Correspondence

Noninvasive Diagnosis of Fetal Anemia by Doppler Ultrasonography

N Engl J Med 2000; 343:66-68July 6, 2000

Article

To the Editor:

The results of the analysis of Doppler ultrasonography for the diagnosis of fetal anemia caused by maternal red-cell alloimmunization reported by Mari and colleagues (Jan. 6 issue)1 are promising, but a note of caution must be sounded. The investigators determined the accuracy of ultrasonography in the same group of fetuses in which they determined the threshold values for the peak systolic velocity in the middle cerebral artery at various gestational ages (>1.5 multiples of the median). The potential danger in this approach is obvious: since only 35 fetuses had moderate or severe anemia and only 23 of these 35 did not have hydrops, it would not be surprising if a substantial number of fetuses in a different population had anemia and would benefit from further intervention despite having values for peak velocity of systolic blood flow below the threshold set. This may be especially true for fetuses at either extreme of the range of gestational ages in the population studied. It would therefore seem premature to use their data to make clinical decisions.

David A. Nagey, M.D., Ph.D.
Johns Hopkins University, Baltimore, MD 21287

1 References

1
Mari G. Noninvasive diagnosis by Doppler ultrasonography of fetal anemia due to maternal red-cell alloimmunization. N Engl J Med 2000;342:9-14
Full Text | Web of Science | Medline

To the Editor:

Mari and colleagues report that measurements of the peak velocity of systolic blood flow in the middle cerebral artery have 100 percent sensitivity for the detection of moderate or severe anemia in fetuses. Reference ranges for fetal hemoglobin concentrations indicating both moderate and severe anemia were defined as multiples of the median gestational-age–dependent curves. The use of multiples of the median is tacitly dependent on the assumption that there is homoscedasticity (equality of variation) in the normal distribution of both these variables about their median curves on a logarithmic scale. Was this checked?

The investigators collected data on 265 normal fetuses and propose that fetal hemoglobin concentrations follow a log-normal distribution. The concentrations expressed as a function of gestational age were best filled by use of the following exponential function: hemoglobin concentration=e(2.84–8.55/GA), where GA is gestational age. They cite the work of Nicolaides et al.,1 who proposed that fetal hemoglobin concentrations increased linearly from 11 g per deciliter at 17 weeks' gestation to 15 g per deciliter at 40 weeks' gestation, with a standard deviation of approximately 1 g per deciliter. Mari and colleagues do not indicate why they rejected such a linear model. Moreover, they assert that they followed the methods of Royston,2 but he used log-polynomial models to develop reference ranges, not their proposed log-hyperbolic model.

Furthermore, Mari and colleagues do not provide us with an estimate of the degree of dispersion of the normal fetal hemoglobin values about their median curve, which would correspond to the standard deviation of 1 g per deciliter provided by Nicolaides et al.1 This also applies to the dispersion about the median curve for peak velocity. These omissions limit our ability to use their new models in clinical practice.

The investigators' results depend on their assertion that transfusion is not indicated in fetuses with mild anemia, defined by them as a “hemoglobin concentration from 0.84 to 0.65 times the median for gestational age.” This recommendation differs significantly from those of others, such as Nicolaides et al.,1 who assert that transfusion is required if the hemoglobin deficit is 2 g per deciliter or more below the mean. For example, Mari and colleagues would not recommend a transfusion for a fetus with a hemoglobin value of 8.6 g per deciliter at 24 weeks, representing a deficit of 3.4 g per deciliter. Only by redefining the standard definition of anemia are they able to claim such a high sensitivity for their test.

Mitchell P. Laks, M.D., Ph.D.
Tamara Cohen, B.A.
Woodhull Medical and Mental Health Center, Brooklyn, NY 11206

2 References

1
Nicolaides KH, Soothill PW, Clewell WH, Rodeck CH, Mibashan RS, Campbell S. Fetal haemoglobin measurement in the assessment of red cell isoimmunisation. Lancet 1988;1:1073-1075
CrossRef | Web of Science | Medline

2
Royston P. Constructing time-specific reference ranges. Stat Med 1991;10:675-690
CrossRef | Web of Science | Medline

Author/Editor Response

Dr. Mari and colleagues reply:

To the Editor: We agree with Dr. Nagey that it is premature to use measurements of peak systolic velocity in the middle cerebral artery to make clinical decisions. Therefore, we are prospectively testing the feasibility of serial measurements of peak systolic velocity in the middle cerebral artery as a means of timing a cordocentesis. We have studied more than 100 fetuses at risk for anemia. Cordocentesis was performed only when the trend in the values for peak systolic velocity deviated from the trend of the reference range obtained in fetuses without anemia and with values that were appropriate for gestational age.1 All but one of the fetuses with moderate or severe anemia were correctly identified by this method. We therefore believe that this method can be offered as an option in pregnancies complicated by red-cell alloimmunization.

In response to Dr. Laks and Ms. Cohen: the values for both the hemoglobin concentration and the peak systolic velocity of the middle cerebral artery are normally distributed. We determined the reference range for peak systolic velocity for fetuses without anemia in an earlier study.2 In the current study, we showed only the reference range for hemoglobin concentrations in 265 normal fetuses; the individual values are shown in Figure 1Figure 1Reference Range for Hemoglobin Concentrations in 265 Normal Fetuses.. The graph shows that the values change exponentially rather than linearly, as was previously suggested.3 We believe this curve better represents the changes in fetal hemoglobin concentrations throughout gestation, because it is based on hemoglobin concentrations in fetuses of all gestational ages. The reference range of Nicolaides et al. was based on hemoglobin concentrations measured in fetuses before 36 weeks' gestation (the individual data were not provided) and in 10 infants after birth.3

Nicolaides et al. suggested that a hemoglobin value that is 2 g per deciliter lower than the normal mean is an indication of the need for fetal transfusion at any gestational age. However, the use of this cutoff point would mandate transfusion in fetuses with only mild anemia. As an example, a fetus with a hemoglobin concentration of 11.5 g per deciliter at 35 weeks' gestation would receive a transfusion. On the basis of our data, mild anemia (≥0.65 times the median) would be defined as a hemoglobin deficit of 2.0 g per deciliter at 18 weeks' gestation and 2.6 g per deciliter at 36 weeks. We think that it is acceptable to delay transfusion in fetuses when the hemoglobin deficiency is so mild.

Giancarlo Mari, M.D.
University of Virginia, Charlottesville, VA 22908-0712

Roland Zimmerman, M.D.
Universitätsspital, Zurich, Switzerland

Utku Oz, M.D.
University of Mersin Medical School, Mersin, Turkey

3 References

1
Mari G, Collaborative Group for Doppler Assessment of the Blood Velocity in Anemic Fetuses. Non-invasive alternative to cordocentesis for detection of fetal anemia -- a prospective multinational trial. Am J Obstet Gynecol 2000;182:S21-S21 abstract.

2
Mari G, Adrignolo 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
CrossRef | Web of Science | Medline

3
Nicolaides KH, Soothill PW, Clewell WH, Rodeck CH, Mibashan RS, Campbell S. Fetal haemoglobin measurement in the assessment of red cell isoimmunisation. Lancet 1988;1:1073-1075
CrossRef | Web of Science | Medline

Citing Articles (1)