Bilirubin and Serial Auditory Brainstem Responses in Premature Infants

DISCUSSION

Serum TB remains the primary biochemical measure used to evaluate and treat premature newborns with hyperbilirubinemia, although there is substantial evidence that serum TB levels correlate poorly with bilirubin-induced neurotoxicity in the premature infant.^7,24–29 Moreover, institutional variations in the levels of bilirubin at which phototherapy and exchange transfusions are initiated in jaundiced premature newborns indicate that the current management of hyperbilirubinemia in these infants is not evidence based.³⁰ Our study demonstrates that UB is a more sensitive predictor of bilirubin-induced auditory toxicity as evaluated by ABR changes than either B:A MR or serum TB levels in newborns of 28 to 32 weeks' GA.

The results of this study and several others provide UB concentrations in premature newborns associated with either overt kernicterus or transient bilirubin encephalopathy as indicated by ABR changes. A progression of bilirubin neurotoxicity seems to correlate with UB levels. In term newborns, UB levels must increase to 1 to 2 μg/dL before bilirubin-induced ABR changes are likely.¹² In the premature infant, ABR changes appear as the UB concentration exceeds 0.5 μg/dL, and overt kernicterus becomes likely above a UB of 1 μg/dL^8,20,21 (Table 6).

Bilirubin binding is a complex function of the concentrations of TB, UB, and albumin. Although the B:A MR is a measure of bilirubin binding and was a significant predictor of ABR abnormalities in the subgroup of premature neonates who had UB measurements, the ratio did not significantly predict abnormal ABR maturation in the entire population (P = .19). The trend, however, was better than that for the TB (P = .98). Although this supports the measurement of UB as the best predictor of bilirubin toxicity, the data do illustrate the potential value of at least considering the B:A MR along with the TB when UB measurements are not available.

Several in vitro and in vivo studies demonstrated that bilirubin-induced neurotoxicity involves changes in energy metabolism, alteration in membrane function, decreased membrane potential, alteration in enzyme function, and inhibition of protein and DNA synthesis.^31,32 It also seems from in vitro studies that immature cells are more sensitive to bilirubin toxicity than differentiated cells, supporting the clinical experience that premature neonates are more susceptible to bilirubin-induced neurotoxicity.⁶

One of the tools commonly used to investigate bilirubin-induced neurotoxicity in neonates is the ABR. ABR has been studied in term and to a lesser extent in premature newborns as a direct, noninvasive neurophysiologic assessment of reversible bilirubin neurotoxicity.^10,13,14,33 These studies have demonstrated reversible changes in ABR wave III and wave V latencies induced by bilirubin. However, until now, there has been a paucity of information on sequential maturational changes in ABR during the first postnatal week in newborns ≤32 weeks' gestation for the ABR to be useful to identify bilirubin-induced encephalopathy in these infants. The ABR at this gestation was shown recently to have rapid sequential maturational changes during the first postnatal week of life.²²Although at these ages waves III and V are not always detectable and therefore sequential measurements of latencies are not always possible, ABR waveforms can be categorized into response types, which also show progressive maturational changes during the first postnatal week.²² These patterns can now be used to assess auditory toxicity. Our data demonstrate a strong temporal correlation of peak UB with abnormal maturation of the ABR.

Various clinical factors, such as hypothermia, hypoxia, acidosis, hypercarbia, asphyxia, sepsis, severe intraventricular hemorrhage, and hemolysis have been postulated to explain the occurrence of bilirubin neurotoxicity at much lower levels of serum TB. These factors are thought to increase the risk of kernicterus by affecting serum bilirubin-albumin binding, bilirubin entry into the brain, or tissue uptake of bilirubin. Ritter et al²¹ in a prospective study identified a degree of hypoxia and acidosis that occurred before maximum UB level and that was present in newborns with kernicterus and not present in those without kernicterus. However, in 2 separate and independent retrospective analyses, Kim et al²⁴ and Turkel et al²⁵ failed to identify a single risk factor or combination of risk factors to be useful in predicting development of kernicterus in premature infants. As in earlier studies, we also found no statistically significant differences in clinical risk factors between the normal and bilirubin encephalopathy groups.

Various biochemical factors are involved in the pathogenesis of bilirubin encephalopathy. In healthy term newborns, it may be possible to estimate the UB from the B:A MR, as binding properties are fairly constant.³⁴ Amit and Brenner⁵ reported dependency of bilirubin toxicity on the B:A MR in vitro using fetal rat glial cells. Similarly, the B:A MR has been shown to be more predictive of bilirubin encephalopathy than the serum TB as evaluated by ABR in term infants with hyperbilirubinemia.¹² However, in the premature newborn, the MR's usefulness is not known. Kim et al²⁴ found no difference in MR between kernicteric and nonkernicteric premature newborns in a retrospective analysis. In the premature infant, bilirubin binding to albumin may be affected by the presence of competitors for bilirubin binding, releasing UB at even lower B:A MRs. The studies of Kim, Ritter, and Turkel all were performed in an era in which kernicterus was very prevalent in some nurseries but not in others. This may have been associated with the common use of benzyl alcohol as a preservative in drugs and injectable saline solutions at this time. The metabolite of benzyl alcohol, sodium benzoate, is a potent displacer of bilirubin from albumin creating elevated UB while having a slight depressing effect on serum TB. A recent study also demonstrated considerable confusion about using the B:A MR by trying to correlate the ratio and the bilirubin with “bilirubin binding ability.”^35,36

Previous studies suggested that the neurologic outcome of hyperbilirubinemia correlates better with UB than serum TB levels.^11,20 Our study also found that UB is a more sensitive predictor than either serum TB or B:A MR of abnormal ABR maturation and hence bilirubin toxicity in premature newborns with hyperbilirubinemia. This suggests that these neonates with bilirubin-induced transient auditory toxicity have altered bilirubin-binding capacity. Our findings suggest a significant relationship between unbound levels and incidence of abnormal ABR maturation with significant increase in the incidence of bilirubin-induced auditory toxicity with UB levels >0.5 μg/dL. The slight overlap in UB levels between neonates with and without bilirubin encephalopathy suggests that other factors such as duration of hyperbilirubinemia, blood-brain barrier status, and neuronal susceptibility probably play an important role in some patients. All study participants passed a screening hearing test before being discharged from the nursery; however, the possibility of auditory neuropathy cannot be excluded. This transient bilirubin-induced auditory toxicity may have short-term implications. An association between abnormal ABR maturation and an increased incidence of apnea/bradycardia in these patients has been observed.³⁷Long-term implications require additional investigation.