SEARCH

SEARCH BY CITATION

Keywords:

  • Liver transplantation;
  • obesity;
  • outcomes research

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

The impact of obesity on outcomes following liver transplantation has been difficult to determine, in part due to the confounding effects of ascites on BMI. We evaluated the impact of pretransplant recipient obesity on outcomes following liver transplantation using the NIDDK Liver Transplantation Database. Pretransplant BMI, corrected for ascites, was categorized as underweight (BMI <18 kg/m2), normal weight (BMI 18–25 kg/m2), overweight (BMI 25.1–30 kg/m2), Class I obese (BMI 30.1–35 kg/m2), Class II obese (BMI 35.1–40 kg/m2) and Class III obese (BMI >40 kg/m2). Primary outcomes were patient and graft survival. Secondary outcomes included days in hospital and days in ICU. Data from 704 adult liver transplant recipients from the NIDDK LTD and a further 609 patients from the Mayo Clinic were analyzed. Early and late patient and graft survival was similar across all BMI categories. Correcting for ascites volume resulted in 11–20% of patients moving into a lower BMI classification. The relative risk for mortality increased by 7% for each liter of ascites removed. We conclude that corrected BMI is not independently predictive of patient or graft survival. Obesity, within the ranges observed in this study, should not be considered to be a contraindication to liver transplantation in the absence of other relative contraindications.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

Two-thirds of adults in the United States were reported by the National Health and Nutrition Education Survey (NHANES) study to be overweight or obese (1). Based on the increasing prevalence of obesity, nonalcoholic fatty liver disease (NAFLD), a common complication of obesity, is likely to affect up to 30 million people in the United States alone, with an increasing number progressing to cirrhosis and end-stage liver disease (2–5). Data from the United Network for Organ Sharing Database, NAFLD is the only indication for liver transplantation that has increased in the United States between 2000 and 2006 (http://www.UNOS.org) (5). Despite the high prevalence of obesity and the parallel increase in the frequency of obesity-associated liver diseases as an indication for liver transplantation, comparatively little is known regarding the impact of obesity on posttransplant outcomes.

Although it has been convincingly shown that obese patients are at increased risk for perioperative morbidity and mortality for a range of surgical procedures (6,7), studies of the impact of obesity on postoperative outcomes following liver transplantation have been inconsistent in their estimations of the impact of obesity on posttransplant morbidity and mortality. In the largest study reported to date, Nair et al. observed that postoperative mortality at 5 years was greater among recipients with class II (body mass index [BMI]>35 kg/m2) or class III (BMI >40 kg/m2) obesity (8). A potentially important limitation of the study by Nair et al., however, was an inability to correct BMI for ascites. If recipients with greater amounts of ascites, which often occurs in the context of more severe liver disease and renal dysfunction, are at increased risk for postoperative morbidity and mortality, then the increased risk for postoperative morbidity and mortality that has been reported to be associated with higher BMI may in fact reflect the impact of ascites on postoperative outcomes. The potential confounding effect of ascites on BMI-related outcomes is particularly important as the analysis of the Scientific Registry of Transplant Recipients (SRTR) by Nair et al. concluded that ‘morbid obesity should be considered a relative contraindication to liver transplantation.’

We evaluated the morbidity and mortality in obese patients undergoing liver transplantation using a large database, which had been prospectively collected, including ascites volume present at the time of transplantation.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

The objective of this study was to evaluate the impact of pretransplant recipient obesity on morbidity and mortality in patients undergoing liver transplantation using the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Liver Transplantation Database. It was established to prospectively collect data surrounding patients being assessed for and undergoing liver transplantation. It was implemented in January 1990 at three clinical centers: Mayo Clinic Rochester, University of Nebraska Medical Center and University of California, San Francisco, with coordination through the University of Pittsburgh (9). Data were collected through April 15, 1990, to June 30, 1994. Consecutive adult liver transplant recipients undergoing primary liver transplantation who consented to participate in the NIDDK Liver Transplant Database study were included in this analysis. There were no other exclusion or inclusion criteria (e.g. etiology of live disease). Data were collected using uniform instruments across all of the participating centers. This included the routine recording of volume of ascites fluid removed at transplantation. In order to provide data from a more contemporary but similar cohort, consecutive adult recipients undergoing liver transplantation at Mayo Clinic between December 1998 and October 2006 were also included in the analysis for primary outcomes. Eligibility for the study required informed consent, no prior liver transplant, and availability to be studied. Patients were excluded if they were under the age of 16 or had factors needed to calculate BMI missing. Data capture instruments for the Mayo cohort were similar to those used in the NIDDK Liver Transplant Database.

For the purpose of this review, data collected from the database on each recipient included age, sex, height, weight (all on the day of admission for surgery), amount of ascites at time of transplant, etiology of liver disease, postoperative complications, treated infections, treated rejection, primary graft nonfunction, need for retransplantation, total days in hospital, total days in intensive care unit, mortality with cause (at 30 and 120 days, 1, 2 and 5 years posttransplantation), as well as length of follow up. For each recipient we also recorded height and weight at 30 and 120 days, 1, 2 and 5 years posttransplantation if data were available.

Our primary outcomes included (1) mortality at 30 days, 1, 2 and 5 years and (2) graft survival at 30 and 120 days, 1, 2 and 5 years. Secondary outcomes included days in hospital, days in ICU, early and late complications.

Immunosuppression

Induction immunosuppression at Mayo Clinic consisted of cyclosporine, prednisone and azathioprine. The University of Nebraska used cyclosporine and prednisone. The University of California at San Francisco utilized antilymphocyte globulin followed by cyclosporine, prednisone and azathioprine. All centers participated in the FK506 Primary Immunosuppression Trial, resulting in a subgroup of 92 recipients receiving a tacrolimus-based regimen. In the mayo Clinic cohort after 1998 immunosuppression consisted of tacrolimus, prednisone and mycophenolate mofetil (MMF). Prednisone was tapered and stopped between 3 and 6 months posttransplantation. Patients who developed biopsy-proven acute cellular rejection (ACR) were treated with three i.v. boluses of methylprednisolone (1000 mg) given on alternate days. Following CMV infection, mycophenolate dosing was reduced or withdrawn.

Statistical Analysis

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

The cohort was described using estimates of central tendency (means, medians) and spread (standard deviation, range) for continuous data and frequencies and percentages for categorical data. Groups were compared using the chi-square test for differences in proportions (categorical data) and the Wilcoxon rank-sum test (continuous data).

Outcomes modeling

The purpose of this analysis was to determine whether true pretransplant BMI, adjusted for ascites, predicts posttransplant mortality or graft failure. Graft survival was modeled using logistic regression. To identify potential predictors of patient or graft failure other than BMI, Cox proportional hazards models were constructed. Only pretransplant characteristics were used in the models. These included BMI, recipient age, race, gender, international normalized ratio, total bilirubin, creatinine, CMV IgG serostatus, donor age, race and gender. Obesity was reflected by the BMI calculated by weight in kilograms (using pretransplant weight minus amount of ascites, with 1 L equaling 1 kg) divided by height in meters squared. BMI was calculated on each recipient at time of transplant and at each follow-up interval. The cohort of patients was analyzed according to their pretransplant weight category: nonobese (BMI 18.5–25 kg/m2), overweight (BMI 25.1–30 kg/m2), Class I obese (BMI 30.1–35 kg/m2), Class II obese (BMI 35.1–40 kg/m2) and Class III obese (BMI >40 kg/m2). Those who were considered underweight (BMI <18.5 kg/m2) were removed from the reference class for the purposes of analysis. They were described using estimates of central tendency (means, medians) and spread (standard deviation, range) for continuous data and percentages for categorical data. Mortality and graft survival were calculated for 30 and 120 days, 1, 2 and 5 years comparing all groups and represented as percentages of each category. Kaplan–Meier analysis was used for comparing survival between the groups.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

There were a total of 704 patients from the NIDDK dataset and 609 patients from the extended Mayo Clinic Rochester dataset, which met eligibility criteria and used in our analysis. Of these, 561 (43%) were normal weight, 405 (31%) overweight, 178 (14%) Class I obese, 69 (5%) Class II obese and 33 (3%) Class III obesity. A total of 280 patients (22%) had a BMI >30 kg/m2, and 67 (5%) were considered underweight with BMI <18.5 kg/m2.

As seen in Table 1, the mean age was between 48.1 and 53.9 years (p = 0.0022). The proportion of females was higher in the underweight group (49%) compared with the Class III obese (30%) (p = 0.0052). The most common indications for transplant were alcohol-related liver disease, chronic hepatitis C and hepatocellular carcinoma. The diagnosis of NASH was not evaluated in the NIDDK dataset but was seen with increasing frequency in the contemporary Mayo clinic dataset as BMI increased above 30 kg/m2.

Table 1.  Clinical characteristics by ascites-corrected weight class
 Underweight (<18.5) (n = 67)Normal (18.5–25) (n = 561)Overweight (25.1–30) (n = 405)Class I obese (30.1–35) (n = 178)Class II obese (35.1–40) (n = 69)Class III obese (>40) (n = 33)
  1. °Values reported are mean ± standard deviation or percent.

  2. *Data not recorded, BMI units are kg/m2.

Age (yr)° 48.1 ± 13.5 49.5 ± 11.6 51.6 ± 10.1 52.2 ± 10.353.9 ± 9.149.9 ± 9.1
Sex (female)49%44%33%43%39%30%
Ascites (liters)° 4.1 ± 4.3 2.2 ± 3.0 1.7 ± 2.7 1.5 ± 2.5 1.2 ± 2.2 0.6 ± 1.4
BMI (ascites corrected)°17.1 ± 1.222.2 ± 1.727.3 ± 1.432.0 ± 1.337.0 ± 1.543.5 ± 3.0
Etiology % (NIDDK/Mayo)
 Alcohol22/2218/1324/1324/836/1510/17
 Hep C7/811/1320/1220/1618/1760/13
 Hep B9/82/24/26/10/00/0
 PBC18/817/710/47/109/00/4
 PSC16/1522/1312/108/50/610/4
 NASH*/8*/1*/13*/10*/9*/17
 Cryptogenic6/011/112/518/814/210/4
 HCC2/82/134/213/260/230/17

The primary outcomes included patient and graft survival at 1 month, 1 year, 2 years and 5 years as described in Table 2. All cause mortality was similar across all BMI categories after correction for amount of ascites removed at the time of transplant, at 1 month, 1 year, 2 and 5 years (see Figure 1). In the normal weight group at 1 and 5 years the mortality rates were 11% and 20%, respectively. For Class I obese mortality was 8% and 12%, and for Class III obese 3% and 20%. Results were similar for graft survival, with no difference seen comparing all weight categories.

Table 2.  Primary outcomes by ascites-corrected weight class
 Underweight (<18.5) (n = 67)Normal (18.5–25) (n = 561)Overweight (25.1–30) (n = 405)Class I obese (30.1–35) (n = 178)Class II obese (35.1–40) (n = 69)Class III obese (>40) (n = 33)
  1. *p = 0.79.

  2. †p = 0.70.

  3. BMI units are kg/m2.

Mortality*
 30 day 2% 2% 4% 3% 3% 0%
 1 year11%11%11% 8%10% 3%
 2 years18%14%16%13%18%10%
 5 years31%20%23%12%18%20%
Graft survival†
 30 day96%96%94%96%93%100% 
 1 year85%84%85%90%82%91%
 2 years76%82%81%85%77%88%
 5 years61%76%71%75%77%74%
image

Figure 1. Patient mortality for recipients in each weight class at 1 month, 1, 2 and 5 years posttransplantation is shown. There was no statistically significant difference in survival between the groups at any of the follow-up intervals (log-rank test p = 0.79).

Download figure to PowerPoint

There was no statistically significant difference in survival between the groups at any given time interval (log-rank test p = 0.79). Calculated hazard ratio (HR) for mortality in the overweight group was 1.14 (log-rank test p = 0.41). In the Class I obese group the HR was 0.92 (log-rank test p = 0.41) and in the Class III obese group the HR was 0.63 (log-rank test p = 0.26).

Impact of ascites volume on posttransplant outcomes

Correcting for ascites volume at the time of transplant resulted in 11–20% of patients with BMI >25 kg/m2 moving into a lower BMI classification (see Figure 2). Only 6% of originally normal weight patients moved into a lower category. The number of patients in the normal group actually went up after correcting for ascites, from 521 to 561. However, in groups with BMI >25 kg/m2, all lost patients. Class III obese went from 37 to 33 patients, Class II obese from 81 to 69 patients and Class I obese from 196 to 176 patients. Ascites volume was independently associated with increasing relative risk for mortality with a HR of 1.07 (1.03–1.11, p ≤ 0.01) for each liter of ascites removed. Essentially, for each liter of ascites removed there was a relative risk for increase in mortality by 7%. A total of 10 L of ascites removed intraoperatively would thus have been associated with a 70% increase in risk for mortality during the follow-up period. There was also a significantly increased relative risk of graft failure with a HR of 1.06 (6%) (1.02–1.09, p < 0.01).

image

Figure 2. Percentage of recipients who moved down at least one weight class after correcting BMI for ascites volume at the time of transplantation is shown. Mortality was significantly higher among patients who moved down in weight category, with ascites volume independently associated with increased relative risk for mortality (HR of 1.03–1.11, p ≤ 0.01, for each liter of ascites removed).

Download figure to PowerPoint

Table 3 shows the secondary outcomes in various weight groups. This information was only available for the NIDDK patients (n = 704). Median values for number of ICU days did not differ. There appeared to be longer lengths of stay in the Class III obese compared to all other classes (27 vs. 16–17 days) but this was not significant with a p-value > 0.1. There was also no difference in early and late complications.

Table 3.  Secondary outcome by ascites-corrected weight class for the 704 NIDDK patients
 Underweight (N = 54)Normal (N = 352)Overweight (N = 194)Class I obese (N = 72)Class II obese (N = 22)Class III obese (N = 10)Total (N = 704)
  1. *p < 0.05.

ICU days
 Median3.54.04.03.04.03.54.0
 Range (1.0–59.0)  (1.0–175.0)  (1.0–108.0) (1.0–92.0)  (2.0–115.0) (2.0–20.0)  (1.0–175.0)
Hospital days
 Median  17.016.016.017.016.527.017.0
 Range (1.0–79.0)  (1.0–224.0)  (1.0–161.0)  (1.0–104.0)  (3.0–299.0)  (11.0–114.0)  (1.0–299.0)
Early complications
 Median0.00.00.00.00.00.50.0
 Range(0.0–6.0)(0.0–8.0)(0.0–6.0)(0.0–3.0)(0.0–3.0)(0.0–3.0)(0.0–8.0)
Late complications
 Median0.01.01.00.00.01.51.0
 Range(0.0–7.0) (0.0–15.0) (0.0–10.0) (0.0–11.0) (0.0–11.0)(0.0–8.0) (0.0–15.0)

Over the course of the study period, the proportion of patients classified as obese increased with approximately 15% in the early 1990's to just over 25% since 2002 (Figure 3). Using a linear model from 1990 to 2006, there is an increase in weight of approximately 1 kg/year. This would be consistent with the overall trend toward increasing rates of obesity in the general population. During the same time interval however, the amount of ascites at the time of transplant had decreased (0.07 L/year).

image

Figure 3. Variation in mean percentage of recipients who were obese at the time of liver transplantation over time is shown. Line indicates trend of means. Using a linear model from 1990 to 2006, there is an increase in weight of approximately 1 kg/year.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

At the core of selecting patients for liver transplantation is the ability to make a projection of the relative risk and benefit to the potential recipient. We are fortunate in having ample data about the influence of many recipient and donor variables, such as age and etiology of liver disease, on posttransplant outcomes. The negative impact of advanced recipient age is, for example, routinely used in the recipient selection process. For some biological variables, such as mean pulmonary artery pressure, quite precise limits are known, above which liver transplantation is rarely performed. In order for transplant centers to develop a rational approach to the management of obese patients, the effect of high BMI on patient and graft survival needs to be determined. The international epidemic of obesity, with the likely attendant increase in the prevalence of obesity among liver transplant recipients, makes a fuller understanding of the impact of recipient BMI on posttransplant outcomes critical. The most important finding of our current study is that, in a multicenter, prospective, long-term follow-up analysis, corrected BMI is not independently predictive of patient or graft survival. Indeed, a trend was seen for superiority in outcomes among recipients in the highest BMI category.

Our observations regarding patient and graft survival among obese and nonobese recipients should be compared to those reported by other groups. Our findings are broadly similar to those of a case-control study of 121 consecutive patients who underwent liver transplantation at a single center by Sawyer et al. (10), who found no effect of obesity on overall short-term patient and graft survival. Similarly, in a small, single center analysis that attempted to correct BMI for ascites, Keefe et al. (11), found that 1-year patient survival was slightly superior in obese when compared to nonobese patients. In contrast, a Danish study reported an association of obesity with increased posttransplant mortality (12). Other, smaller studies have observed no significant effect of obesity on overall mortality (10–15). Each of these aforementioned studies was limited not only by size and the single center nature of the reports but also, with one exception, by the failure or inability to account for the contribution of ascites to recipient weight at the time of transplantation. The SRTR is often utilised, by virtue of sheer size, to provide definitive answers to questions regarding the impact of recipient or donor variables on posttransplant outcomes. In what is, by far, the largest study to date, Nair et al. conducted an SRTR-based review of the impact of recipient BMI on posttransplant patient and graft survival (8). In contrast to our analysis, the SRTR-based review by Nair et al. found that Kaplan–Meier survival was significantly lower in morbidly obese patients, and that morbid obesity was an independent predictor of mortality. The number of recipients analysed by Nair et al. was over 23 000. Does the sheer weight of numbers of the SRTR analysis lend its contrasting conclusions more credibility than our analysis? The authors of the SRTR analysis state that, ‘It is probable that we may have overestimated obesity, because many patients with endstage cirrhosis had fluid overload; however, our comparison between groups is still valid because fluid overload would have affected all groups (equally)’. One of the most important findings of our study is that fluid overload does not affect all groups equally. We found that correcting for ascites volume at the time of transplant resulted in 11–20% of patients with BMI >25 kg/m2 moving into a lower BMI classification while only 6% of originally normal weight patients moved into a lower category, for example (Table 3). The number of patients in the normal group actually went up after correcting for ascites. This would not affect results if patients who changed BMI category based on adjusting for ascites had similar outcomes to patients who did change BMI category. What we observed, however, was that ascites volume is independently associated with increasing relative risk for mortality, with a HR of 1.07 (1.03–1.11, p ≤ 0.01) for each liter of ascites removed. In other words, a patient with 6 L of ascites removed at transplantation would have >40% excess mortality during a given period of follow-up than a patient who had no ascites but was otherwise similar (including BMI). The effect of ascites on BMI calculation may have entirely accounted for the negative association between BMI and posttransplant outcomes observed in the SRTR analysis. The same is also true of other studies that reported an association of high BMI and poor outcomes. Thus, while the SRTR is an important source of information, it cannot, in our opinion, be used to estimate the impact of recipient BMI on overall posttransplant outcomes. Although the NIDDK Liver Transplant Database is a relatively older cohort of liver transplant recipients, it has the advantages of completeness and of all data being captured with a uniform instrument across participating centers.

We also measured the impact of BMI on other outcomes, including the frequency of treated infections, treated rejection, primary graft nonfunction, total days in hospital and total days in intensive care unit. We observed greater lengths of stay for recipients with Class III obesity when compared to all other classes (27 vs. 16–17 days) but this difference narrowly failed to reach statistical significance. An association of Class III obesity and greater length of hospital seems probable and it is likely that the small number of recipients in this category prevented the difference from reaching statistical significance. Unfortunately, a definitive answer regarding the impact of BMI on secondary posttransplant outcomes remains elusive. A significant negative association of increasing BMI on primary or secondary posttransplant outcomes is improbable, based on our observations. Our results should not, however, be interpreted as indicating that patients with higher BMI can routinely be transplanted safely. The obese liver transplant recipients described in this study almost certainly represent a highly selected group. Given the increased technical difficulties presented by patients with Class I–III obesity, it is likely that the tolerance of comorbidities was low. All of the transplant centers participating in this analysis routinely screen for ischemic and other heart diseases in potential liver transplant recipients. It is possible that acceptable parameters for cardiac ejection fraction and of pulmonary hypertension were different in a patient with class III obesity than they were in a potential recipient with a normal BMI, for example. It is also likely that there were differences in selection criteria and other nuances in candidate selection between participating centers.

With these caveats in mind, three conclusions should be drawn from this study. Firstly, outcomes following liver transplantation for selected patients with class I–III obesity are similar to nonobese recipients. When interpreting these results it is important to remember that the good outcomes achieved in obese recipients in this analysis were in the context of rigorous selection and the likely relative absence of other medical comorbidities. Secondly, an effect of recipient obesity on non-life- or graft-threatening posttransplant complications cannot be excluded. This paper should not be construed as stating that obesity is not a potential risk for adverse posttransplant outcomes but that, in selected obese patients, satisfactory outcomes can be achieved. Lastly, when other cohorts of liver transplant recipients are being analyzed to determine the impact of pretransplant BMI in our cirrhotic populations, the BMI must be corrected for the amount of ascites—regardless of endpoints being studied.

The last important observation of this study is the increasing mean recipient BMI that occurred over time. The proportion of recipients with obesity, calculated with correction for ascites, increased from approximately 15% in the early 1990's to just over 25% since 2002. Between 1990 and 2006, the mean increase in recipient weight was ∼1 kg/year. This increase appears to have been due to body fat as the mean amount of ascites at the time of transplant had decreased during the same period. The increasing mean BMI of the US population in general is clearly reflected in the mean BMI of patients undergoing liver transplantation.

In conclusion, the prevalence of obesity among liver transplant recipients appears to be increasing. When BMI is corrected for the amount of ascites present at the time of surgery, outcomes following liver transplantation for selected patients with class I–III obesity are similar to nonobese recipients. If recipient selection committees decide to alter their guidelines for inclusion or exclusion of obese candidates based on these findings they will need to monitor carefully for any change in outcomes—particularly increases in posttransplant vascular complications, diabetes, metabolic syndrome, hypertension, etc. In patients who are otherwise satisfactory candidates for liver transplantation, however, BMI, within the ranges seen in this study, should not be considered to be a contraindication to liver transplantation. The impact of recipient obesity on nonlife or graft-threatening posttransplant complications remains to be determined.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References

This work has been supported by Public Health Service grant NIDDK RO1 DK069757-01 and GCRC RR00585.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Statistical Analysis
  6. Results
  7. Discussion
  8. Acknowledgment
  9. References