Arch Gynecol Obstet (2008) 278:27–31

DOI 10.1007/s00404-007-0522-2

ORIGINAL ARTICLE

Prediction of fetal anemia by Doppler of the middle cerebral artery and descending thoracic aorta

David Pares · Paulo A. Chinen · Luiz Camano · Antonio F. Moron · Maria R. Torloni

Received: 19 August 2007 / Accepted: 15 November 2007 / Published online: 8 December 2007© Springer-Verlag 2007

SummaryBackground Despite signiWcant advances, perinatalhemolytic disease has not been eradicated and is still asso-ciated with signiWcant morbidity and mortality, especiallyin developing countries. An accurate method for antenataldiagnosis and quantiWcation of fetal anemia is a crucial stepprior to the performance of invasive procedures, which arenot risk-free. The middle cerebral artery peak systolicvelocity (MCA-PSV) is currently the most popular non-invasive diagnostic method used to quantify fetal anemiaand determine the exact timing for transfusion The com-bined use of MCA-PSV and descending thoracic aortamean velocity (DTA-MV) has not been previously evalu-ated. Since this strategy has the potential to decrease falsepositive results it could reduce unnecessary cordocenteses,therefore minimizing fetal risks. This study aimed to evalu-ate the accuracy of MCA-PSV associated with DTA-MV inthe prediction of fetal anemia.Method Retrospective analysis of 90 cordocenteses per-formed on alloimmunized pregnancies in a single hospitalMCA-PSV and DTA-MV were performed immediatelybefore cordocentesis. The Dopplervelocimetry Wndingswere compared with fetal hemoglobin concentration. Sensi-tivity, speciWcity, positive and negative predictive valuesand accuracy of MCA-PSV and DTA-MV in the predictionof moderate/severe fetal anemia were calculated.

Results The individual sensitivity of MCA-PSV and DTA-MV was 100 and 95.7%, speciWcity was 65 and 100% andaccuracy was 92.2 and 96.7%, respectively. The combinationof both vessels produced a higher overall accuracy (98.6%).Conclusion MCA-PSV is more sensitive and DTA-MV ismore speciWc in the detection of fetal anemia. The highestdiagnostic accuracy was obtained by the association of bothDoppler evaluations.

Keywords Doppler · Fetal anemia · Middle cerebral artery · Aorta · Ultrasound

Introduction

For centuries, perinatal hemolytic disease (PHD), an immu-nological disorder due to feto-maternal blood incompati-bility, was an important cause of perinatal mortality.Notwithstanding the signiWcant advances in fetal medicineand neonatology in the last Wve decades, this disease stillposes signiWcant perinatal risks and perinatal morbidity andmortality remain high [10]. Although the incidence of PHDhas greatly declined worldwide due to the implementationof preventive measures, it is far from being eradicated,especially in Brazil, in part due to the lack of medical infor-mation and also due to the high cost of anti-D prophylaxis.And the antenatal management of this condition requireshighly trained specialists and considerable resources.

Fetal intravascular transfusion, a life-saving treatment incases of severe PHD, is not devoid of risks. Due to the pas-sage of fetal blood into the maternal circulation during theprocedure, it may in fact cause exacerbation of the maternalimmune response in up to 66% of the cases [13]. Additionalcomplications, such as fetal bradycardia, premature ruptureof membranes, bleeding from the umbilical cord and fetal

D. Pares · P. A. Chinen · L. Camano · A. F. Moron · M. R. TorloniFetal Medicine Unit, Obstetrics Department, São Paulo Federal University Medical School, São Paulo, Brazil

M. R. Torloni (&)Rua Ilimani 125 Cidade Jardim, São Paulo, SP 05602-030, Brazile-mail: ginecologia@terra.com.br

123

28 Arch Gynecol Obstet (2008) 278:27–31

death, are reported in up to 2% of intravascular transfusion[8]. And despite intrauterine treatment, approximately 15%of the non-hydropic infants and up to 50% of the moreseverely aVected fetuses do not survive [10].

Therefore, an accurate method for the antenatal predic-tion of fetal anemia and its severity is a crucial step prior tothe performance of invasive procedures, such as fetal bloodsampling, either for diagnostic or therapeutic purposes. Inthe last twenty years, several non-invasive diagnostic strat-egies have been proposed to determine the severity of fetalanemia and the exact timing for fetal transfusion. Advancesin the Weld of ultrasound equipments have greatly contrib-uted to optimize the management of these cases.

Starting in the 1980s, Doppler ultrasound contributed toa better understanding of fetal hemodynamics and the Wrstpublications on the use of this method in the detection offetal anemia showed encouraging results. In 1986, Right-mire et al. [12] were the Wrst to suggest a possible relationbetween blood Xow velocity and the fetal hematocrit inRhesus isoimmunization. These authors observed aninverse relation between fetal hematocrit and the meanblood Xow velocity in the descending thoracic aorta (DTA-MV) of anemic fetuses, regardless of gestational age. How-ever, due to technical diYculties in the insonation of thisvessel investigators started to evaluate other fetal vessels,such as the middle cerebral artery. In 1995 Mari et al. [7]reported that the middle cerebral artery peak systolic veloc-ity (MCA-PSV) was 100% sensitive and 50% speciWc inthe prediction of fetal anemia. A recent collaborative multi-centric study that evaluated the performance of MCA-PSVin alloimmunized pregnancies, reported a false-positive ratewas 12%. Other publications since then [3, 4, 10, 11] havereinforced the conclusion that the MCA-PSV is a valid anduseful tool in the diagnosis and management of fetal ane-mia in alloimmunized pregnancies, with a high sensitivity(83–100%) and a somewhat lower speciWcity (81–93%).

The combined use of Doppler velocimetry of the middlecerebral artery along with the descending thoracic aorta forthe prediction of fetal anemia has not been previously evalu-ated. Since this strategy has the potential of decreasing thefalse-positive rate, it could result in a reduction in the need fordiagnostic cordocenteses and therefore minimize fetal risks.

The objectives of this study were to evaluate and com-pare the accuracy of the middle cerebral artery peak sys-tolic velocity (MCA-PSV) and the mean velocity of thedescending thoracic aorta (DTA-MV) in the prediction offetal anemia, used individually and in association.

Methods

This was a retrospective study of all alloimmunized preg-nancies (20–34 weeks of gestational age) managed at the

Fetal Medicine Unit of the Obstetrics Department of SãoPaulo Federal University Medical School (UNIFESP-EPM)between June 1997 and January 2005. All patients gavewritten informed consent prior to entering any managementprotocol and the study was approved by the institution’sreview board.

All singleton pregnancies with reliable gestational age(sonographic examination <20 weeks) and a positive indi-rect Coombs test (¸16) who were submitted to cordocente-sis due to suspected fetal anemia were considered eligible.

According to the management protocol in use until 1999,all pregnant women with hemolytic antibodies (either anti-D or atypical antibodies) and a positive Coombs test wereevaluated every 4 weeks through 2D fetal ultrasound,Doppler studies (MCA-PSV and DTA-MV) and amniocen-tesis but the decision to proceed with cordocentesis wasbased exclusively on the results of amniotic Xuid spectro-photometry. The Wrst amniocentesis was scheduled to beperformed around 26–28 weeks, or earlier, according toobstetric history, and repeated subsequently usually within2 weeks, according to the results of the amniotic Xuid spec-trophotometry, using Lilley’s graph [5]. Fetuses <34 weeksgestation with high bilirubin concentrations in their amni-otic Xuid (zones II-B and III on Lilley´s graph) were sub-mitted to cordocentesis to conWrm moderate to severeanemia and receive intrauterine transfusion. Starting in2000, amniocenteses were abandoned and the decision toproceed with cordocentesis was based exclusively on theresults of cerebral Doppler evaluation. Fetuses with MCA-PSV >1.5 MoM [12] were considered as potentially anemicand scheduled for cordocentesis to conWrm the diagnosisand transfusion.

Throughout the study period, immediately before cordo-centesis, Doppler evaluation of the descending thoracicaorta and middle cerebral artery were repeated in all casesand these measurements were compared with the umbilicalhemoglobin levels, which were considered the gold stan-dard. Fetuses with hemoglobin level < 0.84 multiples ofmedian for gestational age received the diagnosis of moder-ate to severe anemia [7].

All Doppler exams, cordocenteses and fetal transfusionswere performed by a single examiner (D. P.). Pregnantwomen were positioned in dorsal decubitus and Dopplermeasurements were performed 15 to 30 minutes before cor-docentesis using pulsed duplex Doppler with a 3.5 MHzconvex transducer using a Toshiba 6000 Power Visionequipment (Toshiba Co., Tokyo, Japan). The Wlter was keptat 50 Hz and the insonation angle between the Dopplerbeam and the aorta axial axis at 55°. Each examinationlasted 15–20 min and all measurements were performedin the absence of fetal activity or breathing. Dopplerevaluation of the descending thoracic aorta was performedin a sagittal scan, placing the sample window in the aorta

123

Arch Gynecol Obstet (2008) 278:27–31 29

immediately above the diaphragm, which served as a repro-ducible reference parameter. The mean blood Xow velocityof two consecutive cardiac cycles was automatically calcu-lated by the equipment software and reported in centimetersper second (cm/sec). The results of the DTA-MV wereclassiWed as normal or altered according to Nicolaideset al. [9]. Fetuses with velocities ¸2SD for gestational agewere suspected of having moderate to severe anemia. Thetechnique for obtaining the MCA-PSV followed the recom-mendations of Mari et al. [7]. The angle of insonation waskept at zero. Fetuses with measurements ¸1.5 MoM werediagnosed as possibly anemic.

All cordocenteses were performed by the Wrst authorunder ultrasound guidance, preferably at the placentalcord insertion and fetal blood samples were immediatelysent to the laboratory for determination of hemoglobinlevel, using the Abbot CELL-DYN 3500 (Abbot lab, Ari-zona, USA).

Fetuses were diagnosed as normal or anemic based ontheir hemoglobin levels, according to the classiWcationsystem proposed by Mari et al. [7]. BrieXy, anemia wasconsidered moderate to severe when fetal hemoglobinconcentrations were < or =0.64 multiples of the medianfor gestational age. Fetuses with moderate to severe ane-mia were immediately submitted to intrauterine bloodtransfusion.

Statistical analysis

Initially, variables were analyzed descriptively. Sensitivity,speciWcity, positive and negative predictive values, andaccuracy were calculated for the three tests (DTA-MV,MCA-PSV and both velocities) in the prediction of moder-ate to severe fetal anemia. The gold standard was the levelof fetal hemoglobin obtained at cordocentesis, immediatelyafter the Doppler studies.

Sensitivity indicated the proportion of anemic fetusesshowing increased blood velocity (aortic Xow velocitiesabove 2SD or MCA above 1.5 MoM) measured throughDoppler ultrasound. SpeciWcity was deWned as the propor-tion of non-anemic fetuses with normal Xow velocities.Positive predictive value was deWned as the proportion offetal anemia in cases with increased Xow velocity. Nega-tive predictive value was deWned as the proportion offetuses without anemia in cases with normal Xow veloci-ties. Accuracy was deWned as the sum of true positives(increased velocity and anemia) and true negatives (nor-mal velocity and absence of anemia) over all fetusesexamined.

Kappa’s concordance index was used to compare bothmethods in the prediction of fetal anemia. According to thisindex, values <0.45 indicate marginal reproducibility,

values from 0.45 to 0.75 indicate good reproducibility andvalues >0.75, excellent reproducibility.

A value of P < 0.05 was considered signiWcant.

Results

During the study period, a total of 46 women fulWlled theinclusion criteria. Mean maternal age was 24.9 (range 19–37 years). Median gravidity was 6 (range 1–11). Therewere three alloimmunized primiparous women. One ofthese was a former intravenous drug user and two had his-tories of blood transfusions. One of the forty-six womenhad anti-Kell antibodies and all others had exclusively anti-D antibodies. The indirect Coombs test results ranged from1/32 to 1/8,182. A total of 90 cordocenteses were per-formed on 46 fetuses at-risk for anemia and the individualhemoglobin value of each cordocentesis was correlatedwith the DTA-MV and PSV-MCA. Mean gestational age atWrst cordocentesis was 26.8 weeks (range 20–34 weeks).Table 1 presents more details of the 90 cordocenteses per-formed on the 46 pregnancies.

The sensitivity, speciWcity, positive and negative predic-tive values and accuracy of MCA-PSV, DTA-MV and theassociation of both tests in the prediction of moderate/severe fetal anemia is presented on Table 2. The associationof the Doppler of both arteries was more accurate than indi-vidual Doppler evaluations in the prediction of fetal ane-mia. It also resulted in a higher speciWcity and positive

Table 1 Timing, number of procedures and results of 90 cordocente-ses performed on 46 alloimmunized pregnancies due to suspected fetalanemia on Doppler evaluation

GA gestational age

Characteristics n (%)

Timing (GA in weeks) of 90 cordocenteses

20–24 14 (15.5)

25–28 33 (36.7)

29–32 41 (45.6)

33–34 2 (2.2)

Number of cordocenteses per patient

1 24 (52.2)

2 13 (28.3)

3 3 (6.5)

4 3 (6.5)

5 3 (6.5)

Fetal anemia on cord blood

Absent 7 (7.8)

Mild 13 (14.4)

Moderate 15 (16.7)

Severe 55 (61.1)

123

30 Arch Gynecol Obstet (2008) 278:27–31

predictive value than the isolated use of MCA-PSV.Table 3 presents the concordance index between both arter-ies in the prediction of fetal anemia.

Discussion

Perinatal hemolytic disease still causes signiWcant morbid-ity and mortality. In developed countries, due to the univer-sal screening for red cell antibodies in pregnancy and theadministration of anti-D immunoglobulin when necessary,there has been a steady decline of the most severe forms ofthe disease, however the incidence of the mild and moder-ate forms is increasing [14].

In developing countries, such as Brazil, all forms of thedisease are still frequent, including cases of fetal hydropsdue to severe anemia. This can be attributed in part to thelack of basic equipments and material needed for the ade-quate management of sensitized pregnant women (ultra-sound equipment and laboratory resources) but also to acertain lack of professional care and knowledge about the

correct use of eVective preventive measures. Indeed, almostall the patients in this study (97.7%) were sensitized to theD antigen, the only type of preventable alloimmunization,and 96.5% of them had been sensitized due to a previouspregnancy, reXecting the lack of adequate preventive mea-sures in this population.

The population of the present study diVers from otherpublications in various aspects. Despite their relativelyyoung age, 60% of these women reported four–six previouspregnancies, indicating they had become sensitized due tothe lack of adequate preventive therapy in previous gesta-tions. Only three of the patients were primiparous and hadbecome sensitized due to non-obstetric events. Further-more, over 50% of the cordocenteses were performed earlyduring pregnancy (<28 weeks) indicating the severity offetal anemia in this group of patients. These Wndings shouldalert Brazilian obstetricians to the need for early detectionof fetal anemia in their patients, in order to oVer adequatetreatment before severe anemia leads to fetal hydrops andhigh mortality rates.

In the United States, the proportion of fetuses at risk foranemia because of maternal alloimmunization to red-cellantigens has been estimated to be 35 per 10,000 live births[7]. Approximately 90% of these will be unaVected orslightly anemic and 10% will need interventions to treatanemia before 34 weeks. In Brazil, there are no nationalstatistics on the subject, but the Wndings of the presentstudy tend to indicate the percentage of severely anemicfetuses must be higher.

The correct diagnosis of moderate to severe fetal anemiabefore 34 weeks is important in order to oVer adequatetreatment (intra-uterine transfusion) to these fetuses beforehydrops sets in. Non-invasive methods of detecting fetalanemia rely on the increased velocity of blood Xow in ane-mia, due to decreased viscosity. The ideal diagnostic testshould be highly accurate, allying at the same time a highsensitivity and a high speciWcity. Therefore, this ideal testwould be positive (abnormal) in almost all the fetuses withmoderate to severe anemia, which would consequently besubmitted to life saving intrauterine transfusion, and thetest would be negative (normal) in almost all fetuses with-out signiWcant anemia, thus sparing them from the risks ofunnecessary invasive procedures.

Bilardo et al. [1], the Wrst to investigate the mean veloc-ity of the descending thoracic aorta in alloimmunized preg-nancies, reported the method was 88% accurate in theprediction of fetal anemia. Steiner et al. [15] reported thepositive and negative predictive values of the DTA-MV tobe 73 and 66%, respectively. In 1995, Mari et al. [6] stud-ied the performance of the MCA-PSV in the prediction offetal anemia in 56 cordocenteses and reported a sensitivityof 100% and a speciWcity of 50%. In a subsequent collabo-rative involving nine centers, the same authors reported the

Table 2 Performance of three non-invasive diagnosis tests in theprediction of fetal anemia

Cut-oVs: DTA-MV ¸2 SD, MCA-PSV ¸1.50 MoM

DTA-MV mean velocity of the descending thoracic aorta, MCA-PSVmiddle cerebral artery peak systolic velocity, Pos PV positive predic-tive value, Neg PV negative predictive value, SD standard deviation,MoM multiple of median

DTA-MV MCA-PSV DTA-MV and MCA-PSV

n (%) n (%) n (%)

Sensitivity 67/70 (95.7) 70/70 (100) 61/62 (98.4)

SpeciWcity 20/20 (100) 13/20 (65.0) 11/11 (100)

Pos PV 67/67 (100) 70/77 (90.9) 61/61 (100)

Neg PV 20/23 (86.9) 13/13 (100) 11/12 (91.7)

Accuracy 87/90 (96.70) 83/90 (92.2) 72/73 (98.6)

Table 3 Concordance between the results of the aortic and cerebralDoppler in the prediction of fetal anemia in 90 cases of alloimmuniza-tion submitted to cordocentesis

DTA-MV mean velocity of the descending thoracic aorta, MCA-PSVmiddle cerebral artery peak systolic velocity, SD Standard deviation,MoM Multiple of median

DTA-MV MCA-PSV

(1.50 MoM <1.50 MoM Total

¸2 SD 61 9 70

<2 SD 8 12 20

Total 69 21 90

Kappa = 0.463; P < 0.001

123

Arch Gynecol Obstet (2008) 278:27–31 31

same sensitivity but fail to mention the speciWcity. A recentmulticentric trial [17] reported a sensitivity and a negativepredictive value of 98% and a positive predictive value of53% for the MCA-PSV. Teixeira et al. [16] tested Mari’scurve in 26 fetuses, using 2 or 4 multiples of the median ascut-oVs and reported a sensitivity of 64 and 100% and aspeciWcity of 93 and 83%, respectively. Similarly, DelleChiaie et al. [2] in a study involving a small number ofcases reported a sensibility of 73% and a speciWcity of81%, Dukler et al. [4] obtained 100% sensitivity and speci-Wcity and Pereira et al. [11] reported a sensitivity of 100%and a positive predictive value of 71%.

The Kappa index (0.463) indicates adequate reproduc-ibility of both methods (Table 3), thus suggesting that fetalanemia may be detected by any of the two Doppler evalua-tions. However, the best results were obtained with theassociation of both vessels (Table 2). with a sensitivity of98.1%, speciWcity of 100%, and accuracy of 98.6%. This isprobably the most important Wnding of this study.

The present Wndings, as other studies before, conWrm theaccuracy of Mari’s [7] cutoV point (1.5 MoM) with a 100%sensitivity in the detection of moderate to severe anemia.The accuracy of Nicolaides curve [9] was also conWrmed inthis study, with an adequate sensitivity and a high speciWc-ity. As previously demonstrated, the MCA-PSV is veryaccurate in the detection of fetal anemia but somewhat lessprecise in the detection of uncompromised fetuses and inthe follow-up of those already submitted to intrauterinetransfusions. The present study emphasizes the importanceof the evaluation of both vessels in optimizing the manage-ment of alloimmunized pregnancies.

This study has several strong points. First, all the 46cases were managed at the same Fetal Medicine Unit. Also,all the 90 cordocenteses and respective Doppler evaluationswere performed by a single investigator. Finally, it is theWrst study to evaluate the association of Doppler velocime-try of these two diVerent vessels in the prediction of fetalanemia in Rh-alloimmunized pregnancies.

The results of this study suggest that the association ofMCA-PSV and DTA-MV increases the accuracy andoVers a higher speciWcity in the prediction of moderate tosevere fetal anemia in alloimmunized pregnancies. TheseWndings need to be corroborated in a larger populationand multiple centers before they can be incorporated inthe management protocol for all alloimmunized pregnan-cies. The higher speciWcity of this protocol could poten-tially lead to a reduction in the number of unnecessaryinvasive procedures and also possibly reduce the numberof pregnant women referred to tertiary centers, which con-sequently would only receive cases that would probablyneed intrauterine transfusions.

References

1. Bilardo CM, Nicolaides KH, Campbell S (1989) Doppler studiesin red cell isoimmunization. Clin Obstet Gynecol 32(4):719–27

2. Delle Chiaie LD, Buck G, Grab D, Teride R (2001) Prediction offetal anemia with Doppler measurement of the middle cerebralartery peak systolic velocity in pregnancies complicated by maternalblood group alloimmunization or parvovirus B19 infection.Ultrasound Obstet Gynecol 18(3):232–236

3. Deren Ö, Önderoglu L (2002) The value of middle cerebral arterysystolic velocity for initial and subsequent management in fetalanemia. Eur J Obstet Gynecol Reprod Biol 101(1):26–30

4. Dukler D, Oepkes D, Seaward G, Windrim R, Ryan G (2003) Non-invasive tests to predict fetal anemia: a study comparing Dopplerand ultrasound parameters. Am J Obstet Gynecol 188(5):1310–1314

5. Lilley AW (1961) Liquor amnii analysis in the management of thepregnancy complicated by rhesus sensitization. Am J ObstetGynecol 82(3):1359–1370

6. Mari G, Adrignolo A, Abuhamad AZ, Pirhonen J, Jones DC,Ludomirsky A, Copel JA (1995) Diagnosis of fetal anemia withDoppler ultrasound in the pregnancy complicated by maternalblood group immunization. Ultrasound Obstet Gynecol5(6):400–405

7. Mari G, Deter RL, Carpenter RL, Rahman F, Zimmerman R,Moise KJ Jr, Dorman KF, Ludomirsky A, Gonzalez R, Gomez R,Oz U, Detti L, Copel JA, Bahado-Singh R,Berry S, Martinez-Poyer J, Blackwell SC (2000) Noninvasive diagnosis by Dopplerultrasonography of fetal anemia due to maternal red-cell alloim-munization. Collaborative group for Doppler assessment of theblood velocity in anemic fetuses. N Engl J Med 342(1):9–14

8. Newnham JP, Phillips JM, Stock R (1992) Intrauterine intravascu-lar transfusion for fetal haemolytic anemia: the Western Australianexperience. Med J Aust 157(10):660–665

9. Nicolaides KH, Bilardo CM, Campbell S (1990) Prediction of fetalanemia by measurement of the mean blood velocity in the fetalaorta. Am J Obstet Gynecol 162(1):209–212

10. Oepkes D (2000) Invasive versus non-invasive testing in red-cell al-loimmunized pregnancies. Eur J Gynecol Reprod Biol 92(1):83–89

11. Pereira L, Jenkins TM, Berghella V (2003) Conventional manage-ment of maternal red cell alloimmunization compared with man-agement by Doppler assessment of middle cerebral artery peaksystolic velocity. Am J Obstet Gynecol 189(4):1002–1006

12. Rightmire DA, Nicolaides KH, Rodeck CH, Campbell S (1986) Fe-tal blood velocities in Rh isoimmunization: relationship to gesta-tional age and to fetal hematocrit. Obstet Gynecol 68(2):233–236

13. Schumacher B, Moise KJ Jr (1996) Fetal transfusion for red bloodcell alloimmunization in pregnancy. Obstet Gynecol 88(1):137–150

14. Spong CY, Porter AE, Queenan JT (2001) Management of isoim-munization in the presence of multiple maternal antibodies. Am JObstet Gynecol 185(2):481–484

15. Steiner H, SchaVer H, Spitzer D, Batka M, Graf AH, Staudach A(1995) The relationship between peak velocity in the fetaldescending aorta and hematocrit in rhesus isoimmunization.Obstet Gynecol 85(5 pt 1):659–662

16. Teixeira JM, Duncan K, Letsky E, Fisk NM (2000) Middle cere-bral artery peak systolic velocity in the prediction of fetal anemia.Ultrasound Obstet Gynecol 15(3):205–208

17. Zimmermann R, Carpenter RL, Carpenter RJ Jr, Durig P, Mari G(2002) Longitudinal measurement of peak systolic velocity in thefetal middle cerebral artery for monitoring pregnancies compli-cated by red cell alloimmunization: a prospective multicentre trialwith intention-to-treat. BJOG 109(7):746–752

123