 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
REVIEW ARTICLE |
a Division of Neonatology, Department of Pediatrics, University of Washington, Seattle, Washington
b LW Ligand, LLC, Vashon, Washington
c Division of Neonatal and Developmental Medicine, Department of Pediatrics, Stanford University School of Medicine, Stanford, California
d Section on Newborn Pediatrics, Pennsylvania Hospital, University of Pennsylvania Health System, Philadelphia, Pennsylvania
e Department of Neurology, Medical College of Virginia, Richmond, Virginia
 |
ABSTRACT |
|---|
 TOP ABSTRACT DETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
METHODS. A review of literature was undertaken to define basic principles of bilirubin transport and brain uptake leading to neurotoxicity. We then reviewed experimental and clinical evidence that relate TSB or Bf to risk for bilirubin toxicity and kernicterus.
RESULTS. There are insufficient published data to precisely define sensitivity and specificity of either TSB or Bf in determining risk for acute bilirubin neurotoxicity or chronic sequelae (kernicterus). However, available laboratory and clinical evidence indicate that Bf is better than TSB in discriminating risk for bilirubin toxicity in patients with severe hyperbilirubinemia. These findings are consistent with basic pharmacokinetic principles involved in bilirubin transport and tissue uptake.
CONCLUSIONS. Experimental and clinical data strongly suggest that measurement of Bf in newborns with hyperbilirubinemia will improve risk assessment for neurotoxicity, which emphasizes the need for additional clinical evaluation relating Bf and TSB to acute bilirubin toxicity and long-term outcome. We speculate that establishing risk thresholds for neurotoxicity by using newer methods for measuring Bf in minimally diluted serum samples will improve the sensitivity and specificity of serum indicators for treating hyperbilirubinemia, thus reducing unnecessary aggressive intervention and associated cost and morbidity.
Key Words: bilirubin–albumin binding • brainstem auditory evoked potentials • bilirubin • hyperbilirubinemia • kernicterus
Abbreviations: TSB—total serum bilirubin concentration • AAP—American Academy of Pediatrics • BBB—blood-brain barrier • Bf—free bilirubin concentration • AB—albumin-bound bilirubin concentration • A—serum albumin concentration • ABR—brainstem auditory evoked response • AN—auditory neuropathy • AD—auditory dyssynchrony • G6PD—glucose-6-phosphate dehydrogenase
THERE IS SUBSTANTIAL evidence that unconjugated bilirubin is neurotoxic and that high serum levels in newborns can produce brain damage (kernicterus).1,2 In the absence of obstructive jaundice, the serum concentration of unconjugated bilirubin is best estimated by measuring the total serum bilirubin concentration (TSB). Calculating the "indirect fraction" (TSB minus the "direct reacting" bilirubin concentration) can be misleading in most newborns, because high levels of unconjugated bilirubin can produce an elevated direct fraction (
10% of TSB), which does not represent nontoxic conjugated bilirubin.
Recently published management guidelines for jaundiced term and near-term infants by the American Academy of Pediatrics (AAP) are based on the premise that TSB is the best available predictor of risk for kernicterus.3 However, clinical evidence indicates that TSB, beyond a threshold value of 20 mg/dL (342 µmol/L), is a poor discriminator of individual risk for bilirubin toxicity. Kernicterus may rarely occur in healthy term infants with a TSB of <25 mg/dL (428 µmol/L),4–7 the level currently recommended by the AAP for aggressive intervention, and most infants with a TSB of >25 mg/dL escape without recognized permanent sequelae.7–10 How can this be? Can the very poor sensitivity and specificity of TSB in predicting bilirubin toxicity be attributed solely to variations in the blood-brain barrier (BBB) or neuronal susceptibility?
In addressing this problem, this review examines (1) determinants of TSB, (2) mechanisms involved in bilirubin movement into the brain, (3) predictability of bilirubin toxicity using TSB and/or the free-bilirubin (not bound to serum proteins) concentration (Bf), and (4) the challenge of developing a more rational management protocol for infants with severe jaundice. To avoid semantic confusion, the term "kernicterus" is restricted to patients with autopsy-proven kernicterus and patients with permanent bilirubin-induced brain injury. "Early bilirubin toxicity" refers to patients with transient or reversible clinical evidence of toxicity in the neonatal period.
|
DETERMINANTS OF TSB |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
The distribution of the bilirubin pool between the various compartments depends on the number of binding sites in each compartment and their affinity for binding bilirubin. The bilirubin-binding affinity is quantified by a binding constant, K, which equals the ratio of association and dissociation-rate constants (k1/k–1) of any specific receptor.
 |
H+ + OH–).11 Despite the ability of bilirubin to accumulate in rather high concentrations in both plasma and tissue, the Bf in the body (and plasma) is very low, rarely exceeding 3 to 4 µg/dL even in cases of marked hyperbilirubinemia. The low Bf is mandated in part by the very low solubility of bilirubin, which has been estimated to be 7 nmol/L12 on the basis of dissolution of bilirubin crystals in water or 70 nmol/L (4.1 µg/dL) on the basis of partitioning between water and chloroform.13 The latter value is more consistent with Bf reported in infant sera. Higher metastable concentrations can exist but would tend to aggregate and precipitate in tissue.
The relationship between Bf, albumin-bound bilirubin concentration (AB), and serum albumin concentrations (A) at clinically relevant TSB and A can be expressed as
 |
TSB, and the equation can be expressed as
 |
Unfortunately, the albumin-bilirubin binding constant, K, is not "constant" but varies considerably among newborns. It is also lower in sick infants and may increase with postnatal age. Furthermore, the effective concentration of albumin, A, can be reduced by the presence of drugs or compounds that bind to the same locus as bilirubin such as sulfonamides or benzoate, the metabolite of benzyl alcohol. Both drugs have been associated with clusters of kernicterus in premature infants.14,15
The equation shown above considers only 1 binding site on albumin and assumes that 1 molecule of albumin binds a single bilirubin molecule. However, albumin has 1 to 2 additional bilirubin-binding sites with lower K values, and apolipoprotein D also binds bilirubin in plasma.16 Thus, the measured Bf in plasma reflects the equilibrium of bilirubin with multiple binding loci, and TSB is distributed among the various plasma protein-binding sites according to their relative concentrations and binding constants. Interindividual variations in K, effective albumin concentration, and alternative binding sites in plasma limit the usefulness of estimating Bf by measuring the TSB/A ratio. Even in cases of kernicterus, TSB does not exceed the "binding capacity" of plasma, although the Bf (Bf/TSB) increases more rapidly as the TSB/A molar ratio approaches unity (9 mg of bilirubin per g of albumin).
Similar equations govern the distribution of bilirubin in tissue; at equilibrium, Bf will be the same in all compartments. However, in a system involving multiple compartments, some of which may involve nonreversible metabolic processes (eg, bilirubin oxidase, conjugation by glucuronyl transferase), a true equilibrium will never be achieved, but the movement of bilirubin will be toward compartments with lower Bf approaching "steady-state" levels of Bf. This movement of bilirubin between compartments has sometimes been referred to as "the free-bilirubin theory." It is not a theory but a law: the law of mass action.
|
ENTRY OF BILIRUBIN INTO BRAIN |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
When the BBB is intact, the rate of bilirubin uptake by brain is determined by (1) the Bf, (2) the permeability and surface area of the capillary endothelium, (3) the transit time through the capillary bed (time available for the extracted bilirubin to be replenished from the albumin reservoir), (4) the dissociation rate of bilirubin–albumin, k–1, (how quickly the Bf will be replaced by albumin-bound bilirubin as free bilirubin leaves the vascular space), and (5) the blood flow (number of capillaries recruited in a given region).21 The BBB is quite permeable to free bilirubin, with single-pass uptakes estimated as high 28% in rats.22 Bilirubin uptake may be increased by alterations in BBB permeability to bilirubin or albumin (eg, hyperosmolality, severe asphyxia), prolonged transit time (eg, increased venous pressure), an increase in blood flow (eg, hypercarbia), or an increase in the albumin-bilirubin dissociation rate (eg, altered binding in sick infants).
Because the transit time is short, 1 second, and the bilirubin–albumin dissociation rate is slow (eg, compared with the dissociation of H2O), there is little time to replenish the unbound bilirubin that is transported across the brain capillary. At a given Bf, brain uptake of bilirubin could theoretically be facilitated by a high TSB (ie, a high A) simply because there would be a larger plasma reservoir to replenish the extracted free bilirubin. Using published dissociation rate constants obtained in dilute albumin solutions,11 Robinson and Rapoport23 concluded that the slow dissociation rate precludes rapid replenishment of bilirubin removed and that brain uptake is governed largely by the TSB until the high-affinity bilirubin-binding sites on albumin are saturated. Unfortunately, their analysis did not consider that significant bilirubin is bound to secondary sites (probably with higher dissociation rates) long before this occurs. Furthermore, subsequent evidence indicates that the binding constant (ratio of association/dissociation rates) is an order of magnitude lower in undiluted serum and extremely variable between infants,24,25 which calls into question their assumption of a low k–1 being rate limiting for brain uptake of bilirubin.26
Clinically, bilirubin neurotoxicity, which occasionally is irreversible, will occur before primary sites are saturated. Still, because of the slow transport of bilirubin across the BBB, time of exposure to a high Bf may be critical in determining the magnitude of brain load and toxicity. In primates, several hours of exposure to a very high TSB and Bf are usually required to create neurotoxicity, and even longer exposure is needed to produce nuclear staining.27,28 Most cases of kernicterus in term infants occur at several days of age and often after the TSBs have been very high for extended periods of time. This "exposure-time factor" complicates outcome studies that examine only the peak (or a single readmission) TSB.
Net transport of bilirubin across the BBB also may be influenced by the energy-dependent multidrug-resistant transporters MDR1 and MRP1. MDR1, or P-glycoprotein, is one of several transporters that are involved in cellular efflux of xenobiotics and is expressed in capillary endothelial cells of the BBB, astrocytes, and the choroids plexus.29 Unconjugated bilirubin is a weak substrate for MDR1.30 In a study by Watchko et al,31 brain uptake of bilirubin in Mdr1a–/– knockout mice infused with high concentrations of bilirubin was twice that of controls (Mdr1a+/+). Inhibition of P-glycoprotein potentiates bilirubin-induced apoptosis in a human neuroblastoma line32 and increases bilirubin content in brains of young adult rats (measured at very high TSB).33
The multidrug-resistance–associated protein, MRP1, performs similar functions. MRP1 is highly expressed in choroid plexus epithelium, astrocytes, (rat) neurons, and placenta trophoblast but has minimal expression in capillary endothelium in whole brain.34 It is upregulated in cultured cells that are exposed to unconjugated bilirubin and mediates ATP-dependent cellular export of bilirubin.35 Inhibition of MRP1 in cultured astrocytes potentiates apoptosis induced by low concentrations of bilirubin.36 Other potential cellular defense mechanisms include mitochondrial bilirubin oxidase,37,38 other transporters, and antiapoptosis factors.39
The extent to which these cellular defense mechanisms contribute to variations in sensitivity to the toxic effects of bilirubin remains uncertain, but all of these protective mechanisms can be overwhelmed by the rapid infusion of a binding competitor. Under steady-state conditions, these endogenous metabolites and exogenous drugs effectively decrease the albumin concentration. However, peak drug levels after intravenous infusion can have a devastating effect, and like the splash from a boulder thrown into a pond, there is a transient surge in Bf, an immediate decrease in TSB, and an acute increase in brain bilirubin content, with potentially severe neurologic consequences.27,28,40,41
The cellular mechanism(s) of early reversible bilirubin toxicity (eg, behavioral changes and prolongation of interwave latencies in the brainstem auditory evoked response [ABR]) are unknown, but exposure of neurons to high levels of Bf will produce apoptosis42–44 and/or necrosis associated with mitochondria dysfunction, possibly produced by bilirubin disruption of the proton gradient required for oxidative phosphorylation.44
In summary, the distribution of bilirubin in the body can be described by using established pharmacokinetic principles. Although much is known about the physiology of bilirubin transport, brain uptake, and toxicity, transformation of this "bench" knowledge22,34,39,45 to clinical research and bedside strategy has been very slow.
|
ASSESSING RISK FOR NEUROTOXICITY AND KERNICTERUS BY USING TSB |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
ABR abnormalities may reverse or improve after phototherapy or exchange transfusion47 but sometimes require months to normalize.53 This slow recovery may be testimony to the plasticity of developing injured brain rather than truly "reversible" bilirubin toxicity. Permanent neurosensory hearing loss may be the only clinical manifestation of kernicterus. A similar sequence of ABR changes has been observed in jaundiced Gunn rat pups given an albumin-binding displacer (sulfadimethoxine).54–56 The abnormalities could be partially reversed with rapid intervention.54
Hyperbilirubinemia can also produce changes in electrocortical activity (increased delta frequency and decreased frequency and amplitude of theta, alpha, and beta waves recorded at TSB levels ranging from 16 to 33 mg/dL),57 delay electrocortical maturation in term newborns,57 and delay ABR maturation in premature infants.58 Premature and near-term infants may be particularly sensitive to bilirubin-induced ABR changes and residual hearing loss.59,60 An association of apnea and bradycardia during the first 2 weeks of life with increased TSB, Bf, and ABR changes was reported recently,61 suggesting that brainstem toxicity may extend beyond auditory pathways in premature infants. Apnea is a common manifestation of toxicity in premature primates that are infused with bilirubin and may precede changes in the ABR.27
In 1996, Starr et al62 described a syndrome of auditory neuropathy (or auditory dyssynchrony) (AN/AD)63–65 characterized by an absent or severely distorted ABR, normal otoacoustic emissions (ie, normal hair cells), normal cochlear microphonic responses (auditory nerve), and variable hearing impairment. The same constellation of findings had been observed previously by Chisin et al66 in 1979 in children with hearing loss caused by hyperbilirubinemia. In sparing the inner ear and acoustic nerve, AN/AD is quite distinct from most causes of hearing loss.
Children with AN/AD have difficulty understanding speech in the absence of significant hearing loss and may have delayed speech development, behavioral problems, and learning disability.67 AN/AD may account for 11% of children with permanent hearing deficits,67,68 with a reported prevalence of 5.3% to 14.8% in infants discharged from NICUs.69 Hyperbilirubinemia and prematurity are significant risk factors for AN/AD,68,70–72 accounting for more than half of the patients with the syndrome.73 When associated with hyperbilirubinemia, AN/AD is more likely to improve over time. Although the incidence of AN/AD in the premature population and its relationship to TSB and Bf remains to be firmly established, the syndrome seems to be an important component of the bilirubin-induced brain-injury spectrum involving auditory pathways.61
Among healthy term or near-term infants without isoimmune hemolytic disease, there is no evidence that a TSB concentration maintained at <20 to 21 mg/dL will produce significant permanent neurologic sequelae,5–9 but how well does TSB discriminate infants at risk for kernicterus when the TSB exceeds 20 mg/dL? How many patients with severe hyperbilirubinemia escape injury, and what is the TSB in patients who develop kernicterus?
Data relating TSB to kernicterus in otherwise healthy term infants are extensive, but conclusions are sparse because of the rarity of the event, logistic and ethical constraints in conducting prospective outcome research with controlled intervention, inadequate documentation of TSB before readmission with symptoms, and varying methods of assessing outcome (eg, acute bilirubin-induced neurologic dysfunction versus permanent injury; subtle deficits; focal or severe damage). Newman and Maisels8 conducted an extensive analysis of available data in 1990 that was dominated by a reanalysis of outcome data from the Collaborative Perinatal Project, a multicenter cohort study of >54000 pregnancies between 1959 and 1965. Their conclusion that "infants without hemolysis are not at risk of mental or physical impairment until serum bilirubin levels rise well above 20 mg/dL" was instrumental in liberalizing the standard of care in the 1994 AAP practice parameter.8,74,75 In their analysis, Newman and Maisels assumed a linear relationship between TSB and outcome that, as described above, is unlikely to occur. In a subsequent study of outcome at 7 to 8 years of age,9 a significant association of hyperbilirubinemia and abnormal or suspicious neurologic abnormalities was found when comparing groups who had neonatal TSBs of <10 mg/dL and those who had TSBs of >20 mg/dL. More recent studies have also reported that mild neurologic abnormalities are more frequent in infants with high bilirubin levels who are evaluated at ages ranging from 12 months to 15 years,74,77 although the functional significance of abnormalities in all 3 studies is uncertain.
Three retrospective chart reviews of infants who were readmitted with severe hyperbilirubinemia provide an insight to the degree of risk. Newman et al10 reviewed the outcome of all infants with TSBs of >30 mg/dL (513 µmol/L) who were readmitted to a large health care maintenance organization over a 4-year period. All 11 patients received aggressive bilirubin-reduction interventions. The TSB ranged from 30.7 mg/dL (525 µmol/L) to 45.5 mg/dL (778 µmol/L). None had documented acute symptoms of encephalopathy, and 9 were reported to have normal neurologic examinations when they were evaluated at ages ranging from 18 months to 5 years. One died of sudden infant death syndrome, and a second was receiving speech therapy but was otherwise normal.
Ahlfors and Herbsman78 examined 8 term infants who were readmitted with TSBs ranging from 28.3 to 34.2 mg/dL. One infant with a TSB of 33.1 mg/dL (566 µmol/L) failed an "Algo" hearing screen, and a second, who was admitted with a TSB of 31.7 mg/dL (542 µmol/L), had acute symptoms of encephalopathy, seizures, and apnea and died with pathologic kernicterus that was found at autopsy. The infant had glucose-6-phosphate dehydrogenase (G6PD) deficiency but no evidence of hemolysis or infection. The remaining infants were asymptomatic, but follow-up was not reported. All surviving infants received exchange transfusion.
Harris et al79 reported 6 readmissions of infants with TSBs of >25 mg/dL and acute signs of encephalopathy in their hospital but did not state the number of those who were admitted with TSBs of >25 mg/dL without symptoms during the same time interval. Admission TSBs ranged from 26.4 to 36.9 mg/dL (451–631 µmol/L). In contrast to the report by Newman et al,10 5 of the 6 infants were symptomatic on admission. MRI was abnormal in 3 of 4 infants tested (including an asymptomatic infant with a TSB of 26.4 mg/dL), and 2 of 5 infants had abnormal ABRs. Follow-up was normal for 4 infants, 1 had residual hearing loss, and 1 had severe cerebral palsy and mental retardation, which is atypical of kernicterus but consistent with global bilirubin encephalopathy observed in animals with a permeable BBB.
These 3 studies are summarized in Fig 1, which illustrates the wide spectrum of response to severe hyperbilirubinemia. Because the denominator of asymptomatic readmissions is uncertain in the latter 2 studies, Fig 1 should reflect the maximum risk for kernicterus in this TSB range. Most infants had no acute or residual effects from severe hyperbilirubinemia, and TSB did not discriminate which infants would or would not develop kernicterus (ie, low specificity). Most of these infants received aggressive phototherapy and/or exchange transfusion, which may relate to the good outcome.
|
32 mg/dL; range: 17.7–51 mg/dL). The mean TSB and lowest TSB in these infants were 5 mg/dL lower than in kernicteric infants with idiopathic hyperbilirubinemia.
|
25 mg/dL. Fifteen percent of the patients with kernicterus had a TSB of <30.1 mg/dL (515 µmol/L), and 50% had a TSB of <38.5 mg/dL (658 µmol/L).
|
In summary, kernicterus is a rare complication of neonatal unconjugated hyperbilirubinemia, and most healthy term infants with a TSB of 25 to 40 mg/dL escape without significant permanent damage. Most of these infants received intensive therapy (phototherapy and/or exchange transfusion), so risks for kernicterus with prolonged exposure and without intervention is unknown. On the other hand, 8% to 9% of reported kernicterus cases occurred with a TSB of <25 mg/dL, with documented admission TSBs as low as 20.7 mg/dL (354 µmol/L). The recommendation for aggressive intervention in term infants with a TSB of 
25 mg/dL balances the very low specificity of TSB for predicting kernicterus with an "acceptable" sensitivity that includes 92% of patients with kernicterus.
|
ASSESSING RISK FOR NEUROTOXICITY AND KERNICTERUS BY USING Bf |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
During the next 2 decades, a number of studies demonstrated positive correlations between Bf measured by Sephadex gel filtration and kernicterus,92–94 and 2 small prospective studies suggested that binding tests might improve predictability of developmental outcome.95,96 This flurry of interest subsided with the introduction of Rh immune globulin and phototherapy to prevent or treat hyperbilirubinemia, and the ensuing decreased prevalence of severe hyperbilirubinemia in term infants made studies to validate the predictive value of Bf logistically impossible. At the same time, several publications15,17,23 (discussed in this review) questioned the value of binding tests, clinician's concerns about jaundice were transformed from fear to annoyance, and the measurement of Bf was relegated to a few research laboratories. The recent resurgence of kernicterus in both the United States and Europe and the recognition that TSB has limited predictive power for kernicterus have challenged us to reexamine whether measurement of Bf might improve our clinical management of hyperbilirubinemia.
Several in vitro and animal studies, which most often use the peroxidase method, have compared the effects of TSB and Bf (or binding) on bilirubin uptake or toxicity. Bf, but not TSB, correlated with bilirubin uptake and toxicity in all studies regardless of whether they used cultured cells,97–99 measured neurotoxicity (electroencephalogram) in animals with osmotic opening of the BBB,18 or recorded ABR changes in Gunn rats.100
Clinical application of the peroxidase method was facilitated by the development of a US Food and Drug Administration–approved dedicated instrument (Arrows Co, Ltd, Osaka, Japan) that can measure both TSB and Bf using a 25-µL serum sample. A Bf of 20 nmol/L (1.2 µg/dL) was hypothesized to be a risk threshold for bilirubin toxicity in term infants on the basis of the highest Bf measured after titrating cord sera to a TSB of 20 mg/dL (R.P.W., unpublished data). Similar conclusions were made by a historically based analysis of kernicterus by Ahlfors.101 These estimates were supported by 2 clinical studies of term infants49,50 in which Bf correlated well with marked changes in the ABR (Bf range: 0.9–1.9 µg/dL), whereas TSB and TSB/A did not discriminate infants with and without toxicity.
Murki et al102 prospectively studied 64 term infants without hemolytic disease who were readmitted for hyperbilirubinemia over a 1-year period. Fourteen infants had clinical evidence of acute toxicity with TSBs ranging from 17.5 to 46 mg/dL (299–787 µmol/L) and Bf levels ranging from 12.6 to 46 nmol/L (0.73–2.7 µg/dL). The mean Bf, TSB, and TSB/A were all statistically higher in the group of infants who were symptomatic. The authors did not report how many of the affected infants had residual damage.
In premature infants, Bf but not TSB correlated with ABR changes, impairment of ABR maturation,58 and clinical signs of acute bilirubin toxicity,103 although thresholds for toxicity seem to be somewhat lower than in term infants. In a study of 138 premature infants, Nakamura et al104 identified 12 infants with clinical evidence of bilirubin toxicity. TSBs ranged from 10.4 to 27.6 mg/dL (178–472 µmol/L), and Bf ranged from 0.8 to 1.76 µg/dL. A Bf of 1.0 µg/dL best discriminated toxic from asymptomatic low birth weight infants (1500–2500 g) with a sensitivity of 100% and specificity of 98%. In very low birth weight infants (<1500 g), a Bf of 0.8 µg/dL correctly identified 4 symptomatic infants with similar accuracy. One of these, a 644-g premature infants with a TSB of 11.5 mg/dL (197 µmol/L) and a Bf of 1.7 µg/dL at 14 days of age, died on day 15 with kernicterus proven at autopsy. This was the only 1 of 12 symptomatic infants who did not receive an exchange transfusion.
Bf in a group of 9 term infants who were readmitted for hyperbilirubinemia (TSB > 28 mg/dL) was evaluated recently with a modified peroxidase method using 2 peroxidase concentrations to correct for rate-limiting dissociation of bilirubin from albumin.78 This modification yields a higher Bf than the standard method. One infant with classic kernicterus had a TSB of 31.7 mg/dL (542 µmol/L) and a Bf of 7.6 µg/dL, a value far exceeding the reported aqueous solubility of unconjugated bilirubin.13 Eight additional infants with TSBs ranging from 28.3 to 34.2 mg/dL had Bf ranging from 1.8 to 4.4 µg/dL. One infant, with a Bf of 4.4 µg/dL, failed an Algo hearing screen, whereas the remaining infants were reported to be asymptomatic. Long-term follow-up was not reported.
The standard peroxidase method,90 using a 1:41 serum dilution, has several limitations. First, serum dilution yields falsely low Bf results in infants with elevated serum levels of competing ligands (eg, sulfonamides, free fatty acids) that require high serum concentrations to effectively displace bilirubin from albumin. Second, serum dilution increases the affinity of albumin for bilirubin, possibly by decreasing albumin dimerization. The binding constant for very dilute human serum albumin is >108 L/mol, whereas K is closer to 5 x 106 L/mol in undiluted serum.24,25,105,114 The Bf measured in minimally diluted serum is 5- to 10-fold greater than that measured by the standard method. Furthermore, the standard method uses a single peroxidase concentration, which underestimates Bf, an error that worsens as the Bf concentration increases. Accurate determination of Bf requires that the dissociation rate of bilirubin from albumin be much greater than the rate of oxidation, a condition that requires analysis using dilute peroxidase when the Bf is high. Finally, conjugated bilirubin, if present, will oxidize rapidly, which gives a falsely high value for unbound unconjugated bilirubin (because it is a direct spectrophotometric assay that measures the loss of yellow pigment). Ahlfors105 recently developed an improved peroxidase assay that minimizes the problems noted above. The method uses minimally diluted serum (1:1 dilution), 2 enzyme concentrations, and a diazo reaction to exclude conjugated bilirubin.
Notwithstanding these limitations, Bf measured by the dilute-peroxidase method has correlated remarkably well with various indicators of bilirubin toxicity,* probably because dilution yields a rather consistent systematic change in bilirubin–albumin binding and apparent Bf in normal infants. This is not true when binding competitors are present. Symptomatic infants reported by Murki et al102 had elevated free fatty acid levels compared with well infants (1.58 vs 0.88 mmol/L), which suggests that the Bf concentration may have been even higher than measured, because free fatty acids are known to alter bilirubin-albumin binding.44
Ritter et al15 measured Bf in 30 premature infants who were autopsied during the benzyl-alcohol era in which elevated plasma levels of benzoate displaced bilirubin from albumin. Seven of these infants had kernicterus with a mean Bf of 18.2 nmol/L (1.1 µg/dL) and mean TSB of 7.3 mg/dL compared with mean values of Bf and TSB in nonkernicteric infants of 10.9 nmol/L (0.64 µg/dL) and 6.1 mg/dL, respectively. Although most Bfs in the kernicterus group were in a range predicted to place the infant at risk, the differences in Bf were not significant when using the Mann-Whitney U test (P = .07); they were significant when using the Student's t test. However, yellow staining was more intense in brains exposed to higher Bf. It was not documented whether these infants received benzyl alcohol, but the high mean Bf relative to mean TSB (expected Bf would be <8 nmol/L) in kernicteric patients is consistent with the presence of strong competitors and/or marked alterations of binding constants. Cashore and Oh103 reported similar findings in 13 autopsied premature infants. Mean TSB and Bf values in 5 infants with kernicterus were 8.6 mg/dL and 27 nmol/L (1.6 µg/dL), respectively, compared with 8.0 mg/dL and 13 nmol/L (0.76 µg/dL), respectively, in 8 infants without kernicterus (P < .05).
In summary, pharmacokinetic principles, in vitro and animal experiments, and limited clinical data in both term and premature infants with clinical evidence of early encephalopathy or frank kernicterus consistently support the propositions that Bf is superior to TSB in predicting risk for bilirubin toxicity and that clinical management might be improved significantly by measuring both variables. Biologically, this is not surprising, because TSB is not the principal determinant of bilirubin entry into brain but rather is a dependent variable, the effect of which is greatly modified by variations in serum binding.
|
THE CHALLENGE: DEVELOPING A MORE RATIONAL APPROACH TO HYPERBILIRUBINEMIA |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
A resurgence of kernicterus in the past decade107 has forced a reevaluation of the 1994 practice parameters. The current AAP guidelines,3 which promote greater parent education and closer predischarge and postdischarge surveillance for jaundice, address the need to prevent infants with hyperbilirubinemia from "falling through the cracks."4 However, the 2004 guidelines continue to base intervention strategies on TSB even while acknowledging that clinical evidence providing a specific risk threshold is weak.
The challenge in defining national guidelines for the management of patients with hyperbilirubinemia is to minimize both the risk for bilirubin encephalopathy and the need for intervention.74 It has been estimated that between 1 and 3 per 100000 otherwise well newborns would develop kernicterus if untreated108 (35–105 cases per year in the United States with 3.5 million term deliveries). Approximately 70000 term newborns (2% of births) will develop a TSB of >20 mg/dL.109 It is clear that a health care policy that demands intervention at a TSB yielding 100% sensitivity (ie, 20–21 mg/dL) would result in an enormous number of unnecessary interventions. Approximately 5000 infants (0.15% of term infants) will exceed AAP guidelines (TSB: 25 mg/dL) each year and require intervention with aggressive phototherapy or exchange transfusion to prevent 92% of kernicterus cases (32–96 cases; 8% will have a TSB <25 mg/dL).
Available clinical data indicate that Bf or Bf combined with TSB should improve risk assessment for neurotoxicity in both term and premature infants. If Bf rather than TSB were used as an indicator for aggressive interventions,109 Ahlfors and Wennberg105 have estimated that the number of interventions, including exchange transfusion with its associated mortality and morbidity,7,110–112 would be greatly reduced. If verified by clinical trials, guidelines based on Bf could yield significant savings in health care dollars, neonatal morbidity, possibly mortality, and family stress.
Although both the AAP3 and the National Institute of Child Health and Human Development113 have identified the need for studies of Bf to improve intervention criteria, the current guidelines based on TSB effectively preclude conducting such studies in patients who are at risk for kernicterus. How then do we proceed? Ahlfors and Wennberg109 suggested a stepwise approach for clinical evaluation of Bf in infants. First, it is important to obtain population data regarding interindividual variations in binding in different clinical settings. This does not require obtaining serum from jaundiced infants but can be done by titrating cord serum with bilirubin in vitro and measuring Bf at TSB levels thought to be "at risk." A second step would be to evaluate day-to-day intraindividual variation in binding. Unpublished clinical experience using the standard peroxidase method suggests that, in the absence of intervening severe illness, binding parameters change little in a given patient. A systematic assessment of day-to-day variation in the Bf level linked to age-specific changes in the TSB6 might improve predictability of severe hyperbilirubinemia, because the Bf should be a better indicator of bilirubin load.
Although controlled, randomized trials that test the efficacy of TSB versus Bf in predicting outcome cannot be performed ethically or logistically, it is possible to compare TSB and Bf as predictors of "reversible" toxicity using bilirubin-induced neurologic dysfunction (BIND) scores,6,59 MRI, or the ABR as dependent variables. The ABR may be the best clinical indicator that we have for evolving bilirubin neurotoxicity53,58 and can identify patients in need of close follow-up for hearing loss or auditory processing difficulties. Because deafness is a common manifestation of kernicterus, it would seem incumbent on those who dismiss the ABR as a transient bilirubin "effect" to show that an abnormal ABR is independent of progressive neurotoxicity leading to hearing loss or auditory processing difficulties.
The greater challenge continues to be in relating Bf and/or TSB to long-term outcome. Because kernicterus is a rare complication of idiopathic hyperbilirubinemia, outcome studies measuring TSB and using kernicterus as the primary outcome variable have produced inconclusive results even in large collaborative projects. Likewise, long-term outcome studies conducted within large health care maintenance organizations have thus far only determined that the risk for permanent sequelae in healthy term patients with high TSBs is very small.10,114 It is unlikely that data allowing meaningful sensitivity, specificity, and receiver operating characteristic curve calculations in this population will be forthcoming without a national reporting system for infants with very high TSB (eg, >25 mg/dL) and available laboratories that can measure Bf and relate it to outcome. It may be more feasible to collect such data for premature infants in newborn intensive care settings in which the sporadic occurrence of kernicterus in sick low birth weight infants with low TSB continues to be reported,115 the effect size of TSB on neurodevelopmental outcome is small,116 and a national consensus for intervention criteria is (correctly and fortunately) not established.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address correspondence to Richard P. Wennberg, MD, Division of Neonatology, Department of Pediatrics, University of Washington, Box 356320, 1959 NE Pacific St, RR542 HSB, Seattle, WA 98195. E-mail: rpwennberg{at}hotmail.com
Financial Disclosure: Dr Ahlfors has a financial interest in an automated system for free-bilirubin–concentration determination.
* Refs 47, 49, 50, 58, 60, 102–104, and 106. 
|
REFERENCES |
|---|
|
TOPABSTRACTDETERMINANTS OF TSBENTRY OF BILIRUBIN INTO...ASSESSING RISK FOR NEUROTOXICITY...ASSESSING RISK FOR NEUROTOXICITY...THE CHALLENGE: DEVELOPING A...CONCLUSIONSREFERENCES |
|---|
