Effects of Hemi-Portocaval Shunts For Inflow Modulation on the Outcome of Small-for-Size Grafts in Living Donor Liver Transplantation


  • This paper was orally presented at the Annual Meeting of the American Transplant Congress, May 30–June 4, 2003, Washington, D.C.

*Corresponding author: Roberto Troisi, roberto.troisi@ugent.be


Graft hyperperfusion in small-for-size grafts (SFSG) is considered the main causal factor of small-for-size syndrome (SFSS). We compared SFSG with a graft-to-recipient body ratio ≤0.8, with and without graft inflow modulation (GIM) by means of a hemi-portocaval shunt (HPCS). Thirteen patients underwent adult-to-adult living donor liver transplantation (AALDLT): G1, n = 5 [4 right livers (RL) and 1 left liver (LL)] without GIM, and G2, n = 8 (4 RL and 4 LL) with GIM. In G2 patients, portal vein flow (PVF) was significantly reduced by HPCS: 190 ± 70 mL/min/100 g liver in G2 vs. 401 ± 225 ml/min in G1 (p = 0.002). One- and 6-month post-transplantation graft volume/standard liver volume (GV/SLV) ratio was of 72% and 79.5% in G1; 80% and 101% in G2 (p = ns). SFSS was observed in three G1 recipients (who were retransplanted), but in none of the G2 patients. At 1-year, patient and graft survival was respectively of 40% and 20% in G1, 87.5% and 75% in G2 (p = 0.024 and 0.03).

It is concluded that drastic reduction of PVF by means of HPCS improves overall patient and graft survival by averting the occurrence of SFSS. Graft inflow modulation through HPCS reduces the risk of complications when transplanting SFSG in adult recipients.


The introduction of adult-to-adult living donor liver transplantation (AALDLT) has opened new perspectives in liver transplantation. The procedure was widely practiced and achieved an exponential growth between 1997 and 2002. However, emerging ethical concerns regarding donor mortality rate and the most recent post-transplantation results suggesting that in AALDLT both patient and graft survival are inferior to those from cadaveric transplantation, have hampered its rapid expansion in the western countries. National surveys have failed to display the true impact of complications (biliary and vascular problems) that are frequently observed in AALDLT (1,2). Indeed, technical and functional complications may lead to graft loss, especially when small-for-size grafts (SFSGs) are transplanted into severely ill recipients (3–5). The transplantation of SFSGs may cause an unbalance between liver regeneration and an increased demand of liver function, which leads to a severe graft dysfunction known as small-for-size syndrome (SFSS) (6). Although its mechanisms are not yet fully understood, SFSS appears to be primarily linked to graft overperfusion (7). We describe a new technique of graft inflow modulation with a permanent-calibrated hemi-portocaval shunt (HPCS) and its impact on the outcome after transplantation of small-for-size grafts in adults.

Patients and methods


Among 50 AALDLT performed between September 1999 and July 2003, 13 (25%)—8 male/5 female, mean age 55 (± 5) years—had a graft-to-recipient body weight (GRBW) ratio ≤ 0.8. The main indication for AALDLT was post-necrotic HCV-related end-stage liver cirrhosis (n = 12, including 4 patients with coexisting hepatocellular carcinoma). One patient was transplanted because of a secondary biliary cirrhosis. From June 2002 onward, 8 out of 13 recipients with a GRBW ratio ≤0.8 were given prospectively a permanent calibrated HPCS in order to decrease portal graft inflow. For clinical and statistical analysis all patients were divided into two groups: G1, n = 5 recipients (transplanted before June 2002) with a mean age of 52 (± 2) years, receiving four right livers (RL) and one left liver (LL), mean GRBW ratio of 0.73 ± 0.10 (range 0.58–0.80), and G2, n = 8 recipients transplanted with a permanent HPCS, with a mean age of 56 (± 5) years, receiving 4 RL and 4 LL, with a mean GRBW ratio of 0.71 ± 0.10 (range 0.56–0.80). Pre- and post-transplant graft volumes were measured by multidetector CT-scan. Vascular and biliary anatomy was assessed with MRI and cholangio-CT (meglumin iotroxalate, Biliscopin® Schering AG, Germany).

Surgical technique

The donor hepatectomy was performed as described elsewhere (8). Following systemic infusion of 150 U/kg of heparin, the harvested grafts were weighed routinely before ex situ portal, arterial and biliary back table flushing. During the recipient hepatectomy the caval flow was preserved and a temporary end-to-side portocaval shunt was placed to decompress and drain splanchnic flow. This was taken off during the implantation and portal anastomosis in G1. In G2 patients, by contrast, either the left or the right portal branch was anastomosed in a permanent fashion ventrally to the inferior vena cava (IVC) with a PDS 6/0 running suture. According to the type of graft procured, either the right or left portal vein branch was shunted (left HPCS for RL and right HPCS for LL). In RL transplantation the venous outflow of the anteromedial sector was reconstructed on the back table with cryopreserved or autologous saphena conduit. When a consistent outflow was seen during back table flushing, additional accessory right veins were also separately anastomosed to the IVC. The policy to drain the anteromedial sector in RL grafts was scrupulously followed since the beginning of our ALDLT program whenever a significant outflow was found during back table flushing. This sector was drained with a venous conduit in two RL grafts in G1 and in three RL grafts in G2. The RL grafts without drainage of segments 5–8 displayed minimal outflow at the transection plane, homogeneous parenchymal perfusion at back table and a main drainage of this sector through the right and accessory hepatic veins. In these grafts indeed, two large right accessory veins more than 5 mm in diameter each were directly anastomosed to the recipient IVC to ensure optimal outflow. Left livers harvesting included always the caudate lobe and the middle hepatic vein (segments 1 to 4). They were engrafted in a piggy-back (end-to-side) fashion to the recipient IVC at the site of the confluence of the recipient's left and middle hepatic vein. The main vein draining the caudate lobe was anastomosed to the IVC through a small cavotomy in two cases, and was included in a common patch with the donor's left and middle hepatic vein in the two other cases. The portal flow drainage into the IVC through the HPCS was not discontinued in G2 during portal anastomosis (PDS 6/0 running suture) but temporarily cross clamped at the time of reperfusion to measure the basal portal flow values. Arterial anastomosis was done with interrupted stitches prolene 8/0. The biliary tree was reconstructed either with a Roux-en-Y loop or with a duct-to-duct (DD) anastomosis with a PDS 6/0 depending on the anatomic variations encountered in the grafts. DD anastomosis was preferred even in the presence of multiple (i.e. 3) biliary ducts. In all cases external biliary drainage was provided through a tiny feeding tube (Kendall Argyle Ch5, Tyco Healthcare, Gosport UK) inserted in the biliary tree and pushed transanastomotically. The immunosuppressive regimen was based on tacrolimus (Prograft®, Fujisawa) with reduced initial dosages (1mg each day for the first 2 days), mycophenolate mofetil (Cell Cept®, Roche) and a short course of steroids (withdrawn within 1 month) (8).

Peroperative systemic and hepatic hemodynamic assessment

Systemic hemodynamic parameters were continuously measured in the recipients by the thermodilution method (Swan Ganz catheter, Baxter Belgium). Flow measurements were performed intraoperatively to assess hepatic hemodynamics in both donors and recipients using the ultrasounds principle (transit time flow measurement by Medi-Stim AS, Oslo, Norway). The values were expressed as total mL/min per 100 g graft and as mL/min/cardiac index (CI). Following cholecystectomy and cholangiography, the right or the left portal vein flow (PVF) and the right or the left hepatic artery flow (HAF) were measured in the donors before starting the transparenchymal dissection. In the recipients, serial PVF and HAF readings were recorded after portal and arterial reperfusion, before and after biliary anastomosis, and before skin closure. In G2 patients, the hemi-portocaval shunt was calibrated (when required) with a 2/0 nonabsorbable ligature during continuous monitoring and after serial readings. The ligature encircled the portal vein branch anastomosed to the IVC with the intention to decrease excessive shunting and avoid a steal phenomenon. The goal of banding was to restitute a flow to the graft around two times the one recorded in the respective donor. Post-transplant graft blood flow velocity was assessed by US Doppler method twice a day during the first week, then twice a week and, subsequently, when indicated. It was expressed as mean velocity (Vm) and corrected by the angle (30° to 60°) between the long axis of the portal and arterial vessels and the Doppler beam. Graft regeneration was assessed by multidetector CT-scan using Philips Easy Vision, release 4.3 software (Philips Medical Systems, Best, The Netherlands). Volumetric assessment was obtained on post-operative days (POD) 10 and 30 and subsequently at 6 months. Variations in pre- and post-operative graft volume (GV) were measured and correlated to the standard liver volume (SLV) measured according to the Urata's formula (9).

Statistical analysis

Data were expressed as mean ± standard deviation (SD). Statistical analysis was performed using the SPSS 12.0 for Windows program. Parametric and nonparametric tests were used according to the distribution of the population. Dichotomous variables were tested by the Fisher's exact test. Associations between graft characteristics, hepatic hemodynamics, biochemical parameters, ascites leak and serum creatinine in both groups were tested by the Mann-Whitney U test. The graft and patient survival was determined by Kaplan-Meier survival analysis, and groups were compared by log-rank test. The statistical significance level was set at α= 0.05.


Overall median follow-up was 22.5 ± 11 months. No primary-non-function occurred and both groups were comparable in terms of age, MELD and UNOS score, and the degree of portal hypertension (Table 1). As for the graft characteristics, both groups were equivalent according to several anatomical and technical parameters. The only difference between the groups was the HPCS technique used for portal flow diversion in G2. The mean GRBW ratio was also similar (the lowest ratio was 0.58 in G1 and 0.56 in G2) (Table 2). The estimated pre-operative graft volumes measured by multidetector CT-scan were compared to the graft weight measured at the end of the donor hepatectomy and showed a difference of -16% and -14.2% in mean values in G1 and G2 respectively (overestimation of graft weight essentially due to the blood content).

Table 1.  Patient's demographics
VariableG1 (n = 5)G2 (n = 8)p*
  1. *Mann-Whitney U test; ** Fisher's exact test.

Mean age (years ± SD)52.0 ± 1.956.6 ± 5.20.093*
Median FU (months ± SD)10 ± 1317 ± 90.222*
UNOS Status 2(n)5/57/81.0**
MELD score (±SD)20 ± 718 ± 50.435*
Esophageal varices gr. 2–3581.0**
Right liver (n)440.565**
Left liver (n)140.565**
Table 2.  Graft characteristics (mean ± SD) (range)
VariableG1 (n = 5)G2 (n = 8)p*
  1. *Mann-Whitney U test; ** Fisher's exact test; GW = graft weight; GRBW = graft-to-recipient body weight ratio; RIHV = right inferior hepatic vein; CIT = cold ischemic time; WIT = warm ischemic time; PART = portal to arterial reperfusion time.

Mean GW (g)662 ± 152 (464–890)545 ± 188 (289–836)0.435
Mean GRBWR0.73 ± 0.10 (0.58–0.80)0.71 ± 0.10 (0.56–0.80)1.0
RIHV anastomosed2/5 (40%)4/8 (50%)1.0**
V5–V8 drainage2/5 (40%)3/8 (37%)1.0**
OP time (min)692 ± 98 (600–848)684 ± 172 (540–960)0.432
Blood Loss (mL)4300 ± 6037 (1000–15000)3857 ± 2462 (500–8000)0.530
CIT (min)191 ± 26 (160–210)189 ± 75 (75–300)0.833
WIT (min)58 ± 18 (45–90)68 ± 25 (41–100)0.724
PART (min)46 ± 10 (30–55)40 ± 11 (25–55)0.354

Systemic and hepatic hemodynamics

As expected, peroperative donor PVF values were overall significantly higher in the right livers than in the left livers. However, when expressed as volume per unit liver mass, no differences between right and left livers were found (Table 3). After graft reperfusion in the recipient, the average blood flow values recorded at the level of the right or left portal vein branch (vascularizing either the RL or the LL) were at least four times higher than those observed in their corresponding donors before hepatectomy (mean overall PVF values of 435 ± 282 mL/min/100 g liver). Left lobes in both groups showed absolute highest values when post-reperfusion PVF values were compared to the real graft weight (range 690–903 mL/min/100 g liver). Mean overall baseline donor's PVF values measured at the site that portal vein branches will be resected were of 120 ± 47 mL/min/100 g for LL and 117 ± 18 for RL (p = 0.53) (Table 3). Graft hyperperfusion was similar in both study groups or during the test of temporary occlusion of the HPCS in G2 patients before inflow modulation (p = ns). After removal of the occlusion clamp on the HPCS, the portal flow to the grafts fell to half of initial PVF values. This corresponded to approximately double the one recorded in the donors before hepatectomy. Whenever the portal flow after partial shunting fell under this limit (which is what could be assumed after a partial hepatectomy), the shunt was narrowed in order to reach a ‘more physiological’ portal flow expected after the reperfusion of an hemiliver. The target was indeed to obtain a PVF reaching two times the values recorded in the respective donor graft. After calibration the mean portal vein flow was of 190 ± 70 mL/min/g in G2 and 401 ± 225 mL/min/g in G1 (p = 0.002) (Table 4). These measurements were subsequently related to the recipient's cardiac index, which confirmed the statistically significant differences between the two groups.

Table 3.  Mean donor portal vein flow (PVF) values according to the right and left livers
VariableLeft Livers (n = 5)Right Livers (n = 8)p*
  1. *Mann-Whitney U test.

PVF (mL/min)462 ± 84785 ± 2290.010
PVF/GW (mL/min/100 g liver)120 ± 47117 ± 180.530
Table 4.  Hepatic hemodynamics in G1 and G2 and influence of the HPCS on portal and arterial graft inflow
VariableG1 (n = 5)VariableG2 (n = 8)p*
  1. *Mann-Whitney U test; a L/min/m2; b Graft weight (mL/min/100 g liver); c Cardiac index; ** PVF/GW vs. PVF2/GW.

  2. PVF = portal vein flow to the graft; PVF 1 = portal vein flow to the graft with hemiportocaval (HPCS) shunt cross-clamped; PVF 2 = portal vein flow to the graft after opening of the HPCS; HAF = hepatic artery flow to the graft; HAF = hepatic artery flow to the graft after opening of the HPCS.

Cardiac Index a4.8 ± 1.9Cardiac Index4.6 ± 1.00.876
PVF/GWb401 ± 225 (258–797)PVF1/GW537 ± 220 (303–903)0.143
 -PVF2/GW190 ± 70 (15–260)0.002**
PVF/GW/CI c82 ± 27 (52–102)PVF2/GW/CI42 ± 25 (3–71)0.018
HAF/GW9 ± 9 (2–23)HAF/GW32 ± 19 (10–63)0.030
HAF/GW/CI2 ± 2 (1–4)HAF/GW/CI7 ± 3 (2–16)0.030

Post-operative blood flow velocity

As stated above, US evaluations were routinely performed. These evaluations, each of which took approximately 20 min to be performed, were often difficult in G2 patients, particularly in those who received a left lobe, due to the small volume of the graft, its position and the presence of the HPCS. For this reason, in one patient, because of a suspicion of hepatofugal flow, a cavography on POD 1 was performed to confirm that PVF was normal. In G1 patients, portal vein Vm values were twice as high as those recorded in G2, while a poor arterial signal was also observed in this group (Table 5). These statistically significant differences persisted after the first post-operative month. Post-operative encephalopathy did not occur in the shunted recipients (prophylactic therapy with lactulose was not used). During the follow-up period, no signs of thrombosis of the shunts were observed (HPCS was easily recognized by CT-scan).

Table 5.  Post-operative mean portal vein and hepatic artery velocity on US Doppler evaluation in G1 and G2
Portal vein flow (cm/sec.)G1 (without shunt)G2 (with shunt)p*
  1. *Mann-Whitney U test; POD: post-operative day.

POD 173.4 ± 18.032.4 ± 17.70.006
POD 770.8 ± 9.531.0 ± 12.50.004
> 1 month65.0 ± 7.137.8 ± 9.00.095
Hepatic artery flow (cm/sec.)
 POD 132.6 ± 7.562.0 ± 25.10.065
 POD 731.0 ± 4.058.8 ± 20.20.004
 > 1 month30.0 ±14.161.8 ± 12.50.095

Graft regeneration rates

CT liver volumetric evaluation showed that shunting of the portal flow did not impair post-operative graft regeneration. Pre-operatively GV/SLV ratio was of 60% and 53% respectively for G1 and G2 (p = ns). One- and 6-month GV/SLV ratio increased to 72% and 79.5% in G1, 80% and 101% in G2 (p = ns). The lower the preoperative GV/SLV ratio, the more pronounced the regeneration was found to be.

Clinical and biochemical features, and outcome

One-year patient and graft survival was of 40% and 20% in G1, 87.5% and 75% in G2 (p = 0.024 and 0.03 respectively) (Figures 1 and 2). Overall per-operative hemodynamic features were characterized by enhanced graft portal hyperperfusion. In G1, these features persisted post-operatively on US evaluation. Massive post-operative ascites leak was observed in 4 (80%) out of 5 patients in G1. Occurrence of graft dysfunction characterized by a prolonged post-operative hyperbilirubinemia, increased prothrombin time and the appearance of renal insufficiency (SFSS) was found in three G1 patients, for whom retransplantation was needed (GRBW ratio of 0.58; 0.68 and 0.78) (Figures 3–6). Sepsis following retransplantation was the cause of death in two of these three patients. In the fourth patient in G1 (GRBW ratio of 0.8), arterial hypoperfusion was responsible for an anastomotic biliary ischemic leak. IVC thrombosis occurred one month after AALDLT leading to fatal massive pulmonary embolism. The last patient in G1 which was transplanted with a RL after starting the policy of inflow modulation for the first two LL by HPCS (GRBW ratio of 0.8) presented prolonged post-operative hyperbilirubinemia, which was successfully treated by four sessions of MARS dialysis. In G2, none of these functional post-operative disorders occurred. In this group, one right lobe—GRBW ratio of 0.8—and one left lobe—GRBW ratio of 0.59 were nevertheless lost. The six remaining grafts (GRBW ratio of 0.56; 0.63; 0.69; 0.77; 0.8; 0.8) recovered successfully. Although initial baseline post-reperfusion PVF values of G2 patients were similar to those recorded in G1 during occlusion of the shunt, no SFSS occurred after hemodynamic modulation of the PVF. In this group, in one recipient of a left liver (GRBW ratio of 0.63) massive ascites production (mean of 5 L/day) was concomitantly observed at the time of the onset of an acute cellular rejection (grade 2 to 3 according to the Banff's classification) on post-operative day 7. This complication was successfully treated with steroid pulse therapy and i.v. infusion of vasopressin (Pressyn® Sering, Toronto-Canada) to decrease the amount of ascites leak. Rejection rates in G1 and G2 were 40% and 37% respectively (p = ns). In one G2 patient (RL), occurrence of duodenal perforation and peritonitis led to secondary hepatic artery thrombosis. Fatal fulminant sepsis from staphylococcus aureus was recorded in another G2 patient, transplanted with a LL a few days following the treatment of steroid pulse therapy an acute rejection episode. In this patient the, GV/SLV ratio was increased from 43% to 62.9% within 2 weeks.

Figure 1.

Overall patient survival.

Figure 2.

Overall graft survival.

Figure 3.

Evolution of total bilirubin.

Figure 4.

Evolution of prothrombin time.

Figure 5.

Evolution of ascites leak.

Figure 6.

Serum creatinine at POD 21.


Since the late 1990s, AALDLT has gained rapid and widespread application in Western countries. It has offered a new solution to the problem of cadaveric organ shortage (1,2,7–11). In contrast to pediatric living donor transplantation, more than 1000 AALDLT procedures were performed in the United States and more than 500 in Europe before 2002. Today it is considered one of most effective alternative ways of reducing the waiting lists for selected adult recipients (10,12). However, from 2002 onwards its exponential growth has decreased markedly in Western countries, possibly as a consequence of a rising concern for donor safety and the published series of recipient survival outcomes and reported complications. AALDLT is a demanding procedure that still lacks standardized surgical approaches and has a high complication rate (i.e. biliary and vascular problems), which has been underestimated and whose true impact national surveys have failed to display (2). The technical and functional complications leading to graft dysfunction are frequently encountered, especially after transplantation of SFSG. One of the presumed causing factors is the concomitant increase in liver regeneration and high metabolic demand of small grafts. The liver function is hampered and the so-called SFSS develops with progressive cholestasis, persisting portal hypertension, ascites production associated with renal failure and coagulopathy (6). Inferior graft survival and higher mortality related to septic complications are common in this setting. In case of a GRBW ratio of > 1, SFSS may also occur in critically ill patients (UNOS status 2a) (3–5,13,14). The syndrome has also been observed after transplantation of aged and steatotic donor livers, in cases of insufficient venous drainage or prolonged warm ischemia (15–17). Its pathogenesis is still debated, though the syndrome appears to be primarily linked to graft overperfusion. In a rat model of SFSG transplantation, progressive damages linked to portal overperfusion have been described resulting in sinusoidal congestion, mitochondrial swelling, disruption of the sinusoidal lining cells and collapse of the space of Disse with impaired bile secretion and severe cholestasis (18). Significant increase of PVF in the post-reperfusion period is manifest when comparing donors and recipients' hemodynamic features in the clinical setting (7,19). Graft inflow modulation was first advocated as a way to overcome graft hyperperfusion and reduce portal pressure with SFSG (7). In a previous paper, we described the effects of splenic artery ligation (SAL) by decreasing the portal inflow and improving arterial signal on US Doppler evaluation. This simple procedure led to clinical benefits in term of the resolution of refractory ascites production, better liver function and eventually improved patient survival (20). SAL is sufficient to relieve portal hyperperfusion when PVF after transplantation do not exceed three to four times the flow recorded in portal vein branch of the donor liver lobe. When the post-reperfusion PVF values are more than four times higher than the one measured in the donor, as frequently seen in terminal liver failure, other approaches to liver hemodynamic modulation are needed. Partial portal blood flow diversion by temporary mesenteric caval shunt has been suggested (21). In our center, systematic partial portal blood flow diversion was performed since June 2002 using permanent HPCS whenever the PVF at reperfusion exceeded 3–4 times the one recorded in the donor. This technique allowed us to adjust the portal graft inflow to what would be expected in liver surgery after an equivalent liver resection. In normal circumstances (absence of cirrhosis and portal hypertension, cardiac index below 3 L/min/m2) a 60% hepatectomy results in a doubling of the portal flow in the 40% of remnant liver. This kind of hepatectomy can be followed by a transient minor degree of SFSS which usually resolves spontaneously within a few days. Remnant liver/body weight ratio after standard hepatectomy (equivalent to the GRBW ratio in the transplant setting) less than 0.8 as obtained after a 75% partial hepatectomy is followed by more pronounced and long lasting SFSS with much higher morbidity and mortality. The patient population in G1 and G2 was similar for most parameters, including MELD scores and general hemodynamic features. Hyperdynamic circulation with increased systemic and portal flow is known to persist initially after liver transplantation, especially in case of advanced post-necrotic liver cirrhosis with high portal hypertension and major splenomegaly (22–24). In this setting, post-operative US values of the PVF reach mean values of >70 cm/s (25,26). The use of HPCS results in a significant and persisting reduction of the portal inflow to more than 50% of the initial values. However, portal flow shunting could, in itself, constitute a risk of hepatofugal flow, which could in turn lead to graft dysfunction, liver atrophy or portal thrombosis. This is why peroperative hemodynamic detailed study with precise flow measurements is essential to determine whether an additional HPCS is needed on top of the existing spontaneous systemic variceal shunts. Peroperative monitoring is essential to calibrate the shunt and avoid graft hypoperfusion. During our follow-up period we recorded normal flow patterns without any steal phenomenon. Moreover, in our experience, HPCS did not jeopardize graft function nor regeneration. Theoretically, graft regeneration may be impaired by a side-to-side portocaval shunt (27). In our G2 patients, mean overall regeneration rates were even higher in the shunted patients than in those without portal flow diversion. Even in very small grafts regeneration rate was not impaired. While shear stress has been suggested as determinant for the initiation of liver regeneration in experimental models, overperfusion can be deleterious. In contrast to what might have been expected, portal blood flow diversion improved graft function without hampering regeneration (28). In the majority of G1 patients, graft dysfunction with persistent cholestasis and ascites leak with high PVF and poor arterial inflow was observed. This could have been the cause of the lower regeneration rate (26,28). Long-term outcome figures regarding the use of HPCS have not been reported so far. Hepatofugal flow and thrombosis of the shunt have to be looked for (29). The favorable evolution in G2 patients prompted us not to close the shunt. Conditions such as post-operative hemodynamic instability (i.e. acute right heart failure) and acute cellular rejection in the grafted liver are well known to induce portal hypertension which could than result in hepatofugal flow and graft dysfunction (30). The question remains whether and at which time point the HPCS should be removed. Recent reports have suggested that patient outcome of AALDLT in Child C patients is inferior compared with cases of less pronounced liver insufficiency (14,16,31). Analysis of failures in adults shows a high incidence of vascular and biliary complications as well as sepsis, mostly connected with small-for-size syndrome (32). Exclusion of patients with advanced liver disease or high MELD scores has therefore been recommended (33). The patient population in our study had high MELD scores, and disappointing results were indeed observed in G1 recipients who showed all the main features of SFSS with a very low 1-year patient and graft survival rate. One G2 patient receiving a LL exhibited unfortunately a fulminant sepsis the 3rd post-operative week due to staphylococcus aureus infection triggered by the steroid pulse therapy given to treat an acute cellular rejection grade 2. In this patient AALDLT had been technically uneventful and good graft regeneration was recorded at 2 weeks till the lethal infection episode. Small grafts are known to be susceptible to early sepsis, presumably because of the insufficient hepatic clearance of microorganisms by the reticulo-endothelial system during the early phases of hepatocyte regeneration (34). This can be considered as an important drawback of this technique if on a prospective basis similar infective complications would not be exceptional. Is there any reason to propose the use of SFSG (i.e. left lobes) in adults? The technical complexity of the use of right-lobe grafts (necessity of adequate venous drainage, frequent vascular and biliary anatomical variations complicating the implantation technique), and the presumed reduced donor morbidity are arguments which could advocate the use of left-lobe grafts in adults (35). This advantage of LL donation has been shown by an Asian survey, but not confirmed in the Western countries, presumably due to the small number of procedures performed (11,35–37). The European Liver Transplant Registry displayed indeed only a 1-year overall graft survival rate of 41% for left lobes vs. 73% for right lobes but, to our knowledge, no graft inflow modulation was used in these patients (12).

Hyperdinamic state as seen in end-stage liver disease can be responsible for reduced graft hyperperfusion after transplantation. Combination of both hyperdynamic state and small grafts transplantation can explain the less favorable results obtained in AALDLT in end-stage liver disease even with optimal venous outflow reconstruction. Hemi-portocaval shunt is an efficient and simple method to avoid post-reperfusion sinusoidal damages and parenchymal congestion. Whether this technique will allow more frequent use of left livers with equivalent results as obtained with the right lobes still need to be proven on a prospective basis. Under strict peroperative monitoring, hemodynamic modulation through HPCS allow the transplantation of small grafts in severely ill patients with acceptable graft and patient survival.