Cytomegalovirus-related fetal brain lesions: comparison between targeted ultrasound examination and magnetic resonance imaging




To evaluate the relative contributions to the diagnosis of fetal brain abnormalities of targeted ultrasound examination and magnetic resonance imaging (MRI) in fetuses infected with cytomegalovirus (CMV).


This was a retrospective analysis of targeted brain ultrasound examination and fetal brain MRI performed in fetuses diagnosed with CMV infection following proven maternal primary infection. The prenatal findings were compared with findings from postnatal transfontanellar ultrasound examination during the first week following delivery or from postmortem when the pregnancy was terminated.


Both targeted prenatal ultrasound and MRI were performed on 49 fetuses. Brain abnormalities were present in 15/49 (30.6%) cases at postnatal/post-mortem follow-up. Fetal cerebral abnormalities were observed in 19/49 (38.8%) cases by ultrasound and/or MRI. The most frequent cerebral lesions induced by CMV and seen on ultrasound and MRI, respectively, included ventricular dilatation in nine and five cases, subependymal cysts in two cases each, microcephaly in five and three cases and periventricular calcifications in five cases on ultrasound only. Termination of pregnancy was performed in 10/49 cases. Sensitivity, specificity and positive and negative predictive values for the presence of cerebral lesions were 88.9%, 93.3%, 88.9% and 93.3%, respectively, when both prenatal ultrasound and MRI findings were abnormal, 85.7%, 85.3%, 70.6% and 93.5%, respectively, for ultrasound alone, and 42.9%, 91.2%, 66.7% and 79.5%, respectively, for MRI alone. Prenatal ultrasound, MRI and postnatal or postmortem examinations were concordant with the presence of brain abnormalities in six cases; however, their conclusions were exactly concordant in only two (33.3%) of these cases. In cases without cerebral abnormality, the results of prenatal and postnatal/postmortem examinations were concordant in 28/34 cases.


The addition of MRI to ultrasound increases the positive predictive value for the diagnosis of fetal brain abnormalities in fetuses with CMV. The two techniques appear to be complementary and should not be mutually exclusive in high-risk fetuses. Their high predictive value for the presence or absence of cerebral lesions provides a useful tool for appropriate counseling since current evaluation of the prognosis is based mainly on the presence of fetal brain lesions. The lack of concordance between ultrasound and MRI should stimulate standardization of the interpretation of both ultrasound and MRI prospectively. Copyright © 2008 ISUOG. Published by John Wiley & Sons, Ltd.


Cytomegalovirus (CMV) is the leading cause of congenital infection in neonates1. Around 50% of pregnant women are not immune at the onset of pregnancy and 1% will develop primary infection, whereas the rate of secondary infection is around 5% in immune pregnant women2. The vertical transmission rate is around 30% after primary infection and around 3% after secondary infection1. At birth, 10% of infected fetuses are symptomatic with cytomegalic inclusion disease and/or already established neurological abnormalities3. Around one third of these fetuses will die and more than half of the survivors will develop sequelae. Infants who are asymptomatic at birth can also develop sequelae, mainly neurosensory hearing loss4. The main difficulty in the prenatal assessment of CMV-infected fetuses is therefore that of accurate prognostic evaluation8. The presence of fetal brain abnormalities at ultrasound examination is currently the most specific predictor of a poor prognosis5, and leads to termination of pregnancy (TOP) in most cases.

Recent progress in the field of magnetic resonance imaging (MRI) has contributed to the development of detailed fetal imaging. Its use has gained widespread acceptance based on the assumption that it could provide additional information to ultrasound alone6, and fetal brain examination is the most common indication for fetal MRI7. However, its actual contribution to the diagnosis or prognosis of fetal disease is poorly evaluated, especially in fetal infections.

The aim of our study was to review the conclusions of prenatal ultrasound and MRI examinations for the presence of fetal brain abnormalities in fetuses with CMV when compared with postnatal or postmortem findings.

Subjects and Methods

We analyzed retrospectively all cases of fetal CMV infection reported in our database, covering two fetal medicine units over a 7-year period (2000 to 2007), and selected the cases that underwent fetal cerebral MRI together with serial ultrasound examinations, as well as transfontanellar ultrasound (TFU) examination at birth. Autopsy findings were considered instead when the pregnancy had been terminated. Seventeen cases treated with valaciclovir were included in this study, 13 of which have been reported previously9.

Cases were referred to our unit when maternal seroconversion was diagnosed or when ultrasound findings suggestive of fetal infection were observed fortuitously. Serological screening for CMV is performed during pregnancy in around 30% of pregnancies in France9, although this has been advised against by public health authorities due to the low risk and lack of treatment9. Congenital infection was considered certain when the viral genome was detected by PCR of the amniotic fluid retrieved by amniocentesis10 and confirmed at birth by the detection of the CMV genome either in fetal tissue following TOP or in neonatal urine samples within the first 2 postnatal weeks11.

Our prenatal management in all cases of CMV-infected fetuses includes serial targeted transabdominal or transvaginal brain ultrasound examination every fortnight, from the time of diagnosis until delivery. Fetal blood sampling is offered in all cases for analysis of hematological and biochemical parameters and fetal brain MRI examination is performed in all cases. Results of fetal blood parameters in this series have been reported elsewhere12.

The decision to continue or terminate the pregnancy was made by the parents after receiving extensive information on the fetal status, based on the findings of ultrasound and brain MRI examinations and of any biological abnormalities if fetal blood samples had been analyzed. According to French law, a parental request for TOP was accepted up to term if adverse outcome could be expected as a result of infection-induced abnormalities. In pregnancies continued to term, clinical, biological and audiological assessments as well as TFU were performed systematically at birth. In all pregnancies that were terminated, postmortem examination was performed and both macroscopic and microscopic results are reported. Assessment of the severity of the disease was based on the extent of brain involvement and the likelihood that neurological abnormalities would have developed if the pregnancy had continued.

Prenatal ultrasound and magnetic resonance imaging

High-frequency probes were used for the transabdominal (4–8 MHz) and transvaginal (5–9 MHz) ultrasound examinations (GE Voluson 730, GE Medical Systems, Ultrasound and Primary Care Diagnostic, Gif sur Yvette, France). The targeted neurosonographic examinations were performed by experienced operators, and included serial transverse, sagittal and coronal views obtained through the fontanelle (by TFU). The transvaginal approach was used unless the presentation was breech. Screening for fetal CMV-induced cerebral lesions was based on previously described sonographic features13.

A harmony 1-Tesla MRI machine (MRI Siemens Syngo, Erlangen, Germany) was used for the fetal brain MRI examination, using a phased-array abdominal coil. The characteristics and MRI set-up included T1-gradient echo with water excitation, T2-half-fourier single-shot turbo spin echo (HASTE) and, when necessary, gradient echo T2 (T2*) and diffusion echo planar imaging (EPI) and weighted imaging (DWI). MRI was performed in three orthogonal planes.

Statistical analysis

Sensitivity (proportion of true positives identified correctly by the test), specificity (proportion of true negatives identified correctly by the test), positive predictive value (proportion of patients with positive test results who are diagnosed correctly) and negative predictive value (proportion of patients with negative test results who are diagnosed correctly) of prenatal ultrasound and MRI examinations of the fetal brain were calculated as described previously14, 15. To investigate interobserver variability, a second experienced radiologist unaware of the ultrasound findings reviewed 25 arbitrarily chosen MRI examinations.


We identified 49 fetuses presenting with confirmed CMV infection over a 7-year period which fulfilled all inclusion criteria. A further five cases were excluded due to incomplete data. The mean gestational age at CMV diagnosis was 31 (range, 27–38) weeks.

Ten of the 49 (20.4%) cases underwent TOP following diagnosis of fetal brain abnormalities at ultrasound and/or MRI examination, after extensive counseling and parental request. Autopsy was performed in eight of these and was declined by the parents in two cases. Disseminated CMV infection was confirmed in all cases in which autopsy was performed. The remaining 39 fetuses were liveborn. One neonate died, but no adverse event occurred in the neonatal period in the other 38 cases.

Cerebral fetal lesions were observed by prenatal ultrasound and/or MRI examination in 19/49 (38.8%) cases. Valaciclovir treatment was used in 17/49 (34.7%) cases and fetal brain abnormalities were observed by prenatal ultrasound and/or MRI in seven of these cases8. Features of fetal brain lesions are reported in Table 1. Among the 38 CMV-infected newborns, fetal brain abnormalities were observed at TFU in four (10.5%). The distribution of the 49 fetuses based on prenatal imaging findings at ultrasound and MRI and their outcome is reported in Table 2.

Table 1. Numbers of cytomegalovirus-related fetal cerebral abnormalities observed by ultrasound (US) and by magnetic resonance imaging (MRI) examinations in 49 affected fetuses
Fetal cerebral lesionUS (n)MRI (n)
  • *

    Microcephaly defined as cephalic perimeter < 5th percentile for gestational age.

Ventricular dilatation95
Subependymal cyst22
Intraventricular abnormality (synechiae, hemorrhage)02
Parenchymous abnormality (white matter and periventricular abnormalities except calcifications)02
Gyration, cortical or neuronal migration abnormality01
Periventricular calcification50
Linear lenticulostriatal echogenicity10
Cerebellar abnormality01
Hyperechogenicity of thalami or ventricular walls30
Table 2. Distribution of 49 cases infected with cytomegalovirus in relation to prenatal imaging findings (ultrasound (US) and magnetic resonance imaging (MRI)) and outcome
 Total n (TOP; live birth)
  1. Values are all n. +, presence of fetal brain abnormality on imaging; −, absence of fetal brain abnormality on imaging (normal); TOP, termination of pregnancy.

US+9 (5; 4)9 (5; 4)18 (10; 8)
US−1 (0; 1)30 (0; 30)31 (0; 31)
Total10 (5; 5)39 (5; 34)49 (10; 39)

Comparison between targeted ultrasound and MRI (Table 3)

Normal ultrasound and normal MRI: 30/49 (61.2%) cases

In 30/49 cases, no abnormality was diagnosed on ultrasound or MRI. All 30 were delivered after 37 weeks of gestation. Postnatal TFU examination was normal in 28/30 (93.3%) cases. In two cases, hyperechogenicities were seen in the wall of the lenticulostriate vessels.

Table 3. Prenatal findings on brain imaging (ultrasound (US) and magnetic resonance imaging (MRI)) and postmortem/postnatal examination results in 49 fetuses infected with cytomegalovirus
CasePrenatal cerebral findingsOutcomePostmortem/postnatal cerebral findings
USMRIFetal brain autopsyTF-US
  1. –, absence of fetal brain abnormality on imaging; +, presence of fetal brain abnormality on imaging; BL, bilateral; HLSV, hyperechogenicity of the wall of the lenticulostriatal vessels; NND, neonatal death; PV, periventricular; SEC, subependymal cyst; TF, transfontanellar; TOP, termination of pregnancy; UL, unilateral; VD, ventricular dilatation.

US − / MRI−     
 1–2800Live birth 0
 29, 3000Live birth HLSV
US − / MRI +     
 31 MicrocephalyLive birth 0
US + / MRI−     
 32Microcephaly,  PV calcifications0TOP 28 weeksBrain calcifications, neuronal heterotopia 
 33UL VD 11 mm0Live birth 0
 34Thalamus hyperechogenicity0TOP 36 weeksAutopsy declined;  postmortem US:  thalamus hyperechogenicity 
 35Microcephlay0NND 32 weeksBrain calcifications 
 36SEC0Live birth 0
 37HLSV0Live birth 0
 38UL VD 10 mm,  PV calcifications0TOP 31 weeksVD, brain calcifications 
 39VD, hyperechogenicity of the ventricular walls0TOP 29 weeksVD, PV calcifications 
 40BL VD0TOP 24 weeksBL VD 
US + / MRI +     
 41VD, hyperechogenicity of the ventricular wallsBL SEC, cerebellar hypoplasia, microcephalyTOP 30 weeksMicrocephaly, UL SEC 
 42UL VD 11 mmUL VD 12 mmLive birth 0
 43Microcephaly,  PV calcificationsMicrocephalyLive birth HLSV
 44BL VD 11 mm,  PV calcificationsBL VD 12 mm, PV hypersignal (T2)TOP 36 weeksBL VD, necrosis and calcifications around the ventricles, inflammatory foci in the white matter and inside the midbrain 
 45Microcephaly,  PV calcificationsVD, cortical development abnormality, PV hypersignal (T2)TOP 34 weeksMicrocephaly, nodular heterotopia calcifications, neuronal development delayed 
 46SECBL SECLive birth 0
 47MicrocephalyMicrocephalyLive birth Microcephaly
 48BL VDBL VD 13 and 11 mm, intraventricular hemorrhageTOP 38 weeksBL VD, intraventricular hemorrhage 
 49BL VDBL VD, intraventricular hemorrhageTOP 36 weeksAutopsy declined; no postmortem US 

Abnormal ultrasound and abnormal MRI: 9/49 (18.4%) cases

In nine cases, fetal brain abnormalities were observed at both ultrasound and MRI examinations. Five of these cases underwent TOP on suspicion of severe brain damage. Autopsy confirmed disseminated CMV infection and brain lesions in four cases but was declined by parental request in one case; this case was confirmed on amniocentesis. The pregnancy continued to term in the other four cases. Postnatal TFU examination was normal in two cases, although unilateral ventricular dilatation was seen in one case and the presence of sub-ependymal cysts was seen in the other at both ultrasound and MRI examinations. In the third case, which had periventricular calcifications on prenatal ultrasound, neurosensory hearing loss was diagnosed at birth but TFU examination was normal. The fourth neonate presented with microcephaly, which had been suspected on both ultrasound and MRI, and severe deafness and was diagnosed with incontinentia pigmenti, a rare X-linked dominant neurocutaneous syndrome in which microcephaly and bilateral deafness are frequently observed. This disease led to difficulties in determining whether the postnatal findings were associated with the CMV infection or the genetic disease.

Abnormal ultrasound and normal MRI: 9/49 (18.4%) cases

In nine cases, abnormalities were observed only at ultrasound examination. TOP was performed in five of these cases, in all of which disseminated CMV infection was confirmed. One neonate died following cordocentesis and fetal distress at 33 weeks of gestation, indicating emergency Cesarean section. This fetus had microcephaly at prenatal ultrasound examination. The autopsy confirmed brain involvement with encephalitis. Three pregnancies continued to term and all three neonates showed no abnormality on TFU examination; their prenatal ultrasound features included unilateral ventricular dilatation (11 mm) and hyperechogenic bowel in one case, subependymal cysts in another and hyperechogenicities of the wall of the lenticulostriatal vessels in the third.

Normal ultrasound and abnormal MRI: 1/49 (2%) cases

In one case, only fetal MRI suggested microcephaly. The pregnancy was continued to term and no fetal brain abnormality was observed at postnatal TFU examination.

Table 4 summarizes the sensitivity, specificity and positive and negative predictive values of ultrasound and MRI for the prenatal diagnosis of cerebral lesions. The best positive predictive value (88.9%) was obtained with a combination of abnormal ultrasound and abnormal MRI findings and the best negative predictive value (93.5%) was obtained by normal ultrasound findings.

Table 4. Sensitivity, specificity and positive (PPV) and negative (NPV) predictive values of ultrasound examination (US) and magnetic resonance imaging (MRI) in predicting the presence of fetal brain abnormalities (confirmed at autopsy or postnatal transfontanellar ultrasound examination) in 49 fetuses affected with cytomegalovirus
Method of detectionSensitivity (%)Specificity (%)PPV (%)NPV (%)
  • Cerebral abnormality detected:

  • *

    on US irrespective of MRI findings;

  • on MRI irrespective of US findings;

  • on both US and MRI (including cases with discordant features of the cerebral findings);

  • §

    on US and/or MRI.

US and MRI88.993.388.993.3
US and/or MRI§86.784.872.293.3

Interobserver variability in the interpretation of fetal brain MRI in 25 cases

In the 25 cases used for investigation of interobserver variability in the interpretation of fetal brain MRI, pregnancy continued to term in 20 and TOP was performed in five. Brain abnormalities were present at postnatal TFU or postmortem examination in seven (28%) of the cases. Overall, both radiologists were in agreement with postnatal and/or postmortem examination in 16/25 (64%) cases. However, they reached different conclusions in four cases with an abnormal brain at birth and there was no strict correlation between their conclusions and the postnatal or postmortem findings in these cases.


The prognosis of fetuses infected with CMV is difficult to establish. Counseling is based mainly on prenatal imaging and the decision to terminate an affected pregnancy is based mainly on the presence or absence of cerebral lesions on targeted serial ultrasound examinations. The results of this series of 49 CMV-infected fetuses investigated by both TFU and MRI examinations suggest that the addition of MRI increases the positive predictive value of ultrasound alone for the diagnosis of fetal brain abnormalities. The two techniques appear to be complementary and should not be mutually exclusive. However, normal prenatal TFU examination had the best negative predictive value and isolated abnormalities suspected on MRI alone should be considered with great caution.

Guerra et al. recently reported that the ability of ultrasound examination in fetuses known to be infected with CMV to predict symptomatic newborns had a positive predictive value of 78.3% in a cohort of 154 infected fetuses5. This was based on overall ultrasound findings, not only on cerebral findings. However, there was a lack of concordance in the description of cerebral lesions on ultrasound and on MRI, although the examinations were performed within days of each other. One limitation of Guerra's study is that only a few pregnancies with CMV-infected fetuses with brain abnormalities were not terminated and the correlation of prenatal images with postmortem examination findings is difficult. The relatively poor concordance between ultrasound, MRI and postnatal/postmortem findings can be explained by the different approaches to the brain anatomy by ultrasound, MRI, macroscopic and microscopic examinations. Calcifications, one of the most typical CMV-induced cerebral lesions, can be depicted easily by ultrasound, while calcifications observed later, at postmortem microscopic examination, are probably much smaller and therefore harder to observe. Cortical development abnormalities were picked up by MRI and confirmed at postmortem examination although some ultrasound examinations were normal. Ultrasound examination of the cortex is limited by the necessity to visualize the brain through bone windows such as the fontanels, and this is increasingly difficult as gestation advances. Comparison between the three methods is therefore methodologically limited and the diagnostic value of prenatal imaging overall is more relevant than a precise description of the lesions retrieved by each method.

Several studies have reported the superiority of MRI over ultrasound to identify various kinds of fetal brain anomalies. However, very few of these cases were explored by MRI in the context of infection when compared with ultrasound of the fetal brain16–18. Malinger et al.19 compared dedicated neurosonography and fetal brain MRI examination in 42 cases with fetal brain abnormalities, and compared the results with the final diagnosis obtained by autopsy after TOP or by neuroimaging at birth. The sensitivity, specificity and positive and negative predictive values for neurosonography and MRI, respectively, in predicting postnatal outcome were 96% and 85%, 87% and 80%, 93% and 88%, and 93% and 75%. However, only one case of CMV fetal infection was reported in this series. Overall, they concluded that dedicated neurosonography was equal to MRI in the diagnosis of fetal brain abnormalities, and considered the major role of MRI to be one of reassurance for the parents regarding the presence or absence of brain abnormalities.

Our study underlines the need for sonographic features to be known by the radiologist at the time of MRI examination. The best concordance between observers was observed for microcephaly and ventricular dilatation, which were described by both radiologists. In two cases periventricular calcifications were observed by ultrasound but only the radiologist aware of the ultrasound findings reported a T2 hypersignal at the same location. The limitations of the prognostic value of imaging methods alone have led us to evaluate the value of hematological and biochemical markers in fetal blood, to be interpreted together with imaging findings12.

We suggest that in CMV infection, MRI is a reliable method for fetal brain examination, and can be performed to confirm or rule out the presence of CMV-related fetal brain abnormalities. Further studies are lacking, mainly because of the low prevalence of the disease. These must be undertaken to determine the predictive value of subtle or isolated cerebral features; it is not known whether mild or unilateral ventricular dilatation, isolated calcification and isolated subependymal cysts correlate consistently with a poor outcome.