Background
Doppler ultrasound measurements of the peak systolic velocity of the middle cerebral artery can be used to noninvasively diagnose fetal anemia but are less precise following fetal blood transfusion and in late gestation. We have previously demonstrated the feasibility of estimating fetal hematocrit in vitro using magnetic resonance imaging relaxation times. Here we report the use of magnetic resonance imaging as a noninvasive tool to accurately detect fetal anemia in vivo.
Objectives
This study has 2 objectives: (1) to determine the feasibility and accuracy of magnetic resonance imaging in estimating hematocrit in anemic fetuses and (2) to compare magnetic resonance imaging and middle cerebral artery Doppler in detecting moderate to severe fetal anemia.
Study Design
Fetuses undergoing fetal blood sampling or transfusion underwent magnetic resonance imaging examinations prior to and following their procedures at 1.5 Tesla (Siemens Avanto). A modified Look-Locker inversion pulse sequence and T 2 preparation sequence were applied for T 1 and T 2 mapping of the intrahepatic umbilical vein. Estimated fetal hematocrit was calculated using a combination of T 1 and T 2 values and compared with conventional hematocrit obtained from fetal blood samples and middle cerebral artery Doppler measurements.
Results
Twenty-three fetuses were assessed during 33 magnetic resonance imaging scans. The mean absolute difference between the laboratory and magnetic resonance imaging–estimated hematocrit was 0.06 ± 0.05 with a correlation of 0.77 ( P < .001) determined by a multilevel, mixed-effects model adjusting for the repeated measurements from the same participants, multiple gestation pregnancies, and the scan type (ie, before or after transfusion scan). Bland-Altman analysis revealed a systematic bias of –0.03 between the magnetic resonance imaging and fetal blood sampling measurements. Magnetic resonance imaging and middle cerebral artery Doppler had similar sensitivities of approximately 90% to detect moderate to severe anemia. However, magnetic resonance imaging had a higher specificity (93% [13/14], 95% confidence interval, 66–100%) than Doppler (71% [10/14], 95% confidence interval, 42–92%).
Conclusion
Moderate to severe fetal anemia can be detected noninvasively by magnetic resonance imaging with high sensitivity and specificity. Our results suggest an adjunct role for magnetic resonance imaging in fetuses with suspected anemia, particularly following previous transfusion and in late gestation.
Fetal anemia is an uncommon pregnancy complication, with potentially serious adverse outcomes if not effectively diagnosed and treated in utero. When moderate to severe, it may result in widespread tissue hypoxia, leading to hydrops fetalis and ultimately fetal demise. Fortunately, ultrasound-guided fetal blood sampling and transfusions confer a survival rate of >90% in nonhydropic fetuses.
Why was the study conducted?
The specificity of middle cerebral artery peak systolic velocity by Doppler is limited following prior transfusions and in late gestation.
Key findings
The specificity and positive predictive value of magnetic resonance imaging for detecting fetal anemia are higher than middle cerebral artery peak systolic velocity by Doppler and approach the gold standard of fetal blood sampling.
What does this add to what is known?
This is the first study to show the utility of magnetic resonance imaging to noninvasively detect anemia by directly estimating fetal hematocrit by exploiting the magnetic properties of blood. The improved specificity over current methods could help to avoid the risk of an unnecessary fetal blood sample, which is associated with a small risk of fetal loss. We propose this method could be used in combination with middle cerebral artery Doppler as a way to confirm or exclude anemia when there is uncertainty.
Fetal blood sampling is the gold-standard diagnostic test for fetal anemia but is associated with risks of significant complications, including a perinatal loss rate of 1–2%. The measurement of peak systolic velocity in the middle cerebral artery (MCA), as obtained by Doppler, has proven to be a reliable screening test for fetal anemia. A subsequent trial comparing this Doppler method with the conventional bilirubin assay by amniocentesis led to the abandonment of this invasive screening method for fetal anemia.
The Society of Maternal-Fetal Medicine subsequently recommended the adoption of MCA Doppler ultrasound to identify fetuses with a sufficient risk of anemia to justify invasive fetal blood sampling. Although now an established noninvasive test, MCA Doppler is significantly less accurate following the first transfusion and later in the third trimester.
We investigated the potential of an alternative method for assessing fetal anemia using magnetic resonance imaging (MRI). The MRI relaxation properties of blood, specifically T 1 and T 2 relaxation times, depend primarily on hematocrit and oxygen saturation. We have previously shown that it is possible to estimate the hematocrit and oxygen saturation of blood in vitro using a combination of T 1 and T 2 values. Therefore, the aim of this study was to investigate the feasibility and accuracy of MRI for estimating hematocrit in vivo in suspected fetal anemia by comparing MRI with MCA Doppler and the conventional hematological analysis of fetal blood samples.
Materials and Methods
Study design and participants
The study was designed as a prospective, cross-sectional study comparing MRI measurements against fetal blood sampling and MCA Doppler measurements in fetuses with suspected anemia. It was approved by the Research Ethics Board at Mount Sinai Hospital and the Hospital for Sick Children in Toronto, Canada.
Pregnant women in their second or third trimester who were older than 18 years of age and were scheduled for an intrauterine fetal blood transfusion or fetal blood sampling for suspected fetal anemia or thrombocytopenia at the Fetal Medicine Unit at Mount Sinai Hospital were invited to participate in our study. Women with contraindications to MRI were excluded. Written consent was obtained from all participants. The main focus of our study was the anemic fetuses, although we included polycythemic fetuses in pregnancies complicated by twin anemia-polycythemia sequence and fetuses with thrombocytopenia whenever they underwent fetal blood sampling.
The primary objective of this study was to determine the accuracy of MRI to estimate hematocrit in the fetuses by comparing the MRI measurements against the laboratory hematocrit from the gold-standard fetal blood sample. Our second objective included testing the accuracy of MRI in detecting moderate to severe anemia compared with MCA Doppler, specifically the sensitivity, specificity, positive and negative predictive values, and the area under curve of the receiver-operating characteristic curves.
Definition of fetal anemia
We adopted previously published reference ranges for hematocrit and Doppler peak systolic velocity in the middle cerebral artery. The expected values of hemoglobin concentration and peak systolic velocity in the middle cerebral artery (MCA-PSV) were calculated as follows:
hemoglobinconcentration(gramsperdeciliter)=e(2.84−8.55/GA)
MCA-PSV(centimeterspersecond)=e(2.31+0.046×GA)
hematocrit=hemoglobinconcentration(gramsperdeciliter)×0.03
To adjust for the expected increases in hematocrit and MCA peak systolic velocity with advancing gestation, the reference ranges were expressed as multiples of the median corrected for gestational age :
multiplesofthemedian=measuredvalue/expectedvalue
MRI imaging protocol
Participants were scanned on a clinical 1.5 Tesla MRI system (Siemens Avanto, Erlangen, Germany). Ungated T 1 and T 2 mapping sequences were acquired in the short axis of the intrahepatic umbilical vein using a modified Look-Locker inversion pulse sequence and a T 2 preparation pulse sequence, respectively. The start time for the T 1 sequence was 100 milliseconds with a 400 millisecond inversion time increment.
The preparation times for the T 2 sequence were 32, 64, 128, 160, and 192 milliseconds, using a 16 millisecond refocusing interval. The recovery time between subsequent inversions or T 2 preparation was 16 beats. According to our previously reported protocol, images of the nonselective T 1 – or T 2 -prepared magnetization were acquired using a single-slice, steady-state free-precession acquisition (MRI sequence parameters are listed in Supplemental Table 1 ). These images were coregistered using nonrigid motion correction and then interpreted using laboratory-developed software written in Python to quantify T 1 and T 2 relaxation times ( Supplemental Figure 1 ).
The relaxation times were measured from the corresponding images with the regions of interest covering the central 60% of the umbilical vein. T 1 and T 2 values were then used to estimate hematocrit with an equation calibrated for in vitro human fetal blood. Individual region-of-interest definition was applied for the T 1 and T 2 images. The scan time for T 1 – and T 2 -measurements was approximately 30 and 24 seconds, respectively. All MRI examinations were accompanied by same-day MCA Doppler examinations.
MCA Doppler, fetal blood sampling, and intrauterine blood transfusion
The investigators making the MRI and Doppler measurements were blinded to each other’s measurements. Standard Doppler ultrasound was performed at Mount Sinai Hospital before and after the transfusions to assess the severity of fetal anemia and to monitor the fetuses after the transfusion. The peak systolic velocity was measured in the middle cerebral artery close to its origin from the internal carotid artery during fetal quiescence.
Under ultrasound guidance, a fetal blood sample was drawn from the intrahepatic umbilical vein immediately before and after the fetal transfusion for fetal hematocrit and hemoglobin concentration measurements. Adult packed donor blood was transfused intravascularly or intraperitoneally to achieve a final hematocrit of 40–55%. Fetal blood sampling and transfusions were done by physicians and laboratory staff and were blinded to all MRI measurements.
Statistical analysis
To evaluate the accuracy of the MRI, we first analyzed the correlation between the laboratory and MRI-estimated hematocrit before and after transfusion in all cases using a multilevel, mixed-effects linear model. We adjusted for the repeated measurements of the same patients over time, the infrequent occurrences of multiple gestation pregnancies, and the scan type (ie, before or after the transfusion) to take into account for the clustering within subjects and within pregnancies. Agreement between MRI and fetal blood sampling measurements was analyzed using a Bland-Altman plot.
The intra- and interobserver agreement for MRI measurements was assessed using Pearson linear regression and Bland-Altman analysis. Next, we generated receiver-operating characteristic curves for the pretransfusion MCA Doppler and MRI measurements to compare their performance in identifying moderate to severe fetal anemia.
The sensitivity, specificity, positive and negative predictive values, receiver-operating characteristic area, and positive and negative likelihood ratios of the pretransfusion MCA Doppler and MRI measurements were calculated using a standard binomial proportion test. The correlation between the pretransfusion MCA Doppler and MRI measurements was also determined using the same multilevel, mixed-effects model mentioned in the previous text. The values in the text are expressed as mean ± standard deviations and the statistical significance level was set at 0.05. The data were analyzed using Stata version 15.1 (StataCorp Inc, College Station, TX) and GraphPad Prism 7 (GraphPad Inc, San Diego, CA).
Results
Participant characteristics
From February 2016 through August 2018, 23 women consented to participate. Of these, 1 participant was excluded for not undertaking a fetal blood sample and 1 participant was lost due to MRI scheduling difficulties. A total of 25 fetuses (5 twin pregnancies and 1 triplet pregnancy) were scanned before and after their first to fourth blood transfusion or fetal blood sampling procedures for an average MRI scan time of 10 minutes.
Multiple MRI scans throughout gestation were performed for 7 fetuses: 4 were scanned twice and 3 were scanned 3 times. Four MRI images from 4 fetuses were excluded because of unacceptable image quality resulting from excessive fetal motion. Therefore, 33 scans from 23 fetuses were included in the final analyses ( Table 1 ; for further details see Supplemental Figure 2 and Supplemental Table 2 ). The average gestational age of the fetuses at the time of MRI was 28.4 ± 4.2 weeks (range, 19–36.6 weeks). Fetal diagnoses are included in Table 1 .
