Qualitative analysis of Doppler assessment used for surveillance of alloimmunized individuals

To summarize image quality variables for alloimmunized women at risk for fetal anemia. To investigate the association between image quality with the highest and median middle cerebral artery peak systolic velocity (MCA‐PSV) at the last visit and fetal anemia based on hemoglobin.


| INTRODUCTION
Alloimmunization occurs when a pregnant individual is exposed to red blood cell (RBC) antigens present in the fetus and not in the mother.
Sensitization usually occurs after a red blood cell transfusion, prior pregnancy or other obstetrical event such as a spontaneous abortion or placental abruption. 1,2 Less frequent causes for sensitization and subsequent alloimmune disease include prior invasive procedures in pregnancy or inadequately treated women after delivery, when fetalmaternal hemorrhage exceeds the volume that can be neutralized with standard immunoglobulin anti-D antibody treatment. This leads to a maternal immune response with antibody formation. In a subsequent pregnancy, depending on the intensity of the immune response caused by the specific antigen, the IgG antibodies can cross the placenta and bind to the fetal red blood cells causing hemolysis with effects that range from mild to severe anemia, to neonatal jaundice, and death. 3 Management of an alloimmunized individual with a fetus known to be a carrier of an RBC antigen not present in the mother involves screening for fetal anemia using spectral Doppler ultrasound to assess blood flow through the middle cerebral artery (MCA) in the fetus. 4,5 Fetal hemoglobin concentration negatively correlates with the peak systolic velocity (PSV) measurement in the MCA. 5 Middle cerebral artery peak systolic velocity (MCA-PSV) measurements can be used to screen for moderate-to-severe anemia when the velocities exceed the 1.5 multiples of the median (MoM) threshold for gestational age. [4][5][6] This value has been shown to have a sensitivity of 100% to predict moderate anemia (hemoglobin concentration 0.65 times the median) or severe anemia (hemoglobin concentration 0.55 times the median), with a 12% false positive rate. 7 Accurate MCA-PSV measurements are crucial as it determines future management, and high values can lead to invasive procedures such as cordocentesis with blood sampling from the fetal umbilical vein or induction of labor. 4 Guidelines have been published to guide the process of measuring MCA-PSV 8 and studies have looked at the reproducibility of the measurement technique, 4 however we have not identified any reports qualifying the images that are being used in clinical management nor their correlation with fetal outcomes. Anecdotally we have observed that there is significant variation in the images being used in our clinical management; we want to determine the relationship between the quality of the image and the outcome.

| Ethics approval
The study was received expedited institutional review board approval because it is research for which patient consent was not required.
Data was collected and stored in such a way that the subject identity could not be ascertained. Women who had declined to allow their data to be used for research were identified in the electronic medical record (EMR) and were not included. The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. The study was performed in accordance with the Declaration of Helsinki. Regarding the quality of pulsed waved Doppler ultrasonography in general, three criteria were identified from ISUOG 9 which included ensuring absence of fetal breathing and body movements (fetal quiescence), the wall motion filter (WMF) being set as low as possible (≤ 50-60 Hz), and an angle of insonation of <30 . An insonation angle deviating from zero leads to velocity error, however angle correction can be used if the angle is >20 , but this can also lead to error.

| Qualitative image criteria
Because there are no established guidelines for what angle is too great, we decided that without angle correction, any angle of insonation <30 from the vector of flow of the MCA would be considered accurate. 4 The ISUOG recommendations include other criteria to maximize accuracy; however, these criteria cannot be assessed from examining a still image and pertain to proper technique by the sonographer. 9 Other criteria, such as quantifying the amount of gain in the image, were also excluded due to the binary nature of our grading system making it difficult to assign a discrete value.
The three criteria from the ISUOG guidelines that were selected to specifically measure the MCA-PSV are: (1) a magnified axial section of the brain showing the sphenoid wings and thalami; (2) the pulsed-wave Doppler gate placed within the proximal third of the MCA near its origin from the internal carotid artery (ICA) due to the systolic velocity increasing the further from the origin and along its branches; and (3) a recording of greater than or equal to 3 and fewer than 10 waveforms. A criterion that was not included was using color flow mapping to identify the circle of Willis because although useful in identifying the MCA, ISUOG states it is not mandatory. The SMFM guidelines includes similar criteria as those listed in ISUOG guidelines, however the SMFM also states that the entire length of the MCA should be visualized to ensure that the measurements are obtained from the proximal third of the vessel, and was added to the quality assessment. In total, the guidelines yielded seven criteria by which to analyze the Doppler images.

| Population
Inclusion criteria for this study included singleton pregnancies of individuals who delivered within our health system, whose pregnancies were between 18 and 35 weeks of gestational age with a confirmed or suspected diagnosis of maternal isoimmunization and risk for fetal alloimmune disease. We did not include twin pregnancies due to alternative causes for anemia among monochorionic twins. Exclusion criteria were fetuses with known congenital abnormalities, infections or fetuses with elevated MCA-PSV without a diagnosis of fetal red blood cell alloimmune disease. Although parvovirus infection can lead to fetal anemia through suppression of erythropoiesis, infections may also lead to anemia by inducing hemodynamic changes such as compromised myocardial function that could alter MCA-PSV values. 11

| Data collection
Data from all women seen in our clinic was obtained from January 2018 to April 2021 from the Viewpoint 6 (General Electric Company, Milwaukee, WI ® ) ultrasound reporting system. We collected patient identifiers (used to link to EMR and abstract fetal and neonatal outcomes when available), gestational age, date of examination, mea- Patients prior to 2018 were not included due to inability to identify the indication given for the visit in the ultrasound database. This yielded 400 ultrasound encounters. Our study objective was to evaluate the quality of the images and to correlate the MCA-PSV used at the last visit leading up periumbilical blood sampling (PUBS), or delivery, with the fetal hemoglobin value obtained during PUBS or neonatal hemoglobin value obtained after birth. We searched the EMR to determine management and outcomes after the last ultrasound MCA-PSV assessment. We identified the initial hemoglobin values obtained at PUBS or first neonatal hemoglobin values after delivery prior to any transfusion. Regrettably, in some cases, only a direct and indirect neonatal bilirubin level was identified and we assumed that normal bilirubin levels indicated a lack of hemolysis with no anemia present.
These cases corresponded to women in whom the fetal antigen status had not been determined prenatally due to patient declining amniocentesis and choosing to proceed with MCA-PSV screening. There were no cases of elevated bilirubin levels. All selection criteria were met in 26 women. We assumed that each woman would have an average of three MCA-PSV measurements taken at their last visit. In order to increase the total number of images and have a more robust image quality analysis, we decided to analyze the previous two visits for each patient. This led to 192 images available for analysis. Only the last visit was used to correlate with fetal or neonatal assessment for anemia.

| Image analysis
Images were analyzed within the Viewpoint ultrasound reporting system. All images were analyzed by one author (C.H.) who was trained by a maternal-fetal medicine physician to identify all the markers of adequate imaging for MCA-PSV assessment. The quality of the work was then verified by a maternal-fetal medicine physician (S.C.). The images and gestational ages pertaining to the last exam before delivery or cordocentesis were obtained, as well as the images from the two visits prior to delivery/cordocentesis. Only the images obtained the visit prior to delivery/cordocentesis were used in calculating accuracy, sensitivity, and specificity; the images from two prior visits were included to characterize the quality of Doppler images. Images were graded based on the seven listed criteria in a binary fashion. Images were weighted equally and assigned a 1 for each criterion they met and 0 for those they failed to meet, giving a total score that ranged from 0 to 7. All images from the exam date were included in the analy- as this value has been demonstrated in the literature to have a high sensitivity and specificity. [4][5][6] To determine fetal quiescence we superimposed a horizontal line at the maximum point above the measured waveforms corresponding to the highest MCA-PSV, and a line through the troughs of the waveforms. If the other waveforms did not meet these horizontal lines or if they overextended them, we determined that the fetus was not in a quiet state due to the variability in the measured PSV that did not allow adequate assessment. All of the images in our database used color flow mapping which was useful in determining if the entire MCA was visualized. To determine if the Doppler spectral gate was within the proximal third of the MCA we utilized two methods. First, we used visual inspection if the gate was clearly within or outside of the proximal third. In images where it was difficult to assess, a ruler was placed against the ultrasound image and we obtained measurements to make the final determination. In the images where the quality was so poor that the MCA could not be visualized, we automatically assigned no points to these assessments.
Details about the WMF settings, angle of insonation, number of waveforms, and presence of the standard anatomical markers including the thalami and sphenoid wings were readily apparent on the digitally saved images.
In order to evaluate the amount of variability between MCA-PSV measurements in a single session we calculated a median value for the MoM at each ultrasound session. We then compared each MoM value to the median value for the session to calculate the interquartile range (IQR) and range inherent in the measurements. We felt that this represents the variability often observed in a busy clinical practice.

| Statistical methods
Image quality variables were summarized for all images from patients' last two visits. We investigated the performance of the highest MCA-PSV and median MCA-PSV of only the last visit to predict fetal anemia (hemoglobin less than 0.55 MoM for gestational age or on average less than 5 gm per deciliter), 7,8,12 using logistic regression. AUC and 95% CIs were obtained. Specificity and sensitivity were also summarized. All reported p values are two-sided and a significance level of 0.05 was used. Statistical analyses were performed using R (version 3.6.2, R Core Team) and SAS (version 9.4, SAS Institute Inc., Cary, North Carolina).

| RESULTS
The mean age and gestational age of individuals in our study were

| Principle findings
Criteria exist that guide how MCA-PSV should be measured, however no studies have been done that qualify the images used in measuring the MCA-PSV and how many of the criteria are being met. Our study shows that only a minority (12.0%) of ultrasound images met the seven criteria, which ideally should always be met.
Our study also showed that using the highest MCA-PSV had a greater sensitivity and better accuracy than the median velocity, however a greater sample size, as well as consistency in measurement technique is needed to elucidate the true nature of this relationship. We also noted substantial variability between measurements in the same ultrasound session.

| Findings
Although we have identified studies that exam variability in technique in measuring the MCA-PSV, 1 no studies to date have analyzed the quality of the actual Doppler images used in management. Our results confirm our clinical observations that there is substantial variability in the quality of images being used and that many are not meeting published standards. [5][6][7] Our study showed a sensitivity to detect moderate to severe fetal anemia that is less than the 100% reported in literature, 4 however due to the small sample size it is unclear how significant this is and if it is due to the poor image quality.

| Research implications
Future research performed prospectively, before and after adequately educating our technicians, will serve to better elucidate the relationship between image quality, false positive and false negative

| Strengths and limitations
The use of neonatal bilirubin as a proxy for neonatal anemia is not ideal, however we think that it is acceptable in this case due to

CONFLICT OF INTEREST
The authors report no conflict of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

ETHICS STATEMENT
The study was received expedited institutional review board approval because it is research for which patient consent was not required.
Data was collected and stored in such a way that the subject identity could not be ascertained. Women who had declined to allow their data to be used for research were identified in the electronic medical record (EMR) and were not included.