Erythropoietic suppression in fetal anemia because of Kell alloimmunization

Erythropoietic suppression in fetal anemia because of Kell alloimmunization Janet I. Vaughan, MD, Ruth Warwick, MD, Elizabeth Letsky, MD, Umberto Nicolini, MD, Charles H. Rodeck, MD, and Nicholas M. Fisk, PhD

London, United Kingdom OBJECTIVE: Our purpose was to test the hypothesis that maternal anti-Kell alloimmunization produces fetal anemia by erythroid suppression. STUDY DESIGN: Erythropoiesis in 11 anemic fetuses from maternal anti-Kell alloimmunization was compared with that in 11 fetuses where the mother was alloimmunized to RhO; each was matched for hematocrit, gestational age, hydrops, and perinatal outcome. Comparisons of the difference were performed by either paired t or Wilcoxon tests. RESULTS: The anti-Kell group had reduced reticulocytosis (p = 0.007) and erythroblastosis (p = 0.045) and lower amniotic fluid bilirubin concentrations (p = 0.02) in comparison with the anti-O group. No correlation was found between hematocrit and reticulocytosis in the anti-Kell group, whereas the anti-O group had a significant linear relationship (r = 0.63, P < 0.05), indicating a progressive reticulocytosis in response to the degree of anemia. CONCLUSION: These findings suggest that erythroid suppression, rather than hemolysis, is the predominant mechanism in producing fetal anemia related to maternal Kell alloimmunization. Fetal blood sampling is the investigation of choice in the evaluation of anemia related to maternal Kell alloimmunization, because reduced hemolysis means amniotic fluid bilirubin concentrations correlate poorly with anemia. (AM J OBSTET GVNECOL 1994;171 :247-52.)

Key words: Kell alloimmunization, fetal anemia, reticulocytosis, Liley's charts

With the decline in the incidence of maternal anti-D antibodies associated with prophylactic administration of anti-D immunoglobulin to Rh-negative women at delivery, the proportion of perinatal deaths because of alloantibodies other than anti-D has increased.' Maternal alloimmunization to the Ken antigen may cause fetal anemia when the fetus is Ken positive. Only 9% of whites are Ken positive, and the majority of maternal anti-Ken antibodies are transfusion induced. 2 Nevertheless, the incidence of maternal anti-Ken antibodies is relatively high at 0.1 % in the obstetric population,2 and the Ken antigen is second only to RhD in immunizing potential.' Maternal anti-Ken has emerged, along with anti-D and anti-c, as a leading cause of fetal and neonatal anemia. 4 Observations suggest that fetal and neonatal disease

From the Royal Postgraduate Medical School, Institute of Obstetrics and Gynaecology, Queen Charlotte's and Chelsea Hospital. Received for publication October 7, 1993; revised December 16, 1993; accepted December 30, 1993. Reprint requests: J.I. Vaughan, MD, Royal Postgraduate Medical School, Institute of Obstetrics and Gynaecology, Queen Charlotte's and Chelsea Hospital, Goldhawk Road, London, United Kingdom W60XG. Copyright © 1994 by Mosby-Year Book, Inc. 00002-9378/94 $3.00 + 0 6/1/53966

related to maternal anti-Ken and anti-D differ. In contrast to hemolytic disease related to anti-D, previous obstetric history is not predictive of outcome in the index pregnancy in disease related to anti-Ken antibodies! Further, there is a poor correlation between antibody titer and outcome, with hydrops fetalis documented with anti-Ken titers as low as 2. 4 - 6 In addition, amniotic fluid spectrophotometric estimation [.:lOD450 ] of bilirubin concentration, although useful for predicting the severity of anti-D disease in the fetus,7 underestimates or correlates poorly with the severity of disease related to anti-Ken!' 8-11 Furthermore, postnatal hyperbilirubinaemia is not prominent in anemic babies with anti-Ken disease. II, '2 We hypothesise that fetal anemia as a result of antiKell antibodies involves erythroid suppression rather than hemolysis alone, as in anti-D. To test this, we compared fetal erythropoiesis in anemic fetuses of mothers alloimmunized to Kell and RhD antibodies.

Material and methods The study group comprised all 11 mothers with Kell anoimmunization seen in our center between 1986 to 1993 whose fetuses were shown at blood sampling to be anemic (hematocrit 2 SDs below our institutional mean for gestation) (Fig. 1). Their past obstetric histories

247

248 Vaughan et al.

July 1994 Am J Obstet Gynecol

50

--

-

Table I. Matched categoric variables for Kell and RhD pairs Kell/RhD

40

:.Ii!

l!.,...

'1: CJ

30

g

0

'Iii E GI CII

20

J:

• •

0

0

o•

•

10

i iii

8

• 0

0

Survival Ascites alone Hydrops Second procedure

e

0 15

20

25

30

35

Gestational age (weeks) Fig. 1. Hematocrit and gestational age at first transfusion in matched Kell (filled symbols) and RhD (unfilled) fetal pairs. Squares, Hydropic fetuses; circles, nonhydropic fetuses; shaded area, 95% data intervals of reference range.

included 10 fetal losses before 24 weeks, three stillbirths, and one neonatal death; only 15 of 29 previous pregnancies had a succ~'ssful outcome. Gestational age ranged from 19 to 32 weeks, and the anti-Kell titers ranged from 128 to 2048. Because fetal hematocrit increases with gestation, the study fetuses were matched for hematocrit and gestational age with 11 fetuses affected by anti-D selected as best matches from > 100 anti-D alloimmunized pregnancies investigated over the same period. Each group was also matched for perinatal outcome, presence of hydrops, and procedure number (Table I). All fetuses underwent ultrasonographic examination (Acuson, Mountain View, Calif.) to confirm gestational age and to document the presence or absence of ascites, either in isolation or with other signs of hydrops (i.e., pericardial effusion, skin edema). Mter the patients gave informed consent, fetal blood sampling was performed to assess the degree of anemia. With a 20-gauge spinal needle blood was obtained by ultrasonographically guided needle puncture of the umbilical vein at either the placental cord insertion I3 or the fetal intrahepatic portion, 14 as previously described. Injection of 1 to 2 ml of normal saline solution enabled confirmation of venous origin by ultrasonic visualization of the direction of turbulent flow. Although all fetuses required intrauterine intravascular transfusion (range one to six), two were not anemic at first sampling (Table I). For this study only samples obtained immediately before the first transfusion were analyzed. In 13 pregnancies amniotic fluid obtained through the same needle puncture at the time of fetal blood sampling was analyzed spectrophotometrically for bilirubin. Because AOD 45o is

No. (n = 11)

%

9 1

82

2 2

9

19

18

inaccurate before 24 weeks' gestation, only values obtained after this age were compared between the groups." As per convention, AOD 45o readings in Liley's zone 3 and the upper 20% of zone 27 were considered indicative of fetal anemia. 16, 17 Fetal blood (0.5 ml) was collected in ethylenediaminetetraacetic acid and diluted I : 50,000 in Isoton II solution (Coulter Electronics, Luton, U.K.) with a Dual Diluter III (Coulter Electronics). The purity and fetal origin of the samples were confirmed at the time by separate red blood cell volume distributions on a particle size analyzer, Coulter Channelyzer C-256 (Coulter Electronics) and later confirmed by the Kleihauer-Betke test. The laboratory spun microhematocrit was measured. The mean cell volume and the red blood cell count were analyzed by a Coulter counter model ZM (Coulter Electronics). The hemoglobin concentration, measured by a Coulter hemoglobinometer (Coulter Electronics) and the total nucleated cell count,' measured on the Coulter Counter model ZM, were estimated after a dilution of 1 : 500 with a Dual Diluter III and the addition of Zapoglobin (Coulter Electronics) to lyse red blood cells. Leishman-stained blood films were examined for the nucleated red blood cell count per 100 white blood cells, then expressed as an absolute nucleated red blood cell count. From new methylene blue-stained blood films, reticulocyte counts were expressed as a percentage of the 500 red blood cells counted. The absolute reticulocyte count was calculated from the percentage reticulocyte and erythrocyte counts. Fetal blood gas and acid-base parameters were determined within 10 minutes of collection with 0.1 ml heparinized blood in capillary tubes (ABL 330, Radiometer, Copenhagen). Values obtained were compared with published reference ranges. IS These are all performed as a part of the standard investigation of fetal anemia. Additional investigations were performed several months after delivery on 1 ml of fetal plasma routinely stored as per clinical practice and included albumin (grams per liter), total protein (grams per liter), and total bilirubin (micromoles per liter) measured by a Technicon RAXT (Bayer, Basingstoke), Published reference ranges were used for comparison. 19. 20 Parametric statistics were used only when histograms

Vaughan et al.

Volume 171, Number 1 Am J Obstet Gynecol

A.

B.

700

i...

i...

600

.!:!.

500

:::JS2'

400

~

80

!~

~S2' 60 • .c

i1

U

300

8~

:i

i1

200

ta:

100


.&

i1

i1

o

~a:

~~

i1

.c

f w

i1

i1

0

.&


20

•

40

[J

20

[J

0 [J

i1

·100 10

•

80

.!:!. ~

100

249

·20

30

10

Gestational age (weeks)

20

30

Gestational age (weeks)

Fig. 2. Difference (a) in (A) reticulocyte count (triangles) and (B) nucleated red blood cell count (squares) of each matched pair, with all but one RhD fetus having greater reticulocyte counts and all but two having greater nucleated red blood cell counts. Filled symbols, Hydropic fetuses; K, Kell.

Table II. Hematologic data Mean/median .d RhD-Kell Hemoglobin (gmIdl) Red blood cell count (X 1012/L) Mean cell volume (£I) Nucleated red cell count (x 109/L) Reticulocyte count (X 109/L) Reticulocyte count (per 100 red blood cells)

0.2 -0.2 13 1.9

95% Confidence intervals -0.7 -0.3 0.7 0.1

161 13

to to to to

0.9 0.2 26 51

68 to 262

5.4 to 28

Significance NS NS P = 0.04 P = 0.045

P= P=

0.007 0.009

NS, Not significant.

of the difference of each pair confirmed normally distributed data. Parametric and nonparametric 95% confidence intervals were computed, and comparisons were made by paired t or Wilcoxon tests, as appropriate. Blocked comparisons were by Fisher exact test. Pearson's correlation coefficient was only used when the least-squares method supported a linear relationship. Results

Median hematocrit in the Kell group was 19.1 %, and in the RhD group was 19.5, and median difference in hematocrit was 0.3 (95% confidence intervals - 1.5 to 2). Mean gestational age in the Kell group was 24.6 weeks and 24.5 in the RhD group; mean difference in gestational age was ~0.1 (95% confidence intervals - 0.6 to 0) (Fig. 1). Mean cell volume, nucleated red blood cell count, and reticulocyte percent and count were all lower in the Kell group (Table II) (Fig. 2). There was no correlation between the degree of anemia and the level of reticulocytosis in the Kell group, whereas there was a correlation in the RhD group (y = 53 - 1.2x, r = 0.63,

p<

0.05) (Fig. 3). In those with .10D45o performed after 24 weeks, none of the five Kell-positive fetuses had values indicating severe anemia, compared with four of the five RhD-positive fetuses (Table III). No relationship was found between the plasma bilirubin concentration and the degree of anemia in either group. Hemoglobin concentration and total red blood cell counts were similar in the two groups (Table II). Before the first transfusion both Kell and RhD fetuses had similar biparietal diameters (median difference 1 mm, 95% confidence intervals - 2 to 5, P = 0.3) and abdominal circumferences (median difference 4 mm, 95% confidence intervals -9 to 16, P = 0.7). No fetus was growth retarded. No significant differences were found in fetal pH, Po 2 , Pco 2 , or base equivalents between the groups (Table IV). In both gr'oups pH values were all >7.3 and base equivalents,all <5.0. Hypoxemia (Po 2 > 2 SD below reference range) occurred in one fetus in the Kell group and two in the RhD group, whereas Pco 2 values were all normal in the RhD group but below the reference range in two Kell group fetuses. 16 Both Kell and RhD groups had similar bilirubin

250 Vaughan et al.

0

m

a:

;I!. II)

~ ()

50

50

40

40

30

0

20

a:

10

'3 () += CD

July 1994 Am J Obstet Gynecol

30

•• • •

•

••

0

•

• • •

10

0 0

30

20

10

0

20

0

20

10

30

Haematocrit (%)

Haematocrit (%)

RhO

K

Fig. 3. Relationship between reticulocytosis and hematocrit in Kell (filled symbols) and RhD (unfilled symbols) fetuses. Squares, Hydropic fetuses; circles, nonhydropic fetuses. Regression line for RhD group (y = 53.2 - 1.2x) indicates progressive reticulocytosis in response to degree of anemia. RBC, Red blood cell.

Table III. Amniotic fluid bilirubin estimation by.10D450

.1.0D450 upper 20% Liley's zone 2

Kell.

RhD

(n = 5)

(n = 5)

o

4

Significance

p

=

0.02

concentrations, total protein, and albumin concentrations with seven (64%) in each group being hypoproteinemic and hypoalbuminemic (Table IV). Comment Maternal RhD alloimmunization causes anti-RhD to cross the placenta and coat fetal erythrocytes, which are hemolyzed in the reticuloendothelial system, producing fetal anemia and jaundice. The pathogenesis of anaemia produced by anti-Kell has been assumed to be similar. 16 Our data show that in spite of equally severe anemia Kell-alloimmunized fetuses had significantly lower amniotic fluid bilirubin concentrations and significantly less reticulocytosis and erythroblastosis than did RhD fetuses, suggesting a different pathogenesis. In their series of 127 alloimmunized fetuses Nicolaides et al. 21 observed absent reticulocytosis and erythroblastosis in three severely anemic fetuses. They suggested endstage hematopoietic failure as the mechanism. In contrast, we consider the more likely mechanism in Kell alloimmunization to be erythroid suppression, because the lack of reticulocytosis and erythroblastosis occurs independently of the degree of anemia and is not related to hydrops. Alternatively, it may be that hemoly-

sis of an early red blood cell progenitor simulates erythroid suppression. Anemia in RhD fetuses produces a compensatory increase in fetal erythropoiesis with the level of the response depending on the degree of anemia. 17 Reticulocytosis as a percentage of red blood cells occurs when the hemoglobin concentration deficit is ;::.: 2 gm/dl, whereas erythroblastosis has been suggested to occur only when the hemoglobin deficit exceeds 7 gmJdp l Our study demonstrated a significantly elevated reticulocyte percentage, absolute reticulocyte count, and erythroblast count in the RhD group, reflecting the known reticulocytosis and erythroblastosis that occurs in RhD disease. In contrast, the Kell group showed little evidence of reticulocytosis and erythroblastosis, with values in the normal range for both reticulocytes and erythroblasts in eight of 11 anemic fetuses. 17 Furthermore, the known linear correlation between reticulocytosis and the severity of anemia21 was confirmed by our RhD group data, but was absent in the Kell group. Our inability to show a difference in plasma bilirubin concentrations between the groups is not unexpected. The cord plasma bilirubin levels in anemic RhD fetuses correlate poorly with .10D 450 7 and with the level of anemia,20 presumably because fetal bilirubin levels rapidly equilibrate with maternal levels across the placenta. In the late second and the third trimesters a .10D450 reading in Liley's zone 3 predicts the presence of fetal anemia with a sensitivity of 95%.22 However, in spite of the small numbers in this series, a significant difference was found between the two groups, with the Kell group fetuses consistently having misleadingly low .10D450 readings. This work supports the anecdotal observations of many authors 2 • 8-12 who have suggested that

Vaughan et al.

Volume 171. Number 1 Am J Obstet Gynecol

251

Table IV. Biochemical data Mean/median L1 RhD-Kell

pH P0 2 (kPa) Pc0 2 (kPa) Base equivalents (mmoI!L) Albumin (gm/L) Total protein (gm/L) Bilirubin (Il-moI!L)

-0.03 -0.7 0.6 0.1 0.3 1.0 11

amniotic fluid ~OD450 readings may be inaccurate in predicting the severity of fetal anaemia in Kell alloimmunization. In the series of Caine and Mueller-Heubach, 2 only one of four severely affected fetuses with ~OD450 readings measured within a week of delivery had levels > 80% of zone 2 and none had readings in zone 3. A role for anti-Kell antibodies in mediating erythroid suppression is plausible, given both its protein structure and its expression both early in fetal life and in early erythroblasts. By means of molecular cloning to detect the amino acid sequence, the structure of the Kell protein has recently been predicted with hydropathy analysis. 23 The Kell protein has a single membrane spanning region and is predominantly extracellular, with the 24 Kell antigen determinants on the exposed portion of the molecule. It has structural homologic characteristics with the protein family of zinc neutral endopeptidases, which process a variety of peptide hormones 2 '; therefore, if Kell is a factor involved in erythroid growth or differentiation, anti-Kell may interfere with this function. Kell antigens are expressed early in the ontogeny of the red blood cell, as evidenced by their expression on the K562 erythroleukemia cell line after induction of hemoglobin synthesis 24 and are well developed on fetal red blood cells, where expression begins by about the tenth week of fetal life. 25 An alternative mechanism for the anemia mediated by Kell antibodies involves the hemolysis of red blood cell precursors at such a primitive stage that the total mass of precursors is small, without significant hemoglobin to produce the anticipated amount of bilirubin in the amniotic fluid, which would occur if the anemia were related to maternal anti-D. Whatever the mechanism mediating the erythroid suppression, our data have significant management consequences for a woman whose fetus is at risk for Kell alloimmunization. Unlike RhD alloimmunization, in which serial amniocentesis is standard management, amniocentesis is an inappropriate investigation in Kell alloimmunization. Women identified by serologic screening to have anti-Kell antibodies and who have serologically proved Kell-positive heterozygous (Klk) partners should undergo fetal blood sampling at 20 to 24 weeks to ascertain the fetal Kell status. Kell-negative (klk) fetuses

95% Confidence intervals

-0.1 -1.6 - 0.1 -0.9 -2.8 -3.0 -15

to to to to to to to

0 0.4 1.3 1.0 3.2 4.0 36

Significance

p= P= P= P=

P=

P= P=

0.06 0.17 0.09 0.97 0.91 0.68 0.29

are not at risk and require no further intervention. Kellpositive fetuses (Klk) identified either by fetal blood sampling or by the partner being homozygous for Kell (KlK) require serial ultrasonography and fetal blood sampling to monitor the development offetal anemia. This is independent of a good obstetric outcome in prior pregnancies because, unlike anti-RhD-affected pregnancies, the obstetric history is not predictive in the index pregnancy. REFERENCES 1. Clarke CA. Mollison PL. Deaths from rhesus haemolytic disease of fetus and newborn. 1977-1987. J R Coli Physicians Lond 1989;23:181-4. 2. Caine ME, Mueller-Heubach E. Kell sensitization in pregnancy. AM J OBSTET GVNECOL 1986; 154:85-90. 3. Giblett ER. A critique of the theoretical hazard of interversus intra-racial transfusion. Transfusion 1961; 1:233. 4. Mollison DL, Engelfriet CP, Contreras M. Blood transfusion in clinical medicine. 9th ed. Oxford: Blackwell Scientific, 1993:581-2. 5. Bowman J, Pollack JM. Manning FA, Harman CR, Menticoglou S. Maternal Kell blood group alloimmunization. Obstet Gynecol 1992;79:239-44. 6. CopelJA, Scioscia A. Grannum PA, Romero R, Reece EA. HobbinsJC. Percutaneous umbilical blood sampling in the management of Kell isoimmunization. Obstet Gynecol 1986;67:285-90. 7. Liley AW. Liquor amnii analysis in management of pregnancy complicated by rhesus immunization. AM J OBSTET GVNECOL 1961;82:1359-70. 8. Leggat HM, GibsonJM, Barron SL, Reid MM. Anti-Kell in pregnancy. Br J Obstet Gynaecol 1991;98:162-5. 9. Berkowitz RL, Beyth Y, Sadovsky E. Death in utero due to Kell sensitization without excessive elevation of the delta OD 450 value in amniotic fluid. Obstet Gynecol 1982;60: 746-9. 10. Barss VA, Benacerraf BR. Green MF, Phillippe M, Frigoletto FD Jr. Sonographic detection of fetal hydrops: a report of two cases. J Reprod Med 1985;30:893-4. 11. Birkenfeld A, Yaffe H. Sadovsky E. Sinusoidal fetal heart rate pattern with severe fetal anaemia. Br J Obstet Gynaecol 1980;87:916-20. 12. Farr V, Gray E. Pregnancy outcome in mothers who develop Kell antibodies. Scott Med J 1988;33:300-3. 13. Daffos F. Capella-Pavlovsky M. Forester F. Fetal blood sampling during pregnancy with the use of a needle guided by ultrasound: a study of 606 conservative cases. AM J OBSTET GVNECOL 1985; 153:655-60 .. 14. Nicolini U, Nicolaidis P, Fisk NM, Tannirandorn Y. Rodeck CH. Fetal blood sampling from the intrahepatic portion of the umbilical vein: experience with 210 cases. Obstet Gynecol 1990;76:47-53. 15. Nicolaides KH. Rodeck CH. Mibash
Mandelbrot et al.

17. Nicolaides KH, Rodeck CH, Millar DS, Mibashan RS. Fetal haematology in rhesus isoimmunisation. BM] 1985;290: 661-3. 18. Nicolaides KH, Economides DL, Soothill PW. Blood gases, pH, and lactate in appropriate- and small-for-gestationalage fetuses. AM] OBSTET GYNECOL 1989; 161 :996-1 00 1. 19. Nicolaides KH, Warenski ]C, Rodeck CH. The relationship of fetal plasma protein concentration and hemoglobin level to the development of hydrops in rhesus isoimmunization. AM] OBSTET GYNECOL 1985;152:341-4. 20. McKenzie IZ, Bowell P], Castle BM, Selinger M, Ferguson ]F. Serial fetal blood sampling in the management of pregnancies complicated by severe rheus (D) isoimmunization. Br] Obstet Gynaecol 1988;95:753-8.

July 1994 Am J Obstet Gynecol

21. Nicolaides KH, Thilaganathan B, Rodeck CH, Mibishan RS. Erythroblastosis and reticulocytosis in anemic fetuses. AM] OBSTET GYNECOL 1988;159:1063-5. 22. Bowman ]M. The management of Rh-isoimmunization. Obstet Gynecol 1978;52:1-16. 23. Lee S, Zambas ED, Marsh WL, Redman CM. Molecular cloning and primary structure of Kell blood group protein. Proc Natl Acad Sci USA 1991;88:6353-7. 24. McGinniss MH, Dean A. Expression of red cell antigens by K562 human leukaemia lines before and after induction of haemoglobin synthesis by hemin. Transfusion 1985;25: 105-9. 25. Marsh WL, Redman CM. The Kell blood group system: a review. Transfusion 1990;30:158-67.

Thrombocytopenia in pregnant women infected with human immunodeficiency virus: Maternal and neonatal outcome Laurent Mandelbrot, MD: Isabelle Schlienger, MD,b Andre Bongain, MD: Alain Berrebi, MD,d Jean Claude Pons, MD,b Nicole Ciraru-Vigneron, MD,e Jean Yves Gillet, MD: IlQd Jean Fran~ois Delfraissy, MDb Paris, Clamart, Nice, and Toulouse, France OBJECTIVES: Our purpose was to evaluate the prevalence of thrombocytopenia related to human immunodeficiency virus among seropositive pregnant women and its impact on maternal and neonatal outcome. STUDY DESIGN: A retrospective survey of all deliveries of women infected with human immunodeficiency virus in 14 maternity units in France over a 6-year period collected data on mothers who had thrombocytopenia < 100 . 10"/L and their infants. RESULTS: Among 890 women, 29 were thrombocytopenic (3.2%, 95% confidence interval 2.1 % to 4.3%). Thrombocytopenia appeared directly related to human immunodeficiency virus infection in 25 of these women. During pregnancy 16 patients were treated for thrombocytopenia with zidovudine, corticosteroids, or high-dose intravenous gamma globulin. Zidovudine was effective in five of seven cases, and intravenous gamma globulin was effective in five of 11 cases. Cesarean sections were performed in 13 of 29 women. Abnormal intrapartum or postpartum bleeding was recorded in five cases. Among 28 infants for whom neonatal platelet counts were available, only one had thrombocytopenia < 100· 10"/L at birth; he went on to have early-onset acquired immunodeficiency syndrome. CONCLUSIONS: The incidence of fetal or neonatal thrombocytopenia appears low and may not justify invasive sampling or routine cesarean delivery. Therapy with zidovudine or intravenous gamma globulin should be considered for women with severe thrombocytopenia, because of the risk of maternal hemorrhage. (AM J OBSlEr GVNECOL 1994;171:252-7.)

Key words: Human immunodeficiency virus infection, acquired immunodeficiency syndrome and pregnancy, thrombocytopenia From the Department of Obstetrics-Gynecology, Hopital Robert Debre: the Departments of Internal Medicine and Obstetrics, Hopital Antoine BecLere,h the Department of Obstetrics-Gynecology, Centre Hospitalier Universitaire Nice,' the Department of Obstetrics-Gynecology, Centre Hospitalier Universitaire Toulouse, a and the Department of Obstetrics-Gynecology, Hopital Lariboisiere.' Received for publication August 26, 1993; revisedJanuary 4, 1994; accepted January 7, 1994. Reprint requests: Laurent Mandelbrot, MD, Maternite, Hopital Robert Debre, 48 Boulevard Serurier, 75019 Paris, France. Copyright © 1994 by Mosby-Year Book, Inc. 0002-9378/94 $3.00 + 0 6/1/54143

252

Thrombocytopenia has been reported in 4% to 20% of human immunodeficiency virus (HIV)-infected, mostly male, adults,I-' but to date there have been only case reports of thrombocytopenia in HIV-positive pregnant women. 4 HIV-associated thrombocytopenia is thought to involve several mechanisms, including infection of megakaryocytes and peripheral destruction by antibodies. 5 The potential for obstetric complications,