The outcome of pregnancy in Kell alloimmunisation


Correspondence: Dr M. D. Kilby, Birmingham Women's Hospital, Edgbaston, Birmingham B15 2TG, UK.


Objective To assess management and outcome of pregnancies with anti-Kell in the West Midlands in the UK over 13 years.

Design A retrospective review of casenotes.

Setting A regional referral clinic for red cell alloimmune disease and fetal medicine unit at a university hospital.

Population Sixty-five pregnancies were identified in 52 Kell-sensitised women with Kell positive partners from the records of the Birmingham Blood Transfusion Centre.

Methods Information from the casenotes was entered on a database and comparisons were made using the SPSS for Windows statistics package.

Main outcome measures Mode of sensitisation, degree of fetal or neonatal anaemia, need for transfusion, gestation at delivery, birthweight and pregnancy outcome.

Results Alloimmunisation was transfusion-related in 29 pregnancies and pregnancy-induced in 33. The cause could not be identified in three cases. There were 22 proven Kell positive fetuses, of which 18 were affected, in which alloimmunisation was pregnancy-related in 12 cases and transfusion-related in five. Antibody titres and amniotic fluid OD450 were not helpful in management. Severe or very severe disease occurred in 50% of the affected pregnancies (9/18). There was no difference in pregnancy outcome between transfusion or pregnancy induced sensitisation.

Conclusions Anti-Kell alloimmunisation is an uncommon cause of serious anaemia in a significant proportion of affected pregnancies. There appears to be no difference between that caused by pregnancy or transfusion. Estimation of fetal haemoglobin concentration by cordocentesis is recommended, as antibody titres and amniocentesis are not helpful.


Red cell alloimmunisation remains the major cause of severe fetal anaemia requiring invasive therapy and preterm delivery. This often results in prolonged intensive neonatal care and sometimes culminates in perinatal death. The incidence of anti-D sensitisation has decreased due to the administration of anti-D immunoglobulin, while the importance of other antibodies has increased1,2. Anti-Kell antibodies appear to cause significant suppression of erythropoiesis rather than red cell destruction3,4 and, unlike Rhesus disease, the outcome is not affected by previous obstetric history5. Also, in contrast to Rhesus disease, the maternal antibody levels and amniotic fluid spectropholomelric estimation (OD450)do not correlate with fetal anaemia5–8.

The frequency of perinatal death in pregnancies with anti-Kell is debatable5,7, although poor fetal outcome is reported to be between 1.5% and 3.9%5,7. Management of these pregnancies remains controversial, but the incidence of anti-Kell immunisation is low in the obstetric population, at 0.1–0.2%5,9 and is reported to be mainly transfusion induced5,7. The incidence of haemolytic disease due to anti-Kell is also low in the white population, where only 9% are Kell positive and 0.2% are homozygous10. Consequently, fewer Kell negative women are likely to carry a Kell positive fetus as compared with Rhesus D negative women.


Subject were obtained from the records of the Birmingham Blood Transfusion Centre, which provides the antenatal screening service to all the hospitals in the West Midlands Region with a population of 5.3 million inhabitants. The study consisted of a retrospective review of all antenatal records of women with anti-Kell booked at hospitals in the West Midlands between 1984 and 1996. Case records of all women with confirmed Kell positive partners were examined. Case notes were obtained from the Medical Records Department at the Birmingham Women's Hospital, or by written request to the referring consultants. Further written and telephone request were made on up to three occasions and some notes could only be reviewed on site. The main outcome measures of the study were the mode of sensitisation, degree of fetal anaemia, requirement for intrauterine transfusion, gestation of delivery, birth-weight, pregnancy outcome and need for neonatal treatment. The Kell group of all these fetuses was determined, in some cases from the amniotic fluid by polymerase chain reaction and confirmed on red cells from cordocentesis samples, or post delivery from cord blood specimens.

A baby was considered to be affected by alloimmune disease if:

  • 1A positive direct antiglobulin test was noted in fetal or neonatal cord blood.
  • 2The fetus or neonate was severely anaemic and died in utero or survived after plasmapheresis and intrauterine transfusion, regardless of direct antiglobulin test result.

The criteria for the severity of haemolytic disease of the newborn were adapted from those used by Leggat et al.6 and developed in Newcastle by Walker, for haemolytic disease of the newborn caused by anti-D11. The need for intrauterine transfusion indicated severe disease and the ‘Stillborn’ category, initially described by Walker as including miscarriage before 28 weeks, is extended in line with current definitions of miscarriage. Phototherapy was not considered to be an objective measurement of severity, as it is used prophylactically.

Severity of disease:

  • 1Direct antiglobulin test positive, normal haemoglobin concentration, no treatment required (Mild).
  • 2Direct antiglobulin test positive, cord haemoglobin concentration > 11 g/dL, requiring postnatal transfusion (Moderate).
  • 3Direct antiglobulin test positive, cord haemoglobin concentration < 11 g/dL or bilirubin >85 μ mol/L (Severe).
  • 4Direct antiglobulin test positive, requiring intrauterine transfusion, cord haemoglobin concentration < 7.5 g/dL, or hydropic with a haemoglobin concentration > 7.5 g/dL (Very severe).
  • 5Stillborn (including miscarriage before 24 weeks).

Information from the patients' records was entered onto a computerised database and comparisons were performed using the SPSS for Windows statistics package (SPSS Inc).


Over the 13-year period, 65 pregnancies were identified, in 52 women, with one twin pregnancy. Of the 66 fetuses involved, 60 were liveborn, one was stillborn after 24 weeks and there were three miscarriages below 24 weeks. There was one instance of pregnancy loss in which the gestation was not defined and one case in which information about outcome were missing. The mean maternal age was 29 years (SD 5, range 17–39 years) and the median parity 2 (SD 1, range 0–6). In those pregnancies that reached viability, the mean gestation at delivery was 37 weeks (SD 4, range 29–42 weeks), with a birthweight of 3119 g (SD 706, range 1370–4350 g). A history of maternal transfusion was documented in 29 of the 65 pregnancies (45%), including the one twin pregnancy. Pregnancy-induced sensitisation occurred in 33 pregnancies (51%) in women who had never been transfused. The mechanism of alloimmunisation was difficult to identify in the remaining four percent of pregnancies. Anti-Kell antibodies were detected in these women, who had all been transfused in the past, but at least one other pregnancy for which records were unavailable had occurred between the transfusion and the previous detection of anti-K.

There were 22 Kell positive infants who were at risk of haemolytic disease, including three in whom antibodies other than anti-K1 were noted; these three infants were all born at term, two were mildly, and one not, affected by haemolytic disease. Fifteen of the 22 fetuses had a positive direct antiglobulin test at some stage, five were negative and two had no test performed (another six infants had a positive direct antiglobulin test, although the Kell status was not determined at any stage). Four of the five infants found to be direct antiglobulin test negative were born with normal haemoglobin concentrations and did not require any treatment. The cordocentesis specimen of the fifth case before the first transfusion was found to be severely anaemic, 6.7 g/dL at 24 weeks of gestation, although the red cells were direct antiglobulin test negative by routine serological methods and by flow cytometry. This case demonstrates that fetal anaemia may occur despite a negative direct antiglobulin test and that the classification adopted to describe the severity of Kell associated disease (after Leggatt) needs revision to exclude direct antiglobulin test positivity as a necessary criterion.

One severely affected fetus was hydropic and died in utero at 21 weeks despite two intrauterine transfusions, having had a haemoglobin concentration of 1.2 g/dL at initial cordocentesis. Parvovirus infection was excluded in this fetus and no other antibodies were detected in the mother. The father was homozygous for Kj and thus the fetus most likely was severely affected by anti-K1 alloimmunisation. Another fetus, treated with plasmapheresis until cordocentesis at 20 weeks showed a haemoglobin concentration of 11.0 g/dL, subsequently underwent six intrauterine transfusions. In this case, although the father was heterozygous for K, the mother again showed no other antibodies. Thus, 18 of the 22 Kell positive fetuses were affected in total. All but one of the affected fetuses were liveborn, eight were only mildly affected, one moderately and nine very severely affected (including the fetal death at 21 weeks). There was no correlation between amniotic fluid OD450 and severity of disease, pregnancy outcome, gestation of delivery, birthweight or initial pre-transfusion fetal haemoglobin. The first dilutional anti-Kell titre correlated negatively with the haemoglobin concentration at initial cordocentesis (P= 0.01), gestation of delivery (P= 0.015) and birthweight (P= 0.007) and correlated positively with severity of haemolytic disease of the newborn (P= 0.015). Table 1 shows the results of cordocentesis performed in the women with Kell positive infants, related to the severity of the disease, cord haemoglobin concentration at first cordocentesis and the number of intrauterine transfusions performed. Figure 1 shows the results of amniotic fluid OD450 estimation in the at-risk Kell positive fetuses in which it was performed, related in Table 2 to severity of disease.

Table 1.  Results of cordocentesis in affected fetuses. IUT = intrauterine transfusion; FBS = fetal blood sample; Hb = haemoglobin concentration (g/dL); GA = gestational age (weeks)
GA at first FBSHb at first FBSGA at first IUTHb before first IUTGA at deliveryNo. of IUTsCord HbDisease severity*
  1. *Disease severity as defined by Walker et al.11.

Figure 1.

Liquor OD450 in affected pregnancies. Both Liley and Whitfield ‘action lines’ noted. Series = individual cases.

Table 2.  Seventy of Kell alloimmunisation in cases in which liquor OD450 was measured.
  1. *Disease seventy as defined by Walker et al.11.


These 22 Kell positive fetuses had a median haemoglobin concentration of 11.0 g/dL (range 1.2–14.3, SD 3.1) at initial fetal or neonatal blood sampling. Of these, 14 (63.6%) were born to mothers with pregnancy-induced anti-Kell and five (22.7%) to mothers with transfusion related alloimmunisation. In three (13.6%) the cause of immunisation could not be clearly identified. At delivery, three infants were anaemic (haemoglobin concentration < 5; SD below gestational mean) of which two were in women with pregnancy-related alloimmunisation. One of these two suffered a pregnancy loss at 21 weeks, having had a cord haemoglobin concentration of 1.2 g/dL at 20 weeks when the fetus was found to be severely hydropic, the other delivered at 30 weeks after having one intrauterine transfusion at 29 weeks. The woman with transfusion related alloimmunisation delivered at 33 weeks after four intrauterine transfusion (initial cord haemoglobin concentration 7.6 g/dL at 24 weeks, but haemoglobin concentration before first transfusion later in the same week was 3.3 g/dL).


Although the number of affected pregnancies in this survey was small, it compares reasonably with other studies. Leggat et al. found 16 affected pregnancies, over a much longer period6 and Bowman et al.2 studying 31 affected pregnancies presenting over 47 years, from 1944–1990. Our study, although including a relatively large number of Kell pregnancies, suffers from the weakness that practice and practitioners will have altered over the thirteen year period. In this study, very severe disease occurred in 14% (9/65) of pregnancies in women with anti-Kell antibodies and Kell positive partners and in 50% (9/18) of the affected Kell positive infants. Eighty-nine percent (8/9) of these infants underwent intrauterine transfusion. This contrasts with the findings of Leggat and colleagues, where only three of sixteen infants (19%) were very severely affected6. There was no statistical difference in clinical outcome between those pregnancies affected by transfusion induced (45%) and pregnancy-induced (51%) anti-Kell alloimmunisation. The use of Kell negative blood for transfusions in all women of childbearing age remains important, but will only eliminate, at most, 45% of cases of anti-Kell alloimmunisation, as pregnancy-induced anti-Kell was found in half of the women in this study. The proportion of pregnancy induced cases is higher than in previous reports, most probably because Kell negative blood was promoted early in this region and was available for transfusion of all women in the child bearing age since 1985. However, the severity of pregnancy-induced Kell disease does not appear worst than transfusion-related alloimmunisation, as compared with reported data13. We would generally expect to see the absolute number and proportion of transfusion induced anti-Kell to decrease as the policy for the use of Kell negative blood becomes universally acceptable and available. Women with pregnancy induced anti-Kell should eventually become the only cause for anti-Kell haemolytic disease.

This study shows that, although relatively rare, anti-Kell alloimmune disease can be a cause of serious fetal anaemia in a significant proportion of cases. In this series five Kell positive pregnancies were found to be direct antiglobulin test negative. Four were clearly not affected. The neonatal red cells were not coated with antibody, their haemoglobin concentrations were well maintained and the neonates did not require any therapy. In the fifth case, the mother had six weeks of intensive plasma exchange because of previous bad obstetric history. Cordocentesis was then performed at 23 weeks of gestation when the fetal haemoglobin concentration was found to be 10.9 g/dL. Before transfusion the following week, the haemoglobin concentration was 6.7 g/dL and the red cells were Kell positive, but the direct antiglobulin test was negative and confirmed by flow cytometry. The most likely explanation was that the antibody level is reduced by plasma exchange and the amount crossing the placenta was just enough to suppress erythropoeisis without being detected on the red cells. This observation has not been reported before and is possible that it was not looked for because, once intrauterine transfusion is given, the majority of red cells will be expected to be Kell negative and direct antiglobulin test negative, direct antiglobulin test positivity should thus no longer be considered a requirement for classifying a fetus as severely affected within the context of modern management of Kell associated disease.

The value of amniotic fluid OD450 measurement in the assessment of severity of fetal anti-Kell disease has been questioned5,7. These investigators reported that serious Kell-induced fetal anaemia was present despite having OD450 values that were significantly lower than those described in equivalent disease caused by anti-D antibodies. Direct estimate of fetal haemoglobin concentration by cordocentesis has been recommended for management of these cases by Vaughan et al. who suggested that anaemia in Kell alloimmunisation might be due to erythroid suppression, rather than haemolysis3. Eleven fetuses anaemic as a result of anti-Kell alloimmunisation were compared with the same number affected by Rhesus disease; the anti-Kell group had a lesser degree of reticulocytosis (P < 0.01), erythroblastosis (P < 0.05) and lower amniotic fluid bilirubin concentrations (P < 0.05)8. The same authors have recently shown that anti-Kell antibodies specifically suppress Kell positive erythroid progenitor cells and postulate this as the mechanism of anaemia in Kell alloimmunisation4. Over the 13-year period, clinical practice nationally has altered in line with these recommendations. The current study supports the suggestion that amniotic fluid OD450 estimation is of little value and that direct measurement of fetal haemoglobin concentration, as has been practised in the Fetal Medicine Unit in Birmingham since 1992, is the appropriate investigation in these cases. It is conceded, though, that even in this relatively large study, the number of cases in which OD450 was undertaken is too small to categorically conclude that fetal blood sampling, although more informative, is superior to OD450 estimates.

Amniocentesis remains useful, allowing for identification of fetal Kell genotype by polymerase chain reaction, which is very reliable. The use of polymerase chain reaction to identify the fetal blood group will enable all pregnancies with Kell negative fetuses to be excluded from follow up and further invasive assessment. Cordocentesis is indicated by previous bad history, homozygous paternal genotype, high amniotic fluid OD450 and a Kell positive fetus. It is possible, of course, that if the quantification methods for anti-Kell are improved and expressed against an anti-Kell standard as for anti-D, the measurement would be more sensitive. The current method relies on crude doubling titration and is not adequate to reflect the subtle changes in antibody activity. McKenna's recent paper indicates, however, that a titre of 1:32 may be a reliable cut off for action14.

Intrauterine transfusion is a life saving treatment in this condition15,16. Copel et al. describe the use of, first, intraperitoneal, followed by intravascular transfusion in the management of a case of hydrops secondary to Kell alloimmunisation16. Intravascular transfusion, when feasible, has superseded intraperitoneal transfusion, as the latter shows significantly lower fetal survival. This is particularly true if the fetus is severely hydropic, moribund and not showing breathing movements as red cells, transfused into the peritoneal cavity, will not be absorbed17,18.

Thus, anti-Kell alloimmunisation is a potentially serious cause of fetal anaemia that, if unrecognised and untreated, carries a high perinatal mortality. The move towards the use of Kell negative blood for transfusion in women of childbearing age should further reduce the prevalence of maternal alloimmunisation (by 45% of cases). The pathogenesis of this condition would indicate that such subjects be managed in a regional centre, with access to a fetal medicine centre for investigation and treatment of the fetus.