase/alanine aminotransferase suddenly rose from 683/586 U/L to 1520/1275 U/L. Doppler US showed patent vasculature, but the main HA RI had increased to 1.00 (PV velocity: 98 cm/second). Venous outflow block and congestive heart failure were ruled out. Octreotide, 50 μg/hour, was started in an attempt to reduce splanchnic perfusion and, consequently, portal flow. After 24 hours of octreotide, Doppler US confirmed patency of all vessels but also showed new onset of reversed diastolic flow in the main and left HAs, with an end-diastolic flow in the HA of −12 cm/second (Fig. 2B).

Figure 2.

Patient 1: Hepatic Artery Doppler signal on POD 1(a), POD 2 (pre-SAE) (b) and POD 3 (post-SAE) (c). The arrow indicates the onset of reverse diastolic arterial flow on POD 2. SAE; Splenic Artery Embolization.

Given the patient's severe portal hypertension pre-OLT and high intraoperative PV flow (>2 cc/g/minute), we believed that PHP and HABR had a role in his HA flow derangement. We therefore proceeded with percutaneous SAE. HA angiography showed no structural lesion involving the anastomosis but confirmed extremely slow filling of the distal left and right intrahepatic arterial branches. Multiple coils were deployed in the splenic artery (SA), resulting in complete occlusion (Fig. 3). Doppler US at 12 and 48 hours after the procedure showed improvement of the RI in the main HA to 0.76 and 0.65 (Fig. 2C), respectively, with PV velocities of 32.6 and 28.7 cm/second (versus 98 cm/second prior to SAE). Following SAE, liver function tests improved and normalized. During this period, no changes in immunosuppression were made, suggesting that increases in RI were not related to rejection. The patient was discharged on POD 18 with normal liver function. At 1 and 2 months after SAE, US showed RI of 0.75 and 0.79, respectively.

Figure 3.

Patient 1: Pre-SAE (3a) and post-SAE (3b) selective HA angiograms. The pre-embolization angiogram is characterized by the lack of peripheral arterial perfusion consistent with severe vasoconstriction. The post-embolization angiogram shows reconstitution of normal peripheral arterial perfusion. SAE; Splenic Artery Embolization.

Patient 2

Patient 2 underwent combined liver/kidney transplantation for cryptogenic cirrhosis and end-stage renal disease secondary to light-chain nephropathy. At 1 hour postreperfusion, PV and HA flows were 2360 mL/minute (162.3 mL/minute/100 g) and 181 mL/minute (12.4 mL/minute/100 g), respectively. The HABR was high, with HA flow augmented by 110% to 387 mL/minute (on 30-second of PV occlusion) With the PV open, the HA waveform showed reversed diastolic flow; with the PV clamped, the waveform normalized.

US at 4 hours posttransplantation showed patency and normal flow direction in all hepatic vessels, with a main HA RI of 1.0 and PV velocity of 80.3 cm/second. US on PODs 2 and 3 showed worsening intrahepatic arterial flow, new onset of reversed diastolic flow in the proper HA (−8 cm/second), and progressive signal loss of the right posterior and left HA. To protect the renal allograft, octreotide was not given. During SAE, angiography showed a severely vasoconstricted HA with lack of opacification of the intrahepatic branches. Gianturco coils were deployed in the proximal portion of the SA with complete occlusion of the vessel. After SAE, angiographic visualization of the distal branches of the HA was dramatically improved. Doppler US at 12 and 48 hours after the procedure showed RIs of 0.8 and 0.72, respectively (PV velocity: 43.1 and 33.3 cm/second) The patient was discharged on POD 11 with normal liver function and no sequela of SAE.

Patient 3

Patient 3 underwent OLT for Laennec's cirrhosis with a Model for End-Stage Liver Disease score of 22 and signs of severe portal hypertension on his preoperative computed tomography scan. Intraoperative hemodynamic data are summarized in Table 1 (PV: 1590 mL/minute, 69.1 mL/minute/100 g; and HA: 132 mL/minute, 5.7 mL/minute/100 g). After reperfusion, US showed reverse diastolic HA flow that normalized after PV occlusion. A 3-fold increase in HA flow during PV clamping was considered an indirect sign of severe portal hypertension and exaggerated HABR. Doppler US studies on PODs 1, 2, and 3 showed persisting waveform alterations (RI 1.0) and increased PV velocity despite an octreotide drip. Loss of peripheral intraparenchymal arterial signals prompted us to perform SAE, after which diastolic flow normalized (RI 0.86) and PV velocity fell from 150 to 70 cm/second. The patient was discharged on POD 21 with perfect liver function.

Patient 4

Patient 4 underwent OLT for primary sclerosing cholangitis. After reperfusion, he had severe PHP (PV: 3200 mL/minute, 134.0 mL/minute/100 g; and HA: 262 mL/minute, 11.0 mL/minute/100 g) with a 20% increase in HA flow after PV clamping. Graft function improved but Doppler US on POD 3 showed significant changes in extrahepatic and intrahepatic arterial waveforms, with minimal diastolic flow, overall decreased systolic flow rate, and rapid systolic upstroke (RI 0.83), presumably related to excessive portal blood flow. With SAE, the waveform normalized (RI 0.68). The patient developed a low-grade fever that persisted for 2 days but disappeared spontaneously 3 days after SAE. He was discharged on POD 9 with normal liver function.

Figure 4 shows pre- and post-SAE HA RIs and PV velocities of all 4 patients.

Figure 4.

Hepatic Artery Resistive Index (RI) and Portal Vein velocity pre- and post-SAE (Splenic Artery Embolization).


Here, we have described attempts to identify and treat patients at high risk for HA complications secondary to PHP and exaggerated HABR. In all patients, the decision to not intervene with SA ligation was established based on our observation that the majority of patients with impaired arterial flow secondary to hyperperfusion do show signs of improvement on immediate postoperative Doppler US follow-ups. In addition to PHP, several factors contribute the production of a transient and reversible postreperfusion HA vasospasm, including the metabolic and hemodynamic state of the recipient as well as the use of marginal grafts. Furthermore, SA dissection and ligation may be technically challenging in patients with severe portal hypertension and peripancreatic collaterals.

In Patient 1, our decision to proceed with SAE was driven by early graft dysfunction in the context of high HA RI, reverse diastolic flow, and failed medical treatment. In patients 2, 3, and 4 the indication was loss of intrahepatic arterial signal on US and high RI despite normalizing liver function.

SASS, initially described in 1991 by Manner et al.,3 affects as many as 6 to 7% of OLT recipients. In the setting of severe hypersplenism, SA flow is thought to “steal” or siphon flow from the HA. Subsequent reports supported this notion, with angiographic evidence of HA hypoperfusion considered diagnostic. In fact, however, there is no real evidence that SASS is due to a siphon effect. In the largest series1 describing SASS and the use of SAE, intra- and postoperative flow data are missing, so conclusions about underlying pathophysiology cannot be drawn.

Lautt4 and Payen et al.5 added an important piece to the puzzle by describing the effect of changes in PV flow on HA flow, now known as the HABR. Adenosine, a potent vasodilator, is released at a constant rate around terminal hepatic arterioles and portal venules. As higher rates of portal flow wash out more adenosine, there is resultant HA vasoconstriction and reduced HA flow.

In partial liver grafts, high portal flow is thought to produce intrasinusoidal damage in 2 ways: first, by the direct effect of high portal pressure on sinusoids; and second, from indirect changes in HA flow; that is, vasoconstriction subsequent to an exaggerated buffer response. Demetris et al.6 described histological changes in partial grafts affected by PHP and small-for-size syndrome. First, PVP causes PV and sinusoidal endothelial cell injury. Second, PVP triggers the arterial buffer response, HA vasospasm, and decreased HA perfusion. Third, poor arterial flow and ischemia manifest in the periphery as centrilobular microvesicular steatosis or infarctions. Evidence for an important contribution from the HABR to PHP/small-for-size syndrome in these cases includes surgical observation of poor arterial flow, angiographic evidence of arterial vasospasm, and clinical observation of a higher incidence of HAT in small-for-size grafts.

Although PVP is well known in partial grafts, similar pathophysiology in whole grafts has not been reported. In cases with uncompensated portal hypertension and overflow, however, this same phenomenon may be responsible for early hepatic dysfunction and HAT. Extremely high portal flow and exaggerated HABR led us to suspect that our patients were at high risk of arterial hypoperfusion and possibly HAT. Close follow-up with Doppler US prompted early intervention with angiography and SAE.

In nontransplant patients, and in 1 transplant case report,7 drugs such as octreotride, terlipressin, and beta-blockers have reduced portal venous system pressure and bleeding.8, 9 Surgical techniques to modify PV inflow in allografts include SA ligation, splenectomy, and mesocaval, portocaval, or inferior mesenteric shunts.10–12 Troisi et al.13 reported that SA ligation produced a 33% mean reduction of PV flow in 10 right lobe recipients. Lo et al.14 salvaged a failing graft using SAL; PV flow fell by 60% and PV pressure from 27 to 15 mm Hg.

The complication rate of SAE is well reported in transplant patients and is much higher when the coils are placed distally. When coils are placed in the proximal SA, collateral flow to the spleen remains intact and the complication rate is negligible.1

In all 4 patients, HA vasoconstriction in response to PHP and exaggerated HABR produced high RI with poor intrahepatic arterial perfusion and reversed diastolic flow in 3 patients. SAE reduced the resistance to distal HA flow by reducing flow in the splenic circulation and consequently in the PV. Immediate post-SAE reduction in the RI and reestablishment of anterograde diastolic arterial flow suggest that HA insufficiency was remedied by reduction of portal overflow and end-organ resistance, as opposed to amelioration of a splenic siphon phenomenon. Although the conventional rationale for SAE in this setting invokes the notion of SA siphon that steals inflow, we show here that the problem is not one of competition for inflow but is instead due to marked increase in outflow resistance. We therefore propose to revise the name of “splenic artery steal syndrome” to “splenic artery syndrome.”

In conclusion, intraoperative measurement of hepatic vascular flow is crucial in identifying patients at high risk for developing major complications such as HAT. PHP with buffered intraoperative HA flows should trigger close Doppler US follow-up to detect changes compatible with splenic artery syndrome. SAE represents an effective option to decrease PHP and improve HA flow.