Neonatal jaundice is one of the first and perhaps the most common problem encountered by the practicing pediatrician. It is a natural phenomenon occurring in the majority of full-term infants and virtually in all preterm infants.1,2 Neonatal jaundice reflects the presence of pigment in the skin and sclera, although little is known about the exact location of the pigment and to what it might be bound in those locations.3 Nonetheless, it is related to hyperbilirubinemia in the transition after birth, which occurs in all babies, except those lacking albumin, which is an extremely rare condition. This transitional phenomenon is usually benign and may have a physiological role in development, but under some conditions bilirubin outside the circulation can be dangerous, such as its accumulation in the brain, contributing to neurologic dysfunction and, sometimes, permanent injury.1,4

The syndrome of neonatal jaundice results from an imbalance between bilirubin production and bilirubin elimination,1,2,5 which is temporarily exacerbated during the transition after birth. This imbalance can be understood by analogy to a sink where the turned on spigot represents the process of bilirubin production and the drain represents the process of bilirubin elimination (Figure 2-1). If the rate at which bilirubin is produced in the body exceeds the rate at which bilirubin is eliminated, then the level in the body increases. In the analogy, the size of the sink represents the capacity of the circulation to contain bilirubin, and this is dependent mainly on the albumin concentration and the affinity of albumin to bind bilirubin. In the newborn, the capacity of the sink is decreased, and thus the likelihood that bilirubin will escape the circulation and move into tissues such as the brain is increased. The situation is worse in this regard in the preterm infant where the capacity is even lower because of a decreased albumin concentration and lower affinity for binding bilirubin, especially in the first days after birth and further compromised by any illness reflected in physiological instability.6 A more general discussion of the physiology of neonatal unconjugated hyperbilirubinemia and the epidemiology of neonatal jaundice is contained in other chapters. However, the biochemistry of bilirubin production is fundamental to the problem of neonatal jaundice, which cannot occur without the existence of the pigment.7

There is a single biochemical source of bilirubin in the body, which is the enzymatic two-step process of heme catabolism.8 The reaction is ubiquitous, occurring in all nucleated cells, and thus in all tissues including the nucleated cells in blood. The substrate for the reaction, heme, is a part of many important proteins, but is present in large amounts in the hemoglobin of red blood cells (RBCs). The first step in the process is catalyzed by heme oxygenase (HO) (Figure 2-2).8 This is the rate-limiting step in the process and is a membrane-bound event with requirements for NADPH donated from the cytochrome P450 system and molecular oxygen and involving a series of oxidations and reductions ultimately resulting in the breaking of the IXα methene bridge of the porphyrin macrocycle, yielding equimolar amounts of carbon monoxide (CO), biliverdin, and ferrous iron (Fe2+). Biliverdin is further reduced enzymatically by biliverdin reductase to bilirubin, which thus is also produced in equimolar amounts with the intermediate products in the same process. The one gaseous product of the first step, CO, diffuses from the cell and is bound to hemoglobin forming carboxyhemoglobin (COHb), which is carried in the circulation to the lungs. CO is excreted in breath in exchange for oxygen, reflecting the total-body production from all sources. Because of the equimolar production of bilirubin and CO in the catabolism of heme, total CO production can be used as an index of total bilirubin production.9,10

The rate of bilirubin formation is highly regulated, in particular by HO. HO is present in tissues in two main isoforms, HO-1 and HO-2.11 The former is inducible and the latter is constitutive. The ratio of these isoforms of HO to each other varies in tissues, with one or the other usually predominating. The expression of HO is also regulated developmentally.12 There are many inducers of HO-1, which has a promoter reflecting its many different roles in biology, probably accumulated over millions of years. Thus, it is important to understand HO as not simply a cause of jaundice, neurologic disturbances, and kernicterus, CO not simply as a toxin capable of causing mitochondrial dysfunction, and Fe2+ not simply as a participant in the Fenton reaction and generator of reactive oxygen species (ROS) production.13 In fact, the same enzymatic system has many potential important biologic effects, some of them clearly beneficial. For example, the biliverdin–bilirubin shuttle14,15 may have important antioxidant,16 anti-inflammatory,14 and antiapoptotic14 effects. CO has a role in many physiological processes.17 It can act directly to cause vessel relaxation and also indirectly through increases in soluble guanylyl cyclase (sGC) and cyclic GMP to cause vessel relaxation17,18 and antiplatelet,19 antiapoptotic,20 and antiproliferative21,22 effects (the latter in vascular smooth muscle cells) (Figure 2-3). It also may have a role in neurotransmission,21,22 and by acting through p38 MAPK, CO may have an inhibitory effect on proinflammatory cytokines.23 Working through increases vascular endothelial growth factor (VEGF),24 CO may have a role in angiogenesis as well. Even Fe2+ under some conditions may participate in antioxidant, anti-inflammatory, and antiapoptotic processes.25 Thus, the process by which bilirubin is produced needs to be understood in the context of other complex interactions between this enzymatic system and others. Such interactions include the biochemical pathways containing hemoproteins, with one of the most notable examples being the nitric oxide synthase (NOS) system.17,26 The resulting interactions are not always predictable or intuitive and are dependent on the timing and context, including developmental time frame. Relevant to the ensuing discussion about the use of CO measurements to estimate total bilirubin formation, it is always important to understand that there are a variety of nonenzymatic sources of CO in the body.27 Under some pathologic conditions these sources can be quite large. Two very important sources, which we have described, are photo-oxidation27–29 and lipid peroxidation.13 Nonetheless, under most conditions encountered in the newborn, except in the case of high supplemental oxygen exposure, severe infection, or intense light exposure in the smallest infants, estimates of total CO production can be used as an index of total bilirubin production.10 In this regard, heme degradation accounts for over 80% of all the endogenous sources of CO with 70% of heme degradation represented by senescing RBCs, 10% by ineffective erythropoiesis, and the remaining 20% by the degradation of other hemoproteins.13,27 Under usual conditions, less than 20% of endogenous CO comes from nonheme source, such as lipid peroxidation and photo-oxidation.13,27

Thus, estimates of total body endogenous CO production (VeCO) can be used as estimates of total bilirubin production. Measurements were made in human infants as early as 1949 by Sjöstrand30 and in 1968 by Fällström.31 These investigators used a flow-through system similar to the one used later by investigators at Stanford in the 1970s (Figure 2-4).32 Most of this earlier work was limited by detector technology, and it was understood that these CO measurements could only serve as an index of total body bilirubin formation because of the inability to account for alternative sources of CO in the clinical setting. In a laboratory, however, the stoichiometric relationship between CO production from heme catabolism and bilirubin production could be validated. This was demonstrated in a rat model using a gas collection apparatus connected to a reduction gas detector, which replaced the earlier infrared CO detectors and had much greater sensitivity.33–35 The new detector depended on a mercuric oxide reaction bed, which would react with a reducing gas, such as CO, to produce carbon dioxide (CO2) and mercury vapor (Hg), the latter measurable using an optical detection system.36 Other reducing gases, such as H2, could also be detected after separation by a gas chromatograph. The detection limit for CO was less than 1 part per billion using such a CO analyzer, making it possible to study CO production by small animals, such as rats and mice, as well as by tissue slices and even collections of cells. The new reduction gas detector could also be adapted for a gas chromatographic HO assay,37 replacing the Tenhunen assay,8 and making it possible to assess directly the activity level of the first and rate-limiting enzyme in the two-step catabolic process of heme in a variety of tissues. In the validation experiment, 100% of heme was recovered as CO in the breath of the animals after injection of a known amount of heme as damaged RBCs over a time frame of 8–12 hours reflecting RBC sequestration, destruction, and heme degradation (Figure 2-5).38

A large number of infants were studied in the 1970s32,39 and 1980s40–42 at Stanford using this older, yet elegant, flow-through system to measure VeCO. This work affirmed earlier reports demonstrating that the term infant produced bilirubin at a rate two to three times higher than the human adult on a bodyweight basis. The technology was also able to distinguish infants with known risk factors for jaundice, such as hematoma and polycythemia (Figure 2-6).43 Nonventilated preterm infants were also observed to have a slightly higher VeCO, probably related to an even shorter RBC lifespan compared with the term infant. Small, ventilated preterm infants had even higher CO excretion rates,44 but it is now known that such infants might also have had a pathologic source of CO related to oxygen exposure and mechanical ventilation, that is, lipid peroxidation in their lungs.45 Although the latter possibility is only speculation at this point in time, circumstantial evidence is supportive of this interpretation. Also, the Stanford group was the first to report that increased bilirubin production was an important contributing cause to the jaundice observed in infants of diabetic mothers.32 In some cases this was related to polycythemia, but in others, because of the absence of erythrocytosis, it was most likely related to ineffective erythropoiesis probably occurring in the liver. Companion studies demonstrated that infants of diabetic mothers also had impaired elimination of the pigment after controlling for bilirubin production.42 Finally, infants with hemolytic conditions were easily identified, with Rh disease demonstrating the highest VeCO measurements (Figure 2-6).43

VeCO measurements conducted using the large chamber or using a smaller hood with a neck seal were cumbersome and limited in application in the clinical setting (Figure 2-4).40 New techniques were developed to measure end-tidal CO concentration corrected for ambient carbon monoxide (ETCOc).9,46–48 Equations predicted a direct relationship between the VeCO and ETCOc as well as carboxyhemoglobin corrected for ambient carbon monoxide (COHbc).40,41 The relationship between COHbc and ETCOc was validated (Figure 2-7), and currently either technique, with appropriate correction for ambient CO, can be used to estimate endogenous CO production and thus total body bilirubin formation.44,49 In fact, a COHbc measurement had been considered the standard approach because it involved a simple blood sample technique and avoided cumbersome technology.27,36,50 Earlier elegant studies conducted by Maisels et al. using a rebreathing system also used a COHb measurement over time to estimate bilirubin production rates.51 Nonetheless, ETCO measurements represent the easiest and least invasive of the techniques for estimating endogenous CO production and can be used to study large numbers of infants, providing for the first time data on the distribution of ETCOc in a large population of infants, most of them normal but some with conditions known to be associated with neonatal jaundice. In fact, the hour-specific bilirubin nomogram52 (Figure 2-8), which is used for decision-making about treatment for late preterm and term infants, is further informed by information about total bilirubin formation, with infants in the highest risk quartile having a much greater likelihood of having increased bilirubin production as an important contributing cause of their neonatal jaundice. Thus, not only does bilirubin help to define risk for the normal population, but it is also now known that hemolysis represents an important risk factor for neurologic dysfunction and permanent injury caused by bilirubin.53 In this regard, the diagnosis of hemolysis is paramount in the management strategy for neonatal jaundice, with intervention being recommended at lower bilirubin levels for a given postnatal age in hours. Because estimates of endogenous CO production can be used to index total bilirubin production, estimates of endogenous CO production can be used to identify infants at high risk for neurologic dysfunction or injury in the presence of jaundice. As a case in point, only about half of infants with a positive Coombs’ test are hemolyzing at the time of the test.54 CO detection technology can identify the half who are hemolyzing and moreover might be useful in gauging the magnitude of the risk not only for jaundice but also for neurologic dysfunction or permanent injury in this group.