Graft weight/recipient weight ratio: How well does it predict outcome after partial liver transplants?



Partial graft liver recipients with graft weight/recipient weight (GW/RW) ratios < 0.8% are thought to have a higher incidence of postoperative complications, including small-for-size syndrome (SFSS). We analyzed a cohort of such recipients and compared those with GW/RW < 0.8% to those with GW/RW ≥ 0.8%. Between 1999 and 2008, 107 adult patients underwent partial graft liver transplants: 76 from live donors [living donor liver transplantation (LDLT)] and 31 from deceased donors [split liver transplantation (SLT)]. Of these, 22 had GW/RW < 0.8% (12 with LDLT and 10 with SLT), and 85 had GW/RW ≥ 0.8% (64 with LDLT and 21 with SLT). The baseline demographics and median length of follow-up were similar. SFSS developed in 3 recipients with GW/RW < 0.8% (13.6%) and in 8 recipients with GW/RW ≥ 0.8% (9.4%; P = not significant). Other early complications were similar between the 2 groups. Inflow modification with splenic artery occlusion was performed in 13 recipients: 7 with GW/RW < 0.8% and 6 with GW/RW ≥ 0.8%. Graft survival at 1 year post-transplant did not differ (91% versus 92%; P = not significant). In conclusion, GW/RW did not appear to be the only determinant of outcome after partial liver transplantation. Using techniques such as inflow modification may help to prevent some of the problems seen with smaller grafts. Liver Transpl 15:1056–1062, 2009. © 2009 AASLD.

Living donor liver transplantation (LDLT) and deceased donor split liver transplantation (SLT) are increasingly being used at some centers to address the shortage of donor organs. Both techniques result in a partial graft that has a reduced overall parenchymal mass compared to the whole organ allograft. Smaller grafts incapable of meeting all of the metabolic, synthetic, and hemodynamic demands of the recipients have been implicated as a cause of early allograft dysfunction. The constellation of persistent ascites, cholestasis, and coagulopathy in the setting of a reduced size graft without an obvious technical cause has been termed small-for-size syndrome (SFSS).

Small graft size is thought to be the major risk factor for developing SFSS and has been associated with inferior outcome. Kiuchi et al.1 analyzed 276 living donor liver recipients and found significantly worse graft survival for patients with a graft weight/recipient weight ratio (GW/RW) < 0.8%. Similarly, Sugawara et al.2 reported that a graft volume/standard liver volume ratio (GV/SLV) < 40% was associated with decreased survival and prolonged recovery of liver function tests. In contrast, a more recent study by Shimada et al.3 found no significant difference in graft survival for recipients with GV/SLV < 40% compared to those with GV/SLV > 40%. This discrepancy may highlight the importance of other factors responsible for graft dysfunction in the small-for-size graft. Local hemodynamic effects such as portal hyperperfusion,4 impaired venous outflow,5 and recipient disease severity6 are also known to contribute to graft injury and therefore influence critical graft size.

Thus, the pathogenesis of SFSS and the various factors that contribute to critical graft size remain poorly defined. This study was undertaken to assess the impact of graft size on the incidence of SFSS and recipient outcomes after partial graft liver transplants.


EtOH, ethanol; GV/SLV, graft volume/standard liver volume; GW/RW, graft weight/recipient weight; ICU, intensive care unit; ILV, ideal liver volume; INR, international normalized ratio; LDLT, living donor liver transplantation; LL-SLT, left lobe split liver transplant; MELD, Model for End-Stage Liver Disease; NA, not available; NS, not significant; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; RL-LDLT, right lobe living donor liver transplant; RL-SLT, right lobe split liver transplant; SAE, splenic artery embolization; SAL, splenic artery ligation; SD, standard deviation; SFSS, small-for-size syndrome; SLT, split liver transplantation.


A retrospective database analysis was performed of all partial graft liver transplants performed in adult recipients at our center between January 1999 and October 2008. Recipients were analyzed in 2 groups: those with GW/RW < 0.8% and those with GW/RW ≥ 0.8%. The 1999 start date was chosen as this was when our living donor and split programs were initiated. A minimum of 12 months' follow-up was available for all recipients. The endpoints of graft and patient survival rates were examined as well as the incidence of specific postoperative surgical complications.

Surgical Technique

Living Donor Transplants

All living donor transplants involved a donor right-lobe hepatectomy (segments V, VI, VII, and VIII) with preservation of the middle hepatic vein in the donor and reconstruction of large (≥5-mm) segment V or VIII outflow vessels in the recipient, for which reversed saphenous vein grafts or preserved vascular grafts were used. Using volume estimates obtained from the preoperative computed tomography scan, we tried to choose donors only if the GW/RW ratio would be >0.8%.

Split Liver from a Deceased Donor

Deceased donor SLT for adult recipients involved in situ splitting of the liver in its midplane, which generated a right lobe graft and a left lobe graft that could be used for 2 adult recipients.7, 8 The middle hepatic vein was retained with the left graft. The implantation of the right and left lobe grafts was performed simultaneously by 2 separate surgical teams. The grafts were usually implanted in a piggyback fashion with the hepatic arterial and portal venous anastomosis performed in the standard manner. We attempted to keep the GW/RW ratio at greater than 0.8% for all such transplants. Excluded from this group and analysis were a very small number of adult recipients that had received an extended right lobe graft (as part of an adult/pediatric split).


Immunosuppression (at the time of transplant and maintenance) was similar in all recipients and did not differ by donor source, type of transplant, or etiology of liver disease. Post-transplant, recipients began a triple immunosuppressive regimen consisting of tacrolimus, mycophenolate mofetil, and prednisone. In most recipients, the steroids and mycophenolate mofetil were weaned rapidly, so the majority of recipients were on tacrolimus monotherapy by 6 months post-transplant.

Postoperative Complications

We looked at the incidence of certain surgical complications in the 2 groups. Specifically, we looked at the incidence of major postoperative infections (minor infections such as urinary tract infections were not included), postoperative hemorrhage (defined by the need for re-exploration or transfusion of more than 4 units of packed red cells), biliary complications (leaks or strictures), and vascular thrombosis (arterial or portal). One specific complication that was examined in detail was the incidence of SFSS. Although there are no well-accepted criteria in the literature for the diagnosis of SFSS, we defined it by the presence of significant cholestasis with serum bilirubin > 10 mg/dL (and continuing to increase) after postoperative day 7, coagulopathy with an international normalized ratio > 1.5, and ascites with drain output > 2 L/day in the absence of an obvious technical problem such as a biliary leak, vascular thrombosis, or stenosis. We chose this definition on the basis of the most consistent findings reported in the literature of hyperbilirubinemia and ascites with this syndrome.9, 10 Ruling out mechanical/technical problems with imaging techniques such as ultrasound, computed tomography scanning, and hepatobiliary iminodiacetic acid scanning was felt to be an important part of establishing the diagnosis. A liver biopsy was obtained in some of these recipients, but after a bleeding complication, we now generally rely on the clinical features to establish the diagnosis.

Statistical Analysis

Categorical variables were analyzed with the chi-square test and, when applicable, Fisher's exact test. Continuous variables were analyzed parametrically with the Student t test. We estimated actuarial patient and graft survival rates by using Kaplan-Meier life table analyses. To compare differences between groups, we used log-rank and Wilcoxon tests.


Between January 1999 and October 2007, there were 107 partial liver transplants performed in adult recipients at our institution. Of these, 76 were from live donors, and 31 were split livers from deceased donors. There were 22 (21%) patients categorized as smaller graft recipients with GW/RW < 0.8% and 85 (79%) with GW/RW ≥0.8%. The number of LDLT and SLT constituents in each of these 2 groups is shown in Fig. 1. There was a slight predominance of SLT in the smaller graft group, but this difference did not reach statistical significance (P = 0.10).

Figure 1.

Study population. Abbreviations: GW/RW, graft weight/recipient weight; LDLT, living donor liver transplantation; SLT, split liver transplantation.

Baseline demographics for the 2 groups are shown in Table 1. The Model for End-Stage Liver Disease (MELD) score reported in the table is the actual calculated value at the time of transplant and does not include exception points or tumor points. This more accurately represents the degree of hepatic dysfunction and clinical condition of the patient at the time of transplant. Differences in mean donor and recipient age, MELD, and median follow-up for the 2 groups were not statistically significant. The degree of portal hypertension (as estimated by the actual MELD score, the percentage of patients with ascites, and the percentage of patients with previous variceal bleeds) seemed to be similar between the 2 groups. None of the grafts in either groups had any significant (>5%) degree of steatosis (which represents a contraindication to SLT or LDLT at our center). The mean GW/RW ratio for the smaller graft group was 0.71 (range, 0.51–0.79) and for the larger graft group was 1.21 (range, 0.81–1.9; P < 0.001).

Table 1. Baseline Characteristics of the Two Groups
 GW/RW < 0.8% (n = 22)GW/RW ≥ 0.8% (n = 85)P Value
  1. Abbreviations: EtOH, ethanol; GW/RW, graft weight/recipient weight; MELD, Model for End-Stage Liver Disease; NS, not significant; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; SD, standard deviation.

Donor age (±SD)29.8 ± 11.134.8 ± 10.80.062
Recipient age (±SD)52.7 ± 7.3549.6 ± 10.50.196
Mean MELD15.717.70.195
Mean GW/RW0.711.21<0.001
Median follow-up (months)55.344.40.116
% with ascites at the time of transplant38%35%NS
Cause of liver disease   
 Hepatitis C36%38%NS
% with a history of variceal bleeding18%22%NS

Patient and Graft Survival

The Kaplan-Meier patient survival curves for the 2 groups are shown in Fig. 2. There was no statistical difference in patient survival rates between the 2 groups, with greater than 90% patient survival in both groups at 1 year post-transplant. Likewise, graft survival rates were no different between the 2 groups. At 1 year post-transplant, graft survival was 91% in the group with GW/RW < 0.8% and 92% in the group with GW/RW ≥0.8% (P = not significant).

Figure 2.

Patient survival curves for recipients based on the size of the partial graft: a comparison of small grafts (GW/RW < 0.8%) and larger grafts (GW/RW ≥ 0.8%). Abbreviations: GW/RW, graft weight/recipient weight; NS, not significant.

Postoperative Complications

The incidence of infections, hemorrhage, bile duct complications, and vascular thrombosis was not different between the 2 groups (Table 2). The incidence of SFSS was also not significantly different between the 2 groups. In the GW/RW < 0.8% group, there were 3 (13.9%) cases of SFSS, whereas in the larger graft size group, there were 8 (9.4%) cases (P = 0.69). The length of the intensive care unit stay for the smaller graft recipients was slightly longer, although this did not reach statistical significance (P = 0.08). The overall hospital length of stay was no different for the 2 groups (P = 0.27). Measures of graft function including serum bilirubin, international normalized ratio, and creatinine were not significantly different between the 2 groups (P = not significant; Table 2).

Table 2. Outcomes for the Two Groups
 GW/RW <0.8% (n = 22)GW/RW ≥0.8% (n = 85)P Value
  1. Abbreviations: GW/RW, graft weight/recipient weight; ICU, intensive care unit; INR, international normalized ratio; SFSS, small-for-size syndrome.

Graft survival at 3 months95.5%96.5%1.0
Graft survival at 1 year90.9%91.8%1.0
SFSS3 (13.6%)8 (9.4%)0.85
ICU stay (days)
Total length of stay (days)17.314.00.27
Labs at 1 month post-transplant   
 Serum bilirubin (mg/dL)
 Serum INR1.21.10.65
 Serum creatinine (mg/dL)
Infectious complication6 (27.3%)29 (45.9%)0.148
Hemorrhage8 (36.4%)23 (27.1%)0.434
Biliary complication3 (13.6%)26 (30.6%)0.177
Arterial thrombosis3 (13.6%)3 (3.5%)0.099
Portal vein thrombosis1 (4.5%)5 (5.9%)1.00

The distribution of GW/RW values for the entire group, including the cases that developed SFSS, are shown in Fig. 3. This demonstrates that although the cases of SFSS seemed to be more clustered around the smaller graft sizes, there were a number of cases of this even when the GW/RW ratio was >1.0%. Examining this in further detail, however, we noted that although the overall incidence of SFSS in patients with smaller grafts (GW/RW < 0.8%) was not that high, patients with very small grafts (GW/RW < 0.7%) had a relatively high incidence of SFSS [3 of 8 recipients (37%)]. As illustrated in Fig. 4, the incidence of SFSS seemed to be highest in the smallest grafts, although there were a number of cases in the larger size grafts.

Figure 3.

Distribution of GW/RW ratios and the occurrence of SFSS. Abbreviations: GW/RW, graft weight/recipient weight; SFSS, small-for-size syndrome.

Figure 4.

Distribution of the incidence of SFSS based on the initial GW/RW ratio. Abbreviations: GW/RW, graft weight/recipient weight; SFSS, small-for-size syndrome.

The outcomes for the 11 patients with SFSS are shown in Table 3 and Fig. 5. Three of the recipients had received split liver grafts, whereas the remaining 8 were right lobe living donor recipients. The mean GW/RW ratio for these 11 recipients was 0.93. No intervention was performed in the first 2 recipients (this was early in our experience and was prior to a clear understanding or description of SFSS in the literature). Both of these recipients died early because of liver failure and ultimately sepsis. The remaining 9 recipients had either splenic artery ligation (SAL) or embolization performed at a mean of 8 days post-transplant for treatment of SFSS. A good response was obtained with improvement of liver function in all except 1 of the recipients, who ultimately required retransplantation at 2 months post-transplant for failure to thrive. The other 8 recipients demonstrated normalization of liver function tests by 1 month post-transplant (Fig. 5) and normal liver regeneration as demonstrated by a volumetric computed tomography scan 3 months post-transplant (Table 3).

Table 3. Clinical Summary of the 11 Recipients with SFSS
PatientGraft TypeGW/RW RatioIntervention for SFSSGraft Size at 3 Months Post-TransplantILV (%)Outcome
  1. Abbreviations: GW/RW, graft weight/recipient weight; ILV, ideal liver volume; LL-SLT, left lobe split liver transplant; NA, not available; RL-LDLT, right lobe living donor liver transplant; RL-SLT, right lobe split liver transplant; SAE, splenic artery embolization; SAL, splenic artery ligation; SFSS, small-for-size syndrome.

1LL-SLT0.66No interventionNANADeath from sepsis at 3 weeks
2RL-SLT1.13No interventionNANADeath from sepsis at 2 weeks
3RL-LDLT1.00SAL169890%Alive and well
4RL-LDLT1.01SAL2062114%Alive and well
5RL-LDLT1.48SAL1620125%Alive and well
6RL-LDLT0.82SAL2248114%Alive and well
7RL-LDLT0.99SAL2516130%Alive and well
8RL-LDLT0.64SAL171588%Alive and well
9LL-SLT0.51SAENANARetransplant at 2 months
10RL-LDLT1.10SAL1918102%Alive and well
11RL-LDLT0.91SAE1542110%Alive and well
Figure 5.

Median values of serum bilirubin (mg/dL), ascites production (L/day), and INR in the 11 recipients with small-for-size syndrome. Abbreviation: INR, international normalized ratio.

Inflow modification by occlusion of the splenic artery (either operative ligation or radiographic embolization) was performed in 9 of the 11 recipients that developed SFSS; additionally, SAL was performed in another 4 recipients in a preemptive manner (ie, it was performed intraoperatively at the time of the transplant in an attempt to try to prevent the development of SFSS). In all 4 of these recipients, the GW/RW ratio was <0.8% (mean GW/RW ratio of 0.71), and none of these recipients in fact developed SFSS. If one excludes these 4 recipients from the analysis, the incidence of SFSS was 16.7% in the group with GW/RW < 0.8% versus 9.4% in the group with GW/RW ≥ 0.8% (P = 0.62)


The increased use of partial graft liver transplants has resulted in the recognition of a unique set of risk factors and complications different from those seen with whole liver grafts, which may have an impact on outcomes. A prime example of this is the potential bile duct problems that may occur early or late after partial graft transplants. The cut surface of the liver and the often complex biliary reconstructions required with these transplants are obvious risk factors for biliary complications.11, 12 Another factor unique to partial transplants that has been shown to be a risk factor affecting outcome is the size of the graft. Usually measured as a proportion to the recipient size (either as the GW/RW ratio or standard liver volume), a small graft has been associated with a higher complication rate and inferior outcomes.9, 10 The reason for this is likely multifactorial, but one important problem with small grafts has been the risk for development of SFSS. The description of SFSS after partial liver transplants is a relatively recent event, and as such, there is no agreed upon definition of the condition. The definition that we chose is an arbitrary one based on descriptions in the literature suggesting that cholestasis and signs of portal hypertension such as ascites are the most important features. If we had chosen a less strict definition with fewer variables, then the incidence might have been higher, but we also would have risked including cases that did not all represent the same pathophysiological condition. It is important to rule out the possibility of mechanical and technical issues that may mimic the symptoms of SFSS but would be most appropriately treated by treatment of the technical problem.

Although there is no good agreement concerning the exact etiology, definition, and pathophysiology of this syndrome, the size of the graft has been shown to be a risk factor in its development (hence the name).13, 14 The size of the graft is likely not to be the only risk factor for SFSS. Other factors such as the degree of steatosis of the graft, degree of portal hypertension, size of the spleen, and illness severity in the recipient all likely play a role. The common pathway for all of these factors, however, is likely to be the degree of portal perfusion as the pathophysiology of this syndrome is believed to be a portal hyperperfusion injury to the hepatic vasculature and parenchyma.13

This analysis sought to answer 2 questions:

  • 1Is the size of the graft indeed a significant risk factor for developing SFSS?
  • 2How much impact does the size of the graft have on patient and graft survival rates?

With respect to the first question, our data showed that the incidence of SFSS was numerically higher in the smaller graft group versus the larger graft group (13.6% versus 9.4%), but this was not statistically different. Looking at Fig. 3, one does get the impression that size affects the incidence of SFSS, as many of the cases seem to be distributed in the lower GW/RW ratio ranges. Also, the incidence of SFSS was very high in the smallest grafts (ie, those with GW/RW < 0.7%). However, we did see some cases of SFSS in grafts with GW/RW > 1.0%, and this suggests that size was not the only factor that played a role in the development of SFSS. One point to keep in mind, though, is that a number of the recipients that received very small grafts were subjected to prophylactic intervention (ie, SAL) in the hope of preventing SFSS. A total of 4 (out of 22) of the recipients with GW/RW < 0.8% had preemptive SAL performed, and none of these recipients developed SFSS. When the cases with preemptive SAL were censored from the analysis, the incidence of SFSS in the group with a GW/RW ratio < 0.8% increased to 16.7% (versus 9.4%) in the larger graft size group; again, however, this difference was not statistically significant.

The other question that this analysis sought to answer was if the graft size ultimately affected graft and patient survival. Previous studies have shown a GW/RW ratio < 0.8% to be associated with significantly lower patient and graft survival rates.1, 2 Graft losses were usually early, secondary to vascular or infectious problems or to the development of SFSS. Our series, however, did not demonstrate a significant difference in patient or graft survival rates in a comparison of recipients with smaller grafts (GW/RW < 0.8%) and recipients with larger grafts (GW/RW ≥ 0.8%). Again, it must be kept in mind that this end result may have been influenced in part by the intervention of SAL, whether it was performed in a prophylactic or therapeutic manner. Nine recipients developed SFSS and underwent SAL; of these, 8 recovered after the intervention, and only 1 ultimately went on to retransplantation. Others have also reported inflow modification (either with SAL or portocaval shunting) to be useful in the setting of potential or established SFSS.15–17 These procedures are felt to work because the underlying pathophysiology of SFSS is believed to be portal hyperperfusion and portal hypertension.18 Hence, diminishing portal flow and pressure can help to avoid or treat this.

As our understanding of SFSS and partial transplants has improved, we have modified our intraoperative approach in these patients to try to prevent SFSS. Initially, we began by identifying recipients that we felt were at high risk for SFSS (on the basis of a small GW/RW ratio) and ligating the splenic artery at the time of transplant. We now routinely measure portal vein pressures at the end of the transplant procedure in partial liver transplant recipients and use interventions such as SAL or portal caval shunts in recipients with measurable high pressures. Although we still try to maintain a minimum GW/RW ratio of 0.8% when possible, we feel that in certain circumstances (such as better compensated patients without severe portal hypertension), it may be possible to use smaller grafts, especially when portal pressure measurements are used to try to prevent portal hyperperfusion.

In summary, graft size plays a role in determining outcomes after liver transplants, but it is not the only factor. The likelihood of SFSS is influenced not only by the size of the graft but also likely by other factors such as the degree of portal hypertension, MELD score, and splenic size. Perhaps a better term than small-for-size to describe this syndrome is small-for-need. In the setting of a high risk for SFSS or established SFSS, the use of inflow modification techniques can have a significant beneficial effect, offsetting the negative impact of small graft size on ultimate outcomes.