Split Liver Transplantation for Two Adult Recipients: An Initial Experience



The shortage of cadaver donor livers has been most severe for adult patients. Split liver transplantation is one method to expand the donor pool, but to have a significant impact on the waiting list, it needs to be applied for 2 adult recipients.

We split livers from 6 cadaver donors, and transplanted 12 adult recipients. All splits were performed in situ with transection through the midplane of the liver, resulting in a right lobe and a left lobe graft. Mean donor age was 19.7 years; mean donor weight was 79.1 kg. Mean recipient age was 41.5 years. Mean weight of right lobe recipients was 89 kg; left lobe recipients, 60 kg. All donors were hemodynamically stable and had normal liver function tests. Mean operative time for the procurement was 7.4 h. Average blood loss during the transection of the liver was 490 mL. Mean GW/RW ratio for all recipients was 0.87%; right lobe recipients, 0.86%; and left lobe recipients, 0.88%. With mean follow-up of 9.3 months, patient and graft survival rates were both 83.3%. There were 2 deaths: 1 after hepatic artery thrombosis (HAT) and subsequent multiorgan failure; the other after HAT, a liver retransplant, and subsequent gram-negative sepsis. The remaining 10 recipients are doing well. We observed no cases of primary nonfunction. Other complications included bile leak and/or stenosis (n = 3), bleeding from the Roux loop (n = 1), bleeding after percutaneous biopsy (n = 1), and incisional hernia (n = 1). In conclusion, split liver transplantation, using 1 cadaver liver for 2 adult recipients, can be performed successfully. Crucial to success is proper donor and recipient selection.


Liver transplantation has become a victim of its own success. With improvements in surgical technique, critical care, and immunosuppression, most centers now report 1-year patient survival rates in the 85% range. This highly positive outcome has resulted in an expansion of recipient selection criteria, leading to longer waiting lists and more debilitated candidates. Even though the number of liver transplants has increased over the last 10 years, it has not kept pace with the rapidly growing waiting list, leading to an ever-widening discrepancy between those waiting and those transplanted (1). In 1988, there were 3000 patients awaiting a liver transplant in the United States; there are now over 15 000. During that same period, waiting time has increased from a mean of 34–477 days, and the number of patients dying yearly on the waiting list has increased from 200 to 1300 (2).

Several methods have been employed to expand the donor pool, including the use of marginal donors as well as innovative surgical techniques, like living donor liver transplants (LDLTs) and cadaver split liver transplants (SLTs). The first successful LDLT was performed in 1989 (3). Since then, close to 2000 LDLTs have been performed – mostly for pediatric recipients (4,5) and more recently, for adult recipients (6–9). Despite encouraging results, considerable concern remains regarding the risks posed to the donor (10).

With SLTs, a whole adult cadaver liver is divided into two functioning grafts. The first SLT was performed by Pichlmayr et al. in 1988 (11). Subsequently, several centers have published their SLT series (12–14). The vast majority of SLTs have been between an adult and a pediatric recipient. The benefits for pediatric recipients have been tremendous, including expansion of the donor pool and a significant decrease in waiting times and mortality rates (15). Splitting an adult liver for pediatric recipients has no negative impact on the adult donor pool, but it does not increase it either. Yet adults now account for 96% of patients dying on the waiting list; in 1988, they accounted for only 70% (16). If SLTs are to have a significant impact on waiting list time and mortality, they must be performed so that the resulting two grafts can also be used in two adult recipients.

Herein, we describe our initial experience with SLTs for adult recipients. We analyzed our donor characteristics in detail in order to develop donor selection criteria. We also analyzed outcomes in our recipients, including surgical complications and graft and patient survival rates.

Patients and Methods

Between September 1, 1999 and March 31, 2001, we performed SLTs for 12 adult-sized recipients using six adult cadaver donors.

Donor characteristics

Characteristics of the six donors are summarized in Table 1. All six donors were young males (mean age = 19.7 years), with a slightly larger than average build (mean weight = 79.1 kg, mean height = 72.1 inches). Cause of death was head trauma for all six. All six had essentially normal liver function tests immediately before procurement and were hemodynamically stable with minimal inotrope use.

Table 1. Donor characteristics
Donor #123456Mean
Age (years)17221923201619.7
Weight (kg)92767688786579.1
Height (inches)73727375707172.1
Days hospitalized2223152.5
Sodium (mEq/L)147149160152140144148
Alanine transferase – ALT (U/L)34475556372843
Bilirubin (mg/dL)
Length of procurement (hours)
Blood loss before crossclamp (mL)400450500600500500490
Graft weight – right lobe (g)800740700760720790752
Graft weight – left lobe (g)470400475580550540510

Donor surgery

All six donors underwent a multiorgan procurement operation, with a thoracic and abdominal surgical team present. The dissection of the liver and transection of the parenchyma were performed in situ. No intraoperative angiography or cholangiography was performed. The hepatic dissection was similar to a living donor right hepatic lobe transplant (6,9). After complete mobilization of the right lobe, the right hepatic vein was isolated. Subsequently, the right hilar structures (right portal vein, right hepatic artery, and right hepatic duct) were isolated. The right hepatic duct was cut at the hilar plate. The parenchyma was then transected along the main portal fissure, generating a larger right lobe graft (segment V, VI, VII, VIII) and a smaller left lobe graft (segment I, II, III, IV), as shown in Figure 1. The middle hepatic vein was retained with the left graft.

Figure 1.

 The split through midplane, with right and left lobe grafts.

Before the vascular structures were divided, the liver and other abdominal organs were flushed with cold University of Wisconsin solution. The liver was then removed. The vessels to the right lobe, which had been previously isolated, were cut to completely separate the right and left grafts (Figure 2). The main hepatic arterial trunk, portal vein, common bile duct, and inferior vena cava were all maintained with the left lobe graft. For the right lobe graft, a segment of donor superficial femoral artery was anastomosed to the right hepatic artery, creating an extension. Both grafts were then weighed using a sterile scale.

Figure 2.

 The right and left lobe grafts after perfusion with University of Wisconsin solution and final separation.

Recipient characteristics

Characteristics of the six right lobe recipients are summarized in Table 2; six left lobe recipients, in Table 3. Mean recipient age was 41.5 years (range = 15–67 years). All recipients had advanced chronic liver disease and were Status 2B according to the United Network for Organ Sharing (UNOS). We had no Status 1 (fulminant hepatic failure) or Status 2A (severely decompensated chronic liver disease requiring intensive unit care) recipients.

Table 2. Right lobe graft recipient characteristics
Recipient #1357911Mean
  1. Alpha1= alpha1-antitrypsin deficiency

  2. PBC = primary biliary cirrhosis

  3. GW/RW = graft weight/recipient weight

  4. INR = international normalized ratio

  5. UNOS = United Network for Organ Sharing

Age (years)49385544474446
Weight (kg)1107070789811189
Liver failureHep CAlpha1cryptoPBCAlpha1PBC
UNOS status2B2B2B2B2B2B
Graft lobeRightRightRightRightRightRight
GW/RW ratio0.73%1.0%1.0%0.97%0.73%0.72%0.86%
Lab results at 1 week
Bilirubin (mg/dL)
ALT (U/L)664168106183300164170
Table 3. Left lobe graft recipient characteristics and outcomes
Recipient #24681012Mean
  1. Auto = autoimmune

  2. PBC = primary biliary cirrhosis

  3. GW/RW = graft weight/recipient weight

  4. INR = international normalized ratio

  5. UNOS = United Network for Organ Sharing

Age (years)67176743151537
Weight (kg)63585265506560
Liver failurePBCCFETOHHep BAutoAuto
UNOS status2B2B2B2B2B2B
Graft lobeLeftLeftLeftLeftLeftLeft
GW/RW ratio0.75%0.78%0.91%0.89%1.1%0.83%0.88%
Lab results at 1 week
Bilirubin (mg/dL)
ALT (U/L)4514310893201125119

Recipient consent

All recipients were informed that they could receive a split liver graft, and proper consent was obtained. The primary recipient (i.e., the one the liver was initially assigned to by the organ procurement organization [OPO]) was informed of the possibility of splitting the liver so that it could be used for another potential recipient as well. The risks of splitting the liver and the potential benefits to the second recipient were discussed with the primary recipient. Only if the primary recipient was agreeable to splitting was the second recipient contacted.

Recipient surgery

In right lobe recipients, hepatectomy was performed, with the inferior vena cava preserved. Portal bypass was used. The orifices of the middle and left hepatic veins were oversewn. The orifice of the right hepatic vein was extended inferiorly on to the inferior vena cava. The donor right hepatic vein was then sewn to the enlarged orifice of the recipient right hepatic vein, creating a large anastomosis. The right portal vein was sewn to the recipient common portal vein. The donor right hepatic artery, with the extension graft, was sewn to the confluence of the recipient right and left hepatic arteries. Biliary reconstruction was performed with a Roux-en-Y hepaticojejunostomy over a small feeding tube, which was externalized.

In left lobe recipients, a standard hepatectomy was performed with resection of the inferior vena cava. The left graft was sewn into place, replacing the recipient inferior vena cava. The hepatic arterial and portal venous anastomosis was performed in the standard manner. Biliary reconstruction was with a choledochocholedochostomy.

Postoperative care

Postoperative care was the same for both right lobe and left lobe recipients. Immunosuppression was with tacrolimus, mycophenolate mofetil, and prednisone. Acetylsalicylic acid (ASA, 325 mg per day) was used for thrombosis prophylaxis; acyclovir, for CMV prophylaxis.


All six donor operations were generally uneventful. All donors remained hemodynamically stable during the liver transection, with no need for significant increases in the dose of inotropes. Mean operative time for the entire procurement was 7.4 h, with blood loss of 490 mL during the parenchyma transection. During the procurement, five of the six donors each received 2 units of packed red blood cells. Arterial anatomy was normal in three donors. Of the remaining three, one donor had an accessory left hepatic artery originating from the left gastric artery, one had a replaced right hepatic artery arising from the superior mesenteric artery (SMA), and one had the common hepatic artery arising from the SMA with an accessory left hepatic artery arising from the celiac trunk. In all six left lobe grafts, there was a single bile duct to anastomose. Two right lobe grafts had a single bile duct; the remaining four right lobes had two separate bile ducts to anastomose.

Graft weight to recipient weight (GW/RW) ratios for all recipients are shown in Tables 2 and 3. Mean GW/RW ratio for all 12 recipients was 0.87%; right lobe recipients, 0.86%; and left lobe recipients, 0.88%. The lowest GW/RW ratio was 0.72% in an obese right lobe recipient.

The intraoperative course was uneventful for 11 of the 12 recipients. However, the first recipient developed a dissection of the native hepatic artery; therefore, the donor right hepatic artery (with an extension graft) was anastomosed directly to the recipient supraceliac aorta.

Mean cold ischemic time for all 12 recipients was 10.0 h; mean warm ischemic time, 55 min. Ischemic time did not significantly differ for left vs. right lobe recipients, because the two transplants were performed simultaneously in all cases.

All 12 recipients seemed to have good initial graft function, as evidenced by falling bilirubin levels and the INR. At 7 days post-transplant, mean INR was 1.3 in right lobe recipients and 1.2 in left lobe recipients (p = ns). Mean serum bilirubin level at 7 days post-transplant was 4.7 mg/dL in right lobe recipients and 1.6 mg/dL in left lobe recipients (p < 0.05). Of the 12 recipients, 11 were discharged from the intensive care unit (ICU) within 3 days post-transplant. However, the first recipient had severe hepatopulmonary syndrome, with progressively worsening hypoxia post-transplant that prevented extubation. This recipient (right lobe graft) eventually suffered hepatic artery thrombosis (HAT) at 4 days post-transplant and required a thrombectomy. She went on to develop multiorgan failure and died at 9 days post-transplant. The other recipient death was in a left lobe recipient who suffered HAT at 3 weeks post-transplant, underwent an urgent retransplant with a whole cadaver graft, but died from gram-negative sepsis soon after the retransplant. With mean follow-up of 9.3 months, the remaining 10 recipients are alive and doing well (patient and graft survival rates of 83%). All 10 have functioning grafts with normal liver function.

Surgical complications in right lobe recipients included HAT (n = 1), relaparotomy for bleeding after a percutaneous liver biopsy (n = 1), relaparotomy for bleeding from the jejunal anastomosis of the Roux-en-Y loop (n = 1), and a bile leak, from the cut surface, that closed spontaneously with drainage alone (n = 1). Complications in left lobe recipients included HAT (n = 1), an anastomotic bile leak requiring hepaticojejunostomy (n = 1), and an incisional hernia at 5 months post-transplant (n = 1).

Medical complications affected three of the 12 recipients (two left lobe, one right lobe), all of whom had tissue-invasive cytomegalovirus (CMV) disease (at 2, 3, and 5 months post-transplant). Only one recipient (right lobe) has had an acute rejection episode, at 6 weeks post-transplant, which readily responded to treatment with intravenous corticosteroids.

Median length of hospital stay for the 12 recipients was 17 days: 17 days for right lobe recipients and 17 days for left lobe recipients (p = ns).


Methods for expanding the donor pool for liver transplants include using marginal donors (17,18), using living donors, and splitting cadaver livers. Most transplant centers already have fairly liberal criteria for accepting a cadaver donor; and using marginal donors more frequently is unlikely to increase the cadaver pool by more than 5% to 10%. Using living donors for liver transplants in adult recipients shows promise; however, concern regarding the short- and long-term consequences for such donors currently prevents their widespread acceptance (19). SLTs have been performed over the last 12 years, but usually for one adult and one pediatric recipient. Yet the vast majority of patients on the waiting list are adults, and 96% of the deaths on the waiting list are of adults. To have a significant impact on the waiting list, SLTs also need to be performed for two adult recipients. Recently, successful splits for two adult-sized recipients have been reported (20,21).

Transection in the midplane divides the liver into the anatomic right lobe (60% of the liver) and the left lobe (40% of the liver). The minimum amount of liver mass needed to sustain life immediately post-transplant is unclear. Some experience with LDLTs suggests that a GW/RW ratio of 0.8% is the minimum (22,23). For cadaver donors, the minimum amount of liver mass may also be influenced by such factors as donor hemodynamic stability and cold ischemic time. In our small series, two of the six left lobe recipients had a GW/RW ratio of less than 0.8% (0.75% and 0.78%); both had good hepatic function immediately post-transplant. Of the six right lobe recipients, three had a GW/RW ratio of less than 0.75%, but all three were obese; therefore, this ratio may not be an accurate indication. One of these obese recipients died early post-transplant. The cause of death was severe hepatopulmonary syndrome with worsening hypoxia, HAT, and subsequent multiorgan failure. Before the thrombosis, however, this recipient had evidence of good graft function.

Nonetheless, a GW/RW ratio of close to 0.8% should likely be the minimum when selecting appropriate recipients. In an average individual, the liver generally comprises 2% of the total body weight: the right lobe, 1.2%; the left lobe, 0.8%. The right lobe tends to be fairly constant in size: from an average-sized donor (70 kg), the right lobe should weigh 800–900 g. That weight should be sufficient for transplantation into most adult recipients. The weight of the left lobe tends to vary more (depending on the size of the lateral segment), but from a 70-kg donor, should be in the range of 400–500 g. That range should be sufficient for transplantation into a small adult recipient (up to 65 kg). All six of our left lobe recipients were smaller, with a mean weight of 60 kg.

Graft size is not the only criterion in selecting donors and recipients. Donors should be medically ideal to minimize the risks of primary nonfunction, especially for left lobe recipients. In a review of SLTs for both adult and pediatric recipients, Busutill et al. made several recommendations regarding suitable donors (15). Likely, the same criteria are appropriate for selecting SLT donors for two adult recipients. Young, hemodynamically stable donors with normal liver function test results should be selected; with such donors, primary nonfunction for the recipients should be uncommon. In cadaver transplants, prolonged cold ischemic time is also a known risk factor for primary nonfunction (24). Therefore, cold ischemic time should be minimized as much as possible in all SLT donors. For this reason, it is preferable to do the actual transection of the parenchyma in situ in the donor. Performing the split on the back table could add up to 2–3 h of cold ischemia. Also, there is likely some warming of the liver on the back table, even if the split is being performed in a cold ice bath of University of Wisconsin solution. Even a warming by a few degrees of the liver may have a negative impact on the outcome (25).

Performing the split in situ also has other advantages. Significantly less bleeding occurs when the organs are reperfused. The two liver grafts can be assessed in the donor immediately after parenchymal transection and before vascular interruption, to ensure no significant ischemia at the borders of the cut edge of the liver. Several retrospective analyses of large SLT series have shown superior results with in situ vs. ex situ liver splits (26,27). For all these reasons, we feel that the actual splitting of the donor liver should be performed in situ.

While donor selection is important, recipient selection is also crucial. Based on poor results with living donor transplants for critically ill recipients with chronic liver disease (UNOS Status 2A), we have chosen to limit our recipients to UNOS Status 2B. Our two deaths were in recipients with very advanced disease, significant chronic debilitation, and multiple other medical problems. Poor results have also been reported with living donor transplants for critically ill recipients with chronic liver disease (19).

Another important aspect of the recipient selection process is adequately informing the potential recipient of the splitting procedure and obtaining informed consent. With the current organ allocation system, the graft is initially assigned to a primary recipient. If the liver is to be split, the second recipient is chosen at the discretion of the center performing the split. This is advantageous for second recipients, who then bypass additional waiting time. For primary recipients, however, there is no significant advantage: we are, in fact, asking them to give a part of ‘their’ new liver to someone else. If the primary recipient is to receive the left lobe (and is therefore likely of smaller size), the issue is not as difficult. A full-size liver graft from a large donor may be difficult to fit into this recipient. Performing the split may allow for a better size match. But if the primary recipient is to receive the right lobe, the issue is more difficult, since such recipients could easily accommodate the whole graft. This specific situation arose in three of the six splits we performed. The primary recipients were fully apprised of the risks of splitting and were informed that the split would only be performed with their consent. In all three cases, there was no hesitation on the part of the primary recipient to provide consent. In fact, two commented that they tremendously appreciated the opportunity to help another individual by agreeing to the split. Nonetheless, with the present allocation rules, it is crucial to fully inform potential recipients about splitting and to obtain informed consent.

Several technical points need emphasis regarding the donor operation, which is very similar to a right lobe liver procurement from a living donor. The transection plane should stay to the right of the middle hepatic vein, so that this structure is retained with the left lobe. Segment IV makes up a crucial part of the left lobe, and hence the middle hepatic vein should be preserved with the left lobe to ensure no congestion. Loss of the middle hepatic vein does not significantly affect drainage of segment V and VIII in the right lobe graft, as these segments drain adequately via the right hepatic vein (28).

Regarding the dissection in the hilum, our preference has been to leave the full length of the hilar structures intact with the left lobe, i.e. the common hepatic artery with the celiac trunk, main portal vein, and common bile duct. The right lobe then retains just the right-sided structures: the right hepatic artery, right portal vein, and right hepatic duct. The right-sided hilar structures are usually larger than the left-sided structures. Therefore, leaving the main vessels intact with the left lobe makes that transplant easier. Performing the split in this fashion will often leave two separate bile ducts on the right lobe, a finding that is supported by the LDLT literature (19).

One crucial technical point for the recipient operation is ensuring adequate venous outflow of the graft to prevent congestion. Doing so is especially important with a right lobe graft, where the outflow almost entirely depends upon the main right hepatic vein. Properly fashioning the anastomosis of the donor right hepatic vein to the recipient inferior vena cava is probably the key factor in avoiding congestion of the right lobe. There should be no narrowing of this anastomosis; opening up the recipient vena cava inferior to the right hepatic vein orifice ensures a large, patulous anastomosis. Reanastomosing any inferior hepatic veins from the graft that are greater than 5 mm in diameter is also important to aid in outflow (29). In the living donor literature, some authors have recommended reimplanting hepatic veins from segment V and VIII using a vein graft, or including the middle hepatic vein with the graft (6,30). Not all favor this approach, however: some feel that these segments drain adequately via the right hepatic vein (28). We have not found it necessary to reimplant these veins in any of our six right lobe graft recipients, and feel that the key is a proper right hepatic vein anastomosis.

Surgical complications are probably more common in SLT (vs. whole graft) recipients. In our series, we had two cases of HAT and three bile leaks. A number of our recipients required relaparotomy because of a surgical complication. The cut surface of the liver and the smaller vessel size in right lobe grafts likely contribute to a higher surgical complication rate.

Overall results in our series were encouraging. Even though follow-up is short, patient and graft survival rates at just over 9 months post-transplant were acceptable: 83%. Despite two deaths, it is important to note that the remaining 10 recipients are currently doing well, with good hepatic function. And, immediately post-transplant, we had no cases of primary nonfunction.

In conclusion, our very small series demonstrates the technical feasibility of reliably splitting one cadaver liver for two adult recipients. More data are needed to better define donor and recipient selection criteria, which are crucial to success. It is difficult to estimate how much impact adult SLTs will have on the donor pool. About 25% of all cadaver donors in the United States are between 15 and 35 years old. If most of these livers could be used for splits, the number of liver transplants would increase by 25%, or by close to 1000. With better preservation techniques and with new methods to accelerate liver regeneration, it may be possible to increase the number of livers that would be amenable to splitting (31). In the near future, this technique will likely become part of every major liver transplant center's repertoire, in order to provide the maximum advantage for their candidates on the waiting list.


Special thanks to Mary Knatterud and DeAnn Ronning for their help in the preparation of this manuscript.