Ascites and serum sodium are markers of increased waiting list mortality in children with chronic liver failure

Authors

  • Renata Pugliese,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Eduardo A. Fonseca,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Gilda Porta,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Vera Danesi,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Teresa Guimaraes,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Adriana Porta,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Irene K. Miura,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Cristian Borges,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Helry Candido,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Marcel Benavides,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Flavia H. Feier,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Andre Godoy,

    1. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Rita Antonelli Cardoso,

    1. STATS Estatistica em Ciências e Negócios
    Search for more papers by this author
  • Mario Kondo,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    Search for more papers by this author
  • Paulo Chapchap,

    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    Search for more papers by this author
  • Joao Seda Neto

    Corresponding author
    1. Hepatology and Liver Transplantation, Sirio-Libanês Hospital and Menino Jesus Hospital
    2. Hepatology and Liver Transplantation, AC Camargo Hospital
    • Address reprint requests to: Joao Seda Neto, M.D., Ph.D., Rua Barata Ribeiro, 414, cj 65, Bela Vista, Sao Paulo, Brazil, 01308000. E-mail: joaoseda@gmail.com; fax: +55-11-32310900.

    Search for more papers by this author

  • See Editorial on Page 1678

    Potential conflict of interest: Nothing to report.

Abstract

Ascites is the most common complication of cirrhosis and in adults it is associated with 50% mortality at 5 years if patients do not receive a liver transplant. The occurrence of hyponatremia in these patients has been associated with increased mortality on the waiting list. The importance of serum sodium levels and the presence of ascites in the pediatric setting remain to be clarified. A retrospective analysis of pediatric patients with cirrhosis on the transplant list was carried out between October 2000 and February 2012. The primary objective of this study was to evaluate the association of pretransplant variables with mortality within 90 days following the inclusion of patients on the waiting list. In all, 522 patients were included in the study; 345 (66%) patients were under 1 year of age; 208 (40%) of the children presented ascites. A multivariate Cox proportional hazards analysis was conducted and total bilirubin (P < 0.001, hazard ratio [HR] = 2.09, 95% confidence interval [CI] = 1.35-3.21), international normalized ratio (INR) (P < 0.001, HR = 9.83, 95% CI = 4.51-21.45), serum sodium levels (P = 0.03, HR = 0.96, 95% CI = 0.92-0.99), ascites (P = 0.001, HR = 2.59, 95% CI = 1.44-4.64), and categorized age (0-1 versus ≥1 year old) (P = 0.025, HR = 2.33, 95% CI = 1.11-4.86) were independently associated with risk of death in 90 days. Malnutrition (Z score height/age, weight/age) and serum albumin (pediatric endstage liver disease [PELD] formula) were not included in the final model. Conclusion: The presence of ascites and serum sodium levels are important variables associated with decreased patient survival while candidates wait for a liver graft. Multicenter studies are necessary to validate these findings in order to improve current allocation policies based on the PELD score. (Hepatology 2014;59:1964–1971)

Abbreviations
GFR

glomerular filtration rate

INR

international normalized ratio

LDLT

living donor liver transplantation

MELD

model for end-stage liver disease

PELD

pediatric endstage liver disease

SBP

spontaneous bacterial peritonitis

Ascites is the most common complication of cirrhosis and in adults it is associated with 50% mortality at 5 years if patients do not receive a liver transplant.[1] In the majority of patients with advanced cirrhosis, hyponatremia develops in the setting of ascites where there is expanded extracellular fluid volume along with increased renal sodium retention.[2] Hyponatremia in these patients determines a 25% probability of survival at 1 year.[1] In the adult, as well as in the pediatric population, the occurrence of hyponatremia has been associated with increased mortality on the waiting list.[3, 4] The debate to improve current liver allocation policies by using serum sodium levels is for the most part restricted to the Model for Endstage Liver Disease (MELD) system, at least if the number of publications on the matter is considered.[5, 6]

The severity-of-illness score developed for children with endstage liver disease (Pediatric Endstage Liver Disease score [PELD]) was derived and statistically tested in a large multicenter database.[7] Before the introduction of the PELD score, other systems developed for assessing disease severity, such as the Child-Turcotte-Pugh score[8] and Malatack et al.'s study,[9] took into consideration subjective parameters such as the presence of ascites and encephalopathy. The latter system was specific for children, but was developed in a smaller single-center patient cohort when compared to the PELD score.

In contrast to the extensive discussion regarding the adult scenario, the importance of serum sodium concentrations and/or the presence of ascites as markers of increased mortality in the pediatric setting remain to be clarified. This article aims to evaluate the factors associated with increased waiting list mortality in patients less than 12 years of age. Among the demographics and laboratory data assessed at the time of patient listing for liver transplantation, the presence of ascites and serum sodium concentrations were confronted with the variables in the formula used to calculate the PELD score.

Materials and Methods

Patients

A retrospective study of pediatric patients with cirrhosis listed for liver transplantation at Hospital Sirio-Libanes, Hospital A. C. Camargo, and Hospital Infantil Menino Jesus, Sao Paulo, Brazil was performed. From October 2000 to February 2012, 553 patients under 18 years of age were listed for transplantation in these hospitals. Patients with fulminant liver failure, liver neoplasms, and metabolic noncirrhotic liver diseases were excluded from the study. Patients under 12 years of age were only included if they had chronic liver failure. Data collection and analysis was undertaken with the approval of the hospitals' Ethics Committees. The patients in these three centers were evaluated and monitored by the same medical team throughout the study period.

Variables analyzed from our database and medical records at the time of listing for transplantation included: age, sex, diagnosis (cholestatic versus noncholestatic liver diseases), presence of ascites, the occurrence of spontaneous bacterial peritonitis (SBP), PELD, albumin, total bilirubin, international normalized ratio (INR), serum sodium concentration, creatinine level, and urea. For patients who were listed and received a transplant before the introduction of the PELD system (July 2006), PELD values were retrospectively calculated using data collected from the time of listing. The PELD score was calculated with the standard formula.[7] All laboratory data were collected within a week of listing.

Hyponatremia was defined as serum sodium concentration persistently (>1 week) <130 mEq/L.[10-12] The patients in this study underwent abdominal ultrasound examination for the diagnosis of ascites the month before being listed for transplantation. The presence of ascites was not quantified, only described as present or absent. Traces of free fluid or minimal ascites were included in the ascites group. Patients presenting with ascites underwent abdominal tapping following the guidelines of the International Ascites Club Consensus, 2000.[13] The diagnosis of SBP was based on neutrophil count in ascitic fluid >250/mm3, as determined by microscopy.

The primary objective of this study was evaluating the association of the described variables with death within 90 days after inclusion of the patients on the waiting list. Patients who met the inclusion criteria were monitored until transplant or death, or else censored at 90 days. The indication for living donor liver transplantation (LDLT) in this study was donor availability once the patient had an indication for liver replacement. LDLT was not restricted to those with features suggesting a poor prognosis, such as hyponatremia, refractory ascites, or patients with high PELD scores. A sensitivity analysis was then performed excluding patients who underwent LDLT.

Statistical Analysis

Univariate and multivariate Cox proportional hazard analysis was conducted. Patients who survived on the waiting list were censored at 90 days. Patients who received a liver transplant were censored at the time of transplantation. Date of registration on the waiting list was used as the onset time. Univariate Cox proportional hazard model of the predictor variables was performed and those found to be significant at the 0.05 alpha levels were selected for the multivariate analysis. The final multivariate model was obtained by a forward selection method (criteria for variable inclusion was P < 0.10). Fisher's exact test was also used in order to compare proportions. The c-statistic was calculated to assess the discriminative ability of serum sodium to predict death. It corresponds to the area under the receiver operative characteristic (ROC) curve. The c-statistic equals 0.50 when the model has no ability to discriminate and equals 1.0 when it has perfect prediction. Survival analysis was conducted according to the Kaplan-Meier product-limit estimates, and patient subgroups were compared using a two-sided log-rank test. SAS v. 9.2 (Cary, NC) was used for statistical analysis.

Results

Of the 553 patients listed for transplantation, 31 were excluded and 522 fulfilled the inclusion criteria (chronic liver failure). The demographic data, as well as the clinical characteristics and laboratory results in this cohort, are summarized in Table 1. It shows the values of each variable for patients who were alive on the waiting list or were transplanted in comparison to those who died waiting for a liver graft. The results represent the laboratory values at the time the patients were included on the liver waiting list. There were 294 (56%) girls and 345 (66%) patients were under 1 year of age. The diagnoses were: 377 (72%) biliary atresia, 54 (10%) alpha-1 antitrypsin deficiency, 40 (7.6%) cryptogenic cirrhosis, 33 (6.3%) autoimmune hepatitis, 11 (2%) type 1 tyrosinemia, 5 (1%) biliary hypoplasia, and 2 (0.3%) Wilson's disease. Two hundred and eight (40%) children presented ascites and 32 patients had at least one episode of SBP. Thirty-one (6%) patients had hyponatremia and the median PELD score was 13 (range −10 to 58). Children with hyponatremia presented higher rates of ascites (79% versus 37%; P < 0.001). Additionally, children with hyponatremia had higher rates of SBP (15% versus 5%; P = 0.022, Fischer's exact test). The status of the patients at the end of the observation period (90 days) is shown in Table 2. The causes of death on the waiting list were: 27 (44.3%) sepsis, 12 (19.7%) progressive liver failure, 10 (16.4%) gastrointestinal bleeding, 10 (16.4%) pulmonary hemorrhage, and for other reasons in 2 (3.3%) patients.

Table 1. Demographics, Clinical Baseline Characteristics, and Laboratory Values in 522 Patients Under 12 Years Listed to Liver Transplantation
  Death Before Transplantation 
  No (n = 461)Yes (n = 61)Overall (n = 522)
  1. SD, standard deviation; SBP, spontaneous bacterial peritonitis; INR, international normalized ratio; PELD, Pediatric Endstage Liver Disease.

Age (m)N46161522
 Mean (SD)20.9 (29.3)14.3 (24.8)20.1 (28.9)
 Median (range)8.6 (0.97-138.9)7.48.3
   (2.1-120)(0.9-138.9)
Age class (yr), n (%)(0-1)295 (64)50 (82)345 (66)
 ≥1166 (36)11 (18)177 (34)
Gender, n (%)Female261 (65.6)33(54)294 (56)
 Male200 (43.4)28 (46)228 (44)
Growth (z-score), n (%)<-2225 (49)35 (57)260 (49.8)
Height or weight/age, n (%)≥-2236 (51)26 (43)262 (50.2)
Cholestatic disease, n (%)Yes383 (83)48 (78.6)431 (82.6)
 No78 (17)13 (21.4)91 (17.4)
Ascites, n (%)Yes167 (36.2)41 (67.2)208 (40)
 No294 (63.8)20 (32.8)314 (60)
SBP, n (%)Yes23 (5)9 (14.7)32 (6)
 No438 (95)52 (85.3)490 (94)
Total bilirubin (mg/dl)N46161522
 Mean (SD)11.4 (8.1)17.1 (9.8)12.1 (8.52)
 Median (range)10.8 (0.2-57.8)15.8 (1.1-44.5)11.4 (0.2-57.8)
Albumin (g/dl)N46061521
 Mean (SD)3.0 (0.7)2.53 (0.81)2.9 (0.7)
 Median (range)2.9 (1.2-5.9)2.40 (1.2-4.8)2.9 (1.2-5.9)
INRN46160521
 Mean (SD)1.3 (0.7)1.78 (0.8)1.39 (0.7)
 Median (range)1.19 (0.4-7.9)1.50 (0.8-5.8)1.2 (0.4-7.9)
Sodium (mEq/l)N43860498
 Mean (SD)135.6 (3.91)131.7 (8.1)135.1 (4.7)
 Median (range)136 (104-146)133 (100-145)136 (100-146)
Sodium (mEq/l), n (%)<13017 (3.8)14 (23.3)31 (6)
 130-135140 (32.1)21 (35)161 (32.4)
 ≥135281 (64.1)25 (41.7)306 (61.6)
Urea (mg/dl)N44761508
 Mean (SD)21.3 (11.2)21.5 (14.1)21.3 (11.6)
 Median (range)19 (3-98)17 (5-68)19 (3-98)
Creatinine (mg/dl)N44961510
 Mean (SD)0.28 (0.16)0.03 (0.14)0.28 (0.16)
 Median (range)0.2 (0.03-1.2)0.28 (0.1-0.6)0.23 (0.03-1.2)
PELD-scoreN46059519
 Mean (SD)12.7 (9)22.7 (8.8)13.8 (9.5)
 Median (range)12 (−10-58)22 (−2-38)13 (−10-58)
Table 2. Patients' Disposition at the End of the Study Period
 n = 522
  1. LDLT, living donor liver transplantation; OLT, orthotopic liver transplant; SD, standard deviation.

Patients underwent liver transplantation, n (%)207 (39.6)
Type of transplant, n (%) 
LDLT188 (90.8)
OLT18 (9.2)
Patients status, n (%)315 (60)
Waiting for liver transplantation254 (48.6)
Death in waiting-list61 (11.4)
Time (days) between inclusion and transplantation 
N207
Mean (SD)49.5 (25.6)
Median50
Min-Max1-90

The univariate Cox proportional hazard model was initially used to evaluate the relationship between each variable and the risk of death within 90 days after inclusion on the transplant waiting list. Categorized age (0-1, and >1 year), ascites, SBP, total bilirubin, INR, albumin, sodium, and PELD score were statistically significantly associated with pretransplant death in the univariate model (Table 3). A multivariate Cox proportional hazards analysis considering all the variables that were statistically significant on the univariate analysis, except the PELD score, was conducted. The results are shown in Table 4. Total bilirubin (P < 0.001, hazard ratio [HR] = 2.09, 95% confidence interval [CI] = 1.35-3.21), INR (P < 0.001, HR = 9.83, 95% CI = 4.51-21.45), sodium (P = 0.038, HR = 0.96, 95% CI = 0.93-0.99), ascites (P = 0.001, HR = 2.59, 95% CI = 1.45-4.64), and the categorized age (P = 0.025, HR = 2.33, 95% CI = 1.11-4.87) held statistical significance in the final model and were associated with risk of death within 90 days after inclusion of the patient on the waiting list. When sodium was studied as a continuous variable, for each unit increase in the serum sodium concentration there was a 4.1% decrease in the risk of death (P = 0.03).

Table 3. Association of the Clinical/Laboratory Variables at the Time the Patients Were Enrolled on the Liver List and Death in 90 Days After Listing (Univariate Cox Proportional Hazard Model)
Variable Hazard Ratio95% CIP Value
  1. SD, standard deviation; SBP, spontaneous bacterial peritonitis; INR, international normalized ratio; PELD, Pediatric Endstage Liver Disease.

Age class(0-1) years2.71(1.37-5.34)0.004
GenderFemale0.94(0.57-1.57)0.820
GrowthStandard height or weight<-21.43(0.86-2.39)0.173
Cholestatic DiseaseYes0.80(0.43-1.51)0.498
AscitesYes3.97(2.32-6.80)0.001
SBPYes3.03(1.44-6.39)0.004
Total Bilirrubin (loge TB+1) 3.02(1.91-4.77)0.001
Albumin (loge Alb+1) 0.03(0.01-0.11)0.001
INR (loge INR+1) 11.01(5.78-20.99)0.001
Urea (loge U+1) 1.01(0.98-1.02)0.877
Creatinine (loge Cr+1) 2.02(0.42-9.64)0.379
Sodium (mEq/l) 2.34(1.40-3.92)0.001
PELD 1.12(1.10-1.15)0.001
Table 4. Multivariate Cox Proportional Hazard Test Showing the Association of Clinical/Laboratory Variables at the Time the Patients Were Enrolled and Death on the Liver Waiting List
VariableHazard Ratio95% CIP Value
  1. INR, international normalized ratio.

Age class2.321.11-4.860.025
Ascites2.601.44-4.640.001
Log Total Bilirubin2.101.35-3.21<0.001
Log INR9.834.51-21.45<0.001
Sodium (mEq/l)0.960.92-0.990.038

The PELD score, ascites, and sodium were then selected for a multivariate Cox regression analysis, to examine if these two variables remained associated with 90-day mortality even after adjusting for the PELD score. Ascites held statistical significance in the model (P = 0.023, HR = 1.97, 95% CI = 1.09-3.5) and sodium presented a borderline significance level (P = 0.057, HR = 0.96, 95% CI = 0.93-1). A further analysis showed that the sodium c-statistics for prediction of death within 90 days from listing was 0.66 (0.57-0.74).

The Kaplan-Meier product-limit estimates of the survival time according to categorized sodium values (<130 mEq/l, 130 to <135 mEq/l, and ≥135 mEq/l) and the presence of ascites are shown in Figure 1A,B, respectively. The survival rates at 40, 60, and 90 days are plotted on the curves. The comparison between groups using a two-sided log-rank test shows a statistically significant lower patient survival (90 days) in patients with hyponatremia (P < 0.001) and in patients with ascites at the time of listing (P < 0.001).

Figure 1.

(A) Kaplan-Meier survival curves: categorized sodium values. (B) Kaplan-Meier survival curves: ascites: present versus absent.

Even though LDLT was not restricted to sicker patients, as stated in the Materials and Methods, a further analysis excluding patients who underwent LDLT was performed. Table 5 shows the demographics of the 344 remaining patients. The same univariate and multivariate Cox proportional-hazard analysis was conducted and the results are shown in Tables 6 and 7, respectively. Categorized age (0-1, and >1 year), ascites, SBP, total bilirubin, albumin, INR, and serum sodium were statistically significantly associated with pretransplant death in the univariate model (Table 6). In the multivariate model, total bilirubin (P < 0.001, HR = 2.07, 95% CI = 1.36-3.16), INR (P < 0.001, HR = 23.17, 95% CI = 8.62-62.27), albumin (P = 0.005, HR = 0.14, 95% CI = 0.03-0.56), ascites (P = 0.022, HR = 2.0, 95% CI = 1.10-3.64), and the categorized age (P < 0.001, HR = 4.30, 95% CI 1.87-9.88) held statistical significance (Table 7).

Table 5. Demographics, Clinical Baseline Characteristics, and Laboratorial Values in 334 Patients Under 12 Years Listed to Liver Transplantation
  Death Before Transplantation 
  No (n = 273)Yes (n = 61)Overall (n = 334)
  1. LDLTs were excluded. SD, standard deviation; SBP, spontaneous bacterial peritonitis; INR, international normalized ratio; PELD, Pediatric Endstage Liver Disease; LDLT, living donor liver transplantation.

Age (m)Mean (SD)24.53 (32.66)14.35 (24.86)22.67 (31.59)
 Median (range)9.53 (0.97-136.2)7.47 (2.13-120)8.77 (0.97-136.2)
Age class (yr), n (%)(0-1)163 (59.7)50 (81.9)213 (63.7)
 ≥ 1110 (40.3)11 (18)121 (36.2)
Gender, n (%)Female153 (56)33 (54)186 (55.7)
 Male120 (43.9)28 (45.9)148 (44.3)
Growth (z-score), n (%)<-2140 (51.2)26 (42.6)166 (49.7)
Height or weight/age, n (%)≥-2133 (48.7)35 (57.3)168 (50.3)
Cholestatic disease, n (%)Yes222 (81.3)48 (78.6)270 (80.8)
 No51 (18.6)13 (21.3)64 (19)
Ascites, n (%)Yes84 (30.7)41 (67.2)125 (37.4)
 No189 (69.3)20 (32.8)209 (62.6)
SBP, n (%)Yes12 (4.4)9 (14.7)21 (6.3)
 No261 (95.6)52 (85.3)313 (93.7)
Total bilirrubin (mg/dl)Mean (SD)10.16 (7.31)17.16 (9.8)11.41 (8.2)
 Median (range)9.7 (0.20-51.2)15.8 (1.1-44.5)10.8 (0.2-51.2)
Albumin (g/dl)Mean (SD)3.09 (0.76)2.53 (0.81)2.99 (0.8)
 Median (range)3.1 (1.20-5.9)2.40 (1.2-4.8)2.9 (1.2-5.9)
INRN27360333
 Mean (SD)1.25 (0.55)1.78 (0.86)1.35 (0.65)
 Median (range)1.14 (0.47-7)1.50 (0.8-5.8)1.2 0.47-7)
Sodium (mEq/l)N25960219
 Mean (SD)135.81 (3.7)131.72 (8.1)135.04 (5.08)
 Median (range)136 (122-146)133 (100-145)136 (100-146)
Sodium (mEq/l), n (%)<13011 (4.03)14 (22.9)25 (7.4)
 130-13581 (29.6)21 (34.4)102 (30.5)
 ≥135167 (61.1)25 (40.9)192 (57.5)
Urea (mg/dl)N26461325
 Mean (SD)22.43 (12.5)21.56 (14)22.26 (12.8)
 Median (range)20 (3-98)17 (5-68)20 (3-98)
Creatinine (mg/dl)N26661327
 Mean (SD)0.29 (0.16)0.3 (0.14)0.29 (0.16)
 Median (range)0.26 (0.03-0.82)0.28 (0.1-0.67)0.26 (0.03-0.82)
PELD-scoreN27359332
 Mean (SD)10.72 (8.27)22.75 (8.8)12.86 (9.5)
 Median (range)10 (−10-42)22 (−2-38)12 (−10-42)
Table 6. Association of the Clinical/Laboratory Variables at the Time the 344 Patients Were Enrolled in the Liver Waiting List and Death in 90 Days After Listing (Univariate Cox Proportional Hazard Test)
Variable Hazard Ratio95% CIP Value
  1. LDLTs were excluded. SD, standard deviation; SBP, spontaneous bacterial peritonitis; INR, international normalized ratio; PELD, Pediatric Endstage Liver Disease; LDLT, living donor liver transplantation.

Age class≤1 year3.02(1.53-5.96)<0.001
GenderFemale0.96(0.57-1.59)0.859
Growth (z-score)Height or weight/age<-21.44(0.86-2.40)0.164
Cholestatic diseaseYes0.85(0.45-1.61)0.622
AscitesYes4.16(2.43-7.12)<0.001
SBPYes2.94(1.40-6.20)0.004
Total bilirubin (loge TB+1) 3.38(2.14-5.33)<0.001
Albumin (loge Alb+1) 0.02(0.01-0.08)<0.001
INR (loge INR+1) 17(8.78-32.92)<0.001
Ureia (loge U+1) 1(0.98-1.02)0.774
Creatinine (loge Cr+1) 1.35(0.27-6.62)0.715
Sodium (mEq/l) 0.91(0.88-0.93)<0.001
PELD-score 1.16(1.12-1.19)<0.001
Table 7. Multivariate Cox Proportional Hazard Test Showing the Association of Clinical/Laboratory Variables at the Time 344 Patients Were Enrolled and Death on the Liver Waiting List
VariableHazard Ratio95% CIP Value
  1. LDLTs were excluded. INR, international normalized ratio; LDLT, living donor liver transplantation.

Age class4.301.87-9.88<0.001
Ascites21.10-3.640.022
Log total bilirubin2.071.36-3.16<0.001
Log albumin0.140.03-0.560.005
Log INR23.178.62-62.27<0.001

Discussion

The development of the MELD/PELD scores respected the idea that patient prognosis is an important consideration in allocating a resource in short supply.[14] The implementation of these scores represented an advance toward allocation of deceased donor livers. Parallel to the use of the mathematical model, the allocation policy in the United States retained the Status 1 designation for both acute and chronic liver disease and established a mechanism to assign increased priority above the objectively calculated score.[15] Many authors addressed the fact that the calculated PELD score determined organ allocation in only up to 50% of the cases[16]; the remaining patients were listed through regional review boards.[19] However, the PELD score has proven to decrease death rates for children waiting for a liver transplant.[18] The other concern with the PELD system is that it appears to stratify the degree of illness in children with similar diseases who are competing for pediatric livers,[16] but it puts children at a disadvantage when they are competing for livers from adult donors.[20, 21] Neto et al.[22] showed strategies to minimize the deficiencies of the PELD score after its introduction into the Brazilian organ allocation system. The waiting list mortality at 6 months and 2 years in the PELD era was 15% and 18%, respectively. In the present study, the 90-day waiting list mortality was 11.6% and most of the pediatric transplants were LDLT. Part of the reason for the high number of LDLT is the low number of suitable donors (size match) for children and suboptimal use of “ideal donors” that are appropriate for liver splitting.[22] This situation makes pediatric liver transplantation in Brazil comparable to what occurs in Eastern countries, where most of the transplants are performed with living donors. In the face of these findings, and based on the discussion of the importance of ascites/hyponatremia in the adult setting, variables that might impact survival on the waiting list were confronted with the ones used for PELD calculation.

The majority of patients with hyponatremia (79%) in this study presented ascites. Indeed, it is difficult to isolate these two variables because they are the expression of the diseased sodium metabolism that occurs with advanced cirrhosis, in adults and in children. The pediatric patients with ascites in this study had a mortality rate of 19.7% as opposed to 6.4% in those without ascites, 90 days after being listed. Even though there is the risk of a “subjective” evaluation of ascites, all the patients in the study underwent an ultrasound examination to check for its presence, without the need for quantification. Malatack et al.[9] identified four risk factors associated with progressive liver disease before transplantation: history of ascites, indirect bilirubin >6 mg/dL, cholesterol <100 mg/dL, and partial thromboplastin time prolonged >20 sec. The statistical analysis presented herein corroborates the importance of three of Malatack et al.'s studied variables, but patients here were censored at 90 days, similar to the PELD study in 2002. On the other hand, McDiarmid et al.[7] did not study the impact of ascites in pretransplant mortality, probably because it was a subjective parameter or because the data were not available in the multicenter study.

Hyponatremia has been associated with hepatorenal syndrome,[23, 24] ascites,[10, 25] and death from liver disease[23, 25, 26] in the adult setting. The physiopathology involved in the development of hyponatremia comprises splanchnic vasodilatation, which leads to decreased effective arterial blood volume and reduced glomerular filtration rate (GFR),[27] nonosmotic hypersecretion of arginine vasopressin with a marked reabsorption of water in the collecting duct, reduced delivery of filtrate to the ascending limb of the Loop of Henle, and decreased renal synthesis of prostaglandins.[28, 29] Consequently, there is impairment of renal excretion of solute-free water and disproportionate renal retention of water compared with sodium retention, resulting in dilutional hyponatremia. Kim et al.[5] showed that the addition of serum sodium values to the MELD score improved survival prediction of patients waiting for a liver transplant. The importance of serum sodium decreased as the MELD score increased. More recently, Leise et al.[6] reported that the impact of hyponatremia was greatest if bilirubin was low, whereas in patients who were already ill with bilirubin >20 mg/dL, hyponatremia did not confer any higher risks.

Carey et al.[4] were the first to report the negative impact of hyponatremia on pretransplant survival in the pediatric population. They showed, in a cohort of 94 patients, that serum sodium concentration <130 mEq/L was associated with an 8-fold increased risk of pretransplant mortality. Interestingly, the PELD score was not predictive of pretransplant mortality in their cohort but, when the authors added hyponatremia to the PELD score, the c-statistic increased and the model became predictive of pretransplant mortality. This showed that the addition of hyponatremia to the PELD score might improve its accuracy in predicting mortality in patients with endstage liver disease. However, this was a publication with a limited number of patients who were not censored at a given timepoint. Similar to that report, where a 5% hyponatremia was observed at the time of listing, the incidence in the present study was 6%. There was 45% mortality at 90 days in patients with low sodium (<130 mEq/L), in contrast to 9.8% mortality in patients without hyponatremia.

Even though the PELD score previously proved to accurately predict the 3-month probability of waiting list death for children with chronic liver failure,[18] some reports did not show this correlation.[4, 16, 17] The findings in this study, however, are not in conflict with the original PELD publication. The strength of serum sodium levels and ascites was tested with the other variables included in the PELD formula[7] within the same observational period (90 days). It is important to point out that the serum sodium concentration was not studied in the original PELD publication, even though it is a continuous variable and less susceptible to manipulation of the results, different from ascites. The prognostic value of hyponatremia may be independent of liver dysfunction and may reflect renal dysfunction or circulatory dysfunction.[4] In children, elevated serum creatinine is a very late finding in kidney dysfunction, secondary to the decreased muscle mass that occurs in malnourished patients. Ascites, serum sodium levels, INR, total bilirubin, and categorized age were independently associated with increased risk of death on the waiting list (Table 5). Malnutrition variables present in the PELD calculation (represented by the Z score height/age, weight/age, and serum albumin) were not included in the final model. The elimination of patients who underwent LDLT in the second part of the study showed that the presence of pretransplant ascites held statistical significance both in the univariate and multivariate models. However, the study of sodium as a continuous variable lost significance in the final model, probably because of the reduction in the number of patients in the analysis. Still, the 90-day mortality rate in patients with hyponatremia was close to 50% in both analyses, indicating that the exclusion of 178 patients submitted to LDLT probably decreased the statistical power of serum sodium values.

The presence of ascites and serum sodium concentration are important variables associated with decreased patient survival while candidates wait for a liver graft. Even though ascites is a subjective parameter during patient evaluation for liver transplantation, its significance cannot be overlooked and previous scores have addressed this issue. Lower serum sodium values usually present in patients with advanced liver disease with refractory ascites may identify patients in extreme urgency to receive a liver graft. Probably these patients were part of those moved to the intensive care unit and transplanted on the basis of petition letters as Status 1, who were not contemplated by the PELD score. However, further prospective multicenter studies are necessary to validate these findings in order to improve current allocation policies based on the PELD score.

Ancillary