STATE-OF-THE-ART

End-tidal carbon monoxide and hemolysis

GF Tidmarsh, RJ Wong and DK Stevenson

Hemolytic disease in newborns can resulIt from a number of conditions, which can place such infants at an increased risk for the development of severe hyperbilirubinemia. Because the catabolism of heme produces equimolar amounts of carbon monoxide (CO) and bilirubin, measurements of end-tidal breath CO (corrected for ambient CO) or ETCOc can serve as an index of hemolysis as well as of bilirubin production from any cause.Elevated levels of ETCOc have been correlated with blood carboxyhemoglobin levels and thus hemolysis. However, the detection of hemolysis can be a clinically challenging problem in newborns. Here, we describe the importance of determining ETCOc levels and their application in identifying infants at risk for developing hyperbilirubinemia associated with hemolysis and other causes of increased bilirubin production.

Journal of Perinatology(2014)34, 577-581; doi:10.1038/jp.2014.66; published online 17 April 2014

INTRODUCTION

A substantial proportion of all newborns are jaundiced at some time around birth.¹ In a vast majority of cases, the jaundice will resolve spontaneously without intervention or sequelae. Some infants will need phototherapy for resolution of the jaundice . A subset of these newborns develops significant hyperbilirubi-nemia.² Even within this subset, most are treatable with photo- therapy and, in rare cases, exchange transfusion. A very small minority of infants with significant hyperbilirubinemia will develop neurodevelopmental abnormalities related to bilirubin-toxicity. These abnormalities range from the subtle changes of bilirubin induced neurologic dysfunction (BIND) and/or auditory abnormalities to the most extreme cases of reversible acute bilirubin encephalo-pathy or the chronic, irreversible changes of kernicterus.^3-5

The relationship between extreme hyperbilirubinemia and kernicterus is well established.⁴ However, it is not clear why only small subset of infants develop the adverse neurodevelopmental outcomes associated with bilirubin deposition in the central nervous system. A small study found lower intelligence quotient (IQ) scores in a subgroup of children with total serum/plasma bilirubin (TB) levels of 25 mg di ^-1 (428μmol I ^-1) or greater who had evidence of hemolysis. However, the sample of children with abnormalities was small (n= 9). ⁶

Kuzniewiczet al. ⁷ re-analyzed the data from the Collaborative Perinatal Project (CPP) to assess whether infants with a positive direct antiglobulin test (DAT) are at a higher risk for bilirubin neurotoxicity using multiple linear and logistic regression models. Of the 54 795 newborns included in the study, the authors analyzed the data of those infants≧36 weeks gestational age with birthweights ≧2000 g and having 7-year follow-up IQ measure-ments (n= 32 808). They found that TB levels alone were not predictive of IQ scores, but infants with a positive DAT and TB levels≧25 mg di ^-1 (428μmol I ^-1) had lower scores (for example,full-scale IQ:-6.7 points).

Besides this empiric association between’suspected’hemolysis (a positive DAT in an infant with a blood group incompatibility) and BIND,there are several other possibilities for how hemolysis might contribute to an increased risk for brain injury in the jaundiced newborn. During  hemolysis, bilirubin is produced at a very high rate. Depending upon the total bilirubin binding capacity (BBC) in the circulation (primarily determined by the albumin concentration and albumin’s binding affinity for bilirubin, which is often reduced in the first several days after birth, especially in preterm infants). there is a maximum level above which bilirubin is much less likely to be retained in the circulation. That is, in the context of hemolysis, bilirubin may be produced in excess of what can be bound in circulation or conjugated by the liver for excretion into bile (a process also compromised temporarily in the newborn) and the unbound pigment will inexorably move into tissues by the process of mass action. Thus, the total body bilirubin load will usually be higher in the setting of hemolysis than might be suggested by the TB alone, if the TB is close to the total BBC, and will likely increase over time. Based on this reasoning, two observations can be made. First, delays in treating jaundice due to hemolysis or waiting until TB levels are high (and close to the total BBC) may be fraught with risk; hence, the more conservative recommendations for phototherapy in hemolyzing infants in the publication of the American Academy of Pediatrics (AAP) clinical practice guideline.⁸ Second, it can be easily appreciated how two infants with approximately the same TB level may be at drastically different risks for BIND. For example, a preterm infant with an intrinsically higher bilirubin production rate lower albumin concentration, and thus lower total BBC than a term infant would be at a higher risk，often at a lower TB.A more instructive example is that of a term breastfed infant with a normal bilirubin production rate and total BBC, and a TB of 25 mg di 1 (close to the total BBC) reached at 7 days of age compared with a term infant with hemolysis, a comparable total BBC at the outset and who presents with the approximately same TB level at 24 h of life, having exceeded the total BBC some time before presentation. Both newborns, of course, would receive photo- therapy, but the latter would be expected to require a longer duration of treatment and experience ‘rebound’ jaundice as bilirubin moves back into circulation by mass action after phototherapy is stopped, and also be at a much higher risk for BIND. This supports the belief that tracking the remaining binding capacity or binding saturation might identify infants at risk for developing BIND before the BBC maximum is reached,⁹ but risk threshold levels have not been well established to date.

However, there are two other reasons why hemolysis itself might impart risk in the context of impaired ability to eliminate bilirubin from the body. First, bilirubin itself can cause oxidative stress and its toxicity may be related to this phenomenon in an antioxidant-deficient newborn. Such oxidative stress is most likely to occur when TB and unbound bilirubin levels are high, and tissue bilirubin levels become excessive. An infant’s vulnerability to the toxic effects of bilirubin may not be strictly related to an increased production of bilirubin arising from hemolysis, but due to a decreased capacity to defend against bilirubin-induced oxidative injury when bilirubin levels are excessive.^10-12 Of course, hemolysis may be the driving force behind excessive TB levels. Although bilirubin is important for maintaining the normal redox state of cells and is a known potent antioxidant ’nothing in excess’ is the rule in biology, and excess hemoglobin and heme in circulation is the state of affairs during hemolysis. Second, little is known about the defenses (such as haptoglobin-related protein)¹³  against hemolytic oxidative stress¹⁴ due to the toxic effects of 

‘free’hemoglobin. It seems likely that, in general, such defenses in the newborn are probably less developed and the potential for oxidative injury is far greater. The liver injury often associated with hemolysis might be a telltale sign of such oxidative stress. Moreover, the ability to upregulate heme oxygenase-1 (H0-1), the rate-limiti『1g enzyme in the heme degradative pathway, in response to hemolysis may vary greatly among newborns, putting some individuals at very high risk for oxidative injury in the context of hemolysis. Recent reports have shown that there is an association between human diseases and inter-individual variations in the expression of H0-1. In the H0-1 promoter regulatory region, a polymorphic (GT)n dinucleotide repeat is present, with longer repeats (≦26) associated with a decreased H0-1 expression.¹⁵ Ironically, in the context of hemolysis, infants with long repeats might be more vulnerable to oxidative stress caused by bilirubin (even though their capacity to produce bilirubin would be relatively less). Conversely, infants with short repeats, and thus higher HO 1 activity, might produce bilirubin at a higher rate during hemolysis, and thus have a higher bilirubin load and an increased risk for BIND. Thus, the nature of a hemolytic neonate’s susceptibility to BIND would depend at least in part upon his or her genetic disposition. In both cases, hemolysis is the backdrop for the increased risk.

Therefore, hemolysis is a key parameter for appropriate risk stratification in jaundiced neonates and the assessment of hemolysis is recommended by the AAP.8 The AAP clinical practice guideline on the management of neonatal hyperbilirubinemia identifies hemolysis as a major risk factor for the development of severe hyperbilirubinemia and therefore identifying newborns with hemolysis is of sign的cant importance. Hemolysis in newborn infants may be due to a number of different etiologies and differ greatly in severity.The causes of hemolysis can be broadly divided into immune-mediated disorders, red cell enzyme deficiencies, hemoglobinopathies and red cell membrane defects (Table 1).

There exists tremendous variability in the extent of hemolysis among these etiologies and even among individuals with the same etiology. Determinants of the extent of hemolysis in immune-mediated disorders include antibody titer, antibody isotype, maternal treatment and target antigen. Among the determinants of hemolysis in non-immune-mediated disorders are heterozygous versus homozygous states and exogenous precipi-tating factors. In summary, the causes of neonatal hemolysis are numerous and diverse, and the extent of hemolysis varies tremendously both among the various conditions as well as within a given diagnosis.

Owing to this variability and to the transition of the neonatal hematopoietic system, the detection of hemolysis can be a clinically challenging problem. Traditional laboratory measures used to diagnose hemolysis may be difficult to interpret or be misleading in the newborn.^16,17 Because carbon monoxide (CO) is produced on a one to one molar basis during heme catabolism and accounts for almost the entire amount (~85%) of CO produced endogenously,^18-20  the measurement of CO in exhaled breath can therefore serve as an index of CO levels in the blood (or carboxyhemoglobin (COHb)) and a marker of hemolysis.The remaining 14% of endogenous CO production arises from non heme processes such as photo oxidation²¹ and lipid peroxi-dation.22 However, in certain pathologic states, these processes may have a significant effect on the ETCO level and its relationship to hemolysis alone. In this review, we summarize the available literature regarding the measurement of end-tidal carbon monoxide (ETCO) and its relationship with hemolysis.

End-tidal carbon monoxide measurements

Numerous publications have reported on the measurement of ETCO and these are summarized in Table 2. The measurement of CO in exhaled breath is a reflection of CO levels in the blood, especially when corrected for ambient CO (ETCOc). Traditionally, ETCOc measurements have been performed with gas chromate-graphy.²³ The first semi-portable bedside instrument was described by Vreman et al.²⁴ and used by Stevenson et al.²⁵ to show that ETCOc measurements may be used as an index of total bilirubin production in healthy term neonates at low risk for developing jaundice.They found that ETCOc measurements using this device strongly correlated (r² =0.94,n= 108) with those made with gas chromatography.²⁴

Vremanet al. ^26,27 subsequently published data regarding values for normal neonates, children and adults as well as those with hemolytic diseases. These researchers used an automated end tidal instrument to collect initial and repeated measurements (n= 17) of ETCOc values of nine neonates.The mean ETCOc values of non-hemolytic and hemolytic neonates were 1.9 ± 0.6 and 7.3 ± 0.6 p.p.m., respectively, and showed that high ETCOc measurements (>3 p.p.m.) correlated with known hemolytic conditions as well as with COHb levels, corrected for inhaled CO (COHbc), measured using gas chromatography, a methodology that is however not available for routine clinical use.26 Although spectrophotometric measurements of COHb by CO-oximetry are performed in the clinical laboratory, the instrumentation does not have the needed sensitivity to accurately measure COHb as a predictor of hemolysis.²⁸

Vremanet al. ²⁷ also used a second-generation automated ETCO device to measure ETCOc values in normal adults and children as well as hemolytic adults and children. They found significant differences between the normal adult and children ETCOc values and the hemolytic (sickle cell disease (SCD) and thalassemias) adult and children ETCOc values (Table 3). These values were well within the range of the published ETCOc values, which have been determined by several different methods (for example, gas chromatography and CO spectrophotometry) and similar devices.²³

To assess the usefulness of ETCOc in healthy, term, Coombs-positive infants, Javieret al. ²⁹ correlated ETCOc values to corrected reticulocyte counts (RCs). At 36 ± 2 h of age, ETC0, and RC were determined in Coombs-positive neonates. 25 of 50 (50%) Coombs-positive neonates had RCs <5% and a mean ETCOc of 1.80 ± 0.34 p.p.m. compared with that of Coombs-negative patients (1.60 ±0.45 p.p.m., n= 50), and thus, did not have hemolysis. However, neonates with RCs between 5 and 8% had a mean ETCOc of 2.77 ±0.68 p.p.m., and those with RCs >8% had a mean ETCOc of 4.52 ± 0.83 p.p.m. Thus, an ETCOc >2.5 p.p.m. appeared to predict significant elevations in RC and confirmed the correlation between elevated levels of ETCOc and the presence of hemolysis.

Stevenson et al.³⁰ evaluated whether ETCOc alone or in combination with TB measurements could predict those infants who will develop hyperbilirubinemia during the first 7 days of life. They found that ETCOc values alone at 30 h of life do not predict hyperbilirubinemia; however, they explained that high positive predictive values were unlikely in this study due to the low prevalence of infants with severe hyperbilirubinemia. In addition, the study did find that ETCOc measurements can identify hemolysis in infants, but did not predict hyperbilirubinemia. This is probably due to the fact that a measure of bilirubin conjugation was not included in the study design.

Herschelet al. ³¹ compared the positive predictive value of the DAT (Coombs test) with ETCOc. They found that the DAT did not identify over 50% of infants with hemolysis that were identified by ETCOc measured at 12 ± 6 h of life. Therefore, early ETCOc testing appears to be more useful than DAT in identifying those neonates with hemolysis.

Using a prototype ETCO monitor, Barak et al.³² showed that ETCOc levels in newborn infants in Israel significantly correlated with red blood cell mass, as indexed by measurements of hemoglobin (Hgb) or hematocrit (Hct) levels within the first 8 h of life. ETCOc correlated linearly with Hct （r² 10.1%, P=0.015) and Hgb (r² = 11%,P= 0.011), with infants with Hct >65% having higher ETCOc levels. However, when the authors corrected for CO production of each infant, the correlation was minimally improved using a parabolic regression curve to Hct （r² 15.9%, P= 0.009) and Hgb (r²= 14.5%,P= 0.013).

In 2002, Herschelet al. ³³ performed a case study of a female, African-American neonate heterozygous for glucose-6-phosphate dehydrogenase (G6PD) deficiency. ETCOc levels were measured at 8 and 25 h of life and were 3.1 and 2.6 p.p.m.，respectively, confirming that the infant was actively hemolyzing.²⁵ Kaplan et al.³⁴ studied hemolysis and the risk of hyperbilirubinemia in G6PD deficient and -normal African American neonates. They found that ETCOc (median and interquartile range) values were significantly higher in G6PD-deficient (n= 59; 2.4 [2.0-2.9] p.p.m.) than those in G6PD-normal infants (n= 363; 2.1 [1.7-2.5] p.p.m.), P <0.001. In a follow-up study, Kaplanet al. ³⁵ demonstrated that among hyperbilirubinemic G6PD-deficient and -normal infants, ETCOc levels were similar and concluded that hemolysis was not predictive nor a major contributor in the pathogenesis of hyperbilirubinemia compared to the contribution of G6PD deficiency.

Sylvester et al.³⁶ assessed whether ETCOc values in children with SCD could be used to identify hemolysis.The authors found that the mean ETCOc values in children with SCD (n= 87,4.9 ± 1.7 p.p.m.) were significantly higher {P<0.0001) than that of children without SCD (n= 26, 1.3 ± 0.4 p.p.m.). In addition, ETCOc values had a positive correlation between COHb (P= 0.007) and TB (P= 0.02). These data showed that ETCOc can be used to identify hemolysis in patients with SCD.

Jameset al. ³⁷ set out to assess whether ETCOc measurements can be used to identify hemoglobinopathies as well as when transfusions would be required in patients receiving chronic transfusion therapy. They performed ETCOc measurements in patients with SCD and thalassemias, which were then compared with healthy controls. They found the ETCOc values to be statistically different in all groups (P= 0.006) with the mean ETCOc values in the SCD non-transfused group (n= 17) being the highest (6.6 ± 0.7 p.p.m.) and the lowest (1.3 ± 11.6 p.p.m.) in the control group (n= 62). In addition, they found that pre-transfusion ETCOc levels significantly correlated with RCs (r= 0.96, P <0.001) and thus may be an indicator of ineffective erythropoiesis. This study not only showed that in patients with hemoglobinopathies, ETCOc values are well above those of healthy controls; but also, that ETCOc measurements can be used to determine the need and proper timing of transfusions.

A recent publication by Bloket al. ³⁸ confirmed thefindings of the Kuzneiwicz re analysis of the CPP data. Their aim was to evaluate the predictive ability of ETCOc and cytokine levels for long-term outcome in 105 preterm infants. At 1, 3 and 5 days of life, measurements of ETCOc, plasma tumor necrosis factor-aand interleukins 6 and 8, and malondialdehyde (a marker of lipid peroxidation) levels were taken and then compared with neuro developmental outcome at 3.5 years of age using the Griffiths Mental Developmental Scales. ETCOc at O 24 h was determined to be higher (2.7 ± 0.7 p.p.m., n= 15) in infants with adverse outcomes (Griffiths developmental quotient <85) compared with infants with favorable outcomes (2.0 ± 0.5 p.p.m.,n= 54,P <0.05).

No differences in malondialdehyde and cytokine levels were found between both groups. Using regression analysis, the authors found that only the ETCOc level correlated with outcome with a 93% sensitivity and an 85% negative predictive value. The authors therefore concluded that an ETCOc <2.0 p.p.m. during thefirst day of life is predictive of a favorable outcome at 3.5 years of age.

CONCLUSIONS

A substantial proportion of all newborns are jaundiced at some time around birth. A subset of these babies develops significant hyperbilirubinemia and a very small minority will develop neuro-developmental abnormalities related to bilirubin toxicity. How-ever, it is not clear why only a small subset of babies will have adverse neurodevelopmental outcomes associated with bilirubin deposition in the central nervous system. Several studies, includ-ing a large cohort analysis, suggest that hemolysis in conjunction with hyperbilirubinemia places newborns at greater risk for adverse neurodevelopmental outcomes. Therefore, hemolysis is a key parameter for appropriate risk stratification in jaundiced neonates and the assessment of hemolysis is recommended by the AAP.

Hemolysis in newborns resuIts from a wide variety of causes and the detection of hemolysis can be a clinically challenging problem. Because CO is produced during the breakdown of heme, measurements of ETCOc can serve as a marker of hemolysis. Elevated levels of ETCOc have been correlated with hemolysis in newborns suggesting that accurate, rapid analysis of ETCOc may provide a clinically useful tool for identifying newborns with hyperbilirubinemia requiring greater attention. Unfortunately, currently there is no commercial device available for clinical use. Exogenous sources of elevated CO must be taken into account when evaluating patients for ETCOc and hemolysis. Nonetheless, the need for an accurate, sensitive and reliable device seems warranted.

CONFLICT OF INTEREST

GFT has afinancial interest in the technology behind a device that Capnia Inc. (Palo Alto, CA) is developing. DKS and RJW are unpaid consultants to Capnia Inc. and have no conflict of interest orfinancial disclosures to declare.

REFERENCES

1 Slusher TM, Zipursky A, Bhutani VK. A global need for affordable neonatal jaundice technologies.Semin Perinato/2011;  35: 185-191.

2 Bhutani VK, J hnson L. A proposal to prevent severe neonatal hyperbilirubinemia and kernicterus.J Perinato/2009; 29(Suppl 1): S61-S67.

3 5hapiro SM. Definition of the clinical spectrum of kernicterus and bilirubin- induced neurologic dysfunction (BIND)J Perinatal 2005;25 : 54-59.

4 Shapiro SM, Bhutani VK, Johnson L. Hyperbilirubinemia and kernicterus. Clin Perinatal 2006;33 :387-410.

5 Shapiro SM, Nakamura H. Bilirubin and the auditory system.J Perinatal 2001; 21 (Suppl 1): S52 S55.

6 Newman TB, Liljestrand P, Jeremy RJ，Ferriera DM, Wu YW, Hudes ES et al. Outcomes among newborns with total serum bilirubin levels of 25 mg per deciliter or more.N Engl J Med 2006;354 : 1889-1900.

7 Kuzniewicz M, Newman TB. Interaction of hemolysis and hyperbilirubinemia on neurodevelapmental outcomes in the callaborative perinatal project. Pediatrics 2009;123: 1045-1050.

8 American Academy of Pediatrics. Management of hyperbilirubinemia in the newborn infant 35 or more weeks of gestation.Pediatrics 2004：114: 297 -316

9 Ahlfors CE, Wennberg RP Bilirubin-albumin binding and neonatal jaundice. Semin Perinatal 2004;28 :334-339.

10 Brito MA, Rosa Al, Falcao AS, Fernandes A, Silva RF, Butterfield DA et al.Uncon-jugated bilirubin differentially affects the redox status of neuronal and astroglial cells. Neurobiol Dis2008;29 : 30-40.