Clinical and Laboratory Medicine-Sci Forschen

Full Text

Research Article
Monitoring Bilirubin Binding Parameters in a Cohort of Japanese Neonates

  Ichiro Morioka1#      Daisuke Kurokawa1#      Sota Iwatani1      Kazumoto Iijima1      Hajime Nakamura1      Angelo A Lamola2      Vinod K Bhutani2      Ronald J Wong2*      David K Stevenson2   

1Department of Pediatrics, Kobe University Graduate School of Medicine, Kobe, Hyogo, Japan
2Department of Pediatrics, Division of Neonatal and Developmental Medicine, Stanford University School of Medicine, Stanford, CA, USA
#These authors contributed equally to the paper

*Corresponding author: Ronald J Wong, Department of Pediatrics, Division of Neonatal and Developmental Medicine, 300 Pasteur Dr, Rm S230, Stanford University School of Medicine, Stanford, CA, USA; Tel: +01-650-498-5264; E-mail:


The status of bilirubin binding to albumin (Alb) is central to personalized management of unconjugated newborn hyperbilirubinemia, especially those at risk of bilirubin-induced neurologic dysfunction. Our objectives were to: (1) compare assays of bilirubin binding using hematofluorometry (Hmf) and the glucose oxidase-peroxidase (GOD-POD) methods; and (2) elucidate the relationships between bilirubin binding parameters and gestational age (GA). 169 blood samples were obtained from 98 Japanese newborns with GAs ranging from 22-40 wks. Apparent serum unbound bilirubin (UB) concentrations were determined by GODPOD. Ratios of bound bilirubin (BB) to reserve Alb binding capacity for bilirubin (RABC) were determined by Hmf. Bilirubin binding capacity (BBC) by Hmf was compared to the calculated BBC (or 8.8 × Alb) using Alb levels as measured by the clinical laboratory. UB was compared with the ratio of total serum/plasma bilirubin (TB) to ((8.8 × Alb)-TB). Linear regression analyses were performed for all comparisons. BBC was lower than that expected when assuming one-to-one binding of bilirubin to Alb and the divergence of (8.8 x Alb) and BBC increased at earlier GAs. A strong correlation between UB and BB/RABC was found (r=0.83; P<0.001) and was independent of GA and unaltered after infants were subcategorized by phototherapy and drug use. UB and TB/((8.8 × Alb)-TB) also strongly correlated (r=0.84), but was significantly GA- dependent. The strong correlation between apparent UB and BB/RABC demonstrates a practical equivalence of the two methods, with UB and BB/RABC well correlated over the entire GA range. While the TB/Alb ratio may provide a reliable assessment of risk of bilirubin neurotoxicity, it appears increasingly unreliable as GA decreases.


Besides its transport and osmotic pressure stabilizing functions, albumin (Alb) can bind unconjugated bilirubin to sequester the toxin within the vascular space and to buffer the level of unbound bilirubin (UB) that can cross membranes and enter cells. Consequently, a low Alb level that is reflective of a low bilirubin binding capacity (BBC), increases an infant’s risk of developing bilirubin-induced neurologic dysfunction (BIND) by reducing an infant’s ability to “tolerate” a given bilirubin load [1-6]. A reduced Alb level (or hypoalbuminemia) is characteristic of neonates at early gestational ages (GAs) [7-9], and considered a major cause for the higher risk of BIND in infants at early GAs at any given total serum/ plasma bilirubin (TB) level [10,11]. But neither the Alb level alone nor the bilirubin: Alb molar ratio (BAMR) appears to be reliable predictors of an infant’s risk for developing BIND [10-13]. Previous studies using various direct assays have shown that the binding of bilirubin in newborns is lower than that expected from one-to-one binding of bilirubin to Alb as reported for adults and that this difference appears to increase as GA decreases [14-16].

Hematofluorometry (Hmf), based upon bilirubin fluorescence, can measure BBC directly in whole blood [15,17- 19]. Furthermore, Hmf specifically determines the capacity of the single strong Alb site to bind bilirubin [15,17-21]. Hmf can also assay the bilirubin bound to Alb (BB) in blood. The difference between BBC and BB (or BBC -BB) is the reserve Alb binding capacity (RABC). Within the limits of the singlesite binding model, the ratio of BB/RABC is proportional to the UB level with the proportionality constant being 1/Ka, where Ka is the affinity constant of the strong site [17,22,23].

The so-called “peroxidase” (POD) method developed by Jacobsen and Wennberg [24] for the measurement of UB in sera or plasma has been in limited use for decades [3,17,22]. Automated colorimetric devices (Arrows Co, Osaka, Japan) for this assay and for TB have been used to determine the clinical utility of this assay for UB. The general conclusion of the various investigators is that UB appears to be a better predictor of BIND than TB alone, especially in the small and early GA infants [2,3,25-27]. Currently, the limited use of the “POD” assay is due to the general unavailability of these systems for routine use outside of Japan where they are widely used. It was, therefore, of interest to compare BB/RABC by Hmf using a prototype device (Bili- 4, Aviv Biomedical, Inc, Lakewood, NJ, USA) and UB by POD using colorimetry (Arrows-2 Analyzer). Although such a comparison was made years ago using the original Aviv Hmf system and the original, non-automated POD assay [23], here we present data from a neonatal population spanning a full range of GAs, using the same blood samples collected for measurements of Alb, BBC, BB, and UB.

Sample collection

169 blood samples from 98 newborns (born from May 2015 to June 2016) were collected in serum separator tubes after phlebotomy for routine laboratory tests and stored at room temperature in the dark until use within 3 hrs. Parental informed consent was obtained prior to blood collections. After sampling for a complete blood count including hemoglobin (Hb), Hmf assays by the Bili-4 Hematofluorometer were performed using whole blood. Then, serum was separated and 20 µl was immediately measured for TB and UB using a UB Analyzer-2. All assays were performed using blood from the same draw and under subdued light.

The study protocol was approved by the ethical committee in Kobe University Graduate School of Medicine (No. 1772) and was in accordance with the current revision of the Helsinki Declaration.

Hematofluorometric (Hmf) assay

The Hmf assay has been described in detail previously [18,23,28]. In brief, 25 µl of whole blood was transferred to a small microfuge vial containing 20-µg bilirubin disodium salt, which was stored at -20°C and warmed to room temperature prior to use. After approximately 30 sec to allow for complete bilirubin to Alb binding, the vial was vortexed. A 15-µl aliquot was then placed on a slide, which was inserted into the hematofluorometer. Another 15-µl aliquot of whole blood (without the bilirubin reagent) was placed directly on another slide inserted into the instrument. The two slides were placed in the instrument within 15 sec of each other to avoid differential settling of red blood cells. BB, BBC, RABC, and BB/RABC values are displayed almost immediately. The Bili-4 Hematofluorometer was calibrated according to each manufacturer’s instructions and as follows: normal adult blood was spiked with known amounts of bilirubin and additionally checked for TB using a Reichert Unistat colorimetric instrument (Buffalo, NY) by the clinical laboratory

GOD-POD method [29-31]

This method is based upon the premise that UB is rapidly oxidized to colorless compounds by POD in the presence of hydrogen peroxide derived from glucose by mediation of glucose oxidase-peroxidase (GOD) and TB concentrations are then determined by direct spectrophotometry at 460 nm for Alb-bound bilirubin. Under these experimental conditions where bilirubin oxidation follows first-order kinetics, the rate constant is determined by measuring the oxidation velocity of bilirubin in the absence of Alb. The initial velocity is estimated from the time required for a 20% decrease in the initial TB concentration. Apparent UB is then calculated from the initial velocity of bilirubin loss and the ratio of the POD concentration to that in the standard assay solution containing Alb-free bilirubin [30,31]. In brief, the procedure is as follows: a cuvette containing 1 ml of phosphate buffer and a metal stirrer is warmed and then placed into the optical unit of the analyzer. 20 µl of serum is then added. Within 30 sec, the TB value is displayed. 25 µl of enzyme solution (GOD and POD, 3.2 units/ sec, respectively) is then added. Within 2 min, the apparent UB value is displayed. The UB Analyzer-2 was calibrated according to each manufacturer’s instructions using manufacturersupplied standard calibrators.

Other assays

Hb levels were measured by the sodium lauryl sulfatehemoglobin method using a XN 9000 instrument (Sysmex, Inc., Kobe, Japan) [32]. Conjugated (direct) serum bilirubin concentrations were measured by the bilirubin oxidase method using IatroLQ D-bil kits (Unitika Co., Okazaki, Japan) [33]. Serum Alb concentrations were measured using the modified bromocresol purple method (Kainos Laboratories, Inc., Tokyo, Japan) [34].


Clinical demographics and sample characteristics

Clinical demographics and characteristics for the 98 enrolled newborns and the 169 blood samples are shown in Table 1. Newborns with various birth weights (BWs) and GAs were included. Of the 169 blood samples, 20 (12%) samples were from newborns treated with phototherapy and 78 (46%) were from newborns treated with drugs. Eighty (47%) samples were from newborns not treated with phototherapy and any drugs. No samples with high conjugated (direct) bilirubin levels were included. Also, there were no samples from newborns treated with lipids or with hemolysis.

    Number (%)
  Birth weight (g) < 1000 8 (8)
1000 to 2499 39 (40)
> 2500 51 (52)
  Gestational age (wks) 22 to 27 9 (9)
28 to 36 37 (38)
37 to 41 52 (53)
Male   48 (49)
Small for gestational age   18 (19)
Delivery mode Vaginal delivery 47 (48)
  Cesarean-section 51 (52)
Nationality Japanese 98 (100)
Hemolysis   0 (0)
Median (range) or number (%)
Postnatal age when sampling (days)   5 (0-147)
Corrected gestational age when sampling (wks)   36 (22-43)
    Laboratory Data TB by GOD-POD method (mg/dl) 10.9 (0.8-20.9)
UB by GOD-POD method (µg/dl) 0.49 (0.01-1.42)
BB by Hmf 9.2 (2.1-18.1)
BBC by Hmf 25.5 (11.0-38.5)
Conjugated (direct) bilirubin (mg/dl) 0.3 (0.1-0.5)
Alb (g/dl) 3.3 (1.9-4.2)
Hb (g/dl) 15.0 (7.8-21.0)
              Clinical Data Infants treated with/without phototherapy 20 (12) / 149 (88)
Infants treated with* / without drugs 78 (46) / 91 (54)
  Infants treated without phototherapy and drugs   80 (47)
Infants given the following drugs*  
Ferric pyrophosphate 31 (18)
Calcium gluconate 31 (18)
Dopamine 17 (10)
Ampicillin 17 (10)
Dobutamine 16 (9)
Amino acid preparation 15 (9)
Amikacin 8 (5)
Caffeine 6 (4)
Activated vitamin D 5 (3)
Fluconazole 4 (2)
Levothyroxine 4 (2)
Furosemide 3 (2)
Phenobarbital 3 (2)
Spironolactone 2 (1)
Indomethacin 1 (1)
Fentanyl 1 (1)
Midazolam 1 (1)
Cefmetazole 1 (1)
Lipid 0 (0)

Table 1: Demographics and clinical characteristics of enrolled newborns (N= 98) and blood samples (N = 169)


To assess precision, a sub-cohort of 14 samples with different concentrations of bilirubin was randomly chosen for precision analyses (7 for Hmf and 7 for GOD-POD) (see Supplementary Table 1). Mean CVs for BB, BBC, RABC, BB/RABC, TB, and apparent UB were 5.2 ± 1.8%, 7.3 ± 3.0%, 10.9 ± 4.5%, 10.6 ± 4.0%, 2.7 ± 1.2%, and 5.5 ± 1.7%, respectively

BBC, Alb, and GA: For one-to-one binding, 8.8 mg of bilirubin can be bound to 1 gm of Alb, and therefore (8.8 × Alb) has been used clinically as an estimate of BBC. Values for (8.8 × Alb) and BBC obtained for the entire cohort were plotted as a function of GAs in completed weeks (cGA) and shown in Figure 1. The level of (8.8 × Alb) and the slope of the regression line (0.38 mg/dl/wk, r = 0.56) are in close agreement with other reports [7,14,16,28]. When the postpartum age (mean of 5 days) was added to the cGA, the slope of the linear regression (data not shown) was nearly identical (0.37 mg/dl/wk). Because most management guidelines for neonatal hyperbilirubinemia are designed specifically for infants greater than 35 wks GA [10], we compared the linear regression lines for infants with cGAs below and above 35 wks (data not shown) and found the slopes to be 0.61 and 0.07 mg/dl/wk, respectively. The linear regression line and correlation coefficient (r = 0.71) for BBC versus cGA (Figure 1) was also consistent with previous studies [7,14,16,28]. The plot of BBC versus cGA + postpartum age (data not shown) exhibited a similar linear regression (slope = 0.72 vs 0.74 mg/dl/wk when corrected for postpartum age). Similar to that observed for the (8.8 × Alb) versus cGA relationship, the slopes of the regression lines for the BBC levels (data not shown) were different for cGAs above and below 35 wks with slopes of 0.11 and 0.93 mg/dl/wk, respectively. All in all, on average, BBC was found to be much lower than (8.8 × Alb) at early cGAs and increases with cGA and comparable with (8.8 × Alb) at term.

Figure 1: Plot of corrected GA vs BBC (mg/dl) by hematofluorometry ( ) and assumed Alb capacity (mg/dl) ( ) calculated by multiplying the Alb level (g/dl) by 8.8 and their linear regressions (r = 0.71 and 0.56, respectively).

By plotting the ratio of the mean BBC to the mean (8.8 × Alb) as a function of cGA, the divergence of BBC from (8.8 x Alb) with decreasing GA is more clearly shown (Figure 2). The linear regression (r=0.93, slope=0.015 mg/dl/wk) indicates that BBC is only 70% of that expected from (8.8 x Alb) at 21 wks GA but increases to almost 100% at term.

Figure 2: The ratio of the mean BBC to the mean (8.8xAlb) at each completed week of gestation (cGA, slope = 0.015 mg/dl/wk; r = 0.93).

Because the various data comparisons with cGA showed little differences from those with cGA + postpartum age, which averaged 5 days, only analyses based on cGA are reported below.

Comparison of UB (by GOD-POD) with BB/RABC (Hmf): The single strong site model for bilirubin-Alb binding predicts a linear relationship between UB and BB/RABC with a zero intercept (see Eq. 2 in Appendix). Figure 4 shows that UB measured by the GOD-POD method and BB/RABC by Hmf correlated well (r = 0.83; P < 0.001) for the entire cohort. Comparable correlations (data not shown) were also observed for four sample subsets: samples from infants treated with (r=0.91) or without phototherapy (r=0.83); and infants treated without phototherapy and with (r=0.81) or without drugs (r=0.83).

The slope of the regression line for the plot of UB (in nM) vs BB/RABC for all samples was 13.7 nM (Figure 3). The y-intercept was very close to zero (–0.01) and indicates no systematic bias in both method and that both methods show negligible signals in the absence of bilirubin. The slopes of the regression lines for the four above-mentioned sample subsets were also comparable (13.7 ± 0.3). When the ratio of (BB/ RABC)/UB (both in nM) versus cGA was plotted (data not shown), the slope was –0.0004/wk, demonstrating that GA did not affect these assays. This was confirmed when we found nearly identical regression lines of the UB and BB/RABC plots when stratified by cGAs above and below 35 wks: for 22 to 27 wks, slope = 11.6 nM) (Figure 4A) and for 38 to 43 wks, slope = 12.2 nM (data not shown).

Figure 3: Linear correlation between the apparent UB (UB Analyzer-2) and BB/RABC (Bili-4 Hematofluorometer) for all samples (r = 0.83, slope = 13.7 nM).

Comparison of UB with TB/((8.8 × Alb) – TB): If we assume that (8.8 × Alb) is equal to the BBC and apply the single strong site binding model, then UB as a function of TB/((8.8 x Alb) – TB) should also be linear. We found that the correlation coefficient (r=0.84, Figure 5) was similar to that between UB and BB/RABC (Figure 4A), and with a y-intercept of 2 nM. However, in the single site model, a y-intercept of zero is expected. Unlike the plot of (BB/ RABC)/UB versus GA, [TB/((8.8 × Alb) – TB)]/UB (both in nM) versus cGA (data not shown) yielded a regression line with a significant slope of 0.04/wk (r = 0.55). This significant dependence on GA was confirmed from our plots of UB versus TB/((8.8 × Alb) – TB) when stratified by cGAs: for 22 to 27 wks, r=0.92, slope=21.2 nM (Figure 4B) and for 38 to 43 wks, r=0.92, slope=11.4 nM, data not shown). While the high cGA regression line was similar to those of both late and early cGA for UB versus BB/RABC, the early cGA regression line has a slope that was nearly two times larger. This difference could be due to either a reduced binding affinity at early GAs or to (8.8 × Alb) being an overestimate of BBC at early GAs, but a good estimate of BBC at later GAs. A quantitative test of the latter can be done by using the measured average ratios of BBC to (8.8 × Alb) at each GA week as shown in Figure 2. Application of this “correction” yields the results shown in Figure 4C, where the slope of the early cGA regression line (11.4 nM) was now very nearly equal to the slope of the “corrected” late GA range plot (11.8 nM) (data not shown) and of the corresponding plots of UB versus BB/RABC. All the correlations between UB and BB/ RABC and “corrected” TB/(8.8 × Alb) - TB) have high r-values, which indicate that UB and BB/RABC and “corrected” TB/((8.8 × Alb) - TB) correlate well for both the early and late GA ranges.

Figure 4: Linear correlation between: A. UB and BB/RABC for infants (slope = 11.6 nM); B. UB and TB/((8.8xAlb) – TB) (slope = 21.0 nM); and C. UB and modified TB/((8.8xAlb) – TB) (slope = 11.4 nM) for infants at cGA between 22 to 27 wks.

Figure 5: Linear correlation between the apparent UB (UB Analyzer-2) and TB/((8.8xAlb) – TB) (r = 0.40).


The premise that BBC does not always correlate with Alb levels (and therefore (8.8 × Alb)) in neonates was based on three different methods for measuring BBC [15,16]. In 2014 [28], BBC and Alb assayed using the bromocresol green method showed that BBC was almost always lower than (8.8 × Alb) and that the BBC/(8.8 × Alb) ratio increased with GA. The correlation of BBC with GA showed an r-value of 0.34 with a slope of 0.11 mg/g/wk, and BBC/(8.8 × Alb) ratios near 0.70 at 26 wks and near 0.90 at 40 wks. In contrast to these past studies, we used the more specific bromocresol purple method [34] to measure Alb, our neonatal population was genetically much more homogeneous, and BBC values were means of at least two determinations. It is, perhaps, for these reasons that the regression line of BBC/(8.8 × Alb) versus GA, while having a similar slope (0.13 mg/g/wk) had a significantly better correlation of 0.62 (P < 0.002), and comparable BBC/(8.8 × Alb) ratios of 0.72 at 22 wks and 0.98 at 41 wks.

There have been concerns that Hmf reports only the bilirubin that is bound to Alb (i.e., BB) at its primary binding site [19-21]. However, it has been shown that BBC values in neonatal sera using Hmf are nearly identical to those obtained by Sephadex column staining or to UB using GOD-POD methods [18].

Other binding sites of Alb have been hypothesized that may affect BBC/(8.8 × Alb) ratios, and therefore not accounted for by Hmf. For example, deficient bilirubin binding in the fetus would allow the transfer of bilirubin across the placental barrier. Is a portion of what is assayed as Alb, a different (fetal) form that does not bind bilirubin well or at all, and diminishes with maturation? Or is there an extremely strong competitor for primary binding, or an allosteric effector that distorts the primary site, that is present at early GAs and decreases with GA. While there are negative findings for a non-binding fetal form of Alb [35], there is evidence for a strong competitor or allosteric effector [36].

Because the UB level is more than 100-fold less than TB, direct methods to measure UB has been elusive. However, several indirect methods have been developed [17,22,23]. The automated POD method that has been used in many studies [2,14,16,22,25-27] and Hmf. An acceptable correlation of results for the two methods has been reported decades ago [23].

The precision analysis showed the CVs of UB and BB/RABC assays to be 5.5% and 10.6%, respectively, with an r-value of 0.83 and comparable (r=0.87) to a previous study [23].

It is documented that the binding affinity of Alb for bilirubin varies with the Alb concentration, especially at low concentrations [37]. The UB Analyzer-2 method measures serum diluted 52-fold so that Alb levels are also 52 times lower than physiological levels and in the range of large variation of binding affinity [37]. Hmf uses undiluted blood samples and, over the range of Alb found in neonatal sera, there is insignificant change in affinity. The variation in the affinity constant with the Alb in diluted sera used in the UB Analyzer-2 method may be one cause of difference between the methods.

BBC measurements can be confounded if the added bilirubin displaces a significant competitor for primary binding. Such known competitors include free fatty acids and some drugs. None of the infants in the study received lipids or known drug competitors. Samples with abnormal levels of unconjugated bilirubin were excluded from the present study, precluding the inaccuracies of both methods for samples with significant levels of direct bilirubin.

The single strong site model for bilirubin-Alb binding is supported by much biochemical and biophysical evidence [18- 20,22]. There are certainly secondary sites on Alb for binding bilirubin, but these have affinities in the order of 100 times smaller. Thus, the relationship of UB to the ratio (BB)/(strong binding capacity) can be well estimated by the relationships shown in the appendix, which explains how the values of Ka obtained from the linear correlations of UB with BB/RABC can be compared (see below) with those obtained from correlations of UB with TB/((8.8 × Alb) – TB).

This study demonstrated that there is a correlation between apparent UB (nM) and BB/RABC. However, the slope of 13.7 nM was different from the slope of 10.5 nM, which was previously observed in the comparison of the early prototype Hmf system [23] and apparent UB determined using the original POD method [23,24], but the difference of 18% was not unreasonable. In those previous studies, apparent UB was measured by the original non- automated POD, and not by the automated GOD-POD method, which uses H2O2 and GOD [31]. Furthermore, the serum dilution factor (52x), substrates, and POD concentrations and protocols were also different from the original POD method [23,24]. To better define the slope (that is, the Ka value that relates UB and BB/RABC), a larger dataset and careful co-calibration of the two methods are required. Presently, it appears that a provisional common scale for UB (nM) and BB/RABC is approximated by multiplying BB/RABC by about 13.

When we calculated the linear correlation of UB with TB/ ((8.8 × Alb) – TB) in our cohort, we found that it was similar to the correlation between UB and BB/RABC. However, the regression line had a significant intercept (2 nM), and thus indicated a bias of UB over TB/(8.8 × Alb – TB). This bias might be due to the divergence between BBC and (8.8 × Alb) at earlier GAs. When the data for TB/((8.8 × Alb) – TB) was stratified by GA, it was observed that the slopes of the regressions of UB vs TB/((8.8 × Alb) – TB) were distinctly different for the earlier (22 to 27 wks, slope=21 nM) and latest (38 to 43 wks, slope=11.2 nM) GA subgroups. That is, assuming that (8.8 × Alb) is a valid proxy for RABC, the mean binding affinity of the early GA group would have to be half that of the late GA group. This result is significantly different from that observed for the early and late GA groups for UB versus BB/RABC for which the slopes were comparable (11.7 ± 0.5 nM). The observed divergence of BBC and (8.8 × Alb) suggests that a “correction” of the (8.8 × Alb) values could resolve this difference. When the “correction” was made for each GA week using the values of the regression line of figure 2, the slopes of the resulting regressions of UB vs (8.8 × Alb - TB) for the early and late GA subgroups were found to be in good agreement (insignificant intercepts, slopes of 11.4 and 11.8, respectively).


It is believed that effective bilirubin binding can vary from infant to infant even in the absence of binding site competitors. Although the affinity of the strong binding site is comparable from infant to infant, the fraction of Alb that is able to bind bilirubin varies from infant to infant and that fraction becomes, on average, significantly larger as GA decreases. Our direct observation of an increasing divergence of BBC from (8.8 × Alb) at early GAs coupled with the solid evidence that Hmf specifically assays the capacity of the strong site, leads us to favor the latter view.

This observation is consistent with what is seen in clinical practice, where there is difficulty in assessing risk for developing BIND in infants at early GAs. Using the Alb level alone as a proxy for BBC, has been used as a rough estimation, but with limitations [10-13]. Apparently, early GA infants suffer two deficiencies that can reduce BBC: (1) they generally possess low Alb levels; and (2) a significant portion of their Alb may not be able to strongly bind bilirubin.

For term and near-term infants where (8.8 × Alb) and BBC appear to be comparable, the use of the TB/(8.8 × Alb) ratio can be used to assess risk. At early GAs where the divergence of (8.8 × Alb) and BBC can be very large, the TB/(8.8 × Alb) ratio is not an accurate measure of risk. The recent consensus [11] action levels for infants <35 wks GA are consistent with our findings and offer a significant improvement in risk evaluation for early GA infants. However, in the age of precision medicine, the variation in individual BBC levels observed at any GA strongly suggests the utility of individual assays of UB and/or BB/RABC.

Future clinical studies are needed to establish guidelines for the management of neonatal hyperbilirubinemia based upon an infant’s bilirubin binding status and not on their TB level alone.

Conflict of Interest

All authors state that they have no conflicts of interest.


Aviv Biomedical, Inc. provided the Bili-4 Hematofluorometer during the study period. Outside the submitted work, Ichiro Morioka has received grants from Japan Blood Product Organization, Daiichi Sankyo Co., Ltd, MSD and AbbVie LLC, lecture fees from Pfizer Japan, Inc., Novo Nordisk Pharma Ltd., Eli Lilly Japan K.K., Shionogi Co., Ltd., AbbVie LLC, and Atom Medical Corp., and a consulting fee from Sanofi K.K. and Japan Blood Product Organization. Kazumoto Iijima has received has received grants from Pfizer Japan, Inc., Daiichi Sankyo Co., Ltd, Japan Blood Product Organization, Miyarisan Pharmaceutical Co., Ltd., AbbVie LLC, CSL Behring, JCR Pharmaceuticals Co., Ltd. and Teijin Pharma Ltd., lecture fees from MSD, ALEXION Pharmaceuticals, AstraZeneca K.K., Meiji Seika Pharma Co., Ltd., Novartis Pharma K.K., Zenyaku Kogyo Co., Ltd., Chugai Pharmaceutical Co., Ltd., Astellas Pharma Inc., Daiichi Sankyo, Co., Ltd., Springer Japan, Asahi Kasei Pharma Corp, Boehringer Ingelheim and Medical Review Co., Ltd., and manuscript fees from Chugai Pharmaceutical Co., Ltd., and consulting fees from Chugai Pharmaceutical Co., Ltd., Astellas Pharma Inc. and Takeda Pharmaceutical Company Ltd.

This work was supported by Grants-in-Aid for Scientific Research (C) (grant number: 26461633) and Young Scientists (B) (grant number: 15K19652) of JSPS KAKENHI, and by Morinaga Hoshi-kai Foundation and Kurozumi Medical Foundation.

The authors thank Dr. Charles E. Ahlfors for his expertise, thoughtful discussions, and critical review of the manuscript, and Aviv Biomedical. Inc, Lakewood, NJ for supplying the Bili4 Hematofluorometer and reagent kits.


The relationship of UB to the ratio (BB)/(strong binding capacity)

UB = (1/Ka)(BB/strong binding capacity), (Eq 1)

where Ka is the strong site affinity constant.

If UB and BB/RABC, as assayed in this study, are accurate, then:

UB = (1/Ka)(BB/RABC) (Eq 2)

and if 8.8xAlb is an accurate proxy for BBC, then:

UB = (1/Ka)(BB/RABC) = (1Ka)BB/(8.8 × Alb - BB), (Eq 3)

or, since on the average BB ~ TB, then:

UB = (1/Ka)[TB/(8.8 × Alb - TB)]. (Eq 4)

Thus, values of Ka can be obtained from the linear correlations of UB with BB/RABC and compared with those obtained from correlations of UB with TB/(8.8 × Alb - TB).


  1. Ahlfors CE (2010) Predicting bilirubin neurotoxicity in jaundiced newborns. Curr Opin Pediatr 22: 129-133. [Ref.]
  2. Ahlfors CE, Amin SB, Parker AE (2009) Unbound bilirubin predicts abnormal automated auditory brainstem response in a diverse newborn population. J Perinatol 29: 305-309. [Ref.]
  3. Ahlfors CE, Wennberg RP, Ostrow JD, Tiribelli C (2009) Unbound (free) bilirubin: Improving the paradigm for evaluating neonatal jaundice. Clin Chem 55: 1288-1299. [Ref.]
  4. Amin SB, Wang H (2018) Bilirubin albumin binding and unbound unconjugated hyperbilirubinemia in premature infants. J Pediatr 192: 47-52. [Ref.]
  5. Watchko JF, Tiribelli C (2013) Bilirubin-induced neurologic damage – mechanisms and management approaches. N Engl J Med 369: 2021-2030. [Ref.]
  6. Wennberg RP, Ahlfors CE, Bhutani VK, Johnson LH, Shapiro SM (2006) Toward understanding kernicterus: A challenge to improve the management of jaundiced newborns. Pediatrics 117: 474-485. [Ref.]
  7. Cartlidge PH, Rutter N (1986) Serum albumin concentrations and oedema in the newborn. Arch Dis Child 61: 657-660. [Ref.]
  8. Lee M, Youn S, Baek KL, Kim JS (2005) Serum albumin concentrations and clinical disorders by gestational ages in preterm babies. Korean J Pediatr 48: 148-153. [Ref.]
  9. Watchko JF, Spitzer AR, Clark RH (2017) Prevalence of hypoalbuminemia and elevated bilirubin/albumin ratios in a large cohort of infants in the neonatal intensive care unit. J Pediatr 188: 280-286. [Ref.]
  10. American Academy of Pediatrics (2004) Management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation. Pediatrics 114: 297-316. [Ref.]
  11. Maisels MJ, Watchko JF, Bhutani VK, Stevenson DK (2012) An approach to the management of hyperbilirubinemia in the preterm infant less than 35 weeks of gestation. J Perinatol 32: 660-664.
  12. Hulzebos CV, Dijk PH, van Imhoff DE, Bos AF, Lopriore E, et al. (2014) The bilirubin albumin ratio in the management of hyperbilirubinemia in preterm infants to improve neurodevelopmental outcome: A randomized controlled trial – BARTrial. PLoS One 9: e99466. [Ref.]
  13. Sato Y, Morioka I, Miwa A, Yokota T, Matsuo K, et al. (2012) Is bilirubin/ albumin ratio correlated with unbound bilirubin concentration? Pediatr Int 54: 81-85. [Ref.]
  14. Bender GJ, Cashore WJ, Oh W (2007) Ontogeny of bilirubin-binding capacity and the effect of clinical status in premature infants born at less than 1300 grams. Pediatrics 120: 1067-1073. [Ref.]
  15. Brown AK, Eisinger J, Blumberg WE, Flores J, Boyle G, et al. (1980) A rapid fluorometric method for determining bilirubin levels and binding in the blood of neonates: Comparisons with a diazo method and with 2-(4’-hydroxybenzene)azobenzoic acid dye binding. Pediatrics 65: 767-776. [Ref.]
  16. Cashore WJ, Horwich A, Karotkin EH, Oh W (1977) Influence of gestational age and clinical status on bilirubin-binding capacity in newborn infants. Sephadex g-25 gel filtration technique. Am J Dis Child 131: 898-901. [Ref.]
  17. Amin SB, Lamola AA (2011) Newborn jaundice technologies: Unbound bilirubin and bilirubin binding capacity in neonates. Semin Perinatol 35: 134-140. [Ref.]
  18. Cashore WJ, Oh W, Blumberg WE, Eisinger J, Lamola AA (1980) Rapid fluorometric assay of bilirubin and bilirubin binding capacity in blood of jaundiced neonates: Comparisons with other methods. Pediatrics 66: 411-416. [Ref.]
  19. Lamola AA, Eisinger J, Blumberg WE, Patel SC, Flores J (1979) Flurorometric study of the partition of bilirubin among blood components: Basis for rapid microassays of bilirubin and bilirubin binding capacity in whole blood. Anal Biochem 100: 25-42. [Ref.]
  20. McCluskey SB, Storey GN, Brown GK, More DG, O’Sullivan WJ (1975) Fluorometric determination of “albumin-titratable bilirubin” in the jaundiced neonate. Clin Chem 21: 1638-1643. [Ref.]
  21. Roth M (1967) [fluorimetric assay of bilirubin]. Clin Chim Acta 17: 487-492. [Ref.]
  22. Ahlfors CE (2000) Unbound bilirubin associated with kernicterus: A historical approach. J Pediatr 137: 540-544. [Ref.]
  23. Wells R, Hammond K, Lamola AA, Blumberg WE (1982) Relationships of bilirubin binding parameters. Clin Chem 28: 432-439. [Ref.]
  24. Jacobsen J, Wennberg RP (1974) Determination of unbound bilirubin in the serum of newborns. Clin Chem 20: 783. [Ref.]
  25. Amin SB, Ahlfors C, Orlando MS, Dalzell LE, Merle KS, et al. (2001) Bilirubin and serial auditory brainstem responses in premature infants. Pediatrics 107: 664-670. [Ref.]
  26. Funato M, Tamai H, Shimada S, Nakamura H (1994) Vigintiphobia, unbound bilirubin, and auditory brainstem responses. Pediatrics 93: 50-53. [Ref.]
  27. Nakamura H, Takada S, Shimabuku R, Matsuo M, Matsuo T, et al. (1985) Auditory nerve and brainstem responses in newborn infants with hyperbilirubinemia. Pediatrics 75: 703-708. [Ref.]
  28. Lamola AA, Bhutani VK, Du L, Castillo Cuadrado M, Chen L, et al. (2015) Neonatal bilirubin binding capacity discerns risk of neurological dysfunction. Pediatr Res 77: 334-339. [Ref.]
  29. Morioka I, Iwatani S, Koda T, Iijima K, Nakamura H (2015) Disorders of bilirubin binding to albumin and bilirubin-induced neurologic dysfunction. Semin Fetal Neonatal Med 20: 31-36. [Ref.]
  30. Nakamura H, Lee Y (1977) Microdetermination of unbound bilirubin in icteric newborn sera: An enzymatic method employing peroxidase and glucose oxidase. Clin Chim Acta 79: 411-417. [Ref.]
  31. Shimabuku R, Nakamura H (1982) Total and unbound bilirubin determination using an automated peroxidase micromethod. Kobe J Med Sci 28: 91-104. [Ref.]
  32. Oshiro I, Tsutsui M, Fujii M, Ueyama M, Takenaka T, et al. (1981) [new manual and automatic method of hemoglobin determination by using SLS (author’s transl)]. Rinsho Byori 29: 203-209. [Ref.]
  33. Kimura S, Iyama S, Yamaguchi Y, Hayashi S, Yanagihara T (1999) Enzymatic assay for conjugated bilirubin (Bc) in serum using bilirubin oxidase (BOD). J Clin Lab Anal 13: 219- 223. [Ref.]
  34. Muramoto Y, Matsushita M, Irino T (1999) Reduction of reaction differences between human mercaptalbumin and human nonmercaptalbumin measured by the bromcresol purple method. Clin Chim Acta 289: 69-78. [Ref.]
  35. Gitzelmann-Cumarasamy N, Gitzelmann R, Wilson KJ, Kuenzle CC (1979) Fetal and adult albumins are indistinguishable by immunological and physicochemical criteria. Proc Natl Acad Sci USA 76: 2960-2963. [Ref.]
  36. Suh B, Wadsworth SJ, Lichtenwalner DM (1987) Demonstration of 2- hydroxybenzoylglycine as a drug binding inhibitor in newborn infants. J Clin Invest 80: 1125-1131. [Ref.]
  37. Ahlfors CE (1981) Effect of serum dilution on apparent unbound bilirubin concentration as measured by the peroxidase method. Clin Chem 27: 692-696. [Ref.]

Download Provisional PDF Here


Article Information

Article Type: Research Article

Citation: Morioka I, Kurokawa D, Iwatani S, Iijima K, Nakamura H, et al. (2018) Monitoring Bilirubin Binding Parameters in a Cohort of Japanese Neonates. J Clin Lab Med 3(1): dx.doi. org/10.16966/2572-9578.119

Copyright: © 2018 Morioka I, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Publication history: 

  • Received date: 02 Apr, 2018

  • Accepted date: 26 Apr, 2018

  • Published date: 03 May, 2018