Measurement of unbound bilirubin by the peroxidase test using Zone Fluidics

Results The CV for unbound bilirubin in the various controls ranged from 11% to 38% (within day) and 12% to 27% (between days). Triplicate CV measurements for newborn plasma measurements ranged from 0.6% to 31% (mean 11%, n = 47). Mean unbound bilirubin by Zone Fluidics was 5-fold higher than that by the commercial method. Conclusion Zone Fluidics can be used to automate the peroxidase test and overcome many of the limitations of the commercially available peroxidase technology. Abbreviations TBC plasma total bilirubin concentration B f plasma unbound (non-albumin bound or “free”) bilirubin Kp rate constant for peroxidase catalyzed oxidation of bilirubin by hydrogen peroxide D dilution of stock peroxidase used for sample analysis R1 phosphate buffer R2 phosphate buffer containing horseradish peroxidase and glucose oxidase R3 phosphate buffer containing glucose and peroxide A460 light absorbance at 460 nm ɛ · b bilirubin extinction coefficient at 460 nm × light path length k 1 bilirubin–albumin dissociation rate constant k 2 bilirubin–albumin association rate constant K equilibrium association constant ( k 2 / k 1 ) Keywords Zone fluidics Peroxidase test Unbound bilirubin Jaundice Newborn Bilirubin binding 1 Introduction The peroxidase test is used to measure the plasma unbound (non-protein bound) or “free” bilirubin ( B f ) in jaundiced newborns [1] . B f is the tiny but extremely important fraction of the plasma total bilirubin concentration (TBC) that plays a fundamental role in the pathogenesis of bilirubin encephalopathy (kernicterus) [2–10] , and recent clinical studies support its use in these patients [5–7,9,10] . B f measurements, however, are rarely offered by clinical laboratories even though the peroxidase test was described for over 30 years ago [1] , and a Federal Drug Administration-cleared commercial instrument is available [11] . We have used the approved peroxidase technology over several years to measure TBC and B f in jaundiced newborns in a neonatal clinical laboratory. Although the test has proved very useful in helping to determine which babies with very high TBCs need treatment [9,12] , the commercial technology has significant shortcomings. These include manual handling of reagents, underestimation of B f unless the results are confirmed by measuring B f at additional peroxidase concentrations [12–15] , and, most importantly, at the 42-fold sample dilution used, intrinsic bilirubin–albumin binding is altered [15,16] and the effects of weak bilirubin binding competitors are attenuated [8,14,17,18] . Iatrogenic kernicterus has occurred in jaundiced newborns following administration of drugs such as sulfisoxazole [19] and benzyl alcohol [17] that are weak bilirubin binding competitors. Even today, drugs interfering with bilirubin–albumin binding such as ibuprofen, which is used to close the patent ductus arteriosus in premature newborns [18] , are used in neonatal practice. Unfortunately, the increase in B f produced by drugs like sulfisoxazole or ibuprofen may be substantially less at the 42-fold sample dilution used for the current peroxidase technology [14,18] . Since B f measurements are desirable in jaundiced newborns, we attempted to make the peroxidase test more reliable for routine clinical laboratory use. Zone Fluidics [20] , a recent extension and expansion of sequential injection analysis [21] , automates complex manual wet chemical procedures in a rapid, precise, and efficient manner using small reagent and sample volumes as well as minimal sample dilution and seemed well suited for the peroxidase test. We therefore investigated the feasibility of performing the peroxidase test using Zone Fluidics. 2 Materials and methods 2.1 Overview of the current peroxidase technology The peroxidase test measures TBC and B f using the light absorbance of bilirubin at 460 nm (A460) corrected for hemoglobin interference [1,11] . The UB Analyzer (Arrows Company, LTD., Osaka, Japan) measures TBC and B f after manually combining 25 μl of sample with 1.025 ml of reagents. The 42-fold sample dilution is necessary to obtain acceptable A460 readings in the 1 cm path length cells typically used for the test [1,11] . Conjugated bilirubin interferes with the test [14] , but the concentration of conjugated bilirubin in newborn plasma is usually so low that the American Academy of Pediatrics does not recommend routine fractionation of the TBC into conjugated and unconjugated bilirubin in these patients [22] . The TBC is calculated from the initial A460 and B f from the decrease in A460 (i.e., the decrease in TBC) over time that occurs following the addition of horseradish peroxidase and peroxide, which oxidizes B f to colorless compounds [1,11] . B f is calculated from the measured change in TBC over time, the concentration of peroxidase used (HRP), and the rate constant (Kp) for the peroxidase catalyzed peroxide oxidation of bilirubin [1] : (1) dTBC d ⁢ t = Kp · HRP · B f . The “gold standard” B f control used to determine the peroxidase Kp is an albumin-free bilirubin solution in which TBC = B f [1] . It has been argued that Kp measurements are unreliable at the TBC ( B f ) and pH typically used (≈ 3 μmol/l and 7.4, respectively) because bilirubin aggregates and becomes insoluble under these conditions. Ahlfors [15] , however, showed that the Kp is the same at TBCs as low as 0.05 μmol/l, which is well within the range of B f values observed in clinical samples. Bilirubin–albumin binding measured using the peroxidase method has been validated using stopped-flow studies [13] . In addition, the bilirubin–albumin binding constant ( K ; see Eq. (2) below) for defatted albumin obtained using the peroxidase method [13,15] is similar to that obtained using fluorescence [23] and ultrafiltration [16] . However, since Kp measurements are typically made using reaction conditions that are substantially different from those used for clinical samples [1] , the Arrows Company provides a TBC and B f control containing albumin (TBC = 18.2 mg/dl, B f = 0.66 μg/dl) specifically for quality control use with the UB Analyzer. The B f measured by the UB Analyzer may seriously underestimate the B f for 2 reasons. The first is that B f is measured at a 42-fold dilution, and sample dilution increases intrinsic bilirubin–albumin binding (i.e., K increases) [14–16] and also attenuates the effects of any weakly binding bilirubin competitors (e.g., ibuprofen) present [17,18] . The second reason is that a single peroxidase concentration is used, and the measured B f may be much lower than the equilibrium B f ( B feq ) for the following reasons [13–15] . B feq is the mass action function of the albumin concentration, the bilirubin–albumin dissociation and association rate constants ( k 1 and k 2 , respectively, K = k 2 / k 1 ), and the TBC (actually, the protein-bound bilirubin, which is nearly equal to the TBC): (2) B feq = k 1 · TBC k 2 · ( Albumin − TBC ) = TBC K · ( Albumin − TBC ) . Since at equilibrium, d B feq d ⁢ t = ( k 1 · TBC ) − { k 2 · ( A − TBC ) · B feq } = 0 , when bilirubin oxidation begins, B feq instantaneously decreases to B fss [12,13] given by: d B fss d ⁢ t = ( k 1 · TBC ) − { k 2 · ( A − TBC ) · B fss } − ( Kp · HRP · B fss ) . B fss is the apparent B feq that is measured by the peroxidase test [13,15] , but B fss will be nearly equal to B feq only if k 1 TBC is extremely fast compared to − Kp·HRP· B fss [13] . Unfortunately, this condition rarely holds at the HRP concentration used in the test as performed by the UB Analyzer [12] . Fortunately, B feq can be obtained by measuring B fss again at a different HRP concentration (we routinely use half the recommended HRP). B feq is the reciprocal of the y intercept of a plot of 1 / B fss versus the corresponding HRP concentrations [12,13] : (3) 1 B fss ∝ HRP + 1 B feq . 2.2 Adaptation of the peroxidase test to Zone Fluidics A Zone Fluidics system (Global FloPro, Global FIA Inc, Fox Island, WA) as shown in Fig. 1 was constructed for automated combination and then delivery of peroxidase test reactants (reagents and sample) to a temperature controlled spectrophotometer flow cell for measurement of TBC and B f from the A460. The system uses perfluoroalkoxy tubing (≈ 0.7 mm ID) to connect in series a reservoir of water containing 60 μl/100 ml of the surfactant Zonyl FSN (DuPont, Wilmington, DE) to prevent air bubbles from forming along the walls of the tubing, a bi-directional stepper syringe pump to move the water–surfactant solution and reactants through the tubing, a 10-port selection valve (Valco Instruments, Houston, TX) to which vials of the reactants are attached for sequential aspiration into the tubing between the pump and valve, a spectrophotometer flow cell with spectrophotometer (Ocean Optics, Orlando, FL), and a waste reservoir. The tubing from the selection valve to the spectrophotometer flow cell passes through the center of a 9 × 2.5 cm (length × diameter) cylindrical aluminum heater maintained at 37 °C, which warms the reactants to 37 °C in about 4 s. A small aperture through the heater 1.5 cm from the detector end of the heater and perpendicular to the tubing direction collimates the spectrophotometer's light source (a white LED ground to remove the lens effect) and detector. Pump speeds of 50 μl/s were used to aspirate and dispense water–surfactant and 5 μl/s to aspirate and dispense reactants through the tubing. Vials containing the three peroxidase reagents and sample are connected via the tubing to individual valve ports, one port is open to air, and one port is used for a 0.1 mol/l NaOH solution to clean the tubing between samples ( Fig. 1 ). The air and reactants are sequentially aspirated into the zone assembly area in the tubing between the pump and valve to form an air–reactant–air “zone” in the tubing ( Fig. 1 ). Once assembled, the zone is advanced to the spectrophotometer flow cell for measurement of TBC and B f . Valve, pump, and spectrometer operations were programmed using FIAlab software (FIAlab, Seattle, WA). Three peroxidase test reagents, designated R1, R2, and R3, were used. R1 is 0.2 mol/l phosphate buffer, pH 7.4 containing 0.2 mmol/l NaCl to maintain physiologic levels of Cl − during B f measurements [16,24] , R2 is R1 containing HRP (EC 1.11.17) as well as glucose oxidase (EC 1.1.3.4), and R3 is R1 containing glucose and hydrogen peroxide. R2 was made by diluting the stock HRP/glucose oxidase reagent for the UB Analyzer 8-fold or 16-fold with R1. The stock HRP/glucose oxidase diluted 24,562-fold at 37 °C had a mean Kp of 0.019 s − 1 (S.D. 0.001, n = 16) as determined using a HP 8452 computer directed spectrophotometer [1] . R3 was made by adding 400 mg of d -glucose and 68 μl of 3% (w/v) H 2 O 2 to 100 ml of R1 (glucose 400 mg/dl, H 2 O 2 600 μmol/l). The peroxide generated from glucose by glucose oxidase prevents peroxide from becoming rate limiting as bilirubin is oxidized. 2.2.1 TBC and B f measurements A sample aliquot of 8 μl was used for each TBC and B fss determination. Since B fss was measured at up to four HRP concentrations, a total sample volume of 45 μl was used. All measurements were made at 37 °C, a 2-fold sample dilution, a reactant volume of 16 μl and a final phosphate buffer concentration of 0.1 mol/l. At the 2-fold sample dilution used intrinsic bilirubin–albumin binding is minimally perturbed [14–16] . The TBC zone was created by sequentially aspirating 10 μl of air, 4 μl R3, 0 μl of R2, 8 μl of sample, 4 μl of R1, and 10 μl of air ( Fig. 1 ). The zone was advanced to the spectrophotometer where the absorbance was recorded twice per s for 10 s at 460 nm, 578 nm to correct for hemoglobin interference [13] , and 688 nm to correct for baseline shifts in absorbance. The TBC was calculated by dividing the corrected A460 by the product of the extinction coefficient ( ɛ ) and path length ( b ). The product ɛ · b was obtained using commercial TBC controls (see below). B fss was typically measured at four HRP concentrations and B feq calculated using Eq. (3) above. The 4 HRP concentrations were obtained by combining 0, 1, 2, or 3 μl of R1 with 4, 3, 2, or 1 μl R2, respectively to obtain a total R1 + R2 volume of 4 μl. The corresponding B fss measurement is designated as B f1 , B f2 , B f3 , or B f4 according to the volume of R2 used. The zones were assembled as described for the TBC using 4 μl of reagent R3, 8 μl of sample and 4 μl of R2 + R1. The B fss at each HRP was calculated from the slope of ln(A460) versus the time needed for a 5% to 20% decrease in TBC, the sample dilution (0.5), the Kp, the dilution of stock HRP/glucose oxidase at which B fss was measured ( D ), and the dilution of stock peroxidase at which the Kp was determined (24,562) [1] : B fss = − 0.5 · TBC · slope · D Kp · 24 , 562 . D ranged from 32 to 256. B feq was calculated as the reciprocal of the y intercept obtained by regressing 1 / B fss versus the corresponding volume of HRP (see Eq. (3) ). 2.3 Calibration and controls 2.3.1 Total bilirubin controls and calibration Commercially available TBC controls (19.6 and 5.3 mg/dl, Sigma; 20 mg/dl Bio-Rad Laboratories Liquichek Pediatric Control, Hercules, CA) were used to calibrate the Global FloPro for TBCs from 0 to 40 mg/dl. The Sigma controls are dissolved in water and the Bio-Rad control is already in solution. The product ɛ · b was obtained by dividing the A460 by the control's TBC. The Bio-Rad control was also used to determine the completeness of zone reactant mixing and to evaluate the imprecision of TBC and B f measurements as explained below. The TBCs of the various controls were verified on the UB Analyzer as well as an HP 8452 computer directed spectrophotometer. 2.3.2 Unbound bilirubin controls As noted previously, the “gold standard” control solution for the peroxidase test is an albumin-free solution of bilirubin that is used to determine the Kp for peroxidase catalyzed peroxide oxidation of bilirubin [1] . We studied the imprecision of the B f measurement using the UB Analyzer control (TBC = 18.2 mg/dl, B f = 0.66 μg/dl at a 42-fold dilution), recognizing that the control could not be used to assess B f accuracy because the zone fluidics system measures B f at a 2-fold sample dilution where the intrinsic bilirubin–albumin binding is likely to be less (i.e., K is lower) than at the 42-fold dilution used by the UB Analyzer. Since B f control solutions containing bilirubin and human albumin would be a more convenient and familiar means of quality control [25] , we prepared human defatted human albumin–bilirubin “controls” to evaluate imprecision and to compare K calculated from the measured TBC, B f , and albumin using Eq. (2) with published values for K determined by other methods [16,23] . These “controls” were made by dissolving bilirubin (Sigma) in 0.1 mol/l NaOH and adding aliquots to a weighed amount of defatted human albumin (Sigma). Water and phosphate buffer (final phosphate 0.04 mol/l, pH 7.4) were then added to obtain controls with TBCs of about 3, 12, and 25 mg/dl with albumin concentrations of about 4 g/dl. The final albumin concentration was verified by the BCP method [26] . The controls were freeze dried in 100 μl aliquots, back-filled with nitrogen, and stored at 8 °C in the dark until reconstituted with 100 μl of water. We also studied the impact of sulfisoxazole on the “controls” by reconstituting with 100 μl water containing 1 mmol/l of added sulfisoxazole, which should about double the B f concentrations [8] . A K value for bilirubin-defatted albumin binding between about 0.3 and 1.0 × 10 7 l/mol at 37 °C is expected based on published data from the peroxidase test itself [15] and other methods measuring K at similar albumin concentrations [16,23] . A range of values is given because some variation in K between defatted albumin preparations is unavoidable due to albumin polymorphism [27] , minor impurities and denaturation during preparation [28] , and the influence of secondary site bilirubin binding [29] . 2.4 Clinical samples Plasma TBC and B f were measured by the UB Analyzer (42-fold sample dilution, 2 peroxidase concentrations) and the Global FloPro Zone Fluidics system (2-fold dilution, 4 peroxidase concentrations) in 15 jaundiced premature newborns participating in a research study of the impact of a drug on bilirubin–albumin binding in jaundiced premature newborns. Plasma was separated from blood samples drawn from indwelling arterial catheters and stored at − 80 °C until shipped overnight on dry ice for TBC and B f were measurements, which were made within 1 h of thawing. Albumin was measured by the clinical laboratory of Antwerp University Hospital using the BCP method on a Vitro-950 (Ortho Clinical Diagnostics). The research study was approved by the Institutional Review Board at Antwerp University Hospital, Belgium. 3 Results 3.1 Calibration and measurements Although the water–surfactant mixture does not come into direct contact with the reactants, preliminary studies showed that zonyl FSN had no effect on bilirubin–albumin binding. The value of ɛ · b was 2.45 × 10 − 3 l/μmol ( n = 20, CV = 0.87%) or 4.19 × 10 − 2 dl/mg (4.19 × 10 − 5 dl/μg) at TBCs between 1 and 20 mg/dl. Above 20 mg/dl, the increase in A460 was proportionally less with each incremental increase in TBC. 3.2 Zone reactant mixing The completeness of reactant mixing was assessed using the Bio-Rad control and substituting R3 (buffer) for R2 to avoid oxidation of bilirubin during the measurements. The assembled zones when moved slowly past the detector did not show a significant increase or decrease in A460 as would occur if mixing were incomplete. In addition, the various reactant volume combinations used for each B fss measurement ( Fig. 1 ) gave A460 readings that were not significantly different. 3.3 Imprecision of TBC and B f zone fluidics measurements (controls) 3.3.1 Bio-rad control Four studies of the imprecision of 5 replicate TBC and apparent B feq determinations at 4 peroxidase concentrations were performed. Similar results were obtained in each of the 4 studies, and the results of one study are given in Table 1 . 3.3.2 Arrows, Ltd. UB Analyzer control The mean measured TBC in the Arrows control of 18.9 mg/dl (S.D. 0.7, range 17.6 to 19.9, CV 3%, n = 10) was not significantly different from that stated for the control (18.2 mg/dl, CV = 2%). However the mean measured B feq of 1.42 μg/dl (S.D. 0.55, range 0.71 to 2.4, CV 38%, n = 10) with a between day CV of 18% ( n = 5) was, as expected, significantly higher than that stated for the control at the 42-fold sample dilution (0.66 μg/dl, CV 5%). 3.3.3 Bilirubin-defatted human albumin controls These controls did not show any significant deterioration in TBC or B f over 4 months. The low bilirubin-defatted albumin control had a mean TBC of 3.4 mg/dl (S.D. 0.4, range 3.2 to 3.6, CV 4%, n = 10), a mean B feq of 1.04 μg/dl (S.D. 0.4, range 0.53 to 1.65, CV 38%, n = 10), and a B feq between day CV of 27% ( n = 5). The mean K of 0.6 × 10 7 l/mol as calculated from the mean TBC of 58.1 μmol/l, mean B f of 0.018 μmol/l, and defatted albumin of 605 μmol/l) was within the range expected. The mid range bilirubin-defatted albumin control had a mean TBC of 11.9 mg/dl (S.D. 0.5, range 10.9 to 12.7, CV 4%, n = 10), a mean B feq of 2.93 μg/dl (S.D. 0.68, range 1.65 to 3.96, CV 23%, n = 10), and B feq between day CV of 12% ( n = 6). The mean K of 1.0 × 10 7 L/mol as calculated from the mean TBC of 203 μmol/l, mean B f of 0.050 μmol/l, and defatted albumin of 605 μmol/l was within the range expected. The high bilirubin-defatted albumin control had a mean TBC of 24.9 mg/dl (S.D. 0.8, range 23.4 to 25.6, CV 3%, n = 10), a mean B feq of 14.2 μg/dl (S.D. 1.6, range 12.2 to 16.7, CV 11%, n = 10), with a B feq between day CV of 12.7% ( n = 3). The mean K of 1.0 × 10 7 l/mol as calculated from the mean TBC of 426 μmol/l, mean B f of 0.242 μmol/l, and defatted albumin of 605 μmol/l was within the range expected. 3.3.4 Effect of sulfisoxazole on bilirubin-defatted albumin controls Sulfisoxazole (500 μmol/l in the reaction), which displaces bilirubin from albumin and has been associated with iatrogenic kernicterus [8,19] increased the B feq in the low, mid range, and high controls to 3.75 μg/dl (275%), 8.50 μg/dl (193%), and 24.5 μg/dl (72%), respectively. However, B f in the mid level control increased only 28% (S.D. 0.8, n = 5) when measured at a 42-fold sample dilution using the UB Analyzer. 3.4 TBC and B f in clinical samples Forty-seven triplicate B fe measurements at 4 peroxidase concentrations (D ranged from 32 to 128) were made using the Global FloPro Analyzer in plasma samples from 15 premature newborns. The CV for triplicate B fe determinations varied from 0.6% to 30.8% (mean 11.1%). The mean TBC was 7.6 mg/dl (S.D. 2.0, range 3.0 to 11.6) and the mean B feq was 4.7 μg/dl (S.D. 5.0, range 0.5 to 21.5). While there was no significant difference in the mean TBC of the samples measured by the UB Analyzer (mean TBC 7.5 mg/dl, S.D. 2.2, range 2.8 to 12.3), the mean B feq was significantly lower at 1.0 μg/dl, S.D. 0.80, range 0.08 to 3.3 ( p < 0.001, paired t -test). A 10-fold range of B feq values occurred over the 4-fold range of TBC values obtained using zone fluidics. This indicates significant variability in binding (i.e., K ) between babies. A binding isotherm of the data obtained by plotting the TBC/albumin molar ratio against B fe [30] gave a best fit value of K determined by non-linear regression for newborn plasma of 1.6 × 10 7 L/mol, similar to the K values obtained for the defatted-albumin controls. 4 Discussion The risk of bilirubin toxicity in the jaundiced newborn remains a significant clinical problem [31] , and both the National Institute of Child Health and Human Development [32,33] and the American Academy of Pediatrics [22] in recent reviews of newborn jaundice mention the need for further exploration of bilirubin–albumin binding measurements in the clinical management of these patients. There is clinical evidence supporting B f measurements by the peroxidase test [5–7,9] , and the test seems better suited for clinical laboratory use than other bilirubin–albumin binding tests that have been developed [3] . Adapting the peroxidase test to Zone Fluidics improves the commercial peroxidase method in that reactant handling is automated, B f is measured using small sample volumes (45 μl), and minimal sample dilution (2-fold) and multiple peroxidase concentrations ensure that B feq is not underestimated. The use of multiple peroxidase concentrations adds to the imprecision of the B f measurement as each B fss measurements contributes to imprecision. Given a CV for each B fss measurement of about 8 or 9%, a CV for replicate B feq measurements in the 32% range would not be unexpected. The mean within-day CV of 11% for the clinical samples was obtained over a fairly narrow range of TBCs, and additional studies of imprecision over the wider range of TBCs encountered clinically are needed. In addition, a substantial error in one B fss measurement may have an inordinate effect on the calculated B feq (Eq. (3) ), and other approaches such as duplicate B fss measurements at two or three HRP levels rather that single B fss measurements at four HRP levels may better detect this type of error and should be evaluated. Dilutional changes in intrinsic bilirubin–albumin binding are minimal at the 2-fold sample dilution used by Zone Fluidics [14,16] , and the effect of weak bilirubin binding competitors are readily detected as seen in our sulfisoxazole results. While sulfisoxazole increased the apparent B feq by about 200% at the 2-fold sample dilution, the increase was only 28% at the 42-fold dilution used by the UB Analyzer. It is important that future studies of B f measurements in jaundiced newborns be performed in minimally perturbed plasma to avoid test conditions that may significantly alter bilirubin–albumin binding. The general problem of bilirubin controls remains a serious issue for clinical laboratories [25] . Fortunately, HRP is readily “calibrated” in albumin free solutions of bilirubin, which provide a reliable “gold standard” control for the peroxidase test. The validity of bilirubin binding measurements obtained using the peroxidase system has been well documented [13,15,16,23] . However, determining the Kp every 8 h for routine quality control would require different instrumentation and an additional layer of quality control making the test inconvenient in the typical clinical laboratory. The bilirubin-defatted albumin “controls” we prepared hold promise as an alternative means for peroxidase quality control assessment that warrants further study. Hemoglobin [34] , conjugated bilirubin [14] , bilirubin isomers [35,36] , and compounds like propyl paraben that alter peroxidase activity but not B f [37] may interfere with the peroxidase test. Hemoglobin levels can be assessed using the A578 absorbance measurement, but excessive hemolysis interferes with the test. Interference from conjugated bilirubin can be overcome by modifying the method [14] , and further studies are needed to assess the effects of bilirubin isomers. Other bilirubin–albumin binding tests have been used clinically [38] , and any clinical studies of the peroxidase test should have other binding tests available to confirm unexpected results or excessive B f levels. Population studies are needed to provide the range of B f values occurring in a large, diverse sampling of jaundiced newborns [39] . The peroxidase test performed using Zone Fluidics provides an automated means for obtaining these data using inexpensive reagents, small sample volumes, and minimal sample dilution. Should these data prove clinically useful, then the peroxidase test performed using Zone Fluidics would be more suitable for routine clinical laboratory use than the current peroxidase technology. References [1] J. Jacobsen R.P. Wennberg Determination of unbound bilirubin in the serum of newborns Clin Chem 20 1974 783 789 [2] C.E. Ahlfors R.P. Wennberg Bilirubin–albumin binding and neonatal jaundice Semin Perinatol 28 2004 334 339 [3] Wennberg RP, Ahlfors CE, Bhutani VK, Johnson LH, Shapiro SM. Toward understanding kernicterus: a challenge to improve the management of jaundiced newborns. Pediatrics in press. [4] J.D. Ostrow L. Pacolo S.M. Shapiro C. Tiribelli New concepts in bilirubin encephalopathy Eur J Clin Invest 33 2003 988 997 [5] H. Nakamura M. Yonetani Y. Uetani M. Funato Y. Lee Determination of serum unbound bilirubin for prediction of kernicterus in low birth weight infants Acta Paediatr Jpn 34 1992 642 647 [6] M. Funato H. Tamai S. Shimada H. Nakamura Vigintiphobia, unbound bilirubin, and auditory brainstem responses Pediatrics 93 1994 50 53 [7] S.B. Amin C. Ahlfors M.S. Orlando L.E. Dalzell K.S. Merle R. Guillet Bilirubin and serial auditory brainstem responses in premature infants Pediatrics 107 2001 664 670 [8] C.E. Ahlfors Unbound bilirubin associated with kernicterus: a historical perspective J Pediatr 137 2000 540 544 [9] C.E. Ahlfors O. Herbsman Unbound bilirubin in a term newborn with kernicterus Pediatrics 111 2003 1110 1112 [10] H. Ihara N. Hashizume N. Shimizu T. Aoki Threshold concentration of unbound bilirubin to induce neurological deficits in a patient with type I Crigler–Najjar syndrome Ann Clin Biochem 36 1999 347 352 [11] R. Shimabuku H. Nakamura Total and unbound bilirubin determined using an automated peroxidase micromethod Kobe J Med Sci 28 1982 91 104 [12] Ahlfors CE, Parker AE. Evaluation of a model for brain bilirubin uptake in jaundiced newborns. Pediatr Res, in press. [13] T. Faerch J. Jacobsen Determination of the association and dissociation rate constants for bilirubin–bovine serum albumin Arch Biochem Biophys 168 1975 351 357 [14] C.E. Ahlfors Measurement of plasma unbound unconjugated bilirubin Anal Biochem 279 2000 130 135 [15] C.E. Ahlfors Effect of serum dilution on apparent unbound bilirubin concentration as measured by the peroxidase method Clin Chem 27 1981 692 696 [16] R.A. Weisiger J.D. Ostrow R.K. Koehler Affinity of human serum albumin for bilirubin varies with albumin concentrations and buffer composition: results of a novel approach J Biol Chem 276 2001 29953 29960 [17] C.E. Ahlfors Benzyl alcohol, kernicterus, and unbound bilirubin J Pediatr 139 2001 317 319 [18] C.E. Ahlfors Effect of ibuprofen on bilirubin–albumin binding J Pediatr 144 2004 386 388 [19] W.A. Silverman D.H. Andersen W.A. Blanc D.N. Crozier A difference in mortality rate and incidence of kernicterus among premature infants allotted to two prophylactic antibacterial regimens Pediatrics 18 1956 614 624 [20] G.D. Marshall D.K. Wolcott D.C. Olson Zone Fluidics in flow analysis: potentialities and applications Anal Chim Acta 499 2003 29 40 [21] J. Ruzicka G.D. Marshall Sequential injection: a new concept for clinical sensors, process analysis, and laboratory assays Anal Chim Acta 237 1990 329 343 [22] American Academy of Pediatrics Clinical practice guideline. Subcommittee on Hyperbilirubinemia Pediatrics 114 2004 297 316 [23] A.A. Lamola J. Eisinger W.E. Blumberg S.C. Patel J. Flores Fluorometric 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 1979 25 42 [24] J. Jacobsen Studies of the affinity of human serum albumin for binding of bilirubin at different temperatures and ionic strength Int J Pept Protein Res 9 1977 235 239 [25] S.F. Lo B.T. Doumas E.R. Ashwood Performance of bilirubin determinations in US laboratories—revisited Clin Chem 50 2004 190 194 [26] A.E. Pinnell B.E. Northam New automated dye-binding method for serum albumin determined with bromcresol purple Clin Chem 24 1978 80 86 [27] F.W. Lorey C.E. Ahlfors D.G. Smith J.V. Neel Bilirubin binding by variant albumins in Yanomama Indians Am J Hum Genet 36 1984 1112 1120 [28] R. Brodersen Competitive binding of bilirubin and drugs to human serum albumin studied by enzymatic oxidation J Clin Invest 54 1974 1353 1364 [29] J. Jacobsen Binding of bilirubin to human serum albumin-determination of the dissociation constants FEBS Lett 5 1969 112 114 [30] I.M. Klotz D.L. Hunston Protein affinities for small molecules. Conceptions and misconceptions Arch Biochem Biophys 193 1979 314 328 [31] Sentinel event alert: kernicterus threatens healthy newborns Jt Comm Perspect 21 2001 10 11 [32] L.R. Blackmon A.A. Fanaroff T.N.K. Raju Research on prevention of bilirubin-induced brain injury and kernicterus: national institute of child health and human development conference and executive summary Pediatrics 114 2004 229 233 [33] D.K. Stevenson R.J. Wong H.J. Vreman A.F. McDonagh M.J. Maisels D.A. Lightner NICHD conference on kernicterus: research on prevention of bilirubin-induced brain injury and kernicterus: bench to bedside—Diagnostic methods and prevention and treatment strategies J Perinatol 24 2004 521 525 [34] W.B. de Carvalho B.I. Kopelman P.S. de Araujo Correlation between free bilirubin and indirect bilirubin in normal newborn infants with non-hemolytic jaundice and effect of hemolysis on free bilirubin measurement by the peroxidase method Rev Paul Med 110 1992 138 144 [35] S. Itoh T. Yamakawa S. Onishi K. Isobe M. Mababe K. Sasaki The effect of bilirubin photoisomers on unbound bilirubin concentrations estimated by the peroxidase method Biochem J 239 1986 417 421 [36] S. Itoh K. Kawada T. Kusaka Influence of glucuronosyl bilirubin and (EA)-cyclobilirubin on determination of serum unbound bilirubin by UB-analyser Ann Clin Biochem 30 2002 583 588 [37] L.F. Rasmussen C.E. Ahlfors R.P. Wennberg The effect of paraben preservatives on albumin binding of bilirubin J Pediatr 89 1976 475 478 [38] J. Kapitulnik T. Valaes N.A. Kaufmann S.H. Blondheim Clinical evaluation of Sephadex gel filtration in estimation of bilirubin binding in serum in neonatal jaundice Arch Dis Child 49 1974 886 894 [39] J.A. Lott L.C. Mitchell M.L. Moeschberger D.E. Sutherland Estimation of reference ranges: how many subjects are needed? Clin Chem 38 1992 648 650