Clinicians have known for decades that newborns with hyperbilirubinemia and excessive bilirubin production or poor plasma bilirubin binding have an increased risk of bilirubin-induced neurologic dysfunction (BIND, kernicterus).1,2,3,4,5 Both conditions increase brain bilirubin exposure, excessive production by increasing the bilirubin load,4, 6,7,8 and poor plasma bilirubin binding by increasing the plasma levels of unbound (free) bilirubin (Bf), which is the bilirubin species moving freely between the vascular and extravascular spaces.1,2,3, 9,10,11,12,13,14

The elevated Bf resulting from poor bilirubin binding forces a greater percentage of the accumulating bilirubin load (bilirubin produced – bilirubin excreted) into the extravascular space, exposing the brain to more bilirubin.2, 5, 9, 10, 12, 14 The vascular bilirubin level and plasma total bilirubin concentration (BT) will be correspondingly lower.1, 3, 9, 10 If Andersen et al.1 had measured Bf in their premature newborns given daily sulfisoxazole or tetracycline, they would have found it to be about twice as high in those receiving sulfisoxazole versus those receiving tetracycline at any BT.9 A greater load of bilirubin was therefore needed in the sulfisoxazole cohort to reach a given BT e.g., >12 mg/dL,9 and the increased extravascular bilirubin lead to BIND occurring at BT well below that at which BIND would ordinarily occur in this population.9, 15 Clearly, both poor bilirubin binding and the size of the bilirubin load were important factors in determining whether newborns receiving sulfisoxazole actually developed BIND.3, 9, 14, 16

The axiom (not hypothesis) emanating from these elderly data is: both BT and Bf contribute important information about the risk of BIND in hyperbilirubinemic newborns. BT reflects the size of the bilirubin load, and Bf provides information about the percentage of the bilirubin load to which the brain is exposed at that BT. Bilirubin binding varies considerably in newborn and adult populations, making it impossible to gauge the Bf from the magnitude of BT.9, 10, 16,17,18

Currently, the hour-specific BT is used to assess the size of the bilirubin load,19 and if coupled with Bf measurements, clinicians would have even more information about the brain bilirubin exposure and the risk of BIND.16, 20 Recent studies suggest the risk of BIND in term newborns with BT ≥ 30 mg/dL is about 10%.21,22,23 Since there is considerable uncertainty as to when an exchange transfusion should be initiated in these newborns,21,22,23,24,25 measuring bilirubin binding would be more helpful in identifying those needing the procedure.

Remarkably, despite the accumulating studies documenting the importance of Bf in assessing the risk of BIND,26,27,28,29,30,31,32,33,34 bilirubin binding has not found its way to the bedside except in Japan,35 where BT and Bf are used routinely and successfully for many years.30,31,32,33 Elsewhere, BT alone guides treatment with phototherapy and/or exchange transfusion using BT thresholds derived from ever-evolving clinical experience and expert consensus.36,37,38 Unfortunately, this approach has resulted in excessive treatment, erratic clinician compliance, and considerable confusion without eliminating BIND.39,40,41,42,43,44,45,46,47,48,49,50,51,52,53 Adding bilirubin binding to the clinical armamentarium is unlikely to resolve all these issues, but it will certainly offer some improvements for the current situation.

The clinical use of bilirubin binding has been effectively blocked because clinicians have been asking the wrong questions. BT was the original and only clinical laboratory (CL) test available for quantifying newborn jaundice, and it was several years before clinically suitable ‘bilirubin binding tests’ were developed.5, 54,55,56 Clinicians then asked for proof that the ‘binding tests’ were superior to BT as predictors of BIND, which was not forthcoming.57,58,59 The better question would have been ‘How well does a jaundiced newborn who has reached a threshold BT bind bilirubin?’ That question provides a natural pathway for bringing bilirubin binding to the bedside that is not dependent on problematic prospective studies comparing how well ‘binding tests’ and BT predict BIND.

In this Review, we argue that the key to bringing bilirubin binding safely and expeditiously to the bedside is to determine the population parameters for Bf at current BT thresholds. The Bf population data will serve as the reference metric60 for quantifying how well a jaundiced newborn binds bilirubin as an additional factor in determining their risk of BIND, irrespective of whether the incidence of BIND is known. Bf population parameters at threshold BT levels can be obtained by using a bilirubin binding panel (BBP) to measure bilirubin binding in about 400 patients from the population of interest.20, 61 In our paradigm, both BT and Bf are included in the BBP.

Two additional variables are needed in the BBP, the maximum BT (BTmax) and its corresponding equilibrium association constant (K). BTmax is an empirical mass action upper limit of BT in a sample that is used with K to calculate Bf at any BT < BTmax (Equation 1). BTmax and K are necessary to calculate Bf at the threshold BT level since the BT of a sample will rarely be the threshold BT. It is important to note that BTmax is not the plasma bilirubin binding capacity (BBC), i.e., the BT level at which ALL the plasma bilirubin binding sites are occupied. There are no tests for determining BBC although the term continues to be mistakenly applied to ‘bilirubin binding test’ endpoints.17, 55 The BBP variables and other binding measurements are summarized in Table. 1.

Table 1 Bilirubin binding variables54, 55, 64
Full size table

The mass action chemical and mathematical relationships hematofluometry, not jematofluometry, could not get system to eliminate J and add H between the BBP components (BT, Bf, BTmax, and K) are shown below.62, 63 BT is the bound bilirubin concentration, since Bf is orders of magnitude less than BT (µg/dL versus mg/dL) and bound bilirubin = BT – Bf ≅ BT. The unoccupied binding sites in the derivation of Equation 1 are BTmax – BT.63

$$\begin{array}{*{20}{l}} {\left( {{B}_{{\mathrm{Tmax}}} - {B}_{\mathrm{T}}} \right) + {B}_{\mathrm{f}}} {\begin{array}{*{20}{c}} {K} \atop\leftrightarrows \end{array}} {{B}_{\mathrm{T}}} \\ \hskip -12pt {K} = {\frac{{{\mathrm{BT}}}}{{{B}_{\mathrm{f}}\left( {{B}_{{\mathrm{Tmax}}} - {B}_{\mathrm{T}}} \right)}}} \end{array}$$
(1)

A practical issue is how the BBP variables are obtained. BT has long been measured by the clinical laboratory (CL), and clinically suitable methods for measuring Bf are available.35, 54, 64 K is calculated from BT, Bf, and BTmax using Equation 1. BTmax, however, is a ‘new’ variable introduced herein to avoid the long-standing misconception that albumin has only one bilirubin binding site that is clinically relevant.17, 55, 64,65,66,67,68 The advantage of using the ‘single-site’ model is that the simplest mass action equations apply (e.g., Equation 1), and measured variables such as the concentrations of albumin or albumin-bound bilirubin that fluoresces55, 68 can be designated as the ‘BBC’ and substituted for BTmax in Equation 1(17,55,64-68). Unfortunately, Equation 1 will not provide accurate estimates of Bf at non-sample BT using the ‘BBC’ and corresponding K provided by these other methods.

Albumin has at least two bilirubin binding sites as shown in Fig. 1 .69 The ~10-fold difference in their affinity for bilirubin as quantified by their respective association constants, Khi > 10·Klo62, 63, 69 is used to justify the assumption that only the high affinity site where bound bilirubin also fluoresces is clinically relevant.18, 54, 67 Fig. 2 shows that both sites will significantly impact the change in Bf as BT increases, invalidating the assumption that only the high affinity site is clinically relevant.

Fig. 1
figure 1

Schematic of the albumin molecule showing two bilirubin binding sites. Bilirubin bound to site I fluoresces and its equilibrium association constant (Khi) is about 10-fold more tightly than site II69

Full size image
Fig. 2
figure 2

The change in Bf as BT increases at an albumin concentration of 3.5 g/dL and assuming a single binding site with Khi = 1.71 dL/µg (100 L/µmol) or two binding sites with Khi = 1.71 dL/µg and Klo = 0.171 dL/µg (see Fig. 1). The individual points were calculated using standard binding equations.62, 63 The lower affinity binding site significantly impacts the change in Bf over the clinically relevant range of BT

Full size image

The inaccuracy of the ‘single-site’ model in predicting Bf at non-sample BT is illustrated using bilirubin binding data from 72 newborns < 35 weeks’ GA.70 Table 2 contains a summary of the data before and after sample bilirubin enrichment. The concentration of binding sites where the bound bilirubin fluoresces (HMF ‘BBC’) was measured using hematofluorometry (HMF).55 Table 3 shows the Bf predicted using the CL albumin (AT) or the HMF ‘BBC’ as BTmax in Equation 1, with the corresponding K’s calculated using the sample Bf before bilirubin enrichment. As expected from Fig. 2, both predicted higher average Bf at the enriched BT compared with the measured Bf, and Bf could not be calculated in several samples because the enriched BT exceeded the BTmax determined using the CL AT or the HMF test.

Table 2 Summary of bilirubin binding measurements before and after sample bilirubin enrichment in 110 samples from 72 newborns 24 to 34 weeks’ GA (median 30 weeks) with birthweights 0.483–2.490 kg (median 1.305 kg). The B/AT molar ratios varied from 0.10 to 0.76 and 0.31 to 1.45 before and after sample bilirubin enrichment, respectively
Full size table
Table 3 Bf calculated at the enriched BT for the 110 samples in Table 2 using Equation 1 and BTmax estimated using either the CL AT (8.8 × AT in g/dL) or the HMF ‘BBC’ and their corresponding K’s determined using Equation 1
Full size table

An empirical BTmax that does not require assumptions about the actual bilirubin binding sites and will provide accurate estimates of Bf at non-sample BT < BTmax can be readily obtained from the BT and Bf measured before and after sample bilirubin enrichment (BT1, Bf1 and BT2, Bf2). BTmax and K in Equation 1 are both considered unknowns, and Equation 1 is solved for BTmax using Equation 2 (70). Enriching samples with bilirubin to BT above the relevant BT thresholds and re-measuring BT and Bf to obtain BTmax and K for accurate estimates of Bf at threshold BT < BTmax will not require substantially more sample volume or time for testing.17, 54, 71

$${B}_{{\mathrm{Tmax}}\,} = \frac{{{B}_{{\mathrm{T1}}}{B}_{{\mathrm{T2}}}\left( {{B}_{{\mathrm{f2}}}{ - B}_{{\mathrm{f1}}}} \right)}}{{{B}_{{\mathrm{T1}}}{B}_{{\mathrm{f2}}} - {B}_{{\mathrm{T2}}}{ - B}_{{\mathrm{f1}}}}}$$
(2)

Once Bf population parameters are available, the BBP can be used to help determine how best to treat a newborn that has reached or exceeded a BT threshold and also as a screening tool72 to individualize the threshold BT as warranted. Table 4 shows current BT threshold guidelines for phototherapy and exchange transfusion in newborns <35 weeks’ GA. Consider a clinician (armed with bilirubin binding) confronting a <28-week GA newborn that has reached the phototherapy BT threshold of 5 mg/dL. Assume the median Bf for the population at BT = 5 mg/dL is 0.30 µg/dL with the 25th percentile 0.15 µg/dL and the 75th percentile 0.45 µg/dL. If the newborn’s Bf is >0.30 µg/dL phototherapy is indicated and if it is >0.45 µg/dL there is considerable cause for concern as the binding is ‘poor’ relative to the population. If the median Bf for the population at the threshold exchange transfusion BT were 1.0 µg/dL,34 BTmax and K from the newborn’s BBP and the Bf of 1.0 µg/dL can be used in Equation 1 to calculate the actual BT at which the Bf would reach 1.0 µg/dL. This approach provides considerably more information about the risk of BIND even though the actual risk is unknown and individualizes care.

Table 4 Hyperbilirubinemia treatment guidelines for newborns <35 weeks’ GA38
Full size table

It is important to note that Bf varies with the local environment, temperature, sample dilution, GA, illness, and the presence of endogenous or exogenous ligands competing for bilirubin binding sites.9, 10, 14, 17, 73,74,75,76,77,78,79,80 Therefore, comparisons of outcome and Bf across institutions will require measuring Bf under standard conditions. BTmax and K, although empirical with respect to specific bilirubin binding sites, are critical binding components required for reliable determination of Bf at relevant threshold BT. The use of the BBP will individualize care and as patient data accumulates may help better clarify the relationships between neonatal bilirubin exposure and BIND. In the meantime, the BBP can provide additional guidance for the clinicians caring for jaundiced newborns.

Summary

BIND is a spectrum of neurological disorders affecting jaundiced newborns and unpredictably related to the magnitude of BT. The BBP (BT, Bf, BTmax, and K) can provide important additional information about the risk of developing BIND and the need for treatment at threshold BT, irrespective of whether the actual incidence of BIND at or above that BT is known. The BBP components can be obtained using minimal modification of well described, clinically suitable methods, and when configured as described in this Review, provide accurate estimations of Bf at BT < BTmax. The BBP can be used to establish Bf population parameters at current threshold BT that are the metric for quantifying how well a newborn binds bilirubin, which provides important additional information the risk of BIND and need for intervention The BBP is useful as a screening tool as well as for determining the urgency of treatment in newborns reaching or exceeding threshold BT. This approach to the clinical of use of bilirubin binding augments current clinical practice within the context of established BT thresholds and helps individualize care.