Bilirubin Toxicity to Human Erythrocytes: A More Sanguine View

In 1962 Watson1 observed that a small fraction of the bilirubin in blood from jaundiced newborns is associated with red cells. A few years later, Cheung et al2 showed that red cells undergo time- and concentration-dependent changes in morphology and metabolism, with eventual hemolysis, when incubated with bilirubin in buffer, particularly at concentrations of >340 μM (20 mg/dL). (In retrospect, it seems highly likely that the bilirubin solutions to which the cells were exposed were supersaturated and may have contained aggregated forms of bilirubin.3–7) The bilirubin-induced morphologic changes were illustrated by striking photographs similar to those published some 30 years later, without reference to the earlier work, by others.8,9 These changes were reversible by addition of serum albumin to the incubation medium and did not occur when albumin was present in the medium from the start. Subsequently, Brito et al10 have confirmed and extended these observations and recently reviewed the toxicity of bilirubin to red cells (again without alluding to the seminal earlier work). For the neonatologist, an important question is whether bilirubin-induced hemolysis is clinically significant in neonatal jaundice or in other unconjugated hyperbilirubinemias. The review implies that it is.10 There is, however, much evidence that it is not: The hypothesis that bilirubin might induce erythrocyte hemolysis is based mainly on studies of the cytotoxicity of bilirubin in vitro. In most of these studies, the pigment was presented to cells as a solution in buffer or culture media in the presence or absence of albumin or other proteins. Because crystalline bilirubin does not dissolve in these solvents, the solutions are invariably made by adding a concentrated solution of bilirubin in strong alkali at a pH of ∼13 (or sometimes in dimethyl sulfoxide or ethanolamine) to the medium, often followed by neutralization with strong acid. It is generally assumed that the structure and properties of bilirubin in the final solutions are identical to those of bilirubin as it is presented to erythrocytes or other cells in vivo. However, that assumption may not always be valid because of the ability of bilirubin to form albumin complexes of different stoichiometry and stereochemistry; its conformational lability and ability to form hydrogen-bonded complexes; and its well-known tendency to aggregate.34–36 Absent from most in vitro studies of bilirubin toxicity are data on the structure, molecularity, and stability of bilirubin in the media used, in both the presence and absence of cells. The preponderance of studies in which significant toxic effects have been observed, including those using displacing agents, were performed under nonphysiologic conditions in which aggregates of bilirubin may have been present. There is, however, no evidence for the presence of bilirubin aggregates in neonatal or Crigler-Najjar red cells or plasma, even at high bilirubin concentrations, and evidence that they are unlikely to occur.34,35 The striking bilirubin-induced erythrocyte damage graphed in Fig 1 of the recent review10 becomes less impressive from a clinical point of view when it is recognized that it was obtained by exposure of the cells to concentrations of bilirubin that were orders of magnitude greater than the free bilirubin concentrations to which red cells are exposed in jaundiced infants. Also, it would not have occurred if physiologic concentrations of albumin had been present in the medium.2 In addition, because the aqueous bilirubin solutions to which the red cells were exposed were most likely supersaturated, their admixture with cells may well have seeded nucleation, aggregation, and precipitation of pigment, resulting in nonphysiologic multiphasic mixtures.3,4,6,37 Few studies of the toxic effects of bilirubin on red cells have been performed in whole blood. Some have been performed with washed red cells reconstituted to a hematocrit level of 10%. It seems not to have been recognized in those studies that the total load of bilirubin presented to the cells for the same plasma concentration of bilirubin is much greater at a hematocrit level of 10% than at the normal ∼45%. Petrich et al38 reported that bilirubin, added to heparinized blood from an adult at a concentration of only 70 to 80 μM (∼4–5 mg/dL), induced hemolysis after 4 hours at 37°C. However, that lone observation is inconsistent with other observations on red cells in vitro, in which no hemolysis was observed at even higher bilirubin concentrations in the presence of serum albumin.2,8 Red cells from just 1 adult were used in the studies, and the possible interference of bilirubin in the spectroscopic method used to estimate hemolysis, a phenomenon noted by Watson,1 was not checked.38From the published clinical and in vitro evidence, with the exception of the article noted above,38 it is hard to escape the conclusion that bilirubin-induced hemolysis is insignificant in vivo even at plasma bilirubin concentrations associated with kernicterus or at levels greater than those at which therapeutic intervention is considered mandatory. There is even evidence that bilirubin can strengthen the red cell membrane and protect red cells from oxidative damage that could lead to hemolysis.8,15,16,39 Of course, it is hard to substantiate a negative. Documentation of mild hemolysis in newborns is difficult, and even a small degree of hemolysis, insufficient to have much effect on hematocrit measurements, could markedly augment the circulating bilirubin pool.40–42 Necheles et al43 and, more recently, Maisels and Kring41 have shown that increased heme catabolism is important in the development of hyperbilirubinemia in normal infants in the first few days after birth. The cause of this is unknown and merits additional investigation, but it would be premature to ascribe it to bilirubin-induced hemolysis.The hypothesis that binding of bilirubin might predispose red cells of neonates to hemolysis or phagocytosis is certainly plausible and might seem to be supported by some in vitro studies.10 It is unquestionable that bilirubin can partition into cell membranes, modifying their properties. However, rash extrapolations from in vitro studies to real life have previously muddied the waters of bilirubin metabolism and led, for example, to suggestions that photodegradation of bilirubin is the most important pathway in phototherapy and that photohemolysis might be a significant adverse effect of the treatment, suggestions now thought to be incorrect.44,45 It seems to have gone unnoticed that bilirubin has the characteristics of a promiscuous inhibitor.46–48 Promiscuous inhibitors are compounds that lead to false-positive results in the high-throughput screening of potential drugs in vitro. Like bilirubin, promiscuous inhibitors are often hydrophobic molecules with a tendency to aggregate, and aggregation is often the cause of their nonphysiologic promiscuous in vitro inhibition. Although the toxic effects of bilirubin on erythrocytes and other cells in vitro may be scientifically interesting with respect to the chemical properties of bilirubin and nonspecific effects of lipophiles and aggregated molecules on membranes, stronger evidence than that recently presented10 or currently available is needed to support the hypothesis that bilirubin-induced damage to red cells is ever sufficient to cause clinical concern.This work was supported by National Institutes of Health grant DK-26307 (to Dr McDonagh).