LIVER TRANSPLANTATION 20:281-290, 2014
ORIGINAL ARTICLE
Additive Effect of Pretransplant Obesity, Diabetes, and Cardiovascular Risk Factors on Outcomes After Liver Transplantation Anna J. Dare,1 Lindsay D. Plank,1 Anthony R. J. Phillips,1,2,4 Edward J. Gane,4 Barry Harrison,4 David Orr,4 Yannan Jiang,3 and Adam S. J. R. Bartlett1,2,4 1 Department of Surgery, Faculty of Medical and Health Sciences, 2Maurice Wilkins Centre for Molecular Biodiscovery, University of Auckland, Auckland, New Zealand, and 3Department of Statistics, Faculty of Science, University of Auckland, Auckland, New Zealand; and 4New Zealand Liver Transplant Unit, Auckland City Hospital, Auckland, New Zealand
The effects of pretransplant obesity, diabetes mellitus (DM), coronary artery disease (CAD), and hypertension (HTN) on outcomes after liver transplantation (LT) are controversial. Questions have also been raised about the appropriateness of the body mass index (BMI) for assessing obesity in patients with end-stage liver disease. Both issues have implications for organ allocation in LT. To address these questions, we undertook a cohort study of 202 consecutive patients (2000-2010) undergoing LT at a national center in New Zealand. BMI and body fat percentage (%BF) values (dual-energy X-ray absorptiometry) were measured before transplantation, and the methods were compared. The influence of pretransplant risk variables (including obesity, DM, CAD, and HTN) on the 30-day postoperative event rate, length of hospital stay, and survival were analyzed. There was agreement between the calculated BMI and the measured %BF for 86.0% of the study population (j coefficient 5 0.73, 95% confidence interval 5 0.61-0.85), and this was maintained across increasing Model for EndStage Liver Disease scores. Obesity was an independent risk factor for the postoperative event rate [count ratio (CR) 5 1.03, P < 0.001], as was DM (CR 5 1.4, P < 0.001). Obesity with concomitant DM was the strongest predictor of the postoperative event rate (CR 5 1.75, P < 0.001) and a longer hospital stay (5.81 days, P < 0.01). Independent metabolic risk factors had no effect on 30-day, 1-year, or 5-year patient survival. In conclusion, BMI is an adequate tool for assessing obesity-associated risk in LT. Early post-LT morbidity is highest for patients with concomitant obesity and DM, although C 2014 AASLD. these factors do not appear to influence recipient survival. Liver Transpl 20:281-290, 2014. V Received June 7, 2013; accepted November 7, 2013.
See Editorial on Page 253 Obesity, diabetes mellitus (DM), and metabolic syndrome [comprising increased waist circumference, hyperglycemia, hypertension (HTN), and dyslipidemia1] are now the most significant public health prob-
lems facing developed countries.2,3 They are encountered in patients in virtually every area of health care, including patients with end-stage liver disease (ESLD) undergoing liver transplantation (LT). The impact of obesity on postoperative outcomes and survival after surgical procedures and its place in the allocation of health resources and expenditures have
Abbreviations: %BF, body fat percentage; BMI, body mass index; CAD, coronary artery disease; CI, confidence interval; CR, count ratio; DM, diabetes mellitus; DXA, dual-energy X-ray absorptiometry; ESLD, end-stage liver disease; HTN, hypertension; ICU, intensive care unit; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease. This work was supported by the University of Auckland Postgraduate Research Fund. The authors have no conflicts of interest to declare. Address reprint requests to Adam S. J. R. Bartlett, M.B.Ch.B., Ph.D., F.R.A.C.S., New Zealand Liver Transplant Unit, Auckland City Hospital, Park Road, Grafton, Auckland, New Zealand. Telephone: 164 9 3074949 or 164 212414647; FAX: 164 9 3754345; E-mail: [email protected] DOI 10.1002/lt.23818 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
282 DARE ET AL.
received considerable attention in the literature in recent years.4-7 Analyzing the impact of obesity on the outcomes of patients undergoing LT has been considered challenging because the conventional measurement of obesity, the body mass index (BMI), may overestimate obesity in patients with ESLD who have concomitant ascites.8,9 Furthermore, BMI obesity cutoff definitions may not be appropriate for nonEuropean ethnic groups, who make up a significant proportion of LT recipients worldwide.10,11 Despite this, morbid obesity, as measured by BMI, is used by some centers as an absolute contraindication to LT,8 and there is anecdotal evidence that recipient obesity continues to influence resource allocation decisions made by physicians.7 Although there has been substantial research into posttransplant obesity and metabolic syndrome, the additive impacts of pretransplant obesity and other pretransplant metabolic risk factors such as DM, HTN, and coronary artery disease (CAD) on postoperative outcomes have received much less attention. LT recipients who develop posttransplant DM or other features of metabolic syndrome have an increased risk of cardiovascular complications.12,13 It seems reasonable to hypothesize that similar risks may exist for patients with pretransplant metabolic syndrome undergoing LT. During the last 3 decades, graft loss and recipient deaths as a result of rejection have declined significantly as a result of improved immunosuppressive regimens. Death with a functioning graft, often a result of cardiovascular disease, is now a significant cause of posttransplant mortality and accounts for up to 42% of non–graft-related deaths.14 There is mounting evidence to suggest that obesity, insulin resistance, and type II DM represent a spectrum of metabolic diseases, with the greatest clinical risk clustering in those with an obese and diabetic phenotype.15-18 Understanding the independent and additive risks of obesity, DM, and metabolic syndrome for outcomes after LT may represent an important aspect of pretransplant management for optimizing both early and long-term outcomes. This is particularly important when we are considering whether allocation decisions based on obesity and other lifestyle diseases are appropriate. The objective of this study was to evaluate the independent and additive risks of pretransplant obesity, DM, HTN, and CAD on early and late postoperative outcomes of LT.
PATIENTS AND METHODS We conducted a single-center, retrospective analysis of outcomes for consecutive patients undergoing LT at the New Zealand Liver Transplant Unit of Auckland City Hospital between November 2000 and May 2010. Patients were excluded if they were 90 mm Hg, or systolic blood pressure > 160 mm Hg), pretransplant CAD (documented myocardial ischemia or coronary angiogram evidence), calculated Model for End-Stage Liver Disease (MELD) score at the time of transplantation, donor type (donation after brain death or living donation), cold ischemia time (minutes), and operative time (minutes). [All LT candidates presenting for an assessment underwent a clinical examination, electrocardiography, and 2dimensional echocardiography as part of a standard cardiac workup; patients with a history of cardiovascular disease or risk factors, such as an age > 60 years, a family history, DM, HTN, hyperlipidemia, nonalcoholic fatty liver disease (NAFLD), or renal failure, or assessment findings suggestive of occult cardiovascular disease were referred for stress testing and further management as appropriate.] Posttransplant factors were obtained through a retrospective review of electronic medical records and included the length of the intensive care unit (ICU) stay and the total
LIVER TRANSPLANTATION, Vol. 20, No. 3, 2014
DARE ET AL. 283
TABLE 1. Modified Clavien Classification System for Surgical Complications Events
Mean
Definition
(n)
(Range)
Any need for a deviation from the normal postoperative course without the need for pharmacological treatment or surgical, endoscopic, or radiological interventions. Allowed therapeutic regimens include drugs as anti-emetics, antipyretics, analgesics, diuretics, and electrolytes. This grade also includes wound infections opened at the bedside. Complication requiring pharmacological treatment with drugs other than those allowed for grade I complications. Blood transfusions and parenteral nutrition are also included. Complication requiring surgical, endoscopic, or radiological intervention Intervention not under general anesthesia Intervention under general anesthesia Life-threatening complication (including central nervous system complications*) requiring intensive care management Single-organ dysfunction Multiorgan dysfunction Death of a patient
242
1.2 (0-5)
275
1.36 (0-4)
28 29
0.14 (0-2) 0.14 (0-2)
16 5 0
0.08 (0-1) 0.02 (0-1) 0
Grade I
II
III IIIa IIIb IV IVa IVb V
NOTE: This table has been adapted with permission from Annals of Surgery.21 The total number of events recorded for each complication grade as well as the means and ranges for event rates per complication grade are shown. *Brain hemorrhage, ischemic stroke, and subarachnoidal bleeding (transient ischemic attacks are excluded).
hospital stay after transplantation, postoperative events occurring within the first 30 days after transplantation, and patient survival. Postoperative events were classified with the validated modified classification of surgical complications by Dindo et al.21 This classification system assigns a grade from I to V on the basis of the severity of a complication and the therapeutic intervention required (see Table 1). We also performed a second outcome analysis based on the type of postoperative complication by etiology with the following categories: infective, surgical, cardiovascular, respiratory, gastrointestinal, renal, neurological, hematological, metabolic, and immunological events (including acute rejection). This study was undertaken in accordance with the study hospital ethics committee requirements and a priori ethics approval was granted.
in the ICU. In this model, a significant positive estimate indicated a longer mean stay per unit change in the variable of interest. Poisson regression analysis was used to model total complication events and events by etiology (infective, surgical, cardiovascular, respiratory, gastrointestinal, renal, neurological, hematological, metabolic, or immunological). In this model, a significant estimate > 1 indicated a higher event rate per unit change in the variable of interest. Kaplan-Meier survival estimates were used to illustrate differences in patient survival between groups, and associated P values were derived from a univariate log-rank test (MedCalc version 12). Statistical analyses were performed with SAS 9.2 (SAS Institute, Inc., Cary, NC) and with R 2.14.1 (R Foundation for Statistical Computing). All statistical tests were 2-tailed at a 5% significance level.
Statistical Analysis Baseline recipient and perioperative characteristics for all patients were summarized with descriptive statistics. Continuous variables were presented as means, standard deviations, and ranges. Categorical variables were presented as observed numbers and percentages. Raw agreement (%) and j coefficients were used to evaluate the agreement between obesity classifications by BMI and %BF overall and within MELD score and ethnicity subgroups. For each outcome measure, the variables of interest were fit one at a time in regression models with adjustments for age, sex, and ethnicity as confounding factors. More specifically, linear regression analysis was used to model days in the hospital and days
RESULTS Two hundred two consecutive patients undergoing LT met the study criteria. The baseline characteristics and the perioperative variables are shown in Table 2. The mean recipient age was 51 years (range 5 19-69 years), and 66.8% of the patients (n 5 135) were male. The most common indication for LT was hepatitis C virus cirrhosis (39.1%). BMI values at the time of transplantation were available for 192 patients: 8 patients (4.2%) were underweight, 52 patients (27.1%) were within the normal range, 50 (26.0%) were overweight, 53 (27.6%) were obese, 18 (9.4%) were severely obese, and 11 (5.7%) were morbidly obese. For the multivariate analysis,
284 DARE ET AL.
LIVER TRANSPLANTATION, March 2014
TABLE 2. Baseline Recipient and Perioperative Characteristics (n 5 202) Categorical Variable Sex Male Female Ethnicity European Maori/Pacific Asian Other Indications for transplant Autoimmune Alcoholic liver disease Hepatitis C virus Hepatitis B virus NAFLD Other Concurrent hepatocellular carcinoma Yes No Missing DM Yes No HTN Yes No CAD Yes No Missing Continuous Variable Age (years) BMI (kg/m2) Total body fat (%) MELD score Cold ischemia time (minutes) Operation time (minutes)
the small numbers of patients who were severely or morbidly obese were grouped with the obese patients. Forty-eight patients (23.8%) had DM at the time of transplantation. More than half of the diabetic patients (58.7%) were also obese according to BMI. Thirty-two patients (15.8%) had HTN at the time of transplantation, and 10 (5%) had CAD. Fifty-three of the patients defined as obese by BMI (64.6%) had 1 additional metabolic risk factor (DM, HTN, or CAD), and 13 of the obese patients (16%) had 2 additional risk factors (DM, HTN, or CAD). One patient was obese and had all 3 metabolic risk factors (DM, HTN, and CAD). CAD and HTN were, therefore, grouped together when we evaluated risk interactions in the multivariate analysis to allow more meaningful comparisons. %BF values were available at the time of listing for 129 patients and at the time of transplantation for 69 patients. Sixty-four of these patients had %BF measurements at both times, so we performed a correlation
n
%
135 67
66.8 33.2
128 46 16 12
63.4 22.8 7.9 5.9
28 17 79 39 14 25
13.9 8.4 39.1 19.3 6.9 12.4
68 132 2
33.7 65.3 1.0
48 154
23.8 76.2
32 170
15.8 84.2
10 191 1
5.0 94.6 0.5
Mean 6 Standard Deviation 51 6 10 28.9 6 6.1 28.69 6 7.56 15.0 6 6.2 434.5 6 161.0 372.6 6 99.4
Range 19-69 17.6-46.3 6.15-51.70 6.4-44.4 23-960 211-1025
analysis and a paired t test for these patients. This showed that %BF values at the time of listing and on the day of transplantation were highly correlated (r 5 0.92, Pearson’s correlation coefficient), and there was no significant difference between the 2 values (paired Student t test). We, therefore, used %BF on the day of listing as a surrogate measure in our analysis when %BF on the day of transplantation was not available. This enabled a more robust data modeling process because of the larger number of patients with %BF data for the day of listing.
Agreement Between BMI and %BF for the Assessment of Obesity The results of the comparison of the BMI and %BF techniques for the assessment of obesity are shown in Table 3. There was agreement between BMI and %BF for the classification of patients into obese or nonobese groups for 86.0% of the study population [j
LIVER TRANSPLANTATION, Vol. 20, No. 3, 2014
DARE ET AL. 285
TABLE 3. Agreement Between BMI and %BF as Measures of Obesity in Patients Undergoing LT: Overall, by MELD Score, and by Ethnicity Overall Agreement
j Coefficient
(%)
(95% CI)
86.0
0.73 (0.61-0.85)
85.7 82.1
0.70 (0.46-0.94) 0.60 (0.31-0.89)
91.4 71 83
0.84 (0.71-0.96) 0.49 (0.21-0.77) 0.56 (0.12-1.0)†
Overall MELD score Low (17) Ethnicity* European Maori/Pacific Asian/other
*A significant difference in the strength of agreement between BMI and %BF was observed for European patients versus Maori/Pacific patients. † There were insufficient numbers to draw accurate conclusions for patients of Asian/other ethnicity.
coefficient 5 0.73, 95% confidence interval (CI) 5 0.610.85]. Europeans were significantly more likely to show agreement between obesity cutoffs (91.4%, j coefficient 5 0.84, 95% CI 5 0.71-0.96) than Maori and Pacific patients (71%, j coefficient 5 0.49, 95% CI 5 0.21-0.77). Patients with MELD scores in the lowest quartile (score < 11) had 85.7% agreement between BMI and %BF cutoffs (j coefficient 5 0.70, 95% CI 5 0.46-0.94). This was not significantly different from the situation for patients with MELD scores in the highest quartile (>17), for whom there was agreement 82.1% of the time (j coefficient 0.60, 95% CI 5 0.31-0.89).
Postoperative Events Postoperative event rates by complication grade are shown in Table 1: 70.2% of the patients experienced at least 1 grade I complication, 70.9% experienced at least 1 grade II complication, 12.4% experienced at least 1 grade IIIa complication, and 12.9% experienced at least 1 grade IIIb complication. A severe (grade IV) complication was experienced by 10.4% of the patients, with 7.9% experiencing a grade IVa complication and 2.5% experiencing a grade IVb complication. Several independent risk factors related to recipients’ pretransplant metabolic risk profiles and features of their underlying liver disease were associated with an increased postoperative event rate in our regression model. Obesity (according to both BMI and %BF cutoffs), DM, and the MELD score were each predictive of a higher postoperative event rate (Table 4). When we evaluated additive risk factors, obesity (by both BMI and %BF) with concurrent DM was the strongest predictor of a higher postoperative event rate [count ratio (CR) 5 1.75, 95% CI 5 1.37-2.25, and P < 0.001 when
BMI was used as the measure of obesity; CR 5 1.63, 95% CI 5 1.17-2.29, and P 5 0.004 when %BF was used as the measure]. The presence of obesity, DM, and a concurrent third metabolic risk factor (either HTN or CAD) did not significantly increase the risk (CR 5 1.65, 95% CI 5 1.16-2.35, P 5 0.005; Table 4). Obesity with cardiac risk factors (HTN or CAD) in the absence of DM also did not significantly increase the postoperative event rate (Table 4). A higher incidence of posttransplant infections (eg, wound infections, bacteremia, and pneumonia) were seen in recipients with higher MELD scores (CR 5 1.06, 95% CI > 1.02-1.10, P 5 0.004), in patients with NAFLD as the indication for LT (CR 5 4.52, 95% CI 5 1.6012.81, P 5 0.005), and in patients who were obese (according to BMI and %BF) with concurrent DM (CR 5 3.56, 95% CI 5 1.52-8.32, P 5 0.005). Higher rates of posttransplant cardiovascular events (eg, arrhythmia and myocardial infarction) were seen in patients with prolonged operative times (CR 5 1.01, 95% CI 5 1.00-1.01, P 5 0.04), in patients who were obese according to BMI (CR 5 1.11, 95% CI 5 1.061.16, P < 0.001), in patients who were obese according to %BF (CR 5 1.05, 95% CI 5 1.03-1.08, P < 0.001), in patients who were obese (according to BMI and %BF) with concurrent DM (CR 5 4.04, 95% CI 5 1.75-9.31, P 5 0.001), and in patients who were obese (by BMI) with HTN or CAD (CR 5 4.78, 95% CI 5 1.37-16.76, P 5 0.01). Higher rates of posttransplant respiratory events (eg, pneumonia and pleural effusion) were seen in patients who were obese according to %BF (CR 5 1.02, 95% CI 5 1.00-1.05, P 5 0.03). Higher rates of acute renal failure were seen in patients who were obese and had DM (CR 5 4.021, 95% CI 5 1.05-15.41, P 5 0.04) but not in patients who were only obese. There was no significant association between gastrointestinal, neurological, metabolic, hematological, surgical, or transplant-related events (including acute rejection rates) with any of the variables in our model.
Length of Hospital and ICU Stay The median posttransplant hospital stay was 10 days (range 5 5-65 days), and the median ICU stay was 1 day (range 5 1-15 days). Table 5 shows the results of multiple regression analyses for hospital and ICU stays. The hospital stay was significantly increased by 6.99 days (95% CI 5 1.60-12.39 days, P 5 0.01) when NAFLD was the primary indication for LT. The length of the hospital stay was increased for patients who were obese and concurrently had DM by 5.81 days (95% CI 5 1.75-9.87 days, P 5 0.01) when BMI was used as the measure of obesity and by 6.82 days (95% CI 5 0.91-12.73 days, P 5 0.03) when %BF was used as the measure. It was also increased with preexisting CAD by 7.62 days (95% CI 5 1.30-13.95, P 5 0.02). The addition of a third metabolic risk factor (HTN or CAD) for patients who were obese and had DM did not increase the length of the hospital stay; paradoxically, this subset of patients actually had shorter stays than patients
286 DARE ET AL.
LIVER TRANSPLANTATION, March 2014
TABLE 4. Total Complication Events CR (95% CI) MELD score Operation time (minutes) Cold ischemia time (minutes) Concurrent hepatocellular carcinoma BMI > 30 kg/m2 %BF: 30% for men and 42% for women DM HTN CAD Risk interactions BMI and DM BMI 30 kg/m2 1 DM BMI 30 kg/m2 only (no DM) DM only (BMI < 30 kg/m2) BMI < 30 kg/m2, no DM %BF and DM* Obese by %BF 1 DM Obese by %BF only (no DM) DM only (nonobese by %BF) Nonobese by %BF, no DM BMI and cardiac risk factor (HTN or CAD) BMI 30 kg/m2 1 HTN or CAD BMI 30 kg/m2 only (no HTN or CAD) HTN or CAD only (BMI < 30 kg/m2) BMI < 30 kg/m2, no HTN or CAD %BF and cardiac risk factor (HTN or CAD)* Obese by %BF 1 HTN or CAD Obese by %BF only (no HTN or CAD) HTN or CAD only (nonobese by %BF) Nonobese by %BF, no HTN or CAD BMI, DM, and HTN or CAD BMI 30 kg/m2 1 DM 1 HTN or CAD BMI 30 kg/m2 1 DM 2 HTN or CAD BMI < 30 kg/m2, no DM, no HTN or CAD %BF, DM, and HTN or CAD* Obese by %BF 1 DM 1 HTN or CAD Obese by %BF 1 DM 2 HTN or CAD Nonobese by %BF, no DM, no HTN or CAD
1.02 1.00 0.99 0.76 1.03 1.02 1.4 1.14 1.07
P Value
(1.00-1.03) (0.99-1.00) (0.99-1.00) (0.63-0.93) (1.01-1.04) (1.01-1.02) (1.16-1.69) (0.90-1.44) (0.71-1.61)
0.02 0.23 0.37
ORIGINAL ARTICLE
Additive Effect of Pretransplant Obesity, Diabetes, and Cardiovascular Risk Factors on Outcomes After Liver Transplantation Anna J. Dare,1 Lindsay D. Plank,1 Anthony R. J. Phillips,1,2,4 Edward J. Gane,4 Barry Harrison,4 David Orr,4 Yannan Jiang,3 and Adam S. J. R. Bartlett1,2,4 1 Department of Surgery, Faculty of Medical and Health Sciences, 2Maurice Wilkins Centre for Molecular Biodiscovery, University of Auckland, Auckland, New Zealand, and 3Department of Statistics, Faculty of Science, University of Auckland, Auckland, New Zealand; and 4New Zealand Liver Transplant Unit, Auckland City Hospital, Auckland, New Zealand
The effects of pretransplant obesity, diabetes mellitus (DM), coronary artery disease (CAD), and hypertension (HTN) on outcomes after liver transplantation (LT) are controversial. Questions have also been raised about the appropriateness of the body mass index (BMI) for assessing obesity in patients with end-stage liver disease. Both issues have implications for organ allocation in LT. To address these questions, we undertook a cohort study of 202 consecutive patients (2000-2010) undergoing LT at a national center in New Zealand. BMI and body fat percentage (%BF) values (dual-energy X-ray absorptiometry) were measured before transplantation, and the methods were compared. The influence of pretransplant risk variables (including obesity, DM, CAD, and HTN) on the 30-day postoperative event rate, length of hospital stay, and survival were analyzed. There was agreement between the calculated BMI and the measured %BF for 86.0% of the study population (j coefficient 5 0.73, 95% confidence interval 5 0.61-0.85), and this was maintained across increasing Model for EndStage Liver Disease scores. Obesity was an independent risk factor for the postoperative event rate [count ratio (CR) 5 1.03, P < 0.001], as was DM (CR 5 1.4, P < 0.001). Obesity with concomitant DM was the strongest predictor of the postoperative event rate (CR 5 1.75, P < 0.001) and a longer hospital stay (5.81 days, P < 0.01). Independent metabolic risk factors had no effect on 30-day, 1-year, or 5-year patient survival. In conclusion, BMI is an adequate tool for assessing obesity-associated risk in LT. Early post-LT morbidity is highest for patients with concomitant obesity and DM, although C 2014 AASLD. these factors do not appear to influence recipient survival. Liver Transpl 20:281-290, 2014. V Received June 7, 2013; accepted November 7, 2013.
See Editorial on Page 253 Obesity, diabetes mellitus (DM), and metabolic syndrome [comprising increased waist circumference, hyperglycemia, hypertension (HTN), and dyslipidemia1] are now the most significant public health prob-
lems facing developed countries.2,3 They are encountered in patients in virtually every area of health care, including patients with end-stage liver disease (ESLD) undergoing liver transplantation (LT). The impact of obesity on postoperative outcomes and survival after surgical procedures and its place in the allocation of health resources and expenditures have
Abbreviations: %BF, body fat percentage; BMI, body mass index; CAD, coronary artery disease; CI, confidence interval; CR, count ratio; DM, diabetes mellitus; DXA, dual-energy X-ray absorptiometry; ESLD, end-stage liver disease; HTN, hypertension; ICU, intensive care unit; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease. This work was supported by the University of Auckland Postgraduate Research Fund. The authors have no conflicts of interest to declare. Address reprint requests to Adam S. J. R. Bartlett, M.B.Ch.B., Ph.D., F.R.A.C.S., New Zealand Liver Transplant Unit, Auckland City Hospital, Park Road, Grafton, Auckland, New Zealand. Telephone: 164 9 3074949 or 164 212414647; FAX: 164 9 3754345; E-mail: [email protected] DOI 10.1002/lt.23818 View this article online at wileyonlinelibrary.com. LIVER TRANSPLANTATION.DOI 10.1002/lt. Published on behalf of the American Association for the Study of Liver Diseases
C 2014 American Association for the Study of Liver Diseases. V
282 DARE ET AL.
received considerable attention in the literature in recent years.4-7 Analyzing the impact of obesity on the outcomes of patients undergoing LT has been considered challenging because the conventional measurement of obesity, the body mass index (BMI), may overestimate obesity in patients with ESLD who have concomitant ascites.8,9 Furthermore, BMI obesity cutoff definitions may not be appropriate for nonEuropean ethnic groups, who make up a significant proportion of LT recipients worldwide.10,11 Despite this, morbid obesity, as measured by BMI, is used by some centers as an absolute contraindication to LT,8 and there is anecdotal evidence that recipient obesity continues to influence resource allocation decisions made by physicians.7 Although there has been substantial research into posttransplant obesity and metabolic syndrome, the additive impacts of pretransplant obesity and other pretransplant metabolic risk factors such as DM, HTN, and coronary artery disease (CAD) on postoperative outcomes have received much less attention. LT recipients who develop posttransplant DM or other features of metabolic syndrome have an increased risk of cardiovascular complications.12,13 It seems reasonable to hypothesize that similar risks may exist for patients with pretransplant metabolic syndrome undergoing LT. During the last 3 decades, graft loss and recipient deaths as a result of rejection have declined significantly as a result of improved immunosuppressive regimens. Death with a functioning graft, often a result of cardiovascular disease, is now a significant cause of posttransplant mortality and accounts for up to 42% of non–graft-related deaths.14 There is mounting evidence to suggest that obesity, insulin resistance, and type II DM represent a spectrum of metabolic diseases, with the greatest clinical risk clustering in those with an obese and diabetic phenotype.15-18 Understanding the independent and additive risks of obesity, DM, and metabolic syndrome for outcomes after LT may represent an important aspect of pretransplant management for optimizing both early and long-term outcomes. This is particularly important when we are considering whether allocation decisions based on obesity and other lifestyle diseases are appropriate. The objective of this study was to evaluate the independent and additive risks of pretransplant obesity, DM, HTN, and CAD on early and late postoperative outcomes of LT.
PATIENTS AND METHODS We conducted a single-center, retrospective analysis of outcomes for consecutive patients undergoing LT at the New Zealand Liver Transplant Unit of Auckland City Hospital between November 2000 and May 2010. Patients were excluded if they were 90 mm Hg, or systolic blood pressure > 160 mm Hg), pretransplant CAD (documented myocardial ischemia or coronary angiogram evidence), calculated Model for End-Stage Liver Disease (MELD) score at the time of transplantation, donor type (donation after brain death or living donation), cold ischemia time (minutes), and operative time (minutes). [All LT candidates presenting for an assessment underwent a clinical examination, electrocardiography, and 2dimensional echocardiography as part of a standard cardiac workup; patients with a history of cardiovascular disease or risk factors, such as an age > 60 years, a family history, DM, HTN, hyperlipidemia, nonalcoholic fatty liver disease (NAFLD), or renal failure, or assessment findings suggestive of occult cardiovascular disease were referred for stress testing and further management as appropriate.] Posttransplant factors were obtained through a retrospective review of electronic medical records and included the length of the intensive care unit (ICU) stay and the total
LIVER TRANSPLANTATION, Vol. 20, No. 3, 2014
DARE ET AL. 283
TABLE 1. Modified Clavien Classification System for Surgical Complications Events
Mean
Definition
(n)
(Range)
Any need for a deviation from the normal postoperative course without the need for pharmacological treatment or surgical, endoscopic, or radiological interventions. Allowed therapeutic regimens include drugs as anti-emetics, antipyretics, analgesics, diuretics, and electrolytes. This grade also includes wound infections opened at the bedside. Complication requiring pharmacological treatment with drugs other than those allowed for grade I complications. Blood transfusions and parenteral nutrition are also included. Complication requiring surgical, endoscopic, or radiological intervention Intervention not under general anesthesia Intervention under general anesthesia Life-threatening complication (including central nervous system complications*) requiring intensive care management Single-organ dysfunction Multiorgan dysfunction Death of a patient
242
1.2 (0-5)
275
1.36 (0-4)
28 29
0.14 (0-2) 0.14 (0-2)
16 5 0
0.08 (0-1) 0.02 (0-1) 0
Grade I
II
III IIIa IIIb IV IVa IVb V
NOTE: This table has been adapted with permission from Annals of Surgery.21 The total number of events recorded for each complication grade as well as the means and ranges for event rates per complication grade are shown. *Brain hemorrhage, ischemic stroke, and subarachnoidal bleeding (transient ischemic attacks are excluded).
hospital stay after transplantation, postoperative events occurring within the first 30 days after transplantation, and patient survival. Postoperative events were classified with the validated modified classification of surgical complications by Dindo et al.21 This classification system assigns a grade from I to V on the basis of the severity of a complication and the therapeutic intervention required (see Table 1). We also performed a second outcome analysis based on the type of postoperative complication by etiology with the following categories: infective, surgical, cardiovascular, respiratory, gastrointestinal, renal, neurological, hematological, metabolic, and immunological events (including acute rejection). This study was undertaken in accordance with the study hospital ethics committee requirements and a priori ethics approval was granted.
in the ICU. In this model, a significant positive estimate indicated a longer mean stay per unit change in the variable of interest. Poisson regression analysis was used to model total complication events and events by etiology (infective, surgical, cardiovascular, respiratory, gastrointestinal, renal, neurological, hematological, metabolic, or immunological). In this model, a significant estimate > 1 indicated a higher event rate per unit change in the variable of interest. Kaplan-Meier survival estimates were used to illustrate differences in patient survival between groups, and associated P values were derived from a univariate log-rank test (MedCalc version 12). Statistical analyses were performed with SAS 9.2 (SAS Institute, Inc., Cary, NC) and with R 2.14.1 (R Foundation for Statistical Computing). All statistical tests were 2-tailed at a 5% significance level.
Statistical Analysis Baseline recipient and perioperative characteristics for all patients were summarized with descriptive statistics. Continuous variables were presented as means, standard deviations, and ranges. Categorical variables were presented as observed numbers and percentages. Raw agreement (%) and j coefficients were used to evaluate the agreement between obesity classifications by BMI and %BF overall and within MELD score and ethnicity subgroups. For each outcome measure, the variables of interest were fit one at a time in regression models with adjustments for age, sex, and ethnicity as confounding factors. More specifically, linear regression analysis was used to model days in the hospital and days
RESULTS Two hundred two consecutive patients undergoing LT met the study criteria. The baseline characteristics and the perioperative variables are shown in Table 2. The mean recipient age was 51 years (range 5 19-69 years), and 66.8% of the patients (n 5 135) were male. The most common indication for LT was hepatitis C virus cirrhosis (39.1%). BMI values at the time of transplantation were available for 192 patients: 8 patients (4.2%) were underweight, 52 patients (27.1%) were within the normal range, 50 (26.0%) were overweight, 53 (27.6%) were obese, 18 (9.4%) were severely obese, and 11 (5.7%) were morbidly obese. For the multivariate analysis,
284 DARE ET AL.
LIVER TRANSPLANTATION, March 2014
TABLE 2. Baseline Recipient and Perioperative Characteristics (n 5 202) Categorical Variable Sex Male Female Ethnicity European Maori/Pacific Asian Other Indications for transplant Autoimmune Alcoholic liver disease Hepatitis C virus Hepatitis B virus NAFLD Other Concurrent hepatocellular carcinoma Yes No Missing DM Yes No HTN Yes No CAD Yes No Missing Continuous Variable Age (years) BMI (kg/m2) Total body fat (%) MELD score Cold ischemia time (minutes) Operation time (minutes)
the small numbers of patients who were severely or morbidly obese were grouped with the obese patients. Forty-eight patients (23.8%) had DM at the time of transplantation. More than half of the diabetic patients (58.7%) were also obese according to BMI. Thirty-two patients (15.8%) had HTN at the time of transplantation, and 10 (5%) had CAD. Fifty-three of the patients defined as obese by BMI (64.6%) had 1 additional metabolic risk factor (DM, HTN, or CAD), and 13 of the obese patients (16%) had 2 additional risk factors (DM, HTN, or CAD). One patient was obese and had all 3 metabolic risk factors (DM, HTN, and CAD). CAD and HTN were, therefore, grouped together when we evaluated risk interactions in the multivariate analysis to allow more meaningful comparisons. %BF values were available at the time of listing for 129 patients and at the time of transplantation for 69 patients. Sixty-four of these patients had %BF measurements at both times, so we performed a correlation
n
%
135 67
66.8 33.2
128 46 16 12
63.4 22.8 7.9 5.9
28 17 79 39 14 25
13.9 8.4 39.1 19.3 6.9 12.4
68 132 2
33.7 65.3 1.0
48 154
23.8 76.2
32 170
15.8 84.2
10 191 1
5.0 94.6 0.5
Mean 6 Standard Deviation 51 6 10 28.9 6 6.1 28.69 6 7.56 15.0 6 6.2 434.5 6 161.0 372.6 6 99.4
Range 19-69 17.6-46.3 6.15-51.70 6.4-44.4 23-960 211-1025
analysis and a paired t test for these patients. This showed that %BF values at the time of listing and on the day of transplantation were highly correlated (r 5 0.92, Pearson’s correlation coefficient), and there was no significant difference between the 2 values (paired Student t test). We, therefore, used %BF on the day of listing as a surrogate measure in our analysis when %BF on the day of transplantation was not available. This enabled a more robust data modeling process because of the larger number of patients with %BF data for the day of listing.
Agreement Between BMI and %BF for the Assessment of Obesity The results of the comparison of the BMI and %BF techniques for the assessment of obesity are shown in Table 3. There was agreement between BMI and %BF for the classification of patients into obese or nonobese groups for 86.0% of the study population [j
LIVER TRANSPLANTATION, Vol. 20, No. 3, 2014
DARE ET AL. 285
TABLE 3. Agreement Between BMI and %BF as Measures of Obesity in Patients Undergoing LT: Overall, by MELD Score, and by Ethnicity Overall Agreement
j Coefficient
(%)
(95% CI)
86.0
0.73 (0.61-0.85)
85.7 82.1
0.70 (0.46-0.94) 0.60 (0.31-0.89)
91.4 71 83
0.84 (0.71-0.96) 0.49 (0.21-0.77) 0.56 (0.12-1.0)†
Overall MELD score Low (17) Ethnicity* European Maori/Pacific Asian/other
*A significant difference in the strength of agreement between BMI and %BF was observed for European patients versus Maori/Pacific patients. † There were insufficient numbers to draw accurate conclusions for patients of Asian/other ethnicity.
coefficient 5 0.73, 95% confidence interval (CI) 5 0.610.85]. Europeans were significantly more likely to show agreement between obesity cutoffs (91.4%, j coefficient 5 0.84, 95% CI 5 0.71-0.96) than Maori and Pacific patients (71%, j coefficient 5 0.49, 95% CI 5 0.21-0.77). Patients with MELD scores in the lowest quartile (score < 11) had 85.7% agreement between BMI and %BF cutoffs (j coefficient 5 0.70, 95% CI 5 0.46-0.94). This was not significantly different from the situation for patients with MELD scores in the highest quartile (>17), for whom there was agreement 82.1% of the time (j coefficient 0.60, 95% CI 5 0.31-0.89).
Postoperative Events Postoperative event rates by complication grade are shown in Table 1: 70.2% of the patients experienced at least 1 grade I complication, 70.9% experienced at least 1 grade II complication, 12.4% experienced at least 1 grade IIIa complication, and 12.9% experienced at least 1 grade IIIb complication. A severe (grade IV) complication was experienced by 10.4% of the patients, with 7.9% experiencing a grade IVa complication and 2.5% experiencing a grade IVb complication. Several independent risk factors related to recipients’ pretransplant metabolic risk profiles and features of their underlying liver disease were associated with an increased postoperative event rate in our regression model. Obesity (according to both BMI and %BF cutoffs), DM, and the MELD score were each predictive of a higher postoperative event rate (Table 4). When we evaluated additive risk factors, obesity (by both BMI and %BF) with concurrent DM was the strongest predictor of a higher postoperative event rate [count ratio (CR) 5 1.75, 95% CI 5 1.37-2.25, and P < 0.001 when
BMI was used as the measure of obesity; CR 5 1.63, 95% CI 5 1.17-2.29, and P 5 0.004 when %BF was used as the measure]. The presence of obesity, DM, and a concurrent third metabolic risk factor (either HTN or CAD) did not significantly increase the risk (CR 5 1.65, 95% CI 5 1.16-2.35, P 5 0.005; Table 4). Obesity with cardiac risk factors (HTN or CAD) in the absence of DM also did not significantly increase the postoperative event rate (Table 4). A higher incidence of posttransplant infections (eg, wound infections, bacteremia, and pneumonia) were seen in recipients with higher MELD scores (CR 5 1.06, 95% CI > 1.02-1.10, P 5 0.004), in patients with NAFLD as the indication for LT (CR 5 4.52, 95% CI 5 1.6012.81, P 5 0.005), and in patients who were obese (according to BMI and %BF) with concurrent DM (CR 5 3.56, 95% CI 5 1.52-8.32, P 5 0.005). Higher rates of posttransplant cardiovascular events (eg, arrhythmia and myocardial infarction) were seen in patients with prolonged operative times (CR 5 1.01, 95% CI 5 1.00-1.01, P 5 0.04), in patients who were obese according to BMI (CR 5 1.11, 95% CI 5 1.061.16, P < 0.001), in patients who were obese according to %BF (CR 5 1.05, 95% CI 5 1.03-1.08, P < 0.001), in patients who were obese (according to BMI and %BF) with concurrent DM (CR 5 4.04, 95% CI 5 1.75-9.31, P 5 0.001), and in patients who were obese (by BMI) with HTN or CAD (CR 5 4.78, 95% CI 5 1.37-16.76, P 5 0.01). Higher rates of posttransplant respiratory events (eg, pneumonia and pleural effusion) were seen in patients who were obese according to %BF (CR 5 1.02, 95% CI 5 1.00-1.05, P 5 0.03). Higher rates of acute renal failure were seen in patients who were obese and had DM (CR 5 4.021, 95% CI 5 1.05-15.41, P 5 0.04) but not in patients who were only obese. There was no significant association between gastrointestinal, neurological, metabolic, hematological, surgical, or transplant-related events (including acute rejection rates) with any of the variables in our model.
Length of Hospital and ICU Stay The median posttransplant hospital stay was 10 days (range 5 5-65 days), and the median ICU stay was 1 day (range 5 1-15 days). Table 5 shows the results of multiple regression analyses for hospital and ICU stays. The hospital stay was significantly increased by 6.99 days (95% CI 5 1.60-12.39 days, P 5 0.01) when NAFLD was the primary indication for LT. The length of the hospital stay was increased for patients who were obese and concurrently had DM by 5.81 days (95% CI 5 1.75-9.87 days, P 5 0.01) when BMI was used as the measure of obesity and by 6.82 days (95% CI 5 0.91-12.73 days, P 5 0.03) when %BF was used as the measure. It was also increased with preexisting CAD by 7.62 days (95% CI 5 1.30-13.95, P 5 0.02). The addition of a third metabolic risk factor (HTN or CAD) for patients who were obese and had DM did not increase the length of the hospital stay; paradoxically, this subset of patients actually had shorter stays than patients
286 DARE ET AL.
LIVER TRANSPLANTATION, March 2014
TABLE 4. Total Complication Events CR (95% CI) MELD score Operation time (minutes) Cold ischemia time (minutes) Concurrent hepatocellular carcinoma BMI > 30 kg/m2 %BF: 30% for men and 42% for women DM HTN CAD Risk interactions BMI and DM BMI 30 kg/m2 1 DM BMI 30 kg/m2 only (no DM) DM only (BMI < 30 kg/m2) BMI < 30 kg/m2, no DM %BF and DM* Obese by %BF 1 DM Obese by %BF only (no DM) DM only (nonobese by %BF) Nonobese by %BF, no DM BMI and cardiac risk factor (HTN or CAD) BMI 30 kg/m2 1 HTN or CAD BMI 30 kg/m2 only (no HTN or CAD) HTN or CAD only (BMI < 30 kg/m2) BMI < 30 kg/m2, no HTN or CAD %BF and cardiac risk factor (HTN or CAD)* Obese by %BF 1 HTN or CAD Obese by %BF only (no HTN or CAD) HTN or CAD only (nonobese by %BF) Nonobese by %BF, no HTN or CAD BMI, DM, and HTN or CAD BMI 30 kg/m2 1 DM 1 HTN or CAD BMI 30 kg/m2 1 DM 2 HTN or CAD BMI < 30 kg/m2, no DM, no HTN or CAD %BF, DM, and HTN or CAD* Obese by %BF 1 DM 1 HTN or CAD Obese by %BF 1 DM 2 HTN or CAD Nonobese by %BF, no DM, no HTN or CAD
1.02 1.00 0.99 0.76 1.03 1.02 1.4 1.14 1.07
P Value
(1.00-1.03) (0.99-1.00) (0.99-1.00) (0.63-0.93) (1.01-1.04) (1.01-1.02) (1.16-1.69) (0.90-1.44) (0.71-1.61)
0.02 0.23 0.37
