Survival among pediatric liver transplant recipients: Impact of segmental grafts

Authors

  • Peter L. Abt,

    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    Search for more papers by this author
  • Rachel Rapaport-Kelz,

    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    Search for more papers by this author
  • Niraj M. Desai,

    1. Department of Surgery, Washington University School of Medicine, St. Louis, MO
    Search for more papers by this author
  • Adam Frank,

    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    Search for more papers by this author
  • Seema Sonnad,

    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    Search for more papers by this author
  • Elizabeth Rand,

    1. Department of Pediatrics, Division of Gastroenterology, Children's Hospital of Philadelphia, Philadelphia, PA
    2. Fred and Suzanne Biesecker Center for Liver Disease, Children's Hospital of Philadelphia, Philadelphia, PA
    Search for more papers by this author
  • James F. Markmann,

    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    2. Fred and Suzanne Biesecker Center for Liver Disease, Children's Hospital of Philadelphia, Philadelphia, PA
    Search for more papers by this author
  • Abraham Shaked,

    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    2. Fred and Suzanne Biesecker Center for Liver Disease, Children's Hospital of Philadelphia, Philadelphia, PA
    Search for more papers by this author
  • Kim M. Olthoff

    Corresponding author
    1. Department of Surgery, University of Pennsylvania, Philadelphia, PA
    2. Fred and Suzanne Biesecker Center for Liver Disease, Children's Hospital of Philadelphia, Philadelphia, PA
    • Department of Surgery, Division of Transplantation, 4 Silverstein, Hospital of the University of Pennsylvania, 3400 Spruce St., Philadelphia, PA 19104
    Search for more papers by this author
    • Telephone: 215-662-6136; FAX: 215-662-7476


Abstract

Segmental liver transplantation with living donor (LD), reduced cadaveric (Reduced), and split cadaveric (Split) allografts has expanded the availability of size-appropriate organs for pediatric recipients. The relevance of recipient age to the selection of graft type has not been fully explored, but could offer the potential to maximize recipient outcome and donor utilization. We conducted a retrospective cohort study among children 12 years of age or less utilizing the United Network of Organ Sharing (UNOS) database. Cox proportional-hazards analysis was used to explore the association of recipient age and graft type to graft and patient survival. Among children <1 year of age and those 1 to 2 years of age, 3-year LD graft survival was superior to whole cadaveric (CAD) organs, Split grafts, and Reduced grafts (for children <1 year of age: 79.4 vs. 61.5, 66.0, and 61.1%, respectively, P = .0003; and for children 1-2 years of age: 79.2 vs 66.9, 57.1, and 63.9%, respectively, P = .02). However, in children 3 to 12 years of age, after controlling for multiple donor and recipient factors, LD grafts failed to offer a survival advantage (hazard ratio = .61; 95% confidence interval = .37-1.02) compared to CAD organs. In an adjusted analysis examining patient survival, there appeared to be minimal association between recipient age and graft type. Much of the difference in graft survival could be attributed to events in the perioperative period. In conclusion, LD liver transplantation provides improved graft survival in children 2 years of age or less. (Liver Transpl 2004;10:1287–1293.)

Liver transplantation has become an accepted modality for the treatment of children with liver disease. The development and adoption of reduced cadaveric (Reduced), split cadaveric (Split), and living donor (LD) liver transplantation as surgical strategies to expand the supply of appropriately-sized hepatic allografts in pediatric patients has filled a void, previously unmet by whole cadaveric (CAD) organs. As these techniques have been adopted, it has been suggested that waitlist mortality has decreased.1–3

These segmental allografts are associated with a myriad of unique technical, logistical, and physiological considerations that may ultimately impact short- and long-term graft and patient survival. From a technical standpoint, each involves similar issues that are associated with the division of hepatic parenchyma, subsequent biliary and vascular reconstruction, and adequate donor-to-recipient graft size.4 From a logistical vantage, the acuity of the patient, regional availability of a cadaveric organ or potential for a LD, and surgical experience of the transplant team are factors in the decision to utilize different graft types. Inherent to graft outcome are donor demographics, hemodynamics, and ischemic times, which vary with the utilization of LDs or cadaveric donors, both whole and segmental.5

Ultimately, many mitigating factors impact outcomes. Among these, discerning the relative importance of graft type to graft and patient outcome would influence decisions regarding graft selection. Theoretically, average graft life could be extended and the need for retransplantation diminished if 1 graft type, relative to others, demonstrated superior survival. Single-center studies usually lack adequate power to detect these differences and bias toward a specific graft type may exist within an individual center. The present study utilized a national database to analyze the relevance of graft type to graft and patient survival among pediatric hepatic transplant recipients, and evaluated the importance of recipient age to graft outcome.

Abbreviations

LD, living donor; UNOS, United Network of Organ Sharing; CAD, whole cadaveric; Reduced, reduced cadaveric; Split, split cadaveric.

Materials and Methods

We performed a retrospective cohort study using data from the United Network of Organ Sharing (UNOS) database. The initial dataset provided by UNOS consisted of all hepatic transplants performed between April 1987 and December 6, 2001. We limited our study cohort to all children 12 years or less who received a primary hepatic transplant between January 1, 1991 and June 6, 2001. Follow-up data was collected until December 6, 2001. Those patients without follow-up, who were unable to be assigned to a graft category, or were missing covariates of interest, were removed from the analysis. Subsequent to all exclusions, our final cohort consisted of 3,125 patients.

Graft survival was considered from the time of initial transplant to graft loss, patient death, or last follow-up. Patient survival was defined from time of 1st transplant to patient death or last known follow-up. Graft type was defined based upon 2 variables within the UNOS dataset. These were: donor type (living or cadaveric) and the liver organ type (whole, reduced/living donor, or split). Primary endpoints were graft failure and patient death. For the purpose of this analysis, recipient age was categorized into 3 periods (<1 year old, 1-2 years old, and ≥3 years old) based upon the frequency of graft type utilization.

Donor covariates included age, race, and gender. Recipient covariates included age, race, gender, cause of end-stage liver disease, preoperative total bilirubin, prothrombin time, albumin, creatinine, and recipient medical status prior to transplantation (i.e., home, in the hospital, or in the intensive care unit). Medical status served as a surrogate for UNOS status, which underwent modification during the time period included in the study. Other variables included degree of ABO blood group match, region, and year of transplant. Etiology of 90-day graft failure was derived from those patients who were assigned a cause of graft failure and who lost their graft within 90 days of transplant.

Correlation coefficients were examined between all covariates using the Pearson statistic for continuous variables and Spearman's rho for categorical variables. Statistically significant correlations were defined by P < .05. Chi-square analysis and analysis of variance were used for comparison of proportions and means between groups. Kaplan-Meier analysis was used to compute overall graft and patient survival. Comparison of Kaplan-Meier survival between groups was performed using the log-rank statistic. Cox proportional-hazards analysis was used to identify factors independently associated with graft and patient survival. Those factors found to be independently significant (P ≤ .1) were then analyzed in a multivariate model. To clarify the contribution of perioperative events to long-term graft outcomes, a conditional survival analysis was conducted for those grafts with survival of more than 90 days. Statistical analysis was performed using SPSS Graduate Pack 11.0 for Windows (SPSS, Chicago, IL).

Results

Recipient Groups

Children under the age of 1 year, 1 to 2 years of age, and between 3 and 12 years of age represented 33.4, 29.4, and 37.1% of the study cohort, respectively (Table 1). Compared to CAD grafts, a larger percentage of LD, Split, and Reduced grafts were utilized in children <1 year of age (28.2 vs. 52.1, 40.4, and 42.2%). Recipients of Split and Reduced grafts were more likely to suffer from acute liver failure, have an elevated bilirubin, and reside in the intensive care unit prior to transplant when compared to LD and CAD recipients. In contrast, LD recipients were least likely to be located in the intensive care unit (15.9%) prior to transplant.

Table 1. Donor and Recipient Characteristics
 LDCADSplitReducedP value
N334 (10.7%)2,174 (69.6%)198 (6.3%)419 (13.4%) 
Recipient age (yr)1.7 (2.8)3.4 (3.8)2.1 (3.0)2.1 (3.0)<.0001
Recipient age at transplant (yr)    <.0001
 <1174 (16.6)614 (58.8)80 (7.7)177 (16.9) 
 1–291 (9.9)633 (68.8)72 (7.8)124 (13.5) 
 3–1269 (5.9)927 (79.9)46 (4.0)118 (10.2) 
Recipient gender (% male)46.448.548.054.7.09
Recipient race (%)    .004
 White84.174.573.773.3 
 Black11.119.317.219.8 
 Other4.86.39.16.9 
Diagnosis (%)    <.0001
 Acute9.010.211.620.0 
 Chronic, cholestatic64.459.262.656.3 
 Chronic, noncholestatic25.728.021.221.7 
 Other.92.64.51.9 
Location at time of transplant (%)    <.0001
 Home56.954.842.940.8 
 Hospital (not ICU)27.220.817.721.2 
 ICU15.924.439.438.0 
Creatinine mg/dL (SD).37 (.29).49 (.73).43 (.58).49 (.8).02
Total Bilirubin mg/dL (SD)11.7 (10.3)12.2 (11.3)13.3 (11.5)14.4 (12.1).001
Prothrombin time (sec)17.2 (7.9)16.8 (7.4)17.4 (7.4)18.9 (8.9)<.0001
Albumin mg/dL (SD)3.1 (.72)3.2 (.9)3.1 (.7)3.1 (.7).22
ABO match (%)    .07
 Identical83.283.379.880.4 
 Compatible14.112.618.715.0 
 Noncompatible2.74.11.54.5 
Donor age (yr)31.0 (6.8)6.7 (10.6)24.4 (12.1)16.9 (15.1)<.0001
Donor race (%)    <.0001
 White85.080.486.984.0 
 Black10.517.68.614.1 
 Other4.52.04.51.9 
Donor gender (% male)38.957.271.257.5<.0001

Graft and Patient Survival

Graft survival at 3 years following transplant was greatest for the LD recipients (77.1 vs 69.3, 63.8, and 63.3% for CAD, Split and Reduced, respectively; P < .0001) (Fig. 1A). In a multivariate adjusted analysis, graft type was significantly associated with graft survival (Table 2). The risk of graft failure was 33, 91, and 55% higher among CAD, Split, and Reduced grafts, respectively, when compared to LD grafts.

Figure 1.

A: Three-year graft survival following liver transplantation. LD (thick dashed line) = 77.1%, CAD (thick solid line) = 69.3%, Split (thin solid line) = 63.8%, and Reduced (thin dashed line) = 63.3%. P < .0001 for all 4 groups. LD vs. CAD, P = .006; LD vs. Split, P = .0003; LD vs. Reduced, P < .00001; CAD vs. Split, P = .03; CAD vs. Reduced, P = .005. B: Three-year patient survival following liver transplantation. LD = 82.3%, CAD = 76.8%, Split = 76.7%, and Reduced = 71.0%. P = .001 for all 4 groups. LD vs. CAD, P = .02; LD vs. Reduced, P = .0001; CAD vs. Reduced, P = .003

Table 2. Graft Survival: Adjusted Analysis for Graft Type Among All Patients and the 3 Age Groups*
 Hazard ratio95% confidence intervalP value
  • NOTE: Age expressed as years.

  • *

    Adjusted for year of transplant, albumin, prothrombin time, donor age, graft type, medical condition at transplant, creatinine, bilirubin, region, recipient and donor race, cause of liver disease, and ABO match.

  • Analysis of variant between 4 groups.

All patients (N = 3,125)  .001
 vs LD   
  CAD1.331.0–1.78.05
  Split1.911.35–2.72<.0001
  Reduced1.551.15–2.10.004
Age < 1 (N = 1,045)  .002
 vs LD   
  CAD2.551.54–4.24<.0001
  Split2.131.22–3.71.0008
  Reduced2.521.54–4.12<.0001
Age 1–2 (N = 920)  .02
 vs LD   
  CAD2.051.13–3.71.02
  Split2.631.40–4.96.003
  Reduced1.941.09–3.48.025
Age 3–12 (N = 1,160)  0.1
 vs LD   
  CAD.61.37–1.02.06
  Split1.05.51–2.16.91
  Reduced.77.44–1.35.36

Patient survival was greatest among LD recipients (82.3 vs. 76.8% for CAD, 76.7% for Split, and 71.0% for Reduced; P = .001 for all 4 groups) at 3 years in a univariate analysis (Fig. 1B); however, when LD was directly compared to Split, the difference did not quite reach significance (P = .08). In a multivariate analysis (Table 3) patient survival was not different among graft types, although this approached significance when recipients of a Reduced graft were compared to LD recipients (hazard ratio = 1.38; 95% confidence interval = .99-1.93).

Table 3. Patient Survival: Adjusted Analysis for Graft Type Among All Patients and by Age Group*
 Hazard ratio95% confidence intervalP value
  • NOTE: Age expressed as years.

  • *

    Adjusted for year of transplant, albumin, prothrombin time, donor age, graft type, medical condition at transplant, creatinine, bilirubin, donor and recipient race, cause of liver disease, region, and ABO match.

  • Analysis of variant between 4 groups.

(N = 3,125) .15
 vs LD   
  CAD1.12.84–1.50.45
  Split1.30.85–1.99.23
  Reduced1.38.99–1.93.06
Age <1 (N = 1,045) .26
 vs LD   
  CAD1.42.92–2.21.12
  Split1.10.56–2.16.77
  Reduced1.58.96–2.58.07
Age 1–2 (N = 920) .14
 vs LD   
  CAD1.35.78–2.32.28
  Split1.94.96–3.92.07
  Reduced1.77.96–3.26.07
Age 3–12 (N = 1,160) .64
 vs LD   
  CAD.7.37–1.33.28
  Split.71.25–2.0.51
  Reduced.86.41–1.79.69

The relationship between recipient age and graft survival was investigated in a stratified analysis (Fig. 2). Among children <1 year old, and those 1 to 2 years of age, LD grafts offered improved survival compared to CAD, Split, or Reduced grafts (79.4 vs. 61.5, 66.0, and 61.1%, respectively, P = .0003; and 79.2 vs. 66.9, 57.1, and 63.9%, respectively, P = .02). Even after controlling for multiple donor and recipient factors, LD grafts retained a survival advantage over the other graft types in both age groups (Table 2). In contrast, the 3-year graft survival among children 3 to 12 years of age receiving an LD graft was inferior to CAD, although this did not quite reach significance in univariate (68.2 vs 76.1%; P = .09) or multivariate analysis (hazard ratio = .61; 95% confidence interval = .37-1.02; Table 2). In this age group, survival of LD grafts was equivalent to Split and Reduced organs (68.2 vs. 69.1 and 66.2%, respectively; Fig. 2C).

Figure 2.

A: Three-year graft survival in children <1 year of age. LD (thick dashed line) = 79.4%, CAD (thick solid line) = 61.5%, Split (thin solid line) = 66.0%, and Reduced (thin dashed line) = 61.1%. P = .0003 for all 4 groups. LD vs. CAD, P < .00001; LD vs. Split, P = .009; LD vs. Reduced, P = .0001. B: Three-year graft survival in children 1 to 2 years of age. LD = 79.2%, CAD = 66.9%, Split = 57.1%, and Reduced = 63.9%. P = .02 for all 4 groups. LD vs. CAD, P = .02, LD vs. Split, P = .003, LD vs. Reduced, P = .02. C: Three-year graft survival in children 3 to 12 years of age. LD = 68.2%, CAD = 76.1%, Split = 69.1%, and Reduced = 66.2%. P = .01 for all 4 groups. CAD vs. Reduced, P = .007

In the stratified analysis, patient survival among LD recipients exceeded CAD or Reduced recipients in children <1 year old and Split and Reduced recipients in children 1 to 2 years old (Table 4). These differences tended to dissipate after controlling for other variables, although there was a trend toward inferior patient survival among younger recipients of a Reduced organ (Table 3).

Table 4. Patient Survival Stratified by Recipient Age
Age category (years)LDCADSplitReducedP value
  • *

    P = .0009 vs. CAD.

  • P = .0008 vs. reduced.

  • P = .03 vs. split.

  • §

    P = .03 vs. reduced.

  • P = .05 vs. reduced.

<183.1%*70.1%78.2%68.4%.002
1–280.7%§73.9%67.9%68.9%.1
3–1283.1%83.3%86.7%77.4%.25

Analysis of Graft Failure

To determine if the differences in survival among the graft types occurred in the perioperative period or at later time points, a conditional survival analysis was conducted. Among the LD, CAD, Split, and Reduced grafts that survived beyond 90 days, similar 3-year survival was obtained (88.9, 84.9, 87.6, and 85.7%, respectively; P = .4). This finding held in separate analyses of the 3 age groups by graft type (<1 year old: 88.7, 82.3, 90.1, and 87.3%, for LD, CAD, Split, and Reduced, respectively, P = .16; 1-2 years old: 87.9, 82.7, 81.2, and 81.7%, respectively, P = .74; ≥3 years old: 90.6, 87.9, 91.2, and 87.9%, respectively, P = .96).

For those grafts that failed within the first 90 days of transplantation, the incidence of failure by graft type was 13.2, 18.3, 26.8, and 26.0% (for LD, CAD, Split, and Reduced, respectively) and by patient age was 23.5, 19.2, and 15.5% (for <1 year old, 1-2 years old, and ≥3 years, respectively). The incidence of 90-day graft failure for LD was highest in older children (24.6%), while among CAD recipients the youngest children had the highest frequency of graft loss (Table 5). Approximately 60% of patients with graft failure within 90 days of transplant have 1 or more reasons for graft failure recorded in the database. Of the graft failures, primary nonfunction was cited as a cause in 20.7% of LD, 35.3% of CAD, 48.6% of Split, and 46.3% of Reduced, while thrombosis was found in 48.3, 45.1, 40.5, and 32.8%, respectively.

Table 5. Incidence of Graft Failure Within 90 Days of Transplant, Stratified by Recipient Age and Graft Type
Recipient ageLDCADSplitReduced
<1 year10.3%25.1%26.3%29.9%
1–2 years9.9%19.0%29.2%21.8%
3–12 years24.6%13.3%23.9%24.6%

Discussion

The development of segmental hepatic grafts has expanded the supply of size-appropriate organs, allowing children who otherwise would have died on the waiting list the opportunity to undergo transplantation. The association between graft type and graft survival has been relatively undefined, in particular the relationship between graft type and recipient age. This knowledge could allow better selection of an appropriate graft type for a specific recipient, thereby improving outcomes.

We have demonstrated that among children <3 years of age, LD grafts provide superior graft survival compared to CAD, Split, and Reduced grafts, and that the older children may benefit from CAD grafts. These findings were maintained after controlling for multiple donor and recipient variables, including diagnosis and medical condition. The choice of graft type had little association with patient survival, although the clinical impact associated with retransplantation is not quantifiable from this analysis.

We recognize that this type of multicenter database study has several limitations. When compared to a single-center study, issues related to compliance with data submission and a lack of detailed data limit this analysis. In addition, residual confounding of factors unable to be considered in the analysis may occur. This includes the inability to account for center-specific experience with specific graft types and surgical volume..

The benefit gained from the use of LD grafts compared to whole or technical variant cadaveric grafts in younger children is most likely multifactorial. Brain death is associated with hemodynamic and endocrine changes that can cause organ injury.5 This damage, when combined with the increased cold ischemic times observed in CAD, Reduced, and Split grafts, leads to greater ischemia-reperfusion injury than in LD grafts. Following transplantation with Split grafts, liver enzymes in the early postoperative period are elevated compared to LD grafts, and the incidence of graft failure at 1 month is increased.6 The higher rate of primary nonfunction in the Reduced and Split groups may be related to longer ischemic times and preservation injury with in situ and ex vivo manipulation and graft preparation.6, 7

Among infants, the higher incidence of graft loss in CAD compared to technical variants has been attributed mostly to vascular thrombosis related to the small size of the vessels.8–10 In a study of factors affecting survival after liver transplants in infants, patients <1 year of age experienced inferior survival if they received a full-size vs. reduced-size organ. Hepatic artery thrombosis was responsible for over 50% of the grafts lost in these young children and accounted for a large portion of the decreased survival.8 Mazzaferro et al.11 demonstrated a higher incidence of hepatic artery thrombosis if donor arteries were less than 3 mm diameter. Technical improvements and the use of perioperative heparin or dextran may contribute to decreasing frequency of hepatic artery thrombosis in recent years.11, 12 As shown in this study, LD grafts also experience significant vascular complications.

In older children, the relative benefit of CAD compared to LD grafts may be related to 2 trends. The 1st, as previously described, is the improved outcome of CAD in older children due to fewer vascular complications. In children 3 years of age or older compared to those less than 1 year of age, 3-year CAD graft survival declined from 76.1 to 61.5%, likely due to an increased incidence of hepatic artery thrombosis. The 2nd is the decline in LD survival in older vs. younger children. Three-year graft survival in those 3 years of age and older was 68.2 vs. 79.4% in those <1 year of age. The specific reasons for these trends are not discernable from the UNOS database; however, the difference appears to be related to perioperative circumstances, as 90-day graft failure was 24.6% in those 3 years of age or older and only 10.3% in children <1 year of age. Furthermore, there was no difference in long-term survival for those grafts that survived longer than 90 days (90.6 vs. 88.7%, for ≥3 vs. <1 year old). It may be hypothesized that the difference in outcome for LD recipients may be secondary to a small-for-size phenomenon in the older children, the majority of whom most likely received left lateral segment grafts. This circumstance has been typically described in older and larger recipients, compared to those that are younger and smaller.4, 13 Theoretically, the advantages of LD grafts should extend to older recipients if graft volume is adequate. Developing a more detailed understanding of this possibility is difficult with the UNOS database, due to the absence of donor weight and graft size.

The results of this analysis are in part supported by data from the SPLIT Research Group. In a multivariate analysis using data collected between 1995 and 2000 from 29 centers, survival of Split grafts, but not Reduced grafts, was found to be inferior to CAD grafts. In contrast to the present study, LD graft survival did not differ significantly from CAD organs.14 Some of the differences between the studies may be attributed to the fact that unlike the UNOS database, the Studies in Pediatric Liver Transplant (SPLIT) database does not capture data from all transplant centers. An earlier unadjusted analysis of UNOS data also demonstrated that cadaveric split and reduced grafts were inferior to LD grafts.15 In contrast to these analyses with multicenter data, several single institutional studies have demonstrated excellent outcomes utilizing cadaveric segmental grafts.16–18 With an ex situ splitting technique, the Kings College group (London, UK) has produced an 89.7% 1-year graft survival. Others have published equivalent graft survival rates between CAD and technical variant cadaveric grafts, as well as between split and living related grafts.19, 20 These single-center reports reflect the results that can be obtained in highly experienced centers.

The findings in this study are based on a national average, and as such reflect a broad practice pattern. Individual center experience will vary and may not apply to these findings. Given the improved survival in very young children following LD transplantation, we suggest that this modality be pursued as an option if possible. In older children, it appears that CAD organs may offer better outcomes. While reduced and split grafts produce overall inferior outcomes in the national experience, they remain an important and necessary tool in a pediatric center's armamentarium. They provide appropriately-sized grafts in children without a suitable LD and for those for whom no cadaveric pediatric donor is available, and can have excellent outcome in experienced centers. It is apparent that the technical complexities and perioperative events surrounding these procedures have significant impact on outcome, as demonstrated by the dissolution of differences in survival if grafts lost in the 1st 90 days are eliminated from analysis. This emphasizes the importance of experience, attention to continued technical refinement, and judicious selection of appropriate donors for specific recipients.

Ancillary