*Division of Gastroenterology, Hepatology and Nutrition
†Division of Bone Marrow Transplantation and Immunodeficiency, Cincinnati Children's Hospital, Cincinnati, OH
‡Department of Pediatrics, Section of Pediatric Gastroenterology, Hepatology and Nutrition, University of Colorado School of Medicine and Children's Hospital Colorado, Aurora, CO
§Division of Gastroenterology, Hepatology and Nutrition, Hospital for Sick Children, Toronto, ON, Canada
||Department of Epidemiology
¶Division of Gastroenterology, Hepatology and Nutrition, Children's Hospital of Pittsburgh of UPMC, Pittsburgh, PA.
Address correspondence and reprint requests to John C. Bucuvalas, MD, Professor of Pediatrics, Cincinnati Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229 (e-mail: email@example.com).
Received 6 August, 2012
Accepted 22 October, 2012
Supported by the Division of Digestive Diseases and Nutrition within the National Institute of Diabetes and Digestive and Kidney (NIDDK) Diseases of the National Institute of Health (NIH) (5U01 DK072146) and NIH/NCRR Colorado CTSI Grant Number UL1 RR025780. Contents are the authors’ sole responsibility and do not necessarily represent official NIH views.
The authors report no conflicts of interest.
Pediatric acute liver failure (PALF) is a devastating life-threatening condition, particularly for those children with acute liver failure (ALF) of indeterminate etiology who comprise almost 50% of the disease population (1). Medical decision making for patients with ALF is complex and requires repeated assessment over time. The physician must regularly estimate the probability of survival with native liver compared with the risks of liver transplantation (LT). Existing liver failure scoring systems (2–7) fall short of the ideal prognostic tool (8–10) because outcomes vary among children with similar disease severity scores. Therefore, a better tool is needed to stratify patients by prognosis and target potentially life-saving treatment.
Emerging evidence suggests a subgroup of patients with ALF have clinical findings and plasma cytokine profiles consistent with natural killer (NK) cell dysfunction (11) or systemic inflammatory response syndrome (12). One such example is hemophagocytic lymphohistiocytosis, a disorder of immune regulation often occurring in infancy and early childhood. Here, we sought to determine whether markers of T-cell immune activation were associated with patient outcome.
The PALF Study Group was funded by the National Institute of Diabetes and Digestive and Kidney (NIDDK) Diseases of the National Institutes of Health (NIH). At the time these data were collected, the PALF Study Group consisted of 20 pediatric liver transplant sites, 17 within the United States, 1 in Canada, and 2 in the United Kingdom. Only the 17 of the 20 sites in the United States contributed to the present study.
Data Collection for the PALF Longitudinal Cohort
This was an ancillary study to the PALF study (1). To be enrolled in the PALF study cohort, informed consent was obtained from a parent or legal guardian. Subsequently, demographic, clinical, and laboratory information were recorded daily for up to 7 days, starting on the date of enrollment into the study. Diagnostic evaluation and medical management were consistent with the standard of care at each site. The NIH provided a Certificate of Confidentiality to the study and institutional review board approval was secured at each site before patient enrollment.
Blood samples to assess markers of T-cell immune activation were collected within 48 hours of enrollment into the PALF cohort. Participants were excluded if they were reported to have received corticosteroids or rituximab before sample collection or were <10 kg to avoid excessive blood withdraw in smaller children. We did not collect data regarding administration of blood products. Consequently, we did not exclude subjects based on the timing of administration of blood products.
Blood was collected in ethylenediaminetetraacetic acid (EDTA) and sodium (Na) heparin tubes. For children weighing 10 to 20 kg, 1 mL of blood in an EDTA tube and 4 mL of blood in a Na-heparin tube were collected, and for children >20 kg, 3 mL of blood in an EDTA tube and 7 mL of blood in Na-heparin tube were collected. Collection days were Monday through Thursday with samples shipped at room temperature by priority overnight for delivery on weekdays only with Friday deliveries received by 11:00 AM Eastern time. Samples were shipped to the Diagnostic Immunology Laboratory at Cincinnati Children's Hospital. Blood samples were processed within 2 hours of receiving the specimen. If blood collected in the EDTA tube was insufficient, then measurement of soluble interleukin 2 receptor alpha (sIL2Rα) was considered a priority. Specimen processing occurred within 24 hours of collection from the patient according to standard procedure, which was why PALF study sites outside the United States were not included in this ancillary study. The assays were done under the supervision of one of the authors (L.F.) in a laboratory accredited by the College of American Pathologists and by Clinical Laboratories Improvement Amendments. Results were compared to established age-specific reference ranges. Markers of immune activation were categorized to permit comparisons among subjects in different age groups.
Analysis of T-Cell Activation
The T-cell immune activation profile assembled in the present study included NK cell function, flow cytometry, analysis of cytotoxic lymphocytes, and measurement of sIL2Rα levels. Markers of immune activation included sIL2Rα, perforin-expressing NK T cells, and CD56 bright immunoregulatory NK cells. NK cell cytolytic function was used as the primary marker of NK cell function (13–16). Cytolytic function of NK cells was measured by release of radioactive chromium from labeled target cells. In addition, granzyme B and perforin expression in lymphocytes was determined by 4-color flow cytometry after staining with both surface and intracellular monoclonal antibodies (15). Levels of sIL2Rα were measured using the Immunlite 1000 (Siemens, Malvern, PA) method, a solid-phase, 2-site chemoluminescent immunometric assay. Normal ranges for sIL2Rα levels analyzed in pediatric populations are well defined (17). In contrast to adults, sIL2Rα levels are physiologically high in normal children. Levels are highest in children younger than 4 years and taper to near adult levels in adolescents. For this analysis, age-normalized levels for sIL2Rα are reported unit-free (eg, 2 times the upper limit of normal) and measured sIL2Rα levels that are not normalized for age are reported with units (eg, 5000 IU/mL). sIL2Rα levels >2 times the upper limit of normal for age were considered to have potential clinical importance. These cutoff levels were based on confidence intervals from normal populations (17). A sIL2Rα level ≥5000 IU/mL is considered to be markedly elevated, regardless of age.
Final diagnoses were determined by each site's principal investigator. Diagnostic categories for PALF based on previously defined criteria (1) included acetaminophen-associated ALF; autoimmune ALF; drug-induced ALF; ALF due to viral hepatitis; metabolic disease including Wilson disease, mitochondrial disease, or disorders of fat metabolism diagnosis; other; or indeterminate ALF.
Within the indeterminate group are those who were fully evaluated and did not meet specific diagnostic criteria or those were incompletely evaluated, often because of death or LT before a diagnosis being established. Laboratory studies and clinical assessment following site-specific standard of care included white blood cell count, hemoglobin, platelet count, serum aminotransferases, international normalized ratio, prothrombin time, bilirubin, ammonia, and coma stage. These parameters provided assessment of bone marrow function, liver cell injury, and disease severity scores. Coma stage was determined by established criteria (1). Elements of the immune activation profile were assessed to determine their association with death without transplantation, LT, and survival with native liver at 21 days.
For demographic, clinical characteristic and severity scores, percentages were reported if categorical and medians (25th and 75th percentiles) were reported if continuous. NK measures were age-normalized by dividing assay results by the upper limit of the age-specific normal range. Kruskal-Wallis tests were used for comparing each of 15 age-specific normalized NK function measurements among those who died without LT, received LT, or survived with native livers. To adjust the error rate in multiple testing (15 normalized NK measures) and multigroup comparisons (pairwise comparisons when the Kruskal-Wallis met criterion for statistical significance), the Bonferroni-Holm step-down procedure was applied. An exact χ2 test was used to test equality of the probabilities of the 3 outcomes for the 4 sIL2Rα groups.
P values <0.05 (adjusted for multiple comparisons as above) were used to determine statistical significance. Analyses were conducted using SAS statistical software (SAS Institute, Cary, NC).
Samples for T-cell immune activation analysis were obtained from 86 participants. Nine were excluded because samples were not received and processed by the testing laboratory within 24 hours after they were drawn, so the final study group included 77 subjects. Outcomes at day 21 were known for 75; 2 participants discharged alive at days 9 and 10 following enrollment did not have further follow-up and were assumed to be alive at day 21.
The distribution of diagnoses and outcomes are shown in Table 1. Subjects with a defined diagnosis accounted for 65% (n = 50) of the cohort. Three subjects (6%) with a defined diagnosis died and 10 (20%) underwent LT. Of the 27 subjects with indeterminate ALF, 2 (7%) died and 11 (41%) underwent LT. Thirty one percent of the study population was younger than 5 years (Table 2). At presentation, 11% of the subjects had coma grade III or IV. Peak coma grade was grade III or IV in 16% of subjects.
T-Cell Immune Activation Markers
sIL2Rα levels were not determined for 2 of 77 subjects because of laboratory errors; 1 of these subjects died and the other survived with native liver. Of the 75 subjects with measured sIL2Rα levels, 50 (67%) survived with their native liver, 21 (28%) received a LT, and 4 (5%) died (Table 3). Note that, among all the NK function measurements, only sIL2Rα levels differed significantly across the 3 outcomes’ groups after accounting for the multiple comparisons.
Outcome differed significantly (P = 0.04) by level of sIL2Rα (Table 4). Among the 37 subjects with normal sIL2Rα levels, 30 (81%) were alive with native liver, 7 (19%) underwent LT, and none died within 3 weeks of entry into the study. At the other extreme, only 5 of the 15 subjects with markedly elevated sIL2Rα (≥5000 IU/mL) were alive with their native liver, whereas 8 (53%) underwent LT, and 2 died (13%) by 21 days after enrollment.
For the 15 subjects with sIL2Rα levels at least 5000 IU/mL, the final clinical diagnosis was indeterminate for 10 (67%) subjects, hemophagocytic lymphohistiocytosis for 3 subjects, drug-induced disease for 1 subject, and autoimmune disease for 1 subject. For the 60 patients with sIL2Rα levels <5000 IU/mL, 16 (27%) were indeterminate.
No markers of T-cell immune activation other than sIL2Rα differed significantly among the outcomes when adjusting for the multiple comparisons. The association between sIL2Rα and the number of CD8 cells was weak (Kendall τ = 0.16, P = 0.07).
sIL2Rα was associated with outcome in this group of children with PALF. Although none of the 37 children with normal sIL2Rα died, 9% of those with values more than normal but <5000 IU/mL and 13% of those with values at least 5000 IU/mL died. LT was also more common among the groups with higher sIL2Rα.
Individual or small pairings of cytokines have been tested as biomarkers that may inform disease severity or outcome. For example, elevated serum levels of α-fetoprotein, markers of liver regeneration, have been associated with a favorable outcome in patients with acetaminophen-induced liver injury (18), but the rate of rise early in the course may be more important than the absolute value (19). Soluble CD154 (sCD154), a marker of immune activation expressed on activated CD4+ T cells and platelets, was associated with death in adult patients with ALF (20). A panel of pro- and anti-inflammatory markers was assessed in adults with ALF before and after treatment with the molecular absorbent recirculating system (21). Although molecular absorbent recirculating system did not alter circulating cytokine levels, investigators found IL-6 and IL-8 levels were decreased in patients who survived with their native liver, IL-6 was increased in those who either died or received LT, and initial values for sIL-2Rα were higher in patients who died or received a liver transplant relative to those who survived with their native liver.
The rarity and patient heterogeneity of PALF lead to practical and logistical difficulties related to predicting outcomes in PALF. Establishing reliable predictive models are confounded by LT, an intervention that is influenced by local experience and organ availability. Inclusion of death and LT into a single outcome is problematic as the two outcomes are not similar. Those receiving LT constitute a mixed cohort of participants who would have died or survived had LT not interrupted the natural history of PALF. The ideal predictive model will include markers that will reliably predict meaningful outcomes that include survival with native liver, death with native liver, and futility to inform LT decisions.
Our examination of a variety of markers of T-cell immune activation in a small PALF cohort identified only sIL2Rα to be discriminatory between survival with native liver, death, and LT. ALF is associated with myriad processes that variably affect outcome including immune activation, immune inactivation, apoptosis, and liver regeneration. We have previously reported a biomarker profile of hepatic regeneration associated with survival with native liver (22). Given the complex physiologic components associated with outcome in PALF, a mechanistic modeling platform that incorporates these and other components should be pursued to develop a more reliable predictive model for outcomes in PALF (23).
The present study has several weaknesses. This exploratory work was done as a cross-sectional study with sIL2Rα measured at a single point in time, and consequently the findings cannot reflect dynamic changes in immune activation. Those participants who died or underwent transplantation following enrollment in PALF but before the blood sample taken for NK cell analysis were not available for analysis. From these data, we cannot determine whether sIL2Rα levels reflect a response to liver injury or reflect a role in the etiopathogenesis of PALF.
These preliminary data suggest patients with higher sIL2Rα levels were more likely to die or undergo LT within 21 days than those with lower levels. Identifying a subset of patients with evidence of immune activation and at risk for poor outcome may form the foundation for targeted clinical trials with immunomodulatory drugs. In addition, sIL2Rα should be considered as a potential component in future PALF disease severity scores when they are developed. Improved ability to predict hepatic recovery may spare children from unnecessary LT, whereas identifying those who are likely to die will allow more timely decisions to move to LT.
The authors give special thanks to Andre Hawkins, the research coordinator at Cincinnati Children's Hospital who coordinated the efforts across centers. The work was made possible by the collaborative effort of principal investigators and research coordinators at the University of Pittsburgh, Mt Sinai Medical Center, Cincinnati Children's Hospital, Children's Memorial Hospital (Chicago), King's College-London (England), University of Texas Southwestern, Texas Children's Hospital, Seattle Children's Hospital, Children's Hospital Colorado (Aurora), Children's Hospital Medical Center (Boston), St Louis Children's Hospital, Johns Hopkins University, Columbia University, University of California at San Francisco, Hospital for Sick Children (Canada), Columbia University, Riley Hospital for Children (Indianapolis).
1. Squires RH Jr, Shneider BL, Bucuvalas J, et al. Acute liver failure in children: the first 348 patients in the pediatric acute liver failure study group. J Pediatr 2006; 148:652–658.
2. Dhiman RK, Seth AK, Jain S, et al. Prognostic evaluation of early indicators in fulminant hepatic failure by multivariate analysis. Dig Dis Sci 1998; 43:1311–1316.
3. Liu E, MacKenzie T, Dobyns EL, et al. Characterization of acute liver failure and development of a continuous risk of death staging system in children. J Hepatol 2006; 44:134–141.
4. Lu BR, Gralla J, Liu E, et al. Evaluation of a scoring system for assessing prognosis in pediatric acute liver failure. Clin Gastroenterol Hepatol 2008; 6:1140–1145.
5. Pelaez-Luna M, Martinez-Salgado J, Olivera-Martinez MA. Utility of the MAYO End-Stage Liver Disease score, King's College Criteria, and a new in-hospital mortality score in the prognosis of in-hospital mortality in acute liver failure. Transplant Proc 2006; 38:927–929.
6. Rhee C, Narsinh K, Venick RS, et al. Predictors of clinical outcome in children undergoing orthotopic liver transplantation for acute and chronic liver disease. Liver Transpl 2006; 12:1347–1356.
7. Sindhi R, Soltys K, Bond G, et al. PELD allocation and acute liver/graft failure. Liver Transpl 2007; 13:776–777.
8. Katoonizadeh A, Decaestecker J, Wilmer A, et al. MELD score to predict outcome in adult patients with non-acetaminophen-induced acute liver failure. Liver Int 2007; 27:329–334.
9. Polson J, Lee WM. AASLD position paper: the management of acute liver failure. Hepatology 2005; 41:1179–1197.
10. Shakil AO, Kramer D, Mazariegos GV, et al. Acute liver failure: clinical features, outcome analysis, and applicability of prognostic criteria. Liver Transpl 2000; 6:163–169.
11. Yazigi N, Tial G, Filipovich A, et al. Natural killer dysfunction in pediatric acute liver failure. Am J Transplant 2008; 8 (supp s2):327A.
12. Rolando N, Wade J, Davalos M, et al. The systemic inflammatory response syndrome in acute liver failure. Hepatology 2000; 32 (4 Pt 1):734–739.
13. Filipovich AH. Life-threatening hemophagocytic syndromes: current outcomes with hematopoietic stem cell transplantation. Pediatr Transplant 2005; 9 (suppl 7):87–91.
14. Henter JI, Horne A, Arico M, et al. HLH-2004: Diagnostic and therapeutic guidelines for hemophagocytic lymphohistiocytosis. Pediatr Blood Cancer 2007; 48:124–131.
15. Kogawa K, Lee SM, Villanueva J, et al. Perforin expression in cytotoxic lymphocytes from patients with hemophagocytic lymphohistiocytosis and their family members. Blood 2002; 99:61–66.
16. Sullivan KE, Delaat CA, Douglas SD, et al. Defective natural killer cell function in patients with hemophagocytic lymphohistiocytosis and in first degree relatives. Pediatr Res 1998; 44:465–468.
17. Filipovich AH. Hemophagocytic lymphohistiocytosis and other hemophagocytic disorders. Immunol Allergy Clin North Am 2008; 28:293–313.
18. Schmidt LE, Dalhoff K. Alpha-fetoprotein is a predictor of outcome in acetaminophen-induced liver injury. Hepatology 2005; 41:26–31.
19. Schiodt FV, Ostapowicz G, Murray N, et al. Alpha-fetoprotein and prognosis in acute liver failure. Liver Transpl 2006; 12:1776–1781.
20. Zheng YB, Gao ZL, Zhong F, et al. Predictive value of serum-soluble CD154 in fulminant hepatic failure. J Int Med Res 2008; 36:728–733.
21. Ilonen I, Koivusalo AM, Repo H, et al. Cytokine profiles in acute liver failure treated with albumin dialysis. Artif Organs 2008; 32:52–60.
22. Rudnick DA, Dietzen DJ, Turmelle YP, et al. Serum alpha-NH-butyric acid may predict spontaneous survival in pediatric acute liver failure. Pediatr Transplant 2009; 13:223–230.
23. Mi Q, Li NY, Ziraldo C, et al. Translational systems biology of inflammation: potential applications to personalized medicine. Per Med 2010; 7:549–559.