Management of fetal hemolytic disease by cordocentesis

Management of fetal hemolytic disease by cordocentesis II. Outcome of treatment Carl P. Weiner, MD,a Roger A. Williamson, MD,a Katharine D. Wenstrom, MD,a Susan L. Sipes, MD,a John A. Widness, MD: Stanley S. Grant, RN,a and Louise Estle, RN Iowa City, Iowa Forty-eight of 128 pregnancies complicated by maternal red blood cell alloimmunization (49%) received a total of 142 intravascular transfusions (range, 1 to 7) for treatment of severe anemia (hematocrit, :530%). Thirteen fetuses (27%) had hydrops when therapy was initiated. The overall survival rate was 96%. Eighty-five percent of survivors received two or more transfusions before delivery. The mean gestational age at initiation of therapy was 28 weeks (range, 18 to 36 weeks). Bleeding from uterine and umbilical cord puncture sites was not of clinical significance. The most common complication was fetal bradycardia (8%). Simple intravascular transfuson resulted in the replacement of fetal red blood cells with adult red blood cells and suppression of fetal erythropoiesis. By the completion of the second transfusion, on average, <1% of circulating red blood cells were fetal. Within 3 weeks of the second transfusion, the mean reticulocyte count was <1%. The rate at which the fetal hematocrit declined after a transfusion (exclusive of the first) was inversely related to gestational age (r = - 0.84, P < 0.0001), permitting a 4- to 5-week interval between transfusions after 32 weeks' gestation. A total of 78% of surviving neonates were delivered at term. Neonates transfused more than once antenatally required less phototherapy (75.8 ± 54 vs 165 ± 101 hours, P < 0.003) and, when delivered at term, fewer hospital days (4.8 ± 2 vs 8.6 ± 6 days, p = 0.Q1) compared with those transfused once. We conclude that the treatment of fetal anemia by intrauterine simple intravascular transfusion permits a term delivery in the majority of cases and is associated with high perinatal survival and low perinatal morbidity. (AM J OSSTET GVNECOL 1991 ;165: 1302-7.)

Key words: Fetal hemolytic disease, cordocentesis, fetal intravascular transfusion, immune hydrops fetalis, neonatal hyperbilirubinemia The first wide-scale fetal therapy was intraperitoneal transfusion for the treatment of fetal hemolytic disease, the success of which varied from center to center. Fetal intravascular transfusion replaced intraperitoneal transfusion as the method of choice for the treatment of fetal hemolytic anemia at the University of Iowa in 1985. We report our experience with the treatment of 48 severely anemic (hematocrit, ::;30%) fetuses who were identified using cordocentesis. The findings demonstrate that simple intravascular transfusion to maintain the fetal hematocrit consistently in a near-normal range (1) results in sustained suppression of fetal erythFrom the Fetal Diagnosis and Treatment Unit, Division ofMaternalFetal Medicine, Department of Obstetrics and Gynecology," and the Division of Neonatology, Department of Pediatrics,! University of Iowa College of Medicine. Supported in part by grant HD24494 from the National Institute of Child Health and Development (C.P.W.). Presented in part at the Tenth Annual Meeting of the Society for Perinatal Obstetricians, Houston, Texas, january 23-27,1990, and in part at the Thirty-seventh Annual Meeting of the Society for Gynecologic Investigation, St. Louis, Missouri, March 21-24, 1990. Received for publication December 26, 1990; revised April 2, 1991; accepted April 22, 1991. Reprint requests: Carl P. Weiner, MD, Department of Obstetrics and Gynecology, Universty of Iowa, Iowa City, IA 52242. 6/1/30439 1302

ropoiesis, (2) is safe enough to routinely allow a term delivery, and (3) is, as a result of the preceding, associated with low perinatal morbidity and mortality.

Methods Fetuses of alloimmunized pregnant women and in need of transfusion were identified exclusively with cordocentesis, as previously described. I Transfusion therapy was initiated when the fetal hematocrit was <30%, because this value is well below the 2.5th percentile for the normal fetus 2'::20 weeks. Transfusions were performed as a simple infusion of packed red blood cells (70% hematocrit) into the umbilical vein, as previously described. 2 Umbilical venous pressure, pH, Pco 2 , and P0 2 were measured and bicarbonate and base deficit were calculated by standard nomograms. The infusion of blood was halted if the pressure rose significantly above normaJ.3 Because of the duration of the procedure (20 to 30 minutes) and the volume infused (up to 200 ml), each fetus received pancuronium (0.3 mg/kg estimated fetal weight intravenously) to prevent fetal movement and furosemide (3 mg/kg estimated fetal weight intravenously) to maximize venous capacitance

Volume 165 umber 5, Part I

immediately before the transfusion. Our goal was a final fetal hematocrit between 45% and 50%. Initially, hydropic fetuses were treated no differently than nonhydropic fetuses. Hydrops was defined as frank anasarca-edema and fluid accumulation in at least two cavities. The transfused blood was negative for cytomegalovirus, hepatitis B, and human immunodeficiency virus; compatible with mother and fetus; buffy coat poor; washed in saline solution to minimize the risk of viral contamination; irradiated to eliminate the risk of graft-versus-host disease; and resuspended in saline solution. Transfusions were initially repeated at 2- to 3week intervals when the hematocrit was anticipated to be between 25% and 30%, assuming a 120-day adult red blood cell lifespan. Frequent transfusions minimized both the duration of fetal anemia and the stimulus for fetal erythropoiesis. The volume of blood infused was individualized for each fetus. We retrospectively tried several different formulas to estimate the volume necessary but found none were accurate enough. We therefore estimated the required volume on the basis of past experience, infused half the estimated requirement, and then measured the increase in the fetal hematocrit. By knowing the amount by which the hematocrit had increased after the first aliquot of blood, the total volume necessary to reach the desired final hematocrit could be easily calculated. In general, aliquots of 20 ml were used in fetuses <22 weeks, 30 ml for fetuses <26 weeks, 40 ml for fetuses <30 weeks, and 50 ml for fetuses <35 weeks. A transfusion consisted of two aliquots. Laboratory tests monitored during transfusion included a complete blood cell count (Technicon, Tarrytown, N.Y.) and a Kleihauer-Betke stain to determine the percent circulatory fetal red blood cells. These were performed in the central hospital laboratory. Umbilical venous pH, Peoe, and POe were measured in heparinized plasma (Instrumentation Laboratory, Lexington, Mass.). Data were maintained prospectively in a computerized data base. The results are presented as the mean ± SD unless stated. Analyses included paired and unpaired t tests, cross tabulation, and stepwise multiple regression. A p :::; 0.05 was considered to indicate a significant difference.

Results Forty-eight fetuses underwent 142 simple intrauterine intravenous transfusions for the treatment of fetal hemolytic anemia. The overall survival rate was 96%. Thirty-seven fetuses (77%) received two or more transfusions (maximum, seven). The mean gestational age at the first transfusion was 28 ± 5 weeks (range, 18 to 36 weeks). The mean hematocrit before the first transfusion was 20% ± 8%

Cordocentesis and fetal intravascular transfusion

1303

(range, 4% to 30%). In subsequent transfusions, the mean hematocrit before the procedure was 27% ± 6%. The mean volume of packed red blood cells administered was 81 ± 38 ml (maximum, 200 ml). The mean hematocrit at the end of the transfusion was 47% ± 5%. No transfusion attempt was unsuccessful. Thirteen fetuses (27%) were hydropic when therapy was initiated. Although the hydrops resolved before delivery in all cases, the time required for complete resolution varied from 48 hours to 6 weeks. There were no deaths among nonhydropic fetuses, and the survival rate for hydropic fetuses was 85% (11113). After the transfusion of 10 fetuses, the hematocrit was again measured 24 to 48 hours later. In all instances, this second hematocrit was within 3% of the final hematocrit obtained on completion of the transfusion. Our practice was strongly affected by our early experiences. The first fetus transfused was delivered electively at 35 weeks, I week after its first transfusion. This neonate experienced severe hyperbilirubinemia in spite of multiple, double-volume exchange transfusions and is deaf. This episode strengthened our resolve to achieve a term delivery whenever possible. There were no other elective, preterm deliveries of treated fetuses. Two hydropic fetuses died shortly after the first transfusion. Both of these cases led to a change in management. The first was an acidotic (umbilical venous pH = 7.31 before transfusion, 2 SDs from the norm = 7.39), hydropic 22-week fetus treated early in our experience (transfusion No. 26). The hematocrit was increased from 5% to 45%. One hour after the technically uneventful transfusion, a bradycardia developed and the fetus died shortly thereafter. There was no evidence of umbilical cord trauma at delivery 24 hours later. We assumed this death was secondary to volume overload, although we had not monitored intravascular pressure. We therefore added umbilical venous pressure measurement to our protocol for all subsequently treated fetuses. We have recently come to believe that transfusionrelated acidemia also may be a factor in the death of the hydropic fetus. This impression originated with our second loss but was supported by the first. Although the umbilical venous pressure of this acidotic (umbilical venous pH = 7.34) 25-week fetus remained stable during the transfusion and the heart rate was normal, the acidemia became profound (umbilical venous pH = 7.26). A bradycardia developed 10 minutes after the transfusion was completed and did not recover; the fetus died. In light of preexisting symptomatic maternal respiratory compromise (chronic lung disease) and the moribund state of the fetus, an emergency cesarean section was not performed. There was no evidence of umbilical cord trauma at delivery 48 hours later. We then modified our protocol for the first trans-

1304 Weiner et al.

I'\o\'cmhcr 1991 Am.J Obstt't Cynccol

30

% Fetal RBCs After Transfusion

0

r--------------------,

15

20

10

10

5

2

23 15

13 9

•

--.Jo

---'==:JI-.---oii==27 20

% Reticulocyte Prior to Transfusion

9

345

6

4

4 6

Transfusion Number Fig. 1. Effect of intra\'ascular transfusion on fetal reticulocyte count (percent bdiJre transfusion) and percent circulating fetal red blood cells (RBCs) (after transfusion) as estimated with KleibauerBetke stain.

fusion of hydropic fetuses to mmlmlze the risks of volume overload and acidemia. The anemia is corrected in two steps 24 to 48 hours apart. The hematocrit is increased to approximately 25% (if the umbilical venous pressure is acceptable) during the first step and to 45% to 50% during the second step. Should the umbilical venous pH decline below 7.30, we infuse 4.20/, sodium bicarbonate in 1 ml aliquots and recheck the pH 5 minutes later. The acidemia in three hydropic fetuses with a decline in umbilical venous pH <7.30 during transfusion has since been successfully treated with this protocol. There were two nonlethal other complications directly related to the transfusion. In one woman whose fetus was seen initially with hydrops at 18 weeks, StajJh.~loc(jceus ejJidennidis amnionitis developed several days after a transfusion performed at 25 weeks. She was delivered after intractable preterm labor. One woman was delivered at 35 weeks' gestation on an emergency basis because of a prolonged fetal bradycardia after inadvertent puncture of the umbilical artery before the transfusion was initiated. Puncture of the umbilical artery is a recognized risk for fetal bradycardia.' These children are alive and well (18 and 43 months of age, respectively, at submission of manuscript). The overall incidence of bradycardia was 8%. Except for the one episode leading to cesarean delivery and the two described episodes associated with fetal death, each resolved within 10 minutes. All neonates who experienced a bradycardia and survived are said to be developing normally by either the pediatrician or a parent. There were five obstetric complications not temporally related to transfusion. Three women experienced preterm premature rupture of membranes followed by preterm labor 1 and 3 weeks after the last of several transfusions. Two growth-retarded fetuses were deliv-

ered because of abnormal results of heart rate testing 2 to 3 weeks after the last transfuson. These five deliveries each occurred between 32 and 36 weeks and were associated with an uneventful neonatal course. Simple intravascular transfusion was well tolerated by the fetus. Because the infused blood was acidotic (pH, 6.9 to 7.0), umbilical venous pH declined (7.40 ± 0.02 to 7.35 ± 0.03, jJ < 0.00(1) during the transfusion. Severe fetal acidosis necessitating bicarbonate was seen only in fetuses with a hematocrit s 15%. Transfusion improved fetal oxygen delivery by increasing hematocrit; the umbilical venous POe remained stable (39 ± 7 to 40 ± 6 mm Hg, p = 0.16). In spite of the fairly large intravascular volume load, the umbilical venous pressure (corrected for amniotic fluid pressure) of nonhydropic fetuses rose only 5 ± 4 mm Hg (p < 0.00(1) over baseline (range, 0 to 13 mm Hg). It remained within the normal range «10 mm Hg) during 69% of procedures. Four of the five hydropic fetuses in whom umbilical venous pressure was measured had ail elevated pressure before the first transfusion (range, 11 to 15 mm Hg). In three, the umbilical venous pressure rose abruptly (l0 to 12 mm Hg) in spite of limiting the volume transfused to an amount adequate to raise the hematocrit to between 25% and 30%. When the umbilical venous pressure was measured again in these three fetuses just before the second transfusion 24 to 48 hours after the first, it had declined to a mean of 6 mm Hg (p = 0.(7) and remained in the normal range throughout the second transfusion. In the remaining two hydropic fetuses studied, the umbilical venous pressure declined into the normal range during the course of the first transfusion. This sustained decline was not observed in any nonhydropic fetus. The fetal reticulocyte count dropped after the ane-

Cordocentesis and fetal intravascular transfusion

Volume 16:; I\umhcr 5. Pa ... I

13

L!.

12

L!. L!.

L!.ltt.

11

L!.

10

L!. L!.

9 /I

HCr

per week

1305

ltt.

L!.

8

L!.

7

n= L!.

(22/28)

r = -0.34/-0.84

L!.

P - 0.11/ 0.0001

L!.

6

5 4 3 18

20

22

24

26

28

30

32

34

36

38

gestation (w) Fig. 2A. Effect of gestational age on decline in hematocrit after transfusion. O/JI'1i Iriall,,11'5. Decline in hematocrit between first and second transfusions: clused Iriangll',l. decline for all subse
mia had been corrected by transfusion. Within 3 weeks of the second transfusion, the mean reticulocyte count was < 1% (Fig. I). In addition. the percentage of cir· culating fetal red blood cells by the conclusion of the third transfusion was < 1% (Fig. I l. To determine with greater precision how often transfusions should be repeated, we calculated the rate the fetal hematocrit declined between transfusions performed January 1985 to October 1989 (32 fetuses, 91 transfusions) by subtracting the pretransfusion hematocrit from the final hematocrit of the previous transfusion and dividing the result by the intervening weeks. There was an inverse relationship between the rate the hematocrit declined and the gestational age at which the transfusion was performed (Fig. 2A). The scatter was quite wide for the decline in hematocrit per week between first and second transfusions (1' = - 0.34, !J = 0.11). However, the correlation between hematocrit decline and gestational age for all subsequent procedures was very good (1' = - 0.84; !J < 0.000 I; standard error of the estimate 1.082). Fetal blood contributed minimally to the hematocrit after the second transfusion: therefore a differing rate of hemolysis for difkrent populations of red blood cells cannot account for the close relationship between hematocrit decline and gestational age. The relationship between gestational age and the rate of hematocrit decline was confirmed prospectively in the next 13 transfusions performed between November 1989 through mid-January 1990. The rate of decline in hematocrit between procedures closely approxi-

13 12 11 10

9 /I

Hcr

per week

8

7 6 5 4

20

~

~

~

~

30

~

34

36

38

gestation (w) Fig. 28. Prospective confirmation of relationship shown in Fig. ~A with new data. 0Pl'll Iriangles. Decline in hematocrit between first and second transfusions: elliSI'd ITinllgll's. decline for all suhsequent transfusions. Regression line is from Fig. 2A. HCT, Hematocrit. mated the regression line obtained for the first data set (Fig. 2B). The gestational age influence on the rate of hematocrit decline after the second transfusion required a 2- to 3-week interval between transfusions at 20 weeks' gestation while allowing a 4- to 5-week interval after 32 weeks' gestation (assuming similar final hematocrits).

1306 Weiner et al.

November 1991 Am J Obstet Gynecol

Table I. Neonatal outcomes of fetuses treated with intravascular transfusion (N

Survival Hydrops Nonhydropic Term delivery Exchange transfusion

o 1

>1

Simple transfusion

o I

46)

=

No.

%

46

96 85

13 35

100

74

32

35 8 3

76 17

33

72 15

6

13

>1

change transfusion. When delivered at term, neonates antenatally transfused more than once spent less time in the hospital. The mean peak total bilirubin was lower in term neonates whose reticulocyte count at birth was :0;1 % (10.8 ± 6 mg/dl, n = 8 vs 15.8 ± 6 mg/dl, n = 10; P < 0.05). There was a trend toward reduced phototherapy when the amount of circulating fetal red blood cells was :0;5% (61 ± 59 hours, n = 13 vs 133 ± 151 hours, n = 5; P = 0.08). As the number of patients with successful pregnancies grew over the 6-year period, we noticed a change in both patients' and referring physicians' attitudes toward subsequent pregnancies. Over the last 2 years, three women have chosen to undergo another pregnancy. In two of the three, the first treated fetus had presented with hydrops.

7

7

Table II. Neonatal clinical features of fetuses treated with intravascular transfusion Mean

Gestation (wk) Hematocrit at delivery (%) Reticulocyte count (%) Kleihauer-Betke (fetal red blood cells in umbilical cord smear) Duration of phototherapy (hrs) Intravascular transfusion > 1 Intravascular transfusion Double-volume exchange transfusion 1 Intravascular transfusion > 1 Intravascular transfusion Hospital days (term delivery) 1 Intravascular transfusion > 1 Intravascular transfusion

36.7 33 2 10

95

165* 76

I~ 2 8 3 17

78

101 54

57.1 t 20.7 8.6:j:

6

4.8

2

*P <

0.0003. 0.06. :j:p = 0.01.

tp

=

Eight-two percent (36/44) of survlVlng neonates without heart rate abnormalities associated with growth retardation were delivered at term (Table I). Eleven of the 46 survivors (includes nonelective preterm delivery) (24%) underwent one or more double-volume exchange transfusions. Eight of the I I required only one exchange. Two of the eight were performed at outside institutions per routine rather than as treatment for hyperbilirubinemia. Simple transfusion for treatment of anemia was performed after discharge in 13 neonates (28%). Each was associated with a delayed return of reticulocytes to the peripheral circulation rather than early neonatal hemolysis. Neonates transfused more than once antenatally (and therefore delivered at least 9 weeks after transfusion therapy had begun) had a less complicated postnatal course than infants transfused only once (Table II). While not all data were available, they did receive less phototherapy and had a near-significant decrease in the likelihood of receiving a double-volume ex-

Comment

Our experience demonstrates that simple intravascular transfusion for the treatment of fetal hemolytic anemia can lead to an excellent perinatal outcome. It is safe in experienced hands and allows a term delivery in the absence of other obstetric indications. Fetal anemia is the direct cause of many of the sequelae of hemolytic disease. The data suggest that the maintenance of the fetal hematocrit at near-normal levels throughout gestation accounted for the overall good neonatal outcome. The selection of a hematocrit that was actually <2.5th percentile for gestation as the threshold to initiate transfusion therapy was initially a pragmatic decision. Although we were reticent to begin therapy knowing that the fetus could tolerate a lower hematocrit well, we were unsure how rapidly a mild anemia might worsen. In hindsight, our selection appears to have been reasonable because several fetuses evaluated 1 would have been transfused unnecessarily if the 2.5th percentile had been selected for the threshold, and hydrops developed in several high-risk fetuses with normal hematocrits over a 2-week interval. The perinatal survival rate of 96% is similar to that in a slightly smaller series reported by Harman et al. 5 (44 fetuses). However, this group performed their transfusions at closer intervals and delivered at earlier gestational ages. Although it was not our purpose to compare intravascular transfusion with intraperitoneal transfusion for the treatment of fetal hemolytic anemia, Harman et al. did. This is one of the few centers that has a large published experience with both intraabdominal and intravascular transfusion. They found an improved perinatal outcome with intravascular transfusion compared with intraperitoneal transfusion. Although another group reported a similar survival rate with a combination of amniotic fluid studies, intraperitoneal transfusion, and preterm delivery,6 this report seems an exception. 5 . 7-12 In addition, some fetuses

Volume 165 Number 5, Part

Cordocentesis and fetal intravascular transfusion

1307

I

transfused solely on the basis of amniotic fluid studies may not in fact need transfusion and thus may bias survival statistics in a favorable way. In a recent series of untreated isoimmunized pregnancies whose changes in optical density at 450 nm measurements were in zone III, Frigoletto et al.'3 noted that four of II fetuses (37%) had a hematocrit 2:30% at delivery and thus would not have qualified for transfusion by our criteria. Our experience l ' with the relationship of the change in optical density at 450 nm to hematocrit is similar to that of Frigoletto et al. In our study fetuses with hydrops had elevated umbilical venous pressures. We initially assumed this reflected obstructed venous return. However, the umbilical venous pressure characteristically fell into the normal range within 48 hours of the first transfusion. This argues against hepatomegaly and venous obstruction as a sole explanation for the increase. Rather it suggests that the increased umbilical venous pressure in hydrops is associated with myocardial dysfunction, perhaps as the result of tissue hypoxia associated with a critically low oxygen-carrying capacity. Transfusion would improve cardiac function by eliminating hypoxemia. The only two losses were in severely hydropic fetuses. Each had a hematocrit <10% and a preexisting metabolic acidosis. In spite of the critical fetal condition, we do not feel these losses were inevitable. Each led to a modification in the transfusion protocol. The role of volume overload in the first loss is unclear and while fetal exchange transfusion would minimize the risk of volume overload it would not prevent the acidemia and would add greatly to the procedure time. On the basis of the decline in fetal erythrocytes and reticulocytes, it would appear that fetal erythropoiesis can be suppressed within 6 weeks of initiating transfusion therapy. Alternatively, rapid hemolysis of erythroid progenitor cells on or before release from the marrow could account for this observation. However, the finding of erythroid hypoplasia in marrow aspirates of treated neonates argues against the latter possibility as a sole explanation. 15 Further, we observed that neonatal hyperbilirubinemia was directly related to the reticulocyte count at birth and was more severe when only one antenatal transfusion had been performed. Although further study is needed, the 24% incidence of neonatal exchange transfusion in our series is less than that of neonates managed by either intraperitoneal transfusion or preterm delivery or both. 13.•• Cordocentesis and intravascular transfusion are deceptively simple. In practice, they require an experienced team and a laboratory capable of performing a variety of tests on a small volume. Our series was drawn from a delivery base approaching 60,000 per year. It is not feasible from the standpoint of personnel and expense for every hospital to offer a similar service.

These are facilities that must be regionalized to maximize both safety and efficacy. The prior recommendation 7 that a center perform 2: 12 intraperitoneal transfusions per year or refer the patient to an experienced center is applicable to intravascular methodology. In conclusion, treatment of fetal anemia by simple intrauterine intravascular transfusion is associated with a high survival rate, permits term delivery, and would appear to reduce immediate neonatal morbidity.

REFERENCES I. Weiner CP, Williamson RA, Wenstrom KD, Sipes SL,

Grant SS, Widness JA. Management of fetal hemolytic disease by cordocentesis: I. Prediction of fetal anemia. AM J OBSTET GYNECOL 1991;165:546-53. 2. Weiner CP, Pelzer GD, Heilskov J, Wenstrom KD, Williamson RA. The effect of intravascular transfusion on umbilical venous pressure in anemic fetuses with and without hydrops. AM J OBSTET GYNECOL 1989; 161: 1498501. 3. Weiner CP, Heilskov J, Pelzer G, Grant S, Wenstrom K, Williamson RA. Normal values for human umbilical venous and amniotic fluid pressures and their alteration by fetal disease. AMJ OBSTET GYNECOL 1989;161:714-7. 4. Weiner CPo Cordocentesis for diagnostic indications: two years' experience. Obstet Gynecol 1987;70:664-8. 5. Harman CR, BowmanJM, Manning FA, Menticoglou SM. Intrauterine transfusion - intraperitoneal versus intravascular approach: a case-control comparison. AM J OBSTET GYNECOL 1990;162: 1053-9. 6. Watts DH, Luthy DA, Benedetti TJ, Cyr DR, Easterling TR, Hickok D. Intraperitoneal fetal transfusion under direct ultrasound guidance. Obstet Gynecol 1988;71: 84-8. 7. Bowman JM. Hemolytic disease (erythroblastosis fetalis). In: Creasy RK, Resnik R, ed. Maternal-fetal medicine: principles and practice. 2nd ed. Philadelphia: WB Saunders, 1989:613-55. 8. Palmer A, Gordon RR. A critical review of intrauterine fetal transfusion. Br J Obstet Gynaecol 1976;83:688-93. 9. Robertson EG, Brown A, Ellis MI, Walker W. Intrauterine transfusion in the management of severe rhesus isoimmunization. Br J Obstet Gynaecol 1976;83:694-7. 10. Hamtilon EG. Intrauterine transfusion. Safeguard or peril? Obstet Gynecol 1977;50:255-60. 11. Frigoletto FD, Umansky I, Birnholz J, et ai. Intrauterine fetal transfusion in 35 fetuses during fifteen years. AM J OBSTET GYNECOL 1981;139:781-6. 12. White CA, Goplerd CP, Kisker CT, Stehbens JA, Kitchell M, Taylor JC. Intrauterine fetal transfusion, 1965-1976, with an assessment of the surviving children. AM J OBSTET GYNECOL 1978; 130:933-42. 13. Frigoletto FD, Greene MF, Benacerraf BR, Barss VA, Saltzman DH. Ultrasonographic fetal surveillance in the management of the isoimmunized pregnancy. N Engl J Med 1986;315:430-2. 14. Weiner CPo Percutaneous umbilical blood sampling. In: Sabbagha RE. ed. Diagnostic ultrasound applied to obstetrics and gynecology. 2nd ed. New York: JB Lippincott, 1987:419. 15. Giller RH, Widness JA, deAlarcon PA, Weiner CP, Johnson KJ. Postnatal anemia following intravascular intrauterine transfusion for isoimmune hemolytic disease: natural history and possible mechanisms. [Abstract 1571] Pediatr Res 1990;27:265A. 16. Ellis MI. Follow-up study of survivors after intra-uterine transfusion. Dev Med Child Neurol 1980;22:48-54.