anti-kell antibody

Management of Pregnancies Complicated byAnti-Kell Isoimmunization

DAVID S. MCKENNA, MD, H. N. NAGARAJA, PhD, AND

RICHARD O’SHAUGHNESSY, MD

Objective: To assess the efficacy of managing pregnanciescomplicated by anti-Kell isoimmunization using the meth-ods developed for evaluating anti–Rh-D isoimmunization.

Methods: We reviewed 156 anti-Kell-positive pregnanciesseen from 1959 to 1995, which were managed with serialmaternal titers, amniotic fluid DOD450 determination, andfunipuncture. Data on maternal titers, paternal phenotypes,invasive fetal testing and therapies, and neonatal outcomeswere collected and analyzed to determine whether severelyaffected pregnancies were identified in time for successfulfetal and neonatal therapy.

Results: Twenty-one fetuses were affected, eight withsevere disease, and two fetuses in this group died. All of theseverely affected fetuses were associated with maternalserum titers of at least 1:32. A critical titer of 1:32 was foundto be 100% sensitive for identifying the affected pregnancies.The affected group had significantly higher amniotic fluidDOD450 values over the range of gestational ages than didthe unaffected group (P <<< .001). The upper Liley curve wasa specific discriminator for the diagnosis of affected fetuses,and the lower curve was specific for the diagnosis of unaf-fected or mild cases.

Conclusion: Fetal anemia due to anti-Kell isoimmuniza-tion might be due in part to erythropoietic suppression, butit is still largely a hemolytic process. The methods based ona hemolytic process, including use of a critical maternalserum titer of 1:32, serial amniotic fluid analyses when thetiter was exceeded, and liberal use of funipuncture, weresuccessful in identifying severely affected fetuses. (ObstetGynecol 1999;93:667–73. © 1999 by The American Collegeof Obstetricians and Gynecologists.)

Widespread use of anti–Rh-D immunoglobulin (Ig) inpregnant women who are D-antigen negative has led toan increasing proportion of isoimmunization due toatypical, non–Rh-D antibodies. Pregnancies compli-cated by isoimmunization due to atypical antibodies aregenerally managed in the same way as those withanti-D isoimmunization, which is largely a hemolyticprocess.1–3 Isoimmunization to the Kell antigen is acommon cause of fetal and neonatal anemia due to oneof these atypical antibodies.1

The Kell protein is a 93-kd transmembrane metal-lopeptidase important in the processing and metabo-lism of peptide hormones, and is believed to play a rolein erythrocyte growth and differentiation.4,5 The anti-genic nature of this protein can induce a pronouncedisoimmune response in antigen-negative women ex-posed to Kell antigen-positive red blood cells (RBCs)through donor blood transfusions or transplacentalpassage of fetal RBCs during pregnancy. It has beenproposed that the mechanism of fetal and neonatalanti-Kell isoimmune anemia is not solely hemolysis, asis the case with Rh disease, and that erythroid suppres-sion is also important.6,7 The use of maternal history,critical serum titers, and spectrophotometric determina-tion of amniotic fluid (AF) bilirubin in the monitoring offetal anemia is predicated on a maternal alloimmuneprocess that destroys fetal erythrocytes. Because thismethodology relies on markers of maternal immunesystem activity and fetal hemolysis, it may not beadequate for pregnancies complicated by anti-Kell iso-immunization with possible hyporegenerative ratherthan hemolytic anemia.

We analyzed retrospectively the 37-year experience atthe Ohio State University in the management of preg-nancies complicated by anti-Kell isoimmunization.During this period, the management protocol devel-oped for anti–Rh-D isoimmunization was used foranti-Kell, and we assessed its efficacy.

From the Division of Maternal Fetal Medicine, Department ofObstetrics and Gynecology, Ohio State University College of Medicine,Columbus, Ohio.

Dr. McKenna is a major in the United States Air Force. The opinionsand conclusions in this article are those of the authors and are notintended to represent the official position of the Department of Defense,United States Air Force, or any other government agency.

Support for statistical analysis was provided by the National Institutesof Health (NIH) General Clinical Research Center at the Ohio StateUniversity (grant no. NIH-M01-RR-00034).

667VOL. 93, NO. 5, PART 1, MAY 1999 0029-7844/99/$20.00PII S0029-7844(98)00491-8

Materials and Methods

A computerized database containing the records of allwomen with isoimmunized pregnancies who receivedcare at our medical center since 1959 was used toidentify all pregnant women affected by anti-Kell. Ob-stetric care was provided by our staff physicians or byreferring obstetricians from central and southeasternOhio. Women were managed in consultation with theOhio State University isoimmunization program, whichprovided all laboratory testing, interpretation, and sug-gested management. Data were obtained from the com-puterized database, patient charts, blood bank andphysician records, and patient telephone interviews.Only laboratory data from our prenatal reference labo-ratory were included in the analyzed data set.

Data included maternal demographics and preg-nancy history, paternal antigen testing, maternal indi-rect antiglobulin tests, results of optical density at450 nm (DOD450), fetal hematocrit, fetal total bilirubin,and percentages of nucleated fetal RBCs. Neonatal dataincluded gestational age at delivery, birth weight, de-livery hematocrit and total bilirubin, cord indirect anti-globulin test results and Kell antigen status, neonatalmorbidity, and necessary treatment(s). To avoid poten-tial confounding effects, we analyzed only data ob-tained before the initiation of therapy. Initial maternaltiter and paternal antigen data were available for allpregnancies. Complete maternal and neonatal datawere available for all patients who had invasive proce-dures (ie, amniocentesis, funipuncture, and transfu-sion). When patients had more than one anti-Kell iso-immunized pregnancy, only the initial pregnancy wasincluded in the analyses. Patients with incomplete datawere excluded.

Maternal titers were measured every 4–6 weekswhen paternal antigen testing was Kell positive orunknown. Standard tube techniques for the indirectCoombs test, as endorsed by the American Associationof Blood Banks,8 were used to determine antibodytiters. Amniocentesis for spectrophotometric determi-nation of AF bilirubin pigment was done when therewas a history of an affected infant or when maternalantibody titers were greater than 1:16. The results wereplotted on a modified version of the Liley graph thatused the original zones defined by Liley9,10 but that alsoincluded lower gestational ages down to 20 weeks. Theportion from 20 to 28 weeks was developed in the 1960sfor the management of Rh disease and was validatedwith data obtained from patients at the Ohio StateUniversity (O’Shaughnessy, Amniotic fluid spectropho-tometry is useful after 20 weeks’ gestation in the care ofpregnancies complicated by red blood cell isoimmuni-zation [abstract]. Am J Obstet Gynecol 1991;164:256).

Since 1986, funipuncture has been done when thechange in DOD450 was in the upper half of Liley zone IIor in Liley zone III and the gestational age was remotefrom term.

Analyses of fetal blood included hemoglobin levels,total and direct bilirubin, and percentage of nucleatedRBCs. For this analysis, we categorized the affectedfetuses by the degree of neonatal anemia using thresh-olds similar to those defined by Liley: mild (neonatalhemoglobin at least 11 g/dL) or severe (neonatal hemo-globin 10.9 g/dL or less).9,10 Anemia was also definedas severe when hydrops fetalis was present or whenfetal or neonatal blood transfusion was required. Deliv-ery or fetal transfusion was done when severe fetalanemia (hemoglobin less than 10 g/dL) or hydropsfetalis was present. Umbilical cord blood specimenscollected at delivery were tested for bound anti-Kell Igby direct Coombs testing and for erythrocyte Kellantigen status.

Statistical and data analyses were done with JMPStatistical Discovery Software (SAS Institute Inc., Cary,NC) and Microsoft Excel (Microsoft Corporation, Red-mond, WA). Alpha of .05 was considered significant.The Levene test was applied on continuous variables totest for equal variances. When variances were equal,Student t-test was used; otherwise Welch analysis ofvariance was performed.11 Fisher exact test was usedfor comparisons involving two groups of nominal orordinal variables. Analysis of variance for repeatedmeasures was used to compare the mean responses ofgroups when there were multiple observations of acontinuous variable over time.

Results

There were 134 anti-Kell-positive women with 156pregnancies at the Ohio State University from January1959 to November 1995. When a woman had more thanone anti-Kell-isoimmunized pregnancy, we analyzedonly data from the initial pregnancy. Complete data forrace and titer were available for 116 initial pregnancies.Eighty-three women (72%) were white and 33 (28%)were black. Nineteen affected infants were delivered bywhite women and no affected infants were delivered byblack women (P 5 .002). The mean maternal age (6standard deviation [SD]) at delivery was not signifi-cantly different (29.1 6 3.9 years for affected infantsversus 27.6 6 5.8 years for those unaffected; P 5 .17).The mean gestational age at delivery was significantlyless for affected pregnancies (35.8 6 5.0 weeks foraffected versus 38.0 6 3.4 weeks for unaffected; P 5 .02).

Paternal serum testing for RBC Kell typing was donewhen paternity was certain. Paternal antigen status wasconsidered unknown when paternity was not certain.

668 McKenna et al Anti-Kell Isoimmunization Obstetrics & Gynecology

Of the 103 paternal serum samples analyzed for Kellantigen, 34 (33%) were Kell positive. The 53 pregnancieswith unknown paternal antigen status were consideredat risk, for a total of 87 at-risk pregnancies in 75 women.Forty-seven women (63%) at risk had a history of ablood transfusion, 12 (16%) had a known negativehistory of receiving blood products, and 16 (21%) didnot have documentation or could not recall.

One at-risk pregnancy that ended in fetal death(described later) was not included in the followinganalyses because of incomplete data. Twenty-one in-fants (24%), born to 20 at-risk women, were affected.Three of 52 pregnancies (5.8%) with unknown paternalKell antigen status were affected, and 18 of 34 paternalKell antigen–positive pregnancies (53%) were affected.No affected infants resulted from pregnancies with anegative paternal antigen status. Twenty infants fromthe at-risk pregnancies had umbilical cord blood thatwas Kell antigen positive. Eighteen umbilical cordblood specimens collected at delivery had positivedirect antiglobulin tests. Two infants with severe ane-mia had negative cord blood direct antiglobulin tests atdelivery, but both had received fetal transfusions, andone previously had a positive direct antiglobulin testfrom a funipuncture specimen. The other received twointraperitoneal transfusions in 1982, before the use offunipuncture at our institution, and fetal blood was notavailable for direct Coombs analysis in this case. Um-bilical cord blood was not collected from one pregnancythat ended in fetal death at 25 weeks’ gestation becauseof severe hydrops fetalis. This pregnancy was believedto be affected based on the clinical presentation of anelevated maternal serum anti-Kell titer of 1:32 and AFDOD450 in Liley zone II.

Eight infants (9.3%) among the at-risk pregnancieswere severely affected. Table 1 presents the maternaldata and neonatal outcomes of the eight severely af-fected infants. Two cases (2.3%) ended in fetal death,

and among three cases (3.5%) of hydrops fetalis, onefetus died. In this case, from 1979 (Table 1, no. 7), thewoman first presented for prenatal care at 25 weeks’gestation, and an anti-Kell titer of 1:32 was found on herserum screen. Amniocentesis and ultrasound were per-formed 2 days later and revealed hydrops fetalis; thefetus died on the same day. The second fetal deathoccurred after a transfusion and was procedure related.A third fetal death occurred in a woman with anti-Kellisoimmunization (titer 1:1024) at 39 weeks’ gestation.This death occurred in 1966 and we were unable toconfirm the clinical circumstances, so we did not in-clude this pregnancy in our analyses. The neonate ofsubject no. 8 was delivered at 36 weeks and hadumbilical cord hemoglobin of 11.0 g/dL at delivery, butthe neonate’s hemoglobin fell to 8.8 g/dL on the firstday of life and a simple transfusion was given.

The severely affected group contained three nullipa-ras, four parous women with negative histories ofaffected infants, and one parous woman who may havehad two previous mildly affected infants. This lastwoman was gravida 4, para 2, and her two childrenwere delivered at term and needed phototherapy forhyperbilirubinemia, but did not require transfusions.The previous pregnancies were managed at anotherfacility and are not included in this review. Two womenin the severely affected group were followed at the OhioState University in subsequent pregnancies; one had anunaffected infant and the other had a mildly affectedinfant.

Table 2 presents maternal and neonatal data from the13 infants with mild anti-Kell isoimmunization. Five ofthe mildly affected fetuses had at least one funipuncture(range one to seven) for fetal blood analysis but did notrequire fetal therapy. One neonate required one simpletransfusion on the fourth day of life, but the directantiglobulin test on umbilical cord blood was positivefor multiple maternal antibodies in addition to anti-

Table 1. Neonatal Outcomes in Eight Severely Affected Pregnancies

No. G-PBloodtrans Past OB

Hightiter

HighestDOD450

Low fetalHgb No. IUTs

EGA atdelivery

CordHgb

CordDAT Complications

1 2-1 Yes Unaffected 1:256 0.22(III) 2.5 8 35 11.7 41 HF, RDS, HB, XT1, ST12 5-1 No Unaffected 1:256 0.16(II) 6.0 6 36 14.8 Neg RDS, HB3 1-0 Yes N/A 1:32 0.13(II) 7.2 3 36 12.6 11 HB, NEC4 4-2 No HB32 1:64 0.18(III) 7.2 2 33 12.3 11 RDS, IVH, HB, ST35 2-0 Yes N/A 1:512 0.28(III) 6.1 3 23 ND 41 IUFD6 1-0 Yes N/A 1:128 0.32(III) ND 2 (IP) 31 10.6 Neg HF, HB, RDS, XT1, ST27 3-1 Yes Unaffected 1:32 0.14(II) ND 0 25 ND ND IUFD, HF8 5-1 No Unaffected 1:128 0.15(II) ND 0 36 11.0 41 HB, ST1

G-P 5 gravida-para; blood trans 5 maternal history of blood transfusion; past OB 5 outcome in previous pregnancies; Hgb 5 hemoglobin;IUTs 5 fetal transfusions; EGA 5 estimated gestational age in weeks; Cord Hgb 5 umbilical cord hemoglobin (g/dL) at delivery; DAT 5 directantiglobulin test; HF 5hydrops fetalis; RDS 5 respiratory distress syndrome; HB 5 hyperbilirubinemia requiring phototherapy; XTn 5 numberof exchange transfusions; STn 5 number of simple transfusions; N/A 5 not applicable; NEC 5 necrotizing enterocolitis; IVH 5 intraventricularhemorrhage; IUFD 5 fetal death; ND 5 not done; IP 5 intraperitoneal.

VOL. 93, NO. 5, PART 1, MAY 1999 McKenna et al Anti-Kell Isoimmunization 669

Kell, and the anemia could not be considered solely dueto anti-Kell. The mildly affected group included onenullipara, eight parous women with negative historiesof affected infants, two women who might have hadprevious affected infants, and one woman with a con-firmed affected infant in the past. The first two womenhad a history of two infants with hyperbilirubinemiarequiring phototherapy and one infant with a history of“anemia,” respectively. These previous pregnancieswere cared for at another facility and their data werenot included in this analysis. The third woman previ-ously had a mildly affected fetus (Table 2). In the secondaffected pregnancy, she had a funipuncture at 25 weeks’gestation that did not demonstrate marked fetal ane-mia, and she was followed with serial amniocentesesthat remained in the lower half of Liley zone II.

Three hundred forty-six maternal indirect Coombstiters were measured, 81 from pregnancies with con-firmed Kell-negative fathers. There were 265 from at-risk women: 194 from the unaffected group, 43 from themildly affected group, and 28 from the severely affectedgroup. All titers were 1:32 or greater in the severelyaffected group, 12 titers (28%) were 1:32 or greater in themild group, and 76 titers (39%) were 1:32 or greater inthe unaffected group. Five at-risk pregnancies were notincluded in the following analysis because three endedin abortion and two had incomplete data for titers. Onlydata from the initial pregnancy were analyzed when awoman had more than one anti-Kell-isoimmunizedpregnancy. We used the two-tailed Fisher exact test toassess a critical titer of 1:32 as a discriminator. The ratioof maximum titers greater than 1:32 in a specific preg-nancy did not differ significantly between the mildlyaffected group (5 of 11) and the unaffected group (22 of

50) (P 5 1.0). The severely affected group was thencompared with the other two groups combined. Alleight maximum titers were 1:32 or greater in the se-verely affected group, whereas 27 of 61 (44%) were 1:32or higher in the other at-risk pregnancies (P 5 .005). Acritical titer of 1:32 was 100% sensitive for identifyingthe severely affected pregnancies.

Amniocentesis for DOD450 was done in 36 women intheir initial isoimmunized pregnancies. Twelve hadaffected pregnancies and 24 had pregnancies resultingin Kell-negative neonates. Sixty-four amniocenteseswere done in the affected group and 46 in the unaffectedgroup. The mean gestational age at amniocentesis wassignificantly less in the affected group than in theunaffected group (26.0 6 4.5 weeks versus 29.9 6 4.2weeks; P , .001). To consider the effect of gestationalage on DOD450, we stratified the data by gestational ageat amniocentesis into three groups: less than 25 weeks,25–32 weeks, and greater than 32 weeks. The gestationalage groupings were chosen because the clinical signifi-cance of an affected pregnancy is substantially differentfor each of these ranges of gestational age. Analysis ofvariance with repeated measures was performed on thestratified data of log10 of the DOD450 values to ascertainthe difference between the affected and unaffectedgroups, taking into account the repeated measures andthe fact that DOD450 varied over gestational age (R2 50.92). The affected group had significantly higherDOD450 values than the unaffected group over the rangeof gestational ages (P , .001). The gestational age effectwas also significant (P , .001).

Figure 1 contains the AF data obtained prior to theinitiation of therapy with the affected group separatedinto severe and mild cases. The severely affected preg-

Table 2. Neonatal Outcomes in 13 Mildly Affected Pregnancies

G-PBloodtrans Past OB

Hightiter

HighestDOD450

Low fetalHgb

EGA atdelivery

CordHgb

CordDAT Complications

4-3 No Unaffected 1:64 0.18(III) 10.3 38 12.6 11 HB, 7 cordocenteses*5-3 Yes Unaffected 1:128 0.13(II) 10.3 37 12.7 41 HB, 4 cordocenteses4-0 Yes N/A 1:64 0.1(II) 10.7 39 11.2 11 HB, 5 cordocenteses4-2 Yes Unaffected 1:32 0.055(I) 10.4 38 18.7 21 1 funipuncture5-4 No Mild 3 1 ND 0.085(II) 11.0 38 ND 11 1 funipuncture†

3-2 Unk HB 3 2 1:8 ND ND 37 12.2 11 None5-3 Unk Anemia 1:32 0.08(II) ND 40 20.2 21 HB7-4 Yes Unaffected 1:4 ND ND 40 15.8 31 None3-1 Yes Unaffected 1:8 ND ND 40 ND 11 None5-3 Yes Unaffected 1:1 ND ND 35 14.5 11 HB, ST1‡

4-3 Yes Unaffected 1:4 ND ND 40 16.9 11 None4-3 Yes Unaffected 1:8 ND ND 40 Unk 11 None3-1 Yes Unaffected Neg ND ND 39 16.3 11 None

Unk 5 unknown; other abbreviations as in Table 1.* First affected pregnancy.† Second affected pregnancy.‡ Neonate also had indirect Coombs positive for A, B, and Rh-D.


nancies are identified by numbers that correspond tothe patient numbers in Table 1. Serial measurements forthe severely affected pregnancies are connected withsolid lines. Figure 1 also shows the modified Lileycurves used at the Ohio State University for the man-agement of Rh disease. Although the DOD450 is anindirect measure of fetal anemia, the upper Liley curveproved to be a specific discriminator for diagnosingaffected fetuses, and the lower curve was a specificdiscriminator for diagnosing unaffected or mild cases.

Fifty-two funipunctures were done in 17 pregnancies(ten affected and seven unaffected), and 24 fetal trans-fusions were done in six affected pregnancies. Theinitial fetal hemoglobin, total fetal bilirubin, and per-centage of fetal nucleated RBCs were compared be-tween the affected and unaffected groups. Only datafrom the initial funipuncture were used. In cases inwhich a woman had two affected pregnancies, only thevalues from the first affected pregnancy were used. Themean gestational age was not significantly differentbetween the groups: 28.9 6 6.9 weeks in the affectedgroup (n 5 9) versus 23.7 6 4.8 weeks in those unaf-fected (n 5 7) (P 5 .12). The fetal hemoglobin wassignificantly lower in the affected group: 7.8 6 3.0versus 11.0 6 0.7 g/dL (P 5 .012). Total fetal bilirubinwas not significantly different between the groups:2.7 6 1.2 mg/dL in the affected group (n 5 4) versus

1.4 6 0.4 mg/dL in those unaffected (n 5 6) (P 5 .12).The percentage of fetal nucleated RBCs also was notsignificantly different between the groups: 10.0 6 12.11(n 5 8) versus 20.8 6 19.6 (n 5 6) (P 5 .22).

Neonatal umbilical cord blood specimens were col-lected at delivery in 18 affected and 24 unaffectedpregnancies. The mean gestational age at delivery didnot differ between the groups: 37.4 6 2.9 weeks for theaffected group (n 5 18) versus 36.0 6 3.2 weeks in theunaffected group (n 5 24) (P 5 .16). The mean neonatalhemoglobin level was significantly lower in the affectedgroup: 13.6 6 2.6 g/dL (n 5 18) versus 15.9 6 2.6 g/dLin the unaffected group (n 5 24) (P 5 .007). The totalbilirubin levels did not differ significantly between thegroups: 3.1 6 1.4 mg/dL (n 5 15) versus 3.2 6 1.7mg/dL (n 5 15), respectively (P 5 .8).

Discussion

For more than 20 years, Kell antibodies have beenknown to cause hemolytic disease in newborns.12,13 Aswith Rh disease, paternal RBC typing is the first step inevaluating a gravida who has a positive indirect screenfor anti-Kell. Approximately 90% of the population isKell negative.14 Assuming that the positives are het-erozygotes, a father with unknown antigen statuswould be expected to have an affected fetus about 5% of

Figure 1. DOD450 values for preg-nancies severely affected (diamonds),mildly affected (triangles), and unaf-fected (circles) by anti-Kell isoimmu-nization at different gestationalages. Severely affected fetuses areidentified by patient numbers,which correspond to numbers inTable 1. Serial values in individualseverely affected fetuses are con-nected by solid lines. Dotted linesrepresent the upper and lower mod-ified Liley curves used at the OhioState University for the manage-ment of Rh isoimmunization.9 Theportions of the curves from 16 to 20weeks (dots and dashes) are straight-line extrapolations and are notbased on data.


the time. This is consistent with the 5.8% rate observedin our population. Our observed rate of affected preg-nancies when the paternal antigen was known to bepositive (53%) is what we expected with heterozygoustransmission. Eight of the affected pregnancies resultedin severe disease, which is similar to the incidence ofsevere disease due to anti-Kell isoimmunization re-ported in other series.15–18 Pregnancy history is alsoimportant in predicting the severity of disease in Rhisoimmunization. In our series, we did not find thatpregnancy history predicted the outcome in Kell isoim-munization; however, the relatively small number ofaffected infants (n 5 21) might not have had the powerto detect an effect. The fact that all of the affected infantswere delivered by white women is not unexpected; theK1 antigen usually occurs in persons of Europeandescent and is rare in blacks.15

In the 1980s, it was recognized that Kell-isoimmunized pregnancies did not always evolve in thesame manner as those sensitized with anti–Rh-D.13,19

Several other investigators reported large series (morethan 100 pregnancies) of anti-Kell isoimmunization,and all but one study found that severely affectedpregnancies had maternal titers of at least 1:32.15–18

Bowman et al18 reported 31 pregnancies affected withKell isoimmunization. Seventeen pregnancies were se-verely affected, and all but one had maternal serumtiters of 1:32 or greater. In the single exception, thewoman presented with a grossly hydropic fetus at 23weeks’ gestation, had a titer of 1:8, and had no historyof an affected infant.18 In our series, none of the severelyaffected pregnancies had a titer less than 1:32, which wefound to be a sensitive discriminator for pregnanciesthat required amniocentesis.

The reliability of DOD450 values in Kell isoimmuniza-tion has been questioned because of reports of moreserious fetal anemia presenting at lower values.16,19

Figure 1 depicts the trend of higher DOD450 values foraffected fetuses, which is consistent with a hemolyticprocess. Caine and Mueller-Heubach16 and Bowman etal18 found AF analysis to be reliable in most instances,but both groups had difficulty diagnosing the affectedpregnancies when serial values of DOD450 droppedfrom Liley zone II into zone I, particularly at gestationalages over 32 weeks. In our series, DOD450 was specificbut not sensitive in distinguishing affected from unaf-fected fetuses. The values that fell in the middle zone(zone II) were not easily categorized, and those preg-nancies required careful further analysis with serialamniocenteses and often funipuncture. In some affectedpregnancies, DOD450 values dropped into a lower zonewhen serial values were collected, but no severelyaffected fetuses fell into and stayed within zone I. Inunaffected pregnancies, serial DOD450 values decreased

as gestational age increased. Funipuncture for fetalhemoglobin is recommended when the rate of declineof serial DOD450 measurements decreases or reaches aplateau. Caution should be exercised when usingDOD450 values, especially when they fall in the middlezone. The limitations of AF analysis as an indirectindicator of fetal anemia should be recognized.

Our data from the funipuncture and umbilical cordspecimens showed significantly lower hemoglobin lev-els in the affected fetuses and neonates. Total bilirubinlevels did not differ significantly between affected andunaffected fetuses and neonates. The percentage of fetalnucleated RBCs also did not differ between the groups.These findings are consistent with a hyporegenerativepathophysiology. Vaughan et al6 and Weiner and Wid-ness7 reported decreased laboratory indices (fetal re-ticulocytes, bilirubin, and nucleated RBCs) of fetalerythropoiesis and hemolysis in pregnancies compli-cated by Kell isoimmunization, and Vaughan et al5

recently demonstrated, in vitro, a dose-dependent sup-pression of hematopoietic progenitor cells by serumfrom women with circulating anti-Kell antibodies. Ourexperience suggests that although fetal anemia due toanti-Kell might be due in part to suppression of fetalerythropoiesis, the elevated DOD450 levels in the af-fected pregnancies indicate that the mechanism of ane-mia is also a hemolytic process.

Our management approach to pregnancies compli-cated by Rh isoimmunization has proved successful inmanaging more than 150 pregnancies with anti-Kellisoimmunization at the Ohio State University since1959. Using our investigation guidelines, we did nothave a single undiagnosed, severely affected fetus.There was one fetal death directly attributable to isoim-munization; however, the woman presented late forprenatal care with a hydropic fetus. This case (Table 1,no. 7) underscores the importance of early prenatal carein the diagnosis and management of isoimmunization.In addition, the fetus might have survived with modernsonography and fetal therapy. Maternal antibody titerswith a threshold for amniocentesis of greater than 1:16and DOD450 determinations are important in decidingwhich pregnancies need additional investigations suchas fetal blood sampling. Fetal Kell antigen typing fromamniocytes should decrease the number of blood sam-plings done on antigen-negative fetuses.20,21 A highindex of suspicion should be maintained because theevaluations are not uniformly consistent with a purelyhemolytic process. It is possible that erythroid suppres-sion can mask the traditional indirect characteristics offetal anemia. Bowman et al18 suggested adopting alower critical titer of 1:8 for anti-Kell isoimmunization.A lower threshold might increase sensitivity, but our


data do not support this because all of our severelyaffected pregnancies had titers of at least 1:32.

References1. Bowman JM. Hemolytic disease (erythroblastosis fetalis). In:

Creasy RK, Resnick R, eds. Maternal fetal medicine: Principles andpractice. 3rd ed. Philadelphia: WB Saunders, 1994:711–43.

2. Jackson M, Branch DW. Isoimmunization in pregnancy. In: GabbeSG, Niebyl JR, Simpson JL, eds. Obstetrics: Normal and problempregnancies. 3rd ed. New York: Churchill Livingstone, 1996:899–932.

3. Weinstein L. Irregular antibodies causing hemolytic disease of thenewborn: A continuing problem. Clin Obstet Gynecol 1982;25:321–32.

4. Turner AJ, Tanzawa K. Mammalian membrane metallopeptidases:NEP, ECE, KELL, and PEX. FASEB J 1997;11:355–64.

5. Vaughan JI, Manning M, Warwick RM, Letsky EA, Murray NA,Roberts IAG. Inhibition of erythroid progenitor cells by anti-Kellantibodies in fetal alloimmune anemia. N Engl J Med 1998;338:798–803.

6. Vaughan JI, Warwick R, Letsky E, Nicolini U, Rodeck CH, FiskNM. Erythropoietic suppression in fetal anemia because of Kellalloimmunization. Am J Obstet Gynecol 1994;171:247–52.

7. Weiner CP, Widness JA. Decreased fetal erythropoiesis and hemo-lysis in Kell hemolytic anemia. Am J Obstet Gynecol 1996;174:547–51.

8. American Association of Blood Banks. Technical manual of theAmerican Association of Blood Banks. 9th ed. Arlington, Virginia:American Association of Blood Banks, 1985.

9. Liley AW. Liquor amnii analysis in the management of thepregnancy complicated by rhesus sensitization. Am J Obstet Gy-necol 1961;82:1359–70.

10. Liley AW. Assessment of hemolytic disease from amniotic fluid.Am J Obstet Gynecol 1963;86:485–94.

11. Sall J, Lehman A. JMP start statistics: A guide to statistics and dataanalysis using JMP and JMP IN software. 1st ed. Belmont, Califor-nia: Duxbury Press, 1996:115–47.

12. Frigoletto FD, Davies IJ. Erythroblastosis fetalis with hydropsresulting from anti-Kell isoimmune disease. Am J Obstet Gynecol1977;127:887.

13. Barss VA, Benacerraf BR, Greene MF, Phillippe M, Frigoletto FD.Sonographic detection of fetal hydrops. A report of two cases. JReprod Med 1985;30:893–4.

14. Marsh WL, Redman CM. The Kell blood group system: A review.Transfusion 1990;30:158–67.

15. Wenk RE, Goldstein P, Felix JK. Kell alloimmunization, hemolyticdisease of the newborn, and perinatal management. Obstet Gy-necol 1985;66:473–6.

16. Caine ME, Mueller-Heubach E. Kell sensitization in pregnancy.Am J Obstet Gynecol 1986;154:85–90.

17. Leggat HM, Gibson JM, Barron SL, Reid MM. Anti-Kell in preg-nancy. Br J Obstet Gynaecol 1991;98:162–5.

18. Bowman JM, Pollock JM, Manning FA, Harman CR, MenticoglouS. Maternal Kell blood group alloimmunization. Obstet Gynecol1992;79:239–44.

19. Berkowitz RL, Beyth Y, Sadovsky E. Death in utero due to Kellsensitization without excessive elevation of the delta OD40 valuein amniotic fluid. Obstet Gynecol 1982;60:746–9.

20. Lee S, Bennet PR, Overton T, Warwick R, Wu X, Redman C.Prenatal diagnosis of Kell blood group genotypes: KEL1 and KEL2.Am J Obstet Gynecol 1996;175:455–9.

21. Spence WC, Maddalena A, Demers DB, Bick DP. Prenatal deter-mination of genotypes Kell and Cellano in at-risk pregnancies. JReprod Med 1997;42:353–7.

Reprints are not available.

Received June 25, 1998.Received in revised form September 29, 1998.Accepted October 22, 1998.

Copyright © 1999 by The American College of Obstetricians andGynecologists. Published by Elsevier Science Inc.


anti-kell antibody

Documents