Human fetal bilirubin levels and fetal hemolytic disease

Carl P. Weiner, MD

Iowa City, Iowa

The development of secondary fetal anemia in association with maternal red blood cell alloimmunization

requires hemolysis. In specimens obtained at the time of a clinically indicated cordocentesis, total and

direct umbilical venous bilirubin was measured and the indirect umbilical venous bilirubin calculated in 43antigen-positive and 30 control fetuses. Twenty-two (51%) of the antigen-positive fetuses had or

SUbsequently developed severe anemia (hematocrit <30%). Umbilical venous total bilirubin (r = 0.47,P = 0.0008) and direct bilirubin (r = 0.520, P = 0.04) levels each rose with gestation. Indirect bilirubin did

not vary significantly with gestation. Bilirubin was unrelated to hemoglobin. In contrast to the controlfetuses, umbilical venous total bilirubin for antigen-positive fetuses was inversely related to hemoglobin

(r = -0.57, P < 0.0001) independent of gestational age (r = 0.53, P < 0.0001) (multiple R of hemoglobinand gestational age for umbilical venous total bilirubin = 0.76, P < 0.0001). Eighteen of 22 (82%) fetuses

in whom anemia developed had an umbilical venous total bilirubin ;0,97.5 percentile compared with only

eight of 21 (38%) fetuses in whom anemia did not develop (p = 0.009). In longitudinal study the umbilicalvenous total bilirubin frequently rose above normal weeks before the development of anemia. An umbilicalvenous total bilirubin >3 mgldl represented the warning line. Fifteen of 16 (94%) fetuses in whom either

severe antenatal anemia or significant postnatal hyperbilirubinemia developed had an umbilical venoustotal bilirubin >3 mg/dl. We conclude that the normal placental capacity for the transport of fetal bilirubin is

exceeded in the face of enhanced fetal hemolysis. An elevated fetal bilirubin often precedes the

development of antenatal anemia. The antigen-positive fetus with an elevated bilirubin is at high risk todevelop anemia antenatally. (AMJ DasTET GVNECOL 1992;166:1449-54.)

Key words: Fetus, bilirubin, hemolytic anemia, alloimmunization, cordocentesis

Although the fetal response to maternal red bloodcell alloimmunization depends on several factors, in­cluding the quantity of antibody bound to the fetal redblood cell and the rate of red blood cell production,'the development of fetal anemia ultimately requireshemolysis. In a preliminary study our unit observedthat the normal fetal concentration of bilirubin ex­ceeded that of the neonate." This suggested that theplacental capacity for fetal bilirubin clearance was lim­ited. The purposes of this investigation were to deter­mine (l) the concentration of fetal bilirubin across ges­tation, (2) whether antigen-positive fetuses of alloim­munized women have higher bilirubin concentrationsthan control fetuses, and (3) whether a high biliru­bin concentration preceded the development of fetalanemia.

From the Fetal Diagnosis and Treatment Unit, Division ofMaternal­Fetal Medicine, Department of Obstetricsand Gynecology, Universityof Iowa College of Medicine.Supported in part by grant 24494 from the National Institutes ofHealth, National Institute of Child Health and Development.Received for publication May 7, 1991; revised September 6, 1991;accepted November 26, 1991.Reprint requests:Carl P. Weiner, M D, Department of ObstetricsandGynecology, University of Iowa College of Medicine, Iowa City, IA52242.611135284

Material and methods

Cordocentesis was performed in 73 pregnancies withreferral to the University of Iowa Fetal Diagnosis andTreatment Unit. Forty-three antigen-positive fetuses atrisk for hemolytic anemia were sampled (95 specimens)to determine the severity of their disease as describedpreviously.' The remaining 30 fetuses were sampledonce for a variety of indications (most commonly ma­ternal red cell or platelet alloimmunization) and shownto be free of disease.

Laboratory measurements performed in the centralhospital laboratory included total bilirubin, direct bil­irubin, and a complete blood cell count. The indirectbilirubin was calculated. Total and direct bilirubin(mg/dl) were measured in serum by a commerciallyavailable colorimetric assay (Boehringer-Mannheim)on an autoanalyzer (Hitachi 737). Hemolyzed sampleswere excluded. The complete blood cell count was per­formed with an autoanalyzer (Technicon H-l) on bloodanticoagulated with ethylenediaminetetraacetic acid.Because of the limited sample volume, not alI tests couldbe performed on each occasion.

The technique for cordocentesis has previously beendetailed.' Briefly, a 22-gauge spinal needle was held inthe ultrasonographic plane of a 5 MHz sector trans­ducer with a fixed-needle guide (CIVCO, Kalona, la.).

1449

1450 Weiner

TB(mg/dL)

3.0

2.0

1.0

n =30 r =0.473 p =O.0008

TB = 2.53-21.8 /Weeks

May 1992Am J Obstet Gynecol

o18 20 22 24 26 28 30 32 34 36 38

Weeks

Fig. 1. Prediction interval (95%) for umbilical venous serum total bilirubin (TH) in healthy, antigen­negative fetuses (n = 30). There was no relationship between bilirubin and hemoglobin in this group.

The umbilical vein was preferred over the umbilicalartery to minimize the risk of a procedure-relatedbradycardia." Pancuronium (0.3 mg/kg estimated fetalweight) was administered intravenously whenever fetalmovement was felt to pose a risk.

All records were maintained prospectively in a com­puterized data base whose veracity was confirmed ona regular basis. The results are presented as themean ± SD unless otherwise stated. One hundredeleven of 125 (90.2%) specimens were from the um­bilical vein. Only the first measurement from each an­tigen-positive fetus was included in the search for arelationship between bilirubin and both hemoglobinand gestational age. Because the small number of ar­terial specimens precluded an accurate comparison ofarterial and venous samples controlling for gestationalage, the statistical analyses presented here were limitedto specimens from the umbilical vein. The 95% pre­diction interval of bilirubin for control fetuses wasbased on the regression obtained using bilirubin as thedependent variable and gestational age as the inde­pendent variable. Subsequent bilirubin measurementsfrom antigen-positive, not transfused, fetuses weregraphed to reveal the natural history. An analysis ofcovariance was used to compare the bilirubin slopes ofantigen-positive and control fetuses. Receptor-oper­ated curves were applied to determine the umbilicalvenous total bilirubin concentration most predictive ofa fetal or neonatal hemolytic complication. Other anal­yses included independent t test, stepwise multipleregression, and cross tabulation. A p value ~0.05 wasconsidered to indicate a significant difference.

ResultsThe mean gestational age of fetuses at risk for hemo­

lytic anemia was 24.9 ± 3 weeks and ranged from 20to 36 weeks. Twenty-two (51%) developed a hematocrit<30% and were considered candidates for transfusiontherapy.' The mean gestational age of the controlgroup was 27 ± 5 weeks and ranged from 20 to 38weeks.

Umbilical venous total bilirubin (r = 0.47, P<0.0008, n = 30) (Fig. 1) and direct bilirubin (r = 0.52,P = 0.04, n = 15) for the control group rose progres­sively with advancing gestation and were unrelated tothe fetal hemoglobin. Indirect bilirubin did not changesignificantly with gestation (mean 1.4 ± 0.3 mg/dl,n = 15).

In contrast to the control fetuses, umbilical venoustotal bilirubin from antigen-positive fetuses was in­versely related to hemoglobin (r = - 0.57, P< 0.0001)(Fig. 2) independent of gestational age (r = 0.53,P< 0.0001) (multiple R of hemoglobin and gestationalage for umbilical venous total bilirubin = 0.76,P< 0.0001). There was no relationship between he­moglobin and gestational age in the antigen-positivefetuses (r = 0.05, P = 0.75). Although there was a sig­nificant relationship between umbilical venous total bil­irubin and gestational age in both groups, they weredifferent (Fig. 3). Umbilical venous total bilirubin inantigen-positive fetuses rose at a rate significantlygreater (p < 0.01) than that expected for the normalgestational increment. A similar association was alsoobserved for both the direct and indirect bilirubin con­centrations and hemoglobin and gestational age (Table

Volume 166Number 5

Fetal bilirubin 1451

uvTB = 4.932 - 0.249 * Hb (n= 41)

uvTBmg/dL

7.57

6.56

5.55

4.54

3.53

2.52

1.51

.5o

o

.~ .::.:v ..

5

Hb (g/dL)10 15

Fig. 2. Relationship between total umbili cal veno us bilirubin (uvTB) and hem oglobin (H b) in ant igen­positive fetuses. The higher the bilirubin concentration, the lower the fetal hemoglobin. Note thattotal bilirubin of 3 mg/d l approximates hemoglobin of 8 gm /d l (two missing values).

4.0 r-----------------....,

T.B.mg/dL

3.5

3.0

2.5

2.0

1.5

1.0

.5

-Control

- - Antigen positive

o20 22 24 26 28

Weeks30 32 34

Fig. 3. Regression lines for umbilical venous total bilirubin (T.8 .) and gestati onal age in contro l andantigen-positive groups. Antigen-positive fetuses have steeper slope. This demonstrates that severi tyof fetal hemolytic d isease was progressive.

I). Because the relationships between indirect and di­rect bilirubin for hemoglobin in antigen-positive fetu seswere similar to tha t for umbilical venous total bilirubin,we ha ve focused on the latter because of the largersample size.

The umbilical veno us total bilirubin exceeded the97.5 per centile in 18 of 22 (82%) of ant igen-positivefetuses who de velop ed a hematocrit <30%. In eight of10 fetus es with two or more samples, the elevation pre­ceded the development of anemia by several weeks(Fig. 4). In contrast, only eight of 21 (38%) of fetus eswho did not develop severe anemia had an umbili cal

venous total bilirubin above the 97.5 percentile at anytime before delivery (p = 0.009 from fetu ses who re­qu ired tra nsfusion) (Fig. 5). Overall, the positive pre­dictive value of a high umbilical veno us total bilirubinfor the development of severe fetal hemolytic anemiawas 69% (18 of 26), and the negative predictive valueof a low umbili cal veno us total bilirubin for the devel­opment of severe anemia was 76% (13 of 17).

We sought to determ ine whether the clinical appli­cation of the umbili cal venous total bilirubin measure­ment could be further simplified by the identificationof a warning line . An umbilical venou s total bilirubin

1452 Weiner May 1992Am J Obstet Gynecol

36343228 30

Weeks262422

o20

10.0

- REGRESS

8.00

6.0T.e.

mg/dL4.0

00

2.0

Fig. 4. Serial umbilicalvenous total bilirubin (T.B.) concentrations in fetuses who developed severehemolytic anemia. Last sample (asterisk) indicates that hematocrit had declined below 30%. Singlesamples indicate that initial hematocrit was <30% first time bilirubin was measured. Shaded areareflects95% prediction interval for normal range. Note that elevated umbilical venous total bilirubinoften preceded development of anemia severe enough to require transfusion.

-0.57 <0.00010.53 <0.0001

39 0.68 <0.0001-0.53 0.0001

0.46 0.000739 0.74 <0.0001

-0.56 <0.00010.52 <0.0001

Table I. Relationship between bilirubin,hemoglobin, and gestation in first sample fromantigen-positive fetuses

Bilirubin group ~ p C=::::=I p

Total 43 0.76 <0.0001HemoglobinWeeks' gestation

DirectHemoglobinWeeks' gestation

IndirectHemoglobinWeeks' gestation

>3 mg/dl exceeded the 97.5 percentile for all gesta­tional ages. Six fetuses already had severe anemia whenfirst sampled: five of the six (83%) had an umbilicalvenous total bilirubin ?3 mg/dl. The umbilical venoustotal bilirubin ultimately rose above 3 mg/dl in an­other nine fetuses whose initial hematocrit was >30%and whose initial umbilical venous total bilirubin<3 mg/dl. Seven of these nine (78%) developed severehemolytic anemia antenatally. The two fetuses who didnot develop anemia antenatally experienced severe hy­perbilirubinemia postnatally (one required two double­volume exchange transfusions and the other required10 days of phototherapy). Fetuses whose umbilical ve­nous total bilirubin before transfusion was <3 mg/dlhad lower reticulocyte counts than fetuses whose um-

bilical venous total bilirubin was ?3 mg/dl (10.5% ±

4.2% vs 24.7% ± 12.6%, P = 0.0003).We previously described four hematologic patterns

predicting the severity of fetal hemolytic disease on thebasis of fetal hematocrit, direct Coombs test, and retic­ulocyte count. 1 These patterns were prospectively as­signed to each of the 14 fetuses who had or developedan umbilical venous total bilirubin >3 mg/dl. Two fe­tuses were assigned pattern 1 (not at risk to developsevere anemia antenatally), three were pattern 2 (in­termediate risk for anemia), and nine were pattern 3(high risk). Each of three with pattern 2 and eight ofthe nine (89%) with pattern 3 developed severe hemo­lytic anemia. Of the remaining three fetuses, two ex­perienced significant hyperbilirubinemia postnatally(as noted in the above paragraph). One of those twofetuses had been assigned pattern 1 and the other pat­tern 3.

Comment

We have examined the effect of gestational age andfetal hemolytic disease on the concentration of umbil­ical venous bilirubin. Normal fetal umbilical venous to­tal and direct bilirubin (not shown) rose similarly withadvancing gestational age. On the basis of the scatter­gram, it is likely that with a larger sample size we wouldhave also observed a rise in the indirect bilirubin con­centration. Although the number of measurementsperformed on specimens from the umbilical artery wasinadequate to assess placental clearance of bilirubin di-

Volume 166Number 5

Fetal bilirubin 1453

5.0

- REGRESS

4.0

•3.0

T.B.mg/dL

2.0

1.0

020 22 24 26 28 30 32 34 36

Weeks

Fig. 5. Serial umbilical venous total bilirubin (T.R.) measurements of fetuses whose hematocritremained above 30%. Shaded area indicates 95% prediction interval for normal umbilical venoustotal bilirubin. The two fetuses marked by asterisk developed significant postnatal hyperbilirubinemia.

rectly, the observation that normal neonatal bilirubinis much lower" than that of the near-term fetus suggeststhat placental capacity to transport the normal pro­duction of fetal bilirubin to the mother is relatively low.

Moise et aJ.5 reported a single alloimmunized fetuswhose serum was icteric and speculated that it reflectedhemolysis." In antigen-positive fetuses, there was aninverse relationship between hemoglobin and bilirubinindependent of gestational age. The high concentrationof bilirubin in the sera of antigen-positive fetuses wasnot likely secondary to decreased placental clearancebecause there was a significant inverse correlation be­tween fetal hemoglobin and bilirubin in the antigen­positive group that was absent in the control group.This suggests that the greater the quantity of red bloodcell hemolysis, the lower the hemoglobin.

The use of fetal blood instead of amniotic fluid forthe evaluation of maternal red cell alloimmunization isa new concept not yet widely accepted. This unit haspreviously demonstrated that the antigen-positive fetusat risk for the development of severe hemolytic anemiacan be prospectively identified on the basis measure­ments of fetal hematocrit, reticulocyte count, andstrength of the direct Coombs test. I On the basis ofthese three tests, the antigen-positive fetus can be as­signed to one of four hematologic risk patterns. Thehigher the pattern number, the greater the risk of thefetus developing a hematocrit <30%. The current in­vestigation suggests that the addition of umbilical ve­nous total bilirubin measurement to that battery of testsmay be of value, especially in instances where the as­signed pattern suggests the fetus is at no or low risk

for the development of severe hemolytic anemia beforebirth. In our experience, only one of five fetuses withpattern 2 will become severely anemic antenatally. Inthe current series each of the three fetuses with bothpattern 2 and an umbilical venous total bilirubin level>3 mg/dl developed severe hemolytic anemia. Andalthough no fetus with both pattern 1 and an umbilicalvenous total bilirubin level >3 mg/dl developed severeanemia, one of the two did require 10 days of photo­therapy postnatally.

Not all fetuses who became anemic had a high bili­rubin level. This may reflect, in part, variability of pla­cental bilirubin transport. However, there is an addi­tional explanation. We have previously demonstratedthat a sizable group of antigen-positive fetuses who be­come anemic do so gradually in association with a lowreticulocyte count and a weakly positive direct Coombstest.' In the current investigation, fetuses whose um­bilical venous total bilirubin level was <3 mg/dl beforetransfusion had significantly lower reticulocyte countsthan fetuses whose level was :::::3 mg/dl. This suggeststhat the degree of hemolysis in some anemic fetuses isrelatively low but that they fail to respond to red bloodcell destruction with an adequate erythropoietic re­sponse. This observation would also explain why theamniotic fluid ~ optical density 450 of anemic fetusesis not always elevated.

In conclusion, the ability of the human placenta toclear fetal bilirubin is limited. Because of this limitation,fetal hemolytic disease is associated with fetal hyper­bilirubinemia. The higher the bilirubin concentration,the lower the hemoglobin concentration. An elevated

1454 Weiner

fetal serum bilirubin indicates that the fetus is at highrisk to develop severe hemolytic anemia antenatally. Itmay also provide additional information regarding therisk of postnatal hyperbilirubinemia.

REFERENCES

1. Weiner CP,Williamson RA, Wenstrom KD, Sipes SL, GrantSS, Widness J A. Management of fetal hemolytic disease bycordocentesis: I. Prediction of fetal anemia. AM J OBSTETGYNECOL 1991;165:546-53.

2. Wenstrom KD, Weiner CP, Williamson RA. Effect of ges­tational age and disease on fetal liver function [Abstract

May 1992Am J Obstet GynecoJ

199]. In : Proceedings of the tenth annual meeting of theSociety of Perinatal Obstetricians, Houston,Texas,January23-27, 1990. Houston: Society of Perinatal Obstetricians,1990.

3. Weiner CPoCordocentesis for diagnostic indications-twoyears' experience. Obstet Gynecol 1987;70:664-8.

4. Weiner CP, Williamson RA, Wenstrom KD, et al. Manage­ment of fetal hemolytic disease by cordocentesis. II. Out­come of treatment. AM J OBSTET GYNECOL 1991;165:1302-7.

5. Moise KJ, Kirshon B, Carpenter RJ, Deter RL. Icteric se­rum as an indicator of fetal vascular access during intra­vascular transfusion for hemolytic disease. Am J Perinatol1990;7:279-80.

Make researchingeasier with the Ten-Year Cumulative Index (1978-1987) to the AMER­

ICAN JOURNAL OFOBSTETRICS AND GYNECOLOGYThe Ten-Year Cumulative Index is a complete guide to more than 30,000 pages of

original articles, case reports, letters, and editorials published in the AMERICAN JOURNALOF OBSTETRICS AND GYNECOLOGY from 1978 through 1987 (volumes 130 to 157). Num­bering more than 800 pages, the Cumulative Index includes both a subject and an authorindex. This hard-cover volume is $64.50 ($69.00 international). Prices include shipping.Payment mustaccompany all orders and must be in U.S. funds, drawn on a U.S. bank.

Contact Mosby-Year Book, Inc., Subscription Services, 11830 Westline IndustrialDrive, St. Louis, MO 63146-3318 USA; phone (314) 453-4394. In the U.S., call toll-free:1-800-325-4177, ext. 4394 . Please allow six weeks for delivery.