PEDIATRICS Vol. 108 No. 1 July 2001, pp. 175-177

COMMENTARY:
Problems With Prediction of Neonatal Hyperbilirubinemia

The article by Stevenson et al¹ in this issue deserves the attention of practicing pediatricians as well as those involved in clinical studies in response to the demand for evidence-based practice of pediatrics. In a multicenter study (MCS) Stevenson et al¹ tested the hypothesis that a single predischarge (30 ± 6 hours of age) measurement of end-tidal carbon monoxide (CO) corrected for ambient CO (ETCO_C) alone or in combination with the simultaneous measurement of total serum bilirubin (TSB) concentration could predict the development of significant neonatal bilirubinemia (defined for MCS as TSB 95th percentile for age as determined by a previous study²). The implicit practical goal of the study was to establish a reliable and cost-effective management of neonatal jaundice leading eventually to the eradication of kernicterus. The study showed that ETCO_C did not improve the predictive ability of a predischarge TSB, and TSB alone or in combination with ETCO_C did not achieve a clinically useful level of prediction of significant bilirubinemia. The purpose of this commentary is to search for the explanation(s) for these negative results.

ETCO_C is a good index of endogenous CO production.³ CO is generated, in equimolar relation to bilirubin, during the enzymatic degradation of heme through the heme oxygenase pathway.⁴ The bulk of degraded heme and the main source of individual variation is derived from the breakdown of erythrocyteshemolysis(erythrolysis is the etymologically correct term). In healthy mature neonates intravascular bilirubin mass, indexed by TSB, is closely related to the total body bilirubin mass⁵ and for practical purposes TSB can be used in its place. Thus, the TSB at any age t during the first few days of life is related to the cumulative rate, up to age t, of the de novo bilirubin production (a), of the bilirubin enterohepatic circulation (b), and of the bilirubin elimination (c). In the neonatal period the rates of these processes are in constant flux. Direct measurements and indirect evidence indicate that a and b decrease and c increases with age. Moreover, the initial level of a, b, and c as well as their rate of change vary independently and show wide individual variation.^6,7 The neonatal bilirubinemia curve represents the net result of the flux of processes a, b, c over time age. In a typical curve the initial phase of decelerating velocity of TSB increase is followed by a TSB peak ie, balance between a and b and c and then by a phase of decreasing TSB values until adult level is achieved. The sequence of these phases is obligatory for all neonates but the level of peak TSB and the age at reaching it exhibit the well-known wide individual variation.^8,9 The complexity of the interactions between the rates of bilirubin production (a), bilirubin enterohepatic circulation (b), and bilirubin elimination (c) in determining TSB at age t is outlined in the following mathematical model:

(1)

where TSB₀ represents the cord blood TSB.

According to this equation an infinite combination of values for a, b, and c can result in the same TSB_t, and the actual level of TSB_t does not depend on the absolute values of these factors at age t but on the cumulative level of their imbalance up to this age. In the case of MCS, t was the age at predischarge sampling for TSB and ETCO_C and the end-point the hours of age-specific 95th percentile TSB.

As TSB_t incorporates the cumulative effects of all three factors (a, b, and c), determining its level it should not have been expected that a spot estimation of bilirubin production by ETCO_C at age t would have improved the predictive ability of TSB_t.^2,3 This was confirmed by the results of the MCS.¹ The intuitively anticipated positive effect of ETCO_C on the predictive ability of TSB_t³ was based on the contributions of estimates of endogenous CO production in our understanding of the pathogenesis of neonatal jaundice in several clinical groups.^10-14 However, this use of estimates of CO production is distinctly different from the requirements of a prognostic test to be applied on the neonatal population as a whole. Notwithstanding the above considerations, including ETCO_C in the MCS protocol should not be judged as a wasted effort. The study by Stevenson et al provided, in the form of a frequency distribution graph (Fig 2, page 35),¹ values of ETCO_C during a narrow time window (30 ± 6 hours of age) for a large cohort of unselected neonates. The graph shows a unimodal Gaussian distribution indicating a continuum of rates of erythrocyte breakdownhemolysisand bilirubin production. There was no obvious cutoff point to separate the fraction of the population with "hemolysis" from those without it. In the MCS article¹ the distribution of ETCO_C values was collapsed in a binary manner ie, above or below the population mean. The implication that half of the neonates suffered from "hemolysis" surely was not among the intentions of the authors. In view of the fact that, depending on the clinical group, an ETCO_C value at 30 ± 6 hours of age may deviate significantly from the values before as well as after this age, any cutoff point based on the distribution of values of a single predischarge measurement of ETCO_C is bound to be arbitrary and clinically confusing.

Hemolysis goes on continuously as erythrocytes reach the end of a finite lifespan. In healthy adults in steady state there is a constant rate of erythrocyte generation and breakdownhemolysiswith little day to day variation.¹⁵ In contrast the perinatal period is characterized by a pronounced heterogeneity of erythrocyte lifespan and absence of a steady state. The rapid expansion of fetoplacental blood volume in the last trimester of pregnancy results in an increased percentage of young erythrocytes and therefore the fraction that daily reach the end of their lifespan is reduced. However, in the other end of the spectrum, highly sophisticated methods, revealed the presence in the neonatal peripheral blood of a fraction, of approximately 10%, of erythrocytes with very short survival after birth.^16,17 Hemolysis of this fraction of neonatal erythrocytes is similar in every aspect with the hemolysis of old, but otherwise normal, erythrocytes. There are no specific morphologic abnormalities and as the process is self-limited no drop in hematocrit can be detected.¹³ Variability in the total mass of erythrocytes and in the fraction with very short survival as well as in the rate of their removal is the likely explanation for the wide range of endogenous CO production rates in normal neonates.^10,11,13 This nonspecific neonatal hemolysis should be separated from increased erythrocyte breakdown caused by specific hemolytic conditions as blood group isoimmunization, inherited red cell enzyme defects as glucose-6-phosphate dehydrogenase (G6PD) deficiency or membrane skeleton abnormalities as spherocytosis, which can be diagnosed by the appropriate tests. Such tests were not used in the MCS and therefore the distribution of ETCO_C values, particularly in the high range, includes the values from the small minority with specific hemolytic conditions.

Information on the time course in the neonatal period of the progressive decline of the rate and of the variability of bilirubin production is limited.¹⁸ (It is regretful that a repeat ETCO_C measurement was not included in the MCS protocol to provide this valuable information). In Rhesus and similar isoimmunizations hemolysis is independent of erythrocyte age and continues, albeit in declining rates, as long as sensitized erythrocytes circulate. In ABO isoimmunization only the young erythrocytes exhibit sufficient density of A or B antigens to induce antibody-mediated hemolysis. Thus hemolysis, although at times initially marked, is short-lived and seldom causes anemia.¹⁹ Triggered hemolysis in G6PD deficiency is independent of erythrocyte age while in the absence of exposure to an exogenous hemolytic agent or infection spontaneous hemolysis (most likely restricted to those with Class II(B-) mutants) is limited to older erythrocytes and therefore anemia does not develop.^20,21 In spontaneous hemolysis the frequency distribution of peak TSB values is shifted to the right and shows a conspicuous tail in the region of high values. Moreover in 15% to 20% of this group of neonates a second phase of accelerating TSB increase may occur any time in the first several days of life.²² Furthermore, early postpartum discharge exposes the G6PD-deficient neonates to household items containing chemicals with unrecognised hemolytic potential at a time when the balance between bilirubin production and elimination is in a precarious stage.²⁰ The above fully explain why in recent years G6PD deficiency was reported to be a frequent cause of kernicterus out of proportion to the prevalence of this enzymopathy in the North American population.²³ The discussion so far allows the conclusion that ETCO_C may prove helpful in the management of neonatal jaundice only in conjunction with a diagnostic classification of the case and not instead of it as it was done in the MCS. A diagnosis of hemolytic jaundice, based on elevated ETCO_C values, does not convey in either a quantitative or qualitative way any useful information to the clinician.

So far in this commentary the limitations of an early (t) ETCO_C measurement in predicting significant bilirubinemia were discussed. The similar failure of TSB_t will be briefly addressed. For TSB_t to predict the course of bilirubinemia, the level of imbalance between bilirubin production plus enterohepatic circulation and bilirubin elimination that determine the percentile position of individual TSB_t (see equation 1) must remain stable. The proportion of the MCS population that deviated from the above condition and reached the end-point of significant hyperbilirubinemia in relation to the test cutoff point (TSB_t 75th hour of age-specific percentile) determined the overall predictive ability of TSB_t. Of those with a positive TSB_t test, 16.7% moved upwards to reach the end-point (positive predictive value) while of those with a negative TSB_t test 1.9% also moved upwards and reached the end-point (1-negative predictive value).¹ The contribution to these disappointing results of the variability over time of bilirubin production was already discussed. The significance of the bilirubin enterohepatic circulation decreases with age in parallel to the development of enteric flora. The effect of breastfeeding on the quantitative and qualitative aspects of the evolution of neonatal enteric flora may be the explanation for the breastfeeding-related accentuation of neonatal jaundice. A host of factors have been implicated as responsible for the marked variability in the initial level and in the rate of improvement of bilirubin elimination that limit the predictive ability of TSB_t.⁷ Gestational age <38 weeks is such a factor inversely related to the initial level and the rate of activity increase of uridine diphosphoglucuronyl transferase (UDPGT).^7,24 The recently described promoter region variant of the UDPGT1 gene associated with Gilbert syndrome was found to delay the improvement of UDPGT activity and thus prolong the phase of increasing TSB levels.²⁵ This effect is magnified by coexistence of increased rates of bilirubin production and by caloric deprivation.²⁴

The discussion so far has focused on the obstacles to early prediction of significant hyperbilirubinemia inherent to the pathogenesis of neonatal jaundice. Flaws in the protocol, the analysis of the data, and the overall strategy of MCS only marginally could have contributed to the disappointing results. This should not be an excuse for inaction. The pediatric community has the strong obligation to effectively stop the wave of cases of kernicterus associated with early postpartum discharge. The new effort will start with a strong advantage: a transcutaneous bilirubometer (TcB) with intradevice and interdevice precision that exceeds by far the performance of clinical laboratory methods for measuring TSB.²⁶ This device makes it easy to perform serial measurements that will allow the calculation of the hours of age-specific velocity of TSB increase, expected to be superior to isolated measurement as a predictor of the course of bilirubinemia. Similarly it will be easier to enroll a representative sample to compute percentile or quartile tracks and avoid the sampling biases and nonrandom drop-outs apparent in the existing nomogram.^2,27 Having available a cord blood sample from all live births for selective diagnostic tests, if such a need arises, is another step toward noninvasive methods. Finally we need to make the intellectual transition from normal range to threshold for intervention TSB values. The first have no practical meaning for a self-resolving condition and only generate unjustified parental and physician confusion and anxiety. The second stress the need for timely intervention safeguarding life and neurologic integrity while time is gained until the resolution of the problem. In defining the threshold for intervention TSB the available method for intervention ie, exchange transfusion, phototherapy, tin-mesoporphyrin should be taken into consideration.²⁸

In the ideal world, reigned by Prometheus, the issue of dangerous hyperbilirubinemia should have been solved before embarking in a drastic cut of the in-hospital observation of newborn infants. However, even if we are forced to follow Epimetheus, we have a good chance of providing evidence-based management methods to protect the newborn population entrusted to our care from the calamity of kernicterus.

Timos Valaes, MD, DCH
53 Dimitrakopoulou Street
Voula 166 73, Greece

FOOTNOTES

Received for publication Jan 3, 2001; accepted Jan 17, 2001.

Address correspondence to Timos Valaes, MD, DCH, 53 Dimitrakopoulou St, Voula 166 73, Greece.

ABBREVIATIONS

MCS, multicenter study; CO, carbon monoxide; ETCO_C, end-tidal CO corrected for ambient CO; TSB, total serum bilirubin concentration; G6PD, glucose-6-phosphate dehydrogenase; UDPGT, uridine diphosphoglucuronyl transferase; TcB, transcutaneous bilirubinometer.

REFERENCES

1. Stevenson DK, Fanaroff AA, Maisels MJ, Prediction of hyperbilirubinemia in near-term and term infants. Pediatrics. 2001; 108:31-39 [Abstract/Free Full Text]
2. Bhutani VK, Johnson L, Sivieri EM Predictive ability of a predischarge hour-specific serum bilirubin for subsequent significant hyperbilirubinemia in healthy term and near-term newborns. Pediatrics. 1999; 103:6-14 [Abstract/Free Full Text]
3. Stevenson DK, Vreman HJ Carbon monoxide and bilirubin production in neonates. Pediatrics. 1997; 100:252-254 [Free Full Text]
4. Rodgers PA, Vreman HJ, Dennery PA, Stevenson DK Sources of carbon monoxide (CO) in biological systems and applications of CO detection technologies. Sem Perinatol. 1994; 18:2-10
5. Valaes T. Bilirubin distribution and dynamics of bilirubin removal by exchange transfusion. Acta Paediatr Scand. 1963;52(suppl 149)
6. Valaes T Bilirubin and red cell metabolism in relation to neonatal jaundice. Postgrad Med J. 1969; 45:86-106 [Free Full Text]
7. Hansen TWR. Fetal and neonatal bilirubin metabolism. In: Maisels MJ, Watchko JF, eds Neonatal Jaundice. Newark, NJ: Harwood Academic Publishers; 2000:3-20
8. Maisels MJ, Gifford K Normal serum bilirubin levels in the newborn and the effect of breast feeding. Pediatrics. 1986; 78:837-843 [Abstract/Free Full Text]
9. Maisels MJ. The clinical approach to the jaundiced newborn. In: Maisels MJ, Watchko JF, eds. Neonatal Jaundice. Newark, NJ: Harwood Academic Publishers; 2000:139-168
10. Bjure J, Fallstrom SB Endogenous formation of carbon monoxide in newborn infants: I. Non-icteric and icteric infants without blood group incompatibility. Acta Paediatr Scand. 1963; 52:361-366 [CrossRef]
11. Fallstrom SP, Bjure J Endogenous formation of carbon monoxide in newborn infants. Acta Paediatr Scand. 1968; 57:137-144 [Medline]
12. Maisels MJ, Pathak A, Nelson NM, Nathan DC, Smith CA Endogenous production of carbon monoxide in normal and erythroblastotic newborn infants. J Clin Invest. 1971; 50:1-8
13. Necheles TF, Rai US, Valaes T The role of haemolysis in neonatal hyperbilirubinaemia as reflected by carboxyhaemoglobin levels. Acta Paediatr Scand. 1976; 65:361-367 [Medline]
14. Bartoletti AL, Stevenosn DK, Ostrander CR, Johnson J Pulmonary excretion of carbon monoxide in the human infant as an index of bilirubin production. I. Effects of gestational and postnatal age and some common neonatal abnormalities. J Pediatr. 1979; 94:952-955 [CrossRef][Medline]
15. Bloomer JR, Berk PD, Howe RB, Berlin NI Interpretation of plasma bilirubin levels based on studies with radioactive bilirubin. JAMA. 1971; 218:216-220 [Abstract/Free Full Text]
16. Bratteby L-E, Garby L, Groth T, Schneider W, Wadman B Studies of Erythro-Kinetics in infancy. XIII. The mean life span and the life span frequency function of red blood cells formed during foetal life. Acta Paediatr Scand. 1968; 57:311-320 [Medline]
17. Linderkamp O, Fraderichs E, Meiselman H Mechanical and geometrical properties of density-separated neonatal and adult erythrocytes. Pediatr Res. 1993; 34:688-693 [Medline]
18. Balaraman V, Pelke S, DiMauro S, End-tidal carbon monoxide in newborn infants: observations during the first week of life. Biol Neonate. 1995; 67:182-185 [Medline]
19. Zipursky A, Bowman M. Isoimmune hemolytic diseases. In: Nathan DG, Oski FA, eds. Hematology of Infancy and Childhood. 4th ed. Philadelphia, PA: WB Saunders; 1993:44
20. Valaes T Severe neonatal jaundice associated with glucose-6-phosphate dehydrogenase deficiency: pathogenesis and global epidemiology. Acta Pediatr. 1994; 394:58-76
21. Valaes T. Neonatal jaundice in glucose-6-phosphate dehydrogenase deficiency. In: Maisels MJ, Watchko JF, eds. Neonatal Jaundice. Newark, NJ: Harwood Academic Publishers; 2000:67-72
22. Kappas A, Drummond GS, Valaes T A single dose of Sn-mesoporphyrin prevents development of significant hyperbilirubinemia in glucose-6-phosphate dehydrogenase deficient newborns. Pediatrics. 2000; 108:25-30 [Abstract/Free Full Text]
23. Kaplan M Hammerman C. Glucose-6-phosphate dehydrogenase deficient neonates: a potential cause for concern in North America. Pediatrics. 2000; 106:1478-1480 [Free Full Text]
24. Watchko JF. Indirect hyperbilirubinemia in the neonate. In: Maisels MJ, Watchko JF, eds. Neonatal Jaundice. Newark, NJ: Harwood Academic Publishers; 2000:51-66
25. Roy Chowdhury N, Deocharan B, Bejjanki HR, The presence of a Gilbert-type promoter abnormality increases the level of neonatal hyperbilirubinemia. Hepatology. 1997; 26:370A
26. Bhutani VK, Gourley GR, Adler S, et al. Noinvasive measurement of total serum bilirubin in a multiracial predischarge newborn population. Pediatrics. 2000;106(2). URL: http://www.pediatrics.org/cgi/content/full/106/2/e17
27. Maisels MJ, Newman TB Predicting hyperbilirubinemia in newborns: the importance of timing. Pediatrics. 1999; 103:493-495 [Free Full Text]
28. Valaes T. Pharmacological approaches to the prevention and treatment of neonatal jaundice. In Maisels MJ, Watchko JF, eds. Neonatal Jaundice. Newark, NJ: Harwood Academic Publishers; 2000:205-213