INTRODUCTION

Current management of pregnancies complicated by Rhesus (Rh) isoimmunization involves invasive procedures, amniocentesis, and/or cordocentesis, to assess the severity of fetal anemia. Both procedures may require repetitions and are associated with complications such as infection, rupture of membrane, fetal bleeding, worsening of hemolysis, and fetal loss.^1,2 This has prompted many investigators to develop noninvasive tests to diagnose fetal anemia.^3,4,5,6 Doppler measurement of the peak systolic velocity of the fetal middle cerebral artery (MCA-PSV) has been shown by many researchers to correlate with the severity of anemia.^7,8,9 The validity of this test has not been confirmed and the use of Doppler studies in planning intrauterine transfusions is, to our knowledge, still controversial.

The objective of the current study is to examine the correlation of MCA-PSV with the severity of fetal anemia due to Rh isoimmunization.

METHODS

Over a period of 2 years starting from March 2000, we prospectively followed all consecutive procedures of intrauterine fetal blood sampling performed on pregnancies at risk of fetal anemia due to red cell isoimmunization. Women were referred to our center in the first trimester and fetal blood sampling by cordocentesis was performed with or without blood transfusion. This was determined on the basis of severely affected previous pregnancies, maternal serum red cell antibody titers 1:32, or increasing bilirubin concentrations in amniotic fluid, as detected by serial spectrophotometric examinations indicating a change in optical density at 450 nm.¹⁰ The gestational age was determined on the basis of the last menstrual period and was confirmed by ultrasonography. Fetal hydrops was defined as the presence of fluid in more than one body cavity and was diagnosed by ultrasonography at the time of blood sampling. The hospital board approved cordocentesis and fetal Doppler study, and an informed consent for cordocentesis was obtained from the patients.

Doppler examination of the middle cerebral artery was performed within 24 hours before cordocentesis in all cases. An axial view of the brain, including the thalami and the cavitas septi pellucidi, was obtained. The circle of Willis was visualized using color Doppler (HDI 3000, Advanced Technology Laboratories Ultrasound, Bothell, WA) and the middle cerebral artery of one side was examined close to its origin from the internal carotid artery, with the angle between the ultrasound beam and the direction of blood flow being kept as close to zero as possible. Two operators (MA and KA) performed all Doppler examinations at times of absent fetal body and breathing movements. Each operator performed three measurements of peak systolic velocity, and the highest velocity level was recorded. Intra- and interobserver reliability tests via the intra- and interclass correlation were calculated.

Although the main objective of this study was to explore the association between MCA-PSV and fetal hemoglobin level, it was necessary to preprocess these measurements prior to exploration of such an association for the reasons described below.

As both MCA-PSV and hemoglobin measurements vary with gestational age, they were adjusted for the effect of gestational age using multiples of the median (MoM) method. This adjustment was carried out by dividing the measured value by the corresponding expected value for a given gestational age as determined by the best fitting regression level. This is a procedure that has been adopted by previous authors.⁸

The diagnostic reference standard was the fetal hemoglobin concentration obtained from fetal blood samples. The normal range at 18 to 40 weeks gestation was derived from a previous report by Mari et al,⁸ who categorized degrees of fetal anemia as follows: mild (hemoglobin concentration from 0.84 to 0.65 times the median for gestational age), moderate (hemoglobin concentration from 0.65 to 0.55 times the median), and severe (hemoglobin concentration less than 0.55 times the median). MCA-PSV measurements were plotted against the gestational age of the fetus for all 300 control fetuses to obtain the expected/predicted values of MCA-PSV for varying gestational ages. The correlation between the MoM for the peak systolic velocity and MoM for hemoglobin concentration was explored using the best fitting multiple linear regression equation using maximum R² criteria with an aim of obtaining a parsimonious model. The receiver-operator-characteristic (ROC) curves were used to determine whether cutoff levels of peak systolic velocity could be sensitive and specific enough to predict the risk of moderate to severe anemia in the study population in general and in two subgroups: those in whom cordocentesis was performed for the first time (Group A) and those with repeat cordocentesis (Group B). The following characteristics of diagnostic test accuracy were calculated: sensitivity, specificity, and predictive values with 95% confidence interval.

Statistical analysis and plotting of ROC curves were performed with the SPSS statistical package (SPSS, Chicago, IL) and AccuROC for Windows (Accumetric Corporation, Montreal, Que.) P-Value less than 0.05 indicated statistical significance.

RESULTS

A total of 66 fetal blood sampling procedures with or without transfusion formed the study group. They were performed on 20 singleton pregnancies, with an average of 2.7 procedures per pregnancy. The mean gestational age of all procedures was 28.54.6 weeks. The frequency distribution of hemolytic antibodies in the women was as follows: D=14; Kell=3; D,c=1; D+E=1, and Kell+Jka=1.

Intra- and interobserver correlation coefficients were calculated to ascertain the agreement between two independent operators on Doppler measurements. Intraclass correlation measure was 0.94 with a 95% confidence interval of 0.93 to 0.95. Interclass correlation was 0.89 with a 95% confidence interval of 0.88 to 0.90. These estimates indicate high agreement between the operators in measuring MCA-PSV. Hence, all statistical analyses were carried out after averaging MCA-PSV values over the two operators.

Reference range for MCA-PSV obtained using measurements of 300 control fetuses is shown in Figure 1. This includes the predicted values and 95% confidence intervals. The regression equation for the predicted value of MCA-PSV (cm/second) based on gestational age (GA) in weeks followed an exponential model.

Figure 1.

Predicted median value and 95% confidence intervals of the MCA-PSV as a function of gestational age, derived from a study of 300 control fetuses.

Full figure and legend (29K)

Group A consisted of 20 initial cordocentesis procedures, in which mild anemia was reported in four fetuses, moderate anemia in three, and severe anemia in eight. Hydropic changes were present in six fetuses in this group. Group B consisted of 46 repeat cordocentesis, of which mild anemia was reported in nine cases, moderate anemia in seven, and severe anemia in 11. Fetal hydrops was present in 20 cases in this group. Figure 2 shows the varying MCA-PSV values with gestational age, with the reference range obtained from control subjects superimposed to show that the association between MCA-PSV and gestational age remains the same for the study fetuses.

Figure 2.

MCA-PSV values measured before first and repeat cordocentesis procedures for fetuses at risk of red cell isoimmunization, plotted on the population reference range. Group A represents the measurements before the initial cordocentesis, six fetuses were hydropic (solid triangles) and 14 were nonhydropic (open triangles). Group B represents measurements before the repeat cordocentesis, 20 fetuses were hyropic (solid circles) and 26 were nonhydropic (open circles).

Full figure and legend (31K)

Consequently, the relationships between the MoM of MCA-PSV and the MoM of hemoglobin concentration were evaluated again by the best fitting regression model. The correlation was weak and fitted a linear regression with borderline significance (Figure 3, R²=0.10, p=0.04).

Figure 3.

Fetal hemoglobin concentration as a function of MCA-PSV, both presented as multiples of the median. Group A represents the initial cordocentesis, six fetuses were hydropic (solid triangles) and 14 were nonhydropic (open triangles). Group B represents the repeat cordocentesis, 20 fetuses were hyropic (solid circles) and 26 were nonhydropic (open circles). Vertical lines indicate degrees of fetal hemoglobin levels: normal (0.84), mild (0.83 to 0.65 MoM), moderate (0.64 to 0.55 MoM), and severe anemia (<0.55 MoM).

Full figure and legend (45K)

Figure 4 shows the ROC curves that were generated when different threshold values of MCA-PSV MoM were used to screen moderate to severe fetal anemia in the group with initial cordocentesis and that with repeat cordocentesis. The areaS.E. under the ROC curve was not significant for the initial cordocentesis (0.480.14, p=0.85) and only weakly significant for the subsequent cordocentesis (0.680.09, p=0.04). The difference between the two ROC curves was not statistically significant. Both ROC curves indicated that specificity declined with improving the sensitivity values.

Figure 4.

ROC curves for the prediction of moderate to severe fetal anemia by MCA-PSV at the initial cordocentesis (Group A) and repeat cordocentesis (Group B).

Full figure and legend (13K)

The diagnostic accuracy for predicting moderate to severe anemia in the group with initial cordocentesis or the group with repeat cordocentesis at MCA-PSV 1.30, 1.50, and 1.55 MoM thresholds is shown in Table 1. The sensitivity was generally low, with the highest record being 50% at 1.30 threshold for the group with repeat cordocentesis. The specificity, however, was high reaching 100% for initial cordocentesis and 96% for repeat cordocentesis at MCA-PSV above the 1.50 MoM. The positive predictive value ranged from 60 to 100% and improved with increasing the MCA-PSV cutoff values before initial procedures. The negative predictive values were generally weak at initial procedures (47 to 48%) and somewhat improved with repeat procedures (66 to 74%). However, the negative predictive values did not show a notable change with modifying MCA-PSV cutoff values.

Table 1 - Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value of Three Different Thresholds of MoM for the MCA-PSV MoM in the Predicton of Moderate to Severe Fetal Anemia.

Full table

DISCUSSION

Several studies have reported that Doppler flow studies in the arterial and venous systems correlate with the degree of fetal anemia^3,4 and several reports suggested that thresholds of MCA-PSV could be used to predict anemia.^5,6,8,11

In this study from a single institution, we have shown that fetal MCA-PSV weakly correlates with the degree of fetal anemia in groups with or without previous intrauterine transfusion; however, velocity threshold levels were not sensitive enough to predict most fetuses with moderate to severe anemia. Furthermore, we have shown that abnormal peak systolic velocity results, though uncommon, were highly specific for significant anemia.

Regression analysis of repeat procedures may raise concerns on studying paired observations that do not seem independent. This can be addressed at several different levels. Firstly, the primary objective of using Doppler examination in Rh isoimmunization is to minimize the use of invasive testing, whether required initially or at a later gestational age. Secondly, the average 2-week lag between procedures is likely to be accompanied by cardiovascular changes that result in changes in peak systolic velocity. Thirdly, it would be of interest to examine the effect of previous transfusion on Doppler measurements at subsequent sampling. The weak sensitivity of MCA-PSV in predicting fetal anemia at initial cordocentesis improved somewhat at times of repeat procedures; however, it remained at 50% or below. Cordocentesis was performed on average at 26 weeks gestation for initial procedures and at 29 weeks for repeat procedures. There were three mortalities out of the 20 babies followed up, one at the initial cordocentesis at 25 weeks and two postnatally after repeat transfusions at 27 and 31 weeks. All three babies were hydropic. MCA-PSV and corresponding hemoglobin MoM values were, respectively, as follows: (0.93, 0.40), (0.46, 0.34), and (1.53, 0.45). MCA-PSV did not predict the severity of anemia in the first two babies but it was predictive only in the last baby. Data on blood gases could not be presented as they were not routinely measured in our series.

Different results for sensitivity and specificity of Doppler measurements depending on whether the cordocentesis was being performed for the first time or being repeated were noted. This could be due to several reasons. Firstly, there was a 2.4 weeks difference (26.8 vs 29.2 weeks) between the mean gestational age at Doppler measurements between first and repeat procedures, and it is known that fetal circulation increases linearly with advancing gestation, associated with a decrease in the resistance to flow in cerebral arteries presumably to compensate for the progressive fall in fetal blood PO₂ and increase in PCO₂.^12,13 Secondly, in pregnancies complicated by red cell isoimmunization, fetal oxygenation initially remains within the normal range due to normal placental function. However, when extreme anemia, hypoxia, and acidosis occur, cardiac output and blood velocity in the middle cerebral artery increase in proportion to the degree of fetal anemia, most likely due to decreased blood viscosity, leading to increased venous return and cardiac preload.¹² In severe anemia, there is cardiac decompensation, presumably caused by the associated hypoxia and lactacidemia, which may result in decreased blood velocity in major blood vessels.^14,15 This may explain the reduced sensitivity of different cutoff values of MCA-PSV in nontransfused fetuses and the improved sensitivity following correction of anemia.

Delta optical density 450 nm curves were generally used before initial cordocentesis except in cases with Kell isoimmunization and those at high risk of fetal anemia as suggested by previous history and antibody type and titer. In these cases, we directly proceeded to blood sampling to minimize the number of invasive procedures.

Fetuses at significant risk of anemia due to Rh isoimmunization may require frequent testing to diagnose and treat anemia. Therefore, it would be clinically relevant to study Doppler measurement repeatedly on the same fetus to evaluate its accuracy in prediction of anemia at different times.

References

1. Weiner CP, Okamura K. Diagnostic fetal blood sampling-technique related losses. Fetal Diagn Ther 1996;11(3):169–175. | PubMed |
2. Anandakumar C, Annapoorna V, Chee WY, Chia D, Bongso A, Ratnam SS. Fetal blood sampling and its complications related to the indications for fetal blood sampling. Aust N Z J Obstet Gynaecol 1993;33(3):259–261. | PubMed |
3. Mari G, Moise Jr. KJ, Deter RL, Carpenter Jr. RJ. Flow velocity waveforms of the umbilical and cerebral arteries before and after intravascular transfusion. Obstet Gynecol 1990;75(4):584–589. | PubMed |
4. Hecher K, Snijders R, Campbell S, Nicolaides K. Fetal venous, arterial, and intracardiac blood flows in red blood cell isoimmunization. Obstet Gynecol 1995; 85(1):122–128. | PubMed |
5. Delle Chiaie L, Buck G, Grab D, Terinde R. Prediction of fetal anemia with Doppler measurement of the middle cerebral artery peak systolic velocity in pregnancies complicated by maternal blood group alloimmunization or parvovirus B19 infection. Ultrasound Obstet Gynecol 2001;18(3):232–236. | PubMed |
6. Teixeira JM, Duncan K, Letsky E, Fisk NM. Middle cerebral artery peak systolic velocity in the prediction of fetal anemia. Ultrasound Obstet Gynecol 2000;15(3):205–208. | PubMed |
7. Mari G, Andrignolo A, Abuhamad AZ et al. Diagnosis of fetal anemia with Doppler ultrasound in the pregnancy complicated by maternal blood group immunization. Ultrasound Obstet Gynecol 1995;5:400–405. | PubMed |
8. Mari G, Deter RL, Carpenter RL et al. Non-invasive diagnosis by Doppler ultrasonography of fetal anemia due to maternal red-cell alloimmunization. N Engl J Med 2000;342:9–14. | Article | PubMed |
9. Deren O, Onderoglu L. The value of middle cerebral artery systolic velocity for initial and subsequent management in fetal anemia. Eur J Obstet Gynecol Reprod Biol 2002;101:26–30. | PubMed |
10. Liley AW. Liquor amnii analysis in the management of the pregnancy complicated by Rhesus sensitization. Am J Obstet Gynecol 1961;82:1359–1370. | PubMed |
11. Bahado-Singh RO, Oz AU, Hsu C et al. Middle cerebral artery Doppler velocimetric deceleration angle as a predictor of fetal anemia in Rh-alloimmunized fetuses without hydrops. Am J Obstet Gynecol 2000;183(3):746–751. | PubMed |
12. Soothil PW, Nicolaides KH, Rodeck CH, Clewell WH, Lindrige J. Relationship of fetal hemoglobin and oxygen content to lactate concentration in Rh isoimmunized pregnancies. Obstet Gynecol 1987;69:268–271. | PubMed |
13. Nicolaides KH, Sadovsky G, Cetin E. Fetal heart rate patterns in red blood cell isoimmunized pregnancies. Am J Obstet Gynecol 1989;161:351–356. | PubMed |
14. Nicolaides KH, Bilardo CM, Campbell S. Prediction of fetal anemia by measurement of the mean blood velocity in the fetal aorta. Am J Obstet Gynecol 1990;162:209–212. | PubMed |
15. Nicolaides KH. Studies on fetal physiology and pathophysiology in Rhesus disease. Semin Perinatol 1989;13:328–337. | PubMed |