Predicting Outcomes for Children Awaiting Liver Transplantation: Is Serum Sodium the Answer?

Prioritization for the allocation of deceased donor organs to children awaiting liver transplantation in the USA and many other countries is currently based on the Pediatric End Stage Liver Disease (PELD) score for children less than 12 years of age and the Model for Endstage Liver Disease (MELD) score for older children. The PELD score was derived from a large cohort of children in the Studies of Pediatric Liver Transplantation (SPLIT) database, and was subsequently validated as an accurate measure of 3-month mortality among children waiting for liver transplantation.1 The PELD score includes data elements for age at listing <1 year, international normalized ratio (INR), albumin, bilirubin, and growth failure, in contrast to the MELD score, which is calculated from bilirubin, INR, and creatinine. One strength of these scores was the move away from prediction using variables whose evaluation was in part subjective, including encephalopathy and ascites. Implementation of the PELD score system in the USA in 2002 was followed by decreased deaths among children on the waiting list, and no impairment of outcomes following transplantation.2 A similar benefit was noted for adults waiting for liver transplantation following the introduction of the MELD score for prioritization.3 Inspired by this improvement in waiting list outcomes, investigators working in adult hepatology explored ways to further enhance the predictive accuracy of the MELD score. Hyponatremia was identified as an independent predictor of mortality in cirrhotic adults and inclusion of a hyponatremia element in the MELD score (MELD-Na) was found to improve its predictive accuracy in studies of different patient cohorts in many countries.4-6 The impact of sodium on the calculated score was particularly notable when the MELD score was low but the patient had ascites.4, 7 Predictive modeling and one prospective trial suggested that adoption of a prioritization system based on the MELD-Na score would further reduce deaths on the waiting list.8, 9 However, despite these findings the MELD-Na score has not been adopted by the transplant community in the USA due to several concerns, including the limited number of patients who would benefit from the change to MELD-Na (about 12% of the patients on the waiting list), the reduced predictive role of MELD-Na in the high MELD range, and the potential for manipulation of the serum sodium level by intermittent fluid shifts or diuretic use.4, 7, 9, 10 For children awaiting liver transplantation, data analyzing the effect of hyponatremia on mortality are scarce. In this issue of Hepatology, Pugliese et al.11 provide an important perspective from a large pediatric liver transplant program in Sao Paolo, Brazil. They used data from 522 children under 12 years of age with chronic liver disease listed for liver transplantation over an 11-year period, excluding those with acute liver failure, liver neoplasms, and noncirrhotic metabolic liver diseases. Data about potential predictors were retrieved from medical charts and databases, including ultrasound scan within the month prior to listing during which the presence of ascites was documented. Biliary atresia accounted for 72% of patients, two-thirds were less than 1 year of age when listed, 40% had ascites, and 6% had serum sodium <130 mEq/L. After 90 days on the waiting list, 11% of children had died, 40% had been transplanted, and 49% were still waiting for a donor organ. Univariate analysis showed that death within 90 days was predicted by age under 1 year, ascites, spontaneous bacterial peritonitis (SBP), total bilirubin, INR, albumin, serum sodium levels, and PELD score. Poor growth or weight gain was not predictive. Multivariate analysis excluding the PELD score confirmed the predictive ability of only some of the remaining elements of PELD (age <1 year, INR, and bilirubin), but not albumin or growth. The predictive significance of ascites and serum sodium were maintained in this multivariate model, with a 4% increase in mortality for each unit decrease in serum sodium. A similar trend was also demonstrated in adults with a 5% increase in mortality for each unit decrease in serum sodium.4 The original decision to exclude ascites when generating the PELD and MELD scores was based on the desire to use only those variables that can be evaluated objectively. By relying on ultrasound scan evidence of ascites, and including those patients with “minimal” or “trace”-free fluid in the ascites group, Pugliese et al. endeavored to reduce the subjectivity inherent in the clinical diagnosis of ascites. They claim that their results challenge PELD-based prioritization systems that overlook ascites in spite of it being an important predictor of mortality. From the analyses presented, however, it is not clear whether the presence of “trace-free fluid,” a common finding in stable children with chronic liver disease, carries the same prognostic significance as larger volume ascites. The added value of serum sodium in predicting mortality on the pediatric waiting list is less clear from these new data. Although hyponatremia is significant in multivariate analysis when PELD score is excluded, serum sodium dropped to a level of only borderline statistical significance when PELD, sodium, and ascites were included in a final multivariate Cox regression analysis. The c-statistic for serum sodium in the prediction of 90-day waiting list mortality was only 0.66. Only one previous pediatric publication has explored serum sodium in the prediction of waiting list mortality.12 In a single-center study of 94 children with cirrhosis listed for liver transplant in Cincinnati, nine children died before transplantation (most commonly due to sepsis or multiorgan failure). Twenty-four children either had hyponatremia <130 mEq/L at the time of listing or developed it while listed. Survival on the waiting list among these children was significantly worse than among the other 70 children without hyponatremia in this cohort of patients. Actuarial survival at 100 days after listing was ∼80% in the hyponatremic children compared to 98% in children with normal serum sodium. In multivariate logistic regression analysis, death before transplantation was predicted by hyponatremia (odds ratio [OR] 8.0, 95% confidence interval [CI] 1.4-45.7) and race (OR 6.3, 95% CI 1.25-33.3). Further studies will be required in larger and diverse pediatric cohorts to determine if there is a role for serum sodium in the improvement of prediction of waiting list mortality. The reproducibility of sodium measurements in different laboratories is an additional important concern when considering a “PELD-Na” score. One study from Spain has shown important differences in calculated MELD-Na scores when values from different laboratories are used.13 Systems to identify and minimize this variation will be needed prior to any change in prioritization practice. The additional predictive value of serum sodium may relate to the information it provides about the broader pathophysiology and severity of cirrhosis. Serum sodium is influenced by, among other things, the abnormal fluid balance, renal function, renin-angiotensin-aldosterone activity, and endothelial dysfunction found in cirrhosis. The predictive value of serum sodium in pediatrics may reflect its influence by factors other than poor renal function, because other markers of renal function (e.g., creatinine) are not predictive of mortality in children. One might speculate that inclusion of variables more directly linked to the primary causes of mortality may improve the predictive ability of the PELD score. Infections were the most common cause of death in the Pugliese et al. and Carey et al. studies, and yet the PELD score does not include variables that specifically target the measurement of this risk. Interestingly, hyponatremia has now been associated with an increased risk of SBP in two pediatric studies. Future studies should explore the predictive ability of other commonly available variables that measure the risk of infection in cirrhosis. In order for any modified version of the PELD score to improve prioritization and waiting list mortality, attention may also be needed to other steps in the prioritization process. A large proportion of pediatric liver transplants occur when donor organs are assigned to a child whose prioritization does not reflect the calculated PELD score, either due to prioritization as status 1 or by the granting of extra points as an “exception” to the PELD score, reflecting the specific clinical circumstances of a patient.14 The exception points system reintroduces significant subjectivity into the otherwise objective PELD score. Whether such a subjective system is optimal is the subject of ongoing debate.15 The need for a careful reevaluation of the exception points system has been suggested by the observation that adults granted exception points within the MELD system have a lower mortality on the waiting list compared to other cirrhosis patients without exception points.16, 17 Ideally, a prioritization system would be based purely on a calculated, objective score. However, it is challenging to imagine that a simple score using routine clinical tests will adequately fulfill the required predictive accuracy in all clinical circumstances. For now, as in most areas of medicine, objective tests are combined, when necessary, with expert opinions based on the published literature and individual clinical details to provide the estimate of appropriate prioritization. In conclusion, ascites and hyponatremia appear to be markers of adverse 90-day outcome in children with cirrhosis listed for liver transplantation. The use of these markers in the clinical arena can assist the clinician in counseling parents, prognostication, and daily management of the cirrhotic child. However, the use of this new information for organ allocation decisions must await further confirmatory studies, clearer delineation of the definition and prognostic significance of small- and trace-volume ascites, and better understanding of the impact of a modified PELD-Na score on the outcome of children waiting for liver transplantation. Simon C. Ling, M.B.Ch.B., and Yaron Avitzur, M.D. Division of Gastroenterology, Hepatology and Nutrition The Hospital For Sick Children Department of Pediatrics University of Toronto Toronto, Ontario, Canada

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.