The presence of unexpected alloantibodies (antibodies against red cell antigens) in patient serum found in pretransfusion screening of recipients is 0.7%-1.6%. This averages 9% (range, 6%-36%) in multitransfused patients. Infants less than 4 months old usually do not form alloantibodies against transfused red cell antigens that they lack. After that, age per se does not appear to affect red blood cell (RBC) antigen sensitization. Immunosuppressive therapy can diminish this response. In the case of Rh system D (Rho) antigen, chance of sensitization has some correlation with antigen dose, but this is not exact or linear. Sensitization of D-negative recipients of D-positive cells has ranged from 8% to 70%. Although antibodies to bacterial or many other antigens usually appear in 7-21 days, alloantibodies usually take 3-4 months after transfusion with a minimum (in one study) of 1 month. Once formed, the antibodies remain detectable for variable periods of time, depending to some extent (but not entirely) on the particular antibody. Anti-D is particularly likely to be detectible for many years; anti-C and anti-Kidd are more likely to become nondetectable (50% loss in 5 years in one study). However, nondetectable antibodies can be reactivated by anamnesthic antigen exposure.

Hemolytic reactions may be caused by either complete or incomplete antibodies. In reactions caused by complete antibodies, such as occur in the ABO blood group system, there is usually intravascular hemolysis. The amount of hemolysis depends on several factors, such as quantity of incompatible blood, antibody titer, and the nature of the antibody involved. However, there is an element of individual susceptibility; some patients die after transfusion of less than 100 ml of incompatible blood, whereas others survive transfusion of several times this amount. The direct Coombs’ test result is often positive, but this depends on whether all the RBCs attacked by the complete antibody have been lysed, whether more antibody is produced, and, to some extent, on how soon the test is obtained. If the sample is drawn more than 1 hour after the ABO transfusion reaction is completed, the chance of a positive direct Coombs’ test result is much less. A Coombs’ reagent acting against both gamma globulin and complement (broad spectrum) is needed; most laboratories now use this type routinely. Free hemoglobin is released into the plasma from lysed RBCs and is carried to the kidneys, where it is excreted in the urine. Some of the intravascular hemoglobin is converted to bilirubin in the reticuloendothelial system so that the serum nonconjugated (indirect) bilirubin level usually begins to increase. In reactions caused by incomplete antibodies, such as the Rh system, there is sequestration of antibody-coated cells in organs such as the spleen, with subsequent breakdown by the reticuloendothelial system. With incomplete antibody reactions, RBC breakdown is extravascular rather than intravascular; in small degrees of reaction, plasma free hemoglobin levels may not rise, although indirect bilirubin levels may eventually increase. In more extensive reactions, the plasma hemoglobin level is often elevated, although the elevation is sometimes delayed in onset. The direct Coombs’ test should be positive in hemolytic reactions due to incomplete antibodies (unless all affected RBCs have been destroyed).

Undesirable effects of blood or blood product transfusion

RECIPIENT REACTION TO DONOR ANTIGENS ON DONOR CELLS
Clinical types of reactions
Hemolytic reaction
Nonhemolytic febrile reactions
Allergic reactions
Cytopenic reactions
Tissue-organimmunologic reactions
Anaphylactic reactions

ANTIGEN GROUPS INVOLVED
ABO system
Rh and minor blood groups
Histocompatibility leukocyte antigen (HLA) system
Platelet antigens

INFECTIONS
Hepatitis viruses
Human immunodeficiency virus-1 and 2 (HIV-1 and 2)
Human T-cell lymphotropic virus-1 and 2 (HTLV-1 and II)
Cytomegalovirus (CMV)
Epstein-Barr virus
Syphilis
Malaria
Other

OTHER TRANSFUSION PROBLEMS
Citrate overload
Hyperkalemia
Depletion of coagulation factors
Depletion of platelets
Transfused blood temperature
Donor medications

Hemolytic transfusion reactions are usually caused by incompatible blood but are occasionally caused by partial hemolysis of the RBCs before administration, either before leaving the blood bank or just before transfusion if the blood is warmed improperly. Analysis of fatal cases of hemolytic transfusion reaction reported to the Food and Drug Administration in 1976-1983 shows that almost 65% were due to problems of mistaken identity and about 9% were due to clerical error. Only about 18% were due to error while performing blood bank tests. Frequent mistakes included transfusion into one patient of blood meant for another patient, obtaining a recipient crossmatch specimen from the wrong patient, mixup of patient crossmatch specimens in the blood bank, and transcription of data belonging to one patient onto the report of another.

The great majority of hemolytic transfusion reactions occur during transfusion. Symptoms and laboratory evidence of hemolysis usually are present by the time transfusion of the incompatible unit is completed, although clinical symptoms or laboratory abnormalities may sometimes be delayed for a few minutes or even a few hours. Occasionally, hemolytic reactions take place after completion of transfusion (delayed reaction). In one kind of delayed reaction, the reaction occurs 4-5 days (range, 1-7 days) following transfusion and is due to anamnestic stimulation of antibodies that were already present but in very low titer. The Kidd (Jk) system is frequently associated with this group. The intensity of the reaction may be mild or may be clinically evident, but most are not severe. In a second kind of delayed reaction, the reaction occurs several weeks after transfusion and is due to new immunization by the transfused RBC antigens with new antibody production. This type of reaction is usually mild and subclinical. In both types of reaction, and occasionally even in an immediate acute reaction, a hemolytic reaction may not be suspected and the problem is discovered accidentally by a drop in hemoglobin level or by crossmatching for another transfusion.

Symptoms of hemolytic transfusion reaction include chills, fever, and pain in the low back or legs. Jaundice may appear later. Severe reactions lead to shock. Renal failure (acute tubular necrosis) is common due either to shock or to precipitation of free hemoglobin in the renal tubules. Therefore, oliguria develops, frequently accompanied by hemoglobinuria, which is typically manifest by red urine or nearly black “coffee ground” acid hematin urine color.

Tests for hemolytic transfusion reaction

These tests include immediate rechecking and comparison of patient identification and the unit or units of transfused donor blood. Tubes of anticoagulated and nonanticoagulated blood should be drawn. The anticoagulated tube should be centrifuged immediately and the plasma examined visually for the pink or red color produced by hemolysis. A direct Coombs’ test and crossmatch recheck studies should be performed as soon as possible. A culture and Gram stain should be done on the remaining blood in the donor bag.

Plasma free hemoglobin. The presence of free hemoglobin in plasma is one of the most valuable tests for diagnosis of acute hemolytic transfusion reaction. Most cases of severe intravascular hemolysis due to bivalent antibodies such as the ABO group produce enough hemolysis to be grossly visible. (Artifactual hemolysis from improperly drawn specimens must be ruled out.) If chemical tests are to be done, plasma hemoglobin is preferred to serum hemoglobin because less artifactual hemo-lysis takes place before the specimen reaches the laboratory. Although hemolytic reactions due to incomplete antibodies (such as Rh) have clinical symptoms similar to those of ABO, direct signs of hemolysis (e.g., free plasma hemoglobin) are more variable or may be delayed for a few hours, although they become abnormal if the reaction is severe. Another drawback in the interpretation of plasma hemoglobin values is the effect of the transfusion itself. The older erythrocytes stored in banked blood die during storage, adding free hemoglobin to the plasma. Therefore, although reference values are usually considered to be in the range of 1-5 mg/100 ml, values between 5 and 50 mg/100 ml are equivocal if stored blood is given. Hemolysis is barely visible when plasma hemoglobin reaches the 25-50 mg/100 ml range. Frozen blood has a greater content of free hemoglobin than stored whole blood or packed cells and may reach 300 mg/100 ml.

Additional tests. If visual inspection of patient plasma does not suggest hemolysis and the direct Coombs’ test result is negative, additional laboratory tests may be desirable to detect or confirm a hemolytic transfusion reaction. Procedures that may be useful include serum haptoglobin measurement, both immediately and 6 hours later; serum nonconjugated (“indirect”) bilirubin measurement at 6 and 12 hours; and urinalysis.

A pretransfusion blood specimen, if available, should be included in a transfusion reaction investigation to provide baseline data when the various tests for hemolysis are performed.

Urinalysis. Urine can be examined for hemoglobin casts, RBCs, and protein and tested for free hemoglobin. RBCs and RBC casts may be found in hemolytic reactions since the kidney is frequently injured. If no hematuria or free hemoglobin is found, this is some evidence against a major acute hemolytic transfusion reaction. Abnormal findings are more difficult to interpret unless a pretransfusion urinalysis result is available for comparison, since the abnormality might have been present before transfusion. In addition, abnormal urine findings could result from hypoxic renal damage due to an episode of hypotension during surgery rather than a transfusion reaction.

Hemoglobin passes through the glomerular filter into the urine when the plasma hemoglobin level is above 125 mg/100 ml. Therefore, since hemolysis should already be visible, the major need for urine examination is to verify that intravascular hemolysis occurred rather than artifactual hemolysis from venipuncture or faulty specimen processing. Conversely, unless the plasma free hemoglobin level is elevated, urine hemoglobin may represent lysed RBCs from hematuria unrelated to a hemolytic reaction. However, if sufficient time passes, the serum may be cleared of free hemoglobin while hemoglobin casts are still present in urine. Hemosiderin may be deposited in renal tubule cells and continue to appear in the urine for several days.

Serum haptoglobin. Haptoglobin is an alpha globulin that binds free hemoglobin. Two types of measurements are available: haptoglobin-binding capacity and total haptoglobin. Binding capacity can be measured by electrophoresis or chemical methods. Total haptoglobin is usually estimated by immunologic (antibody) techniques; a 2-minute slide test is now commercially available. The haptoglobin-binding capacity decreases almost immediately when a sufficient quantity of free hemoglobin is liberated and remains low for 2-4 days. Stored banked blood, fortunately, does not contain enough free hemoglobin to produce significant changes in haptoglobin-binding capacity. The total haptoglobin level decreases more slowly than the binding capacity after onset of a hemolytic reaction and might not reach its lowest value until 6-8 hours later. Haptoglobin assay is much less helpful when frozen RBCs are transfused because of the normally increased free hemoglobin in most frozen RBC preparations. Bilirubin determinations are not needed if haptoglobin assay is done the same day.

In common with other laboratory tests, serum haptoglobin levels may be influenced by various factors other than the one for which the test is ordered. Haptoglobin levels may be decreased by absorption of hemoglobin from an extravascular hematoma and also may be decreased in severe liver disease, hemolytic or megaloblastic anemia, estrogen therapy, and pump-assisted open-heart cardiac surgery. Haptoglobin is one of the body’s “acute-phase reactant” proteins that are increased by infection, tissue injury or destruction, widespread malignancy, and adrenocorticosteroid therapy. Therefore, a mild decrease in haptoglobin level could be masked by one of these conditions. In order to aid interpretation, it is helpful to perform the haptoglobin assay on a pretransfusion serum specimen as well as the posttransfusion specimen.

Sensitivity of tests in hemolytic reaction. Data regarding frequency of test abnormality in transfusion reaction are difficult to obtain. In one series, comprising predominantly hemolytic reactions not due to ABO antibodies, free hemoglobin was detected in plasma or urine in 88% of cases involving immediate reactions and in 52% of cases involving incomplete antibodies or delayed reactions. Serum haptoglobin levels were decreased in 92% of the patients in whom it was assayed. Various factors influence this type of data, such as the amount of incompatible blood, the antibody involved, the time relationship of specimen to onset of reaction, and the test method used.

Nursing station action in possible hemolytic reaction

Transfusion should be stopped at the first sign of possible reaction and complete studies done to recheck compatibility of the donor and recipient blood. If these are still satisfactory, and if results of the direct Coombs’ test and the studies for intravascular hemolysis are negative, a different unit can be started on the assumption that the symptoms were pyrogenic rather than hemolytic. Whatever unused blood remains in the donor bottle, the donor bottle pilot tube, and a new specimen drawn from the patient must all be sent to the blood bank for recheck studies. Especially dangerous situations exist in transfusion during surgery, where anesthesia may mask the early signs and symptoms of a reaction. Development during surgery of a marked bleeding or oozing tendency at the operative site is an important danger signal. A hemolytic transfusion reaction requires immediate mannitol or equivalent therapy to protect the kidneys.

Hemolytic disease of the newborn (HDN)

The other major area where blood banks meet blood group hemolytic problems is in hemolytic disease of the newborn (HDN). It may be due to ABO, Rh, or (rarely) minor group incompatibility between fetal and maternal RBCs. ABO and the Rh antigen D (Rho) are by far the most common causes. The Rh antigen c and the blood group Kell antigen are next most important. Hemolytic disease of the newborn results from fetal RBC antigens that the maternal RBCs lack. These fetal RBC antigens provoke maternal antibody formation of the IgG type when fetal RBCs are introduced into the maternal circulation after escaping from the placenta. The maternal antibodies eventually cross the placenta to the fetal circulation and attack the fetal RBCs.

Hemolytic disease of the newborn due to Rh. Hemolytic disease of the newborn due to Rh incompatibility varies in severity from subclinical status, through mild jaundice with anemia, to the dangerous and often fatal condition of erythroblastosis fetalis. The main clinical findings are anemia and rapidly developing jaundice. Reticulocytosis over 6% accompanies the anemia, and the jaundice is mainly due to unconjugated (indirect) bilirubin released from the reticuloendothelial sequestration and destruction of RBCs. The direct Coombs’ test result is positive. In severe cases there are usually many nucleated RBCs in the peripheral blood. Jaundice is typically not present at birth except in very severe cases (since the mother excretes bilirubin produced by the fetus) but develops several hours later or even after 24 hours in mild cases. Diseases that cause jaundice in the newborn, often accompanied by anemia and sometimes a few peripheral blood nucleated RBCs, include septicemia, cytomegalic inclusion disease, toxoplasmosis, and syphilis. Physiologic jaundice of the newborn is a frequent benign condition that may be confused with hemolytic disease or vice versa. There is, however, no significant anemia. A normal newborn has a (average) hemoglobin value of 18 gm/100 ml, and a value less than 15 gm/100 ml (150 g/L) indicates anemia. Anemia may be masked if heelstick (capillary) blood is used, since neonatal capillary hemoglobin values are higher than venous blood values.

Hemolytic disease due to Rh incompatibility occurs in an Rh-negative mother whose fetus is Rh positive. Usually, the mother and fetus are ABO compatible. The mother develops antibodies against the RBC Rh antigen after being exposed to Rh-positive RBCs. This may occur due to pregnancy, abortion, ectopic pregnancy, amniocentesis (or other placental trauma), blood transfusion with Rh-positive RBCs, or transfusion of certain RBC-contaminated blood products such as platelets. The most common maternal contact with Rh-positive RBCs occurs during pregnancy when fetal RBCs escape through the placenta into the maternal circulation. This may happen at any time after the 16th week of pregnancy, and both the quantity of cells and the frequency of exposure increase until delivery. The largest single dose of fetal RBCs occurs during delivery. Not all mothers have detectable fetal RBCs in their circulation, and of those who do, not all become sensitized during any one pregnancy. There is approximately a 10%-13% risk of sensitization in previously nonsensitized Rh-incompatible pregnancies. When fetal-maternal ABO incompatibility (with the mother being group O) is present, the usual risk for Rh sensitization is decreased, presumably because sufficient fetal cells are destroyed that the stimulus for sensitization is reduced below the necessary level.

The first child is usually not affected by Rh hemolytic disease if the mother has not been exposed to Rh-positive RBCs before pregnancy, and full sensitization usually does not develop until after delivery in those mothers who become sensitized. However, occasional firstborn infants are affected (5%-10% of HDN infants) either because of previous maternal exposure (e.g., a previous aborted pregnancy) or because of unusually great maternal susceptibility to Rh stimulus during normal pregnancy. Once maternal sensitization takes place, future exposure to Rh antigen, as during another pregnancy with an Rh-positive fetus, results in maternal antibody production against the Rh antigen, which can affect the fetus.

Current recommendations of the American College of Obstetricians and Gynecologists (ACOG) are that every pregnant woman should have ABO and Rh typing and a serum antibody screen as early as possible during each pregnancy. If results of the antibody screen are negative and the mother is Rh positive, the antibody screen would not have to be repeated before delivery. Theoretically, if the mother is Rh positive, or if the mother is Rh negative and the father is also Rh negative, there should be no risk of Rh-induced fetal disease. However, the antibody screen is still necessary to detect appearance of non-Rh antibodies. If results of the antibody screen are negative and the mother is Rh negative, the father should be typed for Rh and the antibody screen should be repeated at 28 weeks’ gestation. If the antibody screen results are still negative, a prophylactic dose of Rh immune globulin is recommended (discussed later). If the antibody screen detects an antibody, subsequent testing or action depends on whether the antibody is Rh or some other blood group and whether or not this is the first pregnancy that the antibody was detected. The father should be tested to see if he has the antigen corresponding to the antibody. If the antibody is one of the Rh group, antibody titers should be performed every 2 weeks. Titers less than 1:16 suggest less risk of fetal hazard. Titer equal to or greater than 1:16 is usually followed by amniocentesis (as early as 24 weeks’ gestation) for spectrophotometric examination of amniotic fluid bilirubin pigment density. The greater the pigment density the greater the degree of fetal RBC destruction. Antibodies that are not Rh are managed by amniotic fluid examination without antibody titers, since there is inadequate correlation of titers to fetal outcome.

Rh immune globulin. Studies have now proved that nearly all nonsensitized mothers with potential or actual Rh incompatibility problems can be protected against sensitization from the fetus by means of a “vaccine” composed of gamma globulin with a high titer of anti-Rh antibody (RhIg). This exogenous antibody seems to prevent sensitization by destroying fetal RBCs within the mother’s circulation and also by suppressing maternal antibody production. Abortion (spontaneous or induced), ectopic pregnancy, amniocentesis, and platelet or granulocyte transfusions (which may be contaminated with Rh-positive RBCs) may also induce sensitization and should be treated with RhIg. Current ACOG recommendations are to administer one unit (300 µg) of RhIg to all Rh-negative women at 28 weeks’ gestation and a second unit within 72 hours after delivery. A minidose of 50 µg is often used rather than the standard dose when abortion occurs before 12 weeks’ gestation. Although 72 hours after delivery is the recommended time limit for RhIg administration, there is some evidence that some effect may be obtained up to 7 days after delivery. Untreated Rh-negative women have about a 10% (range, 7%-13%) incidence of Rh antibody production (sensitization). When RhIg is administered postpartum it reduces incidence to about 1%-2%, and antepartum use combined with postpartum use reduces the incidence to about 0.1%-0.2%. RhIg will produce a positive Rh antibody screening test result if present in sufficient titer, so that antibody screening should be performed before administration rather than after. The half-life of exogenous gamma globulin (which applies to RhIg) is about 25 days (range 23-28 days). Injected RhIg may be detected in maternal blood after intramuscular injection within 24-48 hours with a peak at about 5 days, and is often still detectable for 3 months (sometimes as long as 6 months). Detection after 6 months postpartum suggests patient sensitization (failure of the RhIg). There have been a few reports that some Rh-positive infants whose mother received antepartum RhIg had a weakly positive direct Coombs’ test result at birth due to the RhIg. The blood bank should always be informed if RhIg has been given antepartum to properly interpret laboratory test results. The standard dose of 300 µg of RhIg neutralizes approximately 15 ml of Rh-positive RBCs (equivalent to 30 ml of fetal whole blood). In some patients the RBCs received from the fetus may total more than 15 ml.

There are several methods used to quantitate fetal RBCs in the maternal circulation, none of which are considered ideal. The Kleihauer acid elution test is a peripheral smear technique that stains the hemoglobin F (Hb F) of fetal RBCs. Other tests that have been used are the weak D (Du) test read microscopically (to detect the mixed field agglutination of the relatively small number of fetal RBCs involved) and the RhIg crossmatch technique. None of these three procedures has proved adequately sensitive. For example, a proficiency survey in 1980 found that about 10% of the laboratories using acid elution or microscopic Du techniques failed to detect the equivalent of 30 ml of fetal RBCs (twice the significant level). When a normal blood sample was tested, about 10% of those using microscopic Du obtained false positive results, and those using acid elution had 40% or more false positive results (especially with one commercial adaptation of the acid elution technique called Fetaldex). Also, false positive acid elution test results can be produced if increased maternal Hb F is present, as seen in beta thalassemia minor and in hereditary persistence of fetal hemoglobin. Newer procedures, such as the erythrocyte rosette test, have proved much more sensitive and somewhat more reproducible. However, the rosette test is not quantitative, so that screening is done with the rosette test (or equivalent) and a positive result is followed by quantitation with the acid elution procedure. If the quantitative test for fetal RBCs indicates that more than 15 ml is present, additional RhIg should be administered. If this is not done, failure rates for RhIg of 10%-15% have been reported; if it is done, the failure rate is approximately 2%. The current American Association of Blood Banks (AABB) recommendation is to administer twice the dose of RhIg indicated by formulas, depending on the percent of fetal RBCs detected by acid elution methods. This is done because of variation above and below the correct result found on proficiency surveys.

ABO-induced hemolytic disease of the newborn

Fifty percent or more of HDN is due to ABO incompatibility between mother and fetus. There usually is no history of previous transfusion. Most cases are found in mothers of blood group O with group A or B infants. Infant anti-A and anti-B production begins between 3 and 6 months after birth. Until then, ABO antibodies in the infant’s serum originate from the mother. If the mother possesses antibodies to the A or B RBC antigens of the fetus, hemolytic disease of the fetus or newborn may result, just as if the fetus or newborn had received a transfusion of serum with antibodies against his or her RBCs. Nevertheless, although 20%-25% of all pregnancies display maternal-fetal ABO incompatibility, only about 40% of these infants develop any evidence of hemolytic disease. In those who do, the condition is usually clinically milder than its counterpart caused by Rh incompatibility with only 4%-11% displaying clinical evidence of disease. There usually are no symptoms at birth. Some infants, however, will suffer cerebral damage or even die if treatment is not given. Therefore, the diagnosis of ABO disease and its differential diagnosis from the other causes of jaundice and anemia in the newborn are of great practical importance.

There are two types of ABO antibodies: the naturally occurring complete saline-reacting type discussed earlier and an immune univalent (incomplete) type produced in some unknown way to fetal A or B antigen stimulation. In most cases of clinical ABO disease the mother is group O and the infant is group A or B. The immune anti-A or anti-B antibody, if produced by the mother, may cause ABO disease because it can pass the placenta. Maternal titers of 1:16 or more are considered dangerous. The saline antibodies do not cross the placenta and are not significant in HDN.

Results of the direct Coombs’ test done on the infant with ABO hemolytic disease are probably more often negative than positive, and even when positive the test tends to be only weakly reactive. Spherocytes are often present but the number is variable. Good evidence for ABO disease is detection of immune anti-A or anti-B antibodies in the cord blood of a newborn whose RBCs belong to the same blood group as the antibody. Detection of these antibodies only in the serum of the mother does not conclusively prove that ABO disease exists in the newborn.

Laboratory tests in hemolytic disease of newborns (HDN)

When an infant is affected by Rh-induced HDN, the result of the direct Coombs’ test on cord blood is nearly always positive. In ABO-induced HDN the direct Coombs’ test result on cord blood is frequently (but not always) positive. The direct Coombs’ test result on infant peripheral blood is usually positive in Rh-induced disease but is frequently negative in ABO-induced disease, especially when done more than 24 hours after delivery. The direct Coombs’ test should be performed in all cases of possible HDN, because incomplete antibodies occasionally coat the surface of fetal RBCs to such an extent as to interfere with proper Rh typing. Cord blood bilirubin is usually increased and cord blood hemoglobin is decreased in severe HDN. There is disagreement whether cord blood hemoglobin level has a better correlation with disease severity than the cord blood bilirubin or infant venous or capillary hemoglobin level. Infant hemoglobin levels tend to be higher than cord hemoglobin levels if blood from the cord and placenta is allowed to reach the infant after delivery.

Laboratory criteria for hemolytic disease of the newborn

Infants with HDN can frequently be saved by exchange transfusion. Commonly accepted indications for this procedure are the following:

1. Infant serum indirect bilirubin level more than 20 mg/100 ml (342 µmol/L) or, in considerably premature or severely ill infants, 15 mg/100 ml (257 µmol/L).
2. Cord blood indirect bilirubin level more than 3 mg/100 ml (51 µmol/L) (some require 4 mg/100 ml).
3. Cord blood hemoglobin level less than 13 gm/dl (130 g/L)(literature range, 8-14 gm/100 ml).
4. Maternal Rh antibody titer of 1:64 or greater, although this is not an absolute indication if the bilirubin does not rise very high.

Bilirubin levels in hemolytic disease of the newborn. Most infants with HDN can be treated effectively enough that exchange transfusion is not needed. The level of infant bilirubin is generally used as the major guideline for decision and to monitor results of other treatment such as phototherapy. An infant total bilirubin level of 12-15 mg/100 ml (205 µmol/L) depending to some extent on the clinical situation (degree of prematurity, presence of anemia or infection, severity of symptoms, etc.) is the most commonly accepted area at which therapy is begun. However, there is surprising variation in the levels quoted in various medical centers. Rarely, kernicterus has been reported in seriously ill infants at bilirubin levels near 10 mg/100 ml (and in one case even as low as 6 mg/100 ml). The rapidity with which the bilirubin level rises is also a factor, as noted previously. To further complicate matters, autopsy studies have shown yellow staining of brain tissue typical of kernicterus in some infants who did not have clinical symptoms of kernicterus.

Bilirubin levels and kernicterus. The most feared complication of Rh-induced hemolytic disease of the newborn is kernicterus, defined as bilirubin staining of the central nervous system (CNS) basal ganglia with death or permanent neuro-logic or mental abnormalities. When this syndrome was first studied in the 1950s, Rh-induced hemolytic disease was the usual etiology, and the nonconjugated bilirubin level of 20 mg/100 ml in term infants (15 mg/dl in premature infants) was established as the level at which the kernicterus syndrome was most likely to develop and thus the level at which exchange transfusion was required. It was also reported that various other factors, such as acidosis, respiratory distress, infection, and very low birth weight could be associated with the kernicterus syndrome at bilirubin levels less than 15 mg/100 ml (several case reports included a few infants whose total bilirubin level was as low as 9-10 mg/100 ml). Eventually the nonconjugated bilirubin level rather than infant symptoms became the center of attention (however, since neonatal bilirubin except in rare cases is almost all nonconjugated, total bilirubin level is routinely assayed instead of nonconjugated bilirubin). As time went on, the advent of RhIg therapy markedly reduced Rh hemolytic disease, and neonatal jaundice became over 90% nonhemolytic. More recent studies have questioned the relationship between total bilirubin level and the kernicterus syndrome. Although phototherapy can reduce total bilirubin levels, there is some question whether in fact this can prevent kernicterus. Therefore, in the early 1990s there is very low incidence of the kernicterus syndrome, the mechanisms and pathologic basis for this syndrome is uncertain, the relationship and interpretation of nonconjugated or total bilirubin levels is being questioned, and the classic guidelines for therapy are being disputed. Nevertheless, while the situation is unclear, the majority of investigators appear to be using 15 mg/100 ml as the level to begin phototherapy (less if the infant is premature and severely ill) and 20 mg/100 ml as the level to consider intensive therapy (possibly, but not necessarily including exchange transfusion). Exchange transfusion appears to be reserved more for infants with severe hemolytic disease; for example, Rh or Kell incompatibility, neonatal glucose 6-phosphate dehydrogenase hemolysis, and sepsis.

In addition, there are certain technical problems involving bilirubin assay. Phototherapy breaks down nonconjugated bilirubin into nontoxic bilirubin isomers, which, however, are measured and included with unconjugated bilirubin in standard bilirubin assays. Finally, it must be mentioned that bilirubin is one of the least accurate of frequently or routinely performed chemistry assays, with proficiency surveys consistently showing between-laboratory coefficients of variation of 10% or more. To this is added variances due to specimens obtained under different conditions (venous vs. capillary or heelstick), state of infant hydration, and differences between laboratories because of different methodologies. At present, the bilirubin measurement within the capability of most laboratories that is considered to best correlate with clinical kernicterus is unconjugated (indirect) bilirubin. In HDN, most of the bilirubin will be unconjugated. Therefore, it usually is sufficient to obtain total bilirubin levels in order to follow the patient’s progress. If there is a question about the diagnosis, one request for conjugated/unconjugated fractionation is sufficient. A definite consideration is the additional blood needed to assay both the conjugated and unconjugated fractions since large specimens may be difficult to obtain from an infant heel puncture.

Albumin-bound bilirubin. Unconjugated bilirubin is presumed to be the cause of kernicterus. However, unconjugated or total bilirubin levels do not always correlate well with development of kernicterus. In the 1980s there was interest in measurement of free nonconjugated bilirubin and bilirubin-binding capacity. Most of the unconjugated bilirubin in serum is tightly bound to serum albumin, with a small portion being loosely bound (“free”). Since the free portion rather than the tightly bound portion theoretically should be the most important element in kernicterus, various methods have been devised to assay free unconjugated bilirubin rather than total bilirubin. Until approximately 1980, direct measurement was difficult, and most attention was given to indirect methods, chiefly estimation of the bilirubin-binding capacity of albumin. In general, when more bilirubin binds to albumin, the residual binding capacity becomes smaller and less binding of serum free bilirubin takes place. Therefore, free bilirubin levels are more likely to increase. Several methods have been proposed to measure albumin-binding capacity; the most popular involved a Sephadex resin column. Sephadex resin competes with albumin for loosely bound bilirubin. When unconjugated bilirubin levels exceed the bilirubin-binding capacity of albumin, the excess binds to the sephadex column. There are conflicting reports in the literature on the value of the albumin-binding capacity assay. Some reports indicated that it was very helpful; others found that results from individual patients were either too often borderline or did not correlate sufficiently well with the clinical picture. More recently, an instrument called the hematofluorometer that measures total bilirubin-binding capacity (TBBC) has become available. Total bilirubin-binding capacity as measured by the hematofluorometer is reported to correlate well with unbound bilirubin. Nevertheless, there does not appear to be convincing evidence that albumin-binding capacity or free bilirubin measurements have shown clear-cut superiority over traditional guidelines. In summary, although a total bilirubin level of 20 mg/100 ml is a reasonably good cutoff point for substantial risk of kernicterus in full-term infants, no adequate cutoff point has been found for sick premature infants; and consideration of risk factors such as sepsis, acidosis, pulmonary distress, hypothermia, hypoalbuminemia, or bilirubin-binding capacity has not produced totally reliable criteria for determining whether to use therapy or not.

Role of amniocentesis in hemolytic disease of the newborn. Amniocentesis has been advocated in selected patients as a means of estimating risk of severe HDN while the fetus is still in utero. A long needle is introduced into the amniotic fluid cavity by a suprapubic puncture approach in the mother. The amniotic fluid is subjected to spectro photometric estimation of bilirubin pigments. Markedly increased bilirubin pigment levels strongly indicate significant degrees of hemolytic disease in the fetus. If necessary, delivery can be induced prematurely once the 32nd week of gestation has arrived. Before this, or if the fetus is severely diseased, intrauterine exchange transfusion has been attempted using a transabdominal approach. The indications for amniocentesis are development of significant titer of Rh antibody in the mother or a history of previous erythroblastosis. Significant maternal antibody Rh titers do not always mean serious fetal Rh disease, but absence of significant titer nearly always indicates a benign prognosis. Also, if initial amniocentesis at 32 weeks does not suggest an immediately dangerous situation, even though mild or moderate abnormalities are present, a fetus can be allowed to mature as long as possible (being monitored by repeated studies) to avoid the danger of premature birth.