Presenting Features and Prognosis of Ischemic and Nonischemic Neonatal Liver Failure

Abstract

Objectives: 

To describe the epidemiological features, clinical characteristics and outcomes of neonates diagnosed with liver failure, as well as determine prognostic factors.

Methods: 

Cohort study conducted at a single tertiary referral and university-affiliated pediatric center. Hospital records of all neonates diagnosed with liver failure between January 2003 and December 2015 were retrospectively reviewed, and data on clinical and laboratory findings, treatment, and outcomes were collected. Survival analysis (Kaplan-Meier) and Cox regression were performed to identify prognostic factors at diagnosis. Liver failure diagnosis was established using the pediatric acute liver failure study group's diagnostic criteria for every patient with coagulopathy and biochemical pattern of liver disease.

Results: 

Forty-five patients were included. In our series, most cases were secondary to ischemia (28.9%). Other causes were neonatal hemochromatosis (17.8%), viral infections (13.3%), and inborn errors of metabolism (13.3%). A total 55.6% (25/45) of the patients died (median age: 16 days; range 1–235 days). Alanine aminotransferase (ALT) at diagnosis was associated with higher mortality or the need for liver transplantation on day 21 after diagnosis (P = .006). For every 500 IU/L increase in ALT serum levels, the mortality/liver transplantation rate increased 1.3 times (hazard ratio 95% confidence interval: 1.1–1.6). Although ischemic neonatal acute liver failure presents with higher ALT levels, these cases appear to have better outcomes. Higher international normalized ratio tended to increase mortality/transplantation (hazard ratio 1.02; 95% confidence interval 0.91–1.2).

Conclusions: 

Neonatal liver failure should perhaps be considered in the differential diagnoses of any coagulopathy. ALT and international normalized ratio levels at diagnosis could predict prognosis in the short term. Ischemic liver failure appears to have a better prognosis.

What Is Known

• Neonatal liver failure is a rare disease with a wide range of clinical presentations.
• Neonatal hemochromatosis is one of the most important etiologies of neonatal liver failure. Some inborn errors of metabolism and viral infections have also been considered frequent causes.
• The prognosis is poorer than that in older children and adults.

What Is New

• Higher alanine aminotransferase at diagnosis predicts a poorer prognosis both in ischemic and nonischemic patients with neonatal liver failure.
• Perinatal hypoxia-ischemia can be a relatively common etiology of neonatal liver failure.
• Neonatal ischemic liver failure appears to have a better prognosis than nonischemic cases.

Acute liver failure has been defined as an altered mental status (hepatic encephalopathy) and coagulopathy in the setting of an acute hepatic disease ⁽¹⁾. Hepatic encephalopathy diagnosis in children is difficult and does not develop until the final stages of hepatic disease ⁽²⁾. Because of this limitation, recent reports have applied a new definition in which encephalopathy is not always required ⁽³⁾. Neonatal liver failure (NLF) is considered a rare condition with a poor prognosis, although it is often difficult to diagnose because of its wide range of clinical presentations and evolution ⁽³⁾. Data regarding the etiology and prognosis of NLF in neonatal patients are scarce in the literature, and ischemic/hypoxic etiology has not been well delineated. Neonatal cases have typically been included in pediatric series, which can limit our knowledge about liver failure in the neonatal age group. The aims of our study were to describe the epidemiological features, clinical characteristics, and outcomes of neonates with liver failure; and to determine prognostic factors.

METHODS

Study Subjects

This is a single-institution cohort study conducted in a third-level reference hospital that is 1 of the 5 pediatric liver transplantation centers in our country. The study period was from January 1, 2003 to December 31, 2015.

We applied the Pediatric Acute Liver Failure Study Group (PALF SG) diagnostic criteria to diagnose NLF for any patient (term infants from 0 to 28 days of life and preterm infants who had reached term, but not more than 28 days corrected age at the time of diagnosis) with coagulopathy and biochemical pattern of liver disease. Patients were retrospectively identified by chart review using a database system.

Diagnostic Categories

We used ICD-10 codes or their respective keywords in our department's databases at discharge: liver failure (570), neonatal hepatitis (774.4), hemochromatosis (275), and coagulopathy (776.3). If a patient had been labeled with one of these, fulfillment of the PALF SG definition was checked. The inclusion of code 776.3 (coagulopathy) enabled us to detect any case otherwise misdiagnosed.

The clinical charts and laboratory findings were reviewed. Data collection included demographic, clinical, and analytical data, outcomes, and treatments.

Viral infections were diagnosed applying serological and molecular methods (polymerase chain reaction) in blood, cerebrospinal fluid, fecal material, urine studies, or necropsy. Bacterial infection was determined by isolation in blood culture samples. A case was considered to be secondary to a bacterial infection only if liver damage (hypertransaminasemia and/or hyperbilirubinemia) and liver-based coagulopathy (extrinsic pathway alteration refractory to vitamin K administration at least twice) were clearly established by the time a blood culture turned positive, to avoid misdiagnosis, for example, disseminated intravascular coagulation (DIC). Inborn errors of metabolism were confirmed by measuring relevant metabolites in urine or plasma, by erythrocytic measurements of enzyme activity (eg, galactose-1-phosphate uridyl transferase activity) or mutation detection, as appropriate. Neonatal hemochromatosis was confirmed when there was extrahepatic iron overload evidence in salivary gland biopsies, magnetic resonance or postmortem studies. Perinatal asphyxia was considered when there were consistent clinical and laboratory data at birth (eg, encephalopathy, low Apgar test scores, low pH of cord blood) after a hypoxic-ischemic event. Other cases were classified as ischemic when low cardiac output or hypotension was present. Given both conditions cause liver failure through ischemic damage, we will hereafter refer to these cases as ischemic hepatitis/liver failure. Hemophagocytic syndrome was diagnosed if the case fulfilled 5 of 8 diagnostic criteria according to hemophagocytic lymphohistiocytosis-2004 guidelines. When an infection or metabolic disease was suspected, but could not be confirmed, and when the case remained without a specific diagnosis after several tests, the etiology was considered indeterminate.

Liver failure resolution was considered when the international normalized ratio (INR) decreased to <1.4 and remained below that point in subsequent determinations, provided liver transplantation had not been previously performed. A case was considered to be primary or secondary depending on the original site of damage (liver or systemic). Indeterminate cases were considered to be primary cases. A case was classified as early liver failure when diagnosed within the first 7 days of life.

Laboratory Data

Cholestasis was considered if conjugated bilirubin was ≥2 mg/dL or represented >20% of total bilirubin. Renal failure was diagnosed if peak creatinine reached >1.5 mg/dL as long as the baby was >24 hours from birth. In our laboratory, the coagulation test is considered as “not clot formation” when prothrombin time is >120 seconds and INR is >10 (in this case, 10 is the assumed value for statistical analysis).

Statistical Analysis

The statistical analysis was performed using STATA 13.1 statistical software (StataCorp, TX). The descriptive data are presented as mean/median (SD or range) and n (%) as appropriate. The t test (with Welch approximation when required) and Fisher exact test were applied to make comparisons as indicated. The Kaplan-Meier survival analysis and Cox's proportional hazards regression were used to determine the association between significant clinical and laboratory factors and death or liver transplantation. We consider death or liver transplantation at the 21st day after diagnosis as a failure event. Survival curves were compared using log-rank and Wilcoxon tests (the log-rank test P value is presented when there is not discordance between tests). Data are presented as hazard ratio (HR, 95% confidence interval [CI]). A P value <0.05 was considered statistically significant.

The La Paz University Hospital Research Ethics Committee approved the study protocol.

RESULTS

Demographics

Forty-five patients (summarized in Table 1) fulfilled the NLF diagnostic criteria: 55.6% (25/45) were born at term (median gestational age at birth: 38 weeks; range 28–41), 27.9% (12/43) were born small for gestational age (median weight at birth 2705 g; range 600–3640 g), and 57.8% (26/45) were boys. Most patients, 73.3% (33/45) were transferred to our hospital or admitted from the emergency department. Median age at diagnosis was 4 days (range 0–61 days); 62.2% (28/45) were diagnosed within the first 7 days after birth and 37.8% (17/45) in the first day after birth. One hemophagocytic syndrome, which affected a preterm baby, showed coagulopathy at the 61st day after birth (37.6 weeks postmenstrual age).

TABLE 1: Patient's characteristics, follow-up, and treatments

Etiology

In our series, most cases were secondary to ischemia and neonatal hemochromatosis (Fig. 1). Ischemia was the subjacent etiology in 28.9% of the patients (13/45). Nine experienced perinatal asphyxia, whereas the other 4 patients had cardiogenic/hypovolemic shock. Neonatal hemochromatosis accounted for 17.8% of the cases (8/45) and was the second most frequent etiology. Viral infection represented 13.3% of the cases (6/45): 3 herpes simplex and 3 enterovirus infections. Inborn errors of metabolism were the etiology in 6 cases (13.3%). Galactosemia was the most common (3/6 cases). The other 3 patients experienced citrullinemia and congenital lactic acidosis/mitochondrial disease. Bacterial infections were responsible for 6.7% (3/45). The agents isolated in blood culture were Serratia marcescens, Staphylococcus epidermidis, and Klebsiella pneumoniae. Hemophagocytic syndrome was identified in 2 patients (2/45, 4.4%). Other etiologies in our series were liver hemangiomatosis, neuroblastoma (stage IVs), and intrahepatic portosystemic fistula (1 case each). Finally, the etiology could not be established in 8.9% of the patients who were classified as indeterminate (4/45). Globally, 66.7% (30/45) were classified as primary liver failure. Apart from the fact that ischemic cases tend to affect the youngest neonates, the patient's characteristics were not statistically significantly different when comparing ischemic with nonischemic cases (Table 2).

FIGURE 1: Etiologies and clinical status on day 21 after neonatal liver failure (NLF) diagnosis (above). Death or liver transplantation from diagnosis to 21st day after NLF diagnosis (below). CI = confidence interval.

TABLE 2: Patients features according to etiology: ischemic versus nonischemic

Follow-up

In 51.1% of cases (23/45), liver function was fully recovered (mean time 9 ± 6.9 days). However, 55.6% (25/45) of our patients eventually died (median age 16 days; range 1–235 days). Of the patients who died, 16% (4/25) showed liver function recovery before death. Bleeding was a common complication; most cases were mild but persistent hemorrhage at umbilical vessels or venipuncture sites, frequent skin hematomas, hematuria, and bloody tracheal aspirates, gastric residuals, or stools. Three patients, however, had brain hemorrhage and 4 had massive hemorrhages involving lung, urinary tract, and digestive bleeding. Hemorrhage risk appeared to be higher if the patient had thrombocytopenia (P = 0.04) rather than severe coagulopathy (neither peak INR nor INR at diagnosis was shown to be associated with bleeding risk). Massive hemorrhage was the cause of death in 24% (6/25). One of these patients died in the operating room during liver transplantation and another had a primary graft failure. In 32% (8/25) of the cases, limitation of the therapeutic effort was applied. In all but one, it was motivated by irreversible brain damage. The remaining case showed refractory renal failure. Other death causes were hemodynamic (5/25), respiratory (2/25), or multiorgan (4/25) failure.

At 21 days from diagnosis, 62.2% (28/45) of the patients were alive, 60% (27/45) with a native organ and 2.2% (1/45) with liver transplantation. Of all the patients, 37.8% (17/45) died (Fig. 1), all without liver transplantation.

Six patients (13.3%) were included on the waiting list for a liver transplant, all diagnosed with neonatal hemochromatosis. Ultimately, 4 (66.5%) underwent liver transplantation. In all the transplanted patients, the liver came from a deceased donor. The other 2 patients were withdrawn from the list due to their poor evolution and concurrent active infectious process.

Laboratory Data

Direct bilirubin was measured in 38 of 45 patients during admission, and 89.5% (34/38) showed cholestasic pattern of liver injury. Cholestasis appeared to be one of the most frequent short-term sequelae. Most cases (76.5%) were detected during liver failure (at least 26/34); however, in 86.7% of the patients (13 of 15 patients with cholestasis who survived), cholestasis persisted or appeared beyond liver failure recovery. Ammonia was measured in 33 of 45 patients and 20 (60.6%) had hyperammonemia (>200 μg/dL), mean peak value: 309 (range 76–2087 μg/dL).

Prognosis

The laboratory and epidemiological data present at the diagnosis and the clinical findings and complications during admission were analyzed in an attempt to identify prognostic factors for death/liver transplantation by the 21st day after acute liver failure diagnosis.

Regarding epidemiological characteristics: sex (P = 0.8), age at diagnosis (P = 0.8), primary versus secondary liver failure (P = 0.8), ischemic versus nonischemic etiology (P = 0.6), and gestational age (P = 0.9) were not found to be clinically useful in predicting death or need for liver transplantation on day 21 after diagnosis.

Laboratory data analyses evaluated the predictive value of alanine aminotransferase (ALT), INR, total bilirubin, total white blood cell count, platelet count, total serum protein, and plasmatic sodium at diagnosis. Only ALT at diagnosis was associated with higher mortality or need for liver transplantation at the 21st day (P = 0.006). For every 500 IU/L increase in ALT serum levels, the mortality/liver transplantation rate increased 1.3 times (95% CI: 1.1–1.6). Higher INR tended to increase mortality/liver transplantation need (HR 1.02; 95% CI 0.9–1.2). When excluding the ischemic/hypoxic etiology from the analysis, again, only higher ALT at diagnosis remains as a prognostic factor (for every 500 IU/L, HR: 1.51; 95% CI: 1.16–1.96; P = 0.002) among the clinical and biochemical parameters. Considering epidemiological factors, the primary etiologies showed better prognosis than the secondary cases (P = 0.013) in this subgroup analysis.

DISCUSSION

NLF is a rare and often fatal disease. The broad etiologic spectrum of this condition explains the great variability in its clinical presentation. Early diagnosis during the neonatal period is frequently a real challenge for the clinician. The diagnosis of acute liver failure is based on the presence of persistent coagulopathy and biochemical pattern of liver damage, according to PALF SG diagnostic criteria.

In our study, PALF SG diagnostic criteria were applied to a wide population of neonates with coagulopathy to optimize NLF diagnosis. Because strict diagnostic criteria were applied to every suspected case, classification bias can be reasonably ruled out.

Published reports focusing on NLF etiology are scarce. Most series include both the pediatric and neonatal population ^(4–7). The traditional liver failure definition ignores secondary causes such as septic or hypoxic-ischemic cases ⁽⁸⁾. This is a controversial point, given liver failure essentially boils down to synthetic function loss (using coagulation as a surrogate), and encephalopathy is neither a constant nor an early finding when present in pediatric liver failure. In agreement with previous authors ^(3,9–12) we included ischemia and sepsis if the patients strictly fulfilled established criteria. In fact, ischemia was the primary pathogenic mechanism found in our series. Although our finding may not be applicable elsewhere because we are a referral center for both hypothermia and cardiac surgery, it is consistent with previous reports that found ischemia to be a common etiology among infants younger than 120 days of age ⁽¹³⁾. Regarding sepsis, it is frequently difficult to differentiate septic acute liver failure from sepsis-related coagulopathy (eg, DIC) without direct liver injury. Indeed, DIC often coexists in patients with liver failure. In our opinion, a case of NLF can be considered as secondary to septicemia only if PALF SG criteria are fully realized at the time of the blood sample culture isolation of the etiologic agent.

Indeterminate etiology is the cause in up to 50% of patients, according to published pediatric series ^(4,5); in neonates, it appears to represent up to 30% of cases ⁽⁸⁾. Nevertheless, in our study, an indeterminate cause was only present in <10% of patients. We hypothesize that this result could be due to the experience of our group, the extensive diagnostic tests systematically performed in our patients, and the relatively high efficiency of these test in newborns, in whom easily detected infections, simple and relatively few inborn errors of metabolism, neonatal hemochromatosis, and some common recognizable clinical scenarios (such as perinatal asphyxia) explain most cases of NLF. Our hospital is a national referral center for pediatric liver diseases and pediatric liver transplantation ⁽¹⁴⁾, and more than 70% of the patients in our series were transferred from other facilities.

Notably, the patients diagnosed with neonatal hemochromatosis in our series showed a high mortality rate (6/8) (75%) beyond the 21st day from diagnosis. The mortality rate remained high (75%) even in those patients who underwent liver transplantation (50%). Only 1 patient with neonatal hemochromatosis survived without transplantation. Concerning neonatal hemochromatosis, our study found a poorer prognosis than recently reported ⁽¹⁵⁾. This outcome could be explained by the fact that, in our series, only 2 patients diagnosed with neonatal hemochromatosis were treated with the recently recommended therapeutic schedule that comprises IV gamma-globulin administration and double-volume exchange transfusion.

We tested only for previous reported clinical and biochemical prognostic factors ^(7,11,16) related to short-term prognosis (death or transplantation at 21 days from diagnosis). The low survival rate (40%) detected in our study is in accordance with previously reported studies ^(10,17). In our study, ALT serum levels at diagnosis were the only biochemical marker significantly associated with a poor short-term outcome. Surprisingly, the higher the ALT levels at diagnosis the poorer the patient prognosis (eg, higher risk of death or need for liver transplantation). This finding is contrary to what Lee et al ⁽¹¹⁾ reported, analyzing the outcome and prognostic factors of acute liver failure (ALF) in childhood, in which an indirect relation between ALT levels and mortality was found. This discrepancy could be explained by the fact that acetaminophen intoxication was a frequent cause of ALF in that study, which usually presents with high ALT serum levels but also has a better prognosis than other causes of liver failure. According to our data, ischemic liver failure presents high ALT levels and survival appears to be better too. Higher ALT at diagnosis, however, appears to be a prognostic marker when separately analyzing nonischemic causes. As previously reported in older patients, INR could be another prognostic factor in NLF. In our study, higher INR at diagnosis tends to be related to a poorer prognosis. The relatively small size of our study cohort may be responsible for the lack of statistical significance found. Other previously reported prognostic factors such as ammonia or specific clot factors (eg, V) and serum levels ^(9,18,19) could not be evaluated due to the absence of data at diagnosis in our series.

Squires et al, analyzing 348 pediatric patients with ALF, described that etiology of liver failure determines mortality and recovery likelihood ⁽⁵⁾. Indeed, it is likely the survival rate of ischemic liver failure is better, compared with nonischemic etiologies. Our data support this hypothesis. Unfortunately, we lack the statistical power to prove it. Patients with ischemic NLF showed a relatively high mortality rate (38.5%) in our study, although it varied depending on the type of ischemic event. On one hand, most (88.9%) of the patients with NLF secondary to perinatal asphyxia survived; on the other hand, all the patients with NLF secondary to cardiogenic/hypovolemic shock died.

In our study, long-term outcome was not investigated. Nevertheless, although patients who recover liver function appear to have good long-term outcome subjectively, some patients can exhibit later secondary cholestasis, that should be study in the long-term.

In conclusion, etiology and clinical presentation of NLF are heterogeneous. The application of strict diagnostic criteria could help correctly diagnose NLF, including ischemic cases, which appear to be a relatively common etiology. ALT and possibly INR serum levels at diagnosis could predict prognosis in the short term, both in ischemic and nonischemic patients. Because of the low incidence of this condition, an NLF multicenter prospective registry could be of paramount interest.